Studies in Chinese Language: Volume 8

Studies in Chinese Language

The Collected Works of M. A. K. Halliday Volume 1: On Grammar Volume 2: Linguistic Studies of Text and Discourse Volume 3: On Language and Linguistics Volume 4: The Language of Early Childhood Volume 5: The Language of Science Volume 6: Computational and Quantitative Studies Volume 7: Studies in English Language Volume 8: Studies in Chinese Language Volume 9: Language and Education Volume 10: Language and Society

Volume 8 in the Collected Works of M. A. K. Halliday

Studies in Chinese Language M.A. K. Halliday Edited by Jonathan Webster

A\ continuum

W L O N D O N • NEW YORK

Continuum The Tower Building 11 York Road London SE1 7NX

80 Maiden Lane Suite 704 New York NY 10038

First published 2005 This collection © M. A. K. Halliday 2006 All rights reserved. No part of this publication may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopying, recording, or any information storage or retrieval system, without prior permission in writing from the publishers.

British Library Cataloguing-in-Publication Data A catalogue record for this book is available from the British Library. • ISBN 0-8264-5874-2 (hardback) Typset by RefmeCatch Limited, Bungay, Suffolk Printed and bound in Great Britain by MPG Books Ltd, Bodmin, Cornwall

CONTENTS

Preface

vii

Acknowledgements x PART ONE

x

MEDIEVAL CHINESE GRAMMAR

Editor's Introduction 1 The Language of the Chinese "Secret History of ' " (1959) the Mongols 1 The text 2 Method of analysis 3 Contextualization 12 4 Graphic analysis 5 Script and language 6 Grammar 7 Lexis 8 Phonology and transcription 9 Appendices A A phonological (prosodic) analysis of the new Chinese syllable (Modern Pekingese) B Specimen analysis and translation C Statistical tables D Lexical translation forms collocation of Mongolian names E Suggested emendations F Chinese syllabary and table of characters

v

3

5 5 8 12 20 24 37 112 116 128 128 135 137 159 160 162

CONTENTS

PART TWO MODERN CHINESE GRAMMAR

Editor's Introduction

175

2 Temporal Categories in the Modern Chinese Verb (with J. O. Ellis, 1951)

177

3 Grammatical Categories in Modern Chinese (1956)

209

4 Some Lexicogrammatical Features of the Dialects of the Pearl Paver Delta

249

PART THREE CHINESE PHONOLOGY Editor's Introduction

273

5 The Origin and Early Development of Chinese Phonological Theory (1981)

275

6 A Systemic Interpretation of Peking Syllables (1992)

294

PART FOUR GRAMMAR AND DISCOURSE Editor's Introduction

323

7 Grammatical Metaphor in English and Chinese (1984)

325

8 Analysis of Scientific Texts in English and Chinese (1993) 334

Chinese (1993) 334

9 On the Grammatical Foundations of Discourse (2001)

346

Appendix

365

Bibliography

368

Index

379

VI

PREFACE

To fully appreciate the papers in this volume, Studies in Chinese Language, one needs to know something of Professor Halliday's experience as a young man in China leading up to his return to England in 1950 to complete his Ph.D. (For a more complete biographical account of his early years, see Webster 2005.) After leaving the army in 1947, Professor Halliday decided to pursue his degree from the University of London externally in China. In that same year, he enrolled as a student at Peking University, where he took classes in the Chinese Department. A year later, 1948, he flew to Nanking, where the British Council had made the necessary arrangements for him to take the University of London examination, which was on Modern Chinese: a combination of language and literature, including the history of Chinese literature from 500 BC to the present day. After completing his London degree, he took a job in China working for the Chinese Industrial Cooperatives. He worked for them for about six months until; as he tells it, in some small village in Gansu a letter arrived which had been chasing him for about three months which said he had been given a scholarship from England for postgraduate study. 'Proceed back to Peking immediately', began the letter, informing Professor Halliday that he had just been awarded a scholarship to undertake postgraduate study back in England. However, he was allowed to spend two more years studying in China before returning to England. The next six months were spent at Peking University, studying under Professor Luo Changpei, who started training him in historical linguistics and Sino-Tibetan studies. After about six months or so, however, it became clear to Luo that Professor Halliday was really more interested in working on Chinese dialects, so Luo sent him to work with Professor vii

PREFACE

Wang Li, then Dean of the Faculty of Arts at Lingnan University in Guangzhou (Canton), who was at that time carrying out a phonetic survey of the varieties of Cantonese of the Pearl River Delta region of central Guangdong province. In addition to contributing to the ongoing phonetic investigation, Professor Halliday also developed a grammar questionnaire which he used to elicit versions of Cantonese sentences in local dialects (see Chapter Three, 'Some Lexicogrammatical Features of the Dialects of the Pearl River Delta' for an account of his findings from that study; also samples of his field notes are included on the accompanying CD-ROM). When Professor Halliday returned to England from China in 1950 to complete his Ph.D., he thought he would be studying at the School of Oriental and African Studies (SOAS) under the supervision of Professor J. R. Firth, and working on the material from his dialect work with Wang Li. But England in 1950 was at the height of McCarthyism, so even though he was not a member of the Communist Party, the fact that he refused to undertake not to become a member prevented him from being admitted to SOAS. Instead, he was admitted to the Chinese Department at Cambridge, where the emphasis was on classical, not modern Chinese. Realizing Professor Halliday's preference for studying modern Chinese dialects, his supervisor at Cambridge, Professor Gustav Haloun, suggested, as a compromise, that he work on the Chinese translation of the Secret History of the Mongols. Haloun also allowed him to golation of the Secret History of the Mongols. Haloun also allowed him to go to London to study with Firth, who had agreed to take him on for informal supervision. This arrangement was soon to change, however, when Haloun suddenly died at the end of the year. Professor Halliday asked Firth if he would be willing to become his official supervisor, if it could be arranged. Firth agreed, and Professor Halliday, although still a student at Cambridge, was allowed to transfer to Firth's supervision, travelling regularly to SOAS. Thus the dedication to Professor Firth in the preface of the book based on his doctoral dissertation, 'The Language of the Chinese "Secret History of the Mongols"' (see Section One, Medieval Chinese Grammar), which reads 'I was very fortunate to have Professor Firth to guide me in the preparation of the present work, and it is a great pleasure to me to be able to dedicate it to him.' This volume is divided into four sections, the first of which, 'Medieval Chinese Grammar', is the book based on his doctoral dissertation, 'The Language of the Chinese "Secret History of the Mongols'". Section Two, 'Modern Chinese Grammar', contains two previously unpublished papers, Chapters Two and Four, in addition to one of Professor Halliday's first published papers, 'Grammatical Categories in Chinese' (1956), a viii

PREFACE paper published just a year after completing his Ph.D., and three years before the publication of his doctoral dissertation. Section Three, 'Chinese Phonology', contains two papers, one offering a historical perspective, Chapter Five, 'The Origin and Early Development of Chinese Phonological Theory' (1981), the other, Chapter Six, 'A Systemic Interpretation of Peking Syllables' (1992), reflects the influence of his early training in traditional Chinese phonology under Luo and Wang, and subsequent experience with prosodic phonology under Firth. The fourth and final section, 'Grammar and Discourse', presents three papers which are more comparative in orientation, looking at the relation of discourse to the lexicogrammar in both English and Chinese.

ix

ACKNOWLEDGEMENTS

We are grateful to the original publishers for permission to reprint the articles and chapters in this volume. Original publication details are provided below, and also at the beginning of each chapter. The Language of the Chinese 'Secret History of the Mongols' 1956. Copyright © Blackwell Publishing. 'Grammatical categories in Modern Chinese', from Transactions of the Philological Society 1956, pp. 177-224. 'The origin and early development of Chinese phonological theory', from Towards a History of Phonetics, edited by R. E. Asher and Eugenie J. A. Henderson (1981). Copyright © Edinburgh Press, www.eup.ed.ac.uk 'A systemic interpretation of Peking syllables', from Studies in Systemic Phonology, pp. 98-121 (1992). Copyright © Pinter, London. 'Grammatical Metaphor in English and Chinese', from New Papers on Chinese Language Use, edited by Beverly Hong (1984). Copyright © Research School of Pacific Studies, The Australian National University. 'The analysis of scientific texts in English and Chinese', from Writing Science: Literacy and Discursive Power. This chapter was originally presented at the International Conference on Research in Texts and Language, Xi'an Jiaotong University, March 1989. With kind permission of the Taylor and Francis Group. 'On the grammatical foundations of discourse', from Grammar and Discourse: Proceedings of the International Conference on Discourse Analysis, edited by Ren Shaozeng, William Guthrie and I. W Ronald Fong (2001). Reprinted with permission from the University of Macau. 'The Secret History of the Mongols' on the accompanying CD-ROM is included with the kind permission of the Shanghai Classics Publishing House. x

PART ONE MEDIEVAL CHINESE GRAMMAR

This page intentionally left blank

EDITOR'S INTRODUCTION

The first section is comprised of Professor Halliday's doctoral disssertation, 'The Language of the Chinese "Secret History of the Mongols TC^^^.'", completed in 1955 for the degree of Ph.D. at the University of Cambridge, and first published by 'The Philological Society' in 1959. His aim was to provide 'a complete synchronic linguistic analysis', i.e. a descriptive grammar, of the Chinese translation of this personal biography of Genghis Khan known as the 'Secret History'. The text was originally written in Mongolian in the thirteenth century AD and published with the addition of a Chinese translation and Chinese Gloss at the end of the fourteenth century. The complete text is included on the accompanying CD, including (i) the Mongolian version, which is in fact the earliest Mongolian document of any considerable length, (ii) the Gloss, i.e. a complete word-for-word rendering of the Mongolian in Chinese, and (iii) the Chinese translation, the earliest known text in Mandarin Chinese. The reason it was in Mandarin, as Professor Halliday explains, was that it had been translated into Chinese to be used as a textbook for Chinese civil servants who had to learn Mongolian, the official language of the civil service. Professor Halliday's account of the 'The Language of the Chinese was intended, initially, to "Secret History of the Mongols be 'purely descriptive', with 'statements valid for this language' arranged in a series of six sections, each dealing with a particular level of descriptive analysis: contextualization, graphics analysis, script and language, grammar, lexis, and phonology and transcription. The Chinese "Secret History of the Mon ed the role of linguistic informant, or, in fact, as Professor Halliday notes, 'an informant who is himself a linguist, in virtue of the plurilingual character of the text'. Professor 3

MEDIEVAL CHINESE GRAMMAR Halliday also envisaged his description 'may form part of a historical analysis composed of a series of descriptive statements arranged in time-sequence; and that the description, or parts of it, may be subjected to comparative treatment in juxtaposition both with other descriptive parts of the same historical statement and with other descriptive statements of languages where comparison will show systematic relationship'. Thus he indicates the categories of units, classes and functions used for making statements about the language may be compared with those 'set up for Modern Pekingese in my "Grammatical Categories in Modern Chinese", pp. 217—18' (see Chapter 3 of this volume).

4

Chapter One

THE LANGUAGE OF THE CHINESE "SECRET HISTORY OF THE MONGOLS

(1959)

SECTIONS 1-5 1 The text 1 1 The "Secret History of the Mongols" is the name given to a composite text, originally written in Mongolian in the thirteenth century AD and published with the addition of a Chinese translation and Chinese Gloss at the end of the fourteenth century. The Chinese title of the whole work, and of the Chinese translation as a separate work, is Yuan-ch'ao pi-shih (1). The Mongolian version probably at first had no title, though the first three words of the present text may originally have been the title of at least part of the work;1 but the Mongolian title now in current use, Mangqol-un muca tobca 'an, of which the Chinese is a translation, was at some time given to the Mongolian version and later figures as the sub-title of the composite text. 1.2 The "Secret History" is a personal biography of Genghis Khan, beginning with an outline of the legendary history of the Mongol people and extending, in the last two chapters (designated in one tradition as "supplementary"), to the first part of the reign of Genghis' successor Ogodai Khan. The exact date of composition, as also that of the Chinese translation, is uncertain, the final sentence states that the book was completed in July of the year "Rat", and this has generally been assumed2 to refer to the year 1240. Recently, however, Hung has shown that the year referred to may well be later by one or more twelveyear cycles, himself regarding 1264 as the most likely date (Hung, 1951: 487—92) The Chinese translation is thought to have been made not less than a century later, probably soon after the fall of the last Mongol The Language of the Chinese 'Secret History of the Mongols', M A K Halliday (1956) Copyright © Blackwell Publishing

5

MEDIEVAL CHINESE GRAMMAR

emperor in Peking (1367) but possibly as late as 1404 (Hung 1951: 449-69; Pelliot 1949 (preface dated 1920); Haenisch 1931: Section 1, Foreword,p. I.3 The composite text was preserved in China in various manuscript copies and prints, ultimately derived from two separate but not widely divergent originals: the (manuscript encyclopedia) Yung-lo ta-tien (2) (1403—08) edition, now lost, and an early Ming print, roughly contemporary and probably taken from the same original version, of which 41 pages have been preserved. The chief point of difference between the two is that the Yung-lo ta-tien version is divided into 15 chapters, while the other tradition has 12 (10 plus 2 supplementary).4 Some details of the various copies made of the text in the subsequent period are given below.5 In 1936 the Commercial Press, Shanghai, brought out an edition of the composite text photo-lithographically reproduced from a copy6 of the 12-chapter version, with the 41 surviving pages of the original print inserted in place of the corresponding pages of the manuscript copy; it is this edition, forming part of the third series of the great Commercial Press library Ssu-pu ts 'ung-k 'an (3), which has been used for the purposes of the present study. 1.3 The composite text contains the Mongolian version, the Chinese version and the Chinese Gloss. The Mongolian version is written in what we may call "Sino-Mongolian" script: that is, by means of a limited number of Chinese characters used to represent the phonological system of the Mongolian more or less syllabically.7 The entire text is thus written in Chinese characters. The Gloss, which is interlinear to the Mongolian version, is a complete word-for-word rendering of the Mongolian in Chinese. The Chinese version, which has also been published separately and a commentary added,8 is a continuous text divided into paragraphs with each paragraph following the Mongolian passage of which it is a translation;9 it is actually an abridged translation, some passages of the original being left out, though more often the abridgment takes the form of precis rather than of direct omission.10 1.3.1 The Mongolian version is the earliest Mongolian document of any considerable length, and contains many passages in verse, much of it doubtless traditional. The Mongolian text, or an earlier version of it, was used as source material for other contemporary histories; but after the fall of the Mongol empire in China, when the Sino-Mongolian script was no longer used, it was preserved only in the composite text and received little or no further attention until the twentieth century. Since 1900, four separate transcriptions have been made into romanized Mongolian, by Pelliot (1949, verse passages are indicated 6

"SECRET HISTORY OF THE MONGOLS' typographically), Haenisch (1937), Kozin (1941) and Shiratori (1942), each, curiously, working independently of the other three;11 and translations have been made by Naka (1907)12 into Japanese, and by Haenisch (1948) into German and Kozin (1941)13 into Russian. There is finally an incomplete French translation by Pelliot (1949).14'15-16'17 1.3.2 The question of the script in which the original version of the Mongolian text was written down is a matter of uncertainty. Palladius (Kafarov), the first European scholar to pay attention to the text, thought that the original version was written in Uighur-Mongolian script and had been lost (Palladius 1910); and this view, perhaps the most widely accepted (Haenisch 1948: 1; Pelliot 1948: 231; Shiratori 1942: Editor's Foreword, p. 1), was reinforced by the discovery by Pelliot in the late 1920s in the library of Ulan-Bator of a manuscript containing large parts of the "Secret History" in Uighur-Mongolian script (Pelliot 1930).18 This manuscript was published in two volumes in Ulan-Bator in 1938 and is regarded by Kozin, who gives a transcription of it in the same volume with his two transcriptions of the "Secret History", as an attempt by seventeenth-century Mongol Buddhist scholars to rewrite the Mongolian text from the Sino-Mongolian script into the then current Uighur-Mongolian script; Kozin (1941: 25) holds the view that the only original form of the text was the Sino-Mongolian version, a view which Franke (1952: 7-10) likewise regards as a possibility. A third view, held by Hattori (1946), is that the original version was in hPags-pa script, while Poppe (1950) envisages the possibility that the UighurMongolian original was first transcribed into hPags-pa and thence into Sino-Mongolian. 1.3.3 Whereas the Mongolian version has for some time occupied the interest of linguists and historians, the Chinese version has until recently been little studied: being a translation, and moreover abridged, its main interest is in the field of Chinese linguistics. Here, however, it is of particular interest, in that it is composed not in the literary language but in what may by opposition be referred to as "colloquial". It has been translated once only, by Palladius (1910);19 but its unique position as a long continuous narrative in "colloquial" language of its date places it in the forefront of source material for descriptive and historical studies of the Chinese language. 1.4 The present work is concerned with the description of the language of the Chinese version. A cursory reading shows that the language is not literary Chinese, though forms taken from the literary language, rare in the first half of the book, appear with much greater frequency in the final chapters; and that it resembles the Mandarin dialect of the 7

MEDIEVAL CHINESE GRAMMAR

present day rather than any of the dialects of the south. It gives the impression of being racy and colourful, but without great stylistic merit (as indeed there was no recognized literary genre to which it could aspire and by which it could be judged). The language clearly permits comparison with Modern Pekingese, and one possible approach to its study would be by means of a detailed presentation of those points in which it differs from Modern Pekingese; but in view partly of the absence of a comprehensive systematic description of any form of the latter, and partly of the extent of the text, this being such as to allow a particular (that is, non-comparative) statement especially of its grammar and vocabulary, it has seemed preferable to attempt an exhaustive description of the language of the text in its own terms, reference being made to Modern Pekingese where points of comparison presented themselves. Those familiar with the language of other examples of "colloquial" Chinese literature, such as the Yuan drama and the Ming novels, will recognize many points of similarity. These have not been separately noted, as to do so exhaustively would require an equally thorough study of the language of such texts. It would indeed be desirable that further descriptive studies should be made of the language of some texts representative of these genres and periods; the present work might then contribute towards a systematic comparison and later a historical account linking the stages of North Chinese through a study of representative documents from the present day up to the time of the "Secret History of the Mongols". 2 Method of analysis 2.1 The present work is a descriptive grammar of the Chinese language of the "Secret History of the Mongols"; "descriptive grammar" is here taken to mean a complete synchronic linguistic analysis of the text. It is customary for a descriptive account to be divided into sections, with headings such as "phonology" and "syntax", dealing with different levels of the analysis. This method of division into sections has been adopted here; and the word "grammar" has been used — also in conformity with traditional usage — to denote one of the sections of the analysis. This ambiguity, in the use of the word "grammar" both in the characterization of the whole as "descriptive grammar" and in the heading of one of the sections, has been tolerated in preference to the use of "syntax" as a section heading, because of the implied opposition of "syntax" to "morphology", an opposition which it is not desired to make here. 8

"SECRET HISTORY OF THE MONGOLS' 2.1.1 The procedures of descriptive grammar are deduced from the principles of linguistic science in accordance with the place of descriptive analysis in the framework of general linguistics. The accustomed distinction into "descriptive" and "historical" (or "synchronic" and "diachronic") is perhaps inadequate to give the complete perspective to either, and an alternative scheme — necessarily one among many possible such schemes — could be drawn up for ordering the disciplines controlled by general linguistics in a two-dimensional framework: one axis would represent the scope of the study in terms of the material, with division into "particular", "comparative" and "universal" (according as the object of study is (part of) one language, a finite number of languages greater than one or all languages); the other would represent the attitude of the linguist to the material, whether "descriptive", "historical" or "evolutionary" (Halliday 1957:56). According to this a historical study is a series of descriptive studies (to which comparative techniques may have been applied) arranged according to a time-construct, whereas an evolutionary study is essentially a study of change. Thus a descriptive study of the type here undertaken will take its place in a historical study to be made up from a series of such descriptive statements. The totality of such statements is historical in the sense that they are ordered in a time-sequence; they can be subjected, in whole or in part, to a comparative analysis which will take this time-sequence into account. The resulting historical statement will not be evolutionary since it will not deal with change in certain isolated forms but with forms functioning as terms in successive synchronic systems. 2.1.2 The descriptive technique is applicable to all texts, of any extent, spoken or written; the description requires only that the text shall be circumscribed so that the statements refer to the language of the text and are made as valid only for that text. In addition to the delimiting of the text it is desirable that the aim and scope of the inquiry should be formulated, so that the statement can be correctly evaluated and that it may be clear how the choice between alternative methods of statement has been guided.20 The description may then be characterized as one method, selected for simplicity and comprehensiveness, of statement of the material, by its systematization into formally marked categories; the categories have no "universality" but are set up to handle the meaning of the text in linguistic (and not, for instance, propositional) terms. 2.2 The procedure to be adopted can be generalized into a system of techniques which have evolved out of the scientific practice of linguists from the ancient world down to the present, to which all cultures and all periods have contributed. Like the techniques of other sciences they 9

MEDIEVAL CHINESE GRAMMAR reflect and, at the same time, influence the scientific trends of their times, and advances made by divergent "schools" of linguistics have become the common ground of all. Progress in recent decades has been considerable, and it is significant that in 1951 an international commission meeting in Paris under the auspices of the International Council for Philosophy and Humanistic Studies of UNESCO was able to draw up certain recommendations, by following which linguists engaged in descriptive grammar could ensure that their work would rest on a sound basis of up-to-date linguistic theory. 2.2.1 Firth (1951), in his introduction to the Report of the Commission, commented on the degree of progress and extent of agreement in linguistics that made the recommendations possible. "During the last thirty years linguistics has made great advances and today may be said to be in the van of the social sciences" (p. 70). "All the various streams contribute in varying volume and strength to present-day schools of linguistics, each of which has well-marked characteristics and specialized interests. But in the fundamentals of descriptive grammar there is general agreement which is to some extent expressed in the recommendations of the Paris symposium which provide the occasion for this article" (p. 73). Basing himself on these "fundamentals", the linguist can approach his material and state his results in such a manner that his work is of the greater value to others in his field. 2.2.2 Of the treatment of the material, Firth says: 'Language text must be attributed to participants in some context of situation in order that its modes of meaning may be stated at a series of levels, which taken together form a sort of linguistic spectrum. In this "spectrum" the meaning of the whole event is dispersed and dealt with by a hierarchy of linguistic techniques descending from social contextualization to phonology' (p. 76). A descriptive grammar of the language of a given text deals with "the meaning of the whole event"; the meaning is not separate from or opposed to the linguistic form but is a function of the whole text. The complete text has meaning in the social context in which it operates, and this is to be stated by the procedure of "contextualization". Commenting on the Report, Firth points out: 'It will be noticed that "linguistic forms" are considered to have "meanings" at the grammatical and lexical levels, such "meanings" being determined by interrelations of the forms in the grammatical systems set up for the language' (p. 85). Each element has meaning 'with reference to the specific system of which it is a "term", "unit" or "member" in a given language' (ibid.). The words (the units of lexical statement) of the text function in interrelations of collocation and in "ordered series"; 10

"SECRET HISTORY OF THE MONGOLS' grammatical forms are ranged as terms in paradigmatic systems and operate in syntagmatic relations in the structure. The comprehensive statement of such relations is the statement of the meaning of the text at these levels. 2.2.3 The whole of the Report of the Commission serves to guide the linguist in his approach to the material. "The recommendations", it is stated in the Report, "are not put forward as a formal scheme of description. In each case the method of analysis and description to be employed should be synchronic in principle and appropriate to the structure of the language under examination. The intention is simply to bring out certain general and necessary requirements for the descriptions of the type desired" (pp. 77—8). A number of the detailed recommendations have a direct bearing on the present work: these need not be listed here, but the closing paragraph of Firth's commentary may well be stressed. "Finally," says Firth, "there is nothing revolutionary or strikingly new in the recommendations. The important thing is that linguists of most schools would, I imagine, agree that any account of a modern spoken language which followed the recommendations would today be considered a good descriptive grammar" (p. 87). The question to be asked here is: what then of the application of these recommendations to an account of the language of a written text some six centuries old? 2.3 If the recommendations are to be valid for this purpose the aim of the description must be congruent with the aims envisaged by the compilers of the Report. The task here undertaken is the presentation, by means of a comprehensive statement in its own terms, of the language of a certain text, the Chinese translation of the "Secret History of the Mongols". The aim of the account is, initially, purely descriptive: to make statements valid for this language. As a wider aim it is envisaged that the description may form part of a historical analysis composed of a series of descriptive statements arranged in time-sequence; and that the description, or parts of it, may be subjected to comparative treatment in juxtaposition both with other descriptive parts of the same historical statement and with other descriptive statements of languages where comparison will show systematic relationship.21 The task is prescribed for the purpose of the statement and in any case limits itself, since we have no other text of the period in the same language ("colloquial" not "literary") and style.22 The approach to the material is thus congruent with the aims envisaged by the Report. 2.3.1 The problem of application of the descriptive techniques as outlined in the recommendations lies in their application in general to a 11

MEDIEVAL CHINESE GRAMMAR

written text and in particular to a written text so complex in history and structure as the "Secret History of the Mongols". The recommendations made in the Report concerning the description at the grammatical and lexical levels present no special problems in their application to a written text as such. The order of analysis and of presentation of the material, where it is felt that the procedure should be from the larger to the smaller unit, beginning with the text as a whole and working towards the features of articulation, not only may be retained but is indeed the more desirable in handling a written text in that it involves moving from the known towards the unknown. At the two "outer" levels of contextualization and phonology, however, special problems arise in connection with a written text, and these will be dealt with in the respective sections (3 and 8). Moreover at least one other level of analysis is required by a written text, for statements to be made concerning the script, punctuation and so forth; and for the "Secret History of the Mongols" it has seemed desirable to discuss this aspect under two headings: "Graphic Analysis", which treats of the outward form of the written text, and "Script and Language", which treats of the script and its relation to the language of the text. The description has thus been arranged in a series of six sections, in each of which statements are made about the language of the text at a particular level of descriptive analysis: contextualization, graphic analysis, script and language, grammar, lexis, and phonology and transcription. 3 Contextualization 3.1 Any text functions in a context, from which can be abstracted certain features relevant to the descriptive analysis. The context of a written text of the past is more complex, and more difficult to evaluate and make abstraction from, than that of a contemporary spoken language text. A written text of the past can be said to have context in two dimensions, arising out of the effect of writing on the place of a linguistic event on the time-track. On the one hand, the text of the Chinese translation of the "Secret History of the Mongols" exists today as a historical work and work of Chinese literature, and as such is read and used by various people for various purposes. On the other hand, the text has been in existence for some six hundred years and has its own "history". It is not necessary to insist on "two contexts"; the point is rather that a written text can be contextualized at any moment from the time of its being written down, so that its context could be presented "diachronically" in a series of "synchronic" statements.23 12

"SECRET HISTORY OF THE MONGOLS' 3.1.1 Of the use of contextualization in linguistic analysis, Firth has said that the concept of context of situation "is best used as a suitable schematic construct to apply to language events, and that it is a group of related categories at a different level from grammatical analysis but rather of the same abstract nature. A context of situation for linguistic work brings into relation the following categories: A The relevant features of participants: persons, personalities. (i) The Verbal Action of the Participants. (ii) The Non-Verbal Action of the Participants. B The relevant objects. C The effect of the Verbal Action. Contexts of situation and types of language function can then be grouped and classified." Context is "parallel with grammatical rules, and based on the repetitive routines of initiated persons in the society under description" (Firth 1950). In the context of a written text the same basic categories require to be brought into relation, since they are related in the event as with a spoken text. There will, however, be differences of emphasis: different features require to be marked, since the factors determining the two types of linguistic behaviour (including their effects) are different. In the situation in which a written text operates the form taken by the linguistic activity is socially determined as with an utterance; but the fact that the one is written and not spoken itself imposes certain requirements on the elements composing the text, which the nature of the script modifies in various ways. The literary form suited to the social role of the text is determined, as also is the style; and literary form and style further impose their own requirements. 3.1.2 The most significant difference in the treatment, however, does lie in the "two-dimensional" contextualization. This can be seen in terms of the relation of the writer to the other participants in the situation.24 Since the creative effect of linguistic activity is indirect, that is, through the action of other human beings as participants, there is no (or very little) non-linguistic action on the part of a writer; the relevant features of the writer as a personality are observable only in his linguistic action (are internal, so to speak, to the text) so that, once the text is in existence, the writer ceases to function in the situation and is replaced by the text as "participant". In this way the linguistic activity of the writer continues to have creative effect long after it has taken place, through participants in an extended situation: extended not only by simple extension in time (for the creative effect of the spoken word can operate for long periods) but by the extension of the actor-participant 13

MEDIEVAL CHINESE GRAMMAR relation into a new dimension. This extension arises out of the nature of writing (though the tape-recorder has now done the same for speech) and has nothing to do with the written "style", although the latter may determine its creative effect. 3.2 The style of a written text is the sum total of the linguistic features through which the text operates in a specific function, this function being abstracted as common to a class of writing and marking it off from other classes. In this sense the "Secret History of the Mongols" is a narrative text. The word "history" did not figure in the original title of the Mongolian version (Hung 1951: 465).25 The Mongolian version is in fact by content a biographical rather than a historical narrative; moreover it includes many passages of traditional wording and of dialogue. Its style may be characterized as "narrative, partly traditional, with dialogue in direct speech".26 Since the Chinese version is a translation, the element of traditional narrative is lost, and the style is simple narrative. The dialogue remains throughout in direct speech. Differences in the language between passages of narrative and those of dialogue do occur; they are differences of style because of context (particularly the special context of dialogue in a narrative text) but must be taken into account in any statistical analysis.27 3.2.1 It is not known whether the Mongolian version was taken down from oral narrative or composed in writing; while the former may be the more probable, in either case in a language with no previous written literature the implication of narrative style is no more than free oral narrative with the influence of traditional narrative (the presence of traditional forms of utterance), much of which is clearly marked here by alliteration and chiming in initial syllables. For the Chinese version, however, the implication of narrative style is somewhat different. First, the Chinese version is indirect narrative: the author experienced the events he was narrating not directly but through the medium of a written text in another language. Second, the Chinese version was composed in writing. Third, Chinese was a language in which a long accumulation of written literature had replaced oral texts and developed a language and traditional style of its own. 3.3 The language of the "Secret History", however, is not "literary Chinese": this is in fact its significant feature, and therefore the language in which it is written is designated, by opposition to "literary", as "colloquial Chinese".28 3.3.1 The use of the term "colloquial" tends to carry its own implications, but it should not be taken in itself to imply a definite relation to any form of the spoken language of the time: indeed the relation of the 14

"SECRET HISTORY OF THE MONGOLS' language of "colloquial" written texts to that of speech has not yet so far as I know been stated clearly either in general or with reference to any particular texts. Nevertheless there is a clear, if partial, implication in the use of "colloquial" here. The linguistically active participant (speaker or writer) in any situation derives the forms of his expression from his own experience, and it may be possible to determine, to a certain extent, what were the limits of his experience. In a language with no written texts his experience must be purely auditory; this does not mean that it has no element of tradition, since there may be traditional oral texts whose language is marked as different from that of normal communication. In a language such as Chinese, however, with (by this time) a long written tradition, a writer tends to draw almost exclusively on his reading experience, the more so (since he must make a selection between the two) as the divergence increases between the forms of what he reads and those of what he hears. Some forms in the "Secret History", particularly in the later chapters, are clearly drawn from written texts; the language as a whole, however, cannot be traced to earlier written texts and must originate in the author's auditory experience. Thus in naming the language of the Chinese version "colloquial" we exclude the "literary language" and imply that it is based on the author's own direct experience of speech; and, since this is a narrative text, the strongest influence should be that of spoken narrative, possibly the style of the story-teller to the extent that the latter would admit of being adapted to a context of translation. 3.3.2 The identification of the language and style of the text provides a framework for its diachronic contextualization. As this contextualization will show, the relation of the "Secret History" to other works in the same style and language is somewhat unusual. The origin and history of the work have been discussed in general terms in Section 1. Much of the early history remains uncertain: it is not known at what point the three versions became a single text nor for how long any version was in existence before this happened. As noted above, Hung (1951: 465—81) has proposed a scheme for putting the text together in its present form. The Chinese version was probably not composed until the Mongolian text had been in existence for about a hundred years (the omission from the Chinese version and the Gloss of some difficult Mongolian words, which may have become obsolete in the interval, has been pointed out by Ch'en (1934: 296). The use of "colloquial" language places it in a contemporary context "outside the pale" of Chinese literature; and if, as has been suggested, it was composed soon after the foundation of the Ming dynasty when a number of 15

MEDIEVAL CHINESE GRAMMAR Mongolian documents were brought from Peking to the Ming court and printed for the use of students of the Mongolian language being trained by the Ming government, this use of "colloquial" would point to its having been prepared for the training of interpreters. On the other hand many of the Chinese documents of the Mongol dynasty were written in a form of "colloquial" so that, if the Chinese version of the "Secret History" was composed before the end of that dynasty, the existence of such documents, together with the one form of "colloquial" literature, the drama, which flourished in the Mongol period, provided a background of non-literary writing as the social context in which the text could operate. 3.3.3 However, in spite of this uncertainty about the origin of the text, there is one aspect of its context in which it can be said with some certainty to be unique. While for the purpose of descriptive grammar a text is circumscribed and statements are made as valid only for that text, it is true nevertheless that with a written text one can set up certain criteria for placing it in the context of a corpus of written texts which together constitute one "language" (etat de langue); statements made as valid for one text in the corpus may then be taken unless disproved to apply to the others. Such a corpus requires a unity, whose limits cannot be generalized, of time, place, language and style. There is no other text which shares with the Chinese version of the "Secret History" its time and place of composition and a common language (colloquial not literary) and style (narrative not documentary or dramatic). In this sense the "Secret History" can be said to present an "etat de langue" in itself. 3.4 One can thus establish a context for the "Secret History" by setting out a series of reconstructions at different points in time at which the text is known to have been the object of attention; the whole will form the "context of situation" of the work in 1955. Here the task of the linguist is not perhaps to trace the history of a particular manuscript or other form of the text considered as a concrete object so much as to treat the text as a linguistic abstraction from the sum total of attested events in which it has participated. In this way the contextual framework will be seen not as a single series of events in chronological sequence but rather as a series of such "chronologies" each one based on the participation of the text in events of one type. An example of a type or "class" of event would be the use of the text as source material by historians. Certain classes of events would by definition be restricted to one event, and therefore to one point of chronology; for instance, the composition of the text. For other classes the chronology might be coextensive with the whole history of the text. 16

"SECRET HISTORY OF THE MONGOLS' For the "Secret History" the following series of chronologies can be recognized: 3.4.1 Composition of the Mongolian (original) version of the "Secret History", probably entitled ^.inggis-qahan-u huja'ur ("The origins of Genghis Khan"). Suggested dates: 1240 (traditional), 1252 (Grousset),29 1264 (Hung 1951: 490): see colophon to the text (XII 58; 282), where the year is specified as one (the year "Rat") in a 12-year cycle.30 3.4.2 Composition of Chinese version of the "Secret History" in abridged translation from the Mongolian original; its compilation into a single work with the Mongolian version in Sino-Mongolian script and with the addition of the Chinese Gloss. This should be divided into three stages, though we cannot be certain of the exact ordering of the events, which might have taken place without interval. (i) First stage: interlinear insertion of the Chinese Gloss to the Mongolian version in Sino-Mongolian script (with or without the Mongolian version in the other presumed "original" script, Uighur or hPags-pa). (ii) Second stage: composition of the Chinese version entitled Yiianch'ao pi-shih (1). (The translator did not necessarily see the Gloss, and (ii) may have preceded (i) in time.) (iii) Third stage: compilation of the text in the present form (interlinear Gloss, intersectional Chinese version, Mongolian version in Sino-Mongolian script only, with certain changes from SinoMongolian script of (i) (Ch'en (1934:28,29). Date: some time, probably extending over a number or years, in the period 1369-1404. (Stage (ii), if it preceded stage (i), may antedate this period.) 3.4.3 Use of the text as linguistic teaching and reference material. It is possible that 2 (iii) should figure here, if the purpose of compilation in this form was solely for the use of interpreters. In any case the text once compiled was probably so used. (i) Use of text in training Chinese as Mongolian interpreters. (ii) First printing of text, probably for inclusion in enlarged Sinoforeign vocabularies (Hua-i i-yu) (4). Probable date:between 1404 and 1418. Use of text in conjunction with Chinese—Mongolian vocabulary (which already formed part of the first edition of Hua-i i-yu (1389) before inclusion of the "Secret History").31 3.4.4

Inclusion of the text in official compilations. A copy (extant) 17

MEDIEVAL CHINESE GRAMMAR

of the Chinese version alone was submitted to the compilers of the (manuscript) encyclopedia Yung-lo ta-tien (2) for inclusion, but in the event the complete text was incorporated: the text was thus officially marked for preservation. (i) Inclusion of text in official compilation Yung-lo ta-tien: in 15 chapters, forming chapters (volumes) 5179-5193 of the encyclopedia. Date of compilation of Yung-lo ta-tien, 1403-08. (ii) A copy (presumed indirect) of the Yung-lo ta-tien edition was made by Sun Ch'eng-tse (5) (1592-1676) as Book 9 of his compilation Yuan-ch'ao tien-ku pien-nien k'ao (6); Sun's whole work included in the official encyclopedia Ssu-k 'u ch 'uan-shu (7) (1773-82). (Since the "Secret History" occurred in Sun's work it was not copied separately from the Yung-lo ta-tien into the Ssu-k (u ch 'uan-shu.} 3.4.5 The text in Chinese scholarship and bibliography. The work was recorded by a number of scholars as being in their possession; some of these scholars wrote colophons to the text. (i)

Inclusion in Sun Ch'eng-tse's Yuan-ch'ao tien-ku pien-nien k'ao (see 4 (ii)). (ii) Wan Kuang-t'ai (8) (1712—50) compiled a volume of selections from the Chinese version in two chapters (chiian) (25) and the preface, entitled Yuanpi-shih liieh (9). (iii) Ch'ien Ta-hsin (10) (1728-1804) wrote the colophon to the 15-chapter MS copy in his possession, (iv) Pao T'ing-po (11) (1728-1814) collated the copy in 15 chapters. (This copy reached Palladius (Kafarov) in 1872 (see 6 (ii)).) (v) Juan Yuan (12) (1764-1849) discovered MS, 15-chapter copy in Hangchow. (vi) Ku Kuang-ch'i (13) (1776-1835) certified a 12-chapter copy (indirect) of the printed text, collating it with Ch'ien Ta-hsin's 15-chapter copy (see (iii)). (vii) Chang Mu (14) (1805-1879) copied the Chinese version only from Yung-lo ta-tien in 1841 and in 1848 printed it in his Lienytin-i ts'ung-shu (15). (viii)Li Wen-t'ien (16) (1834-98) wrote a commentary on Chang Mu's edition, entitled Yuan-ch'ao pi-shih chu (17), published in 1898. (ix) Shen Tseng-chih (18) (1850-1922) copied Chang Mu's version and wrote a commentary; his manuscript was acquired by Kuo 18

"SECRET HISTORY OF THE MONGOLS'

Tse-yiin (19) who printed it in 1945 in his Ching-chi-t'ang ts 'ungshu (20). (x) Wen T'ing-shih (21) (1856-1904) acquired Ku Kuang-ch'i's 12chapter copy (see (vi)) and had it recopied, presenting the recopy to Naito Torajiro (1866-1934) in 1902; the latter recopied and sent it to Naka Michiyo (1851-1908) in Tokyo (see 6 (i)). (xi) Wen T'ing-shih's own copy acquired by Yeh Teh-hui (22) (1864-1927) who published it in 1908 in a wood-block edition entitled Meng-wen Yuan-ch'ao pi-shih (23). (xii) Commercial Press, Shanghai, issued a photolithographic reprint (in Ssu-pu ts 'ung-k 'an (3)) of Ku Kuang-ch'i's certified text (see (vi)), with 41 pages replaced by corresponding pages discovered from the original printed edition, of which Ku's (indirect) copy was a facsimile. 3.4.6 West.

The study of the text in modern China, in Japan and in the

(i)

Palladius (Kafarov) acquired Pao T'ing-po's 15-chapter copy (see 5 (iv)) and published a Russian translation of the Chinese version (printed edition in vol. IV of the "Works of Members of the Russian Church Mission in Peking", 1910). (ii) Naka Michiyo (see 5 (x)) published a Japanese translation on the basis of both the Mongolian and Chinese versions, entitled Chingisu kan jitsuroku (1907). (iii) Through the work of these scholars the text has become generally known, and problems of language (e.g. the "reconstruction" of the Mongolian, i.e. its transcription into Roman or Uighur; grammar and vocabulary), the translation and the history of the text have occupied the attention of such scholars as Haenisch, Pelliot, Kozin, Shiratori and Ch'en. (iv) The text in translation in various languages is now available to historians and others as source material, and has been so used, for example, by Vladimirtsov. (Since the Mongolian version has been interpreted, the Chinese version no longer figures in this context, being itself incomplete.) (v) In contemporary scholarship problems of the text continue to occupy a prominent position, with, for example, Mostaert's detailed discussion of points of translation, Hung's study of the transmission of the text and the discussion between Hattori and Kobayashi concerning the original script of the Mongolian version. 19

MEDIEVAL CHINESE GRAMMAR 3.4.7 The text as literature in modern China. There is little mention of the text in works in Chinese literature and it is not recognized as having literary merit. Because it is unique in its combination of language and style, it does not fall into any currently recognized literary genre: it is not wen-yen ("literary" language) nor can it be classified with the great novels of the "colloquial" literature.32 It can be fairly said that the "Secret History" is not read as literature in China today.33 In the Ssu-pu ts'ung-k'an, the edition used in the present work, it is classified as "history". 3.4.8 The use of the text in the present work. From the point of view of this analysis the present study of the text represents the final point of contextualization. In this event the text participates as linguistic "informant" — in fact as an informant who is himself a linguist, in virtue of the plurilingual character of the text. This final stage in the reconstructed context of situation completes (again for the present analysis) the chain of creative effect of the original event, the writing of the text; this is admittedly a specialized form of context, but a significant one in that it requires that all linguistic events should be recorded and thus by a self-conscious process links the context of a written with that of a spoken text. 3.4.9 The above is offered as one method of contextualization for a written text. It is not a history of the text, but an abstraction of what are, for the linguist, the relevant features of past and present events in which the text has participated in one form or another. Such a scheme is put forward to correspond with the analysis at the level of context envisaged for a spoken text by the recommendations contained in the Report.34 4 Graphic analysis 4.1 In the transmission of the "Secret History of the Mongols" the original format of the text has been preserved with remarkable accuracy. This is undoubtedly because of its form as a bilingual text with a complete Gloss. Apart from textual variants, the only formal inconsistency in the tradition is that two separate divisions of the text into chapters (chiian) (25) have come down to us. In one tradition the text is divided into 15 chapters; this goes back to the Yung-lo ta-tien edition, the "Secret History" comprising chiian 5179-5193 of the great encyclopedia. In the other tradition the text is divided into 12 chapters, actually 10 plus 2 designated "supplementary" (hsii) (25); this goes back to the Ming print, as seen in the fragments, 41 pages in all, that have been preserved. In both, however, the number and division of paragraphs is identical. There are 20

"SECRET HISTORY OF THE MONGOLS" 282 paragraphs in all, and the division into chapters of the two versions of the text is shown in the following table, which indicates the paragraph with which each chapter begins: 12 ch.: 1 69 104 127 148 170 186 198 209 230 247 265 15 ch.: 1 69 97 119 141 154 170 186 198 208 225 239 247 265 277

Four of the chapters are thus the same length in both versions: 1,6 (7), 7 (8) and 11 (13); and the 15-chapter version has a chapter division at the point where the "supplement" in the 12-chapter version begins. 4.1.1 With the Mongolian version as the starting point, the Gloss may be called "interlinear" in arrangement and the Chinese version "intersectional" (by paragraphs).35 Each complete paragraph is given first in Mongolian with interlinear Gloss; this is then followed by the complete paragraph in Chinese, so that the division into paragraphs is fixed and clear. In the Commercial Press edition of Ku Kuang-ch'i's certified copy, which is the edition I have used for the purposes of this study,36 the pages are numbered by chapters, each page having, as always in traditional Chinese book-production, two sides. The number of pages in the 12 chapters is as follows: 1:49, 2:51, 3:50, 4:51, 5:51, 6:54, 7:50, 8:58, 9:49,10:45, 11:53,12:58. In the original Ming print, as shown by the fragments, the pagination is also continuous, so that, for example, page 40 of Chapter 8 is page 396 in the original print.37 4.2 Each page of the text is divided into five columns by vertical lines running between the upper and lower margins. The Mongolian text is written (like the Chinese, vertically) in large characters at the left side of the column, with the Gloss in smaller characters to its right. The position, and size of print, of the Mongolian relative to the Chinese version are those of a text as opposed to a commentary. Unlike a continuous Chinese text, however, the Mongolian has subdivisions whose presence is indicated in the first place by spaces occurring between groups of characters. These spaces vary considerably in extent; in the first three chapters there is a tendency for them to be divided into long and short spaces corresponding to "piece"-divisions and word-divisions, but after Chapter 3 the length of spacing is quite arbitrary, so that for the text as a whole we can recognize simply word-division as a feature of the script of the Mongolian version.38 The number of characters in a word ranges from one to about ten. 4.2.1 To the right of each Mongolian word is the Chinese Gloss to that word, so that the word of the Mongolian can be further recognized by its being the unit of the Gloss. The unit of the Gloss corresponding to one Mongolian word may be referred to as a "gloss", using a small g. The 21

MEDIEVAL CHINESE GRAMMAR number of characters in a gloss ranges also from one to about ten, and the glosses are generally also marked off by spaces one from another, although the setting of the Mongolian (being the primary text) is not altered to make room for a more lengthy gloss so that the spacing of the latter is sometimes omitted. 4.2.1.1 Each gloss stands in one of two formal relations to the Mongolian word, and it is this feature that always permits an unambiguous identification of the latter. Either the two are simply juxtaposed; or, much more frequently, they are marked as linked by means of a bracket.39 The bracket is open to the left, thus grouping together the characters of a Mongolian word (the limits of which, in the Sino-Mongolian script, would of course be unrecognizable to a Chinese who did not know Mongolian without the aid of such a formal device). Bracketed glosses are 'translations' of the Mongolian. Unbracketed glosses are 'designations', and there are only some 20 different designation glosses, almost all indicating some kind of proper name; for example, rin mirj "name of a man", ho3 mirj "name of a river".40'41 4.2.1.2 Thus each word of the Mongolian is glossed by a Chinese gloss, the juxtaposition of the two showing their equivalence; the presence of a bracket (or dash) shows that the gloss stands to the Mongolian in the relation of a translation. Frequently, however, with translation glosses there is a subdivision within the gloss, indicated both by an extra hinge on the bracket in addition to (and in the event always below) the central hinge and also, where possible, by a space within the gloss. The effect is to show that not only does the whole gloss relate to the whole Mongolian word but also a particular part of the gloss relates to a particular part of the word. This division of a gloss into two parts corresponds to a division of the Mongolian word into "stem" and "termination".42 To the terminations correspond certain regularly used characters and combinations such as hir) (for most cases of the noun),z2di (for the possessive declension) and others. 4.2.2 Each paragraph of the Mongolian text, with Gloss, is followed by the version of the same paragraph in Chinese translation. The Chinese version is in characters of medium size, two rows to the column, and beginning at a distance of about an inch from the top margin: this, by the conventions of Chinese book production, gives the Chinese the status of a commentary, particularly one that has become an accepted part of a text (and which may itself be the subject of further commentary). The Chinese text is punctuated; there is one punctuation mark, a small circle or "stop", this being the usual practice with punctuated Chinese texts before the adoption of a form of European punctuation. 22

"SECRET HISTORY OF THE MONGOLS' Each paragraph of the Chinese version is thereby marked off graphically into a number of units, which we shall call "pieces". The piece is that section of the text contained between two stops (or between the beginning of the paragraph and one stop). The question of the validity of the piece as a category for linguistic analysis will be considered separately.43 4.2.3 Thus the major graphic divisions of the whole text are "chapter" and "paragraph". Below the paragraph, the Mongolian has the "word" and the "character". The Gloss follows the Mongolian, with a "gloss" corresponding to the Mongolian word, the gloss itself being sometimes subdivided into parts of which the latter (or last) is usually one of a limited number of forms which we may call "termination glosses"; the smallest unit is again the "character". The Chinese version, below the paragraph, has the "piece" and the "character". 4.3 In addition to characters of the size standard to each version, the Mongolian and Chinese versions (but not the Gloss) contain a number of very small characters, some of which occur to the left of a standard character, the others below to the right of it. In the former position, only two characters occur: jurj "centre" and se3 "tongue". Fourteen characters are found in the latter position.44 These subscript characters are "auxiliary transcription symbols", those on the left being "diacritic", those below on the right "phonetic". In the Chinese text they occur only with transcription characters, that is, characters used in the transcription of Mongolian words;45 thus they form a part of the Sino-Mongolian script and not the Chinese script.46 The place of the auxiliary transcription symbols in the script is discussed below (5.2.1—2). In practice, as a study of the text reveals by its internal inconsistencies (and as is amply confirmed by the transcription into Romanized Mongolian), these auxiliary transcription symbols are frequently omitted in error, the simple character bu5 being the most prone to disappear. Some are already missing in the Mongolian text and more are lost in the Chinese version.47 4.4 Finally a large number of the Mongolian words having the verbal termination ba are followed by a note in Chinese stating that the character ba2 representing this syllable has been substituted for an original be or bie (occasionally for ba4 or ba5). Ch'en (1934: 8-16)48 has found that while this note is appended to 512 out of 1,110 occurrences of ba2 in the text it does not occur at all in Chapter 1 and occurs only three times in Chapter 2, these chapters having respectively 86 and 115 occurrences of ba confirms what is suggested by the distribution 2. This of the phonetic auxiliary symbols,49 that the Sino-Mongolian script of

23

MEDIEVAL CHINESE GRAMMAR the "Secret History" is fluid, particularly in the first two chapters, though tending to become more fixed and consistent thereafter.50 For the relation of the graphic symbols to the languages of the text see below, 5 "Script and Language". 5 Script and language 5.1 In its written symbols the "Secret History of the Mongols", including both versions and the Gloss, is a homogeneous text: it does not contain one written symbol that is not either an attested Chinese character or a character which, while not attested elsewhere, is made up of elements of the Chinese script so combined as to place it within the system of the latter.51 Within the text, however, two scripts are to be distinguished: the Chinese and the Sino-Mongolian. The difference between the two is greater than that separating the French and the English scripts, both of which have as their substance the Roman alphabet; since, in addition to formal characteristics which serve to distinguish the Sino-Mongolian from the Chinese script to an observer unable to read either (the auxiliary transcription symbols of the former),52 there being also formal characteristics which so distinguish French from English,53 the relation of the unit of the script, the character, to the script as a whole is quite different in the two cases. The two scripts are in fact of basically different types, which we can identify by calling the Chinese script "lexical" and the Sino-Mongolian "phonological".54'55 5.1.1 By the time of the compilation of the "Secret History" Chinese characters, in their relation to the forms of the Chinese language, had already constituted a stable system over a very long period. The identification of the character with that unit in the language of which it was the written symbol was well established.56 The single character was not only the written symbol of a lexical unit (or at any rate lexico-phonological, since it is not identical with the unit required for lexical description: see 6B), but also usually analysable into parts of which one (and that, moreover, the basis of lexicographical entry) functioned as a unit in a system of written forms ("radicals") which classified lexically (by meaning at the lexical level) and lexicogrammatically57 but not phonologically. Moreover, lexical forms themselves being "infinite" (that is, not operating as terms in limited systems), the character of the Chinese script is one of an infinite number of written symbols. The character of the Sino-Mongolian script, on the other hand, is the written symbol of a syllable, one of a finite number of phonological forms; the 24

"SECRET HISTORY OP THE MONGOLS" number of characters is also finite, though it is greater than the number of such phonological forms. This greater quantity of characters is a result of not only the existence of a number of variants in arbitrary58 use for one syllable but also a partial lexicalization of the script in an attempt at conformity with Chinese practice.59 5.2 Haenisch (1939: 186-191)60 has listed the characters used in the Sino-Mongolian script of the "Secret History". He gives a total of 476 characters, plus 62 other symbols formed by the addition to a character of one or other of the two diacritic auxiliaries; actually 500 different characters, since 24 of those found with diacritic auxiliaries are never found without. The use of the diacritic auxiliary has definite implications: a character with a diacritic auxiliary has a different value in the phonological system from the same character without one, and Haenisch is thus right in considering the complex of character plus diacritic auxiliary as a separate symbol. In this way we arrive at the number of 538 different symbols. With the phonetic auxiliaries the case is different; they do not change the value of the character to which they are appended but have a distinct value of their own; they should be considered separate symbols, and bring the total of distinct symbols in the script to 552. 5.2.1 The two diacritic auxiliaries jurj and se3 are prosodic markers, indicating features of Mongolian articulation not otherwise represented in the script, jurj indicates a back (velar/uvular) consonant in back vowel syllable, the use of a diacritic suggesting that the point of articulation was felt as significantly different from that of the back consonants in front vowel syllables: possibly post-velar or uvular, as opposed to velar or palato-velar. But the symbol is added (with a few exceptions) not to characters representing front vowel syllables with back consonant but to those representing syllables, back or neutral, with initial (h);61 this gives no distinction of voiced and voiceless, and it is assumed that characters with this diacritic stood for either one of a pair of syllables, one having as initial a voiceless plosive, the other a voiced fricative: there is no trace of this distinction in the Sino-Mongolian script (Street 1957:4). The reason for the choice of jurj ('centre') as diacritic in this function is not clear. But its use in the script is two-dimensional: it marks a back syllable and it marks a back consonant as syllable initial (the syllable final corresponding to which is represented by the phonetic auxiliary hei). se3 ('tongue') combines with characters representing syllables with initial (1), and indicates some form of lingual articulation, presumably trill or roll, which we designate (r); it also combines with r (and its variant f4) representing syllable-final (r) (but only from Chapter 3 onwards62). As the use 25

MEDIEVAL CHINESE GRAMMAR of a diacritic suggests, a distinction is meant here which was not felt to be present in Chinese. 5.2.2 The phonetic auxiliaries indicate the final consonant of syllables ending in a consonant. Of the 14 phonetic auxiliaries found in the text, five occur regularly whenever a certain consonant is to be marked as syllable final: hei (q), ke(1)(k), 3 (t),le and ti bu5 (b). As variants of le for (1) there occur two other characters both having the signific element "water", Ie2 and (once only) Ie3, the latter a very rare character; these and not le occur in words glossed by designations as watercourse, river or lake, and by translation as cuan "river". As variant of ke, ki8 occurs (once only, para. 267 (XII 10b)). Of irregular but frequent use are mu (m) (first occurrence para. 77 (II 10a)) and s5 (s) (first occurrence para. 121 (III 39b)), which vary with the same characters written standard size in the text; mu also varies with zero occurring where the character used in the transcription of the syllable has final -m in Middle Chinese. Of irregular and infrequent occurrence are ci3 (c) (first occurrence para. 88 (II 26b); found only five times, all in Chapter 2 and always the same word) and sin (s) (only occurrence para. 149 (V lb)), the latter with variant si12 "room, house" (only occurrence para. 169 (V 49b), in a word glossed by translation as farjFinally ni (n) occurs 3z "house"). frequently, but not regularly, in one syllable (en) (first occurrence para. 90 (II 30a)), and varies with zero (in Chapter 1) and le (? once only, in para. 104 (III 2b)); ni also occurs three times (in Chapter 10) in the syllable jan, always in the same word. The variants Ie2, Ie3 and si12 show the same tendency towards lexicalization of the script as is found with the standard characters.63 This use of a character to represent only the initial consonant of the syllable of which it is the symbol is an application ofthefan-ch'ieh (28) method in Chinese phonology, whereby two characters are used to indicate the pronunciation of a third. The first of these, the fan-ch 'ieh shang-tzu (29), indicates the niu (30) (class of initial), and only its initial consonant (sheng-mu (31)) is to be read, the final (yun-mu (32)) being disregarded; the second character likewise yields the final, its initial being disregarded.64 The phonetic auxiliaries of the "Secret History" are to be read as the first character of a fan-ch 'ieh, though here there is of course no second character. 5.2.3 While to each syllable in the phonological structure of the Mongolian language corresponds a number of interchangeable symbols in the script, the choice of a variant is in certain instances determined by this process which has been referred to as the "lexicalization" of the script. By this process, in some instances, in place of the character 26

"SECRET HISTORY OF THE MONGOLS' regularly employed in the transcription of a given syllable a variant is chosen whose lexical meaning (in Chinese) is felt as akin to the meaning of the Mongolian word in which it operates. Thus, for example, the name Mt Burqan is transcribed, passim, as bu-f4-han. The character f 4 has as signific element san "mountain, hill" and was probably made up for the purpose of transcription: according to the Kanghsi Dictionary it occurs only in mountain names. Elsewhere, for the Mongolian value represented here by r4, the character f is used. By this device the written symbol is made to convey, as it does in the Chinese script, lexical as well as phonological meaning. It is difficult to estimate the resultant increase in the total number of symbols, since one cannot always decide whether a given variant is lexically determined or not. Where, as in the case of r4, a radical is added, such determination is usually clear; Ch'en instances 91 characters so formed, the majority with radicals suei2 "water", ma3 "horse", niau "bird", keu "opening", mu3 "eye", san "mountain", tu "earth", zu2 "foot", yi9 "clothing", gim2 "metal", mun2 "gate", yiarj4 "sheep", yian5 "word", su2 "rat" and curj4 "insect", etc. Some of these Ch'en calls 'characters invented for the "Secret History'", noting that they are not found elsewhere. Nine other variants listed by Ch'en, though not having "lexicalizing" radicals, seem likewise determined: for example, bu6 "step" in yia-bu6 (yabu "go") instead of regular bu. Others, however, including both single characters (e.g. Je3 not je in je3~sin-gu3, para. 117 (III 31b), glossed yiu3tu3di 'intending') and groups of characters lexicalized in combination (Ch'en notes e.g. si13-E4-wun4, glossed giun2z 'nobleman' (p. 25b)65), often cannot be classified with certainty as either determined or free variants. 5.2.4 If one recognizes only those distinctions made by the SinoMonogolian script (as opposed to those known from other sources, such as "vowel harmony" in syllables where the script is prosodically neutral), there occur an average of about two symbols for each term (syllable or final consonant) in the phonological system recognized by the script;66 while of the 250—300 variants a substantial proportion show lexical determination in one direction these variants never occur without lexical connotation, though they are not always found where such connotation would be appropriate.67 It is this feature of the existence of variants which distinguishes the Sino-Mongolian script from other syllabaries such as the Japanese kana scripts, though it resembles the script used for Old Japanese in for example the Manyoshu. 5.3 The interrelations among the languages and scripts of the

27

MEDIEVAL CHINESE GRAMMAR

"Secret History" may be summarized as follows. The text of the "Secret History" is bilingual in script (Sino-Mongolian and Chinese) and in language.68 The two main versions, named according to the language in which they are written, are the Mongolian version and the Chinese version. The Mongolian version is original; the Chinese version is a translation of it, somewhat abridged. Each language has its own script; the same symbols are used in both but have value at different levels. Each version contains forms from the language of the other; those which are relevant to a study of the language of the Chinese version are discussed in the following paragraphs. Chinese forms found in the Mongolian version can be considered as "borrowings" by the Mongolian language, and thus as Mongolian forms and not as foreign forms in a Mongolian text. Where, however, the Chinese version contains Mongolian forms it must be recognized that, since the Chinese version is a translation, these forms may function not in the Chinese language as a whole but only in the Chinese text. The presence of these "translation forms" does not affect the method of analysis but they do require to be brought out in the statement, since they may not operate within the framework of Chinese systems. 5.3.1 This will apply both to lexical and to grammatical translation forms where such can be established. Lexical translation forms are usually obvious. They fall into two categories: proper names and others. Proper names (marked in the Gloss by designation glosses), which form the great majority, do not need separate statements, since they function in the Chinese system exactly as do Chinese proper names.69 Other lexical translation forms, while functioning in the Chinese grammatical system, require separate lexical statements since they may need to be collocated in the language from which they are "borrowed". It is interesting that the Gloss, which, as said, contains no translation forms, renders such words by translation; yet the Chinese version uses not the translation but the Mongolian form of the original.70 The complete list of lexical translation forms other than proper names is given in Appendix D. The only form which it is doubtful whether to regard as a translation form or not is M. jam (plural Ja-mu-t), which is both glossed and rendered in C. as Jam (Gloss has plural Jam mui). It is known that this word is borrowed from Mongolian into Chinese,71 the process being assisted by the existence in Chinese of the "word Jam, Middle Chinese -m, 'to stand'; but since the Gloss has Jam it will be treated as a Chinese word, not a translation form, the presence in the Gloss of a separate form being taken as the criterion for the presence in the Chinese of a lexical translation form. 28

"SECRET HISTORY OF THE MONGOLS' 5.3.2 With grammatical translation forms the position is more difficult. Haenisch considers that there are many of these, and he instances two types which may be called morphological and syntactical. The former are phonetic borrowings where the actual termination of the Mongolian is taken over into the Chinese system; the latter are grammatical caiques and include instances of Mongolian word-order. In discussing 18 grammatical auxiliaries ("grammatische Hilfsworter") of the Chinese version, he suggests that four of these are morphological translation forms and at least two other syntactical.72 As an instance of Mongolian word-order, where, in his view, the Mongolian influence is most evident, Haenisch quotes phrases and sentences which, he says, violate the principles of word-order in Chinese.73 It seems in general desirable that all grammatical forms found in the text should be accounted for with reference to the grammatical system. Unless it is possible to show that a given grammatical form is irregular in the language of the text (and the criteria for such a statement will be difficult to establish), such that it should be treated extra-systemically, the absence of a form from other texts assigned to the same etat de langue does not in itself demand that the form be afforded separate treatment. The relative infrequency of certain forms may affect their place in the hierarchy of the grammatical system;74 but while this infrequency may in turn be determined by the fact that these are translation forms, this fact is not itself the determining feature. Of the grammatical translation forms listed by Haenisch, that which appears most alien to the known forms of Chinese is hirj; hirj is regular in the Gloss as equivalent of certain Mongolian noun terminations, and it might be that its occurrence in the Chinese text may indeed be a "borrowing" from the Gloss where it is a translation form of the gloss or "label" type. It is, however, regular in the Chinese version (53 occurrences, fifth in order of frequency of postpositive nouns); and it has seemed preferable to treat it, together with the other forms which, while they do show a coincidence at two levels (grammatical and phonological) with Mongolian forms, yet seem less anomalous, as terms operating regularly within the grammatical system of the Chinese.75 Notes for sections 1—5 1. See Hung (1951), see also 3.2. 2. E.g. by Palladius, Haenisch, Pelliot, Kozin and others. 3. Ch'en Yuan, Yuan pi-shih i yin jung tzu k 'ao, suggests 1389—98 for the first printed edition of the whole. 29

MEDIEVAL CHINESE GRAMMAR 4. 5. 6. 7. 8. 9.

10.

11.

12. 13. 14. 15.

16. 17.

18.

19.

20.

See 4.1. See 3.4. The "Ku-certified copy": see 3.4.5. See 5. passim. See 3.4.5. For the graphic arrangement of the two versions and the Gloss, see 4.2. For a comparison of the two versions, see Haenisch (1931: 28—39) which is a (German) translation, by paragraphs in parallel columns, of the Mongolian and Chinese versions of paras 54-78 (I 34a-II 12b). In fact Kozin (1941) gives two romanized versions of the text, one 'transtranscription' (pere-transkribirovanije) of the Sino-Mongolian 'transcription' (script) of the Mongolian, and one a 'direct' transcription of the Mongolian as reconstructed (through the Uighur-Mongolian script) from the Sino-Mongolian. Verse passages as determined by Kozin are printed in verse form. Pelliot completed his transcription in 1920, but it "was not published until 29 years later, after his death. Haenisch's transcription appeared in 1936, after Haenisch had waited for some time for Pelliot's, of which he knew, to be published. Kozin completed his transcriptions in 1935 and was able to take account of Haenisch's work only in the final editing. Shiratori made the first draft in 1917, but after various revisions it was finally published only in 1942, after his death. According to Shiratori (1942: 3), it was Naka's intention to publish a version of the original Mongolian. Verse passages are rendered into Russian verse. The first six (out of 12) chapters are translated. Translations of particular passages, with a comparison of their renderings by Haenisch, Kozin and Pelliot, have been made by Antoine Mostaert (1950,1951,1952). There is also a Turkish translation: Temir (1948). The author made use of both Haenisch's and Kozin's translations. Since this was written, a detailed study of the language of the Mongolian version has been made by J. C. Street (1957). In his bibliography (p. 78), Street refers to 'Cleaves, Francis Woodman (1955). The Secret History of the Mongols.' I have not seen this work of Cleaves. Pelliot regards the manuscript as descended from the original UighurMongolian version of the text, but notes that it contains errors not found in the Sino-Mongolian version. Translators of the Mongolian have of course made use of the Chinese version; and Haenisch (1931: 4) states that he himself made a second translation but did not publish it, regarding it as of little value beside a translation made directly from the Mongolian version. For the suggestion that the aim of the description should determine which

30

"SECRET HISTORY OF THE MONGOLS'

21. 22. 23. 24.

25.

26. 27.

28. 29. 30.

31. 32.

33.

34.

35.

of two (or more) exhaustive and consistent forms of statement should be preferred, cf Spang-Hanssen (1949). For "relationship" as a function of systems, see Allen (1953: 89—95). For the use of the word "style", see 3.2; for "colloquial" and "literary", see 3.3 I ignore here the specialized context of its use by linguists as a basis for linguistic statements, which it shares with all linguistic source material I retain the term "situation" for the whole diachronic framework of events within which the text operates. Readers of the text in 1955 are participants in this situation. The first three words of this version "Cmggis-qahan-u huja 'ur" ("The origin of Genghis Khan") were probably a title of part of the book, but there seems to have been no original title to the whole work. Cf.1.1. There is very little "oratio obliqua", as shown by the pronominal reference of the dialogue. E.g. the interrogative mood (see 6B.2.6 and 6C.2.6) is found only in passages of dialogue, while imperative is frequent in "indirect commands". Cf. Demieville (1951: 47), on the use of the terms "vulgaire" and "litteraire". R. Grousset, L'Empire Mongol, Pans, 1941 (quoted by Hung (1951)). I give no further chronology for the Mongolian version. There is little doubt that it was in fact used as source material for Mongol histories (see Hung 1951, especially the diagram on p. 481). For the translation of the colophon, see e.g. Haemsch (1948). For explanation of references to the text, see 4.1.1 n. For a detailed account of the Hua-i i-yu see Lewicki (1949). Its nearest relative is perhaps the short story, such as those in the collection Ku chin hsiao-shuo (24). It is these short stories with which a systematic linguistic comparison should first be made; together with the Yuan drama, with -which especially the dialogue passages of the "Secret History" might well be compared. The situation seems to be different in Mongolia, where it appears the Mongolian version of the "Secret History" is now well known. This is to be expected in view of the position of Genghis Khan as a national hero, the unifier of the Mongol peoples. Much of the material for the above section is to be found in the article by Hung (1951). It should be stressed, however, that Hung's purpose was the study of the transmission of the text in particular, concrete forms: original print, copies, etc. The data are presented here as facts relevant to context, this providing a broad basis for the treatment of the text as a restricted language. On the arrangement of the text see also Street (1957:1).

31

MEDIEVAL CHINESE GRAMMAR 36. Yuan-ch'ao pi-shih, Ssu-pu ts'ung-k'an edition (3rd series, History), Shanghai, Commercial Press, 1936: 10 chiian. (Chapters 8—9 and 10—11 each form one chiian (bound volume) in this edition.) 37. All references to the Chinese version of the text are given, first by chapter and page, second by paragraph and line. The page is further marked a or b, indicating recto or verso. The number of the line is the number of the piece (the unit of punctuation) in the paragraph, following the Commercial Press edition subject to the correction of a small number of misprints (see Appendix E.) So for example the reference (IV. 38b; 144.10) is to line (piece) number 10 of paragraph 144, which is on page 38, verso, of Chapter 4. References to pages falling within the fragments of the Ming print incorporated in this edition give the continuous pagination in parentheses, thus: (VII. 30a (336a); 194.15), thus showing that the page in question is one of the 41 preserved. 38. In fact even the word is not always identifiable by the spacing, which is sometimes sacrificed for appearance or economy of paper. Nevertheless this spacing is a definite feature of the Sino-Mongolian script, and its omission may be regarded as an inconsistency in the use of the script. In these instances the word can always be identified by other means; see 4.2.1.In. 39. Where the Mongolian word consists of only one character, a dash is used instead of a bracket; this is equivalent to the hinge of a bracket (indicating a relationship between the two sides) without the bracket (which is required only when two or more characters are to be shown as grouped together). It is the bracket (or dash) which serves as the formal mark of the Mongolian word where the spacing is indistinct. 40. The list of designation glosses is as follows: rin mirj "name of a man" niu mirj "name of a woman" fu4 rin mirj "name of a lady" wu4 mirj "name of a shaman" guan2 mirj "name of an office" (with occasional instances of other, specific names of offices) sir) "clan", sir) mirj "name of a clan" Jurj2"tribe", jurj 2 mirj "name of a people" di2 mirj "name of a place" cirj6 mirj "name of a town" fu7 mirj "name of a prefecture" jveu2 mirj "name of a district" ho3 mirj "name of a river" suei2 mirj "name of a watercourse" hai3z mirj "name of a lake" san mirj "name of a mountain" seu3 mirj "name of an animal (species)" 32

"SECRET HISTORY OF THE MONGOLS" ma3 mirj "name of a horse" yiu4 mirj "name of a fish (species)" (and occasional instances of other names of animal species)

41.

42.

43. 44.

45. 46.

47.

48.

There is some inconsistency in the use of brackets with names of species: seu3 mirj and yiu4 mirj appear sometimes without a bracket, sometimes with; while geu2 mirj "name of a dog (species)" occurs only with a bracket. One can think of these as names of species for which the translator did not know any Chinese equivalent; they are the "borderline cases" of proper names. In some instances in the early chapters personal names, where the name has occurred recently above, have no gloss; and two names of (? species of) horses appear without gloss. In the fragments of the Ming print Mongolian words glossed by rin mirj have a thick line in the place of the bracket. Other designation glosses have, as in Ku Kuang-ch'i's certified copy, nothing. The Mongolian words Cirjgis (Genghis), glossed Tai zu (his Chinese reign title), and baatur, "hero", areyiurj unbracketed. Likewise 2si10glossed are some proper names which, having Chinese equivalents, receive translation glosses, e.g. Mongolian Kitat, glossed Ki7dan2 "Kitan". All forms quoted from the "Secret History" are given in special transcription (see 8 and Appendix A). For the Chinese characters, the transcription may be compared with the Chinese syllabary (at end); the subscript figures refer to the corresponding characters in the syllabary. There is considerable inconsistency, not only in the form of the brackets (some having no hinges at all, where one and where two might be expected) but also in the extent of the Mongolian word. For instance in para. 17 (I. lOa) iigei-bolba "was no more, died" is treated as one word (is enclosed in one bracket) and glossed, with two hinges, wu/zuo-liau; in the same column iigei boluqsa(n), "(after he) had died" is treated as two words, the first glossed wu, and the second zuo-liau-di. These inconsistencies, however, are not of such extent as to invalidate the general principle. See 6A. 2-4. See 5.2.2. The character se3 found in this position in para. 123 (3.43b) is a misprint for ke. Chiefly proper names; see 5.3.1. Since the Gloss contains no Sino-Mongolian script (proper names being designated; see 4.2.1.1), the auxiliary transcription symbols do not occur in the Gloss. For example in paras 50 (I. 31a/b) and 54 (I. 34b) we find the name Daritai-otcigin in four versions, one correct (with two auxiliary transcription symbols), the others having the three different possible errors! According to Ch'en Yuan this change in the transcription is another modification of the "lexicalizing" type (see 5.2.3) in the Sino-Mongolian script, the Chinese word written ba2 having a sense comparable to that of the Mongolian termination ba. 33

MEDIEVAL CHINESE GRAMMAR 49. See 5.2.2. 50. Ch'en (p. 29a) concludes that the Sino-Mongolian script of the "Secret History" was a later improvement, imperfectly applied, of that of the Hua-i i-yti (4). His reasons are that the "lexicalizing" variants are not found in the latter (which has, for instance, be bie ba4 and bas for ba2), and that the diacritic auxiliary dirj2 occurring in the Hua-i i-yti for syllable-final (1) has been incompletely replaced by the phonetic auxiliary le in our text — the 15-chapter "Secret History" having relics of dirj2, while the addition of the diacritic auxiliary se3 to the character f is likewise incomplete. 51. See 5.2.3. 52. The spacing/bracketing of the Sino-Mongolian script is another characteristic. While the bracketing might be regarded as a feature of the Gloss, or at any rate not part of the Sino-Mongolian script, it seems that it is in fact the presence of brackets -which has led to, by permitting, an inconsistent use of spacing; and one might justifiably abstract from the two a formal mark of delimitation of a unit "word", nothing corresponding to which exists in the Chinese script. A text in Sino-Mongolian script without a Gloss would presumably furnish us with a unit which we could designate a "word" marked off entirely by spacing. The fact that the Sino-Mongolian script here contains no stop is another formal difference; but the stop cannot be regarded as a regular feature of the Chinese script before the late nineteenth century, and conversely one could insert stops in the Sino-Mongolian text without violating its principles — though in fact the writer did not do so. 53. It is not contended that there is any parallel, in their place in the system of the script, between the French diacritics, with their heterogeneous functions, and the Sino-Mongolian auxiliary transcription symbols. The point is that, to a person not knowing either language, the diacritics (transcription symbols) not only mark off a French (Sino-Mongolian) text from an English (Chinese) one but also frequently (but not always) mark off an element, such as a word, of the former script found in a text of the latter. 54. The lexical nature of the Chinese script was early recognized by Du Ponceau, who held that it was not ideographic but (syllabic and) "lexigraphic". See Du Ponceau (1848: 36, 59 and Section II passim). See also the reference to this work in Firth 'Atlantic Linguistics' (1949: 104). Du Ponceau further (e.g. on p. 113) employed the term "logographic", also used by Chao of the Chinese script. See Chao (1934). 55. J. C. Street's work The Language of the Secret History of the Mongols (1957) gives a comprehensive account of the Mongolian language of the text, and includes a section (pp. 2—8) on phonology in which he discusses the Sino-Mongolian script and its relation to Mongolian. The outline in the following paragraphs is intended merely to fill in the background to the Chinese version and show how it links up with the original Mongolian at different levels.

34

"SECRET HISTORY OF THE MONGOLS' 56. The Chinese word tzii (26) is the modern name for this identity. Following the Chinese practice, we shall use the one word "character" to denote not only the written symbol but also the linguistic unit which it symbolizes. 57. E.g. the majority of characters with radical (no. 64) seuj represent monosyllabic free verbs. 58. The arbitrary use of variants in the Sino-Mongolian script of the Hua-i iyti is queried by Lewicki; see 8.2. n.3. 59. Liu San-wu's (27) preface to the Mongolian Chinese Hua-i i-yti of 1389 gives an interesting view of a contemporary Chinese opinion of the superiority of a script (such as Chinese) on lexical principles over one (such as Sino-Mongolian) on phonological principles. See E. Haenisch (1952:7), where this preface is translated. 60. Lewicki (1949: 29—50) lists the characters used in the Sino-Mongolian script of the Hua-i i-ytt (4) of 1389: 436 + 8 phonetic auxiliaries + 3 diacritic auxiliaries + 64 formed by the addition of a diacritic to another symbol. Lewicki recognizes 110 homophonic variants and discusses (pp. 66,67) possible reasons for their use. 61. On the question of the roman transcription of the phonological system represented by the Sino-Mongolian script see 8.5.1. 62. With one instance in Chapter 2. See Ch'en, (1934:28a). 63. Haenisch (1939) lists bu (b) among the phonetic auxiliaries, but does not list ki8 (k) or ci3 (c). bu is not found in the Commercial Press edition. Pelliot (1949) transcribes ge-de, glossed Je2li "here",in Chapter 1 (e.g.para. 24 (I 15b)) as ada; the gloss suggests that this should rather be anda, with zero variant for the phonetic auxiliary ni. 64. Thus the phonetic auxiliaries are found only in syllables of structure (C)VC, where they represent final C. If, for example, the character ti3 is represented phonologically as initial t, final i, then when this character is used as phonetic auxiliary it has value t as final C in a structure (C)VC. 65. But in the word ni-cu-gun, "naked", cited by Hung (p. 455) as a lexicalizing combination ("you-without-pants"),ni and cu are regular. 66. Haenisch recognizes 293 distinct phonological values. The number arrived at by a transcription of the Sino-Mongolian script based on a phonological analysis of Chinese will of course depend on that analysis, and the system I have used here gives only 273 (see 8). 67. Cf. Ch'en, (1934: 19a): "Ie2 is not used wherever there is water, but wherever Ie2 is used there is water." 68. It might be possible to consider the language of the Gloss as a third language, since it does have a grammatical system of its own. This grammatical system, however, is simply that of Mongolian and does not function outside the text; so that, while the language of the Gloss has some of the characteristics of a pidgin, the systems of which it is made up do not combine into one single system and it must be denied the status of a

35

MEDIEVAL CHINESE GRAMMAR language. (More accurately, part of its grammatical system pertains to Chinese and part to Mongolian; here again the two parts do not combine into a single system.) 69. Having, as it were, unlimited collocability (cf. Harris (1952: 6, n. 4). Mongolian proper names are able the more easily to function in the Chinese in that they are made to fit in to the Chinese phonological system, this in turn being possible because the symbols of the (phonological) Sino-Mongolian script are also symbols in the (lexical) Chinese script. The only exception lies in the use of auxiliary transcription symbols, which the Chinese reader could ignore in the reading of Mongolian names. 70. For example, M. and C. an2da3, G. ki7ho2; M. jamcin (Ja-mu-cin), G. Jamhu4, C. jamci3 71. See Haenisch (1933). 72. Yieu, lai, yieulai and je;3.hirj It is only andof madau course in certain uses that it is suggested that these are translation forms; possibly the same sort of "contamination" is postulated as for jam above; at any rate these uses are such that (with the exception of hig?) the lexical interpretation of the characters is retained and they do not appear merely as phonetic representations of Mongolian forms. Haenisch does not discuss Jo, but would presumably regard this also as a morphological translation from the Mongolian verbal termination cu, ju. For detailed discussion of these forms, see 6C passim. 73. For example (I 43a; 62.3) ni Je2 rz yian6 mir)3 mian guarj yieu; (I 47b; 66. 14) lau zai gia li di H4 wu. See 6C 3.2 and 6C 3.5.5 where these forms are discussed. 74. See especially 6C 3.5.5, where yieu, lai, yieulai and yie are not regarded as forming a distinct dimension of clause classes. 75. I should agree with the view implied by Ota Tatsuo (1953: 5,13) that the "innovations" in the language of the Chinese version of the "Secret History" are native forms not found in earlier written texts, rather than "borrowings" from Mongolian. Some of these forms occur in the Chinese version without parallel occurrence in the Gloss (see e.g. 6C 2.6.1 n.).

36

"SECRET HISTORY OF THE MONGOLS'

SECTION 6: A and B 6 Grammar 6A General 6A. 1 The principle of proceeding from the larger to the smaller unit, applied in the first place in the ordering of the levels at which the analysis is made, will further guide the choice of the order of the units and segments of the text about which grammatical statements are made. It is clear that no grammatical features can be abstracted, no patterns identified, from or between segments of above a certain extent: it is obviously impossible to set up grammatical "chapter-types". The division of the text into chapters, whether we consider the 12-chapter or the 15-chapter version, can be regarded as a convenience of publication having reference to the content of the work but without linguistic significance. The division of each chapter into paragraphs is also made largely with reference to the content; it is noticeable that there is a general tendency for paragraphs to be shorter in the early part, particularly in Chapter 1, and to be considerably longer in the second half of the work. The length of the chapters, however, is roughly constant (see 4.1.1), so that in those chapters where the paragraphs are shorter and the average number of pieces per paragraph smaller, the number of paragraphs is greater. The paragraph as such contains no special linguistic features and thus lacks any formal criteria for classifying linguistically into types. The paragraph is never smaller than the piece, and each paragraph begins and ends with a complete piece. Two paragraphs (215 (IX 19a) and 327 (X 12a)) consist of one piece only. 6A.1.1 It is with a unit below the paragraph that grammatical analysis must begin.1 In the graphic analysis, the unit next below the paragraph is the piece. It will be necessary to determine whether or not the piece can function as a unit of linguistic analysis, and whether, even if it can, some other unit larger than the piece will be required to be established first. 6A.2 An analysis of certain features of the piece reveals that it is necessary for grammatical purposes to set up a unit larger than the piece (but smaller than the paragraph). The Chinese version in the Commercial Press edition contains 5,386 pieces.2 Certain pieces, defined by • position in the paragraph, display features marking them off statistically from the pieces as a whole. If we take the final piece of each paragraph

37

MEDIEVAL CHINESE GRAMMAR and compare the frequency of occurrence of certain elements, commonly found as piece-final, in these 282 pieces with their frequency in the 5,386 pieces of the whole work, we find striking differences. 6A.2.1 The most frequent piece-final, liau, occurs 630 times as piece-final (in 11.7 per cent of the total pieces) but 120 times as paragraph-final (42.9 per cent of the total): it is nearly four times as frequent in paragraph-final pieces as in the pieces as a whole. Another frequent piece-final, je, occurs 124 times (2.3 per cent) as piece-final but 28 times (9.9 per cent) as final of a paragraph-final piece. This suggests that liau and j e are to be thought of as final features of some unit larger than the piece (but smaller than the paragraph, since they occur respectively 510 and 96 times as finals of pieces that are not paragraph-finals). 6A.2.2 On the other hand the characters si2, rja, xian and heu, and the combination sarjteu 156,108,49,20 and 9 times 2, occur respectively as piece-final, but none of them occurs as paragraph-final. This also suggests that we require a linguistic unit larger than the piece, a unit to which these elements are never final, and that the pieces to which they are final are always part of this larger unit. At the same time it confirms the assumption, justifiable on general grounds,3 that the piece has some validity as a linguistic unit and may form the basis of the stage in the analysis next after the larger unit to be set up. 6A.2.3 Frequent also as piece-finals are lai and kiu. These occur 299 times as piece-final and 17 times as paragraph-final, 5.6 per cent and 6.0 per cent respectively. This closeness not only confirms the validity of the piece, by suggesting that lai and kiu function as piece-finals, but, if this assumption is correct, justifies the choice of the paragraph-final piece as a basis for statistical comparison.4 6A.3 We must therefore admit the need for a unit larger than the piece, and the unit to be set up to meet this requirement will be called the "sentence". So there are 282 points where we know a sentence to begin and 282 where we know one to end. In fact there are no such frequently recurring characters or combinations to be met with as pieceinitial, so that in the delimitation of the sentence the above piece-finals can serve as a preliminary guide to what is to be expected. In dealing with a written text it is impossible formally to delimit every sentence without reference to smaller units, though one may establish the category of sentence without such reference. It is quite obvious from a reading of the text that the sense of the single piece is often not "complete"; but this in itself does not necessitate the setting up of a larger unit because one cannot state at which points in the text the sense is completed, and the larger category is of no use if every member cannot 38

"SECRET HISTORY OF THE MONGOLS' be more or less clearly assigned to it. There may be "borderline cases" but the criteria for the demarcation of such a unit need to be reasonably exact in application, and therefore formal rather than notional. 6A.3.1 There are, as has been shown, certain formal criteria which help to differentiate between pieces that are sentences or sentence-final pieces and pieces that do not complete a sentence. It is not maintained that this differentiation is made without reference to "meaning"; on the contrary, the whole analysis is made on the fundamental assumption that the meaning of the text can be stated, this being in fact what the linguist, in applying his techniques, sets out to do. In saying that si2, rja and so on do not occur as sentence finals, we are setting up a type of sentence in which these forms occur in a certain position, namely at the end of a piece which is not final in the sentence; and this classification of sentence types is part of the statement of meaning at the grammatical level. This is quite other than to say that because certain pieces are hypothetical in sense, or because they contain the English translation meaning 'when', 'if, they must be treated as incomplete. The presence of si2, qa and so on is a formal criterion, the choice of which is guided by the assignment of grammatical meaning to these elements: the meaning of rja may be stated in the terms that it forms one member of a certain word-class the members of which make up a system of piece-finals which are not sentence-finals,5 and as such it excludes other members of the class, such as si2, and, as a member of its class, determines the assignment of the piece to a certain type. 6A.3.2 As a corollary to this distinction between "sentence pieces" (pieces which are at the same time complete sentences — of which at least two must exist, since paragraphs 215 and 237 each consist of one piece only) and "non-sentence pieces", if the category of sentence is a valid unit for grammatical statements we must expect to find parallel structures in sentences irrespective of their division into pieces. The most striking examples of this are (II 47b; 66.10-12) and (X llb-12a; 236.1-5 and 237), where in each instance two sentences parallel in structure are juxtaposed, the one in one piece, the other consisting of two and five pieces respectively. The parallelism in structure suggests that the piece may show so great a variety that the statement of grammatical structures, to be exhaustive and simple, can only be based on the sentence. The position of the piece in the grammatical statement is not of course thereby determined in one way or the other. 6A.4 To summarize the problem dealt with in these paragraphs: it will be found necessary in linguistic analysis at the grammatical level to set up a unit, the sentence, which is less than the text and greater than the 39

MEDIEVAL CHINESE GRAMMAR unit of lexical statement, the word. In the present written text there is a unit, the piece, marked off formally by punctuation. Is this graphic unit, the piece, coextensive with the linguistic unit, the sentence? The answer must be: no. The sentence that is set up must be (as a category) larger than the piece, since certain forms which are final to the piece are not final to the sentence. Of the relation between the two we may say so far that: 1, a piece ending in liau or je will normally be final in the sentence; 2, a piece ending in si2, rja, xian, heu or sarjteu2 will normally be nonfinal in a sentence; 3, a piece ending in lai or kiu may be either final or non-final in a sentence. 6A.4.1 In the preceding paragraphs the method of procedure in the analysis has been exemplified with reference to one particular problem: the establishment of the unit of grammatical statement. The descriptive statement as a whole is not of course made to follow the order of procedure in the inquiry, which would involve cumbersome and lengthy accounts of the steps taken to arrive at each statement. The next section (6B: "Grammatical categories") will begin with a summary of all the grammatical categories employed in the statement at this level, followed by an account of each; in 6C and D the categories will be further described by translation and exemplification. (For section 6D, please refer to the CD-Rom provided.) In certain instances some indication is given of the reasons for the choice of a particular form of statement rather than any other.

6B Grammatical categories 1 Scheme of categories

6B.1 The following is a summary of the categories of units, classes and functions into which the material is organized for statement:6 (i) Units: (a) Graphic: Chapter Paragraph (b) Linguistic: (ii)

Piece Sentence

Character (graphic) Clause Word Character (linguistic)

Classes (systemic): (a) Sentence classes: Compound/Simple (b) Clause classes: (1) Free/Bound (2) Verbal/Nominal 40

"SECRET HISTORY OF THE MONGOLS' (3) (a) Voice: Passive/Ergative/Active (neutral) ((3) Aspect: Perfective/Imperfective/Non-perfective (neutral) (y) Mood: Interrogative/Imperative/Affirmative (neutral) Bound clause classes: Conditional/Genitival (c) Word classes:7 (1) Free/Bound (2) Verbal/Nominal/Adverbial (a) Verbal word classes: Verb/Pro-verb Verb classes: Free/Bound Bound verb classes: Prepositive/Auxiliary/ Postpositive (P) Nominal word classes: Noun/Pronoun Noun classes: Free/Bound Bound noun classes: Numeral/Auxiliary/ Postpositive (y) Adverbial word classes: Adverb/Particle Adverb classes: Free/Bound Bound adverb classes: Preverbal/Final/ Conjunctive (d) Character classes: Free/Bound (iii) Functions (structural): (a) Free clause functions: Sentential/Non-sentential (b) Free word functions: Substantive/Attributive 6B.1.1 The graphic units have been discussed above (see Section 4 "Graphic Analysis"). The relation between the graphic and the linguistic units can be stated in terms of extent. The largest linguistic unit, the sentence, is smaller than the paragraph and has no graphic unit corresponding in extent. The clause is not exactly coextensive with the piece, but is the linguistic unit most nearly corresponding to it: the majority of the pieces form one clause. The one-to-one correspondence between the linguistic and the graphic character is implied in the use of the term "character" for both. 6B.1.2 The classes and functions are the grammatical categories of the linguistic units.8 The terms "system" and "structure", as used by Firth, are distinguished in the manner described by Robins (1953): the system is a paradigmatic construct by reference to which each term is defined as excluding all the other terms; the structure is a syntagmatic framework of interrelated elements stated as values, the exponents of 41

MEDIEVAL CHINESE GRAMMAR which are segments of the text. If it is said that a clause, active in voice, may be found with one of three simple structures NVN, NV or VN, this gives a three-term system of clause-structures for the active voice, (N)V(N); it might then be added that the exponent of value V in this structure may be free verb or pro-verb,9 giving a two-term system for V here, (N)V2(N). The classes set up for each unit are systems independent of any structure, whereas the functions can be identified only in the structure; one could of course talk of a "system" of functions, once the latter have been identified, so that it is a characteristic of each system as well of functions as of classes to be exhaustive; for example all clauses are passive, ergative or active (neutral) in aspect, and all occurrences of free nouns are as substantive or attributive. 6B.1.3 Classes and functions are formally marked. The mark of a class may be said to be the "interior form" of its members, the form of the forms themselves; while that of a function is "exterior form", that is the presence and relative position of other elements. Interior form is taken to include lexical form; by "lexical form" is meant the assignment of a "word to a certain class on the basis of the categories and combinations to which it is susceptible,10 and of a character to the class of "free" or "bound" according to whether or not it is a word (i.e. is assignable to a word-class). The presence of a word assigned to a certain word-class may in its turn be the criterion for the determination of a class of clause: the verbal clause is determined in the first place by the presence of a verb. Since other classes of verb can occur only in combination with a free verb, a verbal clause always contains a free verb; the latter may be defined with reference to the verb-group (see 6B.3.1), that is, by certain possibilities of combination. 6B.1.4 The interrelations among the terms sentence, clause, word and character in the system of linguistic units may be stated within the grammar, though there are implications outside the grammatical level: the word, having been established in the grammatical analysis, becomes the unit of lexical statement.11 The sentence, which in a spoken text can be contextualized, in a written text has no independent context of situation; sometimes, however, a form of contextual analysis can be applied to forms at the level of sentence and clause.12 The character (which, however, is here established without the aid of the grammar) is in this text the unit of phonological statement. 6B. 1.4.1 At the same time, sentence, clause, word and character all operate within the grammatical system. Here each term is defined as consisting of one or more complete exponents of the term next in succession. Each unit therefore admits an opposition of "compound" 42

"SECRET HISTORY OF THE MONGOLS" and "simple", the compound containing more than one, the simple containing one of the unit next below. "Compound" and "simple", being functions of the units in interrelation, do not (except in the sentence) constitute "classes" of the units; but the "bound" class of each unit is characterized as unable by itself to form the unit next above: a bound clause cannot by itself form a sentence, and so forth. Only with the sentence does the opposition "compound/simple" enter into the description, since the sentence is defined maximally; it is unlimited in extent, and is considered to extend until marked as closed.13 Thus a compound sentence consists of one free clause plus one or more free or bound clauses; a simple sentence consists of one free clause. It follows from this that the members of the free class of each unit can be subdivided functionally, by exterior form in the syntagm, into those which form a simple member of the unit next above and those which combine with others, free or bound, to form a compound member thereof. Again, however, this opposition only enters into the descriptive statement at the highest level, a free clause being categorized functionally as "sentential" or "non-sentential".14 6B.1.5 It must be emphasized that the totality of occurrences of any one form in a limited text such as the "Secret History" may not be enough to determine its place in the paradigm. Since the statements made are valid only for the language of the text, there may arise a free choice in the assignment of a given form to one or two categories; where this occurs that assignment has been chosen which permits of the simpler statement, and if simplicity provides no criterion, then the form is assigned a place corresponding to the place of some comparable form in the Modern Pekingese system. Sometimes a form occurring once or twice only has to be treated extra-systemically — even though its rarity may be contextually determined. 2 Sentence and clause

6B.2 As has been said above, the sentence in the "Secret History" cannot be delimited by purely graphic criteria. Since the text is divided into paragraphs, and is punctuated with one mark, the stop, these criteria permit a partial delimitation: the sentence never extends across paragraph divisions nor divides within the piece (but see below for the exception of direct, quoted speech). Further, linguistic criteria are required for complete delimitation. A sentence is therefore said to begin with the occurrence of a clause containing a noun15 in pre-verbal position (other than in complex group),16 unless the previous clause is 43

MEDIEVAL CHINESE GRAMMAR marked as not sentence-final; and with the occurrence of any clause, •whether or not containing a noun in pre-verbal position, where the previous clause is marked as sentence-final. There are three qualifications to this: (i) the opening of a passage of quotation in direct speech is always considered to begin a sentence even where (as frequently) internal to a piece; (ii) two clauses, or groups of clauses, parallel in structure are considered to form one sentence, even if the second clause, or group of clauses, contains a noun in pre-verbal position; (iii) a bound clause not itself containing a pre-verbal noun, but followed by a free clause with pre-verbal noun, is considered to begin a sentence. 6B.2.1 A bound clause which itself is initial in a sentence (beginning a paragraph or passage of direct speech, or containing a noun in pre-verbal position) is ipso facto marked as not sentence-final in virtue of its being bound. In fact in the text it rarely occurs that a bound clause ends a sentence: the normal position of a bound clause is that it precedes, or is internal to, a free clause in the same sentence; a conditional bound clause usually precedes a free clause, while a genitival bound clause precedes or is internal to a free clause. The presence of a following bound clause, however, is not considered in itself to mark the end of a sentence. Similarly a clause which is marked aspectivally (that is, perfective or imperfective, not neutral in aspect) is normally not sentence-final;17 it is, however, not considered to be marked as such for the purpose of the delimitation of the sentence. A clause which is marked modally (interrogative or imperative) is always sentence-final. 6B.2.1.1 A sentence is said to be compound if it consists of more than one clause; if it consists of one clause only it is simple. A (free) clause which by itself forms a complete (simple) sentence is said to have sentential function. Any clause, free or bound, which is combined in the structure with another clause or other clauses to form a (compound) sentence is said to have non-sentential function.18 There is no distinction of interior form between a free clause with sentential function and one with non-sentential function; nor between a non-sentential free clause combining with a free clause or clauses and one combining only with a bound clause or clauses. > 6B.2.2 Primary position in the clause is defined relative to the verb substantive or pro-verb. The basic positions are thus pre-verbal and post-verbal. These admit secondary distinctions: pre-verbal may be initial, non-initial or unmarked, post-verbal may be final, non-final or unmarked. No other positions are recognized in the clause as such, further positional reference being to given elements in the clausestructure (for example, preceding or following a word). Positional forms 44

"SECRET HISTORY OF THE MONGOLS' are stated as structures in linear representation using letters to symbolize the operative values, as follows: V v N n A a p

Free verb substantive, verb group or pro-verb Bound verb Free noun substantive, noun group or pronoun19 Bound noun Free adverb or complex group Bound adverb Particle

It is sometimes convenient to symbolize the categories of word-classes which may operate with these values. With regard to V, N and A it is to be assumed that the relevant forms (free, "pro-" or group) may operate here unless otherwise specified; if it is desired to refer to the value V, N or A in a structure where only the free form may operate, the symbol F is used. To refer to a group, the V, N or A is written above a line below which is given the structure of the group, in which the reference of the symbol F will be shown by the group letter (thus F in verb group must = free verb). To specify which class of bound word may operate with the values v, n and a the following symbols are used:20 Pr Au Po Nu Fi Co

Prepositive (verb),preverbal (adverb) Auxiliary (verb, noun) Postpositive (verb, noun) Numeral (noun) Final (adverb) Conjunctive (adverb)

If it is not clear from the presence of a group letter which value is referred to, the symbol will be preceded by a raised v, n or a. Thus aPr = preverbal adverb, which is the word-class operating with value a in the given structure; while ^ would show that here the element with value V is a verb group the structure of which is free verb+postpositive verb. 6B.2.2.1 In a clause with only one exponent of the value N, the order may be NV or VN. Where there are two, the normal order is NVN; but in verbal clauses in ergative or passive voice the order is NNV.21 Where there is an exponent of A in the clause the normal order will be (N)AV(N); in ergative clauses NANV, in passive clauses NNAV.22 In the ergative and passive clause structures the first N (but not the second) may be absent.23 45

MEDIEVAL CHINESE GRAMMAR Less common structures in an active clause are NN(A)V and (N)(A)VNN. The latter is found only with a small group of free verbs as exponents of V The former resembles the ergative and passive clause structure in being contextually determined: see 6C.2.4. A less common position of A, with value "complex group" only, in an active clause, gives structure (N)VNA. 6B.2.3 Verbal and nominal clauses are defined according to the exponence of V: a verbal clause has free verb, a nominal clause pro-verb. A verbal clause contains one, and only one, free verb in substantive function. A nominal clause contains no free verb substantive; it normally contains a pro-verb and at least one exponent of N, but it may be without a pro-verb and even without a noun, its defining characteristic being absence of free verb substantive. Such instances in the text are, however, few, there being only five nominal clauses with no pro-verb. Both verbal and nominal clauses may be free or bound, but they are distinguished from one another in their possibility of combination with other categories (see 6B.2.8). 6B.2.4 Voice, aspect and mood are classes of the clause. Each is ordered in a three-term system of which one term is formally unmarked and is considered as "neutral"; for instance, a clause which is active may also be described as "neutral in voice". In each class the neutral form is statistically the most frequent in occurrence, and is the "nondetermined" form when features of contextual determination can be shown to operate.24 That voice, aspect and mood are considered as categories of the clause and not of the verb is a result of their being marked by formal distinctions in the clause: variations in the structure, and the presence of elements which are not only not part of the verb but often not even juxtaposed with it.25 The three terms in the voice system are ergative, passive and active or neutral. To the unmarked, active form corresponds the basic structure (N)V(N). The basic structure of the marked terms is (N)NV. The determining structure of the ergative voice is (N)vNV, where v has for exponent the prepositive verb ziai), rarely ba; that of the passive is (N)vNVp, where v is prepositive verb si, and p is di. The passive is very rare in the "Secret History", but it has been set up here because it is formally identical with the passive of Modern Pekingese, which shows the same three-term voice system. The ergative of Modern Pekingese is identical except that the relative frequency of ziarj and ba is reversed.26 6B.2.5 The three terms in the aspect system are perfective, imperfective and non-perfective or neutral. Perfective aspect is marked by post-verbal liau; in clauses of structure (-)V-, where there is at least one 46

"SECRET HISTORY OF THE MONGOLS" exponent of value N following the verb, liau may be either non-final (immediately following V) or final.27 A clause of this structure in which liau occurs non-fmaUy is normally not sentence-final, whereas one with final liau is always sentence-final. In a perfective clause of structure (-)V, liau is of course unmarked as to position, and such a clause may or may not be sentence-final.28 There is thus a subsidiary system of perfective clauses, the three terms of which may be designated non-final, final and unmarked. Imperfective aspect is marked by jo which always directly follows the verb.29 Imperfective clauses are normally not sentence-final. 6B.2.6 The three terms in the modal system are interrogative, imperative and affirmative or neutral. Interrogative is marked by the presence of one of the particles ma, madau3 and (once) feu in clause-final position; imperative by one of the particles je, jo, yieje and za in the same position. Clauses with interrogative words (see 6C.3.16), such as suei "who?", are not considered interrogative, being formally identical (qua clause) with other affirmative clauses. Modal particles have absolute final position, all other final forms (e.g. final liau) preceding them.30 6B.2.7 Both verbal and nominal clauses may be either bound or free. The two terms in the system of bound clauses are conditional and genitival. The conditional bound clause is marked adverbially, by the presence of (i) a preverbal adverb, (ii) a final adverb, (iii) a combination of the two or (iv) a combination of preverbal adverb with the particle di.31 The genitival bound clause is marked by the presence of the particle di (very rarely Ji3) in final position.32 6B.2.8 There are certain restrictions in the possibilities of combination of clause classes. In a text of limited extent such as the "Secret History" it is necessary to distinguish between those combinations which can be regarded as excluded on statistical grounds and those whose non-occurrence is predicted on these grounds and therefore cannot be regarded as a feature of the language. Since for example there are only 25 interrogative clauses in the text, and less than ^ of the total clauses are nominal (454 out of 5719), the predicted number of interrogative nominal clauses would not be greater than two, and the fact that in the event there is only one cannot be regarded as excluding or making irregular the combination of nominal with interrogative. Those combinations which can be regarded as excluded, because their non-occurrence is not predicted, are as follows. Nominal clauses exclude marked voice;33 in the aspect and modal systems, nominal excludes imperfective and imperative,34 while the combination of nominal and perfective is rare.35 Nominal clauses are regularly bound.36 Bound clauses 47

MEDIEVAL CHINESE GRAMMAR exclude only marked mood. Of the combination of the voice, aspect and modal systems, that of marked voice with marked aspect is regular, as is that of marked voice with marked mood; combinations of marked mood with marked aspect are, however, rare.37 3 Word and character 6B.3 The category of "word" is set up to meet the need for a linguistic unit smaller than the clause and larger than the character. Although the character is unambiguously delimited in a written text, its grammatical meaning qua character can only be stated in terms of the opposition "bound/free"; it cannot be assigned by interior form to classes operating in clause structures. The unit which can be so classified is greater than the character, consisting of one or more complete characters, and is called the "word" in accordance with the normal practice of descriptive grammar. Like the sentence and clause, the word cannot be delimited by purely graphic criteria. Its delimitation as well as its classification will be on the basis of its interior, including lexical, form.38 6B.3.1 The establishment of the word-classes of verb and noun, and the assignment to these classes of elements delimited as words, proceed through the description of "groups", specifically the "verb group", "noun group" and "complex group".39 A verb group is a group of verbal elements (words) among which there obtain certain interior relations determining the operation of the various elements; the noun group is likewise a group of nominal words, the complex group one of both verbal and nominal words. The group is not a unit in the grammatical system: it does not operate in the unit series sentenceclause-word-character, since this is a system in which each term consists of one or more complete members of the term following, whereas a clause may contain no group at all. The determining feature of the group, on the other hand, is precisely that it is always interchangeable in the structure with a member of the free word class corresponding to it: the verb group is one term in the two-term system of values for V in the verbal clause structure, namely free verb/verb group. 6B.3.1.1 The bound classes of the verb can only operate in a group: prepositive verb only in the complex group, other bound verbs in the verb group or complex group. Only the free verb can stand alone as exponent of V.40 The structure of the verb group is (AU)F(PO)» where Au, F and Po have values auxiliary, free and postpositive verb. In fact if both Au and Po are absent we do not talk of a verb group, so that the structure here symbolized is that of value V in a verbal clause.41 A verb group always contains a free verb. 48

"SECRET HISTORY OP THE MONGOLS" 6B.3.1.2 The bound classes of the noun can normally operate only in the noun-group; while the noun group as such forms part of the complex group, it is only in the complex group that the postpositive noun can replace the noun group; elsewhere only the free noun can stand alone as exponent of N. The structure of the noun group is ((NU)(A^I))F(PO)> with numeral, auxiliary, free and postpositive noun.42 The postpositive noun may be absent; the numeral and auxiliary nouns may both be absent, but if either one is present the other is normally also present.43 There is a subsidiary structure of the noun group, or "minor noun group", thus: N^. Occasionally a noun group may have value A in a structure NAV; such a noun group always contains a postpositive noun. 6B.3.1.3 The complex group has adverbial value in the clause structure, though the elements of which it is composed are nominal and verbal words, the free form being a noun.44 The basic structure is ^5^: prepositive verb followed by noun group (or free noun). A subsidiary structure has postpositive verb: ^N.45 A further feature of the prepositive type of the complex group is that the prepositive verb may be preceded by an auxiliary verb.46 Unlike the verb group, which can only operate in a verbal clause, but like the noun group, the complex group occurs in both verbal and nominal clauses, with the limitations that in the nominal clause (i) only the prepositive complex group is found and (ii) a complex group never contains an auxiliary verb. 6B.3.2 The group is considered as unable to be split, though this is not taken to exclude attribution. A free verb substantive, whether alone or within a verb group, may be immediately preceded by a free verb attributive; and a free noun substantive, whether alone or within a noun or complex group, may be immediately preceded by a free verb or noun attributive or by a genitival bound clause. 6B.3.3 Thus the identification of the verbal and nominal word classes proceeds paradigmatically through the description of the structure of the three types of group. All classes of verb and noun are identified within the group; free verb and free noun are further identified in the clause structure, where their value is that of the corresponding group. Pro-verb and pronoun have identical respective values in the clause structure but are identified by their exclusion from the group. The paradigmatic identification of word classes in the clause structure permits the syntagmatic identification of a word as a member of a certain word class,47 this being further assisted (as said above) by the categories and combinations to which the word is susceptible. By the latter means, a word may be assigned to a word class on the basis of a category which 49

MEDIEVAL CHINESE GRAMMAR itself is a category not of the word but of the clause. The clause categories of voice, aspect and mood not only presuppose the presence of a verb in the clause, but also partially permit its identification: only the verb is directly followed by the aspect markers Jo and non-final liau.48 In other cases the category with identificatory function is of the word: here the prior identification of (some members of) a word class is presupposed, so that we can say that (i) only a verbal word can be directly preceded by a negative adverb, (ii) only a nominal word can be directly followed by the plural particle mui (in the pronoun, mui is not a particle but a bound character within the pronoun49). 6B.3.3.1 There is one further distinction between verbal and nominal words which, although it concerns the lexical structure of the two types of word class (that is, the interrelation among the members of each), may conveniently enter into the grammatical statement. Within the verbal word classes there is a high degree of lexical identification among the members of each class: almost all forms assigned to the classes of prepositive, auxiliary and postpositive verb are assigned also to the class either of free verb or of pro-verb. This can be stated in tabular form as follows, where x, y and z represent the three largest lexical groupings within the verbal word classes:50 Free verb Prepositive verb x x

Auxiliary verb

y

y

z

Postpositive verb x

Pro-verb

z

z

In the nominal word classes, however, there are no groupings with lexical identification except that a considerable minority of members of the class postpositive noun are at the same time free nouns:51 members of the classes of numeral, auxiliary noun and pronoun are lexically distinct from each other and from the members of the other nominal word classes. 6B.3.4 The adverb forms a separate word class not operating in any group but definable by certain positions in the clause structure. Like the verb and the noun, the adverb admits division into lexical (infinite) and lexicogrammatical (finite) classes; the distinction is one of free and bound, but the implications of the categories "free" and "bound" are somewhat different from their implications in the verbal and nominal word classes. The free adverb includes two (out of three) classes, verbal adverb and nominal adverb, the members of which cannot stand alone as a complete clause; they are, however, free in the sense that they operate in 50

"SECRET HISTORY OF THE MONGOLS" clauses of any type. The bound adverb likewise cannot stand alone as a clause, but is grammatically restricted as to the class of clause in which it operates.52 6B.3.4.1 The free adverb is divided into three classes according to that which it modifies, modification being the function of (adverbial) position initial within that (clause) or immediately precedent to that (verb or noun) which is modified. The classes are named "clausal adverb", "verbal adverb" and "nominal adverb". Position may of course be ambiguous in the syntagm (an adverb might be initial to a clause while immediately preceding a free verb); the assignment is paradigmatic, such that, for instance, an adverb classed as verb modifier never precedes a preverbal noun. Of the three classes only the clausal adverb can stand as exponent of the clause.53 6B.3.4.2 Of the bound adverbs, the preverbal and final adverbs both (occurring in the syntagm either alone or in combination) mark a conditional bound clause. They are defined positionally: in a clause structure V-, the positions are aPrV-aFi, in a clause of structure N(N) V-, they are a PrN(N)V-aFi or NaPr(N)V-aFi. The conjunctive adverb is identical in position with the preverbal adverb, but operates in a free clause, normally following a bound conditional clause in the same sentence. The criterion for the identification of a member of this class is its occurrence at least once in the text in a clause immediately preceded by a bound conditional clause.54 6B.3.5 The particle marks the grammatical category of word or clause; unlike the lexicogrammatical word classes which also mark grammatical categories the particle enters into no lexical system and can be exhaustively described at the grammatical level. There are 15 particles, which can be divided into three classes according as they modify (modification here being the function of position final within, or immediately following, that which is modified) a clause, a verb or a noun. Two of the clausal particles are lexically identical with two of the nominal particles (di, Ji3). The clausal particles are (i) aspectival: liau (perfective), jo (imperfective); (ii) modal: ma, madau3, feu (interrogative), je, jo, yieje, za (imperative); (iii) genitival: di, ji3. The verbal particle is comparative: sie. The nominal particles are (i) genitival (possessive): di, Ji3; (ii) plural: mui, derj. 6B.3.6 The character is divided into the classes of "free" and "bound". A free character forms a (simple, monosyllabic) word; a bound character combines with one or more other characters, bound or free, to form a (compound, polysyllabic) word.55'56 No complete statement can be made of the interrelation of the word classes of the dimension

51

MEDIEVAL CHINESE GRAMMAR

"compound" and "simple" with those of verb, noun and so on; but some general tendencies are observable. Words of more than two characters are rare in the "Secret History'.57 Bound verbs are almost all of one character, except for some disyllabic (and even trisyllabic) auxiliary verbs. Of the bound nouns, all auxiliary nouns are monosyllabic, postpositive nouns both monosyllabic and disyllabic. Numerals are of one to three syllables.58 All pro-verbs except one are monosyllabic;59 pronouns have one or two characters.60 The proportion of monosyllabic words among the free verbs is much higher than that among the free nouns. Bound adverbs are of any extent; among the free adverbs the verbal and nominal adverbs tend to be monosyllabic, the clausal adverbs polysyllabic. 6B.3.7 The classification of words as put forward here has much in common with the traditional Chinese classification of (lexical) characters in Old Chinese and in the literary language based on Old Chinese. This recognized "full" and "empty" characters (to retain the unusual English translation: the Chinese is shih tzu (33) and hsu tzu (34)), the "full" being the purely lexical words, noun and verb, the "empty" the partly or wholly grammatical. "Full" characters in turn were distinguished into "live" (huo tzu (35)) and "dead" (ssu tzu (36)), or verbs and nouns respectively. The problem of whether or not the Chinese language recognized word classes, or "parts of speech", has occupied European linguists from Humboldt onwards (Simon 1937). There can be no question that, in the language of the "Secret History" it is not true,61 as sometimes asserted (Margoulies 1943), that a word may "function as any part of speech" — though it is true that a character per se (as distinct from a monosyllabic word) is not assigned to one of a system of classes corresponding to "parts of speech". We have seen that there is some degree of lexical identification between the word classes, and a complete lexicon would certainly show some forms assigned as words to both verbal and nominal classes; nevertheless the majority of the lexical forms would be found assigned to one word class only. We must agree with Simon's conclusion, drawn with reference to Old Chinese and the literary language, that word classes must be recognized.62 Whereas, however, in the description of inflected languages it is possible to proceed with words as elements of structures, the exponents being identified morphologically, by grammatical form and scatter; in the description of Chinese the word and its classes must be identified by other means.

52

"SECRET HISTORY OF THE MONGOLS' Notes for section 6: A and B 1. Harris, (1952) proposes a method for going "past the sentence limitations of descriptive linguistics". This has not been attempted here, though a study of sequences of sentence structures on similar lines would be possible on the basis of the present description. The only extension of the present analysis across sentence boundaries is in the use of "context of mention" (see 6C 1.8). 2. There are 5,380 pieces in this edition as printed. I have made a total of 26 corrections in the punctuation, by the addition of 16 and the deletion of 10 stops, bringing the number of pieces to 5,386. Reference to pieces throughout will be to the text as amended in this way. For details of suggested emendations in the punctuation, as well as in the text itself, see Appendix E. For a table showing the number of pieces in each chapter, and the average number of pieces per paragraph, see Appendix C. 3. Since the text is prose, the stop clearly has some linguistic significance, which is likely to have some relation to grammatical categories. 4. Note that the figures given above are for the occurrence of these forms as piece-final, and not as clause-final, data for which are given in Appendix C. 5. This is not of course the description which will finally be given of this word-class, but a provisional description that could be made at this stage of the inquiry. Actually they are "final adverbs" serving as markers of a "conditional (bound) clause". 6. This may be compared with the Summary of Categories set up for Modern Pekingese in Halliday (1956: 217—18) (this volume, Chapter 3). Categories with the same name are to be regarded as comparable (not identical). 7. The unit "word" is a term in the system of linguistic units set up for grammatical statement: it is thus a category forming part of the object of the description at the grammatical level. At the same time the "word" is the sole object of the description at the lexical level. For lexical statement, however, the word-classes require to be somewhat differently ordered, thus: (1) Lexical ("infinite"): Free verb/Free noun/Free adverb (2) Lexicogrammatical and grammatical ("finite"): Verbs: Prepositive/Auxiliary/Postpositive Pro-verb Nouns: Numeral/Auxiliary/Postp ositive Pronoun Adverbs: Preverbal/Final/C onj unctive Particle of which the particle is grammatical, the others lexicogrammatical. See further 7A. 53

MEDIEVAL CHINESE GRAMMAR 8. The graphic units have of course no classes and functions as such, since they represent only the method of arrangement of the linguistic material. If the piece, for instance, were to be considered as having linguistic validity and therefore having classes and functions, it would ipso facto be classified also as a linguistic (grammatical) unit. 9. Strictly "a member of the class of free verb or pro-verb". The possibility of occurrence of a class with given value in a structure is systemically restricted; that of a member of a class as exponent is restricted by collocation (see 7A). 10. This procedure of paradigmatic identification is in fact no different from that employed in determining word classes even in such a highly inflected language as Classical Latin. Cf. 6B 3.3. 11. In a language with "institutionalized" words, the word as lexicographical unit may not correspond exactly to the word set up in the grammatical analysis: cf. Robins (1953:125). 12. See6C1.8. 13. As in written texts in modern Chinese or in European languages (where the sentence is taken to extend until marked as closed by a full stop). 14. The two complementary oppositions of "compound/simple" and "sentential (clausal, etc.)/non-sentential (non-clausal, etc.)" could of course be stated at the level of each unit, but this would serve no descriptive purpose. 15. Actually free noun or noun-group, or pronoun; see 6B.3.1. 16. For the position in the clause, see 6B.2.2; for the complex group, 6B.3.1.3. 17. A perfective clause with liau in clause-final position may, however, be final or non-final to a sentence, while a perfective clause with liau in what is considered to be marked final position (see 6B.2.5) is always sentencefinal. 18. It is unnecessary to distinguish "main", "co-ordinate" and "subordinate" as separate functions within the category "non-sentential". "Subordination" might be defined as the function of all clauses in a compound sentence except the last, but such a category serves no further purpose. The concept of "subordination" might best be related directly to the class of "bound clause", which is already defined as a class by interior form. 19. N must be taken to include the possibility of a list of nouns substantive, theoretically unlimited in extent. The symbolization N ... is rejected as too clumsy. 20. Only one word-class, the particle, can operate at p. With free adverbs and particles the following further abbreviations have been employed: (cl) Clausal (ve) Verbal (no) Nominal 21. More narrowly: ergative NvNV, passive NvNVp. For the exponents of v and p in these structures, see below on ergative and passive voice, 6B.2.4. 54

"SECRET HISTORY OF THE MONGOLS" 22. More narrowly NAvNV and NvNAVp. 23. I have not considered values having exponent zero, preferring to talk of distinct structures. 24. See 6C.1.8 and 2.4.1,2.5.1. 25. The question is a terminological one: the naming of certain categories of the clause (see 6C.2.4—6). 26. Cf. Y. R. Chao, (1948: 51). ba is Chao's "pre-transitive"; see pp. 49,162. 27. "Final" position may be more strictly defined as "not followed by any other lexical or lexicogrammatical word in the same clause". Occasionally "final" liau is followed by another particle. 28. The type of perfective clause in which liau occurs twice, once postverbally and once finally, frequent in Modern Pekingese, occurs only once in the "Secret History". Chao, Mandarin Primer (1948: 40—1) distinguishes for Modern Pekingese -le (=liau) "word suffix -le for completed action" and "phrase suffix -le for new situation, . . ., for progress in narration, . . ., etc."; later (p. 132) he says of phrase suffix -le that it indicates "a new situation . . ., or a new realization of an existing situation". The formally ambiguous (post-verbal and final) le he regards as the two forms "telescoped into one". As Chao points out, other forms of New Chinese, such as Modern Cantonese, distinguish the two lexically, so that there is no ambiguity, and admit both together if the clause ends with V. 29. There are five instances where jo follows a form that could be analysed as VN. It seems preferable however to consider these forms (e.g. ki5ma3 "(be(ing)) on horse-back") as (compound) verbs, the position of jo being the criterion. The five instances are: ki5ma3jo (IX lOb; 213.21),xuozianjo "polishing arrows" (V 51a: 169.16), jimtujo "(pillowing the ground, i.e.) (sleeping) on the ground" (VI 47b; 183.22), cueiluijo "crying (letting fall tears)" (X 416; 245.27) and cueiyian4]o "with mouth watering" (VII 40b; 195.25). 30. The text has liau madau3 (three occurrences),liau ma (l),liau je (2),liau jo (1). 31. For the special type of conditional clause marked by one of four postpositive verbs, see 6C.2.7. 32. Except (i) di preceded by the prepositive verb si, which marks a passive (free) clause, and (ii) di preceded by a preverbal adverb, which marks a conditional clause. 33. The non-occurrence of passive voice in nominal clauses is predicted; it is, however, excluded in Modern Pekingese. That of ergative is excluded, some 32 being predicted where none occurs. 34. No occurrences; about 23 and 12 respectively predicted. 35. Three occurrences, 86 predicted. 36. 82 occurrences as compared with the predicted figure of 94. The closeness seems to confirm the validity of this distinction.

55

MEDIEVAL CHINESE GRAMMAR 37. This is predictable on other grounds. Marked modal clauses are normally sentence-final, whereas those with marked aspect (except final perfective) are normally not sentence-final. All occurrences of this combination have in fact final or unmarked perfective aspect. See 6C.2.5—6. 38. Cf. Chao (1948:48); and Robins (1953:110): "in the study and analysis of speech the basic and only immediate datum is the sentence, and . . . words . . . must be formally established in each language investigated before they can be used as a basis for further statements". 39. Cf. Meriggi (1953). Drawing examples from Modern Pekingese, Meriggi says: "dans les langues indochinoises modernes, c'est le groupement des mots, avec la subordination phonetique et semantique qu'il entraine avec soi, qui forme leur veritable grammaire, 1'essence de leur type" (p. 189; Meriggi's italics). And later: "Si nous avons affaire a un texte, nous devons, pour le comprendre, commencer toujours par reperer ces groupes qui constituent la phrase; une fausse separation d'elements qui en fait sont lies, ou bien une reunion d'elements qui ne le sont pas, c'est-a-dire, en un mot, un groupement different de celui qui etait concu par le sujet parlant (ou ecrivant), entrainera les pires malentendus" (p. 191). Cf. also H. E Simon, (1953) passim. 40. In the verbal clause. Since the verb group must contain a free verb (and not a pro-verb) it cannot operate in nominal clauses. The only exponent of V in the nominal clause is the pro-verb. 41. Cf. Chao (1948:48). Chao's classification of the verb of Modern Pekingese is presented conceptually — the work being, as indicated in the title, a primer for English-speaking students - but in fact every one of his categories is formally differentiated by its combinatory possibilities and susceptibility to categories (e.g. of sumxation with -le). There is no regular correspondence between the categories here set up and Chao's (conceptual) categories, but Chao's "auxiliary verb" is formally marked as corresponding to my own; his "classificatory verb" is my "pro-verb", treated here as a separate word-class because of its not functioning within the group. Chao (p. 47) has "pro-verb" in a different sense. My "postpositive verb" is not regarded by Chao as a category of the verb, but is referred to as "potential" and "directional complements" (p. 45). My "prepositive verb" is Chao's "preposition" ("pre-positives" on p. 49). What Chao calls "quality verb intransitive, or adjective" falls within my category of "free verb", forming part of the sub-class "intransitive (free) verb" (see 6C.3.1). 42. The terminology and classification here adopted differ from those of H. E Simon (1953) for Modern Pekingese as follows: My "numeral" is Simon's "determinative" minus the "demonstrative"; my "auxiliary" covers Simon's " deter minator" plus some of his "determinate"; my "free" covers Simon's "noun" plus the remainder of his "determinate"; my "postpositive" is Simon's "substantival suffix". Simon's category "determinator" is unnecessary for the language of the "Secret History". For

56

"SECRET HISTORY OF THE MONGOLS"

43. 44. 45.

46.

47.

48.

49. 50.

Simon, "substantive" is a "supra-category" which will "include not only the forms of the two complexes described in this paper [i.e. the elements of the noun group - M. A. K. H.] but also a non-complex category 'pronoun', which remains to be defined" (p. 328). I have preferred to use the term "substantive" for a function of the free noun and the free verb, and to regard the pronoun as a separate word-class because of its not functioning within the noun group. (Contrast here Modern Cantonese, where the pronoun is a form of the "determinative": in a structural comparison of the "Secret History", Modern Pekingese and Modern Cantonese, the first element of the noun group would form a single (related) category of three distinct systems: "numeral" in the "Secret History", "numeral" plus "demonstrative" for Modern Pekingese, these two plus "pronoun" for Cantonese.) For the occurrence of one without the other, see 6C.3.9. Adverbs do not operate in the group. The bound classes of the adverbs are defined in the description of the bound clause. The postpositive type of the complex group is set up to handle clauses where the postpositive verb is separated from the free verb. It would be possible alternatively to account for this form by considering the postpositive verb as part of an extended verb group; but this would involve the splitting of the verb group (by a free noun substantive) and the admittance of two nouns in post-verbal position, one of which (the first) would then require to be restricted by the condition of its irreplaceability by a noun group. The former statement seems preferable for simplicity. See further 6C.3.5.3. As an alternative method of statement one could say that the auxiliary verb operates only in the verb group, which can then be split by a complex group. This has the advantage that it then becomes unnecessary to limit the occurrence of the auxiliary verb in a complex group to verbal clauses (since only a verbal clause can contain a verb group). But it seems simpler to consider that no group can be split and that the auxiliary verb can operate in a complex group as well as in a verb group, occurring initially (to the group) in both cases. The identification of a given word as member of a certain -word class proceeds syntagmatically, in the "Secret History" (and indeed in Chinese as a whole), to the extent that it is identified by its place in the clause structure. While liau may follow a free verb or a postpositive verb, there are only two occurrences of jo following a postpositive verb; in all other instances jo serves to identify a free verb. The particles themselves are (bound) words, not bound characters. The groupings represented by x, y and z do not form a single system, nor are they ordered in series. They do, however, form in sum a majority of the words members of the lexic©grammatical classes (see 6B.1 n.). Thus for

57

MEDIEVAL CHINESE GRAMMAR

51. 52.

53.

54.

55.

56.

example more than half the members of the class "postpositive verb" are also members either of the class "free verb" or of the class "pro-verb". The grouping (x plus y) does not of course form more than a small proportion of the total membership of the lexical class of free verb. Compound (two-character) postpositive nouns such as xianteu2; but not simple postpositive nouns such as xian. Having, so to speak, grammatical function to the clause. Alternatively the distinction of free and bound might be omitted from the description of the adverb, or subordinated (even in the grammar) to the distinction of lexical and lexicogrammatical. Or again the two classes ("verbal adverb" and "nominal adverb") of what we have called the free adverb might be classified as "bound", retaining the criterion of ability to stand alone as a clause. The form of statement preferred, however, has the merit that it makes what is in fact the primary division in the adverbial word classes while retaining the terms "free" and "bound" to name this distinction, albeit with modified implications. In fact in the text this never occurs unambiguously, since, unless marked as sentence final (e.g. by final position in the paragraph), a clausal adverb can always be assigned as modifier to the following clause. But it is frequent in Modern Pekingese, and arises as a possible alternative method of statement in a number of instances in the "Secret History". It would be possible alternatively to assign all adverbs, other then preverbal and final, occurring under conditions other than in a clause immediately preceded by a conditional clause, to the single class of free adverb, thus assuming a high degree of lexical identification of the members of the two classes (free and conjunctive, as here stated). However, the conjunctive adverbs even when occurring under other conditions always show some contextual requirements; moreover over four-fifths of clauses in which a conjunctive adverb occurs are not sentence-initial and of the remainder a large number follow directly on quoted speech (especially clauses with suei, zieu and yiu2si: see 6C.3.14). The terms "monosyllabic", "disyllabic" etc. are used in preference to the more strictly grammatical but cumbersome "mono-characteric" etc. Cf. Chao and Yang, (1947: Introd. p. xxvii). For lexicographical purposes Chao and Yang set up the categories of words: B ("always Bound") and F ("sometimes Free");for them "no word is always free except interjections". The classification here is made for the different purpose of grammatical statement: we must distinguish bound words from bound characters, but need to recognize for characters simply the two classes of free and bound, noting some degree of lexical identification between the two. In this sense Chao and Yang's definition does not operate here; but its implication, that every member of the free class of character is lexically identical with a member of the bound class, is not true for the text of the "Secret History".

58

"SECRET HISTORY OF THE MONGOLS'

57.

58. 59. 60. 61.

62.

On the question of the (lexical) unit to which the classes of "free" and "bound" are to be referred, see Yang, (1949) and Chao, (1945). Most of these are probably adverbs. It is often impossible to determine whether an element with the value of free noun in the structure of a given syntagm should be considered as consisting of one word or two (attributive plus substantive). For the members of this class see 6C.3.8. wuyieu. For suei2, see 6C.3.2. All disyllabic pronouns have mui as second character. That is, to assume the validity, for linguistic purposes, of such a statement would render an exhaustive grammatical description extremely complex. Indeed the remark may be taken to imply the impossibility of a descriptive grammar. Simon, (1937:110—11): "He [Gabelentz: Chinese Grammar, Leipzig, 1881 — M. A. K. H.] declares that the existence of grammatical categories is proved by the fact that Chinese words differ in their syntactical behaviour. Certain words, this obviously means, behave generally as nouns, others generally as verbs, etc. To give an example, two words are not understood in the same manner by the listener or reader if the first is, for instance, a noun and the second a verb, or the reverse. If the noun precedes the verb, it will generally be conceived as the subject; if the verb precedes, the noun will generally be taken as the object. How could this phenomenon be explained if there were no word categories?" (Simon's italics).

59

MEDIEVAL CHINESE GRAMMAR

SECTION 6: C 6 Grammar 6C Translation of categories 1 Translation and comparison

6C.1 In Section 6A and B, the grammatical categories of the Chinese language of the "Secret History" were presented by means of a systematic description. The intention here is to proceed further in the identification of these categories with the help of translation reference. 6C.1.1 The naming of a grammatical category does not imply its identification with a homonymous category set up in the description of other languages.1 I have drawn as far as possible on the terminology in current use in descriptive linguistics, introducing new terms only when no "equivalent" standard term seemed to present itself. The criteria for such "equivalence" are, however, undefined, so that, even where these are formal (contextual or grammatical), terminological identity does not enter into the statement. Since the descriptive method is both formal and universally applicable, I have avoided the use of terms which are generally employed in analyses that are either wholly or partly not formal-grammatical. This is not to say that the logical or conceptual categories of such an analysis "do not apply", or could not be applied, to this as to any other language.2 But in a linguistic analysis such as is required by a descriptive grammar it seems preferable to confine oneself to linguistic criteria. The presentation of the formally established categories in a systematic statement can then be supplemented by translation reference to the language of description. 6C. 1.2 There are thus as it were two poles from which the language which is the medium of the description impinges upon the language under description. In the systematic statement of grammatical categories the terminology of the description is as it were collocated afresh: it becomes itself a restricted lexical system in the "metalanguage" of description.3 At the opposite pole lies the translation of the text in the language of description. The difficulty lies in the search for points of contact between the language of the text and intermediate regions of the language of description. The relating of the forms of the language of the text as exponents of its categories, and of the categories themselves, to forms and categories of the language of description presupposes some analysis of the latter. It does not fall within the scope of descriptive grammar to make a parallel descriptive statement of the language which,

60

"SECRET HISTORY OF THE MONGOLS' with its "metalanguage", is the medium of description. If this is not to be done, then any statement which relates the forms of the text to the language of description at any point intermediate between the two poles depends on some prior identification of grammatical function between the two languages. 6C.1.3 Such identification of grammatical function across languages, highly desirable as it is, cannot be regarded as resting on a sure foundation before much more systematic comparison has been undertaken. If the criteria for identification are themselves to be grammatical, they must be founded on systematic descriptions of the languages to be compared; but the language of description, not being itself under description, must if its forms are to be used as terms in the identification have already been so described. If such a description is not to hand and familiar we should seek criteria other than grammatical for the grammatical identification. Criteria of contextual reference may provide a basis for the identification of at least some part of the grammatical systems of the two languages.4 6C.1.4 Since the relationship implied in the contextual identification of terms in grammatical systems is not in any sense genetic, but (as it may be called) "systematic", such a relationship may be established between the language of the text and any language which presents itself as a convenient point of reference. In this way in a descriptive grammar we may use the language of description, thus establishing contact at a point intermediate between the poles of systematic statement and translation. We may establish "degrees of relationship" (Allen 1953) such that, when enough systems have been compared, the relative closeness of various fields in the two languages may be demonstrated. 6C. 1.4.1 With the "Secret History", besides comparison with English, the language of description, there is the further possibility of comparison with Modern Pekingese. Such comparison is likely to yield higher degrees of relationship: not that in every system where comparison is possible the "Secret History" will necessarily appear more closely related to Modern Pekingese than to English, but that the number of comparable systems will be greater as between the "Secret History" and Modern Pekingese and that, of the systems comparable in all three languages, the majority will show a closer relationship between the two forms of Chinese than between either of these and English. 6C. 1.5 As an instance of (i) contextual identification of grammatical function and (ii) systematic comparison on this basis, and of the place of these procedures in a descriptive grammar, we may compare the personal pronoun systems of the "Secret History", Modern Pekingese 61

MEDIEVAL CHINESE GRAMMAR and English. Here the systems can be identified as comparable between the "Secret History" and Modern Pekingese by grammatical criteria without contextual reference: in both languages (but not, e.g., in Modern Cantonese) the pronoun enters into the system of nominal word classes in simple opposition to the noun; this can be established by its value in the structure and non-operation in the noun group. For a comparison with the English personal pronoun it is necessary to state that both show contextual reference to persons participating in the linguistic event and categorized as "speaker", "addressee" and so on (Firth 1950; see 3.1.1). This reference is then the criterion for identification of single terms in the different pronominal systems. Terms with the same contextual reference will be identified irrespective of the difference in the total number of terms in the system.5 Such reference must be exactly co-extensive.6 6C. 1.5.1 The (personal) pronoun systems of the three languages can be stated and compared as follows: rEFERENCE 1 11 134

133

SA

Mod.Pek

English

wuo

wue

I

weemen

WE

nam wuomui

1133

nammui

12

za

122

zamui

zamen

2 22 223 2233

ni13 nim nim nimui

ni14

3 33

1122

nimen

You

ta

ta

he

tamui17

tamen

nimmui she they

6C.1.5.2 Any two of these systems may be compared and a degree of relationship stated for each pair. Comparing the 12-term system of the "Secret History" with the 7-term system of Modern Pekingese we find that of the total of 19 terms 4 are identical, giving a possible reduction in a comparative statement to 19 — 4 = 15 terms, or a degree of relationship {f. (In the "Secret History" and English a reduction of only two terms is 62

"SECRET HISTORY OP THE MONGOLS' possible, giving a lower degree of relationship f.) In making a terminological identification between these four terms (for example as first, second and third person singular and third person plural) we should be identifying these terms between the two languages (whereas we should not identify, for example, "perfective aspect" in the "Secret History" and in Modern Pekingese until some similar comparative statement had been made).19 6C.1.6 In fact, though the degree of relationship between any pairs of systems in the "Secret History" and Modern Pekingese may vary considerably, the systems of grammatical categories in the two languages, identified grammatically (with lexical support, but without reference to context), are closely related. In naming the categories of the "Secret History" I have used the same names as can be (and are by myself) applied to categories of Modern Pekingese with which they are identified (at any level), for example "voice" and "aspect" as categories of the clause; where formal identification has not been made, I have used terms also employed for Modern Pekingese where they seemed appropriate, a procedure which seems justified not only for terminological economy but because of the obvious (if unformulated) relationship between the totality of the systems of the two languages and the identity of the techniques of description applied by myself to both. The closeness of the systems of grammatical categories is itself the comparative basis (at the grammatical level) for a view of the "Secret History" and Modern Pekingese as two stages of the same language. 6C.1.7 Since the aim and scope of the present work are descriptive and particular rather than historical and comparative, a systematic comparison with Modern Pekingese, which is not the language of description, should be reserved for separate treatment. Moreover the field of application of contextual criteria to grammatical identification across languages is restricted. For this reason in approaching the language of description at a point intermediate between the "metalanguage" of systematic statement and the language of translation it may be desirable to face the latter pole, as it were, rather than the former; that is to say, in referring the forms of the "Secret History" to English we may take as the point of reference the language of an English translation of the text. A form of the "Secret History" will then be regarded as corresponding to (that is, capable of being translated, in the event, by) a form in the English translation.20 This translation reference may be systematized to the extent that the English form can be assigned to some category definable in use; but the description will always face the pole of translation, 63

MEDIEVAL CHINESE GRAMMAR so that such categories carry no implications of systematized statement of English. 6C.1.8 There does exist, however, a possibility of further assisting the identification by translation reference, through the use of linguistic contextualization. This is possible where the choice of a form can be shown to be determined by some feature of the linguistic context. In a written text, where there is no situational context for any unit lower than the whole text, all reference to context from the level of sentence or clause must in fact be to linguistic context; but whereas pronominal reference may be made on the basis of a situation constructed as it were within the text, the features of which are capable of abstraction such as to permit comparison with other situations and thus identification across languages, direct linguistic context, or "context of mention", may function in the description of the single language.21 This could be viewed as an extension of grammatical statement beyond the limits recognized in the description for grammatical determination: that is, beyond the sentence. Since, however, the determining factor is not the grammatical form, but the presence or absence of a lexical form in the linguistic context (or, with a spoken text, of a lexical form or a term - participant, relevant object or action - in the context of situation), the occurrence of a member of a given class in a structure is here not only determined by the possibilities of the structure itself but also related to non-structural features of the larger unit.22 6C. 1.8.1 In describing, for example, the ergative clause, which in the grammatical system operates in the sentence structure with value F (= free clause), we may find that its occurrence is related to the occurrence in the previous linguistic context of a certain lexical form defined by repetition in the ergative clause itself. In this way we may set up a two-term system of context of mention: the "given" and the "new". A lexical form (which may be verb, noun or adverb), operating in a given sentence or clause, is said to be "given" if it has been mentioned in a previous sentence or clause. The extent of operation of context of mention is not defined, though where it can be shown to operate it is to a certain degree self-defining; clearly the definition cannot be made in terms of a fixed number of clauses or sentences, but rather in terms of the relatedness of context. A sentence or clause is said to have "related context" with a previous sentence or clause if at least one lexical form is given in the context of mention. The use here made of context of mention is only very limited; it is not contended that its application extends to the determination of all or even the majority of grammatical forms. For this reason I have 64

"SECRET HISTORY OF THE MONGOLS" subordinated it in the statement to the translation reference, only appending contextual reference where I have been able to establish a correlation of this type. 2 Sentence and clause

6C.2 The sentence, compound or simple, can usually be translated as one English sentence.23 Sometimes, however, (i) a sentence consisting of parallel free clauses is rendered by more than one; (ii) successive short sentences characteristic of passages of narrative are combined in the translation, and (iii) occasionally two sentences are rendered as one sentence of two clauses, the first subordinate. In Modern Pekingese syntagms of type (iii) are frequently marked as one sentence by the punctuation:24 in other respects sentence division closely parallels that of the "Secret History". 6C.2.1 Since pre-verbal position in the clause is associated with the contextual category of the given (see 6C.2.4.1) in a related context, the presence of a pre-verbal noun constitutes "re-mention" and is therefore considered to mark the beginning of a new sentence;25 where all terms are new, and there is no related context, pre-verbal noun may occur there also and for this reason again is considered to mark a new sentence. Occasionally, where the previous clause is bound, pre-verbal noun marks the second clause in a sentence. The number of clauses in a compound sentence shows an average of about 2.3. Some very long sentences are found, especially where there is parallelism between the clauses, often reflecting passages of traditional narrative in the Mongolian; for example, the imprecation of Hoelun on her sons Temujin and Qasar in (II. 12b-13a; 78.8—19), 26 clauses. (Examples of the sentence: 6D.2.1 are on the CD-ROM provided.) 6C.2.2 Like the sentence, the clause can usually be translated equivalently, as one English clause. Sometimes a genitival (bound) clause may require in translation a form less than a clause (a word or phrase). While to the distinction of free and bound clause frequently corresponds one of main or co-ordinate and subordinate clause in English, a non-sentential free clause, when not sentence-final, is often rendered by a subordinate clause; and in particular it frequently occurs that when a clause, free or bound, is followed26 by a free clause with conjunctive adverb, the whole may be rendered by an English subordinate clause followed by a main clause. Thus in a free clause the translation meaning of a conjunctive adverb may be considered to be the subordination of the previous clause. (Examples of free clauses: 6D.2.2 are on the CD-ROM provided.) Types of bound clause are discussed in detail in 6C.2.7—8. 65

MEDIEVAL CHINESE GRAMMAR 6C.2.3 The verbal/nominal dimension of clause classes is not reflected in translation, though some common feature can be abstracted from the translation equivalents of the pro-verb, which marks a nominal clause; and the absence from the nominal clause system of marked voice and imperfective aspect may be to a certain degree reflected in the absence from its equivalents in the translation of the passive form of the verb and of tense-forms "is (was, etc.) -ing". The verbal clause includes the type of those which in English have predicative adjective. (Examples of verbal and nominal clauses: 6D.2.3 are on the CD-ROM provided.) 6C.2.4 The categories of voice, aspect and mood are categories of the clause. The terms here chosen are those regularly employed with other languages in the description of the verb; the translation and contextual reference, however, suggest some reasons for the use of these terms in the naming of certain categories formally established in the clause. In the system of voice there is no regular grammatical translation reference, and what English equivalence can be observed depends on identification by context of mention. It is possible to make some contextual distinction (i) between marked and unmarked voice and (ii) between ergative and passive. If we consider a given syntagm active in voice, structure N(a)VN(b), where (a) and (b) are contextual referents, we can set up two non-commutable variants (having, that is, the same contextual reference) with marked voice where both N(a) and N^ precede V. Of these variants, the ergative will appear as N(a)N(b)V, the passive as

N N V 27

*>6C.2.4.1 «-

There is some correlation between the dimension of voice in the clause and the contextual dimension in which are opposed the given and the new. The tendency is for the referent of the given to precede that of the new within the limits imposed by the basic grammatical structure. This can be expressed in tabular form (where N l5 N2 are the grammatical (positional) categories, (a), (b) the contextual referents): Voice

Structure

Given

New

Active Ergative Passive

Since this type of contextual reference is normally made in spoken English by intonation/stress, and in written English is ignored, there is no grammatical translation equivalent (except insofar as in a spoken 66

"SECRET HISTORY OP THE MONGOLS' language intonation can thereby be considered a grammatical \ ?9 category). There are two subsidiary structures in the active clause, one of which is likewise contextually determined. These are: (i) NNV(N) which, like the passive, has contextual reference N^N^V, with Nb given and V new (occasionally, like the ergative, N^N^V);30 (ii) (N)VNN which is restricted to certain exponents of V, of which the only regular one is yiu "give".31 6C.2.4.2 In most cases the ergative of the "Secret History" is equivalent to that of Modern Pekingese; sometimes, however, for the ziarj of the ergative Modern Pekingese would have prepositive verb na2 (in active clause). There is perhaps some inconsistency in the assignment of ziarj and the si of the passive to the lexicogrammatical word class of prepositive verb, since their meaning in this structure can be stated purely grammatically and they are unrestricted by collocation,32 rather than to the class of particle; their position in the structure is, however, that of prepositive verb and not that of particle (which follows, or is final in, what it modifies), ziarj is frequently found also as a free verb and as a postpositive verb, and occasionally as a verbal adverb;33 ba, which occurs eight times in ergative clause, is found also twice as free verb. The passive clause is to be distinguished from the special instance of the bound genitival (active) clause (of type (ii) and preceded by si) which it resembles but from which it is formally distinct (see 6C.2.8.2). (Examples of voice: 6D.2.4 are on the CD-ROM provided.) 6C.2.5 In the aspect system likewise it is possible to observe a contextual correlation, and the translation of aspectival forms can be systematized to the extent that it depends on this correlation. In general marked aspect denotes the relatedness (imperfective and non-final perfective) or unrelatedness (final perfective) of the following context. An imperfective clause or a non-final perfective clause has at least one referent given to the following clause or itself operates as a term in the context of the following clause: it has context either related to, or the same as, the following clause; while in a clause following a final perfective clause all referents are new. A clause unmarked in aspect is neutral as to contextual reference, and an unmarked (final/non-fmal) perfective clause is as a category similarly neutral, though any given exponent may be syntagmatically assigned to one or the other term. 6C.2.5.1 The terms "perfective" and "imperfective" are chosen (their use being current with regard to Modern Pekingese) because the reference given by an imperfective clause is one of prolongation, while that given by a perfective clause is one of succession.34 That is to say, a 67

MEDIEVAL CHINESE GRAMMAR perfective clause, in its operation (or that of one of its terms) in contextual relation to the following clause, appears as a term in a series or succession, while an imperfective clause appears as one in a hierarchy or progression. For this reason the perfective has the formal dimension of final/non-fmal, the final marking the last of the series, while the imperfective has no such dimension. It thus happens that a clause with non-final perfective or with imperfective aspect is almost never sentence-final (cf. Prusek (1946): function of subordination), while a clause with final perfective aspect is always sentence-final.35 With regard to non-final perfective aspect, four points may be noted, (i) It frequently occurs in a bound clause, the non-finality of which in a succession is implied in its being bound, (ii) It has no direct time reference, occurring in both free and bound clauses where past reference is excluded: here English (which usually employs past tense forms in narrative, the narrative as such in the "Secret History" being neutral in aspect) may have a variety of forms, including compound past tenses, determined in subordinate clauses by structures of tense-sequence, (iii) It is frequent in parallel clauses, where, however, non-final perfective is usual even in the second (last) clause (if the second had final perfective in fact there would be no parallelism), (iv) There are some instances of it in a paragraph-final clause, where it is prima facie unlikely that contextual reference would extend forward; but the majority of these instances are in the genealogies in Chapter 1, which could be regarded either as extended parallelism or as related in context (with e.g. zuoliau. . . sirjsi6 "founded the clan. . ." given); a few other instances do not seem to be accounted for in this description. 6C.2.5.2 As additional features of aspect may be noted: (i) The imperfective particle Jo is frequently found (47 out of a total of 328 occurrences) in prepositive complex group modifying a prepositive verb (where, therefore, the clause is not itself imperfective); here the same prolongation reference within the clause is marked as a feature of prepositive verb. (ii) There seems to be a correlation between marked aspect and structures with monosyllabic final free verb (cf. above, n., Prusek (1950): aspective particle with simple verbs). This is generally observable throughout, and is particularly pin-pointed by one clause (see example 6D.2.5 no.5 on the CD-ROM provided). The only frequently occurring combination of marked categories is indeed that of ergative voice (NNV) with perfective aspect; here, 68

"SECRET HISTORY OF THE MONGOLS" as in other perfective clauses with final V, the perfective clause is marked merely as one term in a series but unmarked as to final/ non-final. (iii) In Modern Pekingese the combination of perfective aspect with negative adverb normally implies contrast with (i.e. nonoperation of the negative adverb in) the previous context, and can usually be translated as "no longer" or "not, after all (as originally intended)". This is not generally so in the "Secret History" where negative perfective is not a separate term in the system but is merely as it were the sum of its parts; there are, however, a few instances resembling Modern Pekingese (see examples 6D. 2.5 nos. 6,7 on the CD-ROM provided). (iv) Occasionally liau modifies a prepositive verb; this is not considered to constitute a separate verbal system, but rather, like the regular use of jo with prepositive verb, to specify a relation normally linking clauses or sentences as operating within the clause. In a verb group with postpositive verb the position of liau and Jo is following the postpositive verb; Jo is, however, very rare here. For the relation of postpositive verb to perfective aspect, see 6C.3.5.1. (v) Combinations of aspective particle with the postpositive verbs (type (ii)) lai, kiu and yieu do not form a separate system. These are discussed in relation to postpositive verbs (see 6C.3.5.4). 6C.2.5.3 The Modern Pekingese aspect system shows certain important differences, in particular: (i) the presence of a term with (postverbal) gus (perhaps best regarded as a specialized form of the perfective; alternatively as postpositive verb); (ii) the presence of a specialized negative system; (iii) the possibility of occurrence of both non-final and final liau in one clause.36 The systems are, however, clearly comparable; no comparison has been attempted here, in view of the detailed discussion of Modern Pekingese aspect which this would entail. As is implied by the contextual reference,, a free clause with imperfective or non-final perfective aspect may often be translated as an English subordinate clause (with "after. . .", "seeing that. . ." and so forth). An imperfective clause often requires English verb-forms "is (was, etc.) -ing". The imperfective particle Jo should be distinguished from the postpositive verb Jau occurring only after certain free verbs (6C.3.5.1.1), and from clause-final Jo marking imperative mood: all are written with the character Jo. (Examples of aspect: 6D.2.5. Please refer to the CD-Rom provided.) 69

MEDIEVAL CHINESE GRAMMAR 6C.2.6 The category of "mood", like that of aspect, is marked by certain particles in the clause, those of mood being always final. The neutral term in this system is affirmative mood, and no change in the clause structure is associated with the marked terms of the modal system. 6C.2.6.1 The regular form of the interrogative has final particle ma; clauses of this type can be translated with English interrogative verb form. Clauses with madau3(15 occurrences) require an exclamatory form, which may or may not be interrogative in English. There is one interrogative clause with feu (XL 17b; 252.17). All these forms exclude the occurrence in the clause of an interrogative word (some, but not all, of which have ma as final syllable; see 6C.3.17); clauses with interrogative word (noun or adverb) are considered affirmative, there being no grammatical distinction between these and other affirmative clauses. Thus English interrogative verb form is the translation equivalent not only of interrogative mood but also of affirmative mood in "information questions" with words such as suei "who?". Modern Pekingese has not only ma (now written ma2) similarly restricted, but also ni2 which can both replace ma and combine with interrogative word; ni2 is absent from the "Secret History". madau3 is absent from Modern Pekingese; we may compare the latter's (pre-verbal) nandau3 which has, however, a contrary implication absent from madau3 ("(surely) you don't mean. . .?");37 and Modern Cantonese me (m turj . . . me? = nandau3 . . . ma2?) contrasted with interrogative ma. Modern Pekingese has in addition a quite distinct interrogative form of an "alternative" type, characterized by the repetition of (at least) the verb, preceded in the second instance by the negative adverb. This form is very rare in the "Secret History"; perhaps the only two occurrences are (II. 45b—46a; 100.15,20), one of which is an answer to the other. Because of its rarity and incomplete identity with the Modern Pekingese, this form is not regarded as interrogative here. (Examples of interrogative mood: 6D.2.6.1 are on the CD-ROM provided.) 6C.2.6.2 In imperative mood the "Secret History" distinguishes a first, a second and a third person imperative, rendered as such in English but, unlike English, all formally marked. First person imperative is marked by final za (five occurrences), formally identical with personal pronoun za.38 Second person imperative has regularly je, sometimes jo (17 occurrences39); there is no observable distinction between the two. Very occasionally je occurs in clauses (e.g. X. 13b; 238.8-9) translated as first person imperative, je and jo regularly collocate with prepositive verb giau "tell, order", requiring "indirect command" in translation. 70

"SECRET HISTORY OF THE MONGOLS" Third person imperative is marked by final yieje, which occurs nine times, four of these following ]i7or ji7dau3 "know", "let it be known".40 Modern Pekingese has no system of imperative mood, though the final particle ba2 (now normally written ba3), clauses with which may be translated now with interrogative, now with imperative verb form, should be recognized as falling within the modal system. (ba2 in the "Secret History" is regularly a postpositive verb, though it occurs once following yieulai (X. 45b; 246.25) (see 6C.3.5.1.1).) The restriction of certain attributive verbs to clauses which might thereby be regarded as imperative in Modern Pekingese is not observable in the "Secret History". Some clauses requiring imperative or "indirect command" in translation are affirmative in the "Secret History"; these include some with prepositive verb giau, some with negative adverbs hieu and mo (prohibitive: "do not. . .") and others with no formal identification. (Examples of imperative mood: 6D.2.6.2 are on the CD-ROM provided.) 6C.2.7 Of the bound clause classes the conditional is marked lexicogrammatically, by a system of adverbs; these, designated by position as "preverbal" and "final", occur either alone or in combination one with the other. In addition, a small number of conditional clauses are marked by (i) preverbal adverb combined with the (final) genitival particle di, and (ii) one of three postpositive verbs, ba2, ki2 and de. The regular translation equivalent of the conditional clause is a subordinate ("adverbial") clause of the types classified as temporal, conditional, causal, concessive etc. and introduced by a subordinating conjunction. Some, including those with final di, require "whatever, whoever" or "if any, if anyone", etc. Almost all those with postpositive verbs ba2 and ki2 have free verb suo "say" and may be omitted altogether in translation, or rendered "when (he) had spoken". The translation of those with de is very varied and is discussed in relation to the position of de in the system of the postpositive verb (see 6C.3.5.1.3). A conditional clause is never sentence-final unless the following clause is marked as sentence-initial other than by the presence of pre-verbal noun:41 if the following clause has pre-verbal noun the conditional clause itself is considered thereby to be marked as sentence initial. The clause following a conditional clause frequently has a conjunctive adverb, and indeed the presence of a conjunctive adverb may be considered regularly to require the subordination in English of the previous clause whether bound or free.42 The conditional clause is comparable with that of Modern Pekingese, which has, however, a more restricted system of final adverbs and 71

MEDIEVAL CHINESE GRAMMAR combinations with the final adverbs. The preverbal and final adverbs are discussed later (6C.3.12—13). (Examples of conditional clause: 6D.2.7 refer to the CD-ROM provided.) 6C.2.8 Genitival clauses have final particle di and are of two functional types, defined by exterior form: 6C.2.8.1 (i) Immediately preceding a noun (rarely noun group,43 never pronoun) (347 occurrences). The clause in which this noun operates may either follow or surround the genitival clause.44 A genitival clause of this type has attributive function; di is the mark of attribution (compare nominal particle di which marks the attribution of a noun), but in fact such attribution is a function of exterior form and there are a few clauses attributed by position alone to the following noun, not having final di and therefore considered free clauses.45 There are four instances of this type of genitival clause with the literary form Ji3 replacing di.46 Otherwise the form is identical with that of Modern Pekingese, translated like the latter by an "adjectival" clause, with the difference that clauses of this type translated by a relative clause between commas (i.e. "non-restrictive") are considerably more frequent in the "Secret History" than in Modern Pekingese.47 Some polysyllabic final adverbs have first syllable di, as disi2fun "when"; these forms are regarded as compound adverbs because they always mark a bound clause, though clauses to which they are final are syntagmaticaUy identical with the genitival clause plus the following noun. 6C.2.8.2 (ii) Not immediately preceding a noun (51 occurrences). This type, which is rarely sentence-final, has substantive function with value N in the clause to which it is adjacent or internal. The Modern Pekingese form is identical but regularly sentence-final; a particular instance of this type, following si, is very rare in the "Secret History" but so frequent in Modern Pekingese as to require a distinct classification.48 The translation usually has a relative clause with pronominal antecedent: "what (= that which)", "those who", "the one who", etc. All this type have final di, and are distinguished from various related forms: the passive voice structure,49 conditional clauses with preverbal adverb and final di,50 and certain polysyllabic free nouns with final syllable di (see 7B.3).51 There are a few bound clauses with final je but, since most of these have preverbal adverb, all are considered conditional and je is regarded as final adverb marking a conditional clause. (Examples of genitival clauses: 6D.2.8.2 refer to the CD-ROM provided.) 6C.2.9 The translation of the verb group, the noun group and the complex group is discussed in relation to the word-classes making up the groups. 72

"SECRET HISTORY OF THE MONGOLS" 3 The word 6C.3 In the consideration of the translation of word-classes, some correspondence with English word-classes can be indicated, with (in the case of the bound word-classes) some reference to the translation of the group in which these operate. In addition translation reference by means of "specimen translation" is given for members of a wordclass which are of particular interest or of frequent occurrence. Reference to context of mention is valid in some instances in relation to the group; and some comparison is made with Modern Pekingese forms. 6C.3.1 A free verb may have substantive or attributive function, and attributive function permits the further distinction of the free verb into the sub-classes "transitive" and "intransitive". 6C.3.1.1 (i) A free verb, which in combination with a following noun forms a verbal clause is said to be "transitive".52 Such a verb never has attributive function, since in structure VN it is always substantive (its "attribution" taking the form of a genitival clause), while there is no unambiguous instance in the "Secret History" of a transitive verb attributive to a verb. Only a transitive verb can have substantive function in a clause with marked voice. A transitive verb is usually rendered in English by a (finite) verb. This may be of the type recognized in English as "intransitive", for example lai "come". 6C.3.1.2 (ii) A free verb which in combination with a following noun does not form a clause but is attributive to it, the combination thus having value N in the structure, is said to be "intransitive". An intransitive verb may thus have substantive or attributive function, but it never has substantive function in a clause with marked voice or with post-verbal N.53 An intransitive verb in substantive function is usually rendered by an English predicative adjective with "is" etc.; in attributive function by an attributive adjective or, if attributive to a verb, by an adverb.54 (Examples of free verbs: 6D.3.1 are on the CD-ROM provided.) 6C.3.2 The pro-verb has value V in the structure of the nominal clause, the order of elements in which is, as in the active verbal clause, (N)V(N). Specimen translations of the eight pro-verbs of regular occurrence in the "Secret History" are as follows: svi yieu wu

"is" "there is; has" "there is not, has not"

73

MEDIEVAL CHINESE GRAMMAR zai mir) fu and s suei2 preceded by numeral nouns

"is at" "is called" "is like" "is . . . (years old)"

Of the 30 occurrences of pro-verb fu, six are in fu ho "what is ... like?, how is?, why is?" and four in fu x "is like this". In Modern Pekingese the category can be established on the same criterion of non-operation in the verb group; Modern Pekingese has rmi (mui2) or maiyigu for wu and (hau)siar) for fu, s. It is doubtful whether to regard yiau as a pro-verb in Modern Pekingese, where, unlike maiyisu, it regularly combines with an auxiliary verb. Or the 118 and 30 occurrences of pro-verbs yiau and wu, seven and one respectively are in structure NV (one NNV) where VN might be expected.55 This is not unparalleled in Modern Pekingese, especially where the N:V relation is that of given:new in the context:compare (1.36b; 55.11) s wuo ban fu4fin yieu yieje "let there be a(nother) woman like me". To this may be related the regular use of yieu as postpositive verb (type (iib)) (see below, 6C.3.5.5). All pro-verbs occur also as either prepositive or postpositive verbs, or both: the identification of some of these is given in the table in 6C. 3.5.3.1. (Examples of pro-verbs: 6D.3.2 are on the CD-ROM provided.) 6C.3.3 Of the bound verb classes, the auxiliary verb operates in both the verb group, where it precedes the free verb, and the prepositive complex group, where it precedes the prepositive verb.56 Those of frequent occurrence in the "Secret History", with specimen translations, are: lai and kiu ko and koyi2 yiau and yiu3 nerj and de ken gam bei bi2 svi58

"to, in order to, came/went and"57 "can, may, is qualified to, it may be that" "will, is going/wants to" "can" "is willing to" "dare" (passive verb form) "must"

lai and kiu frequently occur following another auxiliary verb or a quasicomplex group (see 6C.3.4.2). The system is comparable with that of Modern Pekingese, where however yiu3 and the less common yiu3yiau are absent, while ks (ko),

74

"SECRET HISTORY OF THE MONGOLS" regular (69 occurrences) in the "Secret History", is narrowly restricted in Modern Pekingese which has regularly ksyi (koyi2); the latter form in the "Secret History" (19 occurrences) never follows a negative adverb and in nine instances can be translated "is qualified to". A number (but not all) of the auxiliary verbs occur also as free verbs; some of these and others also as prepositive verbs, ziarj, which also occurs frequently as free verb ("bring, take"), as prepositive verb in ergative clause and as postpositive verb (see 6C.3.5.2), occurs ten times as auxiliary verb, translated as "will": compare Modern Pekingese adverb jiarjlai (ziarjlai) "in the future", which occurs once in the "Secret History" where it could be regarded as auxiliary verb ziarj plus auxiliary verb lai; and "Secret History" (once only, in (II. 32b; 91.4)) ziarjx2, adverb "about to". (Examples of auxiliary verb: 6D.3.3 are on the CD-ROM provided.) 6C.3.4 Prepositive verbs are distinguished into two types according to whether or not they operate in a complex group having adverbial value in the clause structure. 6C.3.4.1 (i) A prepositive verb of this type combines with following N into a complex group which normally has direct pre-verbal position, occasionally initial position, in the clause; the complex group then has adverbial value, so that (vN)V = AV. Prepositive verbs of this type are frequently modified by the imperfective particle jo, recalling the -ing form of words used in their translation. The most frequent prepositive verbs of this type, with specimen translations, are: yiu zai and yiu2 fu duei (collocating with free verb suo "say") bei z2 Ji and dau zuo yiurj Iirj2

"with (accompanying), for (on behalf of)" "at, in, on" "like" "to" "by (with passive verb form)" "from, through" "to, reaching, by (time)" "as, being" "with (using)" "with, leading"

The translation of the complex group is thus usually a prepositional phrase, the prepositive verb (together with the postpositive noun where present) being rendered by a preposition (with the postpositive noun there is often the possibility of a compound preposition: e.g. zai ... heu "behind, at the back of", dau . . . hsu "behind, to the back of").

75

MEDIEVAL CHINESE GRAMMAR The Modern Pekingese system, is comparable; of the forms listed above, some are not usual in Modern Pekingese, their regular equivalents being shown in the following table with occurrences of both in the "Secret History" in parentheses: Secret History yiu (171) z2 (60) ji (47) yiu2 (100) nij (100)

Modern Pekingese gan2 (-),turj (16) "accompanying" gai(-) "on behalf of" xurj (13) dau (37) zai (51) gan2 . . . yiyiarj (-), etc.

Of the occurrences of ru in the "Secret History", 58 are in ruho "how?", Modern Pekingese Z9nma(yiarj), and 12 in fux "like this", Modern Pekingese jayiarj (Je2yiarj), nayiarj. Frequent in Modern Pekingese also is na2 (1), of which the equivalent in the "Secret History" is ziarj in ergative clause. For the relation between prepositive and postpositive complex group, see6C.3.5.3.1. 6C.3.4.2 (ii) A prepositive verb of this type occurs in formally identical combination with following N, but the combination (vN) has no value as such in the structure, and is regarded as forming a "quasi-" complex group.59 Most frequent among this type, which never has imperfective particle jo, are: giau mirj2 and lirj gian si4 cai suo ji7 hiu kurj

"tell (order)" "order" "see" "make (cause)" "send (order . . . to go and)" "say" "know" "permit" "be afraid"

The translation of the quasi-complex group thus takes the form of a verb followed by noun clause or by noun and verb infinitive: "told him (to go)", "saw him (go)", "know that he (went)", etc.; including "say", etc. with reported speech.

76

"SECRET HISTORY OF THE MONGOLS" Modern Pekingese has a comparable system, in which some rather more usual equivalents are as follows: Secret History giau (230) lit) (19) gian (42) si4 (34) cai (30) Ji7 (17) hiu (16) kurj (12)

Modern Pekingese jiau (giau2) (-) mir)2 (43) kanjian (-gian) (-) s (si4) and jiau (giau2) da/a (-) J7dau3 (-) siu (hiu) and far) (-) kurjpa(2),pa(2)

Some prepositive verbs of this type are disyllables of which the second character is regularly a postpositive verb (e.g. tirjde "hear"); since, however, these are relatively few it seems simpler to regard them as compound prepositive verbs than as prepositive verbs followed by postpositive verb (restricting the latter to occurrence following a free verb). Among collocations regular with prepositive verbs of type (ii) is that of giau with imperative final particle ]e.60 6C.3.4.3 A special sub-type within this type of prepositive verb is formed by the three words si, yieu and wu. This might be considered a third type intermediate between types (i) and (ii), or the forms could be assigned to both types, since it is sometimes doubtful whether to regard the group in which they occur as a complex or a quasi-complex group, and sometimes one, sometimes the other seems preferable. In translation, si and yieu are often omitted; if translated, their rendering when preceded by a noun resembles the forms used in type (i): si "(being =) is . . . and" (or passive verb form, e.g. (II. 44a; 99.7); compare passive form), yieu "having, with, has . . . Who/which"; without the preceding noun, the form is more comparable with those of type (ii): si "it is (. . . Who/ which)", yieu "there is (. . . Who/which), some", wu with preceding noun may be translated "without, not having, has not. . . Who/which", without preceding noun as "(there is) no (. . . Who/which), no one/ -thing".61 Modern Pekingese has s and yisu, while wu is replaced by m9i(yiau). yisu, however, is more restricted, the use to which corresponds zero translation reference (for example (I. 23a; 38.3) yieu na fu4rin hueidau3 "the woman answered"), which is very frequent in the "Secret History", not being found in Modern Pekingese.62 Likewise transitional between the two types are the words bei and

77

MEDIEVAL CHINESE GRAMMAR giau, here assigned to types (i) and (ii) respectively, where in most instances they clearly belong. Examples where assignment would be doubtful are: (IV. 21a; 136.6) hogu bei Jurkin rux zuo "why do we let the Jurkin do this to us?"; (IV. 48b (198b); 146.13) Jar)3fu2 gian bei giun yiausa "I saw the soldier going to kill my husband"; (IV. 44b; 145.57) zaixian bei Merkit yiu2 Burqan san kun wuo si2 "that time when I was surrounded by the Merkit on Mt. Burqan"; (IV. 32b; 141.11-13) bei ... Qoridai dau Gurelgu di3mian gau yiu Cirjgis "(their plot was revealed by) Qoridai (who) went to Gurelgu and told Genghis"; (VI. 53b; 185.19—20) wuo yiu2 jirjju bu fin2 giau nim na2 kiu sa liau "I would not let my own lord be taken and killed by you"; (VII. 13a; 189.32) ziarj zaixian lau Worjqan giau ziantur)2 hu3 de zeuculai s3 liau, kanlai ta gam yiau zuo huarjdi6 madau3 "Seeing that Worjqan was so frightened that time by the quivers that he ran away and got killed, does it look as though he had the courage to become emperor?" (Examples of prepositive verbs: 6D.3.4 are on the CD-ROM provided.) 6C.3.5 Postpositive verbs are divided into two main types according to their incidence within or outside the group, while subdivisions within those types depend on position in the group or clause and certain possibilities of combination. Type (i) are those which operate in the verb group or postpositive complex group. These occur either directly post-verbally (sometimes separated from the free verb by another postpositive verb of this type) or following post-verbal N but in this case always followed by N. Type (ii) are those which operate outside the group and take clause-final position, separated (where applicable) from the free verb. 6C.3.5.1 Type (i) sub-type (a) includes the majority of postpositive verbs: these operate in the verb group and always directly follow the free verb. It is impossible to point to an English category as translation equivalent of this type. The comparable class in Modern Pekingese (see 6C.3.5.1.1) has been variously regarded as marking perfective aspect, resultative aspect, causative form and other categories63; and it has been noted (Frei, 1941; cf. 7A.3) that a free verb on the one hand and the same free verb with postpositive verb on the other often require lexical distinction in other languages (e.g. Jau2 and ]au2dau, English "look for, seek" and "find"). With regard to the "Secret History", where the system is closely related to that of Modern Pekingese, the same is true. We may instance this and two other general types of translation reference: (i) use of a distinct lexical form, e.g. wuaqgian "see" (contrast wuarj "look at"); (ii) post-verbal adverb, e.g. (da2)daw2 "(knocked (him)) over"; (iii) use of a 78

"SECRET HISTORY OF THE MONGOLS' lexical form which would itself be the translation equivalent of the postpositive verb as free verb,64 e.g. JQ^ieu "saved" (where je 4 = "protect" and gieu as free verb = "save"). With regard to contextual reference, two forms may be observed: (i) the free verb is given, the postpositive verb new, e.g. (IV. 29a; 140.2—3) . . . s2bo si2 . . . bodau2 "when (they) were wrestling. . . (he) wrestled (him) over (threw him)"; (ii) the whole verb group is (marked by the postpositive verb as) new, this being further suggested by the fact that the postpositive verb is particularly frequent in ergative clauses (and, among active clauses, in those with no post-verbal noun). That the postpositive verb cannot here be regarded as marking perfective aspect is shown by its occurrence, admittedly infrequent, with imperfective particle jo, e.g. (II. 33a; 91.18) Ii3ju2jo luoheu liau "stopped and (stopping) were left behind". 6C.3.5.1.1 Those most frequently met with (with some indication of their specimen translation!) are: de yiu Ju2 dau dau3 gian po

(diau2de "hooked"; diau2 = "fish, angle") (andsee6C.3.5.1.2-3) (huanyiu "gave in exchange"; huan = "exchange") (na2]u2 "grip"; na2 = "take (hold of)") (laidau "came up, arrived"; lai = "come") (wundau3 "asked (saying)"; wun = "ask") (tirjgian "hear"; tirj = "listen") (kampo "clove, broke"; kam = "chop, strike")

This type corresponds to the system of stressed postpositive verbs in Modern Pekingese. In the Modern Pekingese postpositive verb system many of the same forms occur. Of those above, ds (de) is rare in this use (but operates in the negative/potential system: see below), dau being the regular equivalent; yiu is replaced by gsi (which, however, is assigned only to sub-type (c)), while dau3 is absent from the system, the only relic being the compound free verb susdau3 "say". jau is distinguished from the graphically identical Jo marking imperfective aspect, the transcription retaining the Modern Pekingese distinction; there are 22 occurrences, the majority following the free verb yiu6 "meet". ba2 and ki2 occur (15 and six times) only, with one exception each, following suo "say"; they are regarded as marking a conditional clause ("when (he) had spoken"). Modern Pekingese equivalent wuan2 is not found in the "Secret History". 6C.3.5.1.2 There is with this type of postpositive verb a sub-system of negative/potential with four terms in two dimensions, one term 79

MEDIEVAL CHINESE GRAMMAR

unmarked; this is formed by the negative adverb bu and the postpositive verb de65 in position between the free verb and the postpositive verb, thus: FPo, (unmarked); F-bu-Po, negative; F-de-Po, potential; F-bude-Po, negative potential.

Examples from the text of the marked terms are: (negative) (VIII. 48b; 208.12) xsi2 ruo se2bujurj Sarjgun "if you had not hit Sarjgun then" (se2 = "shoot"); (potential) (II. 51b; 103.12) wuo suoyi2 duo2deguo "that was how I was able to escape"; (negative potential) (II. 49a; 102.5) hirjbuderu2 "could not get in". Since adverbial modification is by immediate precedence, it is the postpositive and not the free verb that is modified by the negative: the first example might be (non-contextually) translated "if, when you shot at Sarjgun, you had not hit him". In Modern Pekingese this system is much more extensive than in the "Secret History", but has only three terms, the negative potential — buds - being absent, while the negative term is often translated "cannot" and contrasts with the form with a negative adverb (bu, more often mai (mui2)) preceding the free verb (in the "Secret History" the verbal adverb buxerj, equivalent of mai (see 6C.3.16), likewise precedes the free verb). Modern Cantonese has the four terms FPo, F-m-Po, F-dak-Po and m-F-dak-Po (with dak and the regular negative adverb m) in a comparable system. 6C.3.5.1.3 The postpositive verb de occurs also a regular postpositive verb of this type; it never occurs, however, in potential form (i.e. is never repeated). It occurs in three formally distinct types of syntagm, requiring different translation forms: (i) in negative form or in interrogative clauses or clauses with interrogative word, translated as a potential form of the free verb - (VII. 17a; 190.14) wuo zuobude ni yieu4seu "I cannot act as your right wing"; (ii) in other ("positive") clauses, not clause-final, usually requiring a distinct lexical form (but sometimes possibly "get, catch") - (II. 8b; 76.3) diau2de yi go gim2se yiu4f "hooked a golden fish" (or "was fishing and caught. . ."); (hi) in positive clauses but with clausefinal position, marking the clause as bound (conditional) as sometimes reflected in the translation of this and the following clause as one: (II. 6b; 74.8) yiarj2de rzmui jarjcirj liau "her sons grew up in her care (by her tending)".66 It is sometimes doubtful whether to regard a form as potential (F-de-Po) or as two clauses (F-de/F), e.g. (DC/ 26b;219.8) laideci liau or laide/ci liau "came late" — here the latter is preferable as ci does not occur elsewhere as postpositive verb.

80

"SECRET HISTORY OF THE MONGOLS" In Modern Pekingese the first type is restricted (to a use which may be exemplified by psgbuds (-de) "can't stand knocking about"), de here in general being replaced by other postpositive verbs; in type (ii) it is, as noted above, replaced by others, but a number of Modern Pekingese compound verbs have da as the second character. Type (iii) is regular in Modern Pekingese. 6C.3.5.2 Type (i) sub-type (b) contains 11 postpositive verbs, operating (like those of sub-type (a)) in the verb group following the free verb, but distinguished by certain possibilities of combination one with another. If these eleven are stated in three groups, symbolized x, y and z, combinatory possibilities (where F = free verb) (in addition to Fx67, Fy and Fz) are: Fxz, Fyz, Fxyz. The members of the groups are: group x (1 member) group y (8 members) group z (2 members)

ziag huei, cu, zin2, ru2, ki, hia, guo, zeu68 lai, kiu.

Specimen translations of group y, which usually requires an English post-verbal adverb, are: "went into the forest" (fu2, in (II. 14b; 79.20)), "drive the horses out" (cu, in (II. 31b; 90.45)), "has run away" (zeu, in (II. 19b; 82.3)). Group z will usually not figure in the translation but may affect the form of the verb: "went in pursuit" (not "came") for si2ziar) kiu in (IV 3b; 128.7).69 ziarj, group x, should, it seems, be grouped in some way with lai and kiu, though not into a single system since lai and kiu are mutually exclusive, while neither excludes ziarj. It requires no translation but frequently collocates with free verbs denoting movement where English may have "over", "along", "away", etc.70 The comparable type in Modern Pekingese is that of unstressed postpositive verbs, which likewise have combinatory possibilities, though only in two groups (y and z;jiarj (ziarj) being absent). Modern Pekingese excludes fu 2 from the postpositive verb system, having as equivalent jin (zin2) (sense "in"; the "Secret History" has zin2 in the system (11 occurrences) but in sense "forward"71), and has zsu (zeu) as type (i a) (noncombining, stressed); but includes in this type sag which does not combine in the "Secret History" (it probably should be assigned to this sub-type, especially in view of its special relation to hia (in translation sag = "up", hia = "down"); but since there are no occurrences of sag in combination there is no alternative but to assign it to sub-type (a)).72 6C.3.5.3 Type (i) sub-type (c) contains 12 postpositive verbs operating not, as those of sub-types (a) and (b), in the verb group, but in the postpositive complex group. Three features distinguish the members of 81

MEDIEVAL CHINESE GRAMMAR this sub-type: (i) they may occur separated from the free verb by a noun; (ii) they may follow a postpositive verb of type (i a); (iii) they never occur without a following noun. The translation resembles that of the prepositive complex group, and indeed all the members of this sub-type are also prepositive verbs of type (i) (but not vice versa).73 The 12, with specimen translations (and compare 6C.3.4.1),are: zai and yiu2 yiu z2 ji and dau zuo and wuei ru yieu wu mirj

"at, in, on" "for" "from, through" "to" "as" "like" "with, having" "without, not having" "(who is) called, named"

Modern Pekingese recognizes this type, with some difference in the forms in regular use: yiu2 ("at"), ji and wuei are excluded, zai, dau and zus (zuo) or darj being regular (respective) equivalents: yiu ("for") is replaced by gai, z2 by xurj, ru by e.g. gan (gen2) . . . yiyarj and wu by m.9i(yi9u); in addition to mirj there is the comparable sir) "(who is) surnamed". 6C.3.5.3.1 This lexical identification of postpositive verbs of this sub-type with prepositive verbs extends to the identification of most of the members with either free verbs or proverbs. The following table shows the forms identified, with occurrences in the "Secret History":

yiu2 Z2

yiu ji dau zuo zai ru mirj wuei yieu wu

Prepositive 100 60 171 47 37 26 51 100 9 1474 103 13

Postpositive 19 1 70 33 31 52 26 4 13 21 475

1

82

Free verb

Pro-verb

freq. freq. freq. freq. 35 30 90 4 119 30

"SECRET HISTORY OF THE MONGOLS' There is thus, with these exponents, a formal identity of prepositive and postpositive complex groups. This distribution is again linked with contextual reference, the later position being as usual that of the new: for example (II. 45b; 100.5) giau zuo3 zai heicez li "told her to sit in the black wagon", but (III. 16b (116b); 110.10) Borte zai na baisirj nei tirj jo "Borte heard him from among the folk" where heicez "black wagon" does not occur above while baisirj "folk" does (110.7).76 6C.3.5.4 Type (ii) of postpositive verbs includes two systems, subtypes (a) and (b), of forms occurring in clause-final position; they are, however, positionally distinguished, as those of sub-type (a) precede final (perfective) liau, those of (b) follow it (but precede the modal particles, which have absolute final position). Where the two occur in the same clause, (a) precedes (b). Since they do not directly follow the verb, they might be considered as belonging outside the class of postpositive verb; but since most of the forms are lexically identified with postpositive verbs of type (i), their assignment to this class seems justifiable. The two systems have respectively place and time reference. Sub-type (a) includes only the two forms lai and kiu. These terms operate in a simple dimension of direction towards or away from the speaker or narrator, sometimes other relevant person or object. They thus resemble lai and kiu in type (i) and likewise frequently require no translation; often, however, they may affect the choice of verb (e.g. "come", "bring" in clauses with lai, "go", "take" in clauses with kiu), and sometimes are rendered analytically, as (it might be) "walked up", "send here" contrasted with "walked away", "send there". Modern Pekingese has an identical system. In both the "Secret History" and Modern Pekingese there is a tendency for these forms to occur in sentence-final clauses. 6C.3.5.5 Sub-type (b) has the four terms yieu, yieulai, lai and yie, occurring 15, 20, 24 and 9 respectively. This system has direct time reference, and would if its use were more extended be considered a fourth category of clause classes along with voice, aspect and mood: the terms would be neutral (unmarked), present (yieu and yie) and past (lai and yieulai). Since, however, not many more than one in a hundred clauses (67 out of 5719) would have marked "tense" (if it were so called), it seems preferable to consider it a postpositive verb system with time reference in contrast to the place reference of sub-type (a). In view of the special interest of this sytem (the absence of any comparable system in Modern Pekingese), every instance has been either quoted or referred to in the examples (6D.3.5 on the CD-ROM provided). It may be noted here that of the occurrences of yieu, six are in 83

MEDIEVAL CHINESE GRAMMAR

nominal clauses with pro-verb zai "is at" and of the rest three are in imperfective and one in unmarked perfective clauses; of yie, seven have lai as free verb or postpositive type (i), one of these and the other two are perfective, while four have adverb rugim "now" and one has auxiliary verb ziarj "will, is going to". Of clauses with yieulai, 14 have adverbial past time reference (as with xerj, zaisian, etc.); of those with lai, 15 have adverbial past time reference. All clauses with postpositive verbs of this type are free. There is an interesting correlation with the English tense forms of the translation. The equivalents given here (with the verb "come" as example) are appropriate in every instance and in many seem to be demanded: yieu and yie "comes, is coming", yieulai "has been coming, used to come, was coming; came (after subordinate clause)",lai "had come, came".

yieu shows a preference for nominal clauses and clauses with intransitive free verbs, yie for those with transitive free verbs.77 This system is absent from Modern Pekingese, which has, however, one comparable form laija (laijo), the equivalent of lai here (e.g. in (II. 37b; 94.2) "when Temujin was nine, he had been parted from Dei-secen's daughter Borte-wujin"; Modern Pekingese could certainly have laip): laijs may, with gua (absent from the "Secret History"), be included in the Modern Pekingese aspect system.78 (Examples of postpositive verbs: 6D.3.5 are on the CD-ROM provided.) 6C.3.6 The free noun is regularly rendered by an English noun. The members of this class include the many personal and place names in the "Secret History": the vast majority of these are Mongolian words, but their place in the Chinese system is identical with that of Chinese free nouns. A free noun regularly has attributive function to another (following) free noun; it is sometimes doubtful whether to regard a given syntagm as a single (compound) free noun or as two free nouns, the first with attributive function: in general if both elements are found separately as free nouns substantive the latter statement is preferred, otherwise the former. A free noun may operate in a noun group; personal and place names do so only infrequently. It may be followed by the plural particles mui and der), especially personal names. It may be followed by di, marking its attribution to what follows; there is also a small group of free nouns, polysyllabic with di as the final syllable (see 7B.3).79 6C.3.7 The pronoun has been discussed previously (6C. 1.5.1). (Examples 6D.3.7 are on the CD-ROM provided.) 6C.3.8 The numeral noun occurs as the first element in the 84

"SECRET HISTORY OF THE MONGOLS' structure of the noun group. The forms of the noun group in which it occurs have structure (i) NuAuF(Po), with a rare variant FNuAu; (ii) NuF(Po), and (iii), the "minor noun group" without free noun, NuAu. Forms of the noun group in which it does not occur are (i) FPo and (ii) AuF(Po).80 The numeral nouns are: yi f2 Harj

"-one" "two" (structures (ii) and (iii) only, and in (iii) only in di4f2 "second" and yi ?2 "one or two") "two",

the numbers three to ten and all compound numerals ending in the numbers one to ten, and gi

the unmarked plural numeral which is variously translated as "how many?", "a few", "which (one)?", etc.

To this class corresponds in the system of categories the "determinative noun" of Modern Pekingese, which, however, is more extensive; Modern Pekingese has ja (Je2) "this", na "that", msi (mui) and ga (go3) "each, every", followed by an auxiliary noun: in the "Secret History" these words are not followed by an auxiliary noun.81 In Modern Pekingese the noun group without an auxiliary noun (structure (ii)) is much more frequent with these words than with numerals, while in the "Secret History" the numeral regularly occurs in the noun group without an auxiliary noun. The following table shows the occurrences of numeral nouns in noun groups (i) with and (ii) without an auxiliary noun: Numeral yi "I" r2 "2" liarj "2" (3 to 9) si3 "10" gi

With NAu 128 3 54 81 13 9

Without NAu 79 63 23 119 11 —

Total:

288

295

82

The words ban2 "half", bai "hundred", xian2 "thousand" and wuan "ten thousand" are auxiliary nouns. (Examples of numeral noun: 6D.3.8 are on the CD-ROM provided.) 6C.3.9 The auxiliary noun in the "Secret History" is a small class of 85

MEDIEVAL CHINESE GRAMMAR which only one member occurs more than 25 times (go, 223 occurrences). Two types may be recognized according to the structural type of the noun group in which they occur. 6C.3.9.1 Type (i) are those which may occur in all forms of the noun group, including the minor noun group. Of these only go occurs (29 out of 223 occurrences) without the numeral noun. In the translation some are omitted, their occurrence being determined solely by collocation with a given free noun: ba 5 pf ma3 "eight horses", wu5 j/ 2 ziangan4 "five arrows".

go, which likewise requires no translation, is, however, collocationally unmarked, having the possibility of occurrence with any free noun. Others require the limiting of the free noun by a collective or quantitative: yi xuiJ2 baisirj "a throng of folk", yi WUTJ ma3nai2z "ajar of maresmilk".

The auxiliary noun with numeral "one", and go without numeral noun, is translated as "a, an" or "one". Where the free noun follows the numeral plus auxiliary noun, the translation is unaffected: san2ma3 ba5 pi "eight geldings", gie2yiarj4 yi xian2 bai2tuo2 yi go "a thousand wethers and a while camel".

The minor noun group with auxiliary nouns of this type occurs where the auxiliary noun is given in the context of mention together with a free noun (occasionally where a collocable free noun only is mentioned). The Modern Pekingese system is much more extended, both in the number and frequency of those with restricted collocability (such that the unmarked ga (= go but now written go2), while still preponderant, is less decisively so), and in its frequency of occurrence in the noun group (such that the structure NuF is restricted to a few exponents of F83). 6C.3.9.2 Type (ii) is a small class of auxiliary nouns which occur only in minor noun groups and only in post-verbal position. In practice these can almost always be translated as "time" (where French would have "fois") or, synthetically, as "once" "twice", etc., though sometimes a more specific form is appropriate: liar) x2 "twice", cl2 sam hia "will be flogged three strokes".

The question of the separate classification of this type in Modern Pekingese is a difficult one; but while the "Secret History", being a limited text, permits its formal distinction on the basis of nonoccurrence with the free noun, one form which may favour a separate 86

"SECRET HISTORY OF THE MONGOLS' classification in Modern Pekingese is almost totally absent from the "Secret History": there is only one instance of the type (such as kankan, kanyikan "have a look") so frequent in the former where a free verb is repeated with or without intervening yi "one". The unique example is (II. 25a; 86.4) seu yi seu, translated "search"; alternatively "make a search". (Examples of an auxiliary noun: 6D.3.9 are on the CD-ROM provided.) 6C.3.10 The postpositive noun occurs as final element in the noun group. Whether or not the noun group of which it forms a part is itself part of a complex group, the postpositive noun can usually be translated by an English preposition, the latter being, with the complex group, the translation equivalent of the prepositive (or the postpositive) verb plus postpositive noun. About half of the postpositive nouns in the "Secret History" are simple (monosyllabic) words, the other half compound, disyllabic; many of the latter are formed from a simple positive noun plus following bian2, mian, gian2, or teu2. Frequently the translation of such a pair is identical, and it is unnecessary to regard the two as separate types; it may be noted, however, that compound postpositive nouns (alone among the bound nominal classes) are lexically identifiable with free nouns. Those of frequent occurrence, with specimen translation, are: li, Iiteu2 cu2 nei sag, sag mian, sagteu2 hig ban, yiban bian2 Jug,juggian 2 xian, xianteu2, xianmian durj2bian2 gian2, Iiaggian2 genxian84

"in (inside, into)" (see below) '' among, inside'' "on" (see below) "like" "beside" "amidst, between" "in front of" "to the east of" "between" "in front of" (see below)

Modern Pekingese has a comparable system, in which most of the same forms operate; there are, however, some differences, yiban is usually replaced by yiyiag, Iiaggian2 by Juggian2, Jug often by dag jug. Of the remainder listed above, three (hig, cu2 and genxian) are absent from Modern Pekingese. 6C.3.10.1 hig, cu2 and genxian form a separate sub-system in that, unlike other postpositive nouns, they may occur following pronouns.85 87

MEDIEVAL CHINESE GRAMMAR hir) may be regarded (with Haenisch) as a grammatical translation form, the equivalent (as in the Gloss) of the Mongolian noun-terminations of the accusative and locative (occasionally other) cases. It is possible however to circumscribe the use of hirj, together with cu2 and genxian, in the Chinese version, (i) They follow nouns of human reference (pronouns, personal names, kinship terms and the interrogative word suei "who?") and place names, (ii) The noun group in which they operate, whether or not preceded by the prepositive verb, regularly takes a preverbal position and has the value of a prepositive complex group. They seem in many instances to be interchangeable, and will often not figure in the translation; where they do, the translation may be more or less specific, the more specific being the more differentiated — hirj collocates more readily with free verbs of motion ("to"), cu2 and genxian with those of location ("at"). Sometimes cu2, omitted in English, would require French "chez"; while genxian collocates with suo "say (to)",86 and with place names may be translated specifically as "in front of", "at the foot of".87 Modern Pekingese has no exact equivalents, but would often have postpositive nali ("there") for hir) and cu2 2, sometimes di di2farj for thepostpositive nali ("there") for hir) and cu2 2, sometimes di di2farj for the latter. Modern Cantonese has a (phonological) variant of cu2 as postpositive noun, beside the regular form in other uses. With genxian may be compared Modern Pekingese (especially) ciantgu (xianteu2) in this use.88 (Examples of postpositive noun: 6D.3.10 are on the CD-ROM provided.) 6C.3.11 As a category the free adverb belongs to the lexically defined, "infinite" word classes; but the text of the "Secret History" is limited in extent and in fact contains 296 different members of this class. A few of these are listed in 7C.6—10. Unlike the other free word classes, the free adverb does not admit a distinction of function into substantive and attributive; its class function is attributive. Three types are distinguished according to the unit modified (clause or word) and, if a word, the class of word (verb or noun). Of the three types, only type (i) is free in the sense that the free verb and noun are free.89 6C.3.11.1 Type (i) are clausal adverbs. They take pre-verbal position, initial or non-initial, in the clause which they modify: in general, initial position is that of the given, non-initial that of the new. The majority have time reference, including specific (mirj3ri "to-morrow") and non-specific (nasi2 "at that time"); the remainder, including interrogative adverbs,90 are such as are often designated as adverbs of cause, manner and so forth (ruho "how, why?", yinx "therefore", je2ban "in this way"). 88

"SECRET HISTORY OF THE MONGOLS" 6C.3.11.2 Type (ii) are verbal adverbs. These take immediate preverbal position, which may therefore in a given syntagm be ambiguous: an adverb which cannot occur separated from the verb is considered verbal, but in the text many are left in doubt. Since the distinction is valid for Modern Pekingese, the difficulty may be avoided by a "squint" at the latter; if no reference were to be made to Modern Pekingese, there would be no alternative but to regard all those which never occur separated from the verb as unmarked, limiting type (i) to those found in marked clausal position. The verbal adverbs include negative adverbs and a number with time reference (including complex time-negation forms: see 6C.3.16) and place reference (yi3z2 " already ",Je2li "here"); the remainder are of "number", "extent", "degree" and so forth (du "all", xinz2 "-self",hauserj "very"). 6C.3.11.3 Type (iii) are nominal adverbs. These immediately precede a noun and are rarely ambiguous in the text. One group of nominal adverbs occur between nouns and may be described as "linking" (birj "and"); the remainder may be designated "pointing" (singular reference) (je2 "this", x2 "second, next") and "grouping" (plural reference) (ju3 "all"). These classes may be recognized in Modern Pekingese, with many of the same exponents though with perhaps more lexical variation than in the free verbs and nouns (though less than in the bound adverbs: svee below). As an example, there are 13 compound clausal adverbs with ri ("day") as final syllable in the "Secret History": ten of these remain in Modern Pekingese but with tian replacing ri, the other three retain f (ri) but are infrequent; tian does not occur in adverbs in the "Secret History". On the other hand such distinctive forms as the clausal adverbs guo2fan and yiuanlai (Modern Pekingese gusfan s ni "so it was you! (as predicted)" and yiuanlai s ni "so it was you! (no prediction)") occur in the "Secret History" with exactly the same connotations as in Modern Pekingese.91 (Examples of free adverbs: 6D.3.11 are on the CD-ROM provided.) 6C.3.12 The preverbal adverb, which not only itself is bound but also marks the clause in which it operates as a bound (conditional) clause, is so called because it takes pre-verbal position preceding or following N t in the clause: in general, a noun preceding preverbal adverb is given to the context of the following clause, one following preverbal adverb is not. This class has the possibility of (extended) colligation with the final adverb. 6C.3.13 The final adverb, likewise both bound and "binding", always has absolute final position in the clause. 89

MEDIEVAL CHINESE GRAMMAR Both the preverbal and the final adverb, and the combination of the two, are normally translated by a subordinating conjunction. The preverbal and final adverbs of regular occurrence in the "Secret History", with specimen translation, are as follows: Preverbal adverb

Final adverb (s) with which collocation is regular

yin wuei ruo suei3 gi2

Translation

because, since (it was) because (. . . that) if ("if and when"), if anyone although when, after, since earlier when if (. . .) any, whenever, whoever, whatever every (time, -where that, one who, etc.) (as final adverb) after that which, those who just when

CU4

dan, danxeq, danfan fan ki3 z, suo xai

Final adverb (without pre-verbal adverb) si2, si2gian2, disi2fun gian2, Jurjgian2 heu, diheuteu2 qa sarjteu2 cu2 Je

when while after if because where whoever, anyone who; that which

Some simple preverbal adverbs have alternative compound forms with si as second syllable (ruosi, fansi, etc.); the translation is unchanged. Sometimes this si could alternatively be regarded as a prepositive or auxiliary verb. Some others have yieu as second character, particularly in collocation with the final adverb Je (e.g. ruoyieu wueije "if anyone disobeys"). It is perhaps in the classes of preverbal and final adverb that there is the widest lexical divergence between the "Secret History" and Modern Pekingese. As examples: from those listed above, (i) for both yin and wuei (preverbal) Modern Pekingese has yinwuai, which occurs only once in 90

"SECRET HISTORY OF THE MONGOLS" the "Secret History"; (ii) dan with its compounds (preverbal) is absent from, and has no equivalent in, Modern Pekingese;92 (iii) for si2, si2gian2, etc. (final), Modern Pekingese has equivalent dishau (disi2 heu2), which is not found in the "Secret History"; (iv) Modern Pekingese has no final adverb corresponding to rja (dihua, which can often be translated "if", is more restricted than either the rja of the "Secret History" or the Modern Pekingese preverbal adverbs regularly translated as "if");93 (v) there is no final adverb in Modern Pekingese corresponding to sarjteu2;94 (vi) ki3 does not enter into the Modern Pekingese system.95 (Examples of preverbal and final adverb: 6D.3.12-13 are on the CD-ROM provided.) 6C.3.14 The 'conjunctive adverb occupies pre-verbal position in the clause, always following the pre-verbal noun where one is present. Itself a bound form, it is not "binding': the clause in which it operates is free. It is distinct from the verbal adverb, with which it normally has identical position, in that (i) it regularly occurs in a clause following a conditional clause, where it marks the free form in the sentence structure of bound-free; (ii) it collocates in this relationship with certain preverbal and final adverbs; and (iii) it normally indicates a related or same context with the previous clause even where the latter is not bound. In a clause following a conditional clause the conjunctive adverb is normally untranslated in English; and its translation frequently takes the form of the subordination of the previous clause even when this is free; in general the conjunctive adverb is specifically rendered in English only when the previous clause is not subordinated. Those of most frequent occurrence, with specimen translation, are: bian, zieu, suei and yiu2si suoyi2 kiuo yie and yieu3 si2 heu

"then" "so" "but" "even then, also" "at that time" "afterwards"

Of these, yiu2si and suoyi2 are also clausal adverbs, yie and yieu verbal adverbs and si2 and heu final adverbs. This class constitutes a distinctive feature of Modern Pekingese, which has many of the same exponents. Of those above, suei is missing and bian restricted to writing; kiuo is normally replaced by dans (dansi) or dau2; si and heu are entirely absent;96 while one of the most characteristic conjunctive adverbs of Modern Pekingese, xai (written xai in the "Secret History", more frequently now xai2), which operates in a special

91

MEDIEVAL CHINESE GRAMMAR two-term sub-system in opposition to jiau (zieu), has as equivalent the comparatively infrequent (seven occurrences) far).97 A special instance of the conjunctive adverb is its occurrence in two adjacent clauses which are thereby related to one another, and not, as normally, to the preceding clause. Six pairs occur in the "Secret History", all infrequently: yie .. . yie, yieu3 .. . yieu3, and yie . . . du huo . . . huo yibian2 . . .yibian2 siuan . . . siuan

"both . . . and" "either . . . or" "and . . . at the same time" "... and . . . in turns".98

Finally the conjunctive adverb in the "Secret History" is particularly frequent in occurrence at the resumption of the narrative following a passage of direct speech. Those found regularly in such instances are bian, zieu, suei,yiu2si and suoyi2. (Examples of the conjunctive adverb: 6D.3.14 are on the CD-ROM provided.) 6C.3.15 The particles, being purely grammatical, have no separate, lexical translation equivalents; they are rendered in English through the translation of the clause, verb or noun which they modify. (i) Clausal particles have been discussed above in relation to the classes of clause in which they operate. Strictly speaking, if the particle is characterized as modifying that which it follows or to which it is final, the aspect particles non-final liau (perfective) and jo (imperfective) should be considered verbal particles; they are, however, regarded here as clausal particles because of their function as marking certain classes of the clause. (ii) The single verbal particle is the rare sie (four occurrences). The English equivalent is either "a little" or the comparative form of the adjective, sie occurs likewise in Modern Pekingese, where, however, yicUan is a more usual equivalent; yidian (yidiam) does not occur in this use in the "Secret History". (iii) Of the nominal particles, di (and its rare variant ji3) occurs between two free nouns, or a pronoun and free noun, as a formal mark of the attribution of the first to the second; the use is thus parallel with that of di as clausal particle in genitival clauses of type (i). This use is identical with that of di in Modern Pekingese, and is reflected in English translation with possessive form or "of", mui is a plural particle and is restricted with one exception99 to nouns of human reference: kinship terms (54 occurrences), personal (including tribal, etc.), names (21) and other words (33). 92

"SECRET HISTORY OF THE MONGOLS" In its occurrence with pronouns it is considered as final syllable in a compound pronoun. Modern Pekingese has as equivalent the chuan-yinwo man (mun), somewhat more restricted in use.101 derj, which is also a nominal adverb (62 occurrences; see examples 6D.3.11.3), occurs 51 times, usually following personal names and normally translated "and (the) others"; sometimes, especially following the last in a list of names, it merely marks the plurality and requires no translation. It is much more frequent in the later than in the earlier chapters. In Modern Pekingese newspaper style it is regular in both uses, while other forms of the language have it more usually doubled: dgrjdsrj "and so on". (Examples of particles: 6D.3.15.) As a conclusion to this section dealing with the translation of grammatical categories, it may be useful to consider three groupings of words which, while they do not figure as categories in the grammatical system of the "Secret History", may conveniently be brought into relation for translation reference: negative words, interrogative words and pronominal words. These are exemplified in 6D.3.16—18. 6C.3.16 (i) Negative words. The verbal adverbs bu, buxerj, wuei2 and wuei2xerj (occurrences 288, 59, 17 and 1 respectively) required a negative verb form in English, bu is the regular negative, while buxerj is the negative term in opposition to xerj (47 occurrences102) with past time reference. wuei2 occurs six times in conditional clause with final adverb si2 "when" (5) or Ji3xian "before" (1); here and in at least three other instances it could be regarded as a preverbal adverb ("when . . . not" = "before");103 of the remaining four occurrences, two are in Chapter 12, where the use of wuei2 where buxerj would be expected, may be regarded as a feature of literary language, while the other two are preceding the verb ji7 "know".104 In Modern Pekingese msifyiau) is the regular equivalent of buxerj and wuei2; this form (i.e. mui2) never occurs in the "Secret History". wu occurs as pro-verb and prepositive and postpositive verb, always in opposition to yieu (for translation see 6C.3.2,4,5); Modern Pekingese has likewise mai (yisu). wuyieu occurs twice as pro-verb, variant of wu. The verbal adverb hieu (35 occurrences, with frequent collocation with the prepositive verb giau) is the normal prohibitive form "do not"; mo occurs once (V. 51a; 169.8) with this translation, before suo "say" in an active clause with structure NNV.105 Modern Pekingese has bu yiau or bis here. 93

MEDIEVAL CHINESE GRAMMAR The compound clausal adverbs mosi, mobu, mobusi and mogam (occurrences 1, 4, 8, 1), together with mo as verbal adverb (1), may be translated identically as "must" (= 'it must be that'), or (as frequently) in interrogative clauses as "surely . . . cannot (= it cannot be that . . _)?";106'107 compare Modern Pekingese nandau3. bu and mo occur identically in the clausal adverbs bufu, buruo and mofuo which may be rendered "it would be better if"; Modern Pekingese bum. feu occurs once (XL 17b; 252.17) as final interrogative particle, collocating with xerj in the clause, where Modern Pekingese would have maiyiau (" (did . . .) or not?"). 6C.3.17 (ii) Interrogative words. These include the two free nouns suei "who?" and simma "what?" (occurrences 24, 8), identical in Modern Pekingese, together with the numeral noun gi (9) which sometimes = "how many?" and "which?", and some clausal, verbal and nominal adverbs. fuho "how?, why?" (64) and hogu "why?" (3) are clausal adverbs; Modern Pekingese has wu9is9(n)ma (wueisimma) and Z9(n)ma (zemma), of which in the "Secret History" the latter never occurs, while the former occurs once, but, like wueisim, not before preverbal noun. Verbal adverbs are zemserj (9) and zemserjban (3) "how?, why?";108 wueisim (6), wueisimma (1) and ho (1) "why?";hocu2 (2) "where?", and ki4 (6) "how (could (it) possibly (be that))?". Modern Pekingese has Z9(n)ma(yiarj) "how?, why?"; wu9is9(n)ma "why?"; nali "where?"; and ci (ki4) with similar use but more restricted. Nominal adverbs are simma (13), sim (7) and ho (8) "what?, which?"; and duosau (3) "how many?". Modern Pekingese has sb(n)ma (simma) and dussau (duosau), together with the determinative noun na "which?" (shang-sheng (51) tone; na in the "Secret History" would never be read in this tone). The "Secret History" thus presents a series of negative and interrogative words (including interrogative particles) in which the labial nasal (m) operates as either initial or final. The forms occurring may be summarized here: Form sim simma zem (set)) ma madau3

Word class Adverb (nominal) Noun (free); Adverb (nominal) Adverb (verbal) Particle

94

Negat./Interrogat. Interrogative

"SECRET HISTORY OF THE MONGOLS" Form mo mo (-si, etc.)

Word class Adverb (verbal) Adverb (clausal)

wuei2 wu

Adverb (verbal) Pro-verb; Verb (pre-positive)

Negat. /Interrogat. Negative Negative (freq. in interrog. clause) Negative Negative

Of these, negative adverb mo (and in mosi, etc.) has final -k in the Middle Chinese system, while wuei2 and wu have initial m-.109 6C.3.18 (iii) Pronominal words. The pronoun system proper of the "Secret History" has been regarded as made up of the 12 personal pronouns in two parallel series of six, plus the irregular ru3, bi and Ji3. A small number of other words of infrequent occurrence could be included in the pronoun system, on the grounds that they stand as exponent of value N but do not operate in the noun group: z2 and Z2gi4 (occurrences 6, 2) "self", and the forms x (1) "this" and je 2 and na "this", "that" (pronominal occurrences 2, 1 only). In Modern Pekingese ja (Je2) and na, and zji (z3gi4) especially in combination with a personal pronoun, regularly have value N.110 suei "who?" and simma "what?" might also be considered as pronouns. There is no system of negative or indefinite pronominal or attributive forms; but the use of interrogative words in repetition or in combination with negative or other verbal adverbs (du, yia), characteristic of Modern Pekingese as a regular equivalent of English forms such as "anyone", "no one", "whoever", is almost unknown to the "Secret History". The latter has instead certain terms in the conditional clause system of bound adverbs, occasionally in combination with interrogative words but more often without any form that could be regarded as pronominal.111 Please refer to the CD-Rom provided for Section 6D. Notes for section 6: C 1. Or with any supposedly "universal" category. The possibility of the comparative identification or even the universality of at least some categories is not to be excluded, but the criteria need to be fully defined. Cf. Robins (1952). 2. It may be more difficult to apply to Chinese than to a highly inflected language an analysis that is part logical, part grammatical. There is, for example, no category of "nominative case" to be linked with that of "subject" into a complex grammatical-logical category. A positional analysis could yield a category Nt which could be termed "subject",

95

MEDIEVAL CHINESE GRAMMAR

3. 4.

5.

6. 7.

8. 9.

10.

11. 12.

defined as "the first pre-verbal free noun (or noun group) not forming part of a complex group, or the first pre-verbal pronoun, in a clause"; but if the subject, in this sense, is to be more than a purely grammatical (positional) category, such that the statement "the subject of a clause takes position Nj in the structure" is not a tautology, we require a (universally valid) prepositional logic, not based on any one language, which would account for the fact that, for example, Modern Pekingese na s wu9 sia di is often the contextual equivalent of English "I (7) wrote that". As used by Hjelmslev. See e.g. Martinet (1942-45:33). As proposed by Allen (1953: especially pp. 99—100). The contextual identification of (terms in) grammatical systems may be one way of establishing relationships which, for Allen, is a function of systems (not of languages). The question of the identification of terms in systems with different numbers of terms must depend on the level at which identification is made. If identification is made, as here, by contextual reference, such identification seems permissible, since the two terms can be identified by what is (contextually) excluded as well as by what is included. Until contextual identification is established, however, such identification is much more difficult; grammatical identification would seem to require the exclusion of a like number of terms. If, for example, the Modern Pekingese form (verb plus) gu9 is included as a fourth term in the aspect system, can we still identify the liau of the "Secret History" with the liau of Modern Pekingese? For example English "you" is not identifiable with Modern Pekingese either ni or nimsn. Explanation of reference: 1 = speaker; 11 = speakers (more than one); 2 = addressee; 22 = addressees; 3 = other person; 33 = other persons. The combinations given are those occurring in the "Secret History", though these in fact include most of those possible. 2233, for instance = "more than one addressee plus more than one other person". For the transcription of Modern Pekingese, see Appendix A. Case distinctions in the English personal pronoun system are ignored. For contextual reference, "I/me" forms a single term. Since English is not the language under description, the complex pronouns recognized by some grammarians, such as "you and me", are also ignored. za occurs once as 133, in (V. 35a; 163.3), zamui once as 11, in (V. 16b; 162.20). gam and wuomui cannot be distinguished by contextual reference, gam is more frequent (39 occurrences, wuomui 8); rjammui occurs once only (XII. 31a; 275.25). Including 123,1233, etc. rjam occurs twice and wuomui once where reference includes 2 (22), in (I.21b;35.6), (III.40a; 121.5) and (V.8a; 149.29). za, zamui occur 42 and 18 times respectively; there is a tendency for more than one addressee or other

96

"SECRET HISTORY OF THE MONGOLS'

13.

14. 15.

16.

17.

18.

person (22 or 33) to be included in the reference of zamui (13 out of 18 occurrences), while za is frequently 12 only (25 out of 42 occurrences), za and zamui are also used in the narrative (with the author as speaker, the reader as addressee): (III. 16b (116b); 110.4), (Vl.lOb; 171.19), (XII.7a; 265.52) and (III. 33a; 119.9). According to Lii Shu-hsiang, (1940b), nin (nini) is never singular in the "Secret History", nim does however occur in a few doubtful cases where there seems to be only one addressee: (V. 45a, b; 167.29, 32), (VII. 12b; 189.14), (VII. 40a, 41a; 195.5,59) and (X. 45a; 246.16). nin is excluded from the Modern Pekingese system as being rather narrowly restricted in use. ni occurs sometimes for 22 = two addressees only, usually (but not always) followed by liarjgo 'two' or iyrin 'two people', e.g. in (II. 36a; 93.18), (III. 23a; 113.5) (liarjgo), (IX. 26b;~219.22), (XI. 34a; 255.18,21) (r2rm); also in (IX. 7b; 212.2), (IX. 26b; 219.3), (ni fu3z (mui) 'you (two) father and son'), and in (IV. 21a; 136.21), (VI. 38a; 179.29) with only contextual indication that two addressees are referred to. nimui, nim and nimmui are contextually identical (except for doubtful occurrences of nim, above), occurring 14, 53 and 5 times respectively. ru3 occurs once (XI. 17b; 22.21) as 2, actually singling out one of three possible addressees. All pronominal reference in this system in the "Secret History" is to persons. In Modern Pekingese pronominal reference to inanimate objects is infrequent. ta is used sometimes for 33 = two persons only, e.g. in (VI. 37b; 179.19), (VI. 43a; 181.17) (ta liarjgo), (VII. 4b; 187.13) (ta f2rin). bi occurs once (IV. 44b; 145.43) as 33 with contrasting reference to that of tamui in the same sentence. ji3 occurs nine times, always postverbally, with reference (3 or 33) to a noun in the previous clause of the same sentence or in the previous sentence: (IV. 48b (198b); 146.21), (VII. 20a; 191.13), (XII. 25a; 272.8,12,13 (?), 15,20,21), (XII. 26b; 273.5). Wang Ching-Ju, (1935) gives the following comparison of personal pronouns of the "Secret History" and Modern Pekingese: Secret History wuo za gam ni nim ta tamui

Modern Pekingese wu9 zamsn wu9m9n ni ninign ta tamsn

regarding tamui as the only pronoun regularly found with -mui, thus ignoring wuomui, nimui and nimmui, and suggesting that the -mui in zamui and rjammui was added later. But there are only 12 occurrences 97

MEDIEVAL CHINESE GRAMMAR of tamui, with 389 of ta; zamui, nimui occur 18 and 14 times respectively. (See tables in Appendix C.) Wang's reason for regarding tamui only as regular is no doubt that no mono-syllabic third person plural (33) form corresponding to gam, nim exists; 33 reference is rare altogether (in fact both ta and tamui are almost exclusively confined to the passages of dialogue, rarely occurring in the narrative). Lii Shu-hsiang (1940a and 1940b) agrees in regarding tamui as the primary pronominal use of —mui, extended by analogy to ni and wuo. Ota (1953) gives the complete list of personal pronouns from the "Secret History" and contrasts the total of 12 with 37 found in the Yuan drama and only eight (wuo za ni ta, each with plural -mun) in Rokittai (Lao-ch'i-ta (38)), a Chinese textbook published in Korea in 1423-34. Of historical interest is the alternance in the "Secret History" system of syllabic and non-syllabic -m (—) in plural forms (gam, nim are both -m in Chung-yuan yin-yiin (37), 1324), the syllabic forms alone remaining (as in Modern Pekingese) when -m disappears from the final nasal system, though nin, rjan remain dialectally with -n, no longer plural. Possibly syllabic -m— is an archaism (cf. Demieville (1951)) even though the origin of the non-syllabic -m may itself be syllabic (mui or some other syllable). Compare the alternance in interrogative words in the "Secret History": see 6C.3.16. Lii Shu-hsiang notes that gam is often exclusive; it is in fact (in spite of two definitely inclusive occurrences: see above, n. 6) opposed to za in the system; contrast Modern Pekingese where zamsn is inclusive, wu9m9n neutral. On the origin of za, Lii suggests it is derived from z2gia; for the relation between pronoun za and (imperative) particle za in the "Secret History" see 6C.2.6.2. What appears synchronically as the plurality of forms in the personal pronoun system of the "Secret History" having the same contextual reference suggests diachronically a 'fluid' state of the system. It may possibly be helpful to recognize 'transitional systems' in historical studies, regarding a particular state of affairs as (relatively) unstable, a sort of intermediate stage in the replacement of one system by another. 19. Further reduction is possible if two or more exponents of a term in one system are identified with the exponent of an exactly coextensive term in the other: nim, nimui and nimmui will then be identified with nimS, giving a further reduction of three. If on the same basis za and zamui are identified with zamsn, even though wu9m9n includes all references of zaman, there is a further reduction of two terms. The final degree of relationship would then be high. If this is to be done, however, it would be preferable that two or more terms with identical reference in the same system should be identified as one; nim, nimui and nimmui in the "Secret History" would then be considered as one term, and we should have two seven-term systems reduced in the comparative statement to one of eight terms: degree of relationship 4/7. Otherwise (if nim, nimui and nimmui

98

"SECRET HISTORY OF THE MONGOLS'

20.

21.

22.

23.

24. 25. 26.

27.

are not first reduced to one term) the method becomes such that a degree of relationship higher than V-i is obtainable and the comparison is meaningless. I should not invoke the fact that the Chinese version of the "Secret History" is itself a translated text as the sole justification for this procedure; nevertheless this fact does offer interesting considerations. On the use of criteria of translation in grammatical description cf. Robins (1952: 296-97). Of the translation of the basic categories of noun and verb Robins says: "There is the additional fact that, when two formally differentiated word-classes are established in any language as the basis of its grammatical system, a large proportion, at least, of the words in these two classes can be translated into the noun and verb, respectively, or nominal and verbal phrases, of the analyst's language" (p. 296). "Context of mention" could perhaps be applied also in descriptive comparison. With spoken texts it would probably reveal some correlation between positional structures in Modern Pekingese and intonational structures in English. Thus, whereas in the procedures of collocation and "colligation" the forms brought into relation one with another operate at the same descriptive level, in the use of "context of mention" grammatical forms are brought into relation with lexical forms. In an English translation which I have made of Chapters I—VI of the Chinese version almost all the sentences are coextensive with those of the original. The type is sufficiently rare in the "Secret History" not to be taken into account in the delimitation of the sentence. Whereas the absence of an exponent of the given frequently implies (not related contexts but) one and the same context. "Followed" is to be understood throughout as "followed in the same unit next above", that is, a clause followed by a clause in the same sentence, or a word followed by a word in the same clause. It is because of this "inversion", that is change of position without change of contextual reference, that I have given the name "voice" to this system. The term "passive" is generally associated with an inversion (relative to the "active") of the nominal terms (if active has N(a)N(b) then passive has N"(b)N(a)); mis is usually coupled with a change in the form of the verb, but in Chinese the verb form is identical and the formal changes, even those other than of position, take place in the clause. The term "ergative" has been borrowed from the description of those Caucasian languages (and some so-called "Palaeoasiatic" languages such as Chukcha) (see e.g. Matthews (1952—3: 403—5), where it is the name given to a structure of a transitive sentence with two nominal terms of which N2 (the "object"), and not N l9 takes the case-form of the nominal term of an intransitive sentence (compare the pre-verbal position of N2 in the Chinese

99

MEDIEVAL CHINESE GRAMMAR

28.

29.

30.

31. 32. 33.

34.

ergative). The term "transitive" might be used instead of "ergative" (cf. Chao's use of "pretransitive" for ba in this structure in Modern Pekingese), but I have wished to retain the term "transitive" for a type of free verb in opposition to "intransitive". Wang Li (1946-7: 164), calls this structure ch'u-chih-shih (39) which he translates "execution form". (Since the above was written Frei has come independently to the use of the term "ergative" for this structure; see Frei, 'The Ergative construction in Chinese' (1956-57)). The distribution in passive clauses is very noticeable in Modern Pekingese. In the "Secret History" the passive is, as already noted, very rare; but of the five occurrences, four demonstrate this contextual correlation and in the fifth N2(a), although given in the preceding context, is clearly marked by the structure as new in relation to the context in which it occurs. See examples 6D.2.4 on the CD-ROM provided. This point is illustrated in the examples 6D.2.4 on the CD-ROM provided. The English marked (stressed) falling intonation is often associated with the category of the new. In fact, probably because of its stress correlation, contextual reference to given/new is often termed "emphasis"; cf. with regard to such reference in the postpositive complex group, 6C.3.5.3.1n. In fact in the "Secret History" this structure might be regarded as a subsidiary form of the passive, admitting the criterion of position alone. The difficulty is that it would then be formally indistinguishable from the (much less frequent) variant of ergative. E.g. (II 42a; 97.7) wuo yiuliau ni yi go diau3su, guo3rfu6 yieulai "I gave you a sable swaddling cloth". Cf. for example (II 22b; 84.9) rugim Sorqan-sira ziarj wuo gian liau "now Sorqan-sira (given) has seen (new) me". Strictly speaking, "there exists also in the text a free verb ziarj (etc.) lexically identified with the prepositive verb ziarj marking ergative clause". The simpler form of expression has been preferred throughout. It should preferably not be interpreted as "the word ziarj functions also as . . ."; indeed "the word ziarj" as an entity is indescribable. The term "aspect" has been current for some time in the description of New Chinese, and much has been written on the subject of aspect in the Chinese verb. Mention may be made of some of the works dealing with the subject: (i) Maspero (1937) considers that the Chinese verb has aspect (and not tense); the usual aspects are: determine (with lai, ciu); accompi (with liau); duratif (with j9 (jo)), and aoristique (with gua). Aspect may be marked by either particles or auxiliaries, the latter retaining some of their lexical meaning. (ii) Frei (1941) classifies aspect into perfective (with liau and other forms, my "postpositive verbs") and imperfective (with ni2,laiJ9 and j9 (jo) (as opposed to jau perfective)); in addition certain adverbs are used 100

"SECRET HISTORY OF THE MONGOLS" exclusively with verbs of one or the other aspect. While Maspero considers aspects as undetermined and expressing shades of verbal meaning, Frei regards them as determined, though aspective particles may sometimes perform modal function. (iii) Prusek (1950), who in this article is discussing the language of the colloquial novel Lao Ts'an Yu Chi (40) of 1907, considers aspects to be a special instance of the Chinese tendency of "elargissement des mots" by which the second element of a compound verb, forming a bridge between the general sense of the verb and its concrete application, expresses the outcome of an action and becomes aspective: this applies both to elements retaining a lexical meaning (my "postpositive verb") and to those becoming purely grammatical. Syntactically the function of aspect is to subordinate one clause to the next (cf. my "related context"); j9 (Jo) is especially common in this function, from which is derived its use with prepositions (cf. my "prepositve verb plus jo" 6C.3.4.1). Further, the aspective particle is much more frequent with simple than with compound verbs (cf. my liau, jo with monosyllabic verbs, below). (iv) Wang Li (Wang Liao-i) (1946-7,1947), drawing his examples from the eighteenth-century colloquial novel Hung Lou Meng (41) ("The Dream of Red Chamber"), regards aspect as a category of the clause ("aspect in Chinese cannot be considered as directed to the verb or the narrative; it can only be considered as directed to the whole narrative predicate, since the aspective element does not necessarily follow directly after the narrative") (p. 1.297). From aspect, which is concerned with time reference other than simple past, present or future, should be distinguished the "causative form" (shih-ch 'eng-shih (42) my "free verb plus postpositive verb") which, while historically the origin of aspect, indicates the result, extent or condition of an action (though also indicating perfective aspect, e.g. gua). Seven aspects may be distinguished, each with clearly defined time reference. (v) Lii Shu-hsiang, (1947), and Kao Ming-k'ai (1948). Lii and Kao both recognize aspect, distinguishing various aspective categories; Wang's "causative form" is considered by Kao to be "resultative aspect". 35. But note that final jo occurs in the text as a variant of je in imperative clauses (always sentence-final). 36. See 6C.2.5.2. There is actually one such occurrence in the "Secret History", in (VII. 26a; 193.35); here in the "Secret History" only non-final liau would be expected, though both would be expected in a Modern Pekingese rendering. 37. Contrast nan, as free verb = "difficult, unpleasant", with the formal identity of the first syllable of madau3 and interrogative particle ma. One could translate analytically nandau3 = "unsayable" (cf. "unthinkable" in this use, and "unspeakable" with different use), madau3 "what (do you) say?". For 101

MEDIEVAL CHINESE GRAMMAR

38.

39.

40.

41.

42.

ma = "what?", cf. 6C.3.17. Haenisch points out that in the Gloss madau3is regularly the equivalent of Mongolian kernen, kemeku marking the closing of direct speech, and is thus quite "un-Chinese". It seems, however, that the use of madau3 in the Chinese version, which is much less frequent, should be clearly distinguished as operating within the Chinese interrogative system. Moreover in (II. 2a; 70.9) and (II. 3a; 71.4) madau3 occurs in the Chinese version but not in the Gloss. References to Haenisch (1931) throughout this section are to Section VIIIB (pp. 72—97). The two never occur in the same clause. Personal pronoun za (reference 12, etc.: see above, 6C.1.5.1) is rarely post-verbal. In (V. 35a; 163.3) which is translated as (unmarked, i.e. second person) imperative in English, za occurs in clause-final position apparently as pronoun with reference 133, value N2; this might be regarded as an instance of imperative za; it seems in fact to be intermediate between the two regular forms, the reference being the reverse of that which is normal in either. There is one instance of za following first person wuo (IX. 22b; 218.8). Plus a small number of formally ambiguous instances where jo is both final and directly post-verbal; these have been assigned on statistical grounds (of frequency) to imperfective aspect, though some (e.g. (XII. 25b; 272.32), where there is collocation with giau "tell, order") require imperative in translation. Once where a first person imperative seems required and four times in exhortations which could be rendered in the second person. Exhortations (as here to heaven and to the emperor) may well, however, be considered as, and translated in, third person; and in the former instance (VIII. 23b (379b); 201.35) yieje follows wuo s3 heu "let me, after I die", which could well be rendered as "let my spirit . . .". Haenisch points out the imperative use of je in the "Secret History", regarding this, together with the "voluntative" yieje, as a Mongolism; he compares Mongolian verb-termination ja ("voluntative"). There is no exact correspondence between the use of je and yieje in the Gloss and in the Chinese, yieje for example being rarer in the latter but occurring nevertheless in places (for example (I. 23b; 21.13) and (XII. 31a; 275.30)) where it is absent from the Gloss. The occurrence of je in imperial orders where the Mongolian has (-tuqai keyen) jarliq bolba is indeed frequent (see Chapter 9 of the "Secret History", passim), but it is not confined to such instances, jo, za as imperative particles are not mentioned by Haenisch, who regards za as a personal pronoun derived by ablaut from z2. In fact this only occurs once, in a paragraph-final clause (XII. 26b; 274.21) where the preverbal adverb is yi2 a form taken from literary Chinese and confined to Chapter 12. The conjunctive adverb does not, however, mark the previous clause as bound, since (i) this would rest on a criterion of exterior, not interior, form, which marks functions, not classes, and (ii) clauses with a conjunctive

102

"SECRET HISTORY OF THE MONGOLS"

43. 44.

45.

46. 47.

48.

49.

50.

51.

adverb are sometimes marked as sentence-initial (so that even on this criterion of exterior form a clause preceding another containing, for example, the conjunctive adverb bian would not always be regarded as having "subordinate function"). But commonly in Modern Pekingese. See H.F. Simon (1953: 340). (Cf. example 6D.2.8.1, no. 4 on the CD-ROM provided.) As always where a form is defined as preceding another form with given value in the structure, the latter admits direct attribution, the attributive element thus intervening. Further, between the genitival clause and the noun there may intervene (i) another genitival clause, as in (II. 36a; 93.13) or (ii) a noun (N) itself followed by genitival particle (nominal) di, as in (II. 5a; 73.3). E.g. (III. 44b; 123.16) wu rin yian2di2mian "an uninhabited place". Attribution does not figure in the description as a function of clauses, since these recognize a bound class and attribution would be definable simply as precedence within the sentence. Note however the single instance (IV. 44b; 145.52) where, if it were not unique, Je2 might be regarded as a formal mark of attribution (not, however, of a bound clause, since, like the conjunctive adverbs, it is outside the clause to be considered "bound" by it); compare the frequent alternance in Modern Pekingese of (nominal) di with j92/na ("this", "that") (with or without the auxiliary noun) especially following a pronoun. Three in the more literary Chapter 12. The four are: (IV. 29b; 140.28), (XII. 28a; 274.1), (XII. 50b-51a; 279.6,25). For example (VI. lla; 171.36) "Quyildar, who was injured", or "the injured Quyildar". This is unexpected from the point of view of Modern Pekingese (cf. Chao (1948:57)). Genitival clauses of this type, like attributes in general, are normally new, but if given may be non-restrictive (Quylidar's injury is mentioned in 171.19). Two instances only: (IX. 18a; 214.19) and (XL 27b; 254.17). There are likewise only three instances of this type of clause in sentence-final position: one is (IX. 18a; 214.19), following si; the others, not following si, are (III. 32a; 118.12) and (XI. 34b; 255.58). Related likewise in Old Chinese, which has passive (N)wueiNsuoV (cf. (N) siNVdi), "genitival (ii)" (suo)Vje. The former occurs once in the "Secret History", but in Chapter 12 (XII. 6a; 265.10). These may precede a noun, in which case they resemble type (i). They are considered conditional since the preverbal adverb may be one of a number, while the genitival clause can thus be restricted to di alone; but an alternative statement would recognize these as genitival, or even consider the category of genitival clause as a type of the conditional. Such as farj2yiarj4di "shepherd". These could be considered as genitival clauses; that they should be considered rather as compound free nouns is suggested by their possibility of combination with nominal (plural) particle

103

MEDIEVAL CHINESE GRAMMAR

52.

53.

54.

55.

56. 57.

58.

59.

60.

mui. The assignment of some such elements must nevertheless, for lack of syntagmatic evidence, remain arbitrary. A transitive verb may occur in active clause without a following noun. Such instances, actually relatively infrequent in the "Secret History", do not of course affect its place in the paradigm. In Modern Pekingese this type admits a post-verbal "minor noun group" (Chao's "cognate or quantified object") but there are no instances of this in the "Secret History". There is, however, one instance (1.47a; 66.5) of an intransitive verb followed by a noun followed by a minor noun group, where the intransitive verb da must be considered to have substantive function: da Temujin yisuei2 "(she was) a year older than Temujin" (alternatively this could be analysed as da prepositive verb (only instance), yisuei2 (as regularly) pro-verb). The categories of "transitive" and "intransitive" are not coextensive with Chao's (1948: 48) for Modern Pekingese, since his "action verb intransitive" (e.g. lai "come"), which does not have attributive function in the "Secret History", is classified here as transitive. Such a verb combines with the following noun to form a clause (for example in (VI. 17b; 174.1), lai Cirjgis cu2 "came to Genghis"). Haenisch regards this, together with the use of yieu as postpositive verb, as a Mongolism: see 6C.3.5.5n. But the clause-final position of yieu is not confined to the "Secret History"; cf. Ota, (1953:11). See example 6D.3.2, no. 3 on the CD-ROM provided. That is, with no intervening N. The auxiliary verb may be separated from the free verb by an adverb. See especially examples 6D.3.3, nos. 1—3 on the CD-ROM provided. As in Modern Pekingese, lai and kiu as auxiliary verbs often require in English translation the infinitive form of the verb, sometimes a more specific form such as "so as to", "in order to". si as auxiliary verb does not normally figure in translation. It might perhaps best be regarded as a marked positive form, excluding the negative and contrasting with the unmarked positive, e.g. (V. 51.a-b; 169.24) kiaxai ti2sim2 ni suo di hua si si5 liau "I have just been able to check that what you said was true". See further example 6D.3.3, no. 4 on the CD-ROM provided. It would be possible to regard a prepositive verb of this type as combining with preceding (pre-verbal) N to have adverbial value in the structure (or, where there is no pre-verbal N, as having this value alone): (N)v = A. The position would then be that of clausal adverb, as distinct from the true complex group which normally has the position of verbal adverb. As an alternative form of statement, prepositive verbs of type (ii) could be regarded as free verbs with either the setting up of a separate clause or the recognition of a type of verbal clause containing two free verbs in substantive function. The objection to the former is that there is

104

"SECRET HISTORY OF THE MONGOLS'

61. 62. 63. 64. 65. 66.

67.

68. 69.

70.

never a corresponding division of pieces; and the present form of statement has been preferred to either alternative because those verbs which operate in the quasi-complex group do constitute a finite word class, some of whose members, moreover, are not lexically identifiable with free verbs. See especially examples 6D.3.4, nos. 7—8 on the CD-ROM provided. This use is reminiscent of that of yi9u (yieu2) in modern newspaper style. Cf. above on aspect, 6C.2.5.1. n. "Causative form" is Wang Li, (1947:1,153 (shih-ch'eng-shih (42)). That is, of the free verb lexically identified with the postpostive verb. de might be regarded as an adverb here. It is, however, a regular postpositive verb of this type (see 6C.3.5.1.3). Wang Li (1946-7) distinguishes this type (iii) into (1) with ds (de) or di, a type of the continued form (which includes my quasi-complex group) and (2) with da only, a contracted form of the resultative form (or, from the opposite standpoint, a dilated form of the causative form (as in my type (ii) here)), on the grounds that the former is followed by a ."descriptive predicate", the latter by a "narrative predicate": examples (from Hung Lou Meng) (I) wu9 laida bu ciau liau "I have come at an awkward time" and (2) suoda Lin Daiyiu "pue" di yi ssrj (sirj2) siau, liau "what he said made Lin Tai-yii burst out laughing" (Wang Li 1946—7: vol. 1,190—205). I have not regarded de and di as being non-commutable variants in the "Secret History", but in fact the "Secret History" has de in clauses which on Wang Li's criteria would be classified as (1). While I have not distinguished clauses into classes corresponding to "descriptive" and "narrative", it may be noted that a clause following a bound (postpositive de) clause may be nominal: (. . .serjde) yieu yian3se "pretty" (1.34b; 54.6). Fx only with lai or kiu (postpositive verbs type (ii a)) as clause-final. This could be regarded (since group z here = lai, kiu) as extended (split) Fxz; but it seems preferable to regard clause-final postpositive verbs as a separate type. The unique example of postpositive suolai (1.39b; 57.2) should be analysed as postpositive verb (type (i a)) suo plus (type (ii b)) lai. It is sometimes doubtful whether to assign lai and kiu in a given syntagm to the first of two clauses as postpositive verb or as auxiliary verb to the free verb in the second, e.g. (V. 16b; 162.7) xerj bei Merki Iu3kiu/curj2 duei2 or xerj bei Merki Iu3/kiu curj2duei2 "he had been taken captive by the Merkit and made to pound a quern" (the distinction could be brought out in translation: "had been taken off. . . and had pounded" or "had been taken . . .to pound"). Prusek (1950: 423) notes instances of postpositive ziarjlai ("getting hold (of something) to . . ." = "preparing to . . ."), thus regarding ziarj as comparable with ziarj (lai) in its "future" sense (and compare lai as auxiliary verb). But note hirjziarj lai paragraph-final in (1.4a; 5.6).

105

MEDIEVAL CHINESE GRAMMAR 71. I.e. in the "Secret History" zin2 is the opposite of huei; in Modern Pekingese it is the opposite of cu. 72. The Modern Pekingese specialized combinations cilai (kilai) "start, become" and siaciu (hiakiu) "go on (becoming)" are not found in the "Secret History". 73. Except yieu and wu which are assigned to type (ii), but with one use resembling those of type (i): see prepositive verbs 6C.3.4.3. 74. But wuei as prepositive verb would (except in one instance (XII. 6a; 265.7)) be read by a Modern Pekingese speaker in fourth (ch'u-sheng) (43) tone; eight of these occurrences are in wueisim "what for?, why?". As postpositive verb and pro-verb it would be read in second (yang-p 'ing) (44) tone, translated as (pro-verb) "is", etc., (postpositive verb) e.g. zuowuei "act as". 75. Excluding (15) occurrences as postpositive verb of type (ii b). 76. Contextual correlation of this type is often referred to as "emphasis" (cf. 6C.2.4.1 n.): the difficulty is that it is often not clear which term is to be regarded as emphasized, the given or the new. Cf. Chao, Mandarin Primer, (1948:54): "(Modern Pekingese) Tajuh tzay Beeipyng (45) 'He lives in Peiping', emphasizing the idea that his home is Peiping, whereas Ta tzay Beeipyng juh emphasizes the idea of his maintaining an abode and the locality is only an accompanying circumstance". Frequently, at least, in the first example "Beeipyng" is new — this might be an answer to "where does he live?"; while in the second "juh" is new — this might be an answer to "what is he doing in Peiping (-staying a few days)?" 77. The clause with final yie in (IV. 5la: 147.22—23) is an editorial footnote in classical commentary style and has not been included under instances of yie in this system. 78. The need to recognize lai as a term in the system of both sub-types is formally determined (i) by the occurrence as clause-final of both liau lai, e.g. (II. 36a; 93.9) and lai liau, e.g. (II. 36a; 93.13), and (ii) by the occurrence of final kiu lai, i.e. the juxtaposition of two forms which in any one system are always mutually exclusive, e.g. in (II. 36a; 93.9) and (II. 45b; 100.13). Lii Shu-hsiang (1947, Vol. II, 158), regards lai and laija identically, Iaij9 being the more modern, as hou-shih hsiang (46) ("successive aspect") and instances ni jian ssma Iaij9 "what was it you saw?"; he explains this as "an action having already taken place", an explanation that is certainly more appropriate than Wang Li's "recent past aspect" (chin kuo-ch 'u mao (47)) which Wang compares with French "il vient de. . .". Haenisch regards lai as a phonetic transcription of the Mongolian verb-termination -luqai (-lai) which was then taken over into Chinese colloquial, and thus as distinct from the (type (i)) postpositive verb lai; he further suggests a contamination with Modern Pekingese (final) liau (la), and notes the alternance of liau and jo lai in one instance in the translation of the same

106

"SECRET HISTORY OF THE MONGOLS"

79. 80. 81.

82. 83.

84.

85. 86.

87.

Mongolian form. Haenisch further considers final yieu as a transcription of Mongolian verb-termination -yu (pro-verb yieu in final position being equivalent to Mongolian "copula" baiyu which takes final position, with yieu again a transcription); yieulai is then equivalent to bailuqai or buluqai, with lai transcription, yieu translation (of bai- "sich befinden" or bu- "is", its transcription equivalence in baiyu being as it were transitional). Compare attributive and substantive uses of di in the genitival clause above (6C.2.8). With auxiliary noun go only; see 6C.3.9. Actually ]e2 (172 occurrences) is found once, na (175) four times, followed by the auxiliary noun. This class in Modern Pekingese is H.E Simon's "determinative" (op. at., pp. 330 ff.). The "Secret History" has no "demonstrative determinative" (on the criterion of "colligability" with the auxiliary noun), and it is (partly for this reason) convenient to restrict this class to the "numeral noun" and to regard Simon's "specific determinative" as an adverbial class. The use of Je2 "this" and na "that" before a noun group with a numeral noun is regarded as an instance of a nominal adverb (compare the instance of a verbal adverb preceding a verb group or complex group): see 6C.3.11.3. Including "compound numerals ending with" yi, etc. H.E Simon's "unrestricted determinates". This class is not required in the "Secret History" since it would need to include (apart from a few forms which in the text are bound characters) all free nouns occurring in noun groups structure NuF (of which there are 45, in a total of 293 ocurrences). Simon's "restricted determinates" recall my auxiliary nouns of type (ii) below, a small class of nine members — which might have been eliminated altogether if the "Secret History" had been a longer text: thus cu2, which one might (because of Modern Pekingese) have expected to be restricted to minor noun groups, in fact occurs in noun groups with a free noun, as sam cu2 di2mian "three places", and is therefore classified as type (ii). Written throughout with gen ("root") not gen2 ("heel"). The two characters are now kept distinct, and the latter would therefore be regarded as more correct; speakers of Modern Pekingese are aware of etymological identity between the words. Thus forming an exception to the non-occurrence of pronouns in the noun group. Compare Modern Pekingese g9n ta su9 "said to him", where gsn (gen2) is the prepositive verb; the "Secret History" has never gen here but always duei — which itself collocates with genxian. Compare also the later zusia (zu2hia) as a form of address. Hauer: (1931:177—180) points out that in its original sense genxian should

107

MEDIEVAL CHINESE GRAMMAR

88.

89.

90. 91.

92.

93.

94.

95.

be translated "zu fussen", whence it comes to have the meaning "bei, zu". See especially example 6D.3.10, nos. 8-9 on the CD-ROM provided. Haenisch notes that genxian in the Gloss is equivalent to Mongolian dergede ("bei,zu"). As said in 6B.3.4., the implication of "free/bound" as applied to this class is somewhat different: the bound adverbs, like bound verbs and nouns, are bound by their grammatical form, whereas the free adverbs of types (ii) and (iii) are bound by their function of being attributive to a word. Those of type (i), being attributive to a clause, are free to stand as exponent of a clause. Curiously, there is no interrogative adverb with time reference ("when?") in the "Secret History". E.g. (VII. 50a; 197.51) guo2 fan buxerj bei wu3 "and indeed she had not been violated"; (II. 49a; 102.8) na giunmui yiuanlai si sam jurj2Merki rin "now the soldiers were (turned out to be) men of the three Merkit peoples". I am indebted to Dr. C. Birch, of the School of Oriental and African Studies, for the following example from the Ku chin hsiao-shuo (24), which parallels the use of this form in the "Secret History": Jin jug maifu5jo gurjzianseu, dan kiu cur) Jin di du bei se2huei "within the defence ring bowmen were concealed, and anyone attacking (dan . . . di) was driven back by their fire" (chuan 6 "Ko Ling-kung sheng-ch'ien nung Chu-erh" (48), probably Sung). Haenisch considers rja to be a final particle (emphatic); that I regard it as an adverb is due to the fact that not merely does it operate as one term in the system of forms marking a conditional clause but also it regularly combines with the preverbal adverb. For rja in clauses with interrogative word, see 6C.3.17. Haenisch regards sagteu2 as (with sag) postposition, equivalent to the Mongolian instrumental suffix or to tula "because". I have distinguished between the two uses as final adverb and as postpositive noun (where, however, it can be translated "because of" and collocates in all three occurrences with the prepositive verb wuei). ki3 in the "Secret History" marks a conditional clause the sense of which is determined by the final adverb. ki3 excludes preverbal N from the clause, and this fact (or, from a historical point of view, its pronominal origin) could be accounted for by the ascribing to ki3 itself of value N - or (since this would not account for its restriction to a bound clause) of value N to the complex ki3 + V. In translation, e.g. (VII 20a.; 191.12) ki3siuan2. . . si2 = ""when selecting . . . " or "in the selection of ..." (rather than "in his selection of. . ."). The one occurrence of ki3 without final adverb (VII 25b; 193.27) could likewise be analysed as ki3 + V = N. The inclusion of ki3 in the preverbal adverb system has, however, been preferred for (non-

108

"SECRET HISTORY OF THE MONGOLS'

96.

97. 98. 99. 100.

101.

102. 103. 104.

105. 106.

107.

108.

historical) descriptive statement, since it does not require the (unique) provision for verb substantive occurring at value N. The classification of si2 and heu as conjunctive adverbs in the "Secret History" depends on the punctuation; since they occur (9 and 8 times respectively) following a stop, as piece-initial, they are assigned to the (first) clause in the piece in which they occur. If the stop were ignored, they could be assigned as final adverbs (which they regularly are) to the previous clause. Also xai (3), farjxai (1) and xaifarj (1). Siuan . . . siuan occurs only once inthe "Secret History" (IV. 44a; 145.16). For examples from other texts see Ota (1953: 8-9). san2ma3mui "gelding", in (VI. 43b; 181.27). The term "chuan-yin" (49) was applied in Chinese phonology to pairs of characters (regarded as one derived from the other) distinguished by a simple opposition at one position in the structure; in particular "tui chuan" (50) signified an opposition of nasal final and open final. The term may well be applied to pairs having in final position different terms in the same prosodic system: thus Modern Pekingese m9i/m9n, "Secret History" mui/mun are chuan-yin. See Wang Li, Chung-kuo yin-yun-hsueh, pp. 1.78-79. Haenisch considers mui as coming into the Chinese version, and into spoken Chinese generally, through the translation of Mongolian documents, via the Gloss, where it is regularly used for the Mongolian plural termination (but derived from Chinese nominal adverb mui, not a phonetic borrowing). Its use in the Gloss is considerably wider, not being restricted to human reference. On the use of mui and mun in other Yuan texts, see Ota Tatsuo (1953:9-10). Including once with negative prepositive verb wu: (IV. 44a; 145.13) xerj wu yi fin gian je "there was no one saw him". E.g. (XL 27b; 254.19-20) Joci wuei2 duei, Caadai suo "before Joci had answered, Caadai said". It is interesting that the Modern Pekingese equivalent jdau (Ji7dau3) is similarly rarely found with the usual perfective negative m.9i (when it is, it requires translating "find out"). Thus mo suo here has the final position taken also by the analytically comparable interrogative particle madau3; cf. 6C.3.17 n. E.g. (V. 45a; 167.28 (as amended)) tian mobu bu rjai2hu ma "surely heaven will (it cannot be that heaven will not) protect us?" ; (V. 43a; 166.7) nim mo huan3 yi7jarj4 ta ma "surely it cannot be that you still trust him?". bu here does not function as a negative adverb, as shown by the identity of mosi and mobusi; for this reason these forms are regarded as single (compound) words. And zemrin2 (1) in (X. 25a; 242.25) "how (could (you) bear to)?", and

109

MEDIEVAL CHINESE GRAMMAR zem (1) in (IX. 18b; 214.43) zem nai ta ho "how could you have coped with him?" 109. The single instance of verbal adverb mo ("do not!") is (i) preceding suo "say" and (ii) (with suo) clause-final, which suggests comparison with interrogative particle madau, of which the syllable dau3 as free verb = "say". In Maspero (1914) ma occurs as final particle beside zuoma (-serj) "how?", the latter form suggesting (i) some relation between zuoma and ("Secret History") zem, (Modern Pekingese) z9(n)ma, and (ii) a use of ma = "what?" (cf. "Secret History" simma; Modern Pekingese gan ma "what (are you) doing?" (= "why?"); and Cantonese zou mat, same use). Demieville, (1951:15—22) suggests that ma may originate in an alternative form of question having negative with initial labial nasal combining with final particle rja into a single syllable. He further suggests that the Modern Pekingese pro-verb maiyiau is an archaizing form of wu whose Middle Chinese final (-iu) could only be preserved disyllabically in the Modern Pekingese phonological system (we may note that while Modern Pekingese does contain one syllable mi9u (e.g. in huarj2mi9u "absurd"), this is a unique form, and in ch'u-sheng (43) tone); in this connection we may note the alternance of sim/simma in the "Secret History", of which Modern Pekingese, which excludes final -m, preserves only s9(n)ma. As an example of a similar series in a southern dialect where -m remains in the final system, mention may be made of the Cantonese forms mat "what?", dim "how?" and the negatives m, mei and mou; together with final interrogative particles ma (regarded by Chao (1948:162), as "fusion of" m and a) and me "surely . . . not. . .?". 110. The total occurrences of je 2 and na may be broken down as follows: (a) = nominal adverb, (b) = "numeral" noun (before auxiliary noun), (c) = in , adverb Je2li, nali "here", "there (/where?)", (d) = in adverb je,ban, naban "like this", "like that", (e) = in other adverbs.

je 2 na"

(a) 80 85

(b) 1 4

(c) 18 41

(d) 37 25

(e) 35 20

Total 171 175

111. E.g. (III. 28b; 117.3) danfan zuo 'anda' rja "whoever became sworn brothers" or "any who . . .", Modern Pekingese e.g. ssi zualiau 'anda', (s9i du yiau . . .) with repetition of sai (= suei) "who". The only two instances of interrogative pronominal word with following negative adverb comparable to the Modern Pekingese form are (V. 27b; 157.7) rim simma xai3wu2 du buxerj yiu "did not give any booty at all", and (VIII. 32b (388b); 203.23) rim suei bu hiu wuei4 liau "no one is allowed to disobey". Instances of an interrogative pronoun with a final adverb are: (V. 42a; 166.23) ni siarj2 zuo simma rja "whatever you are thinking of doing"; (VI.37b; 179.5) suei zau ki rja "whoever got up the first" (with no

110

"SECRET HISTORY OF THE MONGOLS" repetition of simma, suei). There is one (and only one) clause with final rja which is not bound (it closes a passage of direct speech), where rja combines with interrogative pronoun: (V. 51a; 169.6) bu ji 7 zemserj sarj2 rja "I do not know how he would reward him (i.e. he might offer any (a very great) reward)".

Ill

MEDIEVAL CHINESE GRAMMAR

SECTION 7: A 7 Lexis 7A General 7A. 1 In lexicology, that is, linguistic statement at the lexical level, the method most commonly employed is lexicography. The statement takes the form of a dictionary, and the meaning is stated either monolingually, in terms of the language under description, or bilingually, in terms of translation. The unit of meaning at this level, the word, is a category which in Indo-European languages is relatively easily defined; as a result the existence of a distinct category susceptible to lexicographical statement is presupposed, and the need explicitly to differentiate between an element having lexical meaning and one having grammatical meaning does not arise. Each word entered in the dictionary is identified in the system of grammatical categories by being assigned to one (or more than one) part of speech; this facilitates the statement of its lexical meaning by translation (especially into a language with a related system of categories) and forms a sort of cross-reference between the dictionary and the grammar. In the description of a language such as Chinese, however, there does not immediately appear a distinct category which, and which alone, is susceptible of lexicographical statement. There is no obvious unit corresponding to the "word" of Indo-European languages. For the present study a category has been established in the grammar to which the name "word" has been given, and the name has been chosen because this is in fact the unit the meaning of which can best be stated by description at the lexical level. It is not, however, the only category that could form the object of lexical statement; indeed the majority of bilingual dictionaries of Chinese enter the character as the lexicographical unit, polysyllabic forms being considered "compounds" and listed as sub-entries. 7A.2 The practice of character lexicography in Chinese has persisted largely because of the nature of the script. It stems, however, from traditional monolingual lexicography which arose to handle Old (and literary) Chinese. In Old Chinese it seems likely that the category "word", the unit between the clause and the character, can be eliminated from the grammatical system, geminations and the few "digraphs" being handled at the character level, so that the use of the character as the lexical unit would conform adequately to the principles of simplicity and 112

"SECRET HISTORY OP THE MONGOLS' economy of categories.1 The same cannot be said of New Chinese, where the character cannot be classified grammatically and leads to absurd lexicographical complexities; for example, for the meanings of the character no. 1999 fu Mathews (1947) gives (1) Clothes, mourning garments; to wear, as clothing (2) To serve, to submit; to be willing (3) A dose of medicine; to swallow (4) The two inner horses of a team (5) To think of (6) A quiver. For the establishment of a system of classes below (and within) the clause in New Chinese the additional category "word" seems indispensable. Mathews for instance, enters the character, but makes no attempt to assign his characters to parts of speech.2 Simon (1947), on the other hand, although for practical reasons (especially in order to achieve the most effective combinations of Chinese and romanized script) lists monosyllabic words distinctly from polysyllabic ones, yet in fact bases his dictionary on the word as the unit, since he gives no translation meaning to bound (in Chao's sense) characters and arranges the compounds in a strict alphabetical order of the romanized script. Simon's is, I think, the only Chinese-foreign dictionary yet to have been compiled in this way. 7A.3 Doubtless the difficulty of delimiting and classifying the word in New Chinese has contributed to this adherence to traditional character lexicography; and here of course grammatical studies must precede lexicography since it is at the grammatical level that the category of word must be set up. The grammatical statement, besides establishing the category of word, sets up a system of word classes or "parts of speech" and, for the "Secret History" at least, distinguishes between those classes of word which are purely lexical (free verb, free noun and free adverb) and those which are lexicogrammatical (the remainder, except the particle which is purely grammatical). A word assigned to a lexical word class cannot be further described at the grammatical level beyond the assignment to its class of a value in the clause structure; it does not form part of a grammatically determined system, so that we can characterize the lexical word classes as "infinite". The lexicogrammatical and grammatical word-classes, on the other hand, are "finite"; the members are terms in a grammatically determined system (or systems) so that, in addition to the lexical statement, grammatical identification of these words in their places in the systems is necessary for their full description.3 For the particles, no further, lexical, information can be added to what is stated about them in the grammar. The fact that a comparatively large part of the Chinese word stock must be stated both lexically and grammatically is linked with the absence of morphological distinction between word classes; there is 113

MEDIEVAL CHINESE GRAMMAR

no morphological grouping of elements into system.4 A system of verb terminations such as the person and tense system of the Latin verb can be exhaustively stated by paradigmatic treatment of the verb; but the Chinese verb has no paradigm. The statement of voice, aspect and mood, like that of the other grammatical classes, is indeed paradigmatic, or systemic; but these are categories of the clause and imply no paradigm within the word. The "group" resembles the word paradigm of inflected languages in that it consists of a free form with determining elements; but the systems made up by the bound elements, such as the postpositive verb, can only be stated lexically and do not constitute paradigms of the free form. 7A.4 The distinction made in the grammatical analysis between the "purely lexical" or "infinite" and the "lexicogrammatical" or "finite" sections of the lexis is of course operative at both these levels of analysis. The classes of words which form the "infinite" section, namely the free verb, free noun and free adverb, are placed in the grammatical system in such a way as to define their structural relations with each other and with the other word classes; but the classes themselves are diachronically "open" (new words entering the language will come into these classes the synchronic correlate of this is what is referred to here as infinite) and their members have purely lexical meaning and cannot be further described grammatically. For the lexical statement of such words the method most suited to a textual description, which will not normally extend to lexicography,5 is that of statement by collocation.6 This method records the occurrences of a word in regular associations with other words. This may be supplemented by a lexical grouping of words into "ordered series". By these two methods words are ordered in lexical series, defined as groups of words having in common some feature or features of lexical context. 7A. 5 The interrelations among the members of the finite word classes are stated grammatically rather than lexically, and these have been accounted for in Section 6. The following two sub-sections therefore are concerned with the classes of free words. 7B gives the principal collocations of the free verbs and free nouns that are important in the text, including "extended" collocations (where the forms collocated are not contiguous). The principle of ordering is that of the thesaurus. Words are arranged under some 20 headings, these being such that words found to operate in identical or closely related contexts can be grouped under one heading. The headings themselves are thus abstracted from the material presented by the text, though they may be fairly well predicted from a knowledge of the social context; it should be stressed, however, 114

"SECRET HISTORY OF THE MONGOLS' that the collocations are linguistic and not sociological - they show the place of a word in the language, not the role of an idea in the society. In 7C the "ordered series of words" are again limited to the free word classes, and the grouping is restricted to words ordered by their interior form (morphologically); it could be more widely interpreted (for example to include purely contextual ordering, as with the kinship terms which I have in fact handled in 7B). The lists are exhaustive in each case, and each word is accompanied by a specimen translation. Please refer to the CD-Rom provided for Sections 7B and 7C. Notes for section 7: A 1. One would then establish character classes of (for example) verb, noun, etc. 2. Mathews tends to restrict the translation forms for one character to one English word class. But exceptions are by no means rare: e.g. no. 3025 i "Idleness, ease. To err. Out of office". 3. The partly grammatical character of a large part of the Chinese lexicon has been remarked on by Frei (1941: 142) with reference to the Modern Pekingese verb system; and even de Saussure's observation that "le type de I'ultro-lexicologique est le chinois" (see de Saussure 1949: 183), while seeming to contradict this view, in fact reflects the same phenomenon (though no doubt with reference to Old Chinese): what de Saussure saw as the relative arbitrariness of Chinese ("les langues ou I'immotivite atteint son maximum sont plus lexicologiques") was the lexicogrammatical exponence of grammatical categories, which when observed from another standpoint is seen to be the partly grammatical character of sections of the lexicon. 4. With some exceptions; see 7C. 5. Cf. Firth (1951a: 80): "In the case of those languages of which a dictionary has not yet been written, the investigator is not, in principle, expected to compile one, but he should endeavour to classify as large a number of vocabulary elements as possible. He should specify which areas of vocabulary are the richest and the most differentiated." 6. For collocation and "collocability", see Firth (1951b: especially pp. 123 ff.) Of statement by collocation, Firth says: "Just as phonetic, phonological and grammatical forms well established and habitual in any close social group provide a basis for the mutual expectancies of words and sentences at those levels, and also the sharing of these common features, so also the study of the usual collocations of a particular literary form or genre or of a particular author, makes possible a clearly defined and precisely stated contribution to what I have termed the spectrum of descriptive linguistics, which handles and states meaning by dispersing it in a range of techniques working at a series of levels" (p. 125). 115

MEDIEVAL CHINESE GRAMMAR

SECTION 8 8 Phonology and transcription 8.1 Research in the field of Chinese historical phonology, by Karlgren, Maspero, Simon, Lo Ch'ang-p'ei and others, has yielded a great deal of information about the structure of the syllable at earlier stages of the language. The most complete account of the syllabic structure of an earlier form of Chinese is Karlgren's "reconstruction" of Middle Chinese,1 the language of Ch'ang-an, the capital, in the seventh century AD. Karlgren gives first a complete inventory of initials and finals, followed by a lexicon of Chinese characters with their readings. The transcription which Karlgren has devised for the purpose is best regarded, perhaps, as an efficient notation for presenting the phonological system as induced from the data available, as in fact a summary of distinctive features which are to be ascribed - provided the external linguistic facts are correct — to one and the same form of Chinese at round about this time. The sources for Chinese historical phonology include the work of the Chinese phonologists themselves, who very early arrived at the analysis of the syllable into initial and final — their work extends from the seventh century Ch'ieh-yun dictionary (which used the fan-ch'ieh (28) method of description; see above, 5.2.2) down to the detailed studies of the Ch'ing scholars2 in the classification and analysis of the syllable; modern Chinese dialects and Chinese loanwords in Japanese, Korean and Vietnamese; comparative historical studies in Sino-Tibetan languages; and transcriptions, the writing of Chinese in foreign scripts (such as hPags-pa) and of foreign languages in the Chinese script. The Sino-Mongolian script of the "Secret History", which dates from a time roughly half-way between the period of Middle Chinese as described by Karlgren and the present day, is one such source; it has been studied, together with the almost identical Sino-Mongolian script of the Hua-i i-yii - the two can be regarded as a single script — from the points of view of Chinese and Mongolian phonology, for example by Pelliot and Lewicki.3 Complementary data are provided by such nearly contemporary sources as the hPags-pa documents studied by Dragunov. The language represented by these documents has been called (by Karlgren, Maspero and Dragunov) "Ancient Mandarin". I propose to substitute, following normal English practice, the term "Old Mandarin", and to add the term "Early New Chinese" whereby to place this form of the language in the general historical linguistic perspective which goes back 116

"SECRET HISTORY OF THE MONGOLS' to Old Chinese (Karlgren's "Archaic Chinese") in the first half of the first millennium BC. The scale is smaller than for most Indo-European languages, but as long as the name "Chinese" is accepted for the whole period this cannot well be avoided. 8.1.1 It is clear from the Sino-Mongolian script that the phonological system of that form of North Chinese represented by it can be more closely related to the system of some form of Modern Chinese than to that of the reconstructed Middle Chinese. The name Early New Chinese is more suitable than Late Middle Chinese, which can more appropriately be applied to (at least the earlier stages of) what Dragunov called Middle Chinese.4 The criteria required for New Chinese will be (i) the absence of final plosives, and (ii) the absence of voiced initial plosives and affricates. These may be expressed descriptively as (i) the occurrence in final position of only the two terms vowel and nasal consonant, and (ii) the occurrence in initial position where C (initial consonant) = plosive or affricate of only the two terms aspirate and unaspirate. In each case historically a three term system has been replaced by one of two terms; and both these features are attested by the Sino-Mongolian script.5 The reference of the phonological system of the Chinese of the "Secret History" to some form of Modern Chinese rather than to the reconstructed Middle Chinese is preferable not only on account of the higher degree of relationship but also because by taking New Chinese as a point of reference one is starting from what is known and verifiable. Which dialect of Modern Chinese forms the most satisfactory point of reference depends on a number of factors which include the availability of phonological data for that dialect and the purpose of the description. There is no one modern dialect which shows all the phonological features attested by the Sino-Mongolian script for Early New Chinese. All such features are found to occur separately in some or other non-Mandarin dialect or non-Pekingese form of Mandarin; but there is no form of Modern Chinese which agrees with Early New Chinese both where the latter differs from Pekingese and in the features shared by these two. While Modern Pekingese would not, at any rate on the most obvious criteria, seem to be the modern dialect most closely resembling Early New Chinese (some form of Mandarin which preserves the two series dental and velar initial before high front vowel would be a first choice), yet the difference in degree of relationship would be slight enough for Pekingese, being the best known of the modern dialects, to be considered a suitable point of departure for a comparative description. 117

MEDIEVAL CHINESE GRAMMAR 8.1.2 In the present study, the primary datum for which is a written text, the main focus of attention is on the grammar and lexis of the Chinese "Secret History". The phonology of the spoken language represented by the text, or at least of forms of Chinese not far removed from it in time and place, has been studied by Dragunov and Lewicki, whose work together with that of other scholars permits the filling in of further details to the general picture presented by Karlgren. I have made no attempt here to make further original contributions to Chinese historical phonology. My aim has been to arrive at a transcription for Early New Chinese by taking into account those features where the Sino-Mongolian script shows the language of the "Secret History" to have differed from Modern Pekingese. The transcription which I have used for the Chinese of the "Secret History" is related to the phonological system of that language by the use of data provided by the Sino-Mongolian script to modify the transcription of Modern Pekingese on which it is based (see 8.4). It is not a reconstruction of Early New Chinese. It should rather be thought of as a tool for the linguistic study of a written text, the Chinese version of the "Secret History", which enables statements to be made about the language of the text without their being invalidated by inconsistency with the known phonological data from within or outside the text. 8.2 The Sino-Mongolian script does not of course contain a complete syllabary of contemporary Chinese. Even if the total number of distinct syllables required in the phonological systems of the two languages had been exactly identical — itself an improbable coincidence — only by a tour-de-force could a one-to-one correspondence have been achieved in carrying over the symbols of one to fit the system of the other: the sort of thing which is sometimes attempted in the creation of an alphabetic script using exactly the 26 letters of the Roman alphabet. In fact the creators of the Sino-Mongolian script made no such attempt. On the one hand, the script fails to reflect some of the distinctions in Mongolian;6 on the other hand a large number of Chinese syllables, even from the starting point of the Modern Pekingese system, are not used: no character representing them occurs in the Sino-Mongolian script. The latter applies both to single syllables, whose omission leaves a gap in the Chinese system, and to systems as a whole; for example there is no initial f.7 And the Sino-Mongolian script shows no trace of the tone system of the Early New Chinese syllable.8 There can therefore be no certainty that all the systemic differences between Early New Chinese and Modern Pekingese are revealed by the Sino-Mongolian script. This is still more true of non-systemic 118

"SECRET HISTORY OF THE MONGOLS' differences in articulation, representing changes which may be assumed to have taken place but which are by nature often inaccessible to investigation. It is impossible to say, for example, whether the articulation of syllables with Modern Pekingese retroflex initial which are assigned to the palatal class in Middle Chinese was such as would recommend their classification as y-prosodic in the system of Early New Chinese. This distinction, between systemic and non-systemic changes in historical linguistics, is one of degree only, since the one begins as the other, and the change is then seen as qualitative as between a particular two in a series of descriptive statements. To the extent that (i) the values of the symbols of the Sino-Mongolian script can be attested from Mongolian, and (ii) the syllables of Early New Chinese are represented in the Sino-Mongolian script, to that extent is the description of the Early New Chinese syllable yielded by the Sino-Mongolian script complete. 8.3 There would be much to be said for making use, if possible, of a transcription already in current use and therefore familiar to sinologists. Unfortunately there are difficulties. The Wade system, the transcription most familiar to Western scholars, is unsuited to linguistic purposes: it has too many inconsistencies (for example (i) chien but chuang, either chian/chuang or chien/chuong being preferable; (ii) chi/chih, ch'i/ch'ih but his/shih) and exists in too many discrepant versions. Another well-known system, much better suited to Modern Pekingese for pedagogical purposes, is Gwoyeu Romatzyh ('GR'). This system, devised by Y. R. Chao, was introduced by W. Simon for the teaching of Chinese at the School of Oriental and African Studies (University of London); it is now widely used by English, and to a lesser extent American, scholars for work not only in Modern Pekingese but even, for example, by Simon in his studies of Old Chinese grammar. But there are difficulties in the way of its use (i) where any phonological statements are made or implied, and (ii) particularly where any form of Chinese other than Modern Pekingese is in consideration.9 The most important of these difficulties are (i) the lack of consistency in the initials (the four retroflex initials being represented as sh, ch, j, r) and (ii) the use of orthographic variation to show syllabic tone (e.g. pian pyan pean piann), these changes affecting sometimes the initial consonant, sometimes the medial vowel and sometimes the final. The Sino-Mongolian script does not yield a system of syllabic tone, and since this system does not have to be taken into account in the description (its role is in any case extremely complex) it has seemed to me most desirable to ignore it altogether. It would have been possible theoretically to use the 'basic form' of GR; this is the unmodified form 119

MEDIEVAL CHINESE GRAMMAR (e.g. plan) and represents the first tone (yin-p'ing (52)), or, with certain initials, the second (yang-p'ing (44)). But this already involves a radical departure from the whole principle of GR, and I have thought it preferable to devise a new system which avoids these disadvantages and which can easily be modified for Early New Chinese.10 This system is based independently on a phonological analysis of Modern Pekingese but has been made to resemble the basic form of GR and should cause no trouble to those familiar with that system. The main differences are those which will permit the modifications required for Early New Chinese, such as the replacement of the palatal series of initials by two series, one dental the other velar, which operate also in non-palatal syllables. With this transcription the text of the Chinese "Secret History" can still be read aloud in Modern Pekingese, as it would be by a Pekingese speaker today.11 8.4 The transcription devised for Modern Pekingese is given in Appendix A, together with the syllabic analysis on which it is based. The modifications of this transcription which are suggested by the SinoMongolian script for Early New Chinese are as follows: (1) The series Fn (nasal final) has three terms: velar, dental, labial (rj, n, m). This gives a six-term final system, where Modern Pekingese has only five (aF(1)y/wF2). (2) The series CP (palatal initial consonant) does not occur. In syllables where Modern Pekingese has C = P (ci), Early New Chinese has C = V (velar) and C = Da, f (dental affricate, fricative) (ki,xi). (3) The series Vm (mid vowel) has two terms: front and back (e, o), where Modern Pekingese has only (9). (In syllables with y/wF, Vm is, however, one term, written (e)). a (4) Syllables of Modern Pekingese type where C = R (retrom y/w

flex) and F = n (can), assigned to the palatal class in Middle a Chinese, are classified as type (cin) (those with F = o remain Y y/w

as in Modern Pekingese); those of type

a a a

h

: c j s f are classified as

a v a v v v v v —: ci ji si ri (but 1 remains —: f).

yy

aa

(5) Syllables of Modern Pekingese type

a "my/w

120

where C = L (labial)

"SECRET HISTORY OF THE MONGOLS' (except f) (ban, bai), together with the syllable lai, assigned to the a labial class in Middle Chinese, are classified as type (bun, wvv y/w / bin).12 One final modification that has been made is a partial lexicalization of the transcription by the addition of subscript figures to distinguish among homophones (i.e. different lexical characters corresponding to one syllable). There is no significance in the ordering, beyond frequency of reference in the present work: the character most frequently met with is transcribed without figures, and corresponds to the first of the graphic characters listed with the syllable in the Chinese Syllabary (Appendix F). The figures serve a dual purpose: (i) to establish a unique correspondence between the transcription and the symbols of the Chinese script, and (ii) to give values to the transcription which are not limited to the level of phonology, since it would be misleading to restrict it to this level, as is suggested by the fact that from the Chinese point of view there is not one character (for example) bu, but many. (The subscript figures are not used, however, in the transcription of the Sino-Mongolian script, where the situation is quite different.) 8.5 For the reading of the Mongolian words in the Chinese version it is not difficult to modify slightly further the transcription set up for Early New Chinese. The resulting transcription (which would be one possible form of representation of the Mongolian version) is not a reconstruction of Middle Mongolian, nor a transcription of a reconstructed version of the Mongolian text in some other script, but a transcription of the SinoMongolian script; in this it is unlike the transcriptions of Pelliot, Haenisch, Shiratori and Mostaert and the second transcription of Kozin, but resembles rather Kozin's first transcription (his 'trans-transcription') — with the difference that Kozin made the modifications necessitated by the Mongolian language directly from the Modern Pekingese transcription values of the syllables, whereas what is done here is first to set up a transcription for Early New Chinese and then to modify this transcription for the purposes of the Mongolian. The resulting transcription favours the incorporation of Mongolian words in the Chinese text, just as the use of the Sino-Mongolian script favoured the incorporation of Mongolian words in the Chinese version in the original text. At the same time it can be kept as close as possible to current transcriptions of Middle and Classical Mongolian. 8.5.1 For the purpose of such a transcription the Sino-Mongolian script is recognized to be a consistent and systematic script in its own 121

MEDIEVAL CHINESE GRAMMAR

right. There are, as has been shown (see 4.3), inconsistencies in its use in the text; for example, the negligence with regard to the auxiliary transcription symbols, whose omission is an error to be rectified, but these are not defects in the script itself.13 The use of alternative characters for the same Mongolian syllable creates no inconsistency. For the Mongolian language which it represents, the Sino-Mongolian script recognizes a definite system marking the following features: (1) Division of words into syllables with structure (C)V or (C)VC, with syllable-final C marked as such by the use of (a) symbols representing syllables with final nasal consonant and (b) phonetic auxiliary symbols.14 (2) A four-vowel system giving CV4(C) with V = a i o u (3) A two-term system in the open vowel, giving Va = a/e; other vowels having one value only: o i u. (4) Five diphthongs, including two pairs with open vowel (a): ai ei au eu ui. (5) Eighteen consonants, including: (a) eight plosives (including affricates), with three pairs of voiceless/voiced: t d c j k g, and two others: b q. Note (i) q excludes e in the syllable, k g excludes a; (ii) t c k q occur also as syllablefinal. (b) ten others: m w s n l r s y h r j . Note (i) rj occurs as syllable-final only; (ii) m 1 n s r s occur also as syllable-final. To arrive at the transcription values given above, it is necessary to make the following modifications in the system set up for Early New Chinese: (a) (b) (c) (d)

ie, uo, ou are replaced by e, o, eu. ia, ua before nasal final are replaced by e, o.15 a after k, g is replaced by e. am, an, aq after initials c, J are replaced by em, en, ei) in syllables of palatal class in Middle Chinese. (e) Syllable f is replaced by (syllable-final) r. (f) Initial r) is omitted. (g) Initial h with diacritic auxiliary jurj is written q.

(h) Initial 1 with diacritic auxiliary se3 3 is written.16

(j) Of the phonetic auxiliaries the initial only is written.16 8.6 The modifications of the Modern Pekingese transcription required in the transcription of Early New Chinese and Sino-Mongolian may be summarized as under (L = labial initial, N = nasal final): 122

"SECRET HISTORY OF THE MONGOLS' Mod.Pek.

(1)-n (2) ci (-) ji (-) si (-) (3) —9 —ia —ua —ai —an —au —ar)17 (4) caN jaN saN raN (5) paN baN maN faN (6) Lai lai (7) c J s r 1 (8) -iaN -uaN (9) kaga— (10) caN jaN (11) g(12) h(13) 1-

EN.C -m-n ki (-) xi (-) gi (-) zi (-) hi (-) si (-) —e —o —ie —io —ue —uo —ei —en —em —eu —erj ciN ceN JiN JeN siN seN riN puN(peN) 19 buN (beN) muN (meN) fuN Lui lui ci Ji si ri r

S-M -m-n ki (-) gi (-) hi (-) si (-) —e —o —e —o —ei —eu18 ciN JiN

buN muN Lui lui ci Ji si

(as Mod. Pek.) (as Mod. Pek.) (as Mod. Pek.) (as Mod. Pek.) (as Mod. Pek.) (as Mod. Pek.)

-r -eNyaN -oN kege— caN ceN JaN JeN

h-q1-r-

8.7 What has been arrived at in the foregoing paragraphs, together with Appendix A, is in effect one transcription, that of Modern Pekingese, and two modified versions of it, one for Early New Chinese and one for the Sino-Mongolian script. The two modified versions have been made for the purpose of the linguistic study of the Chinese version of the "Secret History", that of the Sino-Mongolian being primarily for the transcription of Mongolian words in the Chinese text. They are "derived transcriptions", of a different type from the original 123

MEDIEVAL CHINESE GRAMMAR

Modern Pekingese transcription, since they are at a stage removed from the phonetic data which form the basis of the latter: they are not accounts of phenomena at the level at which they are directly observed by the linguist, but interpretations of phenomena observed and recorded for other purposes. It is probably fortunate that these purposes were at least at one stage linguistic: the study of Mongolian by Chinese officials of the early Ming government.20 The transcribers of the Mongolian version into Sino-Mongolian script can justifiably be regarded as linguists merely in virtue of their doing this work; but they might have been less efficient as linguists if they had been doing it for other purposes. For the study of the language of the Chinese version the efficiency of the Sino-Mongolian script is of secondary importance, since the information which it yields on the phonology of Early New Chinese can no doubt all be found elsewhere. There is, however, a particular interest in being able to use such information as being actually contained in the same text, since, whether or not the Sino-Mongolian transcription and the Chinese translation were made at roughly the same time (and it appears likely that they were), the fact of their inclusion in the same text from an early date meant that the two versions would be read by one and the same person — who would moreover, if he was a Chinese learning Mongolian, pronounce the Mongolian with the phonetics of his own form of Chinese. We cannot know the standards of Mongolian scholarship achieved by the Chinese students of the language; but in spite of its errors,21 and the inconsistencies in its use, the Sino-Mongolian script was not badly suited to the purpose.

Notes for section 8 1. See Karlgren (1940), which also includes a reconstruction of Old Chinese (c. eighth century BC). Karlgren uses the terms 'Ancient Chinese' and 'Archaic Chinese'. 2. See Wang Li (1936). 3. Lewicki (1949:26): 'Nous avons pris pour notre point de depart la langue mongole et, en restituant la prononciation du chinois du XIVe siecle, nous nous sommes laisses constamment guider par les donnees du rnongol de 1'epoque'. 4. Dragunov (1930:630): 'The Chinese language of the IX-XII centuries: the tables of Ssii-ma Kuang, Annamite borrowings'. 5. As the term 'Old Mandarin' implies, these features are not of course to be ascribed to dialects other than North Chinese. Nor can they be assumed

124

"SECRET HISTORY OF THE MONGOLS'

6.

7.

8.

9.

10. 11.

12.

to have been general even to all forms of the northern dialect; what is suggested is merely that any form of Chinese exhibiting these features should be called New Chinese. The classification is moreover entirely phonological; no grammatical or lexical criteria are implied, nor indeed can these be invoked until more has been described of the grammar and lexis of earlier stages of the language. Known from other sources. Cf. Street (1957: 4): 'The graph q represents either of the two post-velar consonants, which are written y and q by Mostaert and Cleaves. The Chinese transcription [i.e. the Sino-Mongolian script — M.A.K.H.] never distinguishes the two, but some other transcriptions of Middle Mongolian do so.' I.e. no character occurs in the Sino-Mongolian script having as initial Modern Pekingese voiceless labial fricative, Middle Chinese labial plosive in the syllable of palato-labial class. Lewicki (1949) considers the possibility that the tone of the Chinese syllable was significant in determining the choice between homophonous characters. His own conclusion, however, seems more probable (p. 67): 'II serait plus juste de conclure qu'un caractere chinois determine employe pour la premiere fois dans un mot, au debut de la page ou du chapitre, se repete par la suite constamment dans le meme mot ou presque constamment a la meme page ou dans le meme chapitre.' It was not intended to transcribe anything but Modern Pekingese; indeed Chao created it primarily as an instrument for the teaching of Modern Pekingese. He has himself however since adapted it to transcribe Modern Cantonese. See Chao's summary of the purposes and merits of various transcriptions of Modern Pekingese in his Mandarin Primer (1948: Introduction, pp. 10-18). See also Karlgren (1928). I have since used the same system in Halliday (1956) Since this was written the Chinese Script Reform Committee in Peking have published their Draft Alphabet in a number of different versions, and one of these has been adopted by the Chinese government as an alphabetic script for Modern Chinese. It has begun to be used for various purposes (such as the teaching of Standard Chinese) and its use will no doubt be gradually extended; whether or not it will eventually replace characters for Modern Chinese will probably not be decided for some time. It is extremely well suited to its purpose and could have been used for the present work if it had appeared in time. The use of characters representing Modern Pekingese syllables type a — with C = R, L (e.g. csrj (cirj), ban (bun)) to represent Mongolian am y/w syllables with high (respectively spread, rounded) vowel suggests that these syllables should be assigned to the palatal and labial classes (as in Middle Chinese, where, however, the basic syllabic prosody cannot be described in terms of the initial consonant and a fourth prosodic position is required) 125

MEDIEVAL CHINESE GRAMMAR

13.

14.

15. 16.

17. 18. 19.

20. 21.

or to open class with y/w prosody in the vowel, and not, as in Modern Pekingese, to the open class with open vowel. The "depalatalization" after retroflex ("supradental") initial and "delabialization" after labial initial (and, partially, after 1) had not yet (fully) taken place; see Karlgren a (1940: esp. pp. 51-52). The question arises whether syllables type a! y/w with C = R, L (can, ban) should not be similarly treated, those assigned in Middle Chinese to the palatal, labial classes (e.g. can, ban) being transcribed as cian, buan (but e.g. Jan, marj remaining), ban, however, represents the Mongolian low back vowel, and not the mid (unmarked front/back) vowel regularly represented by ENC -ua-. For Early New Chinese we therefore regard such syllables as having a- (open) prosody, as in Modern Pekingese; they thus belong to the same syllabic type as cirj, bun, etc., which are aprosodic and not y-, w- prosodic (see Appendix A). Since, however, in Sino-Mongolian the distinction front/back in open vowel (a/e) after "retroflex" initial (that is, where C = R in Modern Pekingese) is only carried by the use of MC palatal class syllables for the front vowel (e.g. Secen is sie3- can) (whereas after all other initials the front vowel is represented by syllables with y-prosody), in a transcription of Sino-Mongolian we should maintain the distinction by modifying the ENC transcription to write cian for can, etc.: see 8.5.1. The use of the same character (e.g. mu) sometimes as standard, sometimes as phonetic auxiliary, for the same phonological form might be regarded as a defect in the script. The tendency is to use the standard character in the early part of the text (up to Chapter 3) and the auxiliary transcription symbol thereafter. The latter in all syllables where final C = non-nasal; and also in a few syllables where final C = nasal by the use of the phonetic auxiliaries mu and ni, but only after Chapter 3. Except that ia is written a after y. The transcription values thus set up for the Mongolian are referred primarily to the system used by Haenisch in his works on the "Secret History of the Mongols". See also Street (1957: 2—8) (which has appeared since the above was written). But for -an, -srj after C = R, see (4); for these after C = L, see (5) and for —si after C = L, 1, see (6). But —cu, —ju in the Mongolian verb termination (regularly written with the characters ceu, jeu). peN, etc. are retained in the transcription of those (few) syllables of this group which are not assigned to the labial class in Middle Chinese, e.g. beg. Cf. Lewicki (1949: 6), according to whom both the "Secret History" and the Hua-i i-yii were 'manuels pour faciliter 1'etude du mongol'. Cf. 3.3.2. Poppe considers it possible that the errors are a result of transcription at 126

"SECRET HISTORY OP THE MONGOLS" two removes from the primary phonetic data (1950: 268): 'mistakes in the Chinese transcription prove that the original was written with Uighur letters, which can be easily confused . . . The original was transcribed according to the phonetic system of the hP'ags-pa script and it is not impossible that the original text was first written with the hP'ags-pa letters and contained mistakes which were automatically reproduced in the Chinese transcription.' The extreme opposite view is that of Kozin, that the Sino-Mongolian may have been the original version (see 1.3.2).

127

MEDIEVAL CHINESE GRAMMAR

SECTION 9: APPENDICES A A phonological (prosodic) analysis of the new Chinese syllable (Modern Pekingese) 9A.1 Pekingese is distinguished, from, other East Asian languages and (to a lesser extent) from other forms of New Chinese, by the extreme limitation of its syllabic structure. The number of distinct syllables (the syllable here being a phonological unit) employed by a typical Pekingese speaker is within a few of four hundred; with their distribution among the four syllabic tones of the language the number is something over 1,250.1 9A. 1.1 In terms of traditional Chinese phonology the Pekingese syllable may be considered as consisting of an initial and a final, the final being further subdivided into three parts. If we designate the four parts in general phonetic terms as consonant, semi-vowel, vowel, and nasal consonant or close vowel, the Pekingese syllable then consists of either (i) these four elements or (ii) certain combinations of three or two of these elements, both in the order given. Possibilities of manner and place of articulation for the four elements are then as follows: (i) (Initial) consonant: plosive, affricate, fricative, nasal (continuant) or liquid; labial, labio-dental, dental, alveolar, palate-alveolar or velar; or zero, (ii) Semi-vowel: front spread, back rounded or front rounded; or zero. (iii) Vowel: open;2 half open — front spread, central, back rounded or back spread; close — front spread, back rounded or front rounded; or apical (sometimes regarded as zero). (iv) (Final) nasal consonant or close vowel: nasal consonant — alveolar or velar; close vowel — front spread or back rounded; or zero. 9A.1.2 Only a very limited number of the theoretically possible syllabic combinations do in fact occur: the totality of such forms constitutes a phonological system, with the syllable as the unit. Below is given a statement of this system in prosodic terms.3 The statement is based on phonetic observations, primarily of the speech of Mr Lien Shihrnin4 but also of that of various speakers, made by myself in Peking in 1947—49. Three points may perhaps be stressed: (i) The statement is made with the syllable as unit: it is not an analysis of speech (contextualized utterance) but of the syllable spoken in isolation.

128

"SECRET HISTORY OF THE MONGOLS' (ii) The aim of the statement is to set up a transcription suitable for the purposes of the present work (that is, for adaptation for use in a descriptive grammar of the "Secret History"). (iii) The statement is offered as one possible method, chosen for simplicity and comprehensiveness, of ordering the material. 9A.2 We may recognize three points or positions in the syllable: these will be represented as C(onsonant) V(owel) F(inal).5 To C corresponds the combination of articulatory features represented in linear phonetic transcription as (initial) consonant plus semi-vowel; to V and F those represented as vowel and (final) nasal consonant or semivowel respectively. Each position admits of three prosodies as general features of articulation: the y-feature (tongue forward and raised), the w-feature (lips rounded) and the a-feature, with neither of these characteristics. The three positions form a hierarchy such that the direction of prosodic force or determination in the syllable is (i) inwards, rather than outwards (C —> and overcomes or' wnere

S = syllabic prosody, A.3. In fact the systems of y- and w-feature syllables are more restricted than is that of neutral syllables. The latter form a seven-term system:

which m a y b e summarized only four terms each:

The former recognize

which may be summarized as Thus (i) if F = a, then V = a; the type CWwFa does not occur; (ii) in close syllables (C = y/w), V = either a or the other term in the close system; the types C^VF, CWVWF do not occur; (iii) in close syllables, a marked prosody in V excludes the other marked prosody in F; the types Cywyypw and CVWVWF do not exist. So that of a possible 27 syllabic types, 15 occur.7 These are: 129

MEDIEVAL CHINESE GRAMMAR

9A.3.1 In each syllable type there is a limited number of possibilities in each position. The total possibilities for each position in all syllabic types are as follows: C: Place of articulation (5): L(ablal inc. labio-dental) D(ental-alveolar) V(elar) R(etroflex) P(alatal). Manner of articulation (5): p(losive) a(ffricate) f(ricative) n(asal) l(iquid) Condition of articulation (C = p a f only) (2): h (= voiceless ((p a f), aspirate (p a))) o .(= unaspirate ((p a f), voiced (f) voiceless (p a))), (ii) V: Level of articulation (3): h(igh) m(id) l(ow). (iii)F: Condition of articulation (Fy, Fw only) (2): o(ral) n(asal). (i)

9A.3.2 The terms in the system of each position for each prosodic type are as follows: C: a-prosodic syllables (26) y-prosodic syllables (11) w-prosodic syllables (21). (ii) V: a-prosodic (3) y-prosodic (1) w-prosodic (1) (iii) F: a-prosodic (1) y-prosodic (2) w-prosodic (2) (i)

A.3.3 These terms, distributed in syllabic types, are as follows: a 1. — : C20V3. C = L(except Lfo) DVR But where C = R(except Rl) Daf there V = V3 in all other instances V = V2 (Vml) 130

"SECRET HISTORY OF THE MONGOLS"

6.

except that where C = Rl, there V = Vl (Vh) and where C = V, there V = V, (VI). :^v: C19 V2. C = L (except Lfo) DVR V = ml.8 a yy/ w : C19 Vj. C = L (except Lf)D(except Daf)P. But where C = Dp, there yF = Ft (Fo). ^/^: C25 Y!. C = LDVRP But where C = L(except Lfo)Pnl, there F = Ft (wFo). ^: Cn V2. C = L (except Lf)D(except Daf)F, V = ml. But where C = LDpPn, V = Vt (Vm). y ay :C n VjFj. C as no. 5,V = 1, F = n.

7.

a w: CnV2. C,V

2. 3. 4. 5.

8. 9. 10.

11.

y

as no. 5. But where C = LD, there V F = V^ (VI, Fo), and where C = P and V = m, there F = Fl (Fo). ^iCyV^.C^DP. 9 w ^: C21V2. C = LfoDVRP, V = ml. But where C = DP, V = Vt (Vm). w 7^: C19V2. C = LfoDVRP (except Pnl), V = ml. But where C = D, there V = Vl (VI) where C = LVR and V = m, there F = F, (Fo). and where C = P, there V F = V, Fj (VI, Fo). w ~: CyVjFj. C = LfoVR(except Rfo), V = 1, F = n. W

12. y^: CuVjFj. C = D(except Dnl)VR(except Rfo).10 9A.4 For the purposes of alphabetic transcription it is simpler to distinguish all possibilities in each position regardless of their restricted distribution in given syllabic types. The symbols required may be broken down as follows: (i) C:27 L:5 h/oLp/f(4),Ln(l)

D: 7 h/oDp/a (4), Df (1), Dn (1), Dl (1) V:4 h/oVp(2),Vf(l),Vn(l) R:5 h/oRa/f(4),Rl (1) P:6 h/oPa/f(4),Pn(l),Pl (1). In addition, syllabic prosody, being abstracted from position C, 131

MEDIEVAL CHINESE GRAMMAR requires distinct symbolization (which may, however, be synthetic with C) for the three terms a / y / w. (ii) V: 5 (including V prosody) a-prosody: 3 h/m/1 y-prosody: 1 w-prosody: 1 (iii) F: 5 (including F prosody) a-prosody: 1 y-prosody: 2 o/n w-prosody: 2 o/n. 9A.5 In the transcription used here as a basis for a transcription of the language of the "Secret History" the following symbols have been employed: (i)

C: L: h/oLp/f: p b f w Ln: m D:h/oDp/a: f:s Dn:n Dl:l V:h/oVp: :h Vn:rj R: h/oRa/f: c j s f Rl: 1 P:h/oPa/f: cijisiyi Pn:ni Pl:li.n Syllabic prosodies: a-prosody: unmarked. y-prosody: addition of i after C 12 w-prosody: addition of u after C. (ii) V: a-prosody: h/m/1: (zero) 9 a y-prosody: i w-prosody: u (iii) F: a-prosody: (zero) y-prosody: o/n: i n w-prosody: o/n: u rj

Notes to the transcription: (i) In syllables with C = P, i is written once only (that is, the second i, indicating y-prosody of either the syllable or V, is omitted); thus ci, cin, cian (not cii, ciin, ciian), etc. y w (ii) Syllabic types w w andv v are not recognized in the transcription,

_y

w

such syllables being regarded and transcribed asa w anda v respectively. (Thus the formal ambiguity of w w syllables with those of 132

'SECRET HISTORY OF THE MONGOLS' o

type ^~w where C = P, caused by the omission of the second i, does (iii)

(iv) (v)

(vi)

ciue bt ciu becomes)00 ciue bt not arise: ( In the articulation of w-prosodic syllables and those with w-prosodic V, where C = P, there is initial lip-rounding; the combination iu thus represents the complex articulation of y- and w-prosodic features, In the articulation of C = Vn (rj), Mr Lien had glottal release.13 C = Vn only occurs in a-prosodic syllables with a-prosodic V. In the articulation of C = Rl Q, there is a prothetic retroflex vowel. The only occurrence of this C is in the syllabic type Rl has been included in the C system (as in one Chinese transcription, the New Latinized Script) in preference to that of F because of the more restricted character of the latter. In the articulation of V = m, at least four distinct qualities may be noted, determined by the prosodic features of C and F: front spread (in y-syllable with aF, in w-syllable with aF where C = P and in a-syllable with yFo), central (in a-syllable with y/wFn, that with yF being slightly front of that with WF), back spread (in a-syllable with aF except where C = L) and back rounded (in w-syllable with aF except where C = P, and in a-syllable with aF where C = L and a-syllable with wFo).

9A.6 The total syllabary is given in the following table, where syllables enclosed in parentheses are those of unique or irregular occurrence, while empty parentheses indicate gaps in the system.

133

MEDIEVAL CHINESE GRAMMAR

The above (15 — 2 =) 13 syllabic types contain respectively 43,65, 73, 20,11,19,43,16,11,25,28,39 and 7 syllables, making a total syllabary of 400.15

134

"SECRET HISTORY OF THE MONGOLS'

B Specimen analysis and translation16 (II. 32b-33a; 91.1-18). Sentence: Clause:

Simple Free verbal

Word: A(cl) N A(ve) buei4neu nnmui iu2siu2 gan2 ziarjlai. / Behind them the people came one after the other in pursuit. Compound: F(B)F Free verbal imperfective (bound verbal genitival)

One man riding a white horse, and with a lassoing pole grasped in his hand, Free verbal

Simple Free verbal

almost caught up with them.

Bowofcu said:

Simple

Simple

Free verbal

Free verbal

"Give me your bow and arrow. Simple Free verbal

I will have a shot at him." Simple Free verbal imperfective

"I am afraid you will be wounded," said Temujin, "and for my sake.

MEDIEVAL CHINESE GRAMMAR 9B continued Simple

Compound: FF

Free verbal

Free verbal Free verbal

I will have a shot at him."

So he went back and shot at him. Compound: F(B)F

Free verbal ergative imperfective (bound verbal genitival)

Free verbal

The man on the white horse pointed his lasso at Temujin and said: Simple

Simple

Free verbal perfective

Free verbal

"Stop there!"

All the robbers' comrades came up from behind them. Compound: F(F)F(B)F Free verbal perfective (free verbal)

v

Pr N V (N) N V p(cl) Gian ri luo huarj3hun tianse hei liau / But the sun had set and the twilight sky grew dark, Free verbal imperfective (bound verbal genitival)

So all the men behind stopped and fell back.

Free verbal perfective

"SECRET HISTORY OF THE MONGOLS"

C 9C.1

Statistical tables Summary of occurrences of principal categories

137

MEDIEVAL CHINESE GRAMMAR 9C.2

Table of occurrences of combination of marked clause classes Nominal Bound

82

Ergative

-

21

Passive

-

-

Perfective

73

73

138

-

Imperfective

-

9

17

-

Perfective

Interrogative

1

-

1

1

4

Imperative

-

-

7

-

3

Bound

Ergative Passive

Imperfective

9C.3 Pro-verb Total I II III IV V VI VII VIII 108 16 7 11 7 7 10 17 6 4 119 17 13 7 7 13 8 13 15 1 3 30 2 9 5 3 3 wu 2 1 1 wuyieu 5 5 35 4 4 4 4 3 zai 3 2 1 13 2 1 4 30 ru 2 1 1 1 5 s 90 76 7 4 mirj 1 sir) 2 suei2+numeral 11 6 1 2 noun 1 3 1 1 yiban 1 1 Je2ban 2 . 2 4 naban 1 hi

si wuei yieu

138

IX X XI 5 7 6 . 1 1 7 5 6 1

3 1 1 2

2 1

.

2 1

XII 9 2 8 3

1

1

"SECRET HISTORY OF THE MONGOLS' 9C.4 ziag ba Total of above yiu yiu. ru duei bei z. zai Ji dau zuo yiug

lig2

yi. sun wuag. tug s ni3

yi8

wuei xug zieu mir) yin2 suei4 CU3

hiag fau yiib yian dai2 wun

gig gai wuei3 gam? bui2 xiu yin

Prepositive verb (i) Total I II III 390 28 45 24 1 1 1 8 398 29 46 25 180 100 100

92 71 60 51 47 37 26 25 25 21 18 17 16 16 15 15 15 13 9 9 9 6 6 6 5 4 3 3 3 2 2 2 2 2 2 2

9 7 4 3 5 2 6

10 4 1 4 5 2 3 1

13 5 6 9 4 12 6 1 8 7 1 1 2 2 1 2

19 8 15 9 13 14 11

2 1 3 3 1 4

2 3 4 2 2 4 4

6 2 2

IV 42 2 44

V 40

40

VI VII VIII IX X 41 43 26 22 30 1 41 43 26 23 30

17 27 10 9 10 14 2 8 10 5 5 13 2 14 10 1 7 2 3 6 4 6 5 3 3 3 2 1 1 2 2 3 9 2 1 1 1 2 4 1 3 2 3 3 4 1 1 2 2 3 2 2 3 1 3

13 5 13 6 5 2 3 6 1 4 1 1 2 1 1 1 1

20 6 7 13 5 2 5 9 1 8 2 1 2 2 1

6 19 10 1 11 11 5 5 20 3 7 4 1 4 2 5 3 7 3 2 4 3 10 2 1 3 1 1 2 1 4 5 2 3 4 7 1 2 6

17 13 5 15 6 3 2

1

1

5 1 1

2

1 1 1

1 1 1

1 1 2

1 1

1 1 1

2 2

1 2 2 1

1 2 1

1 1

1 1

139

2 3

2

1

2

1 1

1

2 1

1 1

4 2 1

4 1

1 1

2

2 1 1 3 1

XI XII 25 24 1 1 26 25

1

2

MEDIEVAL CHINESE GRAMMAR One occurrence only: I II III IV VI VII VIII IX X XI XII

na2 Ii2 giggou jiyiu2 dai kian dag giu tie guo Iai2

par) kau ji8

de

du3

ge

Jau3yi8 bi3

suei2(+NN)

cirj2

jau3

Prepositive verb (ii) Total giau 230 yieu 103 56 si mirj2 43 gian 42 34 si4 cai 30 21 suo 19 Hr) 17 Jiy hiu 16 wu 13 kurj 12 siarj2 8 7 tir) tirjde 6 yiau 6 wun 5 5 de wuarj 4 wuarjgian 4 jau 4 zurj 4 rim 4 yi2wuei 3 3 dau3

I 6

13 6

II III 11 13

7 9

7 13 1 2

'2

IV 18

6 3

8 12 15 1 3 3

2 1 1

3 6 4 1

1

3

2

4

2 1

2 1 1

2 1 1 1

V VI VII VIII IX 9 36 20 21 40

5 9 5 2 2 2 2 1 1 1

1

12 4 2

3

1 2

2 1

1 1

4 2

2 1 1

9 7 6 1 5 7 6 20 11 2 2 1 4 1 2 5 4 3 2 5 4 3 3 1 3 3 2 1 1 3 2 1 2 2 1 2

1

XII 26

1

1 1

2 4 1 1

1 1 1

3

1

2

6 6 3

2

3 1 1 2

4 3 7 2

2 8 1 5 2 2 6 3 2 2

X XI 17 13

2 1 1

1 1

1

2 1

1 1

1 140

2 2

"SECRET HISTORY OF THE MONGOLS" 9C. 4 continued

Tbtal 3 3 3 2 2 2 2 SU02 giuo 2 2 funfu bau 2 yiar)3wuei 2 2 huan2 xiq 2 hiam 2 zug 2 tisuo 2 fun 2 surj 2

rin2de ju 4 lieu Ji7de pa kurjpa

II III IV V VI VII VIII IX X XI XI] . . 1 . . . 2 . . . 1 . . . 1 1 . . . 1 ... 2 . . . . . . . . 1 1 . . . . . 1 1 . 1 . . . 1 . . . . 1 . . 1 1 1 . 1 1 . . . . ... 1 . 1 . . . . 2 . . . . . . 1 . . . 1 . . . . 2 . . I

1

.

1

. 1

2

.

.

. 1 2

.

1

.

1

One occurrence only:

I II III IV V VI VII VIII IX X XI XII

yi6huo2 huanyiu derj heu2 giu2 bauyiu2

yieu2

sig Iiau2 tuei2cir)3

yi5

S4

tue2ix3

yieu6yin2

gian4 yiu3yiau mian2 yieu6 wun2suo

ti2sim2

siuan2 sigkan wueiyfii

Jau4

141

huei,

siuan3bu3

MEDIEVAL CHINESE GRAMMAR

9C.5 lai ko km yiau neg yiu3 koyi2 ken gam bei de bi2 ziag hiu hau yiu3yiau dag Jug guan siag2 yiuan3 ho2 nan gai siuyiau giau duei xig

Auxiliary verb Total 72 69 52 49 34 28 19 18 17 14 14 14 10 7 5 5 4 4 4 3 3 3 3 2 2 2 2 2

I 2 1 4 1 . . 1 . . 2 2 2 1 2 .

II 4 1 3 3 7 2 1 2 1 1 2

III

3 5 1 1 3

1 2

1

3 .

V VI VII ^VIII IX X 6 6 5 6 9 3 2 8 7 11 11 2 11 4 3 5 6 2 5 6 4 2 9 9 5 1 3 6 3 4 2 1 1 4 3 8 2 1 2 5 2 2 1 1 2 3 4 4 1 1 1 4 1 1 2 4 2 2 1 3 1 1 1 2 1 1 1 5 2 2 2 1 1 1 1 3 1 1 1 2 2 2 1 2 1 1 1 1 1 1 1 1 1 2

IV 14

2

1

yiu3daiyiau siag2yiau zieu bui junbui gamyiau huei2 yiuan3yiau Km

2 1

1 1

1 1

1 1 2 1

1 1

1 1

One oc currence only:

II III IV V VI VII VIII IX XI

XI XII 8 9 9 11 9 5 4 3 3 2 2 2 1 1 1 4 4 1

pa

miai ^ fin2 si8 gia2^ ^2 1

yiur

142

1

'SECRET HISTORY OF THE MONGOLS'

9C.6 de JU 2

dau3 gian jau ba2 bu2 po Ie3 ki san dug duan suo sag seu2 dig zin cig yiau ki, S3

dau2 hai Jiy kai ziue luo ZUO2

suei5 sig2 hau sig jug tuei sa da su mie sin man teu bai3 sag4 bi4 xiu

Postpositive verb (ia) Total 103 44 30 24 22 14 12 11 10 10 10 9 8 8 7 7 7 6 6 6 6 6 6 6 6 5 5 4 4 4 4 3 3 3 3 2 2 2 2 2 2 2 2 2 2 2

I II III 12 22 17 3 6 3 7 11 6 4 1 1 4 5 1 1 1 1 6 1 1 1 1 2 2 1

1

1

IV 2 3 3 1 2 5 3

V VI VII VIII IX X 11 10 11 4 6 3 9 5 2 3 3 4 2 1 2 1 5 2 1 4 1 3 4 3 1 1 4 3 1 2 3 3 1 1 2

1

2 1 1 4 1

1 1 1 1

1 1

2 1

1

1

1 3 5 1 1 2

1

1 3 1

2 1 1 1 1

1 1 1 1

1

1

1

1

2 1

2

1

1

2 2 1

2 1 1

1 1 1

4 4

1

2

3 2

1 2 1

2

2 1

XI XII 1 4 2 1 1 1 2 1 1 2

1 1

1

1 1

1

3

1

1 1

1 1 1 1

1

3

4 1

2 1

1

1 1

1 2 1 1

1

2 1

1

1 2 1

1 1 1

1 1 1

1 1

1 2 2 1

1

1

143

2 2 1

MEDIEVAL CHINESE GRAMMAR

9C. 6 continued

One occurrence only:

II IV V VI VII VIII IX X XI XII

bian3 yiue farj2 wuai2 bau2 Ji7dau3 Ji5 diau sir)3 bian4

gieu tuo

JU 4

huai wun3 giande Iiau2 Yiuan4

huan Ji6 gi3

Postpositive verb (i b) Total I II III 109 16 20 9 lai (other than in combination) 23 3 2 3 ziarjlai 4 2 1 ziarjhueilai 7 hueilai 2 6 culai 1 hialai 150 21 25 12 lai (total) kiu (other than 125 9 30 12 in combination) 18 1 4 4 ziarjkiu 2 2 ziarjhueikiu 5 1 hueikiu 1 1 ziagcukiu 1 6 2 cukiu 5 zinokiu. 1 3 ru2kiu 2 2 guokiu 1 1 zeukiu 168 15 37 19 kiu (total) 6 1 ziarj . (other than in combination) 9 1 huei 2 14 cu 1 6 zin2 25 1 ' 4 1 ru2 24 3 10 3 hia 28 3 3 4 guo 1 4 zeu

IV V VI VII VIII IX X XI XII 7 5 13 13 8 1 6 3 8

2 1 2

2

4

2

2 1

1

10 15 20 11 12 12 8 8 2

1

2

2

2

1

2

1 2 1 7 6

8 3 9 5 11 5

9 7

1

3 2

.

.

1

.

1

2

1 5

1

1

14 17 12 10 1 1 1 1

2

144

2 2

2 3

6

5 2 1 2

4

7 1

. 4

1 1 4 1

5

5 1 1

1

11 7 1 . . . . 1 . .

3 1 3 . 1 5

14 1

2 1 1 3

"SECRET HISTORY OP THE MONGOLS" Postpositive verb (i c) 70 7 3 10 5 52 17 4 6 6 2 33 31 8 11 2 26 5 6 1 1 21 5 . 2 4 1 19 1 18 12 3 4 4

yiu* zuo ji dau zai wuei yiu2 mir) fu yieu

13 1 9 1 2 2 4 1 1

4 3 4 2 3 1 3

4 4 6

7 2 4 2 2 3

5 6 3 1 3 1

3 6 1 3 2 1 1

3 1 4 1 3

4 1 2 1

4

1 1

1

2 1

1 1

One occurrence only: I II X

suei2 (+NN) z2 hiag

wu

* including a small number of occurrences as type (i a).

Postpositive verb (ii a)

lai kiu

Total I 27 3 57 6

II III 7 5 14 4

IV V 2 3 2 5

VI VII VIII IX X 1 4 . 1 . 5 3 10 1 5

1

1 2 2 5

XI XII . 1 1 1

Postpositive verb (ii b) yie yieu yieulai lai

9C.7

wuo wuomui nam rjammui za zamui

8 15 20 24

2 4 3

1 3 2 8

2 4 3 1

1 1

1

1

3 3 1

2 1

3 2

1 1

1 1

Pronoun Total I II III IV V VI VII VIII IX X XI XII 412 29 54 40 31 30 43 24 49 39 23 19 31 8 1 2 . 5 39 4 3 3 3 5 5 3 1 10 2 1 1 42 1 3 . 8 6 11 6 2 5 18 3 8 . 4 3

145

MEDIEVAL CHINESE GRAMMAR 9C. 7 continued Total ni 368 nimui 14 nim 53 nimmui 5 ta 389 tamui 12 fu3 1 bi 1 Ji3 9 z2 6 2 z2gi4 x 3 Ie2 2 na 1 9C.8

I II III IV V VI VII VIII IX X XI XII 22 42 19 1.5 19 62 19 75 31 22 21 21 4 1 1 7 1 10 5 3 2 7 6 3 3 3 6 5 2 1 1 1 56 58 22 216 54 36 34 18 22 28 20 15 1 2 3 1 2 1 1 1 1 1 . 1 7 1 . 1 1 2 2 . 1 1 1 2 . 1 1 . 1

Numeral noun

yi (with AN) yi (with FN) yi (otherwise) yi (total) Hag (with AN) Hag (with FN) Hag (otherwise) Hag (total) r2 (with AN) ?2 (with FN) f 2 ( total) 3 to 9 (with AN) 3 to 9 (with FN) 3 to 9 (otherwise) 3 to 9 (total) si3(withAN) sl 3 (withFN) sia (otherwise) si3 (total) gi (with AN)

Total 128 79 61 268 54 23 1 78 3 63 66 81 119 14 214 13 11 1 25 9

I 42 8 46 96 8 4 1 13

II 22 8 1 31 8 5 13

III IV V VI VII VIII IX X XI XII 5 4 4 13 10 7 5 6 4 6 19 8 6 5 1 3 12 1 3 5 7 2 1 2 2 31 17 13 13 7 10 17 5 7 25 6 1 1 9 4 6 7 2 2 1 1 1 . 5 2 4 9

9

8 11

2

7 3 3 5 3 7 7 7 3 3 5 3 7 7 23 10 9 6 1 1 4 10 14 10 10 9 7 7 2 2 6 1 1 39 25 20 16 12 8 13 1 3 2 2 3 1 1 3 1 1 5 3 3 1 1 1 1

2 1 1 2 6 6 1 13 2

2

2 2 1 1 8 6 12 9 6 12 12 5 2 12 14 17 1 25 19 19 5 5 5

6 1 1 2 2 3 5

5 2

"with AN" = "followed by auxiliary noun"; "with FN" = "followed by free noun". Each numeral includes all compound numerals in which it is final: yi includes f2si3yi "21", 3 to 9 includes samsi3xi "37", si includes Iieu2si3 "60", etc. No compound numeral occurs ending in either liaq or f 2 .

146

'SECRET HISTORY OF THE MONGOLS' 9C.9

Auxiliary noun (i)

g° xian2 wuan bai gian3 Ji2 pi xurio •"•"•VZ ban2 kuai xerj2 wurj

(freq.) (20) (15)

I II III IV V 223 68 36 29 11 9 24 9 1 . 19 . 1 2 18 1 13 9 7 1 . 1 9 4 7 7 6 2 3 1 1 1 . 3 2 2 . 1 1 . 2

Total

(5) (9) (1) (-) (-) V /

(4) (-) (2) (1)

CU2

VI VII VIII IX X XI XII 9 9 13 3 15 12 9 3 1 2 12 6 . 1 1 2 5 . 1 2 3 1 5 1 2 1 11 1 . 1 1 1 1 1

1 2

1

.

1

1

Figures in parentheses denote occurrences in the minor noun group (included in the general total, except with cu2, the frequent occurrences of which in the minor noun group have not been taken into account).

One occurrence only:

I II III IV

cirj4 huo.

gir)2 jarj 2 suarj turj3

V VI X

ba tiau gen

cir)3

Auxiliary noun (ii) Total X2

bian3 zau2

hia

23 7 3 3

I 1 2

II 1 2

III 1

IV V 2 .

2

One occurrence only:

II VII XII

xiarj du2 zie

guei jan

147

VI VII VIII IX X 1 6 6 2 1 1 1 3

XI XII 2 1 1

MEDIEVAL CHINESE GRAMMAR 9C.10

Postpositive noun Total I II III 126 21 39 24 99 5 2 11 4 4 2 67 7 14 4 55 4 5 10 53 5 19 7 50 7 7 5 27 1 3 1 27 1 4 4 22 6 17 5 17 6 1 1 16 2 2 1 13 1 2

li CU2

nei sag nig ban bian2 jug xian heu wuai gian2 hia si2

genxian yiban dur)2bian2 Iiaggian2 sibian2 Iiteu2 JTun2ian.T "Al o 2

xianteu2 saqteu2 heumian wuaimian sagmian xianmian bui2heu mianxian jeuwuei3 zuo4bian2

25 9 8 6 4 3 3 3 3 2 2 2 2 2 2 2 2

5 1

5

1

1

2

IV V VI VII VIII IX X 6 5 5 4 3 7 8 4 12 14 7 2 3 24 7 5 8 2 8 10 6 5 2 4 6 6 4 3 8 12 6 4 1 2 1 3 5 3 1 1 1 3 1 2 1 1 . 3 11 . 1 2 3 3 2 3 3 1 1 1 1 1 1 1 2 2 1 2 3 2 1 1 3 1

1

3

5 1

3

2 1

3 1

1 5 1

1

2 1 1 1 1

3 2 1 1

2

1 1

1 1 1 1

2 1 1

1

4 2 3 1 3

1

3

2 3 2

XI XII 2 2 5 10 5 6

1 2 1

1

1 1

1 1

1 2

One occurrence only: III VI VIII

g;endi3 sian g;enyiuan2

zuo4li si

yieu4li yi2lai

148

IX X

bei2bian2 yieu4bian2

xianheu zuo 4yieu4

9C. 11 Preverbal, final and conjunctive adverb (all totals are exclusive) PREVERBAL yin yin yin yin yin yin yin yin yin yin gai2yin cu4yin yinwuei fuo fuo ruo fuo ruo ruo ruo fuo fuo fuo fuo

FINAL sar)teu2 sarjteu2 disar)teu2 disarjteu2 disarjteu2 disi2fun

ga ga ga ga

ga ga si2 si2 si2 si2

CONJUNCTIVE Total 28 1 1 suoyi2 5 zieu 2 suoyi2 1 1 zieu 4 suoyi2 8 guzieu 1 yieu3 1 suoyi2 1 1 25 23 yie 3 bian 2 1 bi2 zieu 2 1 fan2 10 bian 3 kiuo 1 bi2 1

I 1 1

II III IV V 1 5 4

1

1

1

. . . . . . 2 . 1 .

VI 5

VII 3

VIII

IX

1

1

.

1 .

1

2

.

X 2

.

XI

XII 7

. . . 1

.

2 1 1 2

2

.

1

1

.

1

2

1 1.26 1 3 1 4 1 3 5 2 1 1 1 1 1

1

.

. 1

2 1

1

3

. . 1

5 2

1 1 .

.

1

.

4 .

3 1 . . .

.

.

.

2

.

.

.

.

3 1

. 1

. 1

1

. .

.

.

9C. i 1 continued PREVERBAL fuo fuo fuo fuo fuo fuo fuo fuobian tag fuo fuoguo2 fuo . . tag tarj fu fu fu fu fu fu suei3 suei3 suei3 suei3 suei3 suei3 suei3

FINAL

di di je je

r)a r)a si2 ban yiban

r)a di

CONJUNCTIVE Total 1 1 zieubian 6 1 bian 1 yie 2 bi2 1 ze 1 gis 1 1 1 1 5 1 1 2 4 1 ze 7 yie 3 1 2 kiuo 1 du ... kiuo 1 ko 1 yie

I

II

III

IV

V

VI

VII

VIII

IX

X

. 1 1

3

1

.

.

1

XI

XII

1 1

1

.

1

1 1 1

1 1 1 1

1

1 1

1

1

1 1 1

1 1

1

2 .

. .

. 2

1 2 .

1

2 . 1

. . . . 3 . . . . . . . 1 . .

1 1

. . . 1

1

9C. 11 continued PREVERBAL suei3 suei3dag wuei wuei wuei wuei jig wuei gi2 gi2 gia g!2

gi2 g!2

gi2 gi2 gi2 gi3 gi3 bi3gi3 bi3gi3 bi3gi3 bi3 giji giji ziaqgi3 ji

FINAL si2

sagteu2 disar)teu2

CONJUNCTIVE Total 1 fan 1 3 4 3 yinx 1 1 suoyi2sar)

ga si2 heu Ji3heu di

bian zieu kiuo si2

ga si2

si2

30 2 1 2 1 1 1 1 1 4 1 1 1 2 3 1 1 1 3

I

1 1

1

II

III

IV

V

VI

VII

IX

X

.

VIII 1 .

1

.

1 .

. .

. .

4

1

2

XII

2

1 2 2 1 1 1

XI

1 . 1

4

3

1

2

4

7

2

1

1 1

1 1

1

.

1 . 1 1 1 1

. .

.

1

1 . 1

. .

. .

. . .

.

. 1 . . .

. . .

. . .

. . . 1

. . .

1 .

1

.

.

.

1

.

1

1

1 1 1

9C. 11 continued PREVERBAL FINAL si2 Ji si2 Ji CU2 Ji Ji JiJi Jin ziarj CU4 CU4 CU4

si2 Ji3gu

CU4

darjcu4 dan dan dan dan dan dan dan dan danxerj danxerj danfan danfan danfan

si2

na rja si2 CU2

di di di di

ga si2 di

CONJUNCTIVE Total 2 1 kiuo 2 1 fag 1 1 1 3 16 1 yin 1 bian 1 3 1 1 bian 2 1 gie 1 1 bian 2 du 1 1 bian 1 bian 1 2

I

II

1

III

.

IV

.

V

.

VI

VII

. 1 1

VIII

IX

X

XI

XII

2 1 1

2

3

1 .

. 2 1

. .

1 .

. 1 13

1 . 1.

. . 3

. .

1

3 1

1 .

1

1

1

1 1 1

1 1 1

1 1

1

. . .

1 1

.

.

.

1 1 1

1

9C. 11 continued PREVERBAL fan fan fan fan dafan dafan ki3 ki3 ki3 ki3 ki3 ki3 yinki3 wueiki3 Z

2

Z

2

Z

2

Z

2

Z

2

Z

2

xug dai dai xai xai xai

FINAL

CONJUNCTIVE Total 2 si2 2 CU2 2 5 di" 1 ga 1 1 si2 1 gian2 1 di 2 1 Ji3gu diyiuan5gu yin 1 1 diyiuan5gu suei 1 1 si2 1 Ji3gian2 1 heu suei 1 Ji3heu 1 CU2 1 1 si2 1 1 si2 kiuo 3 1 jur)gian2 bian ba2 1

I

II

III

IV

V

"VI

VII

1

VIII

IX

1

1 2 1

X 2 1

XI

XII

. . . . . . 2

.

1 1 1 1 1 1

.^

1 1 1 1 1 1 1

. . 1 1

. .

. .

. .

1 1 1

.

2 1 1

.

. 1 1

. 1

5>C. 11 continued PREVERBAL FINAL Ji4 ji4dau3

suo suo suo suo

yi2

sar)3 (wuei2) (wuei2) (wuei2) (wuei2) CU3 CU3

jig J«)

Jig dag dagwuei (na) yiu2 yiu2 zieu (suei) wuei8 (xeg)

CONJUNCTIVE Total 1 bi 1 7 di 4 2 je 1 Ji3 3 1 3 5 si2 1 ji3sian 1 Juqgian2 1 1 Ji3wuai 3 giana 1 dijur)gian2 1 yieu3 1 si2 1 si2 1 si2 1 si2 3 CU2 1 si2 1 ga 1 ze 1 bian

I

II

1

III

IV

V

VI

VII

VIII

IX

X

XI

XII

2 2

. .

1 1 1

4

1 1

.

.

1

1 1

2

1 1

2

.

1

1 1

.

2 . . .

1 .

1

. 1 1

1

.

1 1

1

1 .

.

1

.

.

.

1 1 I

1 l 1 1

l

1 1

9C. 1 i continued PREVERBAL FINAL si2 si2

CONJUNCTIVE Total 140 2 zieu ze 1 S12 si. 1 yieu3 si2gian2 1 disi2gian2 yie 1 disi2fun 3 gian2 31 1 gian2 suei jur)gian2 4 di]uqgian2 1 ki3gian2 1 diki3gian2 1 heu 13 diheuteu2 2 Ji3heu 3 r)a 63 qa bian 7 1 qa zieu 1 r)a fan2 1 du qa bi2 2 qa qa yie 3 1 ga yieu5 1 r)a dau2 (nasi2)xai 1 r)a

I 14

II 17

III 13 1 .

IV 6

V 14 1

VI 14

VII 10

VIII 14

IX 23

1

X 6

XI 5

XII 4

4

.

1 1 1 1 4

2 5

3

5 1

2

1

2

1

1

2

1

1

1

.

.

1

1

. 2

.

1 1 1 2 1 9 3 1

5

1 .

. 1

9 1

62 . . .

.

. 2

.

1 5

. 8 2

. . _ .. - 2 -—. . ^""l 13 -.5 1

31 . . .

1 1

1 2

1 1

1

.

.

1 1 . . .

2

9C. i 1 continued PREVERBAL FINAL CONJUNCTIVE Total 1 sar)teu2 1 disarjteu2 1 disar)teu2 suoyi2 4 diyiuan5gu 11 CU2 1 hia 2 ban 1 diban 1 yiban 19 je 1 fag bi4 zieu suei yiu2si kiuo yieu3 bian suoyi2 yie si2 heu

yi4

bi2 big2

59 54 40 40 28 22 20 16 9 8 7 7 6

II

I

III

IV

V

1

1

2 2 1 . .

1

1 1

1

VI

VII 1

1 1

VIII 1 1

IX

X

2

1

XI

XII 1 1

2

1

2

1

2

3

5

1

1

3

7 9 4 4 1 2

5 1 2 1 4 1 2

6 1 3

3 6 4 6

1 7 4 4 1

5 6 2 3 2

1

1 2

1 1

1 1 1 1

1 1

15 . . . . 4 . 3 . .

4 1 3 1 4 6 4 . 2 1

3 1 3 3 2 1 4

1

3 1

2 8 8 6 4

2 11 6 6 8 2 2 5 2

4 3

6 5 5 3 1 4 2 1

1 1

1 1 1

3 1

1

1 3 2

1 1

2 1

2

2 3

9C. 11 continued PREVERBAL FINAL

CONJUNCTIVE Total 5 fag 5 yie yie yin 4 3 fanheu ze 3 2 xai 2 du 2 r3 2 gu 2 zi 2 JO yie du 1 1 siuan siuan 1 fan2 xaifarj 1 1 farjxai huo huo 1 si2 farj 1 ko 1 yi2keu 1 1 Sly 1 a3 1 gi3 ru 1 yiu2si suei 1

I

II

III . .

IV 3 .

V 1 I 1 2

VI

VII . l 2 1

l . .

VIII 1 . 1

IX

X

XI

XII

.

1

.

1

3 1

1 1

.

.

.

.

1 2

1 1

1

.

. . .

1 1 . . .

1

1 1 1 1 .

.

.

1 1 1 1

. .

1 1

.

.

.

.

.

. .

.

1 .

.

.

. . 1 1 1

. .

9C.12 (Clausal)

Particle

(non-final) (final) (unmarked) liau (total)*

liau liau liau

Aspect

: J°

Mood
U9 —di s dan

N:VN swus mai fez —di

NN:V fez wud smai—di N:NV fez swus mai —di NN:V dan ta sgdi-wua —di N:NV dan s ta gsi-wu s —di

N:N:V fez swus mai —di N:NV s f e z wus mai —di N:N:V dan s to gsi-wus —di N:NV s dan ta gsi-wus —di

ta gsi-wus dan

NNV fez wus mai

dan ta s gsi-wus

VNN

VN:N mai fez —di swua g9i-wu9 dan -di sta

219

sta gsi-wus dan —di

MODERN CHINESE GRAMMAR

(d) English labels may be attached to the words which are exponents of V, N here as follows; mai 'sold', siau 'is small',gai 'gave'; wud 'I/ me', fez 'car', ta 'he/him', dan 'money'; lai 'has come', ma 'swore at', qai~ma 'got sworn at' (8); (e) in the associative form, the colon marks the break between the given and the new, while underlining indicates the incongruent position of the new. A vertical systematization gives (i) unmarked (neutral)/marked (associative), (ii) within the associative, unmarked (congruent)/marked (incongruent). Even in the neutral form there is some system of congruence, as shown by the fact that mai fez = fez mai (i.e. in both mai is given, fez new) and not = fez mai (fez given, mai new). The horizontal systematization presents the grammatical themes on which the contextual variations are played (whereas in a description for pedagogical or translation purposes these would be presented as grammatical variations on a contextual theme); for each inversion of the elements of the grammatical structure (including, between NVN on the one hand and NNV, VNN on the other, inversion of the nominal terms (in all NVN forms wua precedes fez, in all others fez precedes wud), so that NaVNb becomes NbNaV, VNbNa) there are possible variations of the given/new structure. That the name "voice" should be given to one or another dimension along which this material is systematized is suggested by its characteristic of inversion of the terms coupled with a formal change in the elements of structure; on the same grounds the marked term in such a system may be named "passive". There are various possible groupings; if inversion is to be the main criterion, then NVN would be neutral and others passive, but this permits no classification of the two-element structures and cuts across the formal distinctness of the associative form. (It would be possible to talk of inversion with two-element structures on the criterion of contextual equivalence: mai fez : fez mai :: lai fez : fez lai but:: ma wud : u>ud gai-ma (not: wud ma}', in the neutral form NV cannot always be unambiguously replaced by VN, so that ma —di s u>ud may be the contextual equivalent of either VN ma u>U9 or NV wus ma. It seems preferable to consider the associative form as such as the mark of passive voice, all other structures (with the exception of the ergative; see next paragraph) forming the neutral term in the voice system. There is one further form characterized by inversion of the clause structure with the addition of a grammatical marker: to this I have given the name "ergative" in view of its singling out of the verbal element as new by opposition to all nominal elements. The ergative voice always has 220

GRAMMATICAL CATEGORIES IN MODERN CHINESE structure NV or NNV; where two N are present their order is that of neutral form NVN (that is, NaNbV), and not that of neutral or passive NNV which have NbNaV. The form ba (9) (word class prepositive verb; see below, 4.2) precedes the immediately pre-verbal N. As with the passive, there is an incongruent form with either N marked as new by stress. It may be remarked here that the polysystemic form of the description may justifiably be represented as a simplification and not a complication of the material. In language primers this polysystematization appears when an identical form is classified under a number of different heads; but the question whether, for example, / (6) is 'one word' or half a dozen 'different words' in Chinese does not arise if one says of a linguistic form that it operates at certain places in certain structures. Such a treatment is possible with a multi-dimensional description according to which a form may be systematized (that is, identified as a term in a grammatical system) along a number of different dimensions, sometimes for different purposes. For example, for pedagogical purposes where the language of description is English it is useful to isolate one form of the passive (as here described), the incongruent NNV form, and classify it with the ergative into a two-term marked voice system, because of the existence of a particular 'agentive' form (with by) of the English passive, at the same time grouping the same NNV form (as it were at right angles) with the other associative forms into a separate system comparable on contextual criteria with one system in the field of English intonation grammar, where a certain intonation form marks the category of the new. This is not to imply that there is direct translatability between any one term in the English systems and any one term in the Chinese systems; it means that the systems are identified contextually (not grammatically) as being comparable in the two languages, and that this identification permits the monosystemic statement of the contextual conditions under which particular terms in the systems of the two languages may operate. Since languages differ in their grammatical reflection of contextual categories (both as regards whether or not and as regards in what priority they reflect them), there is not to be expected a one-to-one translation correspondence of grammatical terms. 3.5 Aspect The terms in the aspect system are neutral, perfective and imperfective. Neutral is as always formally unmarked; the marked terms are characterized by the presence of certain forms (of the word class "particle") 221

MODERN CHINESE GRAMMAR

directly following the verbal group. Perfective clauses have h org ua (10) (the occurrence of both together has probability 0+), of which h marks a term in a succession (whether or not the final term is indicated modally; a clause with h in neutral mood has probability 5+ of being not sentence final), gud an isolated term (conceptually a term in a series of which the second and final term is the present). Imperfective clauses have / (11), which likewise has probability 5+ of being not sentence-final in a clause of neutral mood. The grouping of these forms into a single system is suggested by their being mutually exclusive (probability 0 of occurrence in combination); the names chosen for the system and for its terms have been current for some time in the description of this, or more often some overlapping, system.9 (For marked aspect as a feature of the group see below, 4.4.) Associated with the aspect system is a system of polarity (see below, 3.6) with marked aspect, including both negative polarity and the "reflex" positive form. The marks of negative polarity and of the reflex positive form, as given in the next paragraph, are respectively bu and /; these operate, however, only in the neutral aspect, and corresponding forms in the marked aspects are as follows:

Polarity

Neutral

Negative

Reflex positive

Aspect: Neutral Perfective

V

buV maiV mai V gua buzai V (/)

/V

Vb V gU3

Imperfective

v/

/V h or yisu V (la) /V gua or yidu V (gua) zaiV(J)(12)

Constructed examples with structure VN would give:

Neutral

Negative

Reflex positive

mai caz mai-hcaz mai-gus cdz mai-] caz

bu + mai caz mai + mai caz msi + mai-gua caz buzai + mai (-/) caz

s + mai caz s + mai-h caz or yiau + mai (-Is) caz s + mai-gua caz or yiau + mai (-gua] caz zai + mai (-/) caz

222

GRAMMATICAL CATEGORIES IN MODERN CHINESE

3.6 Polarity The terms of the system of polarity are neutral, negative and interrogative, all of which have unrestricted distribution among the primary classes with the exception that an interrogative clause never occurs at X = conditional clause. The neutral term is unmarked; the negative has bu in a clause with neutral aspect, mai with perfective and buzai with imperfective (word class "verbal adverb") (12). There exists by the side of the negative term a type of neutral polarity which may be considered as a marked positive, formed by a sort of reflex from the negative; in this form to bu and mdi corresponds 5"(to mdi sometimes yiau,probability 5—), while to buzai corresponds zai (12). It may be remarked that the form / (6) operates in three similar but distinct structures: (i) word class free verb, sub-class pro-verb, always with di (3), in passive voice; (ii) word class verbal adverb, always preceding V, in the reflex positive form of neutral polarity; (iii) word class free verb, sub-class (c) (see below, 4.2), as V\ in the double-V structure of type (2a) (see below, 3.10). Examples of these are: (i) ta mai-di scaz; (ii) ta /+ mai caz\ (iii) s ta mai caz.

If the form operating at V is a compound verbal group with postpositive verb, the negative adverb in neutral aspect precedes the postpositive verb; otherwise it precedes the verbal group (with probability 0+ of following the auxiliary verb) (see below, 4.2). Interrogative polarity is characterized by the repetition of the group, or of an element within the group, operating at V, the appropriate negative adverb preceding the repeated term and the whole complex having the value of element V; if the repeated term is a verbal group with postpositive verb the negative adverb has position appropriate to such a group. In a passive clause it is always s that is repeated; the combination of interrogative polarity with ergative voice has probability 5— • For translation purposes it is useful to regard neutral polarity as a synthesis of affirmative and imperative; descriptively this arises as a system only at one point, where there is a two-term system imperative/affirmative. This arises only in a clause neutral in voice and aspect with structure V or NV; the imperative is unmarked and has the form of neutral polarity with an attributive verb (see below, 3.9), while the affirmative has the double-V structure of type (1) (see below, 3.10); examples are: (imperative) (ni) kuai + lai; (affirmative) ni lai-dd + kuai (13)

223

MODERN CHINESE GRAMMAR

3.7 Mood The three terms of the modal system are neutral, aspectival and polar mood, of which the marked terms (the two latter) operate only with value O in the sentence structure, and that only (probability 1—) in sentence-final position (with the exception that polar mood occurs with probability 2— at X = conditional clause). Marked modal clauses are characterized by a system of forms of the word class of particle in clause-final position. Aspectival mood is associated with the dimension of aspect and is marked by the forms la, lai /and na (14), occurring either with neutral aspect or in certain combinations with marked aspect, as follows: Perfective

Imperfective

Aspect:

Mood: (including the possibility of combination with the corresponding negative forms of marked aspect). Conceptually the modal forms may be thought of as the modal or subjective equivalents of marked aspect: they are contextualized as it were at one degree of abstraction, stressing the participation of the speaker rather than the totality of the situation. It is for this reason that modal la in a negative clause (neutral or perfective) usually admits the English translation equivalent 'no longer' or 'not, after all': the negative of'the situation has arisen that. . .' is 'the situation has arisen that . . . not . . .'. Aspect h and modal la are phonologically overlapping (in syllabic transcription h is h/liau, la is la/h); where there is positional ambiguity (that is, in a clause with final V) there is only one perfective term, unmarked as to aspect or mood, though the unambiguous forms la, liau exist by means of which one or the other may be specified.10 For pedagogical purposes it is useful to synthesize this modal system with the aspect system, regarding h and la as perfective (respectively non-final and final, that is occurring the one in non-sentence-final clause, the other in sentence-final clauses, with probability 5+), /+ /—na as imperfective (na in sentence-final clauses only) and gua, laij as a separate two-term system characterized as: gua, 'past in present', laij 'present in past'. For comparison with Early New Chinese, as exemplified in the language of the Chinese translation of the "Secret History of the Mongols", one can set up for the latter the following system: (i) 224

GRAMMATICAL CATEGORIES IN MODERN CHINESE

perfective: le (non-final/final), (ii) imperfective: Jo, (iii) tense (nonaspectival): la/yieulai/yieu/yie (15); with this one could compare Modern Pekingese (i) b (/la), (ii) / (/no), (iii) gua/laij. For comparison with Cantonese, one can set up for the latter: (i) aspect: zo/guo (perfective),gan (imperfective), (ii) mood: la/lak/lok, etc. (16); and compare with these Modern Pekingese (i) \9/gu9,J, (ii) la/lai], na and other modal forms. Polar mood is associated with the dimension of polarity and marked by the forms, bu, maiyiau, ma, ni, ba, rja (17). These combine with the terms of the polarity system as follows: neutral polarity may have any of them, negative polarity only ma, ni, ba and rja, and interrogative only ni and rja. Since maiyiau is the regular clause-final form of the negative pro-verb msi/msiyiau, which is lexically identifiable11 with the negative verbal adverb mai, bu and maiyidu, may be regarded as corresponding to negative polarity, ma, ni, ba and rja to interrogative; but in fact all (except ni and rja) exclude interrogative nominal or adverbial forms from the clause (hma, zamayiarj (18), etc.). One then has the combinations: Negative

Interrogative

Polarity:

bu/msi

(V7w/watV) = V

Mooc

Du/m9iyi3u

mam

oa nc

A clause with neutral or negative polarity is marked as a 'yes-°r-n°' question, with varying degrees of modality, by ma, ni, ba and rja, and (with marked aspect) bu and maiyisu; questions already so marked, by interrogative polarity or other interrogative forms, may have modal ni or WThe classification of modality through contextually specified examples (with if necessary class names such as ma: "indignant affirmation") is useful pedagogically; and the combination of mood (with ma, ni and go) and conditional clause (being the low probability occurrence of marked mood at X in the sentence structure) provides a formal link for such classification. Aspectival and polar mood may combine in one clause, the aspectival particle preceding the polar (but na, ni have probability 0 of combination); na and ni are phonologically overlapping (as na/na and ni/na], such that the form na occurs as a synthesis of aspectival and polar mood. For example, jarj-J mantsu na, wua bu + gan-durj na-ga-hua (19) 'I daren't disturb the fire while I'm steaming bread'

225

MODERN CHINESE GRAMMAR All words marking mood (like those marking aspect and the repeated term in interrogative polarity — bu and mai in negative polarity have a stress/unstress system) are unstressed. 3.8 Adverbial structures In addition to the basic clause structures, which are combinations of one element V with one or more element N, three subsidiary forms of clause structure may be recognized. The first of these has the additional element A, the primary class corresponding to which is the adverbial group. The element A occurs in the subsidiary structures AV(~) (including NAV(~)), ANV(~) and (~)VA, including combinations of these with more than one A (e.g. ANVA). In immediately pre-verbal position (AV, NAV, etc.) there is a two-term system with even probability at A of an endocentric and exocentric adverbial group (see below, 4.4); elsewhere the probabilities are uneven, such that at A preceding preverbal N the endocentric group has probability 2+ while at A in final position the exocentric group has probability 1-. The occurrence of structure AN~ or NA~ where A = the endocentric group is determined by the contextual system of given/new, the given as always preceding (e.g. mirjtian wua . . . (20) answers 'what are you doing to-morrow?', u>ud migtian . . . answers 'when are you . . . ?'); the occurrence of structure (~)AV or (~)VA where A = the exocentric group, while partially determined in the same way, is further restricted in that certain forms of the exocentric adverbial group (identified by the occurrence of certain forms operating at v3 (see below, 4.4), e.g. gai (21)) have probability 5+ (1 — in ergative voice) of following a verbal group if the latter is simple. For example, ksyi ba-biyin-gu di zai-bissyin-di+yi-hi (22) 'you can classify the nasals with the plosives' Only the adverbial group operates at A; some of the adverbial word classes do not operate in the group (and therefore never occur at A) but independently in the clause structure, either at [A] (see 3.9) or as marking certain clause classes. 3.9 Attributive structures Attributive structures are subsidiary clause structures incorporating the element [ ]. This is in fact a single element, but since the operative forms are classes of the group, and may be described in relation to the system of 226

GRAMMATICAL CATEGORIES IN MODERN CHINESE group classes set up for the elements V, N and A, it is helpful to state attributive possibilities in relation to elements [V], [N] and [A], noting the occurrence of classes of the respective primary groups in the various attributive structures. The following attributive structures may then be recognized:

[V]V [V]V

[N]N

[A]V [A]N

In [V]V, only the intransitive verbal group can operate at [V], and then only a simple group. For pedagogical purposes it is useful to set up a restricted system in which the form [V]V is opposed to a double-V structure of type (1) (see below, 3.10), the former being imperative, the latter affirmative (see above, 3.7). In [V]N, attribution may or may not be marked by di (3) preceding N. If it is, the attributive form is considered to be an attributive adjectival clause (see above, 3.3) with value at two levels, at X in the sentence and at [ ] (or [V]) in the clause; this form never has passive voice. Without di, the attributive form is restricted as in [V]V, to the simple intransitive verbal group. Since such a group when attributive to N may still have di, a separate two-term system is formed by [V]N/[V] diN where [V] = simple intransitive verbal group, of which the latter is the marked attributive form.12 In [N]N, only the major nominal group can operate at [N]. N may or may not be preceded by di', if it is, [N] may be simple or the compound nominal group; the question of whether or not a compound nominal group may operate at [N] without di depends on the analysis of a clause such as j9-g9-fdn(+) sarjyin da (24) 'this man has a loud voice'. If j9-gd-fsn is attributive (and the fact that in some such clauses13 an alternative form without di is not possible does not preclude the analysis of this as [N]N), then [N] includes the possibility of a compound group; alternatively, however, all such clauses could be described as NNV, this being the only instance of V = intransitive group in NNV with neutral voice. The same system of presence or absence of di operates as at [V]N. The forms operating at [A], unlike those at [V] and [N], are not forms of the corresponding (here adverbial) group but secondary classes of words of the adverb class, the verbal adverb in [A]V, the nominal adverb in [A]N (see below, 4.5). Attributive forms at [A] have probability only 0+ (as contrasted with even probability for [V] and [N]) of occurring internally to a group, immediately preceding element 0 (classes free verb and free noun; see below, 4.2,3).

227

MODERN CHINESE GRAMMAR

3.10 Double- V structures Except in interrogative polarity, where the complex of the verbal group plus negative plus repeated verbal group has the value of a single element V, V excludes more than one verbal group. There are, however, two distinct clause structures containing two elements V. First, structure (/) W, with the form (word class postpositive verb) da (26) occurring between the two V. V\ is transitive (probability 5+; it may be intransitive with certain forms at V2, e.g. lag -da lihai (27) 'terribly cold'); V2 is simple and intransitive. The whole form is affirmative (as opposed to imperative with [V]; see above, 3.9), but may have negative and interrogative polarity (i.e. is not 'positive'); the negative adverb precedes V2 and in interrogative polarity only V2 is repeated. The form has probability 1 — of being neutral in aspect and voice. Alternatively the form could be analysed (i) as two distinct clauses, the first subordinate (cf. (2b) below) — but a single clause structure appears preferable in view of the affirmative/imperative system; or (ii) as a single compound verbal group (see below, 4.2) — but it does not operate with the negative/ reflex positive system of the latter, having (not v: 03/0&w3/0da3 but) negative V -dd bu + V. Second, structure (~)V(N)V(~), with probability |+ that N is present. Here it is convenient to recognize (2a) and (2b). In (2a) V2 is unrestricted, but there is a sub-class of free verbs (sub-class (c); see below, 4.2) occurring in the verbal group operating at Vj. In (2b) both Vj and V2 are unrestricted, and Vj is always directly followed by dd (since VNtb is excluded, where V\ is followed by N it must be repeated, and this is considered to mark a separate clause: e.g. ta sua hua/ sud -dd wudmdn du+bu+durj (28) 'he spoke in such a way that none of us understood'). The form could alternatively be analysed as two distinct clauses, the first subordinate (a special instance of either conditional or adjectival clause, cf. (1) above) - but this would involve either the assignment of da to a different word class outside the verbal group or the recognition of a postpositive verb as marking a subordinate clause. (2a) and (2b) may combine in one clause, giving in fact a treble-V structure: e.g. ta sud -dd mdiyidu fan dug (29) 'he spoke in such a way that no one understood'.

228

GRAMMATICAL CATEGORIES IN MODERN CHINESE

4 The group 4.1 Structure of the group The structure of the group will be discussed in terms of basic and subsidiary structures, the elements set up for both of which are symbolized as o and x. These elements will be further specified in the description of each primary class of group, element x as 1,2 or 3, by opposition to which element o will appear as 0. The systems operating at these elements are systems of word classes: the primary classes operating at o, x respectively are those of free word and bound word, but it is more convenient to set up primary classes according to the specific structures of the three group classes and in this way to recognize (i) the verb, operating at 0,1, 2 in the verbal group and at v3 in the adverbial group; (ii) the noun, operating at 0, 1, 2 in the nominal group and at n3 in the adverbial group; and (iii) the adverb, operating at 0 in the adverbial group (as well as in clause structures independently of the group). Symbols for these three classes will be v, n and a. Of the generalized group structures — (basic) o, xo, ox, xox, xxo; (subsidiary) x, xx — the verbal and the adverbial group admit all except xxo and xx, the nominal group all except ox and xox. The classes operating in the structure of the verbal and nominal groups are respectively words of the classes verb and noun; the adverbial group, however, may (probability j— if simple, 5+ if compound) contain no word of the adverb class; in this case one of the elements at least is of the noun class, and in any case more than one word of adverb class is excluded. Exemplifications of the word classes operating in the group structures, with specimen translations, are given at the end of each section below.

4.2 The verbal group The structure of the verbal group is: (basic) 0,10,02 or 102; (subsidiary) 1. Both transitive and intransitive verbal group may have any basic structure; the subsidiary structure, however, never occurs in a verbal group at V preceding N, so that a verbal group with subsidiary structure in neutral as to transitivity. The secondary word class operating at 0 is that of the free verb. A free verb may be classified as transitive or intransitive according to the transitivity of the verbal group in which it operates at 0. Three sub-classes of free verb may be distinguished:

229

MODERN CHINESE GRAMMAR

(a) The pro-verb is distinguished by the fact that a group in which it operates at 0 has probability 1— of being a simple group. A group with pro-verb at 0 is intransitive but has probability only 0+ of occurring at V in a clause with marked voice. (b) This sub-class is distinguished by the fact that a verbal group in which it operates at 0 combines with following N, the combination occurring at V in the clause structure (cf. above, 3.4). (c) This sub-class operates in a group at V\ in the double-V structure of type (2a) (cf. above, 3.10); a verbal group with this sub-class at 0 is transitive. The class operating at vl is the auxiliary verb; a subsidiary structure of the verbal group has vl only. In a verbal group with structure 10 (~) an attributive form may, and in ergative voice the form ba (9) plus N always does, occur between vl and vO. The class operating at v2 is the postpositive verb. This may be described in two sub-classes, with further subdivision of the first. Phonologically sub-class (1) has probability 1— of being unstressed, sub-class (2) probability 1— of being stressed. Sub-class (1) is limited to nine words classifiable further into (la) and (Ib), with seven and two members respectively; (a) and (b) may combine in the order ab, the combination operating at v2j and (b) may occur following a nominal or adverbial group at post-verbal N or A (alternatively (b) here could be assigned to a distinct word class, with lexical identification). Sub-class (2) is much more extensive and has no combinatory possibilities. The postpositive verb, or rather a number of exemplifications of this class, has sometimes been regarded as marking perfective aspect;14 and indeed there is a probability only 0+ that an imperfective clause will have at V a verbal group containing postpositive verb. But it may be more helpful to regard the postpositive verb contextually as giving extension to the free verb, sub-class (1) giving extension in a direction, sub-class (2) extension to a goal. This is reflected in the high degree of lexical identification between the exemplifications of this class and those of the class of free verb; a group such d&jian-duan (30) (structure v02) 'cut' is on the one hand a term in a system in which jian may be replaced by other free verbs (gd-duan, da-duan, da-duan, etc. (30)) and on the other hand opposed both to jian and to duan as distinct members of the free verb class in simple group structure. The verbal group with postpositive verb shows a distinct system of negative polarity with a reflex positive form: this operates only in neutral aspect. In this system the negative adverb bu (12) occurs between the free 230

GRAMMATICAK verb and the postpositive verb: jian-bu-duan. If bu is replaced by da (26) the result is a marked positive foTm:jian-da+duan. In marked aspect the negative adverb occurs in its usual place preceding the verbal group, and there is no da form. The formal relation between the postpositive verb system and that of aspect in the clause is displayed in the fact that in neutral aspect only the negative adverb has what might be regarded as attributive position not to the free verb but to the postpositive verb. The comparable systems in Cantonese and Early New Chinese each show four terms, having in addition a further "reflex negative" form, with both negative adverb and da (dak, da}; in Cantonese the negative precedes, in Early New Chinese it follows, the free verb (Cantonese m + haag-dak+yap (31),Early New Chinese hig+bu+da+fu (31)). Free verb, sub-class (a): yiau 'there is' zai 'is at' Free verb, sub-class (b) jiau 'call' huan 'give back' jid 'lend' (notjid 'borrow') gaum 'tell' wun 'ask' (33) Free verb, sub-class (c): sud 'say' jiau 'tell (order)' wun 'ask (inquire)' siag 'think' Auxiliary verb: nsq 'can' hu9\ 'can, may, know how to' Postpositive verb, sub-class (1): sag 'up' jin 'in' huai 'back' d 'up' (37) lai 'come' Postpositive verb, sub-class (2): wuan (zuswuan 'finish') faggug (gai&ggug 'get built') dau (zaudau 'arrive') ju (tiaju 'stick fast') dau (kaidau 'reverse')

maiyisu 'there isn't' hausiag 'is like' (32) gai 'give' sug 'present, send' jau 'give (money) in chan§

gausu 'tell (inform)' dg 'ask (request)' jdau 'know' siwuag 'hope' (34)15 kdyi 'can, may' yiggai 'ought to' yiau 'want to' (36) sia 'down' cu 'out' gus 'past' du 'go' (38)

car) (biancag 'turn into') jin (yiurjjin 'use up') tut) (daturj 'force through') dug (kagdug 'carry') guan (tigguan 'am used to listening') (39)16 231

MODERN CHINESE GRAMMAR 4.3 The nominal group The structure of the nominal group is: (basic) 0, 10, 20 or 120; (subsidiary) 1 or 12. The secondary class system of major and minor corresponds to the basic and subsidiary system in the structure: a nominal group with basic structure is major, one with subsidiary structure minor. The major nominal group may operate at N in all clause structures except following V = intransitive verbal group in active voice; the minor nominal group is unrestricted, but in an active clause with intransitive verbal group at V the combination of such a verbal group with a following minor nominal group operates at the simple element V (see above, 3.1). The secondary word class operating at 0 is that of the free noun. Three sub-classes may be distinguished according to their structural possibilities: (a) The pronoun is distinguished by the fact that a group in which it operates at 0 has probability 1— of being a simple group. This sub-class is limited to personal pronouns and to proper names. (b) This small sub-class is distinguished by the fact that a nominal group in which it operates at 0 has probability 1- of excluding n2 (the auxiliary noun; see below). (c) This sub-class is distinguished by the fact that a nominal group in which it operates at 0 exhibits a two-term number system of neutral/plural, the latter marked by the form (of the word class of the particle) man (40). This sub-class is largely limited to nouns of human reference. The class operating at nl is the determinate noun. This class may be described in two sub-classes, (1) and (2), combinable in the order 12, the combination operating at the simple element nl. The class operating at n2 is the auxiliary noun; this class may likewise be described in two sub-classes (1) and (2). Sub-class (1) includes an unmarked member ga (41), together with those words of the class which can be assigned to a single system, each of the terms of which is opposed to ga in a contextual system of specialized/generalized (for example yi-j-gdu 'a dog' (or 'one dog') (specialized context), yi-ga-gau 'a dog' (generalized context) (42)). Sub-class (2) includes the remainder, which have probability 5— of occurring in a nominal group without nl and which may be further sub-divided according to probability of occurrence in a minor nominal group.17

232

GRAMMATICAL CATEGORIES IN MODERN CHINESE Free noun, sub-class (a): wuaman 'we/us' wug 'I/me' juggua 'China' (43) hi 'who?' Free noun, sub-class (b): tian 'day' yia 'night' sag 'province' (44) nian 'year' Free noun, sub-class (c): ran 'man' siussdg 'student' yigsiug 'hero' haiz 'child' (45) Determinate noun, sub-class (1): na 'that' jd 'this' na 'which'? mai 'every' (46) Determinate noun, sub-class (2): Hag 'two'18 yi 'one' ji 'a number of/what ban 'half (47) number of?' Auxiliary noun, sub-class (1): ga (yi- ga- ran 'a man') / (yi- J/ga-gw 'a dog') zua (yi-zua/ga-ciau 'a dig (yi-dig/ga-mauz 'a hat') bridge') (48) Auxiliary noun, sub-class (2): bai 'hundred' dan 'thousand' bai 'cup of wuan 'ten thousand' suarj 'a pair of jin 'pound of Jug 'kind of x 'time' (liag-x 'twice') (49)

4.4 The adverbial group The structure of the adverbial group is: (basic) 0, 10, 02 or 102; (subsidiary) 2; but since the system of classes operating at these elements includes classes of the verb and noun, these may be better stated as specific structures with reference to the operative classes. If 0 is replaced by a/N, 1 by v3 and 2 by n3, the basic structure is then: a, v3N, v3a, Nn3, an3, v3Nn3 or v3an3; the subsidiary — n3. The distinction into the secondary classes of endocentric and exocentric corresponds to the occurrence of a or N: endocentric group structures are those which include element a, exocentric those which exclude a, a being the element of which the operative class is a class of adverb. The distribution of the classes of endocentric and exocentric at A in the clause structure has been discussed above (see 3.8.). The class operating at a is the free adverb, the only class of adverb which operates in group structure, in which it always occurs at the element generalized as 0. Where 0 is represented specifically by N, at this

233

MODERN CHINESE GRAMMAR

element may operate any nominal group, a minor nominal group having probability |— here; in these instances the nominal group is operative in group structure and not in clause structure. The class operating at 1 (specifically v3) is the prepositive verb. Two sub-classes may be distinguished within this class: (a) With this sub-class the adverbial group has probability |+ of being marked as to aspect, most of the members of the sub-class admitting imperfective only; aspect is marked by the addition of/, h (11,10) to the prepositive verb. (b) This sub-class is distinguished as alone operating in the adverbial group at A in the post-verbal place in the clause structure. The class operating at 2 (specifically n3) is the postpositive noun, distinguished into two sub-classes (1) and (2). Sub-class (1) consists of simple words only and has probability only 0+ of occurring in the simple adverbial group. Sub-class (2) consists of compound words and occurs with even probability in the simple and compound adverbial group. Prepositive verb: gan 'with' xurj 'from, through' jiggud 'via' bai 'by (agentive)' (50) Prepositive verb, sub-class (a): wuai-j 'round (rest)' rau-j 'round (motion)' gsn-j 'following' wudi-h 'for' (51) Prepositive verb, sub-class (b): zai 'at' dau 'to' gdi 'for' (52)19 Postpositive noun, sub-class (1): say 'on' hsu 'behind' (54) Postpositive noun, sub-class (2): sarjtau 'on' hsumian 'behind' darjjui) 'among' yiwuai 'apart from' (55)

4.5 Word classes not in group structure The word classes operating at elements in the group structure may be summarized as follows: Element: Group: verbal nominal adverbial

0

1

2

free v. free n. free a. (or nominal gp)

auxiliary v. determinate n. prepositive v. (=v3)

postpositive v. auxiliary n. postpositive n. (=n3)

234

GRAMMATICAL CATEGORIES IN MODERN CHINESE There remain certain classes of word which operate independently of the group, occurring directly in the clause, to which they may have classificatory function. These are the secondary adverb classes of conjunctive, conditional, verbal and nominal adverb and particle. The conjunctive adverb marks the class of conjunctive clause (see above, 3.2), and distinguishes two sub-classes: (1) compound, with probability 1— of preceding pre-verbal N, and (2) simple, with probability 1— of following pre-verbal N. The conditional adverb marks the class of conditional clause (see above, 3.3), and distinguishes likewise two sub-classes: (1) preverbal, and (2) final, according to position in the clause. The verbal adverb occurs at [A] in [A]V, the nominal adverb at [A] in [A]N (see above, 3.9). There is some lexical identification among these classes, sometimes with a concomitant system of stress/unstress; for example, jiau and xai (56) (stressed: verbal adverb; unstressed: conjunctive adverb); suayi (57) (stressed: conjunctive adverb; unstressed: conditional adverb); and also with the free adverb, for example zurjs (58) (stressed: free adverb; unstressed: conditional adverb). The particle occurs as marking the classes of the aspect and modal systems (see above, 3.5, 7), the adjectival clause (3.3) and one form of attribution (3.9), as well as plurality in the nominal group with a free noun of sub-class (3) atO. Conjunctive adverb, sub-class (1): susyi 'so' dans 'but' bigcia 'moreover' bufan 'otherwise' (59) Conjunctive adverb, sub-class (2): jiau (after 'if/when', etc.) xai (after 'unless/until', etc.) ka 'still, though' dau 'yet' (60) Conditional adverb, sub-class (1): fugus 'if jiran 'since' suairan 'although' jid us ' even if (61) Conditional adverb, sub-class (2) dishdu 'when' yihau 'after' dihua 'if (62) Verbal adverb: han 'very' du 'all' faicaqdi 'extremely' bu, s (see 3.6) lai 'in order to' bdi (English passive verb) (63) Nominal adverb: lig 'other' di (ordinal numeral) suayiau 'all' gs 'every' (64)

235

MODERN CHINESE GRAMMAR

5 Conclusion 5.1 Word and character The choice of the term 'word' to designate one rather than any other of the units of the grammatical description reflects the fact that it is this unit rather than any other which it seems should be the unit of lexical description, including lexicography. The unit which is institutionalized in Chinese, both as the basis of traditional lexicography and as that to which a speaker refers in general conversation about the language,20 is the unit here termed 'character'. In Modern Chinese the character is merely the raw material of syntactical classes; it cannot itself be organized into any system of classes operating at places in group structure. It admits only classification into "free" and "bound"; an empirical form of this classification such as that adopted by Chao and Yang ('B means always bound, but F means sometimes free'21) has pedagogical advantages, but it is impossible to proceed directly from the class system of free/bound to the elements of group structure, since in the establishment of the classes operating at these elements the question of whether the form is simple or compound, and of the classification as free or bound of the constituent characters of the compound form, is irrelevant; at most it provides a criterion for certain sub-classification (compare for example the postpositive noun, 4.4 above). For example, zua-cdgguq 'get done', kan-jian 'see' and zujcilai 'get organized' (65) are all analysed as v: 02. For this same reason, even though the character is what is talked about, the larger unit here called "word" is what is talked with. The word can be entered in a dictionary, classified syntactically and its exemplifications collocated and given translation meaning. An entry would read: jian 1. free verb To see (etc.) (with citations) (a) at Vjin double-V structure (with citations) 2. postpositive verb (with citations). A compound word in which the character jian occurs in initial position would constitute a separate lemma, for example: jianjia

free noun

Opinion (etc.) (66).

At the same time it is noticeable that by the recognition of a unit "group" intermediate between the clause and the word a reduction has been effected in the word structure, such that many forms which are often regarded as "compound words"22 appear here as compound groups, made up of simple words. For example, liaij-ga-fan 'two people',

236

GRAMMATICAL CATEGORIES IN MODERN CHINESE zai-daq-sar) 'on the wall', yiau-zua-wuan 'want to get it finished' (67) are all three-word structures of simple words. The word does of course still exhibit compound structure, as exemplified in the structurally identical sf-ga-biqfen 'twelve patients', rauj-fcujz-hdumian 'round the back of the house', kayi- jzau-feggut) 'can get it made' (68). The unit of word structure is the character and this is the lower limit of grammatical description. 5.2 Word structure The term "character" has been chosen to reflect the identity in the Chinese word z (69), which is the name both of the graphic unit in the script and of the linguistic unit of which this is the written symbol.23 If the character is to be recognized as a grammatical unit, the word must exhibit a structure to which character classes bear the same relation as do word classes to group structure, etc. The problem of word structure is, however, complex and demands separate treatment; it is proposed here merely to mention the tendencies in the distribution of simple and compound structure among the various word classes and to suggest one possible line of systematization of the compound forms. The probabilities of simple structure in each of the word classes set up in the group analysis are as follows: Verb:

Free, \+ Auxiliary, 2 Postpositive, 1- (but note compound structure ab with sub-class (1)) Prepositive, |+ Noun: Free, \— Determinate, 1 (except compound numerals; and note compound structures with sub-classes (a), (b)) Auxiliary, 1 Postpositive, \ Adverb: Free, \— Verbal and nominal, 5+ Conjunctive, 2 Conditional, 0+

It is with the class of free noun that one may most easily seek to establish a relation between the elements represented by the component characters; that is, a relation, other than precedence, that is not reciprocal. If we recognize elements symbolized S, G (from the words 'specific' and 'general') to account for structural relations, a compound noun of

237

MODERN CHINESE GRAMMAR

two characters then has probability |+ of exhibiting structure SG. If S and G are considered as relational elements, the operative forms will not be classes of the character, and no element of structure will be exhibited in the simple word; but it may be possible to establish these elements as absolute (i.e. linked to class systems) by the application of combined lexical and contextual criteria, since contextually the SG compound is specific and enters into a contextually defined lexical series one of whose terms is general. If this is admitted then the character operating at S may be classed "specific", while that operating at G in the SG compound, as well as in the general term, may be classed "general". This may be exemplified in the following series: SG (contextually specific) sigua 'watermelon' huaggua 'cucumber' zgua 'seed melon' mugua 'quince' mankau 'doorway' luksu 'street corner' gagksu 'harbour' haksu 'river mouth'

•

siaumai 'wheat' maiz 'corn' yianmai 'oats' daumai 'buckwheat'

G (contextually general) gua 'cucurbitaceae'

kdu 'opening'

damai 'barley'

(70)

The use of either the specific or the general term to refer to an identical object in a context of situation represents different linguistic orderings of the situation; thus, for example, a bus may be variously mentioned as (G) &, (SG) dca or (with a further specifying form, itself a compound verb) guggugdfe (71) according to what is to be excluded: the Chinese for 'Hurry up and get on the bus', if the bus is in front of you, might be ni gankuai+sag fa gal (72). It is not suggested that this analysis is exhaustive (the fact that the same character may be assigned to both classes is not itself objectionable, such lexical identification being regular among the word classes also); many compound nominal forms cannot be analysed in this way, including some which are contextually general; for example, the general term in the series faduqji 'engine', faiji 'aeroplane', dahuaji 'cigarette lighter', 238

GRAMMATICAL CATEGORIES IN MODERN CHINESE Jbuji 'loom' is jid 'machine' (73). The general term does, however, tend either to be simple (and the general character therefore to be free) or to take the form of a suffixed compound (e.g. maiz 'corn', mutau 'wood' (74)). SufHxation itself may be regarded as an extension of SG structure in which the form at G, the suffix, is merely a mark of generality. 5.3 Interrelation of categories Finally it may be considered whether any general conclusions may be drawn concerning the interrelations of the categories, such as to characterize the grammatical structure of the language as a whole; and specifically whether one may set up any relation which obtains among the categories of the various units. No identification has been made across the units except that implied in the classification of the group as verbal, nominal and adverbial and of the word as verb, noun and adverb: what is implied here is actually no more than that each class of word operates in the structure of only one class of group (hence the choice of specific, i.e. classified, group structures as a basis for the establishment of primary word classes), while in general all primary group classes operate in the structure of all classes of clause. The "group", as a concept not necessarily identical but certainly related, has indeed been regarded as a distinctive feature of Chinese (or of a language of a type exemplified by Chinese); and the simplification which the introduction of the group as a middle term in the unit system brings to the description reflects the relatively more 'syntactical' (above the word) form of analysis demanded by a language from which -word paradigms are absent. It might reasonably be asked whether from the syntagmatic and paradigmatic relations of the group might be adduced some generalizations valid for all the units. As regards the group structures, the opposition of a nodal element 0 to as it were peripheral elements which are in some sense (a sense susceptible of contextual definition) qualifiers seems to be reflected at other units, as does the irreversibility of the relation found at some points in group structure of what might be termed "specifier" and "specified" (for example in the given/new system in the clause). But it is doubtful whether any relation so abstracted is left with any more significance than one of simple precedence, "which is in fact an abstraction to zero since structure is by definition syntagmatic. Systemically the presence of free forms among the word classes operating in group structure is paralleled by the presence of a neutral term in a number of the class systems throughout, sometimes with a three-term system which 239

MODERN CHINESE GRAMMAR may be analytically stated as neutral, negative, marked positive. The presence of a neutral term in system and the specifier/specified relation in structure might be synthetized at the contextual level into a feature which might be designated negatively as absence of concord, or economy in the grammatical reflection of contextual categories: there exists at many points in the paradigm a general term, and this alone is required by the structure. In information theory terms, Chinese is characterized by a particularly marked contrast between the high degree of entropy in the grammatical structure and the low degree of entropy in the phonological structure. But it should be stressed that, interesting as such attempts at synthesis are, even in typological studies they must follow descriptive comparison; in a particular description they may contribute to the statement of contextual-grammatical relations but do not constitute a logic either of reality or of thought. No other language reflects reality - that is, orders the relations between context and grammar — in exactly the same way; every other language is susceptible of equally thorough systematization. If we do give a name to some generalized relations, what we are doing is to give a name to the Chinese-ness of Chinese. There are two extremes we may avoid, claiming neither universality for the features described (still less for the categories) nor exceptionality (as opposed to uniqueness) for the language. The Chinese language works, and the task of the descriptive linguist is to show how it works.

6 Summary of categories, examples, and table of Chinese characters 6.1 Summary of categories I.

Units: Sentence / clause / group / word / character. II. Elements: (substantive: unmarked; attributive: [ ]) (1) Sentence structure (elements O, X) O or O/X (. . .n)O, with possibilities at any place of O<X>, X<X>. (2) Clause structure (elements V, N, A [V], [N], [A]) (Basic) V N NV VN NVN NNV VNN. (Subsidiary) (i) to these may be added A, in the following positions as exemplified from NVN structure: ANVN NAVN NVNA 240

GRAMMATICAL CATEGORIES IN MODERN CHINESE

(ii) to V or N may be added [ ], with possibilities: [V]V [A]V [V]N [N]N [A]N (iii) a second V may be present, with possibilities: (~)W (~)V(N)V(~) (3) Group structure (elements o, x: vO, nO, aO, vl/2/3, nl/2/3) (Basic) o xo ox xox xxo } , , r ,, ,0 i - j - \ r by} group r classes as follows: (Subsidiary) x xx J o: xo: ox: xox: xxo: x: xx:

verbal vO vlvO vOv2 vlvOv2

nominal nO nlnO,n2nO

nln2nO nl nln2

adverbial a v3N,v3a Nn3, an3 v3Nn3,v3an3 n3

Attributive element in verbal and nominal group precedes group or elements o: [ ] xo or x[ ]o, etc. (4) Word structure (elements S, G) GSG III. Classes: (free: unmarked; bound: —. The elements at which the classes operate are shown to the right) (1) Clause classes (Primary) Free/subordinateO/X (Secondary) Disjunctive/conjunctive O/O Conditional-XadjectivalX / X, N (in clause) Voice: Neutral/passive/ergative all O or X Aspect: Neutral/perfective/ imperfective all O or X Polarity: Neutral/negative/interrogative all O or X O or X /O / O Mood: Neutral/aspectival/polar (2) Group classes (Primary) Verbal/nominal/adverbial V/ N /A (Secondary) Verbal: Transitive/intransitive V / V or [V] Nominal: Major/minor N or [N] / N Adverbial: Endocentric/exocentric A / A 241

MODERN CHINESE GRAMMAR

(3) Word classes (Primary) Verb/noun/adverb v / n / a (Secondary) Verb: Free/auxiliary-/-postpositive/prepositivev:0/l 72/3 Noun:Free/determinate-/-auxiliary-/-postpositive n:0/ 17 2 / 3 Adverb: Free/ conjunctive-/-conditional-/verbal-/ nominal-/-particle a / - / - / [A] / [A] / 6.2 Examples Each example is one sentence. Clause division is marked by /, group division by a space, word division by —. Attributive forms are linked by +, clause particles by a space followed by —, group particles by —. Sentence and clause structure, and clause classes of the direct secondary dimension, are indicated at the head of each example; small a indicates a class of adverb not operating in group structure. Other categories are to be found illustrated in the examples as follows, with underlined figures indicating the possibility of an alternative form of statement: Marked voice: 7 8 13 14 Marked aspect: 2 5 9 1 0 1 2 1 3 Marked polarity: 261013 15 21 Marked mood: 2 6 9 1 0 1 2 1 3

W structure: (i) 13 22 (ii) 291516 21 22 (VN) = V : 20

Compound verbal group: 1 3 4 5 6 7 8 10 12 13 15 16 19 21 22 Compound nominal group: 1 3 7 11 13 14 15 16 19 20 22 Compound adverbial group: 2 4 7 9 12 16 18 Verbal group, intransitive: 6 9 15 16 22 Nominal group, minor: 1 3 16 17 19 22 Adverbial group, exocentric: 2 4 7 9 12 17 18 Verb, auxiliary: 146121317 Verb, postpositive: 3 4 5 6 7.8 10 14 15 19 21 Verb, prepositive: 2 7 9 1 2 1 7 Noun, determinate: 1 2 3 7 11 12 13 14 16 17 19 20 22 Noun, auxiliary: 3 7 11 12 13 14 16 17 19 20 22 Noun, postpositive: 9 17 18 1. OXO (disjunctive, conditional, conjunctive) VN/aNV/aV yiau-juan jarj/jyiau liayban pirjhdg/jidu kdyi 'If you want to transfer the accounts it's all right provided the two halves balance.' 242

GRAMMATICAL CATEGORIES IN MODERN CHINESE 2. XO (conditional, conjunctive) aNA[A]V/NaVNV zugs tags-Hag-tian bu+lai /wudjisu yiwudi ta s —la 'Everytime he doesn't come for a couple of days I think he's dead.' 3. OO (disjunctive, conjunctive) NVN/a[A]V[A]N tafdn-wuaijidu-bu / dans hai+yiau sis+cufu 'He divides them into nine classes, but there are still some overlaps.' 4. O<X> (disjunctive, adjectival) AVN na / ta zgi+fen-wuai jar) —di / yin lai-zua biaujun 'He takes as standard the sounds he himself considers correct.' 5. XOO (adjective, disjunctive, disjunctive) (N[A]V(di)) = /NVN/VN dicidu zgi+suai-cudu —di / bian-cag-h yiualiag / wuai-j ta 'What the earth itself had thrown offbecame the moon circling round it.' 6. OO (disjunctive, disjunctive) NAV / VA ni jayiag pag-siadu / yiau-zsu-bu+dug —la 'If you go on getting fat like that you won't be able to move.' 7. O (disjunctive) N (s) NAV (di) jd-gs-tu swu3 ti-ta zau-culai—di 'I worked out this chart for him.' 8. O (disjunctive) (ba) NV ba-saqz jid-kai 'Untie the string.' 9. OO (disjunctive, conjunctive) NV / aVNA or NV / aVNVN tian hdi-h / hai fag wua zai-waitau —na 'You even leave me outside when it's dark.' 10. O (disjunctive) [N] NaV ta+jia hai ban-zau-h —msiyiau? 'Has he moved out yet?' 11. O (disjunctive) N[A]VN wua zai+jia yi-bal-juggua+su —la Til borrow another Chinese book.' 12. O (disjunctive) A[A]V[N]N yiug-na-ga-ssu lai+gan-j ni-di+ldsu 'Press your ear with that hand.' 13. O (disjunctive) [A]V(da)[A]V bu+yiau ba-nai-ga-sumuz kan —da tai+s —la 'Don't take those figures too literally.' 14. O (disjunctive) N(s)V(di) jeliag-jian-sdg sdi-ju —di 'These two things conflict.' 15. O<X> (disjunctive, adjectival) AVNV dagai s / husi-bu+lai —di/ cagfan dua 'The chances are that they won't come back.' 16. O (disjunctive) NVN[A]V

243

MODERN CHINESE GRAMMAR xugdan ranjiuada ja-ga han+sanmi 'In the past they thought this was very mysterious.' 17. O (disjunctive) NVA ja-ga-sumu yiggai-sia zai— ja-ga-litau 'This figure should be written inside here.' 18. XO (adjectival, conjunctive) AV(di) / N[A][V]VN dag-sag sua —di /fan du+kuai+cag ma 'The whitewash on the wall is even turning into dust.' 19. O (disjunctive) NV ja-ga rag-diau-la —ba 'Throw this away, shall I?' 20. O (disjunctive) NV[A]N wua gausu-niman ligwuai+yi-ga-dau 'I'll tell you another wrinkle.' 21. O (disjunctive) N[A]W ta bu+guan bu-da+svag-bu-bu+sag 'He doesn't mind whether he can fill up the gaps or not.' 22. OOOO (disjunctive throughout) NV / NVNV / NV / N[A]V (da)VN big dg / ni juag big jug / big jug / ni zai +juag —da jug yi-dian 'If you are slightly ill you pretend it's serious; and if you're seriously ill you pretend it's still more serious.'

6.3 Table of Chinese characters

(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11)) (12)) (13)) (14)) (15)) (16) (17))

244

GRAMMATICAL CATEGORIES IN MODERN CHINESE

(18)) (19)) (20)) (21)) (22)) (23)) (24)) (25)) (26)) (27)) (28)_ (29)) (30)) (31)) (32)) (33)) (34)) (35)) (36)) (37) (38)) (39)) (40)) (41)) (42)) (43)) (44)) (45)) (46)_ (477) (48)) (49)) (50)) (51)) (52)) (53)) (54)) (55)) (56)) (57)) (58)) 245

MODERN CHINESE GRAMMAR

(59)) (60)) (61)) (62)) (63)) (64)) (65))0 (66)) (67)) (68)) (69)) (70)) (71)) (72)) '(73)) (74)) Examples:

(1) (2) (3) (4) (5) (6) (7) (8) (9)

(10)) (11)_ (12)_ (13)_ (14)_ (15)_ (16)_ (17)_ (18)_ (19)_ (20)_ (21)_ (22)_

246

GRAMMATICAL CATEGORIES IN MODERN CHINESE Notes 1. In spite of the lack of indication of tone in the syllable. Syllabic tone has been omitted because its inclusion would (a) complicate the transcription and (b) involve some prejudgement of the role of the terms of the syllabic tone system in the phonological structure, a question which seems better avoided in a grammatical paper. 2. The use of the term 'character' as the name of both the unit of the script and the linguistic unit of which it is the "written symbol parallels the Chinese use of the same term for both. 3. In the structural formulae three types of bracket are used: round brackets () indicate alternatives, that is elements which may or may not be present; diamond brackets < > indicate internal place (i.e. O<X> means that X is internal to O); square brackets [ ] indicate attributive elements, in the element system substantive/attributive. 4. Cf. Chao Y. R. (1948: 208, n. 23): 'A dependent clause comes after the main clause only when it is added as an afterthought.' 5. Figures in parentheses refer to the table of Chinese characters at 6.3. 6. The absence of bound group classes might be a reason for excluding the group from the unit system, thus admitting word classes as operating directly in the clause structure; the parallelism between group classes and word classes would support this. But the complexity of the description of the clause structure is considerably reduced if the intermediate unit 'group' is recognized. 7. In the examples (below, 6.2), words of those adverb classes which operate directly in the clause have been indicated in the structure by the symbol a; this has been done for ease of identification, but does not represent an actual element in the clause structure, these word classes serving merely to mark the class of the clause. 8. I have not used the terms 'subject' and 'predicate'; the given/new system is clearly accounting for features which have often been accounted for by subject/predicate, but the latter terms would not generally be considered applicable, e.g. to the positional system in the conjunctive adverb (see above, 3.3). The actual contextual categories reflected by the grammar in any language are particular to that language; so also may vary the devices (and even the level of the devices) by which such categories are reflected: compare the partly phonological reflection of contextual given/new in English. 9. See for example Maspero, 1937:40.1-42.6; Frei, 1941; Prusvek, 1950. 10 Chao (1948: 41): 'when the two suffixes come into juxtaposition, they are telescoped into one'. For the combination of forms of aspectival mood, and of aspect, with certain adverbs cf. Frei, op. cit. 11 For the syllabic structure of this form cf. Demieville (1951). 12 Kao Ming-k'aj, Han-yti yti-fa lun, Shanghai, Kai-ming (1948: 154), 247

MODERN CHINESE GRAMMAR

13 14 15 16

17 18 19

20 21 22 23

instances duan + ku 'shorts', duan-di + ku 'short trousers' (23); this is comparable with the variation of position of the attributive adjective relative to the noun in Romance languages. E.g. Chao (1948: 35). 'Woo daw. luh sheng (wud (+) daulu sari (25)) "I, the roads are unfamiliar — I don't know my way here".' E.g.byFrei(1941). Also yiau (cf. Chao 1948: 36), maiyiau and 5" (34) in yiau fan yiau, maiyiau ran yiau, maiyiau fan yiau, sja-ga-ran yiau (34). The English translation equivalents of these forms vary to such an extent (according to the free verb in the group) that specimen translations are given here for verbal groups consisting of a free verb together with a postpositive verb of this sub-class. Cf. the auxiliary noun, sub-class (1), below. For a detailed study of forms of the nominal group see Simon (1953). And the numerals to 99. Owing to some lexical identification between the classes of prepositive and postpositive verb, there is sometimes an alternative possibility of statement of a form as either VN or VA: for example surj-dau tusuguan (V(02)N) or sur) dau tusuguan (V A (10)) (53) 'take (it) to the library'. Cf. Chao (1946); also (1948:33). Chao and Yang (1948: xxvii); cf. Chao (1948:33 n). E.g. by Chao (1948:41-4). Chao uses the term 'morpheme' (1948:33).

248

Chapter Four

SOME LEXICOGRAMMATICAL FEATURES OF THE DIALECTS OF THE PEARL RIVER DELTA (1950)

In August 1949 Professor Wang Liao-i (Wang Li) prepared a plan for a phonetic survey of the dialects of the Pearl River Delta region of central Kwangtung. Professor Wang's object was to record the speech-sounds of a number of these dialects in order to gain a complete picture of their phonetic structure, synchronically and diachronically: that is, including the phonology of the present sound-system and the historical development of each syllabic sound-group. On the basis of his knowledge of the Kwangtung dialects (gained partly through investigation and partly through personal contacts1), Professor Wang drew up a list of 36 places of which the dialects were to be recorded. They were Canton , (including Saikwan , C a u s i n g a n d Shekpai ,, Punyu . . . ., Whampoa , Chanchuen I S h i k i u , Shuntak Fatshan , Kowkong , S a m s h u i ' * , Sainarn , Lupao , Koming , Hokshan Hoiping , Sunwui : T o i s h a n , P a k s h a _ „ , T o w s h a n , Chihhsi Kwonghoi .. . _ , Hoingan , Macao ] , Shekkei . . _ . , Siulam , Santong , Tsenghsing : , Tungkun , Sheklung , Cheungmutau , Shamchun „„,.,, Poon , Kowloon P o k l o . , W a i y e u n g , Yungki and 2 Shekwaitong Of these 36 places, all but two were in the Cantonese-speaking area; these two, Poklo and Waiyeung, were situated in Hakka-speaking countryside, and it was desired to know how much Cantonese was spoken in the towns and to what extent the dialects were mixed. The material chosen for the investigation was based on the list of characters prepared by Chao Yuan-ren (Professor Y. R. Chao for the Academia Sinica, and used in the investigation into the Hupei 249

MODERN CHINESE GRAMMAR

dialects carried out in 1936 under Professor Chao's direction.3 Professor Chao's list contained 678 characters arranged under the finals classified according to the Chieh Yun pronunciation, from which according to him4 all the modern dialectal readings apart from a number of irregular single characters can be derived. Professor Wang enlarged this list with some 400 more characters, again adapted from the material used in the Hupei investigation and arranged in two groups: the first by tones and the second by initials, the aim being to facilitate the recording and collating of the phonetic material as a whole. Allowing for duplications, the total number of different characters used was about 1150. In addition Professor Wang added a short list of words and phrases occurring in Cantonese which contain or might be expected to contain phonetic peculiarities. This investigation was carried out by Professor Wang and a research assistant working under him, Chien Sungsheng , with myself assisting for part of the time particularly in the recording and description of the tones, which was left to me entirely. Each subject was given the list of characters and given time to read it through himself in order to become familiar with them, and then asked to read through the list of characters aloud, his pronunciation being taken down in writing (International Phonetic Alphabet), with usually at least two of the three investigators recording independently of each other. Then the subject was asked to read the whole list through again and recorded this time on a wire recorder. As the reading time was 2^ to 3 hours, three one-hour spools of wire were required for each individual reader. The investigators then checked their written records against the wire recorder and verified their notation of the speech-sounds. They then prepared a synchronic analysis of the phonetic structure of the dialect, using a fairly narrow transcription such as would make possible a phonetic comparison with other dialects; and a diachronic analysis of the development of the sounds of Ancient Chinese in the dialect under investigation and the various phonetic conditions of this development. All characters that seemed to show exceptional development were noted and after the complete analysis had been made the subject was summoned a third time and made to read all the doubtful characters. The form of the diachronic and synchronic analysis was again that used in the Hupei investigation with some additional new features, chiefly descriptive interpretation of the diachronic tabulation and generalized tables showing all the possible modern developments in the particular dialect of each broad class of initials and finals.5 No phonological analysis was attempted, though the material relating to the present 250

DIALECTS OF THE PEARL RIVER DELTA

phonetic structure of the dialect is quite adequate for such an analysis to be made. It may be worth noting that there were certain dialects in which the investigators were unable to distinguish by ear the different types of final implosives: final p is of course visible and was clearly seen to exist, and that fact, and far more the fact that each of the subjects said quite definitely that certain pairs of characters otherwise identical were not homonymous, led us to believe that final t and k were in fact distinguished. Nor was this distinction clear on the wire-recorded - until we listened with the wire running backwards. Then the final implosives became perfectly clear voiceless unaspirate t or k — and in Tungkun there was actually found to be a fourth, a palatalized t or t, whose existence had been suspected before but had been impossible to confirm. Some other sounds were also clarified by this method, for example final n and rj. The beginning of the investigation coincided with the southward movement of the civil war, and from the start it was impossible for the investigators to travel to the areas whose dialects were being investigated. That led to the adoption of an alternative, and far less satisfactory, method of investigation: the mountain had to be summoned to the prophet. The majority of the students of Lingnan University, in Canton, come from the various counties of Kwangtung province; and an appeal was made for those who could still speak their native dialects to volunteer to act as subjects for the investigation. It is worth mentioning that dialects in China have none of the social connotations that dialects and 'accents' have in England; they are still purely regional matters. Each Chinese speaks the dialect of his native place (and that means the dialect, not merely the standard language spoken with the accent of the local dialect) no matter what his social status, or the level of his education. Historically the Chinese system of writing has helped to resist the spread of any standard language by means of education, and has thus been one of the greatest factors for promoting linguistic disunity, in that it had enabled every Chinese to go right through his education without being able to speak anything other than his native dialect, at the same time making it as difficult as possible for him to learn a dialect other than his own. In the Hupei investigation, in spite of the fact that it was conducted on the spot, the subjects chosen were all senior secondary or university students, and since they were required to know all the characters involved such a choice was largely inevitable; it was not felt, however, in any way to invalidate the results of the investigation. 251

MODERN CHINESE GRAMMAR

For the Pearl River Delta dialects therefore, students of Lingnan University were the subjects. Those who volunteered were first tested for their knowledge of their native dialect, and it was apparent at once if their memory was faulty. On one admittedly extreme occasion three students from the same county in Chinese) came and sat down together; when one was asked to read all three began to discuss each character before he read it in order to decide on the 'correct' pronunciation! (They were not asked to act as subjects!) Details of the personal history of each applicant were taken and only those who had lived in their native place until their teens and had been in the habit of returning frequently ever since; and finally two subjects were chosen from each place to read separately and were used as an additional check on each other. The material gathered in this way is accurate as far as it goes. Unlike Professor Y. R.. Chao and his collaborators in the Hupei investigation, Professor Wang Li did not attempt to make any inquiry into the lexicon or syntax of the dialects, nor even into the pronunciation of anything except isolated sounds. Because of the structure of the Chinese language it is possible to record the pronunciation, including the tone, of morphemes in isolation, and to gain thereby a fairly accurate idea of the sound of the language; and this especially in the Cantonese dialects, where the sentence-modification is less than in any of the others. But there still exists, even in the Cantonese dialects, a great discrepancy between the sound of the morphemes as pronounced in isolation and as pronounced in sentences; and this is not revealed within the scope of such an investigation. It seemed to me that it would be possible, even under the difficult circumstances then prevailing, to undertake a complementary investigation into the lexicon and syntax of the dialects, at the same time revealing something of the speech-sounds in sentences. Having only a limited time left of my stay in China I decided to make the attempt. Professor Wang Li's investigation afforded the opportunity, as not only were the same people available as subjects but also in listening to their reading of characters and helping in the recording I was becoming familiar enough with the dialects to be able to understand and record short passages of connected speech. From the start there was one great difficulty: the material. What sort of material should be used? And (the question even existed) in what language should it be prepared? To answer this question it was necessary to consider the purpose of dialect investigation and of this inquiry in particular. 252

DIALECTS OF THE PEARL RIVER DELTA

The Academia Sinica (Institute of History and Philology) had formulated in the early 1930s a project for a complete dialect investigation of China together with the preparation of a dialect atlas. The three investigations referred to above, of Kiangsi, Hunan and Hupei, were in fact the first steps in the carrying out of this project. The project as a whole had to be suspended during the war with Japan and could not be resumed after the Japanese surrender because of the almost uninterrupted civil war. Professor Wang Li, who took up the post of Dean of the College of Arts in Lingnan University in 1948, intended to initiate the work of dialect investigation in Kwangtung as soon as he had trained an assistant to help him; not only with the aim of pushing forward the original plan of the Institute of History and Philology, but also because he was extremely interested in the problem of script reform6 and particularly the latinization of Chinese. The 'New Latinization', the form of latinization most likely to be adopted officially (whether for replacing Chinese characters or merely for notation) was favoured precisely because it was capable of being adapted to write any dialect, and Professor Wang wanted to amass all the material necessary for applying it to the dialects of Kwangtung. Cantonese (the dialect of Canton city) had already been Latinized,7 the first draft having been prepared in 1935. City Cantonese is the regional standard for speakers of the Kwangtung dialects (that is, dialects of the Cantonese group): these include most of the area of Kwangtung except the eastern end, where South Fukienese or Amoy dialect is spoken in its Swatow sub-dialect, and the Hakka-speaking interior; and including south-eastern Kwangsi. Within this area there are some subdialects which can still be classified as Cantonese but which are mutually unintelligible8 with city Cantonese, such as Toishan and Kukong. It is clear, however, that city Cantonese has now the status of a regional standard; it is in fact the usual vehicle of communication among people in the Kwangtung dialect region. For the purpose of latinization, city Cantonese was naturally the first to be latinized, and other Cantonese dialects will base their spelling on that of Cantonese, indicating by variant spellings sounds which differ from those of the regional standard. In the same way therefore when investigating the lexicon and syntax of these dialects it is inevitable that Cantonese should be taken as the norm, and a record made of these elements of the dialects which differ from the norm. This question does not arise with the purely phonetic investigation, as each dialect is treated as a linguistic unit, related not synchronically to others but only diachronically to a supposed earlier stage of its own development. Moreover it is a familiar dictum that in 253

MODERN CHINESE GRAMMAR

the investigation of language (le langage) to the linguist all dialects are equally important. This dictum is not being contravened if we say that in the investigation of a language (une langue) all dialects are not equally important; a standard language is more important than a regional dialect and a regional standard than a local dialect. Moreover there are good linguistic reasons for taking Cantonese as the norm for the purpose of the investigation. Obviously the recording of single characters presents no such problem, Chinese writing being mainly ideographic and pictographic; but a lexical-syntactical investigation has to start from material in the form of connected sentences which must be language and therefore in some dialect or other. The discrepancy between Pekingese and the Kwangtung dialects is so great that it would be impossible to present the subject with sentences in Mandarin and ask for his dialect equivalents; moreover it turned out that the majority of the subjects did not know Mandarin. Canton city Cantonese, however, belonged to the same dialect group and could be expected to show no fundamental variations in syntax. Differences in vocabulary could be brought to light by the careful preparation of the material, in such a way that each word to be investigated appeared in context. Accordingly I prepared a body of material consisting of about a thousand phrases and sentences in Cantonese. Apart from the common phrases in everyday use, the material was mainly gathered from three people: either they composed spoken sentences to illustrate the use of certain words or phrases and I took them down in writing, or I took down sentences from their conversation, or they themselves suggested the word or idiom to be discussed; most of the remainder I recorded from conversations, and a very little I composed myself. All of it was checked by at least one person other than the one who composed it and that which I composed myself by two; and finally among the subjects investigated were two people, one from Canton and one from Kowloon, whose native languages were city Cantonese. It was found to represent normal idiomatic conversational city Cantonese quite accurately. It would be rash to say that it was in any way exhaustive; but it was I think representative. This material was in ten sections, as follows: 1. Common words and phrases, including parts of the body, numbers and time-expressions, etc. 2. Unit-words (i.e. classifiers). 3. Common sentence-types, including questions, various types of subordinate clause, etc. 254

DIALECTS OF THE PEARL RIVER DELTA

4. Verbs (monosyllabic) expressing mental and abstract activity. 5. Verbs (monosyllabic) expressing physical activity. 6. Causative and aspective verb-forms. 7. Monosyllabic adjectives. 8. Disyllabic verbs. 9. Disyllabic adjectives. 10. Nouns occurring in Cantonese and not in Pekingese. The first nine sections attempted to be comprehensive and representative; while the tenth was merely a few words chosen at random. To explain the principle on which the selection of material was based it is necessary to consider for a moment the particular nature of the Chinese language. Discussing the modern development of Chinese, Prusek (1945) says that whereas ancient Chinese used prefixes or phonetic changes in the word to form new words, later on words became invariable and the number of morphemes decreased, new terms and new grammatical forms being made up by combinations of existing monosyllabic morphemes. Chinese became, according to Skalicka, a polysynthetic language. The tendency for the number of morphemes to decrease in relation to the total subject matter of language is common to many languages, such as English, in which the trend is towards analysis rather than synthesis; but we can recognize the particularity of Chinese in this respect if we take Prusek's statement at its face value, namely, to mean that the total number of morphemes in conversational use decreases. There is in fact a tendency in Chinese towards analysis and abstraction which has gone further than in any Indo-European language — though whether further than in other East Asiatic languages is open to question, as Malay for instance shows in some words a greater degree of analysis than Chinese. But the real characteristic of Chinese is the specific way in which the process of analysis is carried on. The development is uneven and contrasts specifically with English. Ogden and Richards were able to create Basic English because of one characteristic of the language — the advanced stage of analysis to which the English verb has developed. In Basic there are exactly 18 verbs, called operators, and these are enough to express every action. But English nouns have not attained anything like such a degree of analysis. Basic contains the words 'museum', 'university', 'colony', 'bank', 'programme', and the adjectives 'parallel', 'elastic', 'material', all of which are capable of much greater analysis, as is shown by the fact that every one of these words is an analytical compound in Chinese. 255

MODERN CHINESE GRAMMAR Suppose one tried to make a basic vocabulary of Chinese, one would find the situation directly opposite to that of English. The number of nouns required would be small; the same word would serve to convey the nominal element in 'university' and 'school'; another single word for 'museum', 'restaurant', 'hotel', 'embassy' and 'consulate'; another for 'melon', 'cantaloupe', 'cucumber' and 'pumpkin'. But with the verbs, an immense number is needed to provide an adequate vocabulary. As Wang Li points out with reference to causative forms ,9 the Chinese form expresses the method of the action as well as the result: it is in fact true of all verb-forms in Chinese. Suppose one asks for the Mandarin equivalent of the English 'to carry'. The answer must be tai or ti or kang or duan . or bei ., or tiau or gia , according to where and how the article is being carried. The English 'cut' may be cai or cie or go or kan or pi or zian or even tui • (used for cutting hair) according to the method of cutting and the implement used. This then gives us a clue to the type of lexical variation we may expect to find among Chinese dialects. This absence of analysis is most striking in the verbs of physical action. There is some evidence that this may be connected with a materially backward civilization in which a vast number of operations was performed by hand; we find for instance in Old and Middle English far greater diversity of such verb forms than in Modern, and in our own time we can watch the tendency to ever greater analysis as the use of one generic term for a type of action supersedes the more precise descriptive words, which then become the technical jargon of the specialist. Whether this is true or not, it is here that Chinese is least analytical; the actual number of morphemes in use denoting physical action is very great, and therefore the difference in the lexical forms in the various dialects may be expected to be most considerable. The largest single section in the material which I prepared for this investigation was therefore that dealing with monosyllabic verbs of physical action. It contains 250 sentences and phrases, nearly all complete sentences. The majority of words expressing physical activity are in fact monosyllabic in every Chinese dialect; but there are a number which are not, while the verbs of mental activity seem to be about evenly divided. Of the disyllabic verbs of both kinds, some are subject-object compounds and can therefore be split to form negative and potential forms and with the insertion of pronouns and adverbial phrases; in other words these are generally felt still to consist of verb plus object; while others are inseparable compounds. Comparatively few of the 256

dIALECTS OF THE PEARL RIVER DELTA

disyllabic verbs of physical action, however, are inseparable compounds. It was to be expected that disyllabic verbs would show much less variation than monosyllabic verbs among different dialects: subjectobject compounds because of the stability of the noun element (as in Pekingese bangmang , kitou , etc.) and inseparable compounds because they are mainly abstract. Accordingly verbs of mental activity and disyllabic verbs were treated separately with a separate section each; in this way it would become clear if there were distinct types of dialectal variation. Causative and aspective verb forms play a unique part in modern colloquial Chinese, with no parallels among Indo-European languages; they occupy a separate section of the material. These forms are derived from full verbs, and in classical Chinese they retained the syntactical characteristics of full or finite verbs. Later they became functional, while still retaining a greater or lesser degree of their meaning.10 In modern Chinese they combine with a full verb to express both the method (expressed in the full verb) and the result (expressed in the causative— aspective verb) of an action; but while expressing result they also express aspect. Certain of them with original meanings of completion, attainment, etc., have become completely functional with a definite aspective function: such as kilai , laizh , liao (le) . The majority express a kind of perfective aspect, or (in combination with another functional element) a potential aspect: and they retain their original meanings in which the result of the action is expressed: such are shang . in guanshang , g u a n d e s h a n g , , guanbushang ;diao in todiao , todediao ,tobudiao Since most of the causative—aspective verbs have remained in colloquial use also as full verbs (shang shan . , diao-xialaile lexical variation among the different dialects should be in general no greater and no less than among monosyllabic verbs of physical action. But it might be found that where two synonymous full verbs both existed in a number of dialects, some dialects might have developed one into a functional word and others the other. Moreover there might turn out to be differences in usage and application, affecting the scope of any single one of these causative—aspective verbs or of the speech-pattern as a whole. The section dealing with causative—aspective verbs contains 80 sentences. With adjectives there is for logical reasons much less scope for analysis and abstraction than with either nouns or verbs. Nor can any such generalization be made about the distinction between monosyllabic and disyllabic adjectives, especially considering the large number of 257

MODERN CHINESE GRAMMAR doubled or rhyming adjectival compounds in Chinese. I did, however, divide the adjectives into these two sections, without expecting any decisive differences in treatment to emerge. It was impossible in the short time I had to attempt a systematic preparation of material dealing with nouns; they appear mainly in three sections, the first being a very few common nouns in everyday use, the second one or more examples of a noun with each of the unitwords (classifiers), and the last being a selection of nouns for which the Cantonese uses different words from the Pekingese. Before going on to discuss some of the results obtained in this investigation I should like to add a word about the scope of the inquiry and the method used. The same physical limitations outlined above with regard to Professor Wang's phonetic investigation applied equally of course to this work, which was carried out at the same time and under the same circumstances. It was necessary to find subjects who had not been cut off from their native dialects for any great length of time; while most of the subjects used were the same people used in the phonetic investigation, not all those who could read single characters in their dialect pronunciation could adequately produce sentences in the dialect. I did not attempt to cover all the 36 places that Professor Wang hoped to cover; I had hoped to spend six months after the preparation of the material but was actually reduced to just over two, while Professor Wang was hoping to spend a year or even two years if necessary. So I selected a number of places to be covered, basing the selection partly on the linguistic topography of the area and partly on consideration of what subjects were actually best available. The places chosen were Shuntak, Samshui, Kowkong, Sunwui (two versions), Hoiping, Toishan (two versions), Chungshan (two versions), Hokshan, Tsenghsing, Tungkun and Waiyeung; with two additional places that were known to be Hakka-speaking, Meihsien and Hingning, which I wished to make some inquiry into partly for their intrinsic interest and partly to see whether it was possible to get the required information on the basis of material in a totally different dialect. With the Hakka speakers I went through, not the whole of the material, but about one-half. Of the other places, Tungkun proved to be impossible because only two suitable subjects appeared from the whole university and both fell ill before I could do more than obtain a little preliminary material from one of them; and two others, Hokshan and Tsenghsing, I was unable to finish in the time. I should also add Kowloon; the Kowloon speech is really city Cantonese but I did go through the material with a Kowloon speaker in Kowloon; only one difference from 258

DIALECTS OF THE PEARL RIVER DELTA Cantonese appeared, but that was an interesting one; I shall be referring to it later. As regards the method, it must be borne in mind that in the investigation of lexical and syntactical dialect forms a rather different approach was needed from that made to a phonetic investigation. The dialects under consideration are all so distinct from Cantonese that when the subject is reading isolated characters there is very little chance of the two pronunciations becoming mixed in his mind (or in his mouth). The matter is quite otherwise when the subject is asked to give connected speech and when it is the lexicon and syntax which are the concern of the inquiry. The investigator working under such circumstances can never be sure that the subject will not unconsciously adopt Cantonese forms to the exclusion of those of his native dialect. This means that in order to extract the desired information the investigator must try to avoid bringing the Cantonese forms into prominence and thereby fixing them in the mind of the subject. Since the basic material is in Cantonese the only way to do this adequately is by the use of context. Professor Harold Orton, speaking of the present dialect investigation that is being undertaken in various parts of England with a view to the preparation of a dialect atlas, outlined in a description of the work three forms of question put by the investigator to the subject: the direct question 'what do you call that?', accompanied by pointing to an object; the sentence with one word left out, the subject to supply the missing word, such as 'if you go without drinking water for a long time you get —'; and the introduction of a general topic of conversation in the course of which the subject is likely to use the desired speech-form. The first method is obviously extremely restricted and the third demands much more time than I, at any rate, had at my disposal. The method most likely to succeed is the second, with some variations of which it is capable. If the context of a speech-form can be made clear, the subject can be asked not only to supply a word but also to react, as it were, to a whole situation. You have a prepared sentence, for instance 'I caught a glimpse of him coming so I hid myself; you want to know the dialect forms of 1£ and You describe the situation in Cantonese, using perhaps a less specific word for.,,, such as 'I saw' instead of 'I caught a glimpse', and leaving out ; and then ask the subject to sum up the situation. In practice I found that another method proved highly successful, but it is one which could only be applied in the investigation of dialects of Chinese. Many of the words with which I was concerned, particularly the verbs and adjectives, were colloquial Cantonese and had written 259

MODERN CHINESE GRAMMAR characters that were very little known or else no character at all, in which case I invented them. If I then showed the subject a sentence in writing in which one of these unfamiliar characters occurred, the character did not present any definite phonetic pattern to him; in other words, no Cantonese sound sprang immediately to his mind, and as he was already speaking his native dialect, as he read the sentence he could supply the word that fitted the context. It was then possible to verify that the word did express the meaning by asking for or suggesting the Cantonese equivalent. I was aiming in each instance to note down not merely the single word in which I was interested particularly but also the dialect equivalent of the whole sentence as I had it in Cantonese. With either of the above methods I had to work in two parts: first noting the salient word and then asking for the version of the complete sentence. For this reason, and also because the investigation had to be conducted in a short space of time, I frequently adopted a more rapid method: I gave the subject a copy of the material I had prepared and read out the sentence to him in Cantonese, asked him to digest it and then reproduce the meaning of the whole in his own dialect. In this way the subject kept close to the Cantonese wording. The danger that he would bring in Cantonese words not normally in use in his own dialect could not be entirely eliminated. But various checks could be operated. In the first place much of the vocabulary and all the syntactical forms occurred more than once in the course of the material. In the second place it became possible for me to have some idea of what to expect, to know when dialect forms were likely to differ from my original. In the third place the subjects were at least partly conscious of the aims of the investigation and tried to cooperate. It remains true of course that all that can be said regarding the result of the investigation is that the dialect forms recorded do exist and are correct and in common use; but where the subject has admitted a Cantonese form> even though he said it in the pronunciation of his own dialect consistently with the rest of his pronunciation, there is no unassailable proof that this is in fact a normal or even acceptable form to the native speaker who does not know Cantonese. What I am attempting to describe and evaluate here is one part of the result of this investigation, the material gathered under the section dealing with monosyllabic verbs of physical action. Furthermore I shall in general confine myself to the lexicon of these particular verbs, with only occasional references to other words or to syntactical forms. I shall not of course read lists of words, which is what the material when collated consists of; I shall pick out words and groups of words which have 260

DIALECTS OF THE PEARL RIVER DELTA particular interest and exemplify the general picture of the Pearl River Delta dialects. I shall give the Pekingese equivalent of the words discussed, in order to contrast the difference between Cantonese and related dialects with the difference between Pekingese and Cantonese. We shall then have to decide what relevance, if any, the results obtained here have to the general understanding of the Chinese language.

We may begin with the words corresponding to English 'take, use, using, with'. In Chinese these meanings are often found in the same word, and they can therefore be considered as all one; but in Chinese usage they fall into two, that of 'take' as a finite verb, and that of 'using' or 'with' as a dependent verb almost equivalent to a preposition. I shall speak of the two senses as 'take' and 'use' because these are both verbs in English, though the more normal equivalent in translating the second sense into English would be 'with' (instrumental). Pekingese has the one word na to cover these two uses, with yng used in the latter sense alone; it also has the word ba which has now become purely functional and serves merely to bring into prominence the object of a transitive verb (usually one of physical action) by placing and lo _.. for 'to take', of AND I it earlier in the sentence. Cantonese has nirj which nirj is also used in the sense of'to use', in which sense however the word wan ._, literally 'to look for', is probably more common. The Pearl River Delta dialects show two other words used in the sense of 'take', namely k'ai ... and t'o j^. The first one is pronounced usually k'ai; the second exists only as ho or hu except in Chungshan for which I heard it as t'o; but each of these is consistent with the phonetic developments of t'o in these dialects so I think there is no doubt that it is in fact this word. On the map the two uses 'take' and 'use' are distinguished. In general, dialects use the same word for both, and furthermore they extend the meaning slightly to overlap with the purely functional use of ba in Pekingese. (The type of sentence as in Pekingese, with the object brought forward before the verb, is rare in these dialects, though ziang is sometimes used in conversation in Canton, and was used in a sentence of this type by a subject from Shuntak; but it can occur, with the usual 'take — use' word of the dialect extending its use to that of an auxiliary functional verb equivalent of Pekingese .) But there is some distinction in the Pearl River Delta dialects between the senses 'take' and 'use'. The normal 'take' words in the Canton area are nil) and lo. lo is never found in the sense of'use', but in addition to nirj the word wan 'to look 261

MODERN CHINESE GRAMMAR

for, find' is very frequent in this sense. The Toishan—Sunwui group for 'take' have usually k'ai and t'o (local pronunciation ho, hu) of which only k'ai has the sense 'use', nirj and lo both occur in this area for 'take', but very rarely except for nirj in Sunwui, which is common (pronounced nir), nik); and t'o occurred once in my material for 'use' (Sunwui: Kongmoon). The only normal word for 'take — use' in this area is k'ai, and this word is interesting in two respects. First, it has two variant pronunciations kai and hai found (the first in Hoiping, the second in Kowkong and the two Sunwuis) only in the sense of 'use'. Second, it is found in some dialects on the 'border' of the Canton group and the Toishan— Sunwui group in the sense of'use' and not in the sense of'take': these dialects are Samshui, Kowkong, Sunwui; Kongmoon, Shekkei and, notably, Kowloon. (It will also be clear from these two points together that out of the four places where the variant pronunciation is found, in two of them the word is used only in that pronunciation and only in the sense of use; while in the other two it occurs pronounced k'ai when meaning 'take' and hai when meaning 'use'.) This is the one significant difference which I referred to earlier that I found between Canton and Kowloon. is very common — normal, in fact — in Kowloon speech for 'use'. In Canton I know it is understood, but both in and out of the investigation I never once heard it used except by people I knew to be from the Toishan—Sunwui—Chungshan area. Chungshan in this respect is a border dialect, though actually it merits being considered a group on its own; in Namlong k'ai was completely normal in both senses 'take' and 'use', though both nirj and lo also exist as 'take' only; but in Shekkei k'ai was found only in the sense of'use'. It looks therefore as though k'ai is spreading, roughly north-eastwards we may say, but in a restricted sense, that of 'use'. It is quite reasonable that Kowloon should have received it from Chungshan (perhaps via Macao); it may be found in dialects east of the river which I did not inquire into, but I think this is less likely. At any rate it is not in Tsenghsing and seems not to be in Tungkun. With one single exception the limit of t'o is the same as that of k'ai in the sense 'take'; the exception being that Sunwui: Kongmoon admitted, curiously, t'o but not k'ai; it is possibly the restriction of k'ai to the sense 'use' is there moving back from the east into the area which originally had it in both senses of 'take' and 'use'. In the sense 'take', there is a difference in usage between nirj and I'o: you can say , not on the other hand you can say , not . Speaking very roughly, nirj corresponds to 'bring' or 'take', I'o to 'fetch' in English. The same distinction as between nirj and I'o 262

DIALECTS OF THE PEARL RIVER DELTA exists between k'ai and t'o in Toishan, etc.; k'ai is equivalent to nig and t'o to To. This is worth nothing because it is one instance of a principle of which many more occur in Chinese dialects. Where two dialects use different morphemes (words) in an equivalent sense, the semantic correspondence between the two words is often very exact; that is, the scope of meaning of the two is the same — as contrasted with the equivalent English word which will have quite a different semantic range. This is more common with nouns, however, than with verbs, as is to be expected. The 'throwing' verbs show a great variety of forms, with a rather irregular distribution, and with the same word being used both for 'throw' and for 'throw away'. We find the same thing in Peking, of course, where zheng , diu and even shuai can all be used in both senses. In Canton five words are found, tarn , tig , p'sk , p'au .. and wig ; there is a tendency for tig to be used for throwing with a purpose, wig for throwing away and tarn and p'ek for either; p'au is rarer. Of these p'sk and tarn are found commonly in almost all the dialects; p'au is mainly in the Canton area, and so is wig, though the latter is found in Toishan: Tynfan, and there is a word fen in Hoiping which is almost certainly not connected with wig but is not found anywhere else, tarn is not found at all in the Toishan-Sunwui area, which has instead the Mandarin word tiu read variously tiu, iu, ei, etc. This word is also the common word in Waiyeung (a Hakka dialect) and may have disappeared in the Cantonese area through taboo, being homonymous with the worst of the Cantonese swear-words. Toishan and Sunwui also have another word read Jau in Sunwui, sau in Toishan, and in the tone. Words for 'carrying' and 'holding' in Cantonese, as also in Pekingese, contain in themselves the sense of a particular method or position of carrying. Comparing Cantonese "with Pekingese, we find the four words 'hold (have) inside the mouth' (Ct ham, Pk xaen, both iangping), 'carry (lift) heavy object in arms or on shoulders or on pole between two or more people' (Ct t'oi, Pk t'ai, both iangping), 'carry (usu. child or animal) in arms' (Ct p'ou, iangshang, Pk pao, kysheng), and 'carry under arm' (Ct ka:p, inzhu, Pk tcia, inping) used in the same way in both dialects, except that Cantonese has also the word or t'ok (inky), which has the same meaning as but covers also the sense of Pekingese tuan (inping) 'carry in both hands'. In a further two instances Cantonese and Pekingese use different words but with the same meaning (semantic range). These are 'hold between the lips', Ct .. ta:m, Pk tiao (both inping), and 'carry by a handle', Ct „ _ wa:n (inshang), Pk _._ t'i (inping). 263

MODERN CHINESE GRAMMAR In all other instances the semantic field of the Pekingese and Cantonese are different. For the two Pekingese words k'arj (iangping) 'carry (heavy object) on one shoulder' and t'iao (inping) 'carry on carryingpole (across one shoulder)' Cantonese has the single word fi ta:m (inping); for Pk tirj (shangsheng) 'carry on head' and Pk T=f pei (inping) 'carry on back' Cantonese has no equivalents, in these senses using the nouns for 'back' and 'head' combined with another verb, usually t'ok. The equivalent words in the Pearl River Delta dialects are as follows. All dialects have (tarn, am) in both senses as in Cantonese; all have (t'ok, hok) as Cantonese. All have (t'oi, hoi, hui) (with variant pronunciation toi in Chungshan: Namlong) and (ham) (with variant pronunciation kam (cf. ksrj for , but irregular, cf. hap for ) in Chungshan: Namlong) as in Cantonese. For Cantonese ka:p, Shuntak, Kongmoon, Shekkei and Tsengshing have kip, cf. Cantonese 'pinch' (kip) (related words); other dialects have kap. With 'hold between the lips' it appears that Shuntak, Samshui, Kowkong, Toishan (both), Chungshan: Namlong, Hokshan and Tsengshing have this word, read (tarn, am); Sunwui (both), Hoiping and Chungshan: Shekkei, together with the Hakka dialects, have only (ham) used in this sense as well as in the sense 'hold inside mouth'; it should be noted that is exactly homonymous with (both inping), a fact which would tend to lead to the disappearance of as the less common of the two. Cantonese 'hold in the arms' is only found in Shuntak, Samshui, Chungshan (both), Tsengshing and Sunwui: Kongmoon, together with the Hakka dialects. Sunwui: Sunling, Hoiping and Toishan (both) have lak (iangzhu), Hokshan has kurj (inping or inky, the two having coalesced in this dialect), and Kowkong has ju (inky). Cantonese 'carry by handle' is only found in Shuntak, Samshui, Kowkong, Hoiping and Tsengshing. Sunwui: Sunling and Toishan: Touhuk have lam (iangping), which is the word for 'basket'. The rest used either k'ai (Toishan: Tunfan, Chungshan: Namlong and Hokshan) or nik, nak (Sunwui: Kongmoon, Chungshan: Shekkei), while the Hakka dialects used the ordinary word for 'take', na in Meihsien, tj'a in Waiyeung. 'Eat', Pekingese tsj z_, is in Cantonese sik, or (less common) ja:k (inzhu). Shuntak, Samshui and Kowkong all have sik, as do the Hakka dialects, with jak possible in Shuntak and, rarely, in Kowkong. Chungshan has only jak. Toishan (both), Hoiping and Sunwui: Kongmoon have hek, (inzhu); Sunwui: Sunling hct, which is also the Sunling 264

DIALECTS OF THE PEARL RIVER DELTA pronunciation of and therefore regarded as being the same as Pekingese For 'drink' all dialects including Hakka have as in Cantonese; contrast Pekingese 'Sleep' and 'lie down' are not normally distinguished in the verb in Cantonese, both being fan (inky), with p'a (iangping) used for 'lie down' of animals. In the dialects, p'a for animals was retained in all except Sunwui: Sunling and Toishan: Tunfan which have pok (wok), put (iangzhu) and Chungshan: Namlong which has mau (inping) (the word for 'squat' of human beings). In senses 'lie' and 'sleep' applied to human beings, (fan, fun, uan) is found in Shuntak, Samshui, Kowkong, Sunwui: Kongmoon, Hoiping and Chungshan. In Toishan (both) and Sunwui: Sunling 'lie' and 'sleep' are distinguished, (msn, min) being used for 'lie' and for 'sleep'; and in Waiyeung (Hakka) for 'lie' and soimuk for 'sleep'. Pekingese of course has the distinction between t'arj 'lie' and suei 'sleep'. As a general word for 'rest (vb.)', Pekingese eie, Cantonese has t'au (inshang). This is found in all the other dialects (t'au, hau, hiu) except Chungshan, which has only fan , and Sunwui: Sunling which has a word her). The sense 'tired', Pk lei, is expressed in Canton by kwu:i (iangky), found also in Shuntak, Samshui, Hoiping, Sunwui: Kongmoon and (rarely) Kowkong. Another word, kau, kiu (inping) is found in Kowkang, Sunwui: Sunling and Toishan (both); Chungshan has na:i (shangsheng, inshang and iangshang having coalesced). Waiyeung has k'oi (kysheng), presumably the same as For 'open, shut (the eyes or the mouth)' Pekingese has and The normal word for the former in Canton and the Pearl River Delta dialects is mak, ma:k , but , read with aspiration tj'ar) (inping) is heard in Canton, Kowkong, Sunwui and Toishan, and is the only word in Waiyeung. In the sense 'shut', Canton, Shuntak, Samshui, Kowkong, Sunwui: Sunling, Hoiping, Chungshan (rarely) and Waiyeung (Hakka) have mi, mei (inping); the phrase 'shut your eyes' is in Hakka). Other places have hap, hsp (Toishan (both), Chungshan; Shuntak and Canton (less common)) or hip (Sunwui: Kongmoon), both or cognate. In the senses 'stick; mend; patch' Pekingese has the series t'ie means 'stick (something on to something larger, e.g. wall)', ts^aen 'stick together (transitive)', nien 'stick together (intransitive); be sticky', lisn 'mend by joining together', pu 'mend by patching'. Canton and all dialects have but in many it is felt to be 'refined' 265

MODERN CHINESE GRAMMAR

(because borrowed from literary Mandarin) and the subjects from Samshui and Chungshan: Namlqng rejected it altogether, using in this sense. is used in Cantonese and all the Pearl River Delta dialects except Sunwui: Sunling and Toishan: Touhuk, but the sense is more that of Pk , the sense of Pk being represented by another word (tj'i (inping) in all) found in Cantonese and every one of these dialects, as well as also Hakka; moreover Toishan: Touhuk has the additional word nak found elsewhere only as an adjective 'sticky'. The word is also used in the sense 'mend by joining' (where Pk has ), in which sense the Pearl River Delta dialects show another little group of words with very complicated variations of pronunciation. Thus, Cantonese has (lyn, but read lin in this sense) and na (inping); the latter is not found elsewhere at all, but (lyn, 10n, lin) is used in the other dialects, being the normal word in Shuntak, Sunwui: Kongmoon, Toishan: Tunfan, Chungshan: Shekkei and Hokshan. Chungshan: Namlong has the word nen, len which may be (it would correspond to Cantonese reading lin); but there also exists a word read nan, Ian (iangky) clearly distinguished from by the fact that while occurs in Shuntak, Hoiping, Toishan: Touhuk and Hokshan, Shuntak and Hokshan have as well, the two pronunciations being quite distinct (Shuntak nan: lyn, Hokshan nan 10n). Whether this nan, Ian has any connection with through the Chungshan it is impossible to say. At any rate there are also ma (iangshang) in Samshui and Sunwui: Sunling, na:rj in Toishan: Tunfan and na:p and mar) in Kowkong, all in this same sense. The Hakka dialects use here. Pekingese is retained in Canton and other dialects in the sense 'mend by patching'. For the adjective 'sticky', Pk nien, Canton, Shuntak, Samshui, Kowkong, Hoiping, Sunwui: Sunling, Toishan: Tunfan, Chungshan (both) and Hokshan have nap. Sunwui: Kongmoon and Toishan: Touhuk nak. The Hakka dialects have njasm. The few words which I have quoted have shown I think that there is need for a lexical study of this sort if the dialects are to be fully understood. Nearly all of the three hundred and more verbs in the section from which I have been quoting showed at least one variant among the 12 dialects when taken in context: at least one of the dialects, that is, used a different word (not merely a different pronunciation of the same word) from that used by the Cantonese to express the same meaning in the sentence. Without this kind of an investigation there is no means of getting at these words. I have not spoken about the syntactical side of the inquiry. Some of the features revealed there are equally interesting, particularly the use 266

DIALECTS OF THE PEARL RIVER DELTA Pearl River Delta dialects provisional tone-chart ping in iang Canton

5,53

shang in iang

ky in

21

iang

zhu in 1

in 2

5

3

iang 1 ,2

2

35

Shuntak Samshui

53

42

Kowkong

5

23

35

Hokshan

3

5

34

23

3

2

12

21

3

12

45

3

43

34

12 32

43

Hoiping

32

Sunwui Toishan

3

Shekkei

5,53

42

23

32

2

3

42

23

2

5

21

3

32

5

3

32

21

5

3

5

3

Chungshan Namlong

21

of the different causative—aspective verbs of which as I have said neither the meanings nor the forms are identical throughout the dialects. The broad outlines of sentence structure vary hardly at all among these dialects, as indeed they do not very much even among the major dialect groups of China (though there is more syntactical variation between Pekingese and Cantonese than is often supposed, as it is obscured by the fact that most people when they learn a second dialect transfer the syntactical features of their own to the new one, which sounds odd but is quite intelligible; for example, Pekingese tends to put the object at the beginning of a sentence containing a pronoun subject followed by a potential aspect — Cantonese does not; but if you say it is of course perfectly comprehensible). Between Cantonese and Toishan dialect the difference is even less. But in matters of detail there is a certain amount of difference. Toishan for instance negatives the potential aspect in the Mandarin way ( not as in Cantonese ). It is not possible as a result of this investigation to draw boundary lines between different groups of sub-dialects. In fact, in all dialect investigation, the more one finds out about the dialects, the less clear the popular descriptions become, and it is necessary to discard altogether the notion

267

MODERN CHINESE GRAMMAR

of delimiting dialect areas with accuracy. The exception to this is where there is a clear boundary between two dialects which do not merge one into the other; this is or may be true of the major dialect groups, e.g. Cantonese and Hakka, which result from a double settling of the area - the second wave of settlers arrived in Kwangtung some centuries after the first wave, and the descendants of the first arrivals speak Cantonese while those of the newcomers speak Hakka. Even here it must be inquired into whether there has since been merging, as undoubtedly there has, and how much merging there has been. But for sub-dialects of a dialect continuum, the only valid approach is by the drawing of isoglosses for each salient linguistic feature, in the way I have tried to do on the maps. Popular classification will be found to correspond roughly to the line of the thickest bundle of isoglosses; but there will be may smaller boundaries which diverge from and even cross the main linguistic boundaries. In a country like England, where many regional dialects are disappearing fast, the work of dialect investigation, at which we are far behind the continental countries and which is only now beginning to be done seriously here, has to be done in a great hurry. Professor Orton already records only from speakers over sixty years old and if possible from those over seventy. In China it will be a long time before even the sub-dialects merge and disappear, though even now the tendency exists for local vernaculars to approximate to their regional standards. In any case the material provided by the Chinese dialects contains much that is of great linguistic interest and significance, and it is to be hoped that Chinese scholars like Wang Li will take the subject in hand and fill in the picture, so much of which still remains sketchy, or a blank. November 1950

Notes

1 Professor Wang is a native of Popai 2

3

in Kwangsi; his native dialect is Kwangsi Cantonese. See the maps on the accompanying CD-ROM. Spelling of places as on the War Office map 'Kwangtung', 1:1,000,000,1939. It should be noted that there are inconsistencies, some places being transcribed from Cantonese and others from Mandarin. The results of this investigation were ready for publication in 1938 but could not be sent to press owing to the war; they were actually published in 1948. See 2 vols, Shanghai, Commercial Press, 1948 (p. 1565, + numerous maps). Although this was the third such investigation 268

DIALECTS OF THE PEARL RIVER DELTA

4. 5.

6. 7.

carried out, being preceded by Kiangsi and Hunan, it is the only one so far published. See ibid., Introd., p. 7. The first of these investigations to be published was the Tungkun dialect. See in the Lingnan Journal, vol. X no. 1, December 1949. See for instance his Commercial Press, Changsha, 1940. See Ni Xaishu, (1949). The first draft of Latinized Cantonese appeared in 1936: see published by

8. This is a relative term. Educated speech on serious topics would be mutually intelligible. 9. vol. I, p. 155. 10. See Prusek, ibid. Aspective verbs, says Prusek, emphasize one side of the action expressed by the main verb; they retain something of their original meaning and offer the speaker a range of subtle distinctions from which to choose.

269

This page intentionally left blank

PART THREE CHINESE PHONOLOGY

This page intentionally left blank

EDITOR'S INTRODUCTION

Both papers in this third section deal with Chinese phonology. Chapter Five, 'The Origin and Early Development of Chinese Phonological Theory' (1981) presents a historical overview of the development of phonological theory in China. A more recent paper, Chapter Six, 'A Systematic Interpretation of Peking Syllables' (1992), offers a systemic interpretation of data based on observations made when he was in China in 1959. The 'Origin and Early Development of Chinese Phonological Theory' traces the historical roots of Chinese linguistic scholarship from the Er Ya Treasury of Fine Words, which was said to be the work of a disciple of Confucius, Zi Xia, through to the Shuo Wen Jie Zi 'Explanation of Simple and Derived Characters', which was compiled at the end of the first century AD by Xu Shen. Described by Professor Halliday as "a direct and logical development of the thesaurus system of arrangement as already adopted in the Er Ya", the Shuo Wen's method of classifying characters according to a component of the written symbol, known as the 'radical', helped to establish the lexicographic principle on which many contemporary dictionaries and encyclopedias are based. Another significant milestone in the development of Chinese phonology was the Qie Yun, 'Segmental Rhymer', published in 601, "a phonological dictionary which had the specific purpose of establishing standardized rhyme tables for use in poetry, including the poems composed by candidates for the public service examination". The Qie Yu was based on the fanqie technique, 'turning (i.e. inverting) cutting', which analysed every syllable into two parts, including its initial consonant, and whatever followed. The pronunciation of a character could 273

CHINESE PHONOLOGY then be worked out by reference to two other characters. Professor Halliday illustrated the fanqie technique with the following example: doqg, de hong qie in which, d5ng is (represented by) the segmentation of de hong; i.e. to find the pronunciation of dong take the initial sound of de and the final sound of hong and combine them. Beginning with the Tang dynasty, the influence of Indian phonetics, which tended to be more explicit in orientation, can be seen in attempts to step-by-step describe and classify the initials and finals of the Chinese syllable. As Professor Halliday notes, "By the time of the Sung dynasty (c. 950-1250) the impact of Indian linguistics had led to an extremely rich if sometimes rather confusing era of development within Chinese phonology, a development that continued over a thousand years, right up to the introduction of Western linguistics at the end of the nineteenth century." Throughout its development, Chinese phonology has remained a phonology of the syllable. It is not surprising therefore that the study reported on in Chapter Six, 'A Systematic Interpretation of Peking Syllables', focuses on the syllable in modern standard Chinese, drawing not only on traditional Chinese phonology, as interpreted by Luo Changpei and Wang Li, under whom Professor Halliday studied when a student in China, but also on prosodic phonology as developed by his mentor, J. R. Firth. In fact, as Professor Halliday indicates, "These two approaches are entirely compatible and share a highly abstract view of phonology based on paradigms of (non-segmental) features."

274

Chapter Five

THE ORIGIN AND EARLY DEVELOPMENT OF CHINESE PHONOLOGICAL THEORY (1981)

Most students of classical thought would be likely to reject the view that there was any connection between a particular language and the philosophy or other intellectual modes that developed among its speakers: the idea that, for example, the directions taken by Greek scholarship could in any significant respect be ascribed to the nature of the Greek language. But there is one special exception, in which such a connection is naturally admitted, namely the study of language itself. Linguistics, wherever it has developed autochthonously, has begun as the study of the home language; and it is readily accepted that the nature of this language, whichever it happens to be, has played a significant part in determining the directions linguistic studies would take. In ancient Greece and India, where the languages were rich in word morphology, linguistics first developed as the study of word paradigms; and this, through a search for explanation of the choice of case, eventually led to syntax. The influence of the language on the development of linguistics in China seems no less clear. Classical Chinese had virtually no morphology; morphology is the outward sign of grammar; this, it is assumed, is why there is no study of grammar in Chinese linguistics. Both the general pattern of Chinese linguistics, with its concentration on lexicology and phonology, and the particular methods developed in these two fields, can be reasonably well explained by reference to the nature of the Chinese language and script.

'The origin and early development of Chinese phonology theory', from Towards a History of Phonetics, edited by R. E. Asher and Eugenie J. A. Henderson (1981). Copyright © Edinburgh Press.

275

CHINESE PHONOLOGY

Classical Chinese was largely monosyllabic: most words were (composed of) one morpheme, and most morphemes were (realized as) one syllable. The writing system was, and still is, morphemic: one character represented one morpheme. We may accept Y. R. Chao's observation that 'so far as Classical Chinese and its writing system is concerned, the monosyllabic myth is one of the truest myths in Chinese mythology' (Chao 1968b: 103). The fundamental linguistic unit represented in the consciousness of the speaker is now, and presumably was then, the complex made up of morpheme/syllable/character; this concept is referred to in Modern Chinese as zi.1 Bimorphemic compounds already existed in the classical language, typically synonym compounds, species compounds and expressive reduplications; but the sense of the unity of morpheme/syllable/character was so strong that every disyllabic expression, even one in which neither element could occur without the other, was and always has been interpreted as consisting of two morphemes, and hence felt as belonging to the class of synonym compounds. Not only native words are thought of in this way, such as xishuai2 'cricket'; the principle is extended to include disyllables that began as loanwords, e.g. boli3 'glass'. Like other original traditions in linguistics, that in China sprang from concern about the growing unintelligibility of the ancient texts; in particular, there were many characters in the Book of Odes (songs dating mainly from 900-600 BC, said to have been compiled by Confucius) that had disappeared from current use by the later classical period. The earliest known work of Chinese linguistic scholarship is a thesaurus of classical and pre-classical terms entitled the Er Ya4 'Treasury of Fine Words', traditionally ascribed to a disciple of Confucius, Zi Xia.5 This contains between three and four thousand entries, divided into 19 sections; the first three sections are devoted to the exposition of important or difficult terms of a general nature, nouns, verbs and figurative expressions, while the remaining 16 sections are topical groupings entitled Kin, Buildings, Implements, Music, Sky (i.e. calendar and climate), Land, Hills, Mountains, Water, Plants, Trees, Insects (which includes reptiles), Fishes, Birds, Wild Animals and Domestic Animals. Entries in the first three sections are explained by the use of synonymy: they are single items, or lists of related items, glossed by a synonym, or lists of cohyponyms glossed by a superordinate. In subsequent sections the method varies, but includes mainly i) synonymic gloss, as above; ii) reverse definition; iii) descriptive definition. Some entries are citations from the Book of Odes. No indication of pronunciation is given.

276

CHINESE PHONOLOGICAL THEORY Examples: (i) daodao, tuantuan,. . ., 'you ye daodao, tuantuan, etc. mean 'sad' wang, hou, bi, gong, hou, jun' ye wang, hou, bi, gong, hou are 'the royalty' (ii) 'da qin' wei zhi li 'a large lute' is called li (iii) yayu, lei 'chu', hu zhua, shi ren, xyn zou yayii is a kind of wild cat, tiger-clawed, man-eating, swift-running The thesaurus principle established by the Er Ya continued to be used throughout Chinese history as a method of lexicology; much later its use was extended to Sino-foreign wordbooks and phrasebooks. The work that first gave a systematic representation of how characters are pronounced did not, however, adopt this arrangement; it introduced a very different method of classifying characters, grouping them according to a component of the written symbol. This was the Shuo Wen Jie Zi6 'Explanation of Simple and Derived Characters'. The Shuo Wen Jie Zi classification of characters established the lexicographic principle which has been used in dictionaries and encyclopedias right down to the present day. In fact, however, this method, of classifying by 'radical' (to give it its usual English name), is a direct and logical development of the thesaurus system of arrangement as already adopted in the Er Ya, and results in quite similar semantic groupings. The Shuo Wen (it is usually known by its abbreviated title) was compiled at the end of the first century AD by Xu Shen.7 It contains 10,516 character entries, of which 1,163 are said to be duplicates. Each character is given in its archaic, 'seal script' form, followed by a gloss which is typically in three parts: semantic — a synonym, or brief definition; palaeographic — an explanation of the origin of the character; and phonetic — an indication of the pronunciation. For example: hao, 'hu sheng' ye, cong kou hu, du ruo hao hao means 'noise of a tiger', from kou (mouth) and hu (tiger), read like hao These take up 14 of the 15 sections of the work. The fifteenth section is in two parts, a 'postface' and an index of radicals. The postface is a 277

CHINESE PHONOLOGY

short palaeographic essay containing a classification of characters into six types according to the principle on which they were formed (cf. Chao 1968b: 103-5; Kratochvil 1968:148-51). These are, in the order listed in the Shuo Wen, and with Xu Shen's own examples: Type Translation (1) zhis ( 2 ) xiangxing9 p i c t o r i (3) xingsheng10 phonetic (compA (4) huiyi11 semantic ( c o m p (5) zhuanzhu12 m o d (6) jiajie13 transferred

Examples

Meaning v e , below un, moon )river, river )arms, loyalty eased, old order, grow

Type (1) are non-representational iconic symbols, e.g. (old ) shang 'above'; (old ) xia 'below'. Xu Shen gives very few of these, preferring to interpret in terms of the next wherever possible. Other examples that have been given include ba 'eight', symbol of divisibility, and the numbers er 'two' and san 'three'. Type (2) are pictorial representations, e.g. (old 0) ri 'sun'; (old ) yue 'moon'. Others: ( ) mu 'tree'; ( ) men 'door'; ( ) shui 'water'; ( ) nii 'woman'. Xu Shen includes in this category indirect representations such as da 'big', picture of a man with arms outstretched; gong'skill, artisan', picture of carpenter's square. Type (4) are meaningful combinations, e.g. (modern ) wu 'arms, military', consisting of 'lance' and 'foot' - by Shuo Wen's time meant 'stop', so this character was later explained as 'fighting to stop fighting' (Chao says 'cf. idea of "a war to end all wars" '),but the probable explanation is the simpler one of marching under arms; xin 'loyalty, belief, consisting of a man standing by his word. Other examples are rning 'bright', sun and moon; fu 'submit', kneeling man and hand; dong 'east', sun seen in tree as at sunrise. Type (5) are modifications of another character with related meaning, e.g. kao 'old, deceased, "the late"', interpreted by Xu Shen as a modification of lao 'old' by deflection of one stroke. But according to Karlgren (1940), is a phonetic compound, type (3) below, with radical 'old' and phonetic kao, made to look like a modification of because the element got fused into it. Chao doubts the reality of this class; another traditional example is shi 'corpse', said to be a variant of the character ren 'man (human being)', presumably deformed in some way. In the characters of these four types, there is no phonetic principle at work. In the two remaining, there is. Type (6) are 'phonetic loans', 278

CHINESE PHONOLOGICAL THEORY characters transferred from their original morpheme referent to another one that is homophonous with it. was originally a pictorial representation of a man and a bell, and was the character for ling 'bell'; it was borrowed to represent the homophone ling 'to order'. The two were probably cognate; in Xu Shen's other example, chang 'long' transferred to zhang 'to grow', they certainly were. But this category is now thought of primarily as one involving transference to noncognate homonyms, e.g. lai 'kind of wheat', picture of growing grain, 'borrowed' to represent lai 'come'; kai 'triumphal music', picture of a drum, for qi 'how?' (O.C. k'ar, k'jsr); fan 'sail', picture, for fan 'all'. In type (3), which are the most numerous, the phonetic loan has been backed up by the addition of another character signalling the meaning, resulting in a phonetic—semantic compound, usually referred to simply as 'phonetic compound', e.g. Jiang 'river', with gong as phonetic element and (a reduced form of) ?R 'water' as semantic element; he 'river', RTke as phonetic element and again 'water' as semantic (both pairs were near-homonyms in Old Chinese). Other examples: zhui 'dove' (itself a picture) occurs as phonetic element in chui 'hammer', with semantic 'tree, wood'; in shui 'who?', with semantic 'speech';in wei 'to tie, guiderope', with semantic 'silk';in tui 'to push', with semantic 'hand'; and various others, all near-homonyms in Old Chinese (when there were fewer exact homonyms than in the modern language). Some of Xu Shen's explanations are fanciful; for example, he explains the character ming, meaning 'personal name', which consists of the character for 'mouth' surmounted by that for 'night' (the latter a variant of the 'moon' picture), with the words 'at night it is dark and people cannot see one another, so they use their mouths to call out their names'. Some of his interpretations have been disproved by recent discoveries of early forms of characters in inscriptions and on oracle bones, to which he did not have access; and the meanings he assigns to simple characters that no longer occur alone, but only as elements in others, seem at times to be guesswork based on his interpretation of their origins, without textual evidence. But the vast majority of his analyses are right; and the general theory (which he ascribes to an imperial tutor of former times), with the possible exception of the doubtful type (5), can certainly be accepted as valid. As a means of ordering the characters, Xu Shen adopted the following principle. Out of the ten thousand or more he was able to identify 540 such that all other characters contained at least one of the 540 as a 279

CHINESE PHONOLOGY component part. This set of 540 was then used as 'head characters' for the purpose of classifying the remainder. He arranged them in a reasonably natural order based on what seemed to be the dominant configuration of strokes; this yielded small sets of around 5-10 characters, within which further ordering was achieved roughly by degree of complexity (number of strokes). Under each head character was entered, first itself, and then all other characters identified by having it as their 'radical' component, in apparently arbitrary order. This system is still in use, though the number of radicals was progressively reduced in two dictionaries of the Ming dynasty, first to 360 and then, in the Zi Hui14 of around 1580, to the modern inventory of 214; and the number of strokes became the dominant ordering principle, not only of the radicals themselves but also of the characters listed under each. What determined Xu Shen's choice of radical? In the case of characters of types (1), (2), (4), (5) and (6) the choice was largely arbitrary; but there were relatively few characters belonging to these types that were not themselves among the 540 head characters, where the question of which part was the radical did not arise. The vast majority of characters other than the 540 head characters were of type (3); that is, they consisted of two parts, one semantic and the other phonetic in value. And here Xu Shen's choice of radical was not arbitrary, since he adopted as a principle what had already emerged as a natural pattern in the Er Ya. Looking at the later sections of the Er Ya, one finds that, because the characters in a section share a common meaning, they tend to share a common semantic element. So those in Section 14 typically have the 'tree' radical mu,15 those in Section 17 the 'bird' radical niao;16 those in Section 6 have 'metal', 'bamboo', 'cloth' and so on. Xu Shen turned this into a general principle, always selecting the semantic and not the phonetic element as the radical, under which to list the character. So in paving the way for future lexicographers Xu Shen ensured that there was much less mismatch between dictionary and thesaurus than there is in European languages, because the indexing principle of the dictionary is based on the semantic element in the written symbol. But the consistent choice of the semantic element as radical also has a consistent side-effect: it clearly identifies the phonetic element wherever there is one, since that is the part left over. Wherever possible Xu Shen made use of this element as an indicator of the pronunciation. Let us look at the most typical form of entry in the Shuo Wen. By far the most frequent is the following: 280

CHINESE PHONOLOGICAL THEORY wei, 'bao' ye, cong van, wei sheng wei is 'to announce', from yan (speak), wei sound

The character is made up of the character 'speech' functioning as semantic element (pronunciation irrelevant) and the character wei functioning as phonetic element (meaning irrelevant). Both halves now serve a lexicological function: as the 'radical', for indexing purposes, and as the indicator of pronunciation. By the time the Shuo Wen was written, however, at least a thousand years had passed since the main period of evolution of the script; and many pairs of characters, the one serving as 'phonetic' element in the other, were no longer exact homonyms — if indeed they ever had been. In the case of and they are: both are, in Karlgren's reconstruction, Old Chinese gjwsd, Middle Chinese jwei. Xu Shen clearly tolerated some departure from exact homonymy: he used the same formula 'sound ' not only for instances that did not match for tone but also for others where there was a minor difference in consonant or vowel, for example: pu, 'shui cao' ye, huo yi zuo xi, cong cap, pu sheng pu is 'a water plant, sometimes used for making mats'; from cao (plant), pu sound

where pu17 had voiced initial (O.C. b'wo, M.C. b'uo) and pu18 voiceless (O.C. p'wo, M.C. p'uo). In practice Xu Shen was accepting about the same degree of homonymy as had gone into the making of the script in the first place. Where one character was used as phonetic element in another, the morphemes they represented were often not identical in pronunciation but might differ in a variety of features: voicing/ aspiration, nasality, palatality, type of stricture, vowel quality, or combinations of these (see Karlgren 1940, passim). For Xu Shen, if one character had another one as its phonetic, that was normally enough reason for him to accept the latter as its 'sound'; in other words, the entry under 'sound' was not yet a true indication of pronunciation, and the concept of the syllable as a phonological unit had not yet fully emerged from the complex notion of'character'. But that it was already well on the way can be seen from the other type of pronunciation gloss found in the Shuo Wen. Characters of types (1), (2), (4) and (5) have no phonetic element, and so could not be treated in the same way. Most of those that are themselves head characters, as 281

CHINESE PHONOLOGY well as a few others that are very common, mostly of type (4), are left without any phonetic gloss at all. But in a minority of instances, Xu Shen introduces another formula: du ruo 19 'is read like ', or du yu tong20 'is read the same as '. This is used with less common characters, like the 'noise of a tiger' cited above; and is also occasionally added to the entry for a character which already has a phonetic element where that element is for some reason unsatisfactory as an indication of the sound: because it is very rare, or it has more than one reading — or perhaps because it is felt to be markedly different in pronunciation. Here both forms of gloss are combined in a single entry: du, 'bi' ye, cong pti, du/duo sheng, du ruo du du is 'to close off, from pu (to strike), du/duo sound, read like du where the du reading had final vowel (final voiced stop in Old Chinese, but lost by Shuo Wen time), the duo reading had final consonant (voiceless stop); thus the phonetic element,left by itself, would have been ambiguous as a pronunciation guide. In the following, however, the justification seems to be different: tiao,

'si ji'

ye, cong xue,

zhao

sheng,

du ruo

tiao

tiao is 'frivolous', from xue (cave), zhao sound, read like tiao (the original sense of the character was 'secluded', hence the 'cave' radical). Here the phonetic has only one reading, nor is it at all rare; the reason for the additional gloss appears to be that by this time the difference in pronunciation between it and its 'host' character had become rather striking — probably zhao21 was voiced initial, palatalized, open vowel, while tiao22 was voiceless initial, unpalatalized, mid-vowel. So while on the face of it there is not much difference between the two forms of the gloss,' sound' and 'read like ' (both simply refer one character to another), there is an important difference in principle. The former means 'this part of the character indicates its phonetic value'; it involves no concept of an independent phonetic symbol. The latter means 'this character is pronounced like that one'; the second character is brought in'only as a phonetic notation. The latter is therefore the first step towards an abstract phonological analysis. Xu Shen was not, in fact, the first person ever to think of glossing a character in this way, as being pronounced like another one: there are odd instances in the classical texts, sometimes accompanied by a descriptive note. Wang Li (1936:1, p. 107) quotes the following from the Spring and Autumn Annals: 282

CHINESE PHONOLOGICAL THEORY yan

nai

zhe

nei

er

shen,

yan

er

zhe

wai

er

qian

nai is spoken interior and deep, er is spoken exterior and shallow

recalling Downer's comment that 'such descriptions mean little to anyone except the originator' (1963: 128) — although given that nai was O.C. nag (M.C. nai), er was O.C. njsg (M.C.ngi), it is perhaps not too fanciful to interpret 'interior and deep' as 'open back vowel' and 'exterior and shallow' as 'close front vowel'. But Xu Shen was the first to make systematic use of character homonymy in what was, in fact, a work of linguistic scholarship. In so doing, he set Chinese scholarship along the path of phonological analysis. As a technique of phonological analysis the use of homonymy has severe limitations, since it is confined within an extremely restricted circularity. In the century or two following the appearance of the Shuo Wen it was superseded by a new technique, which eventually became to be known as 'turning (i.e. inverting) cutting', fanqie23 ('countersegmentation' might be an appropriate English rendering). By this method the pronunciation of a character is stated by reference not just to one other character but to two, which then have to be combined. The formula is exemplified by dong, de hong qie dong is (represented by) the segmentation of de hong

(M.C. tun, tsk yurj). In other words, to find the pronunciation of dong24 take the initial sound of de25 and the final sound of hong26 and combine them. This is still, of course, circular; but the circle is very much wider, as in the pronunciation guide in a modern English dictionary. This method is known to have been used in a third century phonological commentary on the Er Ya, the Er Ya Yin Yi27 by Sun Shuran,28 which is now lost. Few fanqie renderings from before AD 600 have been preserved in the context of the works in which they originally appeared. But from 600 onwards there is a wealth of such material, and it is known that many of the fanqie renderings contained in the principal works now extant were taken over from earlier sources without change. The most important single work, which represents the high point of the early phonological tradition, is the Qie Yun29 'Segmental Rhymer'. This was produced by a government research team directed by Lu Fayan30 and published in 601. For many centuries it was lost, as was an eighth century revision of it the Tang Yun31 'Tang Rhymer' by Sun 283

CHINESE PHONOLOGY Mian.32 It was known, however, through an extant further revision, the Guang Yun33 'Enlarged Rhymer' (full title 'Great Sung Revised and Enlarged Rhymer') by Chen Pengnian34 and Qiu Yong,35 published in 1007-8, which was the version used by all subsequent scholars until the twentieth century. In the first decade of this century portions of early manuscript copies of both Qie Yun and Tang Yun were found in the far north-west of China, by Paul Pelliot and Aurel Stein, in the great Buddhist collection that had been preserved in the Tunhuang caves. These showed that the Guang Yun did in fact carefully conserve the original fanqie renderings given by the authors of the Qie Yun. Subsequently further manuscripts were found in other parts of China, making it possible to reconstruct the major portion of the Qie Yun or its earlier, Tang Yun revision. Much of this material was made available in an edition prepared by Luo Changpei36 and Wei Jiangong37 entitled Shi Yun Hui Bian.38 Just a few years later than the Qie Yun is another important work of 'contrasegmental' phonology, the Jingdian Shi Wen39 'Character Analysis of Classical Texts', by Lu Deming,40 who died in 630. According to Karlgren (1940: 8-11), it was quite independent of the Qie Yun, but based on the same dialect, that of Changan,41 the capital at that time (modern Sian); and it often serves to confirm the Qie Yun interpretations. Lu Deming quotes extensively the fanqie renderings of earlier writers. This, together with Lu Fayan's Preface to the Qie Yun which shows that the authors were aware of the phenomenon of diachronic change and had deliberately preserved distinctions recorded by their predecessors, suggests that the fanqie were seen as a method of maintaining traditionally accepted standards of pronunciation. The fanqie technique presupposes an analysis of every syllable into two parts, an initial and a final. Wang Li (1936:1, p. 107) suggests that this originated with a small number of etymologies, known since classical times, by which certain morphemes were explained as the fusion of two others, for example: er

yl wei

er, zhi hu wei

zhu

er yi becomes er, zhi hu becomes zhu (O.C. jijag zjag becomes jijag; tjag g'o becomes tjo), though these 'natural segmentations' were never used as phonetic renderings. The initial segment consisted of just the initial consonant. The final included everything else: vowel, final consonant if any, prosodic value (palatal, labial or open), and tone. The final thus corresponded exactly to 284

CHINESE PHONOLOGICAL THEORY what functioned as the rhyming element in verse, not only in postclassical times but also as far back as the Book of Odes itself (see Karlgren 1940:90-110, for the actual rhymes contained in these songs). By the time of the Qie Yun, this two-part structural analysis had become explicit: that is, the elements 'initial' and 'final' had been recognized, and given names, niu42 and yun.43 The latter in fact means 'rhyme'. The former means 'a knot', and its use in the sense of'initial' has not been explained. The Qie Yun was a phonological dictionary which had the specific purpose of establishing standardized rhyme tables for use in poetry, including the poems composed by candidates for the public service examinations. The characters are therefore classified under their finals, a total of 206 distinct finals being recognized. These are distributed over the four tones, so that the actual number of distinct syllabic rhymes is much less: 61 in all, or 95 if those in -p -t and -k are counted separately rather than as tonal variants (see below). The finals are tabulated at the end of the work; but there is no notation for them, nor is there any explicit analysis in phonetic terms. The only explicit phonetic terminology at this stage is found in the analysis of the tones. In the Nan Shi44 'History of the South (Southern Kingdom)', the biography of Shen Yue45 (441—513) refers to an essay he wrote called Si Sheng Pu46 'Gamut of the Four Tones', and recounts the now wellknown story of the answer Shen Yue gave to the Emperor Wu Ti who asked him to explain them: 'Tian Zi sheng zhe'.47 This illustrated the four tones in the order in which he had listed them (M.C. fien tsi Jjaerj tjaet); it also expressed the loyal sentiment 'The Son of Heaven is holy and wise'. Shen Yue called the tones ping, shang, qu, ru,48 terms which likewise are appropriate in both sound and meaning: they were in fact instances of their respective tones (M.C. bjwarj, gdrj, kjwo, ngsp), but they could also be — and subsequently usually were — understood as descriptive labels, since the words meant 'level', 'rising', 'going' and 'entering': the first were, in fact, level in tone, the second rising, the third falling and the fourth 'stopped', that is, ending in a plosive consonant. Note that these do not correspond to the four tones of Modern Pekingese. Shen Yue thus treated the stopped syllable as a tonal variant of the others; and this became the established interpretation. It makes very good sense provided the stopped syllables are matched with the corresponding syllables with nasal finals; and this is the practice adopted in the Qie Yun, and followed in the subsequent mainstream tradition. Hence any one rhyme had potentially either three or four tonal variants according to whether it was closed or open, for example: 285

CHINESE PHONOLOGY closed type open type

kam ka

karri ka

kam ka

kap

(34 rhymes) (27 rhymes)

Of the 61 rhymes, only 34 were of the closed type; the absence of a fourth, 'entering' tone with the remainder is thus presented as a systematic gap. There are also accidental gaps in the other three tones, since none of the 61 rhymes occurs in all three; there are 57 in level, 55 in rising and 60 in falling tone. This, together with the 34 in stopped tone, makes up the total of 206 yun (finals) referred to above. The number of characters (morphemes) in each rhyme is of course very variable; the most numerous category is the level tone. The initials were not tabulated, since they did not form part of the rhyme; but systematic study of the fanqie renderings by later scholars revealed that 32 initials were kept distinct; and if the interpretation was taken strictly on the basis of the first character used in each rendering, ignoring 'chains' of renderings that arise because the same character may be used to render both an initial and a final, it could be shown that some further distinctions were actually being made, resulting in an inventory of 47 initials in all. The difference arises because 15 out of the 32 initials can occur both with palatalized and with non-palatalized finals; the two are, of course, clearly set apart in the finals themselves, but the distinction is also treated as a feature of the initial consonant, so that in the fanqie rendering of a palatalized syllable the initial character is also regularly one that stands for a palatalized syllable; and likewise with the nonpalatalized. Similarly with the finals: if one takes the fanqie renderings rather than the tabulation it turns out that the number of finals actually distinguished is 290 (90 if the tonal categories are ignored) instead of the 206 (61) appearing in the tables. In other words the theory that is embodied in the fanqie practice makes more distinctions than are recognized in the classification scheme; especially if only the direct renderings are taken into account. The fanqie technique clearly had the potential for being developed into a systematic notation. Theoretically it would have been possible to select a set of 32 characters to represent the initials and 206 for the finals and use these all the time. But instead of this the scholars preferred to vary the characters used in the fanqie renderings. This was partly no doubt, as suggested by Wang Li, to preserve traditional renderings, and also to avoid using any character in the rendering of itself. But the fact is that, in the native tradition, no need was felt for a more systematic notation. This same reluctance to conceive of a character as ever being a 286

CHINESE PHONOLOGICAL THEORY purely phonetic symbol is evident throughout Chinese history; in the fourteenth-century Sino-Mongolian script, in which Chinese characters are used to represent Mongolian, over twice as many characters are used as the actual number of syllables recognized as distinct. The same principle holds in the transliteration of foreign names in Chinese today. So reliable and consistent was the work of the early phonologists that, despite the absence of any concrete phonetic descriptions, the fanqie renderings of the Qie Yun and its revisions have been the primary source for the reconstruction of Middle Chinese. Bernhard Karlgren, the Swedish scholar who first systematically reconstructed the sound systems of both Middle Chinese (AD 500-700; his term is "Ancient Chinese") and Old Chinese (800-600 BC; his term is "Archaic Chinese"), wrote (1940: 3): "By a lucky chance, rich sources of highly varying kinds (rhyme dictionaries, foreign transcriptions, loanwords in Korean and Japanese and the testimony of strongly diverging daughter dialects) combined to throw light on one definite stage of the evolution of the Chinese language: the language spoken in Ch'ang-an [Changan], the capital, in Suei and early T'ang time (sixth and early seventh century AD). This language [is] most fully represented by the dictionary Ts'ie yiin [Qie Yun]." Table 5.1 shows the reconstructed 32 initials of the Qie Yun system, as embodied in the fanqie renderings. If we count as distinct those occurring in palatalized and those occurring in non-palatalized syllables, shown separated by a slash /, the total becomes 47. Karlgren's Ancient Chinese has 37, since he keeps apart tj, etc. and tjj, etc. as separate sets ("supradental" and "palatal" affricates and fricatives), and likewise j and 0, but not the other pairs of palatalized and non-palatalized. Table 5.2 shows the reconstructed 61 finals of the Qie Yun system. These represent the ones that are tabulated. Taking account of all finals normally kept apart in the fanqie renderings, shown by a slash, the total becomes 90. It will be noted that these additional distinctions are largely Table 5.1

P/PJ t

t

ts

tj/tjj k/kj ?/?j 0/j

Reconstructed Qie Yun initials

P'/P'J t' f ts' tJVtJj kVkj

m/mj

bVb'j d' $ dz ds'/ds'j

Ji

sri

JjA)j

n

nsj

287

1/lj s

z

J/JJ 5C/5CJ

3J YJ

CHINESE PHONOLOGY Table 5.2 Reconstructed Qie Yun finals urj,-k / jurj,-k uog,-k jwog,-k or),-k ie/wie i/wi e ei/wei jwo ju uo iei/iwei jaei/jwaei a:i/uaii a:i/wa:i ai/waj uai ai jwai jen,-t/jwen,-t juen,-t jsn,-t jusn,-t jsn,-t an jan,-t/jwan,-t an,-t uan,-t a:n,-t/wa:n,-t an,-t/wan,-t ien,-tiwen,-t jsen,-t/jw3en,-t ieu jaeu (iau) au a ua/jua a/ja/wa jar),-k/jwar),-k ar),-k/war),-k ar),-k/jag,-k/war),-k/jwag sr),-k/wsg,-k jaer),-k/jw3er),-k ieg,-k/ iwer),-k J9i),-k/jw9k 9rj,-k/u9r),-k J9U

9U

i9U

J9m,-p am,-p a:m,-p jaem,-p iem,-p am,-p a:m,-p jam,-p jwam,-p a matter of labialized/non-labialized, except with a back vowel where they are palatalized/non-palatalized; these were later elaborated into a theory of 'grades' (see below). The 16 major groupings shown in the table are the she49 'classes of final' recognized in the later tradition. In the tables, I have used a simplified form of representation designed to illustrate the achievements of the early phonologists, and to give some idea of the distinctions that they were aware of, rather than to give an accurate account of reconstructed Middle Chinese. For Karlgren's work on Ancient (Middle) Chinese, see his Etudes sur la Phonologic chinoise (1915—24) and 'The reconstruction of Ancient Chinese' (1922); for his work on Archaic (Old) Chinese, and on the origins of the characters, see Grammata Serica (1940). A great deal of further work has been done following up Karlgren's achievements and revising and amplifying his conclusions, but his interpretation remains fundamentally valid (cf. Chao 1940). It has been pointed out in the discussion of the initials and finals that there were a number of secondary distinctions embodied in the contrasegmental phonology, though less consistently or explicitly; and that these all relate to variation in the y/w prosody of the syllable. This pattern of prosodic systems was a marked feature of North Chinese in Changan times, and still is today. The intuition of the fanqie scholars was to show y-prosody twice over, as a feature both of initial and of

288

CHINESE PHONOLOGICAL THEORY final (cf. Chao 1940: 207; Downer 1963: 129); in this they show an awareness of the fact that it colours the whole syllable. On the other hand w-prosody can be treated simply as a feature of the final; there are no pairs of initials that have to be interpreted as (independently) labialized/non-labialized. Exactly the same is true in Modern Pekingese, where a systematic interpretation requires palatality as a term in two distinct systems (cf. Halliday 1959: Appendix A): i) a prosodic system of 'open /y/w/yw (11)' in the syllable as a whole ("which can, of course, be represented as a feature of the final), as well as ii) a distinct system of'y/ non-y' in (certain of) the initials - although the actual sets of morphemes falling into these various classes are now quite different. Such was the phonological theory that evolved, first as an adjunct to lexicology and palaeography and then as an interpretation of poetic form, in the native Chinese scholarly tradition. It had almost no labelled categories, but was beginning to make use of tabulation, at the end of the Qie Yun but also already in the Yu Pian50 'Jade Tablet' by Gu Yewang51 which appeared in 543. Although it would be foolish to assert that this tradition was totally free of all outside influences, since aspects of Buddhist thought had been known in China for a long time, it was clearly indigenous to China in all its essentials. During the Tang dynasty (seventh to tenth centuries), however, a major influence came into Chinese phonology from outside China. At this period Indian phonetic theory was introduced, and this led to a considerable shift in orientation. The Indian tradition was explicit: it involved labels for phonetic categories, and also the idea of notation — derived from the nature of Indian writing, which is of course phonological. But Indian linguistics had, naturally, evolved in the study of Indian languages, and these languages make some phonological distinctions not found in Chinese and ignore others that are found in Chinese. The effect on Chinese phonological theory was therefore somewhat mixed. It is beyond the scope of this paper to trace the history of subsequent developments. A summary may be found in Downer's 'Traditional Chinese phonology' (1963, especially pp.131-40). On the basis of Indian concepts, during the Tang dynasty the initials and finals of the Chinese syllable were step-by-step classified and described. The concept of niu was superseded by one of zimu;52 this originally meant simply the character used for writing the niu, but then evolved to mean 'category of initial described phonetically and designated by a standard symbol' (Downer's 'initial matrix'). These are said to have been first enumerated by Shou Wen53 in the Sanshiliu Zimu Tu54 'Charts of the Thirty-six Initials' some time in the ninth century. Shou Wen selected certain of the 289

CHINESE PHONOLOGY

characters used as the initial element in the fanqie renderings and used them as category headings. Since the starting point was Indian phonetic theory, some of the distinctions made in Chinese were ignored, although the 36 initials also incorporated certain new ones, referred to below. Following the Indian categories, the initials were labelled as follows chun yin55 she ym56 chi ym57 ya ym58 i ' - 59 hou ym

'labials' 'linguals' 'dentals' 'molars' 'glutturals'

P t ts k

P' t' ts'

V d' dz

k'

Q

.X

Y

j

m n s r) ?

5 z

r tj

41 tt)'

Ji zm (jLz'e 4>zi3r) b

cjLz u

cjL^'a

4=z'5

4>Z'DO

4,^1 cj,zi3

CJUi

4Z.1 4^uBn u

4z^ Bs 4^ua 4^uu3n ag

2 00

4^?

4^.?i 4^er)

4^e

4z,°u 4zi.u u^n

zy£en 4>zyn

<jlzyffi 4-zy

4>zyaes (jL^y'n

4>zy0 cj,zyur)

(jLzyu

cjU'ug

4^u'n

4z.o

Chart 6.2 Phonetic realization of syllables with palatoalveolar (j-) and retroflex (zh-) initials, showing variation in rendering of the finals. Notes: (1) On-line phonetic symbols are those of the IPA, except that 12 has been used here instead of OE for the open front rounded vowel, as instead of a for the open front spread, with a used for maximally open position. (2) Raised phonetic symbols represent generalized postures towards which, through which or away from which the vowel moves in its glide. (3) Diacritics show ~ 'nasalized', T'lowered', praised', ^'fronted' and _ 'backed' variants of the vowel in question.

299

CHINESE PHONOLOGY Here the effect on the vowel nucleus is more pronounced, since in each case the preceding and the following margins are exerting the same force: both are y-prosodic (i- and -n), or else both are w-prosodic (u- and -ng). The vowel nucleus is further from the open position, and it does not vary significantly in the degree of openness attained. Here there are no corresponding syllables with non-nasal final: there are no forms jiai or zhuao. The syllables in row 30-34 present special problems which I shall return to later in the study; but in respect of the system of posture they are the same. Once we have recognized that -n is y-prosodic, like -/, and -ng is w-prosodic, like -o, we shall expect the nasal finals to exert the same force on the nuclear vowel as the oral finals do. The generalization that the phonetic quality of the nucleus of the syllable — that is, the range within which it can vary — is determined by the posture at the peripheries applied throughout the phonological system, and not only with the vowel written a. It makes no difference whether the final is nasal or oral; because of their y-/w-prosody, -V (final vowel) and -N (final nasal) are identical in their effect on the preceding syllabic nucleus. In terms of posture, therefore, we find nine distinct environments. These are shown in Table 6.1, where the symbols are in phonemic notation. In the next step, we will represent this in prosodic notation; but first we need to note the other significant dimension of variation that occurs just with the nasal finals, namely variation in the way the nasality itself is manifested. In syllables written with final nasal consonant (row's 1, 3, 10, 13, 21, 24 in Chart 6.2), the onset of nasality may take place at any time following release of the initial consonant; and there may or may not be tongue contact. The variant without tongue contact is relatively more frequent with the alveolar nasal than with the velar; but both forms occur both with and without contact, so that the only constant feature is the opposition between nasality and orality. Thus nasal/oral constitute another prosodic system; we will refer to it as the system of RESONANCE (cf. Catford 1982:138). We can now see the difference in what is implied by a prosodic and a segmental interpretation. Consider the syllable zhan, in contrast (i) with Table 6.1

open front back

The nine possible environments of/a/ [in phonetic notation] preceding # i u

a

300

following # i, n u,g

A SYSTEMIC INTERPRETATION OF PEKING SYLLABLES zhai and (ii) with zhang. The pair zhan/zhai may have identical vowel quality, differing only in that one is nasal, the other oral. On the other hand, the pair zhan/zhang may have identical nasal resonance (no tongue contact, hence no segmental realization of alveolar/velar), differing only in that one has a fronted vowel, the other a backed vowel. Likewise, zhao/zhang have the same posture with different resonance; zhao/zhai the same resonance with different posture. It is difficult to explain this in segmental terms, with a segment /a/ followed by the four distinct segments /i/, /o/, /n/, /ng/. But it becomes predictable if we explain these four as the product of two intersecting two-term prosodic systems: posture (y-prosody or w-prosody) and resonance (nasal or oral). Returning to Chart 6.2, however, we have also to accommodate the syllables ending in a, namely jia zha zhua (rows 2,12,23). Here the vowel written a attains its most open quality; and there is very little variation, either within the syllable or even across all three. What variation there is, however, turns out to be similar to that found throughout the set of examples - as a function, not of the final posture but of the posture associated with the vowel onset. Again, we have found this represented in alphabetic writing as a segment: i in jia, u in zhua and neither of these in zha. (For the alternation j/zh see below.) We therefore need to modify the prosodic system of posture in two respects. In the first place, we need to add a third term representing the open posture, neither y-prosody nor w-prosody: let us call this 'a-prosody'. These give us the three basic postures out of which our human speech has evolved and which can be observed in young children's protolanguage: tongue lowered, lips open (a-prosodic); front of tongue raised to front, lips spread (y-prosodic);back of tongue raised to back, lips rounded (w-prosodic). In the second place, we need to recast this in terms of movement through the syllable, so that instead of one choice of posture we need two, one for the beginning of the syllable and another for the end. Table 6.2 The 18 possible syllable types with vowel a, showing the 13 which actually occur (in prosodic notation)

y/ Oral Nasal

y-a

y-y

a/

y-w

a-y

y-w

a-y

301

a-a

w/

a-w

w-y w-a

a-w

w-y

w-w

CHINESE PHONOLOGY Table 6.3 spelling)

The 13 syllable types with vowel a (open aperture) (in Pinyin

y-y y-a Oral Nasal

a/ a-a a-w

y/

M

jia

jian

w/ w-y w-a w-w

y-w

a-y

jiao

zhai zha

zhao

zhuai zhua

Jiang

zhan

zhang

zhuan

zhuang

This means that we now have 18 possible syllable types, defined by the prosodic systems encountered so far: three initial postures by three final postures by two resonances. Table 6.2 shows these 18 prosodic patterns. It also indicates that, of these 18 possibilities, 13 actually occur. Table 6.3 shows these 13 syllable types in Pinyin spelling, using the initial consonants from Chart 6.2. The five syllable types that are excluded are: (1) nasal resonance with open (a-prosodic) posture (three types); (2) oral resonance with stable front (y-y) or back (w-w) posture (two types). The prosodic profile of each of the 13 syllable types that do occur describes the phonetic space-time co-ordinates within which it is found to vary. Figure 6.1 gives a simple network for generating this paradigm, using as entry conditions elements corresponding to initial and final in the Chinese phonological analysis. However, it generates all the 18 syllables envisaged as possible in Table 6.2, whereas as we have seen only thirteen of these occur. We therefore rewrite Figure 6.1 in the form of Figure 6.2, which specifies just the required 13. These are all the possible syllable types that have an a in the spelling, but without taking account of initial consonants - that is, corresponding to one column (or two halfcolumns) in Chart 6.2. In terms of Chart 6.2, they are the first four in Block i, the ji- block; the first five in Block ii, the zh- block; and the first four in Block iii, the zhu- block. But although Figure 6.2 gets the right answer, it is somewhat arbitrary; there seems no pattern in its constraints. We can therefore replace it, in turn, with Figure 6.3. By introducing the system 'stable/shifting', this shows the gaps to be systematic: you can choose between nasal and oral resonance only if you shift either from non-y (a or w) posture to y, or from non-w (a or y) posture to w. Otherwise, you have either oral only or nasal only: oral if ending with open posture (there is no a-prosodic

302

A SYSTEMIC INTERPRETATION OP PEKING SYLLABLES

Figure 6.1 vowel a

Network specifying 18 theoretically possible syllable types with

Figure 6.2 vowel a

Network specifying 13 actually occurring syllable types with

303

CHINESE PHONOLOGY final nasal), nasal if the posture is stable at y-y or w-w (that is, if you maintain the posture you must change the resonance; note in this connection that the initial nasals n-, m- operate in a different system, not one of resonance). (It is natural to ask whether in those dialects which have preserved the three nasal finals -n, -ng, -m the -m carries the aprosodic 'open' posture. But as far as I know none of these dialects shares this Mandarin pattern by which nasal and oral finals are posturally matched.) The network combines four principles of analysis. One is the Chinese phonological principle whereby all syllables are structured simply as initial plus final. The second is the Firthian prosodic principle whereby features such as posture (y/a/w) and resonance (nasal/oral) are treated non-segmentally. The third is the paradigmatic principle whereby features are interpreted as terms in systems, each system having a specified condition of entry. (Note that in Firthian system-structure theory the entry condition is specified syntagmatically, whereas in a system network it is specified paradigmatically: entry to one system depends on selecting a certain term in (at least one) other.) The fourth is the dynamic principle whereby the syllable is envisaged as a wave, a periodic pattern of movement characterized by a kind of 'flow-andreturn'. What this last means is that the syllable is construed as a movement from an initial state to a final state; each of these states is specified as a 'selection expression' (a cluster of features from different prosodic

Figure 6.3

Revised version of Figure 6.2 showing system of'posture shift' 304

A SYSTEMIC INTERPRETATION OF PEKING SYLLABLES systems); and there is variation both temporally, in the extent to which a particular feature persists across the syllable, and spatially, in the route that is traversed from the initial to the final state. The evidence so far, on the basis of the vowel written as a, suggests that the final state exerts the greater force: the shape of the syllable at its peak tends to anticipate where it is going. This is of course implied by saying that all syllables have CV structure, and it will turn out to be true of the Pekingese syllable as a whole. (It is not true, however, of all Chinese dialects; in Cantonese, for example, the features selected at syllable final have little effect on the quality of the vowel.) So far, however, I have considered only those syllables whose trajectory passes through an open, a-like intermediate position. We must now explore the other finals in the Pekingese syllabary, those that have no a in their spelling in the Pinyin transcription. These syllables display a variety of different spellings: see rows 5-9, 15-20 and 25-29 in Chart 6.2. But if we apply the same principles of analysis, we shall find that all except those in the last row of each block (ji, zhi, zhu) have one and the same vowel nucleus. The phonetic value for each syllable type is shown as before in Chart 6.2, together with the observed range of variation. For quicker reference, a typical value for each of the different syllable finals is shown in Table 6.4, with Pinyin spelling above the line and phonetic representation below. The phonetic range covered by these syllables is considerable; they range over most of the upper half of the traditionally recognized vocalic space. This is reflected in the Pinyin spellings, which use e, i, o and «,both singly and in digraph combinations. Some other transcriptions also use trigraphs such as uei and iou. But we can interpret all this variety in terms of the same prosodic systems as we used to explain the values of the a vowel, recognizing one and the same vowel nucleus in different prosodic environments. There are three syllable types which end as a monophthong at a height which we can call half-close (Catford 1982:176): front spread [e], Table 6.4 Typical phonetic values for the 13 finals with half-close aperture (Pinyin spelling above the line, IPA below) -e -ie y

e

-uo

-ei

-en

-ou

-eng

-in

-iu

-ing

-ui

-un

-ong

o

ei

§n

ou

eg

in

iag

fg

uai

usn

ug

305

CHINESE PHONOLOGY back rounded [o] and back spread [Y] . These show no tendency to move towards a 'close' position; in this they are parallel to final -a, having neither y-prosodic nor w-prosodic posture at syllable final. The distinction among the three is determined by initial posture alone, which has a much greater effect here than with the open vowel: the y-prosodic initial gives [e], as in jie; the a-prosodic initial gives [Y], as in zhe; the w-prosodic initial gives [o], as in zhuo. This simply reflects the greater possible variety of tongue and lip configurations available at this degree of vowel closure (cf. the shape of the "vowel triangle/quadrilateral"); there is more space in which to move. Four syllables, those spelt -ei, -en, -ou, -eng, have a half-close vowel as the nucleus. The spelling suggests that two of them form closing diphthongs while the other two are monophthongs followed by nasal consonant. In fact, however, the nasal syllables closely parallel those with the a vowel: the nasality may begin at any point, and there may or may not be consonantal closure (obstruence) at the end — again, it is more common with -ng than with -n. The vowel of the nasal syllables is more central than those of the oral syllables; but it is clearly fronted before -n, often with a glide towards close front, and backed before -ng, sometimes with a (slighter) glide towards close back. Thus the nasals have exactly the same prosodic values as those of the a syllables; the final part of the syllable does not consist of two segments, vowel plus nasal, but is a combination of two final prosodies, nasal resonance with y or w posture. The oral finals tend to begin less centrally, but they show considerable variation towards the centre: -ou, in particular, is often heard as [su], curiously like its British English analogue. Thus zhei, zhen, zhou, zheng are prosodically identical with zhai, zhan, zhao, zhang, but with half-close vowel instead of open: they begin with open posture (a-prosody) and end with y or w, oral or nasal. This leaves the six finals -in, -iu, -ing; -ui, -un, -ong. Of these, the oral pair -iu and -ui are the half-close analogues of the open oral pair -iao and -uai. Both pairs move prosodically from y to w, or from w to y - that is, they shift to the opposite posture; but while -iao and -uai follow a trajectory via the 'open' region of vocalic space, -iu and -ui make the same postural shift but with the trajectory through the 'half-close' region. There is obviously a broad band through which such a traverse can be made; interestingly, the route taken tends to depend on the tone, and -iu and -ui are in fact the only syllable types in Mandarin that display tonally regulated variation, as shown in Table 6.5. (The spellings -iou and -uei are not used in Pinyin, but this systematic tonal variation is recognized in some other transcriptions.) 306

A SYSTEMIC INTERPRETATION OF PEKING SYLLABLES Table 6.5 Typical pattern of tonally correlated variation in the phonetic realization of y-w and w-y finals with half-close aperture (Pinyin and variant spelling on the left, IPA on the right) tone 1 tone 2 tone 3 tone 4

-m

-1OU

-ui

1U

-uei

i3u iou ieo

ui u3i usi use

The nasal finals form two sharply distinct pairs. Two of them, -in and -ong (in jin, zhong), show almost no movement and no variation in vowel quality; they are somewhat opened-up versions of the two close cardinal vowels. The only variation they display is in the nasality: when it sets in, and whether or not there is obstruence; in other words the pattern is the same as we have found with nasality throughout. There is perhaps slightly greater tendency for final tongue contact here than with the -an, -ang nasals, but it is still by no means categorical. The other two nasal finals, -ing and -un, are very different. Here there is both movement and variation. There is the same variation in the nasality as elsewhere; but there is also variation in the transition to the final posture. The vowel in -un is opened (lowered) and then fronted, in varying degrees, so that it sounds more like the vowel in English jewel, or even ruin, than like that in full or rune. Likewise the vowel in -ing is opened (lowered) and then backed, like English young but with the diphthong falling instead of rising (Catford 1982: 216). It is not difficult to see what is happening here, once we interpret in terms of the prosodic system of posture. As always, final'-«' is y-prosodic nasality, and final '-ng' is w-prosodic nasality. In zhen, zheng the postural transition is a -y, a-w. In the four syllables jin, zhong, jing, zhun, on the other hand, the initial posture is either y or w; hence with nasality there is the possibility of either stabilizing the posture (y-y, as in jin; w-w, as in zhong) or shifting (y-w, as in jing; w-y, as in zhun). When the posture is stable, the transition is simple, with little movement or variation. When it is shifting, the transition is complex; there is considerable movement, and hence great latitude for variation in the trajectory adopted. But the initial and final states are what constitute the essence of the syllable. Thus, whereas in English the peak of resonance in the syllable - the vowel nucleus - is also the most 'fixed' part, so that in a set like seen, soon, sing, soong, the vowel posture is projected outwards on to the initial and final consonants, in the Mandarin syllable it is the other way round: the vowel 307

CHINESE PHONOLOGY 'nucleus' is simply a degree of aperture, and the initial and final postures of the syllable are projected inwards to create a movement within this broad band of phonetic space. We have thus recognized a multiple proportionality, (1) within the half-close vowel: ui: iu :: un : ing ::: [both w-y : y-w] ei: ou :: en : eng [both a-y : a-w]; (2) and between all of these and those with open vowel: uai: iao :: uan : uang ::: [both w-y : y-w] ai :ao ::an :ang [both a-y : a-w]. What is striking is that this same proportionality extends right throughout the system; the half-close vowel series shows the same five finals remaining unpaired in respect of nasal/oral resonance: (3) oral only: ie : e : ia a : (4) nasal only: in :ong:: ian : uang

uo ua

::

[y-a: a-a: w-a] [y-a: a-a: w-a] [y-y:w-w] [y-y : w-w]

Thus out of the 18 finals that are theoretically possible as combinations of initial posture, final posture and resonance, exactly the same 13 occur with the half-close "vowel as with the open vowel; moreover the realization of the several terms in both these prosodic systems is entirely analogous throughout. If we represent the two vowel spaces, the open and the half-close, as a third prosodic system of APERTURE, using V for open and 3 for half-close, then for all the 13 prosodic profiles of initial posture, final posture and resonance there would be a constant proportionality such that V 1 :3 1 ::V 2 :3 2 ::V 3 :3 3 ::. . . Table 6.6 shows the syllable finals with the 'half-close' aperture together with their prosodic values for resonance and posture; note the identity between this and Table 6.3. It is not difficult to accommodate the halfclose series within the same system network. All we need to do is to modify Figure 6.3 as Figure 6.4. We have now accounted for the finals in all rows in Chart 6.2 except the last one in each block, namely ji, zhi, zhu (rows 9, 20, 29). There is almost no variation in the pronunciation of these syllables (at least among people brought up in Peking - there is a great deal of variation 308

A SYSTEMIC INTERPRETATION OF PEKING SYLLABLES Table 6.6

g

The 13 syllable types with half-close aperture (in Pinyin spelling)

y-w

Oral

y/ y -y y-a Jie

Nasal

jin

a-y

a/ a-a

a-w

w/ w-y w-a w-w

Jiu

zhei

zhe

zhou

zhui zhuo

Jing

zhen

zheng

zhun

zhong

Table 6.7 Typical phonetic values for the three finals with close aperture (Pinyin spelling above the line, IPA below) -i [in zhi chi shi ri]

-i[mjiqixi]

-u [in zhu chu shu ru]

1

i

u

among non-Pekingese speakers in their rendering of the 'vowel' in zhi!). Here there are three entirely distinct vowel qualities: [i] in ji, [u] in zhu, both very close; and [i] in zhi. It is the last of these, in fact, that provides the clue to their phonological status (see Table 6.7). It will be seen that, in Chart 6.1, there are two series of syllables in row 1 written with the letter /: zhi chi shi ri, and zi ci si. The vowel in all of these is an apical vowel: the tip of the tongue remains in the position of the initial consonant and is relaxed just enough to allow vocalic release. The lips take up a neutral posture at the start of the syllable and remain unchanged throughout. Since zhi chi shi ri are retroflex, while zi ci si are dental, the acoustic effect of this vowel in the two series is very different; but its prosodic profile is the same in both. If we now consider the vowel in ji qi xi (row 12 in Chart 6.1), we can describe it by exactly the same formula. These consonants are palatoalveolar, so the vowel cannot be called apical; but it is entirely analogous to an apical vowel, since it is the vowel that is produced by minimally relaxing the tongue away from the initial consonantal position. When this vowel follows the other initials in that row, iri bi, di, ni, etc. there is of course movement away from the place of obstruence; but the vocalic posture of tongue and lips is established at the beginning of the syllable. In other words, all syllables with [i] (that is, those written with -/ in Pinyin except the retroflexes and dentals zhi chi shi ri, zi ci si) are y-prosodic from the start. 309

CHINESE PHONOLOGY Likewise in all -u syllables (row 21 in Chart 6.1), the w-prosodic posture, with the back of the tongue raised and lips rounded, is established at the beginning of the syllable. The articulatory organs simply remain in place to produce the close back vowel. We can now account for the remaining three finals of Chart 6.2, namely -i as in ji, -/ as in zhi, and -u, in terms of our prosodic systems. In syllables with these finals, there is no prosodic movement: the initial posture is maintained throughout. So instead of nine possibilities there are only three, a, y, w (i.e. a-a, y-y, w-w). Moreover there is no prosodic system of resonance; such syllables are oral only. The vowel represents a third term in the system of aperture, namely 'close', which we can write as T. One could interpret these syllables as having no vowel in their structure; but, apart from destroying the powerful generalization that all syllables have identical structure CV (initial + final), this would lead us to predict that such syllables would be toneless, whereas in fact they display the same system of tone as all the other syllables. More interestingly, we could treat the finals of ji and zhu as the missing y-y and w-w terms in the half-close system, with the idea of

Figure 6.4 Network specifying the 26 syllable types with open and -with half-close aperture 310

A SYSTEMIC INTERPRETATION OF PEKING SYLLABLES eliminating the need for a close vowel altogether. But it would not in fact do that, because we could not account for zhi in this way as there is already an a-a final in both the half-close and the open series (zhe, zha). It would also distort the analogy between the half-close and the open series (Tables 6.3 and 6.6) because (as we have seen) there are no comparable y-y or w-w oral syllables in the latter; and it would make the wrong predictions phonetically — the one kind of variation that is found in the I-series is a slight increase or decrease in the amount of friction generated; here -/ (palatoalveolar, in ji) and -u clearly belong with apical -/, in that they can have considerable friction (especially in the syllables yi and wu), whereas friction is never found elsewhere in the 3 series. So the finals of ji, zhi, zhu constitute a third sub-system, having close vowel but no further prosodic paradigm beyond their initial selection of posture y, a or w (Table 6.8). The system is now as shown in Figure 6.5. Note that syllables with close finals (the I-series) select only in the initial posture system; they can enter no other system in the network — but they pick up the feature 'oral' in the resonance system by default. This network generates the 29 finals we have been considering: with reference to Chart 6.1, all those except Block iv. More importantly, it does so in a way which naturally predicts not only the form of the phonetic output but also the kind of variation that is associated with it. Two points should be stressed here. One is that the network so far specifies only the finals; although it states the prosodic posture (y/a/w) of the initials, it does not yet incorporate the system of initial consonants. Second, it specifies the finals entirely as complexes of features in prosodic systems, without the need for any phonematic units. The question arises whether the initials can be treated the same way, as a network of prosodic systems; but that will have to be explored in another study. Let us turn now to Block iv, rows 30—4 in Chart 6.1. As can be seen, these form a very limited series. Phonetically they begin with a front rounded posture; close, not open (i.e. clearly not a-prosodic), but combining the tongue position of y-prosody with the lip position of w-prosody. We could provisionally consider them as a fourth term in the posture system, labelled TJ. But in order to handle these we shall need to Table 6.8

The three syllable types with close aperture (in Pinyin spelling)

1 Oral

V-

a-

w-

ji

zhi

zhu

311

CHINESE PHONOLOGY

Figure 6.5 aperture

Revised version of Figure 6.4, including syllables with close

take one or two steps towards a theory of the initial consonants. So far we have concentrated on three sets of initials (Blocks IV, V and VI in Chart 6.1): y-prosodic j(i) q(i) x(i)

a-prosodic z cs zh ch sh r

w-prosodic zu cu su zhu chu shu ru

These are obviously not symmetrical, at least on the basis of the spelling; and this does reflect the phonetic facts, in that the dentals (z, etc.) and the retroflexes (zh, etc.) remain constant in their place of articulation whether a-prosodic (open posture) or w-prosodic (labiovelar posture), whereas the palatoalveolars (/, etc.) constitute in this respect a distinct set. This suggests a solution for Box IV/iv in Chart 6.1: that ju qu xu are labialized (w-prosodic) versions of ji qi xi. Which, indeed, they are; there is a proportionality such that jun :jin :: zun : zen. If we now turn to rows 30—4 in Chart 6.2 we shall find that all the variants predicted by this interpretation come out right. Note for example that juan rhymes with zuan, not with jian; and it is the zuan vowel that would be predicted if juan has initial w-prosody. (That is to say,juan is (dzysen), etc. not (d?ysn)). 312

A SYSTEMIC INTERPRETATION OF PEKING SYLLABLES I have never heard the second form from a Pekingese speaker; interestingly, however, it is sometimes heard in the Mandarin of speakers of other dialects, and is reported by Lock (1992) from Singapore. The yu group, Box VH/iv in Chart 6.1, follows exactly the same pattern. This analysis would give us, for the series of affricate/fricative initials, the pattern shown in Table 6.9 (cf. Ladefoged and Wu 1984). The question arises, however, of how to interpret the/- q- x- y- series of palatoalveolar initials in prosodic terms. Assuming that they carry a yprosody, is this inherent in their consonantal make-up, or are they to be interpreted as a variant of one of the other consonantal series in a syllable with y initial posture? Historically, they are a mixed group, partly related to the retroflex (affricate/fricative) series and partly to the velar plosives (Blocks V and III in Chart 6.1). That, at least, is the recent history of the morphemes which now have palatoalveolar initial; but this is simply the latest cycle in a process that has repeated itself at least three times in the known (i.e. reconstructable) history of Chinese phonology. This is far outside the scope of this chapter; but we need to establish some principle for determining their present status. We saw earlier (cf. Tables 6.3 and 6.6) that Cy (y-prosodic initial) excludes the finals, -ai, -ei, and Cw (w-prosodic initial) excludes -ao, -ou. Ifju- embodies some combination of y and w postures, it should exclude all four; and it does. But this suggests that, rather than seeing it as a fourth type of syllable (with a distinct q posture, as formulated above), we should treat it as a combination of y and w postures; so let us rewrite Cq as Cyw. Now, since Cy and Cw each occurs with nine finals, but excluding a different two, that should leave seven for Cyw. In fact, however, there are only five, because in the V group (open aperture) neither -a nor -w is found: there are no syllables jua, qua, xua, yua; and no syllables juang, quang, xuang or yuang. On the other hand, both -a and -w occur in the 3 group (half-close). Thus for the ju- series (including yu-, column 22 in Chart 6.1) the syllabary is as in Table 6.10, having only juan,jue,jun,jiong and ju. Note that all these syllables are Table 6.9 Affricate and fricative initials: place of articulation by initial posture (in Pinyin spelling) POSTURE

non-labialized

labialized

non-palatal (Blocks V, VI)

zh- z-

zhu- zu-

palatal (Block IV)

ji-

Ju~

PLACE

313

CHINESE PHONOLOGY

lip-rounded throughout; this includes that spelt jiong (also qiong, xiong, yong), which is typically realized as [cj,zyur)] — although there is a variant which begins unrounded, [(J^'on] dissimilating the two w postures. In some way or other, therefore, syllables of the ju- series have to be analysed as Cyw, that is, as having a combination of y and w postures. But there are two possible ways of interpreting this: (1) they could be generated as four initial prosodies each combining a term from two distinct systems; (2) alternatively, the entire j-, q-, x-, y-series of syllables (Blocks IV.ii, IV.iv, Vll.ii and Vll.iv in Chart 6.1) could be treated as having an initial consonant which is inherently palatal (and hence y-prosodic), and which then can have associated with it either a-prosody (in ji, qi, xi, yi) or w-prosody (in ju, qu, xu, yu). This is a substantive issue, reflecting a complex history of sound change, as mentioned above: there has always been a skewness between palatalizing and labializing, such that in some way or other palatalization has to be accommodated in the system twice over - in prosodic terms, both as a feature of the initial consonant and as a prosody of the initial element of the syllable — even though the morphemes involved in modern Mandarin are quite different from those in respect of which this pattern is first known to have appeared (Karlgren 1940; Wang 1980). The situation becomes complex at this point; once we start to bring in the initials we face a cluster of intersecting features which can be interpreted prosodically in more than one way, offering different explanations of the patterns involved. I shall not try to present the alternatives here because this would require a detailed treatment of all the initial consonant systems. Here I am taking account just of the problems that are raised by the finals; and when we include those that Table 6.10

The five syllable types with yw posture (in Pinyin spelling) yw/

yw-y

V,

Oral

v,

Nasal

3,

Oral

3,

Nasal

I,

Oral

yw-a

yw-w

juan jue

jiong

jun

Ju

314

A SYSTEMIC INTERPRETATION OF PEKING SYLLABLES need to be explained by postulating a complex initial prosody Cyw we have to bear in mind two factors: (1) Finals occurring with Cyw include only those which would be permitted both by Cy and by Cw (juai is excluded because (like jiai) it would be O/y-y; juao is excluded because (like zhuao) it would be O/w-w). (2) Finals occurring with Cyw are always realized phonetically as either one or the other (that is, like those with Cy or those with Cw), not in some distinct phonetic shape of their own. (But not always the same one: the type with V aperture is Cw-like (juan is like zhuan, N/w-y); whereas the type with 3 aperture is Cy-like (jue is like jie, O/y-a), although there is a Cw-like variant juo (like zhuo, O/w-a) used by some speakers with some morphemes only — i.e. the variation has a dialectal basis.) It seems therefore that we should prefer a three-term system of posture both at C (initial) and at V (final), rather than adding a fourth term for initial position only. Furthermore, since the ju- qu- xu- yu- series exclude Vw marginally more strongly than they exclude Vy (juan, quan, xuan, yuan occur, while juang, quang, xuang, yuang do not), it would seem better to treat them as having Cw (initial w-prosody) mapped on to an inherently y-prosodic consonant rather than the other way round. (This is in fact borne out by the detailed investigation of the initials, which also allows n- and /- to be accounted for in the same way (note the occurrence of Cyw with n and I in Chart 6.1, Block II.iv).) A network for generating the entire syllabary is shown as Figure 6.6. This takes account of all the points raised in the present study. But it should be made clear that it is not the network I would offer as the optimum systemic solution, because it does not take account of all the problems raised by the initials in their own right, and it leaves unexplained some gaps in initial + final (including tonal) combinations which can be shown to be systematic. There are of course some random gaps in the syllabary; I know of no way to explain the absence of syllables at shong or tiu, and I would not want to exclude them from the network. But it is clear that there is a suspicious similarity among some of the 'defective' boxes in Chart 6.1 (especially I.ii, II.ii, Il.iii and Vl.iii), and patterns of this kind seem to call for some explanation. Firth once remarked that if modern phonology had developed out of the tradition in China, we should never have heard of the phoneme with the implication that the world would have been a better place. I think there is an important place for the phoneme, as a potential member 315

CHINESE PHONOLOGY

Figure 6.6

Network specifying total Mandarin (Pekingese) syllabary

316

A SYSTEMIC INTERPRETATION OP PEKING SYLLABLES Notes The initials as represented in Pinyin spelling are specified by the systems in the network as follows (roman numerals refer to blocks in Chart 6.1):

continuant obstruent early late

flattened

pointed

z dental VI cerebral V zh palatal IV, II j

labial I velar III alveolar II

b g d

fricative early late

c ch

r

q

y "

p k t

approximant postures

r

w 0-

nasal lateral i a a

s sh x

n

f

m

VE

ii

1

h n

1

iii iv w w

a a a a

w

y y

w w

Initials n, 1 are interpreted as alveolar in Blocks i and iii, palatal in Blocks ii and iv. (Hence ni-, Ii- are palatal with a-posture, not alveolar with y-posture; and nil-, Iii- are palatal with w-posture.) Block VII are interpreted as voiced fricatives. With those in VH/ii-iv there can be considerable friction, especially with close aperture. Those in VH/i typically have voiced glottalic initiation; the variant with velar nasal, which I observed frequently in my original investigation, seems to be much less common today. Notes regarding unmarked (default) options: ** All 'voiced' (early voice onset) continuants are fricative unless palatal, labial or alveolar, which may be fricative or approximant. 1-1 All 'flattened' initials, and also the velars, if selecting 'narrow' posture can take only w-prosody, not y (either they are palatal already, or they cannot be palatalized). 2 2 — Labial initials, if selecting 'narrow' posture can take only y-prosody, not w (they are labial already). Thus, from 1—1 and 2~2 together, only alveolar initials can select either y- or w-prosody. # # — All syllables which retain the same narrow posture throughout (y-y or w-w) must have nasal resonance. *—* All syllables with close aperture, and all non-close with final a-prosody, must have oral resonance. Thus, from #—# and *—* together, only those syllables which shift into y- or w-prosody from somewhere else can select either nasal or oral resonance.

of a phonological rank scale; and the fact that speakers of certain languages chose to write them with an alphabet suggests that they also felt the same. But the Chinese did not; they developed a charactery instead; and the phonology of their language, while it has very clearly defined syllables with remarkably little indeterminacy, shows no real trace of phonemic structure. It should be said that this last is less true of some of the southern dialects — most notably Cantonese, where there is far less distance between a prosodic and a phonemic interpretation than 317

CHINESE PHONOLOGY there is with Mandarin. In Mandarin the two approaches give very different pictures; here, instead of taking the minimum articulatory segment as prototypical and reducing everything to that, we can take tone as our prototype and explain the entire system as a network of tone-like features. This is, in essence, what the prosodic analysis does; and it is what the Chinese phonologists were doing starting about two thousand years ago — even though it was another five centuries before they took account of tone itself as a distinct feature of the syllable (cf. Halliday 1981). The system network that has been being built up in the course of this study, up to and including Figure 6.5, generates the total set of finals for the syllabary of modern Pekingese, other than the tonal variants. That in Figure 6.6 incorporates the remaining features of the syllabic system but in a sketchy and provisional form. I will try to summarize the theoretical principles on which this systemic interpretation is based. (1) In accordance with Chinese phonological theory, the syllable consists of two loci, initial and final. As these (English) names imply, the two are lineally ordered; but they are not seginental. Rather, the syllable has an initial state, characterized by a syndrome of features; and a final state, characterized by another such syndrome — the movement from one state to the other is continuous throughout. (The Chinese terms do not embody the notion of linear ordering. The word for 'final' is 'rhyme'. The word for the initial is usually regarded as opaque; it meant 'twist' (like the twist in a cord to make a Chinese button), but also 'handle' for lifting with - perhaps a metaphor either of lift-off or of configuration.) (2) There is considerable variation in the enactment of the syllable, both among different speakers and within one and the same individual speaker. This variation is explained as a spatio-temporal dynamic: (i) initial, opening features may be more, or less, prolonged; (ii) final, closing features may start earlier or later; and (iii) there may be more than one route from the one to the other. (3) The initial prosodic systems have been presented only in outline, without explanation, to enable the network to be completed. The systems are: (a) ALIGNMENT (PLACE): pointed : labial / velar / alveolar // flattened : dental / cerebral ('retroflex') / palatal; (b) MANNER: obstruent /// continuant: fricative // approximant: nasal / lateral; (c) VOICE ONSET: early ('unaspirated') / late ('aspirated'); (d) POSTURE:y-prosodic / a-prosodic / w-prosodic. 318

A SYSTEMIC INTERPRETATION OF PEKING SYLLABLES (4) The final prosodic systems are: (e) APERTURE: close [I] / half-close [3] / open [V] (f) POSTURE: y-prosodic / a-prosodic / w-prosodic; (g) RESONANCE: nasal [N] / oral [O]; (h) TONE: high level [1] / mid rising [2] / low rising [3] / falling [4](5) The system of POSTURE figures both at initial and at final; hence a syllable may either retain the same posture throughout or shift from one posture to another. It turns out that this option, in the context of the opposed postures y and w, is best interpreted as itself forming a prosodic system: (i) POSTURE SHIFT: stable / shifting. (6) Not all features are equally variable, in the ways described in (2) above. But it is not misleading to represent even the apparently segmental features of the initial consonants in these prosodic terms. For example, it is often noted that in Mandarin the aspirated syllables 'may be very heavily aspirated', which means there is variation in the timing of the onset of voice. And the evolution of the apical vowel seems to suggest a tendency to prolong the tongue contact at the place of articulation. (7) It is not suggested that in every language the syllable would be best interpreted in entirely 'prosodic' (non-segmental) terms. In English, for example, there seems no convincing argument for saying that all syllables have the same structure or for trying to specify all consonant and vowel features as syllabic prosodies. Matthiessen (1987) presents a systemic interpretation of Akan phonology which includes separate networks for the syllable and the phoneme, with the latter further subdivided into consonant and vowel. It is to be expected that for some languages there would be networks at both syllable and phoneme rank. In Chinese, however (at least in Mandarin), all syllables have the same generalized structure, and there is no value in setting up the phoneme as a separate unit. The one syllable I have not discussed is the maverick syllable er, phonetically [srj, which appears as vowel plus consonant, the 'consonant' being a (very) retroflex frictionless continuant. At one stage in its history this was a palatoalveolar nasal with close aperture, systemically where ni is today; now, however, it has no trace of either nasality or palatalality and might be interpreted systemically in one of two ways: either as the 'close' member of the a-prosodic series with initial semi-vowel (going with yi, wu and yu; this is how it is located in Chart 319

CHINESE PHONOLOGY 6.1) or as the occupant of the empty ra slot. In fact in typical Pekingese speech its vowel tends to be open, rather than close or even half-close, which suggests that it is felt more akin to the latter. It is actually quite like ra pronounced backwards — a variant which is presumably within the limits of what we should expect, if the syllable is interpreted as having no linear segmentation. One major variable among phonological systems that is foregrounded in a prosodic and systemic perspective is where they make contact with the grammar (cf. Hill 1966). The phonological rank scale may include tonal and/or rhythmic units which may be mapped more or less consistently on to clauses or phrases; and in many languages the word is the point of origin for certain phonological systems, either with or without being fully integrated into the overall phonological structure (Matthiessen 1987; Prakasam 1987; and cf. the phonological hierarchy of tagmemic theory). In Chinese the word has hardly any phonological significance: none at all in many dialects, a little in Mandarin because it defines an environment within which the tonal system may be neutralized. But throughout the known history of the language there has always been an overwhelming association of the syllable with the morpheme, and this gives an added significance to the syllabary as the basis of the phonological system. It should be possible to synthesize the Mandarin syllabary from a systemic-prosodic description, with the prosodic features as the parameters and provision for varying their duration and the traversal between one setting and another. It would be interesting to compare these results with those obtained by synthesizing in phonemic and allophonic terms.

320

PART FOUR GRAMMAR AND DISCOURSE

This page intentionally left blank

EDITOR'S INTRODUCTION

This section begins with 'Grammatical Metaphor in English and Chinese' (1984). Both English and Chinese use lexicogrammar to construct scientific reality and signal the taxonomic organization of objects under scientific investigation. Both languages objectify by means of grammatical metaphor, using nominalization to realize processes (events, actions, states) as nouns, thereby giving them object-like status. In Chapter Eight, 'The Analysis of Scientific Texts in English and Chinese', (a paper originally presented at the International Conference on Research on Texts and Language, Xi'an Jiaotong University, March 1989; later appearing in Halliday and Martin 1993) Professor Halliday describes the features of 'scientific language' in English and Chinese. He shows how "the grammar of both Chinese and English, using similar resources (noun compounding, the nominal group, nominalization and grammatical metaphor), creates a form of discourse for codifying, extending and transmitting scientific knowledge," while "text structure establishes the highly valued form of scientific argumentation: what constitutes an acceptable canon of evidence, reasoning and proof. In the final chapter of this section, Chapter Nine, 'On the Grammatical Foundations of Dicourse' (2001), presented at the International Conference on Discourse Analysis held at the University of Macau in 1997, Professor Halliday considers the relation of discourse to lexicogrammar, both in English and Chinese, in terms of the following three questions: (1) How are grammatical units organized internally such that they form "parts" of a textual "whole"? 323

GRAMMAR AND DISCOURSE (2) How does the grammar set up semantic relations that create unity in a text? (3) Do the logical-semantic relations that link grammatical units also figure in relating one phase of a text to another? Those grammatical systems at work at the "micro", or clausal level, such as the theme-rheme and information systems, also may be seen to play a significant role in organizing the discourse as a whole. As Professor Halliday points out, "On the one hand, they are powerful resources for the creation of local meaning; on the other hand, they have at the same time a global significance for the semantic unfolding of the text." Similarly, conjunctive relations in the text "cover essentially the same semantic space as the logical-semantic relations that are construed structurally within the clause complex: the three types of expansion (elaborating, extending, enhancing)". Although, acknowledging the dangers of "modelling semantic patterns in terms of the theoretical concepts developed in lexicogrammar", nevertheless, he still regards it "as more fruitful than sidelining the grammar and setting up semantic categories solely on the basis of some rather vague appeal to meaning".

324

Chapter Seven

GRAMMATICAL METAPHOR IN ENGLISH AND CHINESE (1984)

The title of this conference ('New Papers on Chinese Language Use') refers to Chinese language use; and this implies that one focus of interest is on the teaching of Chinese as a foreign language, or to speakers of other languages. Now in the years when I was still a language teacher — before I turned from teaching languages to the much less daunting task of teaching linguistics — I had become interested in a number of applied linguistic topics which I was unable to follow up for lack of an adequate theoretical base. Some of these I have tried to explore subsequently in other contexts, but I have never lost sight of the aims and interests of a language teacher, and in the course of various other pursuits — such as the study of child language development, the linguistic interpretation of literature, machine translation and so on, I have found the same kind of need for an 'applied' grammatical theory — or one that, if not already applied, was at least applicable to these various goals. Here, therefore, I would like to pick up just one of those threads, and look at some examples of a phenomenon that arises in every language and poses a problem in all these spheres of activity, including for the learner, for whom, however, it is a source both of difficulty and of insight. I will discuss this first in relation to English and then take a sideways glance at Chinese, hoping to learn from others what may have been done in this area in the interpretation of the Chinese language. The phenomenon I would like to talk about has no general name; so let me give it one, and call it grammatical metaphor. It is related to 'Grammatical Metaphor in English and Chinese', from New Papers on Chinese Language Use, edited by Beverly Hong (1984). Copyright © Research School of Pacific Studies, The Australian National University.

325

GRAMMAR AND DISCOURSE

metaphor in the normal sense — in fact it is metaphor in the normal sense; but whereas metaphor as we usually understand it is a lexical phenomenon, illustrated by such familiar transferences as flow inflow of ideas, source in the source of the trouble, current in currents of opinion, tide in stem the tide of inflation, ooze in he oozes geniality and so on, the kind of metaphor I am referring to is grammatical rather than lexical, although in any particular expression we often find both of these coming together. Let me exemplify from English. Suppose we read in a report of a mountaineering expedition: The fifth day saw them at the summit.

There are various things to note about this clause. First, is that at one level it ought to be wrong, ungrammatical, because it contains a verb of mental process see, in active voice, with an unconscious entity as subject, the fifth day. Days cannot see. Second, a competent speaker of English knows that this is metaphorical, and can compare it with its nonmetaphorical agnate on the fifth day they arrived at the summit. A machine translation programme, say from English to Chinese, could demetaphorize it in either of two ways: (i) by replacing it in English with the non-metaphoric form, and then translating it «E*ftfljSII7Uj]I HS^'ftBiP^iS^'J T 1-LlM Diwu tian tamen dddaole shanding, or (ii) by translating it as it stands Jit £ ^ ff JrlfifeJF'j^lJLlJli-t Diwu tian kanjian tamen zai shdndmgshang, *E*»JUHHfcUJ]I± and then rewriting it in acceptable Chinese. There are advantages and disadvantages in either approach; which of the two works better will tend to depend on the socio-semantico-grammatical distance between the two languages (the closer they are, the more likely the second solution is to pay off). Examples of this kind are reasonably clear; their metaphoric nature is obvious, because of internal contradictions (here between day and see). The only question is, why is the metaphoric form used? No doubt this is partly for rhetorical embellishment: such locutions carry a certain value in certain types of context. But there are other more systematic semantic and communicative reasons why this metaphoric form is used: it enables the fifth day to function as the unmarked Theme of the clause — that is, in initial position as Subject, and thus gets the message structured the way the writer wants it. If we put on the fifth day first, as in the nonmetaphoric form, it is still thematic, but it is marked Theme, because it is not the Subject, and hence carries foregrounding that the writer may not want. In other words, the motivation lies in what the Prague school called the "functional sentence perspective". But many instances of grammatical metaphor fall into one or other 326

GRAMMATICAL METAPHOR IN ENGLISH AND CHINESE of two kinds which are rather less tractable, whether to the language student or to the programmer: (1) learned metaphors, typical of written language, especially rather weighty written language — scientific, bureaucratic, etc. — to which we can find non-metaphoric equivalents, even though it is sometimes not very easy; and (2) everyday metaphors, belonging to the system of the language (automatized, or "dead" in rhetorical terms), which have taken over the field and become a natural mode of expression. There is no sharp line between the two, but they may need different treatment in relation to particular tasks. For both these classes of metaphor we can usually find some explanation by relating them to general features of the system of the language, although the explanation may involve a fairly long and complex chain of reasoning. I would like to consider mainly the everyday kind of metaphors, but let me look briefly at the learned ones first. These are expressions like braking distance increases more rapidly at high speeds (to date) this latter proposition has not been pursued into practice advances in technology are speeding up the writing of business programmes sex differences in cognitive style have been observed in a wide variety of groups the analysis of the processes of social structural change requires a historical perspective In order to demetaphorize examples like these we have to sort out which are the processes (actions, events, mental processes, relations), which are the participants in the processes, and which are the circumstantial elements associated with them. So the last one might be 'interpreted' (reworded) as if we want to analyse how the structure of society changes we must study it over a long period of time. Learned metaphors of this kind tend to occur in all learned languages. Presumably we could render this into Chinese as something like Shehul jiegou bianqian chengxu dejenxi xuyao lishishangde guandidn. I do not know whether such patterns evolve independently in different languages, or whether they are borrowed, spreading out by processes of language contact from Western European languages (especially English).But they appear to be generally associated with these registers, and to be felt as reasonably natural

327

GRAMMAR AND DISCOURSE (otherwise people would not need to protest about them, which they do from time to time). Hence to learn the Chinese equivalent is not a major problem for the student; one just takes over the metaphor. The non-metaphoric form would also go readily into Chinese:

In other words, not only are both forms possible in both languages, but there is the same kind of relationship between them. Note that the metaphoric form would not have been possible in the English of Swift or the Chinese of Cao Xueqin; at least I think not. Not all instances are quite so easy to handle as this one. Consider braking distance increases more rapidly at high speeds I find this difficult to translate into English, let alone into Chinese. You can say the faster you're going the longer it takes to stop, but that is not the whole story; what it means is the faster you're going the more the time it takes to stop gets longer. This is one interesting example of a concept, namely 'rate of acceleration', which is simply not coded in the ordinary nonmetaphoric grammar of English. I do not know how to translate that into Chinese of any kind; but I suspect that in Chinese too it would be easier to represent it metaphorically — that is, it would be easier to translate it horizontally, at the same metaphoric level, than diagonally into some non-metaphoric form. Incidentally, it is difficult to find any systematic studies of concepts of this kind, concepts for which the language lacks a coded, non-metaphorical mode of expression. These things tell us very little about the nature of the socio-semantic distance between English and Chinese. They show that both languages are used for learned purposes, which we know already; and they suggest that when a language develops for such functions it will always develop along the same lines, although it will never be possible to prove this. We can never know what would have happened to the Chinese language if the processes of technological and economic change that were starting in the Song dynasty had not been arrested and deflected instead of leading on, as happened in England five hundred years later, to an industrial revolution. But it is not improbable that the changes that take place reflect the vast range of new functions that languages are called upon to serve, the increased demands on their overall semantic resources, a process that we are seeing again today as our languages are further extended to cope with moving into the high technology age. Let me pass on, however, to look at the category of everyday metaphors. There are two kinds of these: the interpersonal, having to do

328

GRAMMATICAL METAPHOR IN ENGLISH AND CHINESE with the mood, modality and key of the clause, and the ideational, having to do with transitivity, the clause in its representational function. An example of an interpersonal metaphor in English is found in the expression I don't think he's coming In fact there are two metaphors here. First, a non-metaphoric form would be something like probably or in my opinion he isn't coming. Now, by one of these metaphors the modality probably has been represented as the primary clause I think. We can tell that this is a metaphor because the tagged form is I think he's coming, isn't he? not I think he's coming don't I? which would be the appropriate form if the clause was really a clause about my thoughts. Second, in addition to the modal metaphor there is a polarity metaphor whereby the negative has gone into the primary clause; so, I don't think he's coming, instead of I think he isn't coming. Compare it isn't likely he's coming instead of it's likely he isn't coming. Note that both of these metaphorical clauses are on the face of it absurd: you cannot have non-thoughts and negative probabilities. If we compare the Chinese pattern, Chinese accepts the first of these metaphors: Wo xiang ta hui Idi as semantically related to Ta ddgai hui Idi. Note here again that if I say and you answer Dm then the question at issue is Ta Idi buldi and not Wo xiang buxidng. But Chinese does not accept the second; you cannot normally say Wo bu xiang ta hui Idi, as far as I am aware. Another type of interpersonal metaphor is the familiar type studied in speech act theory, in which one speech function is represented metaphorically as another; for example, I wouldn't do that if I was you, do that again and I'll clobber you, both usually regarded as embellished forms of'don't do that (again)!' Chinese, like (presumably) all languages, has plenty of this kind; for example ' Ni buru bu zaijixu zheydng zub cdi hdo, m zdici zheydng zuojiu hui di da. It is not their existence that causes a problem to the learner, but the specific principles on which those in the second language are constructed. The student of Chinese who asks for directions and gets the familiar answer Kongpd m yige ren yiding zhaobuddo 'I'm afraid you wouldn't be able to find it on your own' has to guess whether this means 'don't go there', or Til come with you' or 'I don't know the way; ask someone else!' But many of the 329

GRAMMAR AND DISCOURSE examples cited in speech act theory are conventional mannerisms that do not have any general validity as grammatical principles. Many ideational metaphors, on the other hand, are built into the grammar of every language. One common type in English is that exemplified by he has black hair I have a sore arm These the student learns very early on do not go into Chinese, where the process must be represented congruently as an attributive clause, such as Ta (de) ton fa hei, w6 gebei teng, not metaphorically as a possessive clause, such as Ta you hei toufa, wo you yige teng gebei. These have arisen in English so that the pronoun can function as the unmarked Subject Theme; in other words, the clause is about me or him, not about hairs or arms. In Chinese this message structure is achieved without the aid of this grammatical device, since there is no need to use the pronoun as a possessive. A second type common in English is that of she gave a smile, he had a bath, I made a mistake, they did a dance, where the process is represented not as a verb, which would be its congruent form as in she smiled, he bathed, I erred, they danced, but as a noun. The noun functions as what in traditional grammar was labelled a "cognate object", with the verb being one of the semantically "empty" words do, have,give, make, take. Again it is not difficult to find the motivation for this structure; the process is nominalized in order that it can be quantified and/or qualified, as in she gave a welcoming smile, he had a relaxing hot bath, I made two serious mistakes, they did this well-known Rumanian folk dance and so on. This has its parallel in the Chinese constructions like perhaps also ( c f . ) xizao, fancuowu, (tiaole neige hen youmingde Luomdniyade minwu) although not all the same set of processes exist in both languages. In Chinese the process tends to be specified t w i c e : T i a o l e yige wu not Zuole yige wu. We might say Ta geile yige weixiao, but certainly not Ta youle yige zdo. The pattern is more like English play a game, where play and game are both names of playing. Both these types are metaphorical in origin, although now they are fully coded in the system of the language. They have become the unmarked forms. The reason in each case is a textual one: to get the message organized in the way that is appropriate at that point in the discourse. Let us look at one other type in English, a type not 330

GRAMMATICAL METAPHOR IN ENGLISH AND CHINESE

obviously recognizable as metaphorical although in fact it is, where the same motivation emerges very clearly. For example: they called the meeting off they left the best part out where the process is represented as verb plus discontinuous adverb: call + off ('cancel'), leave + out ('omit'). Before we look for an equivalent in Chinese, let us explain this construction in English. Why is it so highly favoured, especially in the spoken language? The answer is to be sought in this discourse structure, and it is entirely systematic. Suppose we have the following piece of dialogue: What are you doing here? I thought you were in Canberra. They cancelled the meeting. / They called off the meeting. There is nothing to choose between the two forms of the response. But suppose the initiating turn had been: What are you doing here? I thought you were at a meeting. If I want to give a non-metaphoric response, using the verb cancel, I have to say: They cancelled the meeting. Note that the focus of information, signalled by tonic prominence, is now on cancel, because meeting has already been mentioned, and therefore is treated as Given. But in English the unmarked place for the tonic prominence is at the end; the result, therefore, is a form that is marked, and marked information focus is typically contrastive ('they cancelled the meeting; they didn't just postpone it'), whereas this is not the meaning that is intended. What is required is a form with the focus unmarked, in other words with 'cancel' coming at the end. There are two ways of achieving this. One is to use the passive, the meeting was cancelled, and the fact that the passive solves this problem is the main reason why it is used as frequently as it is in spoken English. But the passive is marked in another dimension — it is the marked term in the voice system; so another strategy has evolved, that of splitting the verb into two parts. Instead of cancel we use call off, and this allows the off to come at the end, where it can carry the tonic in its unmarked place: They called the meeting off. (This is why, if the object is a pronoun, it almost always comes before the adverb, since a pronoun is by definition 'given'.) Hence the strong

331

GRAMMAR AND DISCOURSE preference for phrasal verbs, which represent the process metaphorically, rather than single word verbs in spoken English. In written English, of course, there is no such pressure, so phrasal verbs are felt as rather colloquial. Chinese is like English in two respects that are relevant here: (1) it is a Subject-Verb-Object (SVO) language, (2) the information focus typically comes at the end. But Chinese has no phrasal verbs. In Chinese, the problem is solved by another metaphor, the metaphor. As in English, this also involves pretending to split the process into two; but in Chinese you first 'grasp' the object, and then do something to it: Tamen ba huiyi quxiaole

Note that there is a proportionality here between Chinese and English, as shown in the following pairs: is to Wo rengdiaole wode po yifu

Wo ba wode po yifu rengdiaole

as

(

discarded ] threw awav I

m

^ °^ clothes

is to

I threw my old clothes away

Like the English, moreover, the Chinese construction is also subject to a phonological restriction: the verb in a clause must contain at least two syllables, to carry the weight of the information. If the lexical verb does not consist of two syllables, the second syllable must be made up by the aspectual marker. Likewise in English, a phrasal verb must have at least two syllables: eat + up, put + out, etc. What lies behind the similarity in these two constructions is the fact that Chinese and English share a common concern with getting the message right. Specifically, each language is concerned to make it clear what the speaker is on about, which it does by identifying a thematic element — signalled by position at the beginning of the clause; and each language is concerned to make it clear what the speaker wants the learner to attend to, which it does by identifying a focal element signalled by tonic prominence, but with favourite (unmarked) position at the end. In English, this feature is part of a very general pattern which pervades the grammar in numerous ways; superficially very disparate but all part of a highly coherent (but at the same time highly unstable) design. I am sure that in Chinese there is an equally coherent pattern, probably also in process of continuous change. Chinese, of course, is very far from having the same semantic system as that of English; but in this particular 332

GRAMMATICAL METAPHOR IN ENGLISH AND CHINESE

respect, the organization of the clause as a message in the context of the discourse, there are striking instances of similarity. These features of grammatical metaphor, therefore, provide some interesting clues to the general fashions of meaning in a language, its overall semantic styles. Those of the everyday kind — and there are certainly many other features of Chinese that belong in this category — have become part of the linguistic system, for one reason or another. The reason is usually, it seems, something to do with the structure of discourse, the way the individual messages are represented in clause structure to give the language its natural texture. Such metaphors are learnt by children very early, as part of the foundation of the linguistic system. Those of a more learned kind, which children do not learn till much later — typically not until early adolescence — and tend to find rather difficult at first, also relate to the overall texture of the language, but in a different way and only in certain registers. These seem to represent tendencies common to the elaborated discourse of science and technology, government and bureaucracy, in all languages. Both types require a great deal of further study, not only to help in the more effective teaching of languages but also as a contribution to our general understanding of grammar, discourse structure and semantic styles.

333

Chapter Eight

ANALYSIS OF SCIENTIFIC TEXTS IN ENGLISH AND CHINESE (1993)

The Language of Science There are practical reasons for analysing scientific texts. The most obvious is educational: students of all ages may find them hard to read, and we know from various research reports that, in English at least, the difficulty is largely a linguistic one. So if we want to do something about it we need to understand how the language of these texts is organized. Of course, if a text is hard to read the difficulty is bound to be in some sense linguistic, since texts consist entirely of language, but in the case of scientific writing it seems that there are certain features of the way meanings are organized, and the way they are 'worded', that present special problems for a learner, over and above the unfamiliar subject matter and its remoteness from everyday experience. So people recognize that there is such a thing as 'scientific language', at least in the written mode; therefore it must be possible, using the theories and methods of linguistics, to say what its special features are. This raises two questions. First, where (at what level, or levels) in the linguistic system do we explain them? Are they a) in the lexicogrammar - that is, in the meanings constructed into sentences by the syntax and the vocabulary? b) in the discourse - the composition of the text and its rhetorical structure? or c) in some higher order - the ideological framework of knowledge, beliefs and value systems that form the cultural

'The analysis of scientific texts in English and Chinese', from Writing Science: Literacy and Discursive Power. This chapter was originally presented at the International Conference on research in Texts and Language, Xi'an Jiaotong University, March 1989. With kind permission of Taylor and Francis Group.

334

ANALYSIS OF SCIENTIFIC TEXTS IN ENGLISH AND CHINESE context of the text? All of these, of course, find their expression in what we call "the words on the page"; but they involve different processes and conditions of interpretation. The answer may be, of course, that the concept of'scientific language' involves all three. The second question that arises is whether the phenomenon of scientific language is universal. That is to say, does each language have its own scientific register(s), so that we recognize scientific English (a special kind of English), scientific Chinese, and so on? Or is there in some sense a single language of science' that is essentially the same no matter what language it is manifested in? Of course English and Chinese use different expression systems and different sentence grammars. But it might be that their scientific registers would have many features in common. If so, this makes the task of translation (including machine translation) much easier, since the transfer can take place at the lower levels of the linguistic system. This paper mainly focuses on the second question, taking English and Chinese as the languages to be explored. Note that if we find (as of course we do) that there are some features in common to scientific English and scientific Chinese, this does not by itself tell us why. It might be that one is borrowed from the other: that in the twentieth century, scientific Chinese has been created by a process of transference (facilitated by actual translations) from English and other European languages. This is what has happened — or is commonly said to have happened — in many parts of the world where new scientific and technical forms of discourse have been created, sometimes in languages that were not previously written down. The other possibility is that the two evolved quite separately, but because of the nature of science, and scientific knowledge, they have grown to be very much alike. The truth may well be some mixture of the two (and either way we can compare them). But to the extent that shared features can be shown to be motivated, as opposed to conventional, to that extent we would expect them to evolve on their own, quite independently, without having to be transmitted by linguistic borrowing. But some decision has to be taken regarding where to move in, in terms of the first of our questions above: do we investigate the lexicogrammar ("sentence grammar"), the text structure, or some higher order abstraction? Here I must take up an explicit theoretical position, since that will determine what the options are. I shall assume that all the features of a text participate in the creation of meaning; thus, whatever is characteristic of scientific writing realizes (dialectically, both determines and is determined by) the structure of scientific knowledge and its 335

GRAMMAR AND DISCOURSE attendant ideological formations. But within this overall semogenic system-and-process, certain components are more accessible, more under conscious direction than others. To give an example from science, editors of scientific journals often give their authors specific instructions about how to write a scientific paper: introduce the topic, present the current state of knowledge, define a problem, describe the approach adopted to solving it, give the results accompanied by tables, figures, equations, etc. where appropriate, draw conclusions, state what needs to be done next. This kind of rhetorical organization of the text, its structure as a piece of discourse, is easy to observe and fairly easy to prescribe. Furthermore it is minimally dependent on which language is being employed; such instructions can be carried out equally well in English, Chinese, Italian, Malay and so on. I am not saying that all scientific papers are written in the same format; on the contrary, there are considerable differences in the favoured construction of articles even in such culturally close languages as English, French and German (Clyne 1991).But any such structure could be implemented in any language, and is readily borrowed from one language to another (thus translators will usually not alter the discourse structure of a text they are translating). It is clear that the evolution of such highly valued text types played a significant part in establishing Newtonian physics as the dominant paradigm for the natural sciences in renaissance Europe (Bazerman 1988). The text structure establishes the highly-valued form of scientific argumentation: what constitutes an acceptable canon of evidence, reasoning and proof. It models the 'macro' movements by which knowledge is extended and the scientific understanding progresses. By contrast, the lexicogrammar models the 'micro' movement, the smaller steps that, taken all together, comprise one episode in the macrostructure of the text. The lexicogrammar also models the conceptual structure, the taxonomies of categories that provide the scaffolding of scientific knowledge, based on the two fundamental semantic relationships of hyponymy (a is a kind of x) and meronymy (b is a part of y). And these components - the conceptual structure and the microargument - are much less accessible to observation and control, because they depend on linguistic patterns that speakers of a language are not aware of unless they engage in some technical linguistic analysis. By the same token, they are more closely tied to the semantic and lexicogrammatical structure of the particular language that is being used. We will illustrate very sketchily some of the similarities and differences between English and Chinese in the way their lexicogrammar constructs scientific reality. We start from the lexical end — the 336

ANALYSIS OF SCIENTIFIC TEXTS IN ENGLISH AND CHINESE vocabulary - because, within the lexicogrammar, it is the vocabulary, or lexicon, that is nearer the surface of consciousness. It is easier to pay attention to the vocabulary than to the grammar.

Taxonomies All languages create taxonomies, based especially on hyponymy ('is a kind of); and children begin to learn them from a very early age. For example,/ra/£ is a kind of food; a berry is a kind of fruit; blackberry, strawberry, raspberry are kinds of berry. In the last stage in this example the language signals the taxonomic organization by including the word berry as a constituent morpheme of the word blackberry, etc. English does not do this consistently; carp, herring, cod/codfish, gemfish are all equally kinds of fish. Chinese on the other hand does construct hyponymic sets consistently in this manner: every kind of fish ends in yu . Thus for a Chinese child the taxonomic structure of things in the environment is made very explicit; for an English child much less so. An Englishspeaking child does not learn cucumbers, melons, squashes, pumpkins and marrows as a related set; in Chinese they are all kinds of gud (donggud, xigud, ndngud, wogua, hudnggud, etc.). When it comes to constructing technical and scientific taxonomies, therefore, Chinese maintains the same principle — all gases are a kind of (i.e., compounded with) qi , all pointed instruments a kind of zhen ; and it extends also to abstract terms, so that liebidn 'fission' is a kind ofbidn'change', tdisheng 'viviparity' is a kind of sheng'giving birth', pinlu 'frequency' is a kind of lu 'rate', and so on. The English words fission, viviparity, frequency give no signal of their taxonomic status. The more highly technical terms in English are more likely to be constructed as taxonomic compounds; but if so, the components will be borrowed from Greek or Latin, e.g., chemotaxis, phototaxis, thermotaxis (Chinese quyaoxing, qugudngxmg, qurexing} There are three consequences of this difference: (1) The specific taxonomic organization is more explicit in Chinese than in English. (2) The general principle of taxonomy is therefore more accessible (that is the way the Chinese world is organized). (3) There is less distance between the world of science and the everyday world than there is in English. Both languages construct basic scientific taxonomies; but in English the general principle is less explicit, the particular structure is less explicit,

337

GRAMMAR AND DISCOURSE and (this aspect of) scientific knowledge is further removed from experience of common sense. When it conies to more elaborated taxonomies, both languages operate on the same principle: the expansion of the noun into a nominal group by premodification. So in both English and Chinese we find the same patterns: plastic foam plastic thermal plastic engineering plastic phenol plastic

suliao paomo suliao resu suliao gongcheng suliao fenquan suliao

The semantic resource is the same in both languages, and both tolerate about the same degree of expansion without introducing further structural mechanisms.

Objectification The prototypical form of a scientific taxonomy is a hierarchy of classes of objects, like the taxonomies of the plant and animal kingdoms. But many of the entities named by nouns in scientific discourse are not, in fact, objects, but processes or properties \ikefission and frequency mentioned above. In both Chinese and English the noun is a clearly defined grammatical category; in both languages its core meaning is that of an object — an object possessing certain properties and capable of participating in certain processes. Thus in diamond is energetically unstable/transforms into graphite diamond is an object; diamond (has the property:) is energetically unstable; diamond (participates in the process:) transforms into graphite. Here unstable, representing the property, is an adjective; transforms, representing the process, is a verb. Both these may, however,be represented as nouns: the energetic instability of diamond, the transformation of diamond into graphite. Instability, transformation are now 'objectified', that is, they are presented by the grammar as if they were classes of objects. And typically, just as in the history of the language these things were first adjectives or verbs and then became nouns (the process may have taken place at any time from early Greek science to the present day), so in the history of a text we usually meet them in the same order.

338

ANALYSIS OF SCIENTIFIC TEXTS IN ENGLISH AND CHINESE For example, . . . (diamond is) energetically unstable . . . (its) energetic instability (diamond) transforms into graphite . . . (its) transformation into graphite These noun-forming resources (derivational morphology) developed in ancient Greek, were taken over and extended in Latin, then taken over and further extended in English and other European vernaculars. Chinese also objectifies. So from buwen 'unstable' we have buwenxlng 'instability'. In Chinese, however, when a word changes its grammatical class in the syntax, it is not usually signalled morphologically; so bianhua 'transform' is also 'transformation'. Thus while Chinese speakers often regard the large number of nominalizations ending in xing as foreignisms, it is not the nominalization itself they are reacting to but its formal marking — the way the noun xing 'quality' has become delexicalized through excessive use. It is often said that Chinese does not go in for nominalizations the way that English does; it prefers verbs. It may very well be that in other registers this process of nominalization has not gone so far as in English; but as far as the discourse of science is concerned there is no noticeable difference. For example, Hang dianzi-di jingdian chili shineng wei zheng zhi, qie yu liangzhe-di jianju cheng fanbi the electrostatic repulsion potential of two electrons is a positive value, and is in inverse proportion to the distance between them We refer to this process of objectifying as "grammatical metaphor". A semantic feature that is typically realized by one grammatical means comes instead to be realized by another. Thus processes (events, actions, states) are typically realized, in both languages, by verbs; but by grammatical metaphor they come to be realized as nouns. There are very many kinds of grammatical metaphor in both English and Chinese; but the one that dominates the language of science is nominalization. Phenomena which are not objects become 'objectified' by being realized as nouns. Since the core meaning of the category of noun is a class of object, when anything is realized as a noun it takes on some of this object-like status. Complex nominalization In an expression such as energetic instability the above features are combined: 339

GRAMMAR AND DISCOURSE

(1) unstable has become an 'object' instability; (2) this noun functions as 'Thing' in a nominal group, expanded by a premodifying word energetic, (3) energetic instability contrasts with kinetic instability in a taxonomic relation - two kinds of instability; (4) and energetic itself embodies grammatical metaphor: grammatically it is not an Epithet, as in an energetic person, but Classifier '(unstable) with respect to energy' vs. kinetic '(unstable) with respect to movement'. I do not know these terms in Chinese, but I imagine that I could coin nengliang buwenxing for energetic instability and yundong buwenxing for kinetic instability, using exactly the same grammatical resource — grammatical energy, or yufa nengliang, one might say. The next step is to nominalize not only the process but also any participants and circumstances that go with it: grammatically, expanding the nominal group still further to include some or all of the elements of a clause. For example, dianzi de guidao yundong the orbital motion of an electron (see the clause an electron moves in an orbit, dianzi shun guidao er dong ). Here by a complex grammatical metaphor the process 'move' has become a noun (Thing), the actor in that Process 'electron' has become its possessor (Qualifier), and the circumstance 'along an orbit' has become the property of the noun (Classifier). Such nominals can of course become very long; the actual Chinese example was: dianzi-di zixuan yundong he guidao yundong xianghu ouhe chengzongdi yundong 'the combined motion of an electron resulting from the coincidence of the orbital with the rotational motion'

At this point, English and Chinese begin once again to diverge. (Remember they were different in step 1, the taxonomic structure of the lexicon, but similar in step 2, the expansion of the nominal group, and grammatical metaphor.) This time, however, the relationship is reversed: Chinese is less explicit than English. Let me explain this. When a process

340

ANALYSIS OF SCIENTIFIC TEXTS IN ENGLISH AND CHINESE is represented as it typically is — in the form of a clause — the semantic relationships among the components are made explicit by the grammar; for example, an electron moves around a nucleus in a circular orbit. As the process comes to be nominalized these semantic relationships become less explicit: orbital motion is a kind of motion having something to do with an orbit, but is not made explicit whether it is motion of an orbit, in an orbit, caused by an orbit or what. In movement of an electron, we are told that the electron is a participant, but not whether it is Actor or Goal (does it move or is it moved?). Now, in English, a typical nominalization of this kind includes both premodifiers, which leave the relationship entirely implicit (see electrical energy, electrical coil, electrical calibration, electrical potential, where the semantic function of electrical is different in each case), and postmodifiers with prepositional phrases, where the preposition gives some semantic information. A typical example would be the overall enthalpy charge for the conversion of graphite to carbon dioxide. In Chinese, on the other hand, there is no postmodification; all modifiers precede the Head noun. This does not mean that all the transitivity relations of the process are lost in nominalization, because both embedded clauses and (the Chinese equivalent of) prepositional phrases occur in this premodifying position, and such clauses and phrases retain their transitivity structure. Here is a text example: yiner, duiliuceng daqi an qi wuli xingzhi zai shuiping fangxiang-shang-di chayi fenwei ruogan qituan for that reason the tropospheric atmosphere is divided into several air masses according to differences in its physical properties on the horizontal plane (note an 'according to', zai. . .shang . . . . 'on'). Most of the time, however, these relationships are left implicit, and the only structure markers introduced into the premodifier are di showing hypotaxis, and he showing parataxis, equivalent to generalized 'of and 'and'. Here is an example: hanzi bujian jiegou character component structure fangfa he yuanze method 'and'

canshu prarmeter

fenxi analysis

di 'of

neirong, content

principle

content, method and principle of the parametric analysis of the composition and structure of [Chinese] characters 341

GRAMMAR AND DISCOURSE

In fact there are numerous other possible English equivalents of this nominal group, some of which do not alter the semantic relationships but others of which do; for example, analysis of the componential and structural parameters of Chinese characters. Where English leaves out all ands except the last (a, b, c and d), but inserts all ofs, Chinese leaves out the pre-final ofs as well; hence any sequence a b c . . . di n permits 2n—1 different bracketings; and while most of them may not make sense, there is often a subset which do, and a student (whether a foreign student of Chinese, or a Chinese student of science!) may be hard put to it to know which to reject. Furthermore, while English can distribute the elements of a nominal group into pre- and postmodifier ([a cow is] a cud-chewing quadruped which eats grass, a grass-eating ruminant with four legs, a four-legged herbivore which ruminates, etc.}, in Chinese all must precede, so that even the fact that something is a modifier may not be made explicit until right at the end. The sentence following the 'troposphere' example (see p. 341) above is: [qituan ziushi] zai shuiping fanwei ke da shuqian gongli er wendu, shidu deng wuli xingzhi zai shuiping fangxiang-shang-di chayi bu da di yi da kuai kongqiceng

[an air mass is] a large atmospheric layer in which such physical properties as temperature and humidity display little variation in a horizontal direction over a horizontal range of up to several thousand kilometres — where everything outside the square bracket is a single nominal group with a Head noun kongqiceng'layer of air' at the end. Let O C oi o> / me make it quite clear that there is of course no virtue in putting the modifying elements at one side of the Head noun or the other. What is relevant is that, whatever the resources in a particular language for creating indefinitely expandable nominalizations of this type, they will be put under maximum pressure by the requirements of scientific discourse. Each language will then respond in its own way; and in Chinese the effect is to create a high level of ambiguity and to leave implicit many of the semantic relationships among the various elements of the process.

The functions of nominalization Our last point, then, will be to ask: why does this pressure arise? Why does scientific discourse demand this very high degree of nominalization with all the grammatical metaphor that this involves? 342

ANALYSIS OF SCIENTIFIC TEXTS IN ENGLISH AND CHINESE There seem to be two main reasons. One concerns the structure of scientific argument, as developed in Europe by Isaac Newton and his successors. The core of a scientific text was the development of a chain of reasoning (ultimately based on experiments) in which each step led on to the next. But in order to lead on to the next step you have to be able to repeat what has gone before and is now being used as the springboard for the next move. Notice how the grammar does this, with an English and a Chinese example: . . . both ethyne and nitrogen oxide are kinetically stable . . . The kinetic stability of nitrogen oxide shows . . . tianqi shi shunxiwanbian-di; dan tadi bianhua shi you iding guilu-di the weather is constantly changing; but its changes have a definite pattern In both languages the grammar 'packages' what has gone before by nominalizing the Process (attribute or event), and making the Medium of that Process a 'possessive' modifier. This enables it to function as the Theme of the succeeding clause. This is the simple, most basic form. Sometimes, however, the new step in the argument has to include not just the previous one but a great deal that has gone before (and sometimes to anticipate much of what is to follow); and this may require a more complex package, picking up a number of related motifs. For example, The great reactivity of fluorine in these reactions with non-metals [is explained . . . ] putong yingguangguan suo fashe-di guang shi you zhezhong ziwai yueqian suo jifa-di tuliao-di yingguang the light that is emitted by the usual fluorescent tube is a fluorescence of the coating stimulated by this kind of ultraviolet transition

The noun is the only syntactic class that can accept thematic 'packages' of this kind; or, to put this another way round, the process of packaging turns (metaphorizes) all events and attributes into textual objects. I have discussed this phenomenon in another paper (see Halliday and Martin 1993: Chapter 3) and will not try to elaborate it here. The second reason for nominalizing has to do with the structure of scientific knowledge. While the argument has to be dynamic (hence the flow of the text), the edifice that is constructed by it is a static one — or let us say that it 343

GRAMMAR AND DISCOURSE embodies a synoptic rather than a dynamic representation of reality. Newtonian science has to hold the world still, to anaesthetize it so to speak, while dissecting it — if you are trying to understand something, then in the early stages of your enquiry it is helpful if it does not change while you are examining it. Of course, reality does change while you are looking at it, and twentieth-century science is coming to terms with this; so the grammar of scientific discourse in the next 500 years will probably be very different from what it is now — perhaps more like that of spoken language with its clauses rather than expanded nouns. To begin with, however, scientists had to create a universe that was made of things. The concept of a taxonomy, that we started with, is first of all a taxonomy of objects; it is only later that other kinds of phenomena come to be classified in this way, and again in order for this to happen the grammar has to turn them into nouns (they are types of change, motion, etc.; see Halliday and Martin 1993: Chapter 8, p. 341). So the grammar of both Chinese and English, using similar resources (noun compounding, the nominal group, nominalization and grammatical metaphor), creates a form of discourse for codifying, extending and transmitting scientific knowledge. With this discourse, the argument — the rhetorical movement — is made very clear and explicit; while the content — the conceptual structure and internal relationships — may be left highly implicit, in one way or another (each language in its own way). We are not saying that for the meaning to be implicit in this way is something undesirable; but this kind of grammar does make demands on both writer and reader: demands on the writer to ensure that the text provides the semantic information that the reader needs in order to construct the taxonomies, decode the metaphors, and follow the argument; and demands on the reader to be alert enough to receive and make use of this information. What seems important to me, as a linguist and as a teacher, is that we should understand both how the grammar of science works and why it works the way it does. For this we need to examine scientific discourse, analysing the grammar in terms of its different functions ('metafunctions', in the systemic sense; especially the ideational and the textual); and also to trace its history, how it evolved the way it did. I asked at the beginning: would scientific Chinese have been different if modern science had evolved first in China? Obviously we can never answer this kind of question with certainty. But technology did evolve in China, and was ahead of technology in Europe throughout most of history; so we can compare the earlier forms of scientific discourse that developed in classical Chinese and in medieval Latin (and also the early texts 344

ANALYSIS OF SCIENTIFIC TEXTS IN ENGLISH AND CHINESE in English and other European vernaculars). Of course, although we can ask why things happened in the past, we cannot change them. We can, however, adopt practices that will have an effect on the future. As linguists, and discourse specialists, we have a responsibility to look ahead: to assist those who are learning science either in their mother tongue or through some foreign language; to advise on language planning — whether scientific language should be 'engineered' in some way (as is done with chemical terminology, for example; but not limited to the creation of technical terms); and, not least, to raise issues of the role of science, and of scientists, in our modern society. If we examine the discourse of science, we can become more aware of the ideology that is enshrined in the way scientific language construes the world.

345

Chapter Nine

ON THE GRAMMATICAL FOUNDATIONS OF DISCOURSE (2001)

When I first studied linguistics, Haifa century ago, there was no discipline of "discourse analysis". Linguists typically investigated phonology and morphology, or, less often, syntax. They might make use of texts — written texts — as sources for their examples; but a text was not an object of study in its own right. If a text happened to be the only source for finding out about a particular language, in the way that the Chinese version of the Secret History of the Mongols was for the earliest phases of modern Mandarin, it might be analysed in considerable detail as I did in my doctoral thesis (Halliday 1959); but even here the aim was to describe the system of the language, not to elucidate the text. J. R. Firth was unusual among linguists in regarding "typical texts . . . in ... contexts of situation" as a proper domain for linguists to study (Firth 1968: 13). In general, analysing text was a literary preoccupation, and it figured as an activity within linguistics only where the two disciplines could merge, as in the work of certain scholars of the Prague school. Any discursive tendencies that linguists might have shown were suppressed by Chomsky, with his insistence on treating language as a formal system and his consequent contempt for data. So when discourse analysis was taken up as an academic pursuit, it tended to be divorced from linguistic theory; and the myth was perpetuated that "linguists have no interest in discourse". Those who studied discourse were not, for the most part, grammarians or phonologists; Conversational Analysis, for example, pays little attention either to grammar or to rhythm and 'On the grammatical foundations of discourse', from Grammar and Discourse: Proceedings of the International Conference on Discourse Analysis, edited by Ren Shaozeng, William Guthrie and I. W. Ronald Fong (2001). Reprinted with permission from University of Macau.

346

ON THE GRAMMATICAL FOUNDATIONS OF DISCOURSE intonation. And undeniably, one can gain much insight into a text from many angles without analysing it in grammatical or phonological terms. But many linguists do, in fact, engage with discourse; and the linguistic analysis of text is an essential component in its overall interpretation. After all, a text is made of language; it gets its meaning, and its value, from being an instance of some linguistic system. More specifically, a text is made of lexicogrammar: it comes to us in the form of wording. But the relation of discourse to lexicogrammar, and of discourse organization to lexicogrammatical organization, is highly complex. It involves three distinct vectors: part to whole, form to meaning, instance to system. Grammar creates structure up to the clause complex (the written sentence); texts tend to be more inclusive (e.g., a conversation, a speech, a book). Text is organized as semantic structure: what we call "a text" is really a unit of meaning. And, as noted above, a text is an instance, whose properties derive from a system, the total "meaning potential" that the lexicogrammar makes possible. All these vectors are implicated when we produce and understand a piece of discourse. In this paper I want to consider just three aspects of this complex relation of discourse to lexicogrammar (which I shall usually refer to simply as "grammar" for short): (1) How are grammatical units organized internally such that they form "parts" of a textual "whole"? (2) How does the grammar set up semantic relations that create unity in a text? (3) Do the logical-semantic relations that link grammatical units also figure in relating one phase of a text to another? I have not tried to be comprehensive; my aim has been rather to illustrate the general principles, and to suggest how a grammarian may approach the analysis of discourse. 1 The "textual" organization of the clause I myself became aware of this problem when investigating spoken language. I have always been particularly interested in spoken language, and also I had had to teach spoken language to foreign language students: first teaching Chinese to speakers of English, and later on teaching English to speakers of Chinese. In the course of this experience I had become aware that there were certain resources in the grammar, both in Chinese and in English, which were specifically "textual" in function: 347

GRAMMAR AND DISCOURSE that is, they were resources for organizing the discourse as continuous flow, maintaining coherence with whatever had gone before. For example, I wanted to explain to English-speaking students of Chinese when to say xihao yifu and when to say ba yifu xihao ; when to say wo xi yifu and when to say yifu shi wo xi-de , xi yifu-de shi wo wo xi-de shi yifu and so on. Clearly these options had something to do with the flow of the discourse: the clause was being organized as a message, based on a distinction between what is being presented as news to the listener and what the listener is being presumed to know already — between the Given and the New. At any given point in a dialogue, there seemed to be some sort of balance, a kind of oscillation between these two poles; and the most "newsworthy" item was given some kind of prominence - either it was put last in the clause, or it was predicated (introduced by shi ), or both. Y. R. Chao brought to our notice, in his Mandarin Primer (1948), the contrast between ta zhu zai Beiping and ta zai Beiping zhu explaining it as a difference in what the speaker wanted most importantly to convey. From this point of view it became clear that the ba construction, which at first seemed quite mysterious, was a device for getting the verb to the end of the clause, so that it would carry prominence as new information. Otherwise, if there are two participants (nominal elements) in the clause, unless one uses ba the verb cannot come in final position. English has the same problem, but solves it by means of phrasal verbs. Compare they expelled the leader and they threw the leader out: when these are spoken naturally, in the former the leader is New, whereas in the latter threw . . . out is New (Halliday 1985,1992). Clearly it is the speaker who decides what to treat as Given and what to treat as New; but he does this (unconsciously, of course) in the light of all that has been said before. Wang Li (1946-47:164) called the b a c o n s t r u c t i o n chuzhishi which he translated into English as "execution form"; and in my early writings on Chinese grammar I labelled it the "ergative voice", on the grounds that it was "singling out the verbal element as new by opposition to all nominal elements" (Halliday 1956: (= 1976: 44); cf. 1959:56,80 (and n.l)). The Swiss grammarian Henri Frei independently called the ba construction "ergative" (Frei 1956-57). I included both the ba construction and all the other variants mentioned above (yifu shi wo xi-de and so on) in the Chinese system of "voice". In doing so, I was treating "voice" as a system which organizes the flow of discourse. The problem with the category of voice as 348

ON THE GRAMMATICAL FOUNDATIONS OF DISCOURSE described by classical European grammarians is that, because it always involved changes in the form of the'verb (active/passive, as in Greek and Latin), it was seen as a verbal system; whereas in fact it is a system of the clause, and it may or may not be accompanied by changes in verbal morphology. In Chinese, obviously, the verb does not undergo any change; but there is still a fully grammaticized system of voice. What is common to all voice systems is that they are resources for assigning variable status in the discourse to participants, and other elements, within the clause. In other words, they allow the speaker to control the information flow. Thus the passive, in English, enables the Actor to be presumed (e.g. these clothes have been washed}; or, alternatively, to be placed in a position of prominence, at the end of the clause (these clothes were washed by your mother). The particular options in the Chinese system are not, of course, identical with those in English; but the semantic space construed by the two systems is the same. The one term in the Chinese system that most resembles a typical Indo-European passive is that with bei so the bei construction was at first the only one in Chinese that was recognized by most Western grammarians as being a category of voice. But the variation in both Chinese and European languages is governed by the same principle: that of arranging the elements of the clause in all possible ways in relation to the context of the discourse — the context as the speaker inherits it, and as he wants to shape it afresh in his turn. Reordering the elements of the clause is not the only way of varying their status in the discourse flow. Both in Chinese and in English the positional prominence is complemented by phonological prominence (pitch movement and stress), and may in fact be overridden by it. This happens in Chinese with bei or gei , as in zhe yifu gei ni muqin x i h a o l e w h e r e phonological prominence i s assigned to ni muqin thus overriding the positional prominence of xihaole . Note that there is no way in Chinese of getting the Actor in final position other than xi zhexie yifu-de shi ni muqin but this adds another semantic feature — the construction X-de shi Y X Y identifies Y exclusively as 'the one(s) who X', and the speaker may not want this particular nuance of meaning. Similarly in English we can say your mother washed these clothes, with phonological prominence on your mother. Both languages make use of this phonological device, in their own different ways, to get around the constraints imposed by the grammar on the order in which the elements occur. This does not mean that certain orderings are impossible — that seldom happens! What it means is that other orderings carry other meanings, and these may not correspond to what the speaker wants to say. 349

Among these other meanings are some that are also textual: other choices in the organization of the discourse, independent of the Given and the New. Thus, both in Chinese and in English there is choice in the order of elements not only in the final part of the clause but also at the beginning. Let us say, for example, that your mother/ni muqin is in a position of prominence, either at the end of the clause or following shi /be; and let us hold her constant in that position. We still have a choice with regard to what comes first: in Chinese, either zhe yifu or xi (zhe yifu shi ni muqin xi-de , xi zhe yifu-de shi ni muqin ), in English either these clothes or wash (these clothes your mother washed/were washed by your mother, the one who washed these clothes was your mother). I had not explored this feature in my work on Chinese; but in analysing spoken English I found it essential to account for the speaker's choice of what to put first in the clause. It was clearly a meaningful choice; and it seemed to belong to the same general domain, the organizing of the discourse flow. But whereas the prominence associated with clause-final position, and with predication by the verb be, directed the listener's attention to what was newsworthy — new, or unexpected, in the discursive context, or at least constructed this way by the speaker — the significance of initial position was a prominence of a different kind. When I had looked at this in Chinese, trying to teach my students how to start off a Chinese clause, I had thought it was simply the 'Given' part of the Given + New configuration. But that did not seem to explain things very well; for one thing, the s h i . . . de ... constructions are reversible — one can have both zhe sanjian yifu shi wo xi-de and wo xi-de shi zhe sanjian yifu When I investigated a quantity of spoken texts in English, while investigating English intonation in 1958-63,1 found that, in order to explain the choice of word order in English one had to postulate that there was another system at work: a system of Theme whereby the speaker selected some element in the clause as the point of departure for the message, realizing this selection by putting this element first. This accounted for the difference, in English, between your mother washed these clothes and these clothes were washed by your mother, in both, your mother is New, but in the first of the two she is also thematic, whereas in the second example these clothes is selected as the Theme. At the time, I was unaware of the work of the Prague school linguists in this area; I later found that they had explored it very fully in a number of languages, English being one. I was not sure whether there was any comparable system of Theme in Chinese — or whether, even if there was, the function of Theme was 350

ON THE GRAMMATICAL FOUNDATIONS OF DISCOURSE necessarily realized by first position in the clause. This problem was addressed by Fang Yan When Fang Yan was in Sydney she analysed various texts in Chinese from this point of view, including some spoken xiangsheng texts (of which I still have a copy!); and she found that Chinese does make use of initial position in the clause in a way that is quite similar to English. This is not very surprising, because Chinese and English are quite similar in the range of functions for which they use word order (that is, in the differences in meaning that are realized by different arrangements of the elements in the structure of the clause). So Fang Yan identified this initial element as Theme (Fang 1993, 1995); and she has also co-authored a paper with Edward McDonald and Cheng Musheng (Fang et al. 1995), entitled "On theme in Chinese: from clause to discourse". Let me draw attention particularly to the second half of that title. In what way is the speaker's choice of Theme relevant to the unfolding of the discourse? We have seen that, in the case of Given + New, the listener-oriented prominence of the New has an important — if fairly local — discourse function. This becomes apparent, for example, in dialogue, where it provides a major resource for steering the topic of conversation. Given + New can be interpreted as a kind of questionand-answer framework: the New is giving the answer in response to a particular question. So zhexie yifu shi ni muqin xi-de implies the question shi shei xi-de ?; these clothes were washed by your mother implies the question "who washed these clothes?". I am not suggesting that this is the answer that would have been given if such a question had actually been asked; if it had been, the answer would have been much more likely to be elliptical, e.g. Your mother. What the speaker is doing, in assigning the Given + New structure, is setting up the message as the answer to an unspoken question. The remainder, the Given part, is what is presumed to be "in the air" already; in this case, the listener is presumed to know that the discourse is about clothes and their getting washed. But the Theme + Rheme structure is very different in meaning. The Theme is a kind of prominence construed from the point of view of the speaker — it is the element the speaker chooses as the foundation on which to construct a message; and it would seem at first sight to have little significance beyond the particular clause in which it occurs. It is true that the Theme is, very often, selected by the speaker from within the Given part of the Given + New configuration; but in that case its discursive significance might be simply because it is Given, not because it is Theme. However, it is not always selected from within the portion that is Given; sometimes the Theme is

351

GRAMMAR AND DISCOURSE also presented as New. So the question that arises is: has the clausal Theme got any function in the wider context of the discourse? A number of years ago Peter Fries showed that, in written English, the choice of Theme in the clause played a very significant discursive role (Fries 1981). The elements chosen as clausal Theme throughout the course of a text were not, in fact, a random selection of items which varied clause by clause; rather, they displayed a considerable regularity of patterning. The particular kind of pattern formed by the progression of the Themes depended, as Frantisek Danes had predicted some time earlier (1974), on the register, or functional variety, of the text: narrative, for example, tended to display a different kind of thematic organization from that of expository discourse. In some registers, there would be a certain motif, or set of motifs, which would figure most of the time as clausal Theme; while other motifs, although equally common in the text, would only seldom be given thematic status. This suggested that the Theme in this technical grammatical sense, as an element in the structure of the clause, was related to "theme" in its meaning as the topic or concern of the text as a whole. To take a simple example, in a short biographical text, which is a life history of a particular individual such as one finds on the back cover of a book introducing the author, the majority of clause Themes are likely to be either the individual in question (by name, or referred to by a pronoun) or some expression of location in time. In the more than 15 years since Fries' original paper there has been a great deal of further work on thematic systems in grammar, both in English and in various other languages. Fries himself showed that the patterning of the non-thematic portion of the clause, particularly in combination with the New (the "N-Rheme", as he labelled it), was also highly relevant to the progression of the discourse. In subsequent studies he has explored the relationship between the selection of Theme and the goals of different types of text; and he has demonstrated that the type of thematic patterning will vary, not only among different registers, but also at different stages in the unfolding of a single text, in ways that closely relate to its functional context (Fries 1992,1994, 1995). Martin (1992), extended the grammatical notion of Theme to encompass larger units than the clause. Martin's "hyper-Theme" and "macro-Theme" relate to the structure of the paragraph and beyond; he proposes that, analogous to the Theme of a clause, there is a hyper-Theme of the paragraph and a macro-Theme of the text (1992: 437). This provides the essential bridge between the "micro" grammar of the clause and the "macro" semantics of the text. Matthiessen (1995a) showed how the continuity of clausal 352

ON THE GRAMMATICAL FOUNDATIONS OF DISCOURSE Themes throughout a text was the primary resource for organizing the text's ideational content — for enabling the text to function in the construction of knowledge. And in the paper I referred to earlier Fang et al. have established in Chinese a principle of thematic organization linking the clause to the clause complex, and begun to explore how the Themes that appear in these grammatical environments maintain a systematic progression through the discourse. Thus the grammatical systems that organize the flow of discourse at the "micro" level - the thematic (Theme + Rheme) system and the information (Given + New) system within the textual component of the grammar — do play a significant part in organizing the discourse as a whole. On the one hand, they are powerful resources for the creation of local meaning; on the other hand, they have at the same time a global significance for the semantic unfolding of the text. If we look at discourse "from below", as it were (that is, from the perspective of the lexicogrammar), it appears as the product of the interaction between Theme + Rheme and Given + New. In this connection it is interesting to note what Kress and van Leeuwen say about "Given and New" in relation to semiotic systems other than language (Kress and van Leeuwen 1995:186—92). They find that the construction of Given + New is one of the fundamental parameters of the organization of visual images, and in this way is central to the production and comprehension of all forms of pictorial art. 2 The creation of semantic unity within a text Whereas the thematic and information systems are realized through structural configurations in the grammar (Theme + Rheme, Given + New), there are other systems in the "textual" domain which are also realized by the grammar although not by structural means. These are the various systems of "cohesion". An account of such systems in English was given by Ruqaiya Hasan and myself in our book Cohesion in English (1976). These systems fall into four general categories: reference (tracking participants or other semantic elements), substitution/ellipsis (retrieving items of wording), conjunction (building up logical-semantic sequences) and lexical cohesion (by repetition, synonymy, hyponymy, etc.). Systems of cohesion in Chinese were first identified and described by Hu Zhuanglin (1981). The categories of cohesion in English and other languages have been widely discussed in educational and other contexts, and I shall not try to summarize them here. My aim is just to give a brief illustration 353

GRAMMAR AND DISCOURSE of their function in creating discourse. For the present purpose I shall confine myself to the systems of reference and conjunction. All discourse depends on the ability of language to refer: to construe categories of human experience - things, events and so on - and to keep track of them as the discourse proceeds. Since things, or "entities" (people, animals, objects), are on the whole more stable than phenomena of other kinds, most types of discourse depend heavily on tracking certain particular entities through the text, making it clear that the same persons and objects are still around. Every language has resources for doing this; for example, in Chinese, pronouns such as ta, tamen demonstratives zhe, na as well as items that are semantically more complex like geng (geng da implies some earlier reference to the quality da compare English comparatives such as bigger). Such features serve to create discourse, because the reference has to be resolved within the text. The cohesive device of reference is clearly related to the structural device of Given + New, since it would seem that anything previously referred to in the text must be inherently "given". But the two concepts are not identical. Something that has been referred to before may be foregrounded when it reappears on a different occasion, being presented as unexpected or contrastive; for example, zhe yifu bushi ta xi-de where the ta is coreferential with some earlier expression (e.g. ni muqin ) but is New in this particular context. The fact that something has been mentioned before does not restrict its potential in the flow of the discourse: it can still be given prominence as the New. There is an important difference between Chinese and English at this point. In English, in almost all environments where some person or other entity is presumed from the preceding text, it will be realized by the presence of a reference item; for example, Your mother has given me these clothes. She washed them. This is connected with the way English expresses the categories of transitivity and mood, which depends on the presence and location of such items in the clause; in this case, the position of she realizes declarative mood, and the presence of them realizes "effective" in transitivity (Halliday 1985/1994). Thus these two items cannot be ellipsed. (There is ellipsis in English; but it is a "copying" device, not a referential one.) In Chinese on the other hand coreference is typically established without being signalled by any pronoun; for example, Ni muqin gei wo zhejijian yifu. Xihao le (This may also be called "ellipsis"; but if so it is important to point out that it means something different — that in Chinese ellipsis 354

ON THE GRAMMATICAL FOUNDATIONS OF DISCOURSE is a referential device, which in English it is not; cf. Fang et al. 1995.) In formal grammars this phenomenon in Chinese is often referred to as "pro-drop" (dropping the pronoun), but this is a disgracefully anglocentric term: it implies that something that is really present (or else ought to be present!) has been dropped, or got left out. But it is English, not Chinese, which is exceptional; there are relatively few languages in which the reference item is obligatory in the way that it is in English. A clause in Chinese like xihaole , which displays this referential ellipsis, has exactly the same cohesive power in the discourse that English she's washed them has. (Of course, the reference in such a clause might be to the context of situation; but that is true of reference in general, not a special feature of any particular wordings in either English or Chinese.) In addition to creating cohesion by tracking participants, referential expressions also serve to maintain continuity of Theme. (Again, this is so no matter whether the reference is signalled by pronouns or by ellipsis.) Both in Chinese and in English the REFERENCE system intersects with the THEME system to produce a powerful discursive effect: that of an entity being "chained" along the text both as a participant in a range of different processes and as the point of departure for a sequence of different messages. Because such an entity is semantically complex (it is both a participant in some process and a Theme), the effect of its persistence through the discourse is especially striking. The other cohesive system I want to mention here is that of CONJUNCTION. The conjunction system sets up logical-semantic relations between one piece of text and another: relations of equivalence, addition, alternation, adversity, comparison and contrast, cause, time, condition and concession (Halliday and Hasan 1976: ch. 5; Halliday 1985/1994 ch. 9; Martin 1992: ch. 4). This system is unique in that it operates both structurally and cohesively. Structurally, it takes the form of a grammatical relationship between clauses in a clause complex: either with one clause dependent on the other ("hypotactic") or with the two clauses having equal, mutually independent status ("paratactic"). Thus for example in English: hypotactic: paratactic:

I put my coat on because I was feeling cold I was feeling cold, so I put my coat on.

The paratactic type is less fully grammaticized; it is in a sense intermediate between the hypotactic and the purely cohesive: cohesive:

I was feeling cold. That's why I put my coat on.

355

GRAMMAR AND DISCOURSE The expression that's why is an expression of the logical-semantic relation of "cause"; but, while it creates cohesion, it does not set up any structural relationship in the grammar. There is the same range of conjunctive resources in Chinese: hypotactic, paratactic, and cohesive. There is an interesting contrast between the two languages in the realization of the hypotactic: in English, the dependent clause must be marked by a conjunction, whereas the dominant clause may or may not be: hypotactic:

If you're feeling cold, (then) put your coat on.

In Chinese, it is the other way round: the dominant clause must be marked by a conjunction, the dependent clause may or may not be: hypotactic:

Wo

(yinwei)

juede leng jiu

chuanshang

paratactic:

Wo iuede leng, suoyi chuanshang dayi

dayi '

A systemic study of the clause complex in Chinese was undertaken by Ouyang Xiaoqing (Ouyang 1986). As in English, the system of conjunction in Chinese also functions cohesively (cf. Hu 1981): cohesive:

Wo juede leng. Jieguo shi, wo chuanshang dayi

In fact, both in English and in Chinese there is some difference between the grammatical categories of conjunction in the clause complex and the categories that set up non-structural, cohesive relations. But they cover more or less the same logical-semantic space. The major distinction that is made within this space is that between expansion and projection: in expansion, one term in the relationship elaborates, extends or enhances the other term; in projection, one term is quoted or reported by the other. Let us focus on the relationship of expansion. Typical logical-semantic relations within this category in both languages are: (elaborating) 'that is', 'in other words', 'for example'; (extending) 'and', 'or', 'but'; (enhancing) 'when', 'because', 'if, 'although', 'then', 'so', yet'. (The examples given above were of the "enhancing" type.) Now, these types of expansion turn up all over the grammar; in Chinese, for example, we find that they define the three basic types of relational process: elaborating, with shi as in ta shi jiaoshou 'he is a/the teacher'; extending, with you as in ta you qiche 'he has a car'; enhancing, with zai as in ta zai ketang 'he is-at the classroom'. But the question that concerns us here is their significance 356

ON THE GRAMMATICl foundations of discourse for the construction of discourse. We have seen that, in the form of cohesion, they establish semantic links between larger components, those that are beyond the reach of grammatical structure. But will we find logical-semantic relations of this kind defining - or at least contributing to — the structure of the text as a whole? Let me clarify this question a little further. When we are considering the relation between grammar and discourse, we will certainly find that different texts are distinguished from each other, and one part of a text is distinguished from another part of the same text, by their selection of features in the grammar. This point has been brought out by Ren Shaozeng (Ren 1992). Typically we find differences in the relative frequency of terms in grammatical systems: one text, or one part of a text, may show a preponderance of certain process types in transitivity, or certain selections in modality; and if this turns out to be a regular feature characterizing a whole class of texts, where those texts also have a common context of situation, then we recognize that grammar has a role in defining discourse types, or registers. Now, the logical-semantic relations that make up the conjunction system display regularities of this kind just as other grammatical systems do: we find certain kinds of conjunction favoured in certain texts or portions of texts. So, for example, expository texts tend to contain a large number of elaborating clause nexuses, whereby the speaker (or, more often, the writer) attempts to clarify some general proposition by repeating it in different words and giving examples. In this sense the conjunction systems contribute to the characteristic properties of a text by virtue of the way they are deployed. But this is not the question I am asking here. The question is, rather, whether the conjunctive relations which link one clause with another, and in so doing form a clause complex, also serve to link one piece of discourse with another, and in so doing form them into a text. Is there any deployment of conjunction analogous to the way the system of reference constructs a narrative by the "chaining" of participants? If we come back for a moment to conjunction as a cohesive relation, joining two passages in a text which are not related in grammatical structure: Martin has argued that a cohesive relation of conjunction should be recognized on semantic criteria alone, even when there is no actual conjunction present - no marking of any kind (Martin 1992: ch.4). This does present certain problems; in particular, it then becomes necessary to explain the difference in meaning between the marked and the unmarked variants; for example, between I felt cold. So I put on my coat. and I felt cold. I put on my coat. But Martin's point — which is a valid 357

GRAMMAR AND DISCOURSE one — is that in both cases the relationship is construed in the lexicogrammar; the difference is simply that in the second case it is not authenticated by a conjunction. Thus with I felt cold. I put on my heavy coat, the meaning 'therefore' is clearly present in the wording of the two successive clauses, even though the causal relation itself is not explicitly marked. This principle is embodied in Rhetorical Structure Theory as developed by Mann et al. (1992). Rhetorical Structure Theory is used to chart the progression of the argument through a paragraph or longer segment of a written text. This starts, usually, with relations between single clauses; but these relations are then extended across larger and larger portions of text, so that the text appears as a hierarchy of semantic units. The relations that are set up between pairs of units, at any rank, do not depend on any specific form of marking. Where Rhetorical Structure Theory differs from the cohesive system of conjunction, however, is that the relational categories that it invokes — such things as evidence, motivation, background, solutionhood — are set up purely on semantic grounds, without being derived from systemic features of the grammar (but see Matthiessen forthcoming). This does not mean they are invalid; but it does mean that the criteria for assigning them are not explicit, and hence the analysis tends to be rather subjective. For the same reason they do not illustrate the point that I am raising here, which is: do the logicalsemantic relationships that are explicitly construed in the grammar play any part in the larger construction of discourse? This will be the topic of the final section. 3 Logical-semantic relations in discourse structure We interpret the structure of grammatical units as configurations of functions. For example, in English, a clause is a configuration of Actor + Process + Goal, and also of Theme + Rheme; a nominal group is a configuration of Deictic + Numerative + Epithet + Classifier + Thing, and so on. There is an important analogy, in this respect, between grammatical units and discourse-semantic units (texts), in that they are built up on similar principles. Hasan has shown that this same notion of a functional configuration can be applied to the structure of a text (Hasan 1978,1984,1994). Some notion of this kind was implicit in theories of literature, and the idea of narrative structure has been familiar since the work of Vladimir Propp on the Russian folk tale. What Hasan proposed was the notion of a "generic structure potential"; and she demonstrated that a register could be defined by its generic structure potential, having 358

ON THE GRAMMATICAL FOUNDATIONS OF DISCOURSE some obligatory and some optional elements, some ordering that was fixed and some that was fluid, some possibility of iterative patterns closely analogous to structure as found in grammar. Thus the structure of the nursery tale in English could be represented (in slightly simplified form) as: [«Placement»AInitiatingEventA]SequentEvent[A(Finale)A(Moral)] where the element enclosed in ( ) is optional, + indicates fixed order, °fluid order, and [ ] define the limits within which the ordering may vary; the symbol ^ indicates that the element in question may be iterated (Hasan 1984: 80). Hasan has presented the generic structure potential for various other registers, such as a retail shopping encounter and a dissertation defence. And just as in identifying grammatical structures we need to have the relevant semantic information, so also "for the construction of the [generic] structure potential, we need to have sufficient contextual information" (Hasan 1994: 167). In other words, we are always interpreting linguistic structure, whether grammatical structures or discourse structures, "from above" as well as "from below". Now, in grammar we find two different kinds of structure, with a basic typological distinction between them. One is the configurational type, like the clause consisting of Actor + Process + Goal, where each element has a different function with respect to the whole. The other is the iterative or "tactic" type, like the clause complex made up of an indefinitely long sequence of clauses with each nexus linked by a relation of expansion or projection. It seems that the same distinction may be found between two kinds of structure in discourse. Consider the various structures associated with narrative, such as the English nursery tale that Hasan analysed. The simplest kind of narrative is that which Martin and Rothery (1981) refer to as a "recount", which is a series of episodes linked by the meaning 'and'. A recount is an iterative structure; we might refer to it as an "episode complex", where the logical-semantic relation is, and remains throughout, one of the "extending: additive" kind of expansion. When children first make narratives, this is how they typically do it. The conventional narratives of adults have a more complex structure. This can be represented configurationally, as Hasan does for the nursery tale; compare Labov and Waletsky's (1967) analysis of narratives of personal experience: (Abstract)+[(Orientation+)Complication]+[Evaluation°Resolution](+Coda)

359

GRAMMAR AND DISCOURSE (I have shown this using Hasan's notation.) These are labelled as functionally distinct constituents. But notice that most of them occur in a fixed order (and see Martin 1992: 556-9 for a critical assessment of the concept of "evaluation"); and their meaning is dependent on the order in which they occur. Thus Resolution makes no sense unless there has first been a complication to resolve; and Complication must have been preceded by something that it is a complication of. And the Orientation must come first - or at least before the narrative action starts. It might be helpful to analyse this as an iterative structure, one that is based on the same logical-semantic relations that make up the grammatical system of conjunction. As in the recount, each segment is an expansion of something that has gone before; but the sequence is no longer constructed out of simple extension by "and". Instead, the narrative switches to the adversative kind of extension "but" — that is the • Complication; and when it comes to the Resolution, this is no longer extending but enhancing. The Resolution embodies a concessive relation "although": the meaning is "despite the complication, matters have been resolved". I am not suggesting that it is wrong to label the components of a narrative as a configuration of functional elements; this is one way of looking at them —just as the components of a conditional clause nexus, the 'if ... then' relation, have traditionally been labelled as "protasis + apodosis". But we may gain further insight into it if we also interpret it in terms of the logical-semantic relations of conjunction. Whereas the configurational elements that we can recognize in narrative and other discourse genres seem to have no counterpart in grammatical structure, the analysis of a text as a sequence of phases linked by types of expansion suggests that even the "macro" organization of discourse is construed along lines laid down in the grammar (cf. Martin's concept of macro-Theme, above). Let me recapitulate a little. One of the resources out of which speakers construct connected discourse is the system of conjunction, which establishes semantic relations between portions of a text that are not (or may not be) related by grammatical structure. These semantic relations are construed by the lexicogrammar in the form of conjunctive expressions, whose own grammatical status may be quite varied: in Chinese, as in English, they may be adverbs or adverbial groups, "prepositional" phrases, or clauses. These items themselves enter into the grammatical structure; but they do not create structure — they create cohesion. Conjunctive relations cover essentially the same semantic space as the logical-semantic relations that are construed structurally within the 360

ON THE GRAMMATICAL FOUNDATIONS OF DISCOURSE clause complex: the three types of expansion (elaborating, extending, enhancing). Same examples from Chinese are: [elaborating]

na jiushi shuo huan yiju hua shuo zong er yan zhi , piru [extending] hai you erqie huozhe buguo buran-de hua [enhancing] tongshi zheyang yilai jieguo yuanlai wulun ruhe When we come to consider the relations which obtain between the elements of a discourse structure, in most types of discourse we will not find these explicitly marked. But they may still be linked by the same kinds of semantic relations. Thus, a new element might elaborate on what has gone before — repeating it, clarifying it or exemplifying it; it might extend what has gone before — adding something further, or stating some alternative, adversative or reservation; or it might enhance what has gone before — in terms of time, cause and effect, condition and consequence, comparison, concession and so on. In other words, the same logical-semantic relations that are construed in the grammar, both structurally and cohesively, by the conjunction system are also available for constructing discourse: not realized by specific lexicogrammatical markers, but nevertheless construed by the progressive unfolding of the text. They are present in the broader semantic patterns created by the wording. Not all types of discourse would display patterns of this iterative kind. Narratives, or most of what we recognize as different types of narrative, probably do; and so also does conversation (cf. Hasan 1984, 1994; Hasan (ed.) 1985;Eggins 1990;Eggins and Slade 1997). But many types of discourse are essentially configurational in their structure, like guidebooks, manuals, or advertisements for jobs. Texts of this kind could often be reordered, or set out in different ways on a printed page, without disrupting the relationship among the parts. It is not the case that texts of a given register belong clearly to one kind or the other; rather, the analysis in terms of conjunctive relations will be more revealing with some varieties of text while having little or no relevance to certain others. It is always difficult to illustrate points of text analysis within the limits of a short paper (for obvious reasons!). But I am appending here two short specimens of Chinese text which will perhaps bring out the difference between text that is organized more configurationally and text that is organized more by relations of expansion. Text (a),Bai She Zhuan 361

GRAMMAR AND DISCOURSE (the Legend of White Snake), is a narrative, taken from a programme of a Peking opera performance. Text (b), Deng Wanxia (Tang Yuen-ha) and Cai Zhengren is a biography of two of the leading performers, taken from the same source. By and large, my analysis would work also with the English translations; but it is the Chinese original that I am considering here. Text (a), the narrative, has an Initiating Event, the encounter and subsequent marriage of Bai Suzhen and Xu Xian (lines 1—5). This is followed by three rounds of Complication + Resolution: (1) Fahai plots and brings about Xu Xian's death (lines 5-8); Bai brings Xu Xian back to life (lines 8—9); (2) Fahai captures Xu Xian and refuses to release him (lines 9-12); Bai Suzhen enlists help and Xu Xian escapes (lines 12-16); (3) Xiaoqing is angry with Xu Xian and tries to kill him (lines 16-17); Bai restrains her and effects a reconciliation (lines 17—19). There is no separate constituent of Evaluation; but there is a positive appraisal in the final clause nexus of the last Resolution ('the three of them become reconciled and are once again united', where hehao ru chu and t u a n y u a n b o t h carry the semantic feature 'desirable'). We can identify the logical-semantic relations whereby this text is organized: (a) between Initiating Event and Complication (1) extending : adversative (b) between Complication and Resolution enhancing : concessive (c) between one Complication and the next enhancing: temporal Text (b), on the other hand, which is biographic, displays a more configurational type of structure. Let us consider just the first of the two biographies, that of Deng Wanxia (Tang Yuen-ha). This has three fairly distinct elements in its structure, which we might call Recognition, Qualification and Position. The Recognition consists of two parts, the awards and honours she has received (lines 1-2, and also lines 7-8), and the stars who have taught her and praised her work (lines 3—5). The Qualification gives details of the roles she has taken and her achievements in these roles (lines 6—7). The Position is an account of her present responsibilities (lines 9—11). These three elements are not logically related, and their function does not depend on the sequence in which they occur: they could appear in any order and the overall configuration would be unaffected. Much the same pattern recurs with the second biography below. 362

ON THE GRAMMATICAL FOUNDATIONS OF DISCOURSE I am not suggesting that texts of every kind exclusively favour one or other of these two modes of structure. What we recognize as discourse will typically embody some features that can best be interpreted as configurations and some that can be explained as types of expansion — as envisaged in Hasan's (1984) concept of "generic structure potential". Much of the discourse that we encounter in the course of our lives is a mixture of patterns, some larger some smaller, some clearly visible on the surface and some more subtle and obscure. My concern is just to point out that the logical-semantic relations which are systemic in the lexicogrammar, and appear in a discursive form as relations of cohesion, also appear figure on a more "macro" scale in the generic organization of texts. Again, I do not want to press the analogy between grammatical structure and discourse-semantic structure too far. But I think that, in our experience, modelling semantic patterns in terms of the theoretical concepts developed in lexicogrammar, while it has its dangers, is on the whole more fruitful than sidelining the grammar and setting up semantic categories solely on the basis of some rather vague appeal to meaning. This has been, as I argued at the beginning, a problem with discourse studies in general: they have tended to be pursued without reference to the study of the other strata of language. What Hasan, Martin, Matthiessen, Fries and others have done is to draw attention to those aspects of the grammar which provide a point of entry to theorizing about discourse in explicit linguistic terms. There is a very good opportunity in Chinese linguistic studies for maintaining a close association between theory of grammar and theory of discourse, such that each contributes significantly to the other; and at the same time for exploring the differences, and the similarities, between discourse patterns in Chinese and in other languages.

363

This page intentionally left blank

APPENDIX

Tang Yuen-ha The only Hong Kong artist to have won the Chinese Theatre Plumblossom Awards, the most prestigious award for Chinese theatre workers. Tang was taught personally by the renowned Kun Opera master Yu Zhenfei, who valued her as one of his most talented and accomplished students. Gifted with a full and rounded voice and backed by rigorous musical training, Tang excels in both singing and acting. She has an extensive repertoire in both Beijing operas and Kun operas. She is especially renowned for the detailed understanding and refined expression of her characters, which results in moving dramatic representations. Tang is acclaimed as "the Pearl of Beijing Opera and Kun Opera of Hong Kong". Tang is Chairman of Hong Kong Jing Kun Arts Association and Executive Member of the Chinese Opera Performance Society of China.

365

APPENDIX

Cai Zhengren First Rank National Artist. Winner of the Fourth Chinese Theatre Plum-blossom Award and the Fifth Shanghai Theatre White Orchid Award. Graduate of the first Kun Opera Class of the Shanghai Academy of Chinese Opera. Cai specializes in Male Role. He is most renowned for Hatted Male Role, usually representing young scholars. He has learned his art mainly from the Kun master Yu Zhenfei. He has a wide repertoire in both Kun operas and Beijing operas. He excels in delivering arias and speeches in a full and strong voice, combined with a charming stage presence. Cai has toured his performances and delivered lectures extensively.He is the present Director of Shanghai Kunju Theatre, Member of the Chinese Dramatists Association and Vice-chairman of Shanghai Dramatists Association.

The Legend of White Snake Bai Suzhen (the white snake goblin) longs for romance and leaves her hidden abode on Mount Emei with Xiaoqing (the blue snake goblin) after their transformation into human shape. Meeting at the West Lake on a rainy day, Bai and Xu Xian fall in love with each other. Xu lends his umbrella to the ladies and arranges to visit them to get it back. Subsequently he marries Bai on that day. Monk Fahai of Jinshan Temple learns about the match and disapproves. He visits Xu and instructs him to persuade Bai to drink Xionghuang wine on the day of the Dragon Boat Festival, which turns her back to her snake shape. Xu is scared to death, but is brought back to life with the magic ganoderma Bai steals from the Magic Mountain. Now recovered, Xu goes to worship the gods 366

APPENDIX

in Jinshan Temple and is kept in custody by Monk Fahai. Bai pleads for his release but her request is refused. Bai and Xiaoqing gather the water goblins to seize and flood the temple. Monk Fahai asks for help from the Heavenly Army to fight back. Since Bai is pregnant, she cannot hold out and retreats to the Broken Bridge, where Xu also manages to escape to with the help of a junior monk. Xiaoqing cannot forgive him for abandoning Bai and wants to kill him. Bai convinces her to accept Xu's apology and the three are reconciled. Bai Suzhen — Tang Yuen-ha Xu Xian — Cai Zhengren Xiaoqing - Wei Huili Monk Fahai - Wang Wenzhi Jialan — Zhang Siquan Junior monk — Huang Baixue Crane fairy — Zhou Long Deer fairy — Huang Jingping Water goblins and the Heavenly Army —Joint-troupe Ensemble

The texts were taken from a programme for a Chinese opera at City Hall, Hong Kong,July 3,1997.

367

BIBLIOGRAPHY

Aleksejev, V. M. (1932) The Chinese Script and Its Latinization, Izd. AN SSSR, Moscow. (AneKceeB, B. M. (1932) Kumaucmx uepoejiu-

(fiujiecKdH

nucbMeHHOcmb u ee jiamunu3aif.uH, H^j\. AH CCCP,

JleHHHrpaA.) Allen, W. S. (1953) 'Relationship in Comparative Linguistics', Transactions of the Philological Society. Bazell, C. E., Catford, J. C., HaUiday, M. A. K. and Robins, R. H. (eds) (1966) In Memory ofj. R. Firth, London: Longman. Bazerman, Charles (1988) Shaping Written Knowledge: the genre and activity of the experimental article in science, Madison, WI: University of Wisconsin Press [Rhetoric of the Human Sciences]. Catford, J. C. (1977/82) Fundamental Problems in Phonetics, Bloomington, IN: Indiana University Press/Midland Books. Chao, Y. R. (1934) 'The non-uniqueness of phonemic solutions of phonetic systems', Academia Sinica, Bulletin of the Institute of History and Philology, iv 2. Chao, Y. R. (1940) 'Distinctions within Ancient Chinese', Harvard J. Asiatic Studies 5. Chao, Y. R. (1945) 'The logical structure of Chinese words', Language, xxii 1. Chao, Y. R. (1947) Cantonese Primer, Cambridge, MA: Harvard University Press. Chao, Y. R. (1948) Mandarin Primer, 2 Vols, Cambridge, MA: Harvard University Press. Chao, Y R. (1968a) A Grammar of Spoken Chinese, Berkeley and Los Angeles: University of California Press. Chao, Y. R. (1968b) Language and Symbolic Systems, Cambridge: Cambridge University Press. 368

BIBLIOGRAPHY

Chao, Y. R. and Yang, L. S. (1947) Concise Dictionary of Modern Chinese, Cambridge, MA: Harvard University Press. Chavannes, E. (1904-08) 'Inscriptions et pieces de chancellerie chinoises de 1'epoque mongole', T'oung Pao, v, 1904, vi, 1905, ix, 1908. Cleave, F. W. (1949) (review of) 'Erich Haenisch, "Die geheime Geschichte der Mongolen"', HJAS xii. Chen Yuan (1934) Yuan pi-shih i yin yung tzu k'ao (The Chinese transcription of the "Secret History") Academia Sinica, Institute of History and Philology, Peiping. ( ) Clyne, Michael (1991) 'Directionality, rhythm and cultural values in discourse', in Frances Christie (ed.) Literacy in Social Processes: papers from the Inaugural Australian Systemic Network Conference, Deakin University, January 1990, Darwin: Northern Territory University Centre for Studies of Language in Education. Danes, Frantisek (1974) 'Functional sentence perspective and the organization of text', in F. Danes (ed.) Papers on Functional Sentence Perspective, The Hague: Mouton. Demieville, P. (1951) 'Archaismes de prononication en chinois vulgaire', T'oung Pao XL. Downer, G. B. (1963) 'Traditional Chinese phonology', Transactions of the Philological Society. Dragunov, Aleksandr Aleksandrovich (1930) 'The /zPhags-pa script and Ancient Mandarin', IzvAN (Otdelenije gumanitarnych nauk), vii 9— 10. Dragunov, Aleksandr Aleksandrovich (1940) Researches into Dungan Grammar, I: Categories of Aspect and Tense, Izd. AN SSSR MoscowLeningrad. (flparyHOB, A. A. (1940) MccjiedoeanuH e o6jiacmu dymancKOU ^paMMamuKU (I: Kamezopun euda u epeuenu e dymancKoM H3UKe) H3^. AH CCCP, MocKBa-JIeHHHrpa/i:.) Dragunov, A. A. (1952) Studies in Modern Chinese grammar, I : Parts of Speech, Izd. AN SSSR, Moscow-Leningrad. (^paryHOB, A. A. (1952) Hccjiedoeanue no ^paMMamuKe coepejweoHHoeo KumaiicKoeo H3bma (I: *lacmupemi} H3A, AH CCCP, MocKBa-JleHMHrpa^.) Eggins, Suzanne (1990) 'Conversational Structure: a systemic-functional analysis of interpersonal and logical meaning in multi-party sustained talk'. Ph.D. thesis, University of Sydney. Eggins, S. and Diana Slade, Analyzing Casual Conversation, London: Cassell, 1997. Enoki Kazuo (1951) 'Gencho Hishi kankei bunken kammoku' (A short bibliography concerning the "Secret History of the Mongols") 369

BIBLIOGRAPHY Toyo

Gakuho, xxxiii

3—4. (

Fang Yan (1989) 'A tentative study of Theme and Rheme in Chinese', Journal ofQinghua University 2. Fang, Yan (1993) 'A contrastive study of Theme and Rheme in English and Chinese', Keqi Hao, Hermann Bluhme and Renzhi Li (eds) Proceedings of the International Conference on Texts and Language Research, 29-31 March 1989,Xi'an: Xi'an Jiaotong University Press. Fang, Yan, McDonald, Edward & Cheng, Musheng (1995) 'On Theme in Chinese: from clause to discourse', in Ruqaiya Hasan and Peter H. Fries (eds), On the Subject and Theme: a discourse functional perspective, Amsterdam and Philadelphia: Benjamens. Fawcett, Robin P. (1988) 'Towards a systemic flowchart model for discourse analysis', in Robin P. Fawcett and David J. Young (eds) New Developments in Systemic Linguistics, Vol.2: theory and application, London: Pinter. Firth, J. R. (1937/53) [with B. B. Rogers], 'The structure of the Chinese monosyllable in a Hunanese dialect (Changsha)', Bulletin of the School of Oriental & African Studies 8 (reprinted in Firth, 1957). Firth, J. R. (1948/57) 'Sounds and prosodies', Transactions of the Philological Society (reprinted in Firth, 1957). Firth,J. R. (1949) 'Atlantic Linguistics', ArchL, I 2. Firth, J. R. (1950) 'Personality and language in society', Sociological Review, xlii 2. Firth, J. R. (195la) 'General Linguistics and Descriptive Grammar', Transactions of the Philological Society. Firth,J. R. (1951b) 'Modes of Meaning', Essays and Studies of the English Association. Firth, J. R. (1957) Papers in Linguistics 1934-1951, London: Oxford University Press. Firth, J. R. (1968) 'Linguistic analysis as a study of meaning', in F. R. Palmer (ed.) Selected Papers of J. R. Firth 1952-59, London: Longman. Franke, O. (1952) Geschichte des chinesischen Reiches, V Band, Berlin: Walter de Gruyter and Co. Frei, H. (1941) 'Un systeme chinois des aspects', Acta Linguistica, II.3, 137-150. Frei, H. (1956—57) 'The ergative construction in Chinese', Gengo Kenkyu 31-2. Fries, Peter H. (1981) 'On the status of Theme in English: arguments from discourse', Forum Linguisticum 6.1 (reprinted in Janos Petofi 370

BIBLIOGRAPHY and Emel Sozer (eds) Micro and Macro Connexity of Texts, Hamburg: Helmut Buske, 1983). Fries, Peter H. (1992) 'The structuring of information in written English text', Language Sciences, 14.4. Fries, Peter H. (1994) 'On Theme, Rheme and discourse goals', in Malcolm Coulthard (ed.) Advances in Written Text Analysis, London: Routledge. Fries, Peter H. (1995) 'Themes, Methods of Development, and Texts' in Ruqaiya Hasan and Peter H. Fries (eds), On Subject and Theme: a discourse functional perspective, Amsterdam and Philadelphia: Benjamens. Gao, Mingkai (1948) Xanjy Yfa Lun (Shanghai, Kaiming Shudian .) Haenisch, E. (1931) 'Untersuchungen iiber das Yiian-ch'ao Pi-shi', Abhandlungen der philologisch-historischen Klasse der sachsischen Akademie der Wissenschaften iv, Leipzig. Haenisch, E. (1933) 'chan "Halteplatz" und tiao "Kampfturm", zwei Fremdworter' MSOS (Mitteilungen des Seminars fur orientalische Spracheh) xxxvi (Ostasiatische Studien), Berlin. Haenisch, E. (1937) Manghol un Niuca Tobca'an (Yuan-ch'ao Pi-shi), Leipzig: Otto Harrassowitz. Haenisch, E. (1939) Worterbuch zu Manghol un Niuca Tobca 'an (Yuan-ch 'ao Pi-shi}, Leipzig: Otto Harrassowitz. Haenisch, E. (1948) Die geheime Geschichte der Mongolen (2nd edn, revised), Leipzig: Otto Harrassowitz. Haenisch, E. (1952) Sino-mongolische Dokumenten vom Ende des 14. Jahrhunderts, Berlin: Akademie-Verlag. Halliday, M. A. K. (1956) 'Grammatical categories in Modern Chinese', Transactions of the Philosophical Society. Halliday, M. A. K. (1957) 'Some aspects of systematic description and comparison in grammatical analysis', Studies in Linguistic Analysis (Special Volume of the Philological Society), Oxford: Blackwell. In Collected Works, Vol. 1, chap. 1,2002. Halliday, M. A. K. (1959) The Language of the Chinese "Secret History of the Mongols", Oxford: Blackwell (Publications of the Philological Society 17). This volume, chap. 1. Halliday, M. A. K. (1976) System and Function in Language: selected papers G. R. Kress (ed), London: Oxford University Press. Halliday, M. A. K. (1981) 'The origin and early development of Chinese phonological theory,' in R. E. Asher and Eugenie Henderson (eds), Towards a History of Phonetics, Edinburgh, Edinburgh University Press. 371

BIBLIOGRAPHY

Halliday, M. A. K. (1982) 'How is a text like a clause?', in Sture Allen (ed.) Text Processing: text analysis and generation, text typology and attribution, Stockholm: Almqvist and Wiksell [Proceedings of Nobel Symposium 51]. In Collected Works, Vol. 1, chap. 9,2002. Halliday, M. A. K. (1985) 'It's a fixed word order language is English', ITL Review of Applied Linguistics 67—68. In Collected Works, Vol. 17, chap. 7,2005. Halliday, M. A. K. (1988) 'On the language of physical science', in Mohsen Ghadessy (ed.) Registers of Written English: situational factors and linguistic features, London and New York: Pinter Publishers. In Collected Works, Vol. 5, chap. 5,2004. Halliday, M. A. K. (1992) 'Systemic grammar and the concept of a "science of language", Waiguoyu 1992.2 [78]. Halliday, M. A. K. (1994) An Introduction to Functional Grammar, 2nd edn, London: Edward Arnold [1st edn 1985]. Halliday, M. A. K. and Hasan, Ruqaiya (1976) Cohesion in English, London: Longman. Halliday, M. A. K. and Martin J. (eds) (1993) Writing Science: Literacy and dicursive power. London: Palmer Press. Halliday, M. A. K. and Matthiessen, Christian (1999) Construing Experience through Meaning: a language-based approach to cognition, London: Cassell. Harris, Z. S. (1952) 'Discourse Analysis', Language, xxviii 1,4. Hartman, L. M. (1944) 'The segmental phonemes of the Peiping dialect', Language, xx 1. Hasan, Ruqaiya (1978) 'Text in the systemic-functional model', in W. U. Dressier (ed.) Current Trends in Text Linguistics, Berlin: de Gruyter. Hasan, Ruqaiya (1984) 'The nursery tale as a genre', Nottingham Linguistic Circular 13. Hasan, Ruqaiya (1994) 'Situation and the definition of genre', in Allen Grimshaw (ed.) What's Going On Here? Complementary studies of professional talk, Norwood, NJ: Ablex. Hasan, Ruqaiya (ed.) (1985) 'Discourse on Discourse', Applied Linguistics Association of Australia [Occasional Papers 7]. Hasan, Ruqaiya and Fries, Peter H. (eds) (1995) On Subject and Theme: a discourse functional perspective, Amsterdam and Philadelphia: Benjamins. Hattori, Shiro, (1946) Gencho Hishi no mokogo wo arawasu kanji no kenkyu (A study of the Chinese transcription of Mongolian in the "Secret History") .Tokyo. ( ) 372

BIBLIOGRAPHY Hattori, Shiro (1951) 'Gench5 Hishi onyakubon genten hPags-pa-ji honsetsu ni tsuite' (The theory that the original transcription of the "Secret History" was in hPags-pa script) Gengo Kenkyu xix-xx. ( ) Hauer, E. (1931) (review of) ' "Untersuchimgen iiber das Yiian-ch'ao Pi-shi" von Erich Haenisch', MS OS xxxiv (Ostasiatische Studien), Berlin. Henderson, E.J. A. (1966) 'Towards a Prosodic Statement of Vietnamese Syllable Structure', in Bazell, C. E. et at, In Memory of J. R. Firth, London: Longman. Hill, T. (1966) 'The technique of prosodic analysis', in C. E. Bazell et al, In Memory ofj. R. Firth, London: Longman. Hockett, C. E (1947) 'Peiping phonology',Journal of the American Oriental Society,Ixvii 4. Hu, Zhuanglin (1981) 'Textual Cohesion in Chinese', MA Hons thesis, University of Sydney. Hu, Zhuanglin (1993) 'A semantic-functional approach to word order in Chinese', Keqi Hao, Hermann Bluhme and Renzhi Li (eds) Proceedings of the International Conference on Texts and Language Research, Xi'an: Xi'an Jiaotong University Press. Hu, Zhuanglin, Zhu Yongsheng and Zhang Delu (1989) Xitong Gongneng Yufa Gailun [A Survey of SystemicFunctional Grammar], Changsha: Hunan Educational Publishing House. Hung, W. (1951) 'The transmission of the book known as "The Secret History of the Mongols"', Harvard}. Asiatic Studies, xiv, 3,4. Kao, Ming-Kai, (1948) Han-yu yu-fa lun (Chinese grammar), Shanghai: Kai-ming. ( ) Karlgren, B. (1913) A Mandarin Phonetic Reader in the Pekingese dialect, Stockholm: Archives d'etudes orientales. Karlgren, B. (1915—24) Etudes sur la Phonologic chinoise, Stockholm: Archives d'etudes orientales, 4 volumes. Karlgren, B. (1922) 'The reconstruction of Ancient Chinese', T'oung Pao 21. Karlgren, B. (1928) The Romanization of Chinese, London: The China Society. Karlgren, B. (1940) Grammata Serica: script and phonetics in Chinese and Sino-Japanese, Stockholm: (Museum of Far Eastern Antiquities (= Bulletin of the Museum of Far Eastern Antiquities 12). Karlgren, B. (1951-52) 'Excursions in Chinese grammar', Bulletin of the Museum of Far Eastern Antiquities xxiii, 1951, xxiv, 1952. 373

BIBLIOGRAPHY Kobayashi Takashiro (1954) Gencho Hishi no kenkyu (Researches in the "Secret History"), Tokyo. ( ) Kotvic, V. L. (1925) 'Towards an edition of the Yuan-ch 'ao pi-shih', Papers of the College of Orientalists i. (KoTBHH, B. Jl (1925) 'K H3,n;aHHK) K)am> nao 6n ma,' SanucKu Koji/ie3uu BocmoKoeedoe i.) Kozin, S. A. (1941) The Secret History (Yuan-ch 'ao pi-shih), Izd. AN SSSR (Inst. of Oriental Studies), Moscow-Leningrad. (KO3HH, C. A. (1941) CoKpoeennoe cKa3anue (lOam uao 6u iuu] Hs^. AH; CCCP (HHCT. BocTOKOBeAeHHfl) MocKBa-JIeHnnrpaA.) Kratochvil, P. (1968) The Chinese Language Today: features of an emerging standard, London: Hutchinson (Hutchinson University Library). Kress, Gunther and Van Leeuwen, Theo (1995) Reading Images: the grammar of visual design, London: Routledge. Labov, William and Waletzky,Joshua (1967) 'Narrative analysis: oral versions of personal experience', in J. Helm (ed.) Essays on the Verbal and Visual Arts, Seattle: University of Washington Press. Ladefoged, P. and Wu, Z. (1984) 'Places of articulation: an investigation of Pekingese fricatives and affricates \Journal of Phonetics 12. Lewicki, M. (1949) La langue mongole des transcriptions chinoises du xive siecle. Le Houa-yi yi-yu de 1389, Prace Wroclawskiego towarzystwa naukowego (Travaux de la societe des sciences et des lettres de Wroclaw) A 29. Lock, G. (1989) 'Aspects of variation and change in the Mandarin Chinese spoken in Singapore', Australian Journal of Linguistics, 9,2. Lock, G. (1992) 'Non-segmental phonology and variable rules: investigating variation in Singapore Mandarin nasal finals', in Paul Tench (ed.), Studies in Systemic Phonology, London: Pinter. Long Rijin (1981) 'Transitivity in Chinese', MA Hons. thesis, University of Sydney. Lii Shu-hsiang (Lii Hsiang) (1940a) 'Han-yii ti-san-shen tai-tz'u shuo' (Third person pronouns in Chinese), Studia Serica, i 2.

( ) Lii Shu-hsiang (Lii Hsiang) (1940b) 'Shih "nin", "an", "tea", "tsan", fu lun "men" tzii" (The pronouns nin, an, tsa, tsan; and the syllable men), Studia Serica, i 2. ( ) Lii Shu-hsiang (Lii Hsiang) (1947) Chung-kuo wen-fa yao-lueh (Outline of Chinese grammar) (3 vols), Shanghai: Commercial Press. ( ' ) Mann, William C., Matthiessen, Christian and Thompson, Sandra A. (1992) 'Rhetorical structure theory and text analysis', in William C. Mann and Sandra A. Thompson (eds) Discourse Description: diverse

374

BIBLIOGRAPHY linguistic analyses of a fund-raising text, Amsterdam and Philadelphia: Benjamins. Margoulies, G. (1943) La langue et I'ecriture chinoises, Paris: Payot. Martin, J. R. (1992) English Text: system and structure, Amsterdam: Benjamins. Martin, J. R. (1993) 'Genre and literacy - modelling context in educational linguistics', Annual Review of Applied Linguistics 13. Martin, J. R. and Rothery, Joan (1981) 'The ontogenesis of written genre', University of Sydney Linguistics Dept.: Working Papers [Writing Project Report 2]. Martinet, A. (1942-45) 'Au sujet des "Fondements de la theorie linguistique" de Louis Hjelmslev', Bulletin de la societe de Hnguistique,^dii 1. Maspero, H. (1914) 'Sur quelques textes anciens de chinois parle', Bulletin de I'Ecolefranfaise d'Extreme-Orient,idv 4. Maspero, H. (1937) 'Les langues de 1'extreme orient,' Encyclopedie fran$aisel,40A-42.6. Matthews, R.H. (1947) A Chinese-English Dictionary, American edn revised by Y. R. Chao and M. Y. Wang, Cambridge, MA: Harvard University Press. Matthews, W. K. (1952-53) 'The Ergative construction in Modern Indo-Aryan', Lingua iii. Matthiessen, C. M. I. M. (1987) 'Notes on Akan phonology: a systemic interpretation', (mimeo). Matthiessen, C. M. I. M. (1992) 'Interpreting the textual metafunction', in Martin Davies & Louise Ravelli (eds) Advances in Systemic Linguistics, London: Pinter. Matthiessen, C. M. I. M. (1995a) 'THEME as enabling resource in ideational "knowledge" construction', in Mohsen Ghadessy (ed.) Thematic Development in English Text, London: Pinter. Matthiessen, C. M. I. M. (1995b) Lexicogrammatical Cartography: English systems, Tokyo: International Language Sciences Publishers. Matthiessen, C.M.I.M. (forthcoming) 'Textual semantics: RELATIONAL EXPANSION - Rhetorical Structure Theory'. McDonald, Edward (1992) 'Outline of a functional grammar of Chinese for teaching purposes', Language Sciences, 14.4. Meriggi, P. (1933) 'La structure des langues "groupantes"', Journal de psychologie, (Psychologie du langage). Mostaert, A. (1950-52) 'Sur quelques passages de 1'Histoire secrete des Mongols', Harvard}. Asiatic Studies, xiii, 1950, xiv, 1951, xv, 1952. Mullie, J. (1937/1970) The Structural Principles of the Chinese Language, Vols I—II, (Peiping, Peit'ang Lazarist Press).

375

BIBLIOGRAPHY

Murayama Shichiro (1953) '" Onyaku Mobun Gencho Hishi" ni tsuite' (On "The Mongolian version of the Secret History"), Toyo Gakuho, xxxv 3—4. ( ) Naka Michiyo (1907) Chingisu kan jitsuroku (The life of Genghis Khan), Tokyo. ( ) Ni, Xaishu (1949) Latinxua Sin Wenz Gailun, Shanghai. Ota Tatsuo (1953) 'Rokittai no gengo ni tsuite' (The language of the Rokittai), Chugoku gogaku kenkyukai ronshu (Journal of the Chinese Language Society) i, Nara. ( ) Ouyang, Xiaoqing (1986) 'Clause Complex in Chinese', MA Hons thesis, University of Sydney. Palladius (Kafarov) (1910) 'An ancient Mongolian history of Genghis Khan', Works of Members of the Russian Church Mission in Peking iv,

Peking. (IlAJiJiAAHyc (Ka(|)ApOB) (1910) 'CxapHHHoe MOHTOJibcicoe CKasanne o ^HHrncxane,' Tpydu HJICHU POCCUUCKOU ffyxoenou Muccuu e TIsKune, iv, HSKHH.) Palmer, E R. (ed.) (1970) Prosodic Analysis, London: Oxford University Press. Pelliot, P. (1913) 'Le titre mongol du Yuan tch 'ao pi che', T'oung Pao xiv. Pelliot, P. (1925) 'Les mots a H initiale, aujourd'hui amuie, dans le Mongol des xiiie et xive siecles',J/l clvi. Pelliot, P. (1930) 'Un passage altere dans le texte mongol ancien de 1'Histoire secrete des Mongols', T'oung Pao xxvii. Pelliot, P. (1948) 'Le Hoja et le Sayyid Husain de 1'Histoire des Ming', T'oung Pao xxxviii. Pelliot, P. (1949) Histoire secrete des Mongols, Oeuvres posthumes, Paris: Librairie d'Amerique et d'Orient, Adrien-Maisonneuve. Ponceau, P. S. Du (1848) A Dissertation on the nature and character of the Chinese system of writing, Philadelphia: American Philosophical Society. Poppe, N. (1950) (review of) 'Paul Pelliot, "Histoire secrete des Mongols"', Harvard], Asiatic Studies xiii. Prakasam, V. (1987) 'Aspects of word phonology', in M. A. K. Halliday and R. P. Fawcett (eds) New Developments in Systemic Linguistics, Vol. 1, London: Pinter. Prusek, J. (1945) 'Quelques remarques sur les aspects en chinois,' Archiv Orientdlni, Prague. Prusek, J. (1946) 'La fonction de la particule di dans le chinois medieval', Archiv Orientdlni xv. 376

BIBLIOGRAPHY Prusek, J. (1950) 'Quelques remarques sur les aspects en chinois,' Archiv Orientalni, XVIII 1—2. Ren, Shaozeng (1992) Yujing zai xushi yupian-zhong-de yuyan tixian [The linguistic realization of context in narrative discourse], Waiguo. Robins, R. H. (1952) 'Noun and verb in universal grammar', Language, xxviii 3. Robins, R. H. (1953) 'Formal divisions in Sundanese', Transactions of the Philological Society. Saussure, F. de (1949) Cours de linguistique generale (4th edn), Paris: Payot. Scott, N. C. (1956/70) 'A phonological analysis of the Szechuanese monosyllable', Bulletin of the School of Oriental and African Studies 18 (reprinted in Palmer (1970)). Shiratori, Kurakichi (1942) Onyaku Mobun Gencho Hishi (The Mongolian version of the Secret History), Toy 6 Bunko Publications C 8, Tokyo. Simon, H. F. (1953) 'Two substantival complexes in Standard Chinese,' Bulletin of the School of Oriental and African Studies, XV, 2. ( ' }

Simon, W. (1937) 'Has the Chinese language Parts of Speech?', Transactions of the Philological Society. Simon, W. (1947) A Beginners' Chinese-English Dictionary of the National Language (Gwoyeu), London: Lund Humphries. Skalicka V. (1946) 'Sur la typologie de la langue chinoise parlee', Archiv Orientali XIV, Prague. Spang-Hanssen, H. (1949) 'On the Simplicity of Descriptions', Travaux du cercle linguistique de Copenhague 5,61.70. Street, J. C. (1957) The Language of the Secret History of the Mongols, American Oriental Series 42, New Haven, CN. Tarn, Mo-Shuet (1979) 'A grammatical description of transitivity in Mandarin Chinese, with special reference to correspondences in English based on a study of texts in translation', Ph.D. thesis, University of London. Temir, Ahmet (1948) Mogollarin Gizli Tarihi (Vol. I, Translation), Ankara: Turk Tarih Kurumu Basimevi. Wang, Ching-Ju (1935) 'Chiu Yuan pi-shih i-wen so chien chih Chung-kuo jen-ch'eng tai-ming-tz'u' (Personal pronouns in the Chinese translation of the "Secret History"), Academia Sinica, Bulletin of the Institute of History and Philology, v 4. ( ' )

377

BIBLIOGRAPHY

Wang, Li (1936) Zhongguo Yinyunxue ' (Chinese Phonology), Shanghai: Commercial Press. Wang, Li (1946—7) Zhungguo Ytifa Lilun (Theory of Chinese Grammar), Vols. I-II, Shanghai: Commercial Press Wang, Li (1947) Chung-kuo hsien-tai yti-fa (Modern Chinese grammar) (2 vols), Shanghai: Commercial Press. ( ) Wang, Li (1980) Hanyu Shigao (Outline history of the Chinese language) (3 vols.), (Vol. 1: Phonology), Beijing: Zhonghua Publishers. Wang, Li (1981) Zhongguo Yuyanxue Shi (History of Chinese Linguistics), Taiyuan: Shanxi People's Publishing House. Webster, Jonathan (2005) Continuing Discourse on Language, A Functional Perspective, London: Equinox. Wignell, Peter, Martin, J. R. and Eggins, Suzanne (1989) 'The discourse of geography: ordering and explaining the experiential world', Linguistics and Education 1.4. Wong, Helen (1953—54) 'Outline of the Mandarin phonetic system', Word, ix, 1953, x, 1954. Yang, L. S. (1949) 'The concept of "free" and "bound" in spoken Chinese', Harvard J. Asiatic Studies xii.

378

INDEX

attribution 49,72, 84,92,227,235 attributive 41-2,49, 52,71-4, 84-5, 88-9,95,214-15,217,223,226-8, 230-1,240-2,330 auxiliary 23-5,27-8,41,45,48-50,52, 72,74-5, 81-2, 84-7, 90,95,121-2, 142,146-7,189,202,223,230-4, 237,242,261 auxiliaries 25-7,29,68,122,200

adjective 49,66, 73,92,180,182-4, 196-8,228,235,243,255,257-9, 266,338 adjectival 21-6,227-8,243-4,258 adverb 39,41,45,47-52,64-5,67-73, 75-6,78-81,83,85,88-95,113-5, 135,149,180-1,186,189-90,194, 200-3,216-8,223-5,227-31,233, 235,237,239,242,331,360 adverbial 41,49,51,75-6,80,84-5, 215,225-7,229-30,233-4,239, 241-2,360 affirmative 41,47,70-1,181, 223,227-8 affricate 117,120,122,128,287,290,313 agglutinative 202 allophonic 295,298,320 alphabet(ic)(al) 24,113,118,120,131-2, 162,212,250,301,317 alveolar 128,300-1,318-19 aperture 302,305,307-11, 313,315,319 apical 128,309,311,319 appraisal 115,362 articulation 12,25,119,128-30,132-3, 212-13,292,312-13,319 articulatory 129,290,310,318 aspect 12,41-2,44,46-8,50,62,66-71, 78-9,83-4,92,114,137,149, 177-203,221-6,228,230-1,234-5, 241-2,257,267,338 aspectival 51,67,224-5 aspective 68-9,185-9,192,194,198, 201,255,257

ba 23-4,46-7,66-7,71,79,180,182, 189-90,197,201,221,225,230, 243,261,348 bilingual 20,28,112 bound adverb(s) 41,45,51-2,89,95 character 50-2,87 class(es) 48-9,52,72,214 clause 39,41,43-4,47,49,51,65,68, 71-2,91,137 word 45,50,52,229 Buddhist 7,284,289 Canton 180,191,195,249,251,253-4, 261-3,265-6 Cantonese 47,49,62,70,80,88,95,120, 177,186-7,191,194-6,198-9, 210-11,225,231,249-50,252-68, 305,317 causative 78, 80,255-7,267 Chungshan 258,261-2,264-6 Chao,Y.R. 25,47-9,52,66,72-4,83, 379

INDEX 95,113,119-20,181-97,200,2023,215,224,228,231,236-8,24950,252,276,278,288,348 clause adjectival 215-18,227-8 bound 41,43-4,47,49,51,65,68, 71-2,91,137 conditional 48,51,71-2,79,89,91, 93,95,214,216-17,223-4,235,360 conjunctive 217,235 ergative 64,75-6 final 39,45,47,69,71,74,78,80-1, 83-4,95,224-5,350 free 41,43-4,48,51,65,69,137, 213,216 genitival 72-3,85,218 imperfective 67—9,230 interrogative 70,223 nominal 46,49,66,73 passive 45-6,67,217,223 perfective 45,47,67-9 subordinate 65,69,84,181,185,194, 214-16,228,254 verbal 42,46,48-9,66,73,77 cohesion 359,355-7,360,363 cohesive 354—8 colloquial 16,68,84,210,257,259,332 conditional 41,44,47-8,51,71-3,79-80, 89,91,93, 95,137,181,214-17, 223-5,228,235,237,242-3,360 conjunction 71, 89,353—61 conjunctive 41,45,51, 65,71-2,91-2, 149,214-18,235,237,242-4, 356-7,360-1 connotation 27,89,194,201,251 consonant(s) 25-7,117-22,128-9, 281-2,284-6,300,302,306-7,309, 311-12,314-15,319 consonantal 306,309,313 contextualization 10,12-3,15,20,64

256-68,284,287,291,304-5,313, 317,320 dialectal 135,250,257,315 dictionary 27,52,112-14,116,181,236, 277,280,283,285,287 discourse 210,293,330-1,333-6,338-9, 342,344-55,357-61,363 disjunctive 214-6,241-4 'disyllabic(ally) 52,87,95,180,195, 255-7,276 disyllables 77,276 ellipsis 354-5 endocentric/exocentric 226,233, 241-2 English 24,39,52,61-71,73,75,78,81, 84,87-8,91-3,95,113,116,119, 178,186-8,192-4,200,202-3,218, 220-1,224,231,235,255-6,261-3, 277,283,293,306-7,318-19, 325-32,334-45,347-60,362 entropy 240 ergative 41,45-6,48,64,66-8,75-6, 79,136-8,216-18,220-1,223,226, 230 evolution 186,281,287,291,319,336 exponent(s) 41-2,45-9,51-2,63,65,67, 82,89,91,95,161,220,292 FangYan 351,353,355 Firth, J.R. 10-11,13-14,25,41,62,114, 294-5,298,315,346 Firthian prosodic 292,304 phonology French 7,24,84,86,88,115,336 fricative(s) 25,119-20,128,133,287, 290,313,318-19 Fries, Peter 352,363 German 7,336 Germanic 203 grammar 8-12,16,19,37,42,48,51-2, 60-64,112-14,118-19,129,178,209, 215,218,221,240,275,293,320, 328,330,332-5,337-8,341,343-4, 346-9,352-3,355-61,363 grammatical analysis 9,13,37,42-3,114

dentals 290,309,312 determinate(s) 87,180,195,232-4,237, 242 ' diachronic 14-15,250,284 diacritic(s) 24-6,122,132,295,298 dialect(s) 7,95,116-18,177,180,183,187, 191,193,195,202,210,249-54,

380

INDEX Japanese 7,19,27,116,189,253,287

categories 39-41,51,60,63,67,93, 112,114,178,200,209,212,338, 356 description 52,63,211,236—7 form(s) 11,29,52,64-5,89,114, 177,218,255 function 51,61,162,180 level 39,41-2,51,63,113,220 meaning 39,48,112 metaphor 325-6,333,339-40,342, 344 reflection 211,218,221,240 statement(s) 37,39-41,50,52,64, 211,213-14 structure(s) 39,66,178,218,220, 239-40,357,359-60,363 system(s) 10,28-9,42,48,61,64,93, 112,114,221,353,357,360 translation 28-9,87 form(s) units 237,347,358 Greek 192,275,337-8,339,349

Korean 116,287 Kowloon 249,254,258,262 Kozin,S.A. 6-7,19,121,124

Haenisch, Erich 6-8,17,19,25,27-9, 70-1,74,84,87-9,91-2,121, 123 Hakka 249,253,258,263-6,268 Hasan, R. 353,355,358-61,363 Hunan 250,253 hyponomy 336—7,353 hypotaxis 341 hypotactic 355—6 imperative 14,41,44,47, 51,63,68-71, 77,137-8,191,200,202,223,227-8 imperfective 41,44,46-7,66-9,71,75-6, 79,83,92,135-8,179-80,188-9, 191-2,194-5,199,201-3,221,23, 230,234,241 inchoative 181,183,196 Indo-European 112,117,178,195,201, 203,255,257,349 interrogative 14,41,44,47,51,63,70-1, 80,88-9,91,93-5,137-8,180, 190-1,194,201,223-8 intonation 66-7,183,221,347,350 intransitive 49, 66,73-4, 84,182-4,192, 216,218,227-30,232,242,265

381

labial(s) 94-5,119-23,128,213,284, 290,318-19 labialized 312-13 labio-dental 128,130,290 labiovelar 298,312 language Chinese 7-8,24,28,52,60,117,178, 202,240,252,255,256,275,287, 290,293,320,325,328 Indo-European 178,255,257 literary 7,52,93 modern 202,279 Mongolian 16,25,26,28,121 restricted 20,210-11 scientific 335,345 spoken 11,12,14,118,177,201, 202,331,344,347 standard 251,254,293 written 327 lateral 290,318-19 Latin 43,114,178-9,181,202,337,339, 344,349 latinization/latinized 180,253 lectal 295,298 lexicon/lexical 10,12,24-5,27-9,40-2, 47-8,50-2,60, 63-5,68,78, 80,82, 89-90,92,112-14,116,118,120, 131,159,162,180,191,198,201, 203,211,230,235-6,238,252-1, 256-7,259-60,266,326,332, 336-7,353,360 form(s) 24,42,52,64-5,78,80,256 identification 50-2, 82,230,235, 238 meaning 27,68,112,114,201,203 statement 10,28,40-2,112-14 translation form(s) 28 variation 89,198,256-7 lexicalization/lexicalized 25-7,120 lexicogrammar/lexico-grammatical 41, 47,50-1,67,71,113-14,177,201, 334-7,347,353,358,360-1,363

INDEX lexico-phonological 24 lexicography/lexicographical 24,43, 112-14,236 linguistic analysis 8,13,23,37,39,60,336,347 context 64 feature(s) 14,37,209,268,293 material 42,209,212 scholarship 276,283,292 system 333—5,347 unit(s) 25,38,40-2,48,212,237, 253,276 linguistics 7,9-10,12,25,37,60-1,114, 119,177,210-11,275-6,289,291, 293-4,334,336 literary 7-8,13,16,20,52,72,93,112, 114,186,266,293 literature 8,12,14-16,20,325,358 loanwords 116,276,287 logical-semantic 347,353,355-63

morpheme(s) 180,195,252,255-6,263, 276,279,281,284,286,289,313-15, 320,337 morphology/morphological 29,113—14, 275,339,346,349 narrative 7,14-6,63,65,68,92,183,352, 357-62 nasal(ity) 63,92, 94-5,117,120,122, 128-9,133,226,281,285,290,298, 300-2,304,306-9,318-19 negative(s) 50,69-71,75,79-80,89, 93-5,183-7,190,195,197,199200,202,211,222-6,228,230-1, 240,256,267,329 neutral 25,27,41-2,44,46-7,63,67-8, 70,83,129,178-80,195,200,203, 214,217-18,220-5,227-31,239-40, 309 nominal 40-1,45-52, 62,64, 66,72-3, 80,83,84-5,87,89,92,94-5,115, 137-8,149,215-17,220,225,227, 229-30,232-5,237-9,241-2,256, 338,340,342,344,348,358 nominalization 37,341,342,344 noun auxiliary 50,72,85-7,95, 231-2 free 44,49, 52, 61, 84,86-7,92, 113-14,227,232,235-7 minor group(s) 73, 86—7,147 group 45,486-91,61-2,72-3,84-8, 95,147 numeral 84-6,94 postpositive 75, 87-8,91,234,236 pre-verbal 44,51,65,71,91,94 numeral(s) 41,45,49-50,52,74,84-6, 94,146,233,237,319

Mandarin7,116-18,129,133,177,193-4, 201-2,249,254,256,263,266-7, 290,294-5,298,304,306-7, 313-14,318-20,346 Primer 47-9,72,74,83,120,181, 215,231,236-7,348 Matthiessen, C. 319-20,352,358,363 McDonald, E. 351,353 meaning 9-11,24,27,39,48,65, 67-8, 88,112-14,180-2,184-91,195-6, 201,203,211,213,236,257,260-3, 266,278-81,285,293,331,333, 335,338-9,344,347,349,351-53, 357-60,363 metaphor 210,293,318,325-33,33940,342,344 modal(ity) 47-8,51,68,70-1,83,183-4, 186-7,191,200-1,224-5,293, 328-9,337 Mongolian 5-8,14-17,19-23,25-29,65, 70-1,84,87,89,91-2,116-19,121, 123-4,159,287 monosyllabic 25,51-2,68,87,113, 180-1,195,202,255-7,260,276 mood 41,46,48, 50,66, 69-71, 83,114, 137,149,180,186,200,222,224-6, 241-2,328,354

obstruence 306-7,309 obstruent 318-19 palatal 119-22,132,212-13,284,290, 313-14,318-19 palatoalveolar 298,309,311-13,319 Palladius (Kafarov) 6-7,18-19 Palmer, ER. 292-3 parataxis/paratactic 341,355—6

382

INDEX predicate 68,80,180-3,186-7,189, 191-5,197,200 predicative 66,73,181-4,187-8,191 preposition(s) 68,75,87,261,341 prepositive 41,45-6,48-50, 67-71, 73-78, 82,87-8,90-1,93,137, 139-40,221,234-5,237,242 preverbal 41,45,47-8, 51,71-3, 88-91, 93-4,149,216,235 pro-verb 41-2,44-6,49-50,52,66, 73-4,82-4,93-5,138,218,223, 225,230 pronominal 14,62-4,72,91,93,95 prosodic 25,92,121,128-30,133,284, 288-90,292-5,298,300-2,304-11, 313-14,317-20 prosody 121,129-32,288,300,314-15 punctuation 12,21-2,38,40,65,92,161

passive 41-2,45-6,48, 66-7,72-5,77, 137-8,188,216-18,220-1,223, 227,235,331,349 Peking 6,16,120,128-9,177,180,193, 210,263,294,298,308,362 Pekingese 8,41,43,46-9, 51,61-7, 69-95,114,117-23,128-9,132-5, 162,178,180,184,186-7,189, 191-4,199,202-3,210-11,225, 254-5,257-8,261,263-7,285,289, 293,295,298,305,313,318-20, Pelliot, P. 6-7,19,27,116,121,124,284 perfective 41,44-8,51,67-9,78-9,83, 92-3,136-8,179-89,191-2,195, 197,200-1,203,221-5,230,257 phoneme(s) 129,298,315,319 phonemic 129,295,300,317,320 phonetic auxiliary(-ies) 23—7,122 element 279-82 quality 295,300 phonetics 124,129,275,293 phonology 10,12,25-6,92,116,118, 121,124,128-9,212,249,275,284, 288-94,313,315,317,319,346 phonological analysis 28,120,211,250,282-3, 295,302 prominence 349 rank scale 317,320 structure 26,212,240,320 system 6,25,27-8,95,116-18,128, 292,298,300,320 theory 275,289,291-2,318 plosive(s) 25,117,119,122,128,285, 290,313 plural 28,50-1,63,73,84-8,92,149 plurality 63,92,235 polarity 222-8,230,241-2,329 polysyllabic 51-2,72, 84,112-13,202 positive 75, 80,87,94,186,194-7,199, 202,222-3,228,230-1,240,339, 362 possessive 22,51,92,330 postpositive 29,41,45,48-52,67-9,71, 74-84,87-8,91,93,114,137, 143-5,148,223,228,230-1,234-7, 242

reduplication 199,201 referential 354—5 register(s) 293,327,333,335,339,3550, 357-9,361 resonance 300-2,304,306-8,310-11, 319 retroflex 119-21,132-3,290,298,309, 312-13,319 rheme 351,353,358 rhyme(s) 285-7,291-2,312 rhyming 258,285,298 Russian 7,19,178-9,185,188,192,203, 358 Sanskrit 202-3 de Saussure, F. 114,178 scientific -9-10,327,334-8,342-5 script Chinese 23-5,27-8,116,120-1 Sino-Mongolian 6-7,17,21-5, 27-8,116-21,123-4,287 Uighur-Mongolian 7 semantic 180-1,202-3,263-4,277-81, 326,328,332-3,336,338-9,341-2, 344,347,349,353,357-60 semi-vowel 128—9 sentence-final 39,44-5,47-8,65,68, 71-3,83,214,222,224 sentence-initial 51,71-2,214

383

INDEX Slavonic 177-8,189-91,195,201 speaker 15, 62-3, 82-3,120,128,178-9, 190,196,201,224,236,257-8,260, 276,294-5,313,318,326,332, 348-51,357 speech 14-15,43-4,48,51-2,70,76,92, 112-13,115,128,202,252-3, 258-9,262,291,294-5,298,301, 320,329-30,347 suffix(es) 47,91,181-2,187,189,201-2, 224,239 syllabary 22,118,121,133-4,162,295, 298,305,313,318-20 syllabic 24,63,116,119-21,128-34,212, 224-5,249,285,300,318-19 syUable(s) 14,23-7,52,63,70,72,84, 89-90,92,95,116,118-23,128-35, 149,162,180,202,211-12,276, 281,284-7,288-9,291-5,298, 300-2,304-11,313-15,317,319, 332 synchronic(ally) 8-9,11,63,114,177, 188-9,191,200,211,249-50, 253 synonym 276—7 syntactic(ally) 29,60,181,236,249,257, 259-60,266-7,343 syntagm(atic) 11,41,43,49,51-2,66,73, 80-1,84,88,239 syntax 191,252-4,259,275,334,339, 346 systemic 40,114,118-19,212,293-4, 315,318-20,344,356,358,363 taxonomy/taxonomic 336—8,340,344 tense 68,84,114,178-87,189-95, 198-203,225 theme 326,330,343,350-3,355,358 thesaurus 276-7,280 tone(s) 82,94-5,118-20,128,131,180, 191,198,212,250,252,263,281, 284-6,291-2,295,306,310, 318-19

tonal 285-6,292,295,306,315,318, 320 transitive 66,73-4, 84,181-4,216,218, 228-30,261,265 transitivity 229,329,341,354, 357 Uighur-Mongolian 7 Ulan-Bator 7 velar 25,117,120,128,132-3,300,313, 318-19 verb(s) Chinese 68,114,178,186,200-1 compound 47,68,81,178,192, 195-7,199,203,238 disyllabic 180,255-7 free 25,42,45-6,48-52,66-9,70-1, 73-5,77-81,83-4,87-9,95, 113-14,149,223,227-31,236 intransitive 73,182-4,192 passive 74-5,77,235 phrasal 332 postpositive 45,48-50,67,69,71, 74-5,77-84,87,93,114,223,228, 230-1,235-6 prepositive 46,48-9,67-71,73-8,82, 88,91,93,137,221,234 transitive 73,181-2,184,261 Vietnamese 116,295 voice 41-2,45-8,50,66-8,72-3,83,114, 137,201,216-8,220-1,223,226-8, 230,232,241-2,318-9,326,331, 348-9 voiced 25,117,130,133,281-2,290,319 voiceless 25,119,130,251,281-2,290 vowel(s) 25,117,119-22,128-9,133, 212-3,281-4,287,291-2,294-5, 298,300-2,305-12,319-20 word-class(es) 39,41-2,45,48-9,72-3, 113 word-order 29

384