A Phonological History of Chinese 1107135842, 9781107135840

The phonological history of Chinese can be traced back to two main traditions: one starting with the Qieyun of 601, and

732 48 39MB

English Pages xxxvi+404 [442] Year 2020

Report DMCA / Copyright

DOWNLOAD FILE

Polecaj historie

A Phonological History of Chinese
 1107135842, 9781107135840

Table of contents :
A Phonological History of Chinese
Contents
List of Figures
List of Tables
Foreword by William S-Y. Wang 王士元
Preface
Major Periods of Chinese History
Locations of Ancient Capitals of China
Part I: The Keys to Traditional Phonology
1 An Introduction to Chinese Historical Phonology
1.1 Defining Chinese
1.2 The Nature of Phonological History
1.2.1 Chinese Phonology through Written Record
1.2.2 Phonological Categories in Logographic Systems
1.2.3 Terminology of Traditional Chinese Phonology
1.2.4 Syllable Structure
1.2.5 Nonsyllabic Phonological Characteristics
1.3 Yùnshū 韻書 ‘Rhyme Dictionaries’
1.3.1 The Qièyùn 切韻 (601)
1.3.2 Fǎnqiè 反切
1.3.3 Zhíyīn 直音 Notation
1.3.4 Shēngdiào 聲調 ‘Tones’
1.3.5 Yùnxì 韻系 ‘Rhyme Series’
1.4 Yùntú 韻圖 ‘Rhyme Tables (Rhyme Charts)’
1.4.1 The Thirty-Six Initials
1.4.2 The Qīngzhuó 清濁 System
1.4.3 Yīndiào 陰調 and Yángdiào 陽調
1.4.4 Děng 等 ‘Division’ and ‘Rank’
1.4.5 Kāihé 開合
1.4.6 Chóngniǔ 重紐 and Chóngyùn 重韻
1.4.7 Sìhū 四呼
1.4.8 Wàizhuǎn 外轉 and Nèizhuǎn 內轉
1.4.9 Shè 攝 ‘Rhyme Groups’
1.4.10 Yīnshēng Yùn 陰聲韻, Yángshēng Yùn 陽聲韻, and Rùshēng Yùn 入聲韻
1.5 Nonanalytical Source Materials
1.5.1 Poem Rhyming
1.5.2 Foreign Transcriptions and Sino-Xenic Pronunciation
1.5.3 Xiéshēng Characters 諧聲字
1.6 The Availability and Reliability of Source Materials
1.7 Periodization
1.8 Reconstruction
1.9 Transcriptions
Part II: Old Chinese
2 Old Chinese
2.1 Introduction
2.1.1 The Tibeto-Burman Connection
2.1.2 Methodology and Working Principles
2.2 Source Materials
2.2.1 Shījīng 詩經 Rhyming
2.2.2 Xiéshēng Characters 諧聲字
2.2.3 Partially Reduplicated Words
2.2.4 The Qièyùn 切韻 System
2.3 Reconstruction
2.3.1 The Reconstruction of Syllable Structures
2.3.2 The Reconstruction of Onsets
2.3.2.1 Simple Initials
2.3.2.2 Cluster Initials
2.3.2.2.1 Initial Clusters with l-
2.3.2.2.2 Initial Clusters with s-
2.3.2.2.3 hN- Clusters or Voiceless Nasals
2.3.2.3 Post-Stem Elements
2.3.2.3.1 The Post-Stem -r- and Middle Chinese Rank-II Syllables
2.3.2.3.2 The Post-Stem -r- and Middle Chinese Rank-IIIb Syllables
2.3.2.3.3 Structure of Initial Consonant Clusters
2.3.3 The Reconstruction of Rhymes
2.3.3.1 The Codas of the Rhyme Groups
2.3.3.2 The Relationships between Rhyme Groups
2.3.3.3 The Absence of Medials
2.3.3.3.1 The Deletion of Middle Chinese -w- Medial
2.3.3.3.2 The Short Vowels and Middle Chinese Rank-III Syllables
2.3.3.4 Main Vowels
2.3.3.4.1 The Number of Main Vowels
2.3.3.4.2 Four-Vowel System
2.3.3.4.3 Six-Vowel System
2.3.3.5 The Codas
2.3.3.5.1 The Post-Coda Segments and Tonal Developments
2.3.3.5.2 Segmentals to Suprasegmentals
2.3.3.5.3 The -ʔ Becomes Middle Chinese Shǎng Tone
2.3.3.5.4 The -s and the Middle Chinese Qù Tone
2.3.3.5.5 The -r Coda
2.3.4 The Reconstruction of Syllable Structures
2.3.5 Recent Advances
2.3.5.1 Syllable Structures
2.3.5.2 Pharyngeals
2.3.5.3 The Uvular Series
2.4 Examples of Old Chinese
Part III: Middle Chinese
3 Middle Chinese: The Qièyùn 切韻
3.1 Special Notes on the Qièyùn 切韻 and Fǎnqiè 反切
3.1.1 The Fǎnqiè of Rank-III
3.1.2 The Information Contained within the Qièyùn
3.1.3 Some Main Issues of the Qièyùn
3.1.4 A Single System or a Composite System
3.1.5 The Qièyùn Rhymes and the Nature of the Yùn 韻
3.1.6 The Revisions of the Qièyùn and the Splitting of Rhymes
3.2 The Nature of the Rhymes
3.2.1 New Approaches
3.2.2 A Possible Solution
3.3 Complementary and Near-Complementary Relationships
3.4 Syllable Structure
3.4.1 Suprasegmental Tones
3.4.2 Segmentals
3.4.3 Initials (I)
3.4.4 Medials (M)
3.4.5 Main Vowels (V)
3.4.6 Codas (E)
3.5 Phonological Characteristics of Middle Chinese
3.5.1 The Tripartite Distinction of Stop and Affricate Initials
3.5.2 The Labials
3.5.3 The Retroflex Stops
3.5.4 The Four Ranks (Divisions)
3.5.4.1 The Status of Rank-III Syllables
3.5.5 The Kāihé 開合 Contrast
3.5.6 The Contrast of Chóngniǔ 重紐 Syllables
3.5.7 The Contrast of Chóngyùn 重韻 Rhymes
3.5.8 Final Types
3.5.8.1 Rhymes jì 祭, tài 泰, guài 夬, and fèi 廢
3.6 Modern Evidence of the Phonological Contrasts
3.6.1 The Three-Way Contrast of Phonation Types
3.6.2 The Coda System
3.6.3 The Chóngniǔ 重紐 Contrast
3.6.4 The Chóngyùn 重韻 Contrast
3.6.5 The Contrast of Rank-III and Rank-IV
3.7 Examples of Middle Chinese Phonology
4 Middle Chinese: The Tang and Song Dynasties
4.1 The Tang Dynasty (618–907)
4.1.1 Xuánzàng’s 玄奘 Inscriptions
4.1.2 Simplification and Tóngyòng 同用 Marking
4.1.3 The New Standard
4.1.4 Devoicing of Voiced Stops and Affricates
4.1.5 Labiodentalization
4.1.6 Later Middle Chinese
4.2 The Five Dynasties and Ten Kingdoms (907–960)
4.2.1 Ěryǎ Yīntú 爾雅音圖 (tenth century)
4.2.2 Northwest Dialects
4.3 Northern Song (960–1127)
4.3.1 Guǎngyùn 廣韻 (1008)
4.3.2 Jíyùn 集韻 (1039) and Lǐbù Yùnlüè 禮部韻略 (1037)
4.3.3 Shēngyīn Chànghè Tú 聲音唱和圖
4.3.3.1 Initials
4.3.3.1.1 Devoicing of Voiced Obstruents
4.3.3.1.2 Labiodentals
4.3.3.1.3 The Status of Middle Chinese zhī 知and zhào 照 Initials
4.3.3.1.4 The Loss of the Nasal Quality of the Middle Chinese rì 日 Initial
4.3.3.2 Finals
4.3.3.2.1 Loss of Stop Codas
4.3.3.2.2 Simplification of the Vowel System: The Merger of the Middle Chinese zēng 曾 and gěng 梗 Rhyme Groups
4.3.3.2.3 Centralized Vowels, i > ɨ
4.3.3.3 Tones
4.3.4 Summary
4.4 Southern Song (1127–1297)
4.4.1 Hangzhou Dialect
4.4.2 The Sìshēng Děngzǐ 四聲等子
4.4.3 The Qièyùn Zhǐzhǎngtú 切韻指掌圖
4.4.4 The Qièyùn Zhǐnán 切韻指南
4.5 Examples of Tang Poetry
Part IV: The Beginnings of Mandarin
5 The Chinese of the Liao, Jin, and Xixia Dynasties
5.1 The Liao Dynasty (916–1125)
5.1.1 The Khitan Materials
5.1.2 Diphthongization of Middle Chinese Syllables with Coda -k
5.1.3 Devoicing of Middle Chinese Voiced Obstruents
5.1.4 Labiodentalization
5.1.5 The Merger of the Middle Chinese zhī 知 and zhào 照 Initial Series
5.1.6 The Loss of the Middle Chinese Stop Codas -p, -t, -k
5.1.7 The Status of the Middle Chinese Velar Nasal Initial ŋ-
5.1.8 The Middle Chinese Bilabial Nasal Coda -m
5.1.9 Middle Chinese má-III (麻) Syllables
5.1.10 The Foundations of Mandarin
5.1.11 Examples
5.2 The Jin Dynasty (1115–1234)
5.2.1 The Gǎibìng Wǔyīn Jíyùn 改併五音集韻 1212)
5.2.1.1 The Merger of Rhymes
5.2.1.2 Chóngniǔ 重紐 Syllables
5.2.2 The Jurchen Materials
5.2.2.1 Diphthongization of Middle Chinese Syllables with Coda -k
5.2.2.2 Devoicing of Middle Chinese Voiced Obstruents
5.2.2.3 Labiodentalization of Middle Chinese Bilabial Initials
5.2.2.4 Loss of Middle Chinese Stop Codas -p, -t, -k
5.2.2.5 Retention of the Middle Chinese Velar Nasal Initial ŋ-
5.2.2.6 Loss of the Middle Chinese Bilabial Nasal Coda -m
5.3 The Xixia Dynasty (1038–1227)
5.3.1 The Tangut Script
5.3.1.1 Basic Phonological Characteristics
5.3.1.2 Devoicing of Middle Chinese Voiced Obstruents
5.3.1.3 Labiodentalization
5.3.1.4 The Merger of the Middle Chinese zhī 知 and zhào 照 Initials Series
5.3.1.5 The Loss of Stop Codas
5.3.1.6 The Loss of Nasal Codas
5.3.1.7 Centralization of High Front Vowels
5.3.2 Phonological Characteristics and Differences of Northern Dialects
5.3.2.1 A Comparison of Two Northern Dialects
Part V: A New Standard
6 Old Mandarin: The Ménggǔ Zìyùn 蒙古字韻
6.1 The ḥP’ags-pa Script
6.1.1 Adaptations between Tibetan and ḥP’ags-pa Script
6.1.2 The Syllable Block
6.1.3 The Problems of the Script
6.1.4 Phonological Contrasts That Cannot Be Reflected
6.2 The Phonological System of the Ménggǔ Zìyùn 蒙古字韻
6.2.1 The Fifteen Rhyme Groups
6.2.2 The Phonetic Values of the ḥP’ags-pa Letters
6.2.3 The Spelling of Finals
6.2.4 The Subsystems of Finals
6.2.5 Tonal Marks
6.3 Sound Changes in the Initial System
6.3.1 The Loss of Voiced Obstruents
6.3.2 Labiodentalization of Middle Chinese Bilabial Initials
6.3.3 The Loss of the Middle Chinese yí 疑 ŋ- Initial
6.4 Sound Changes in the Final System
6.4.1 The Medial
6.4.2 The Reduction of the Vowel System
6.4.3 Chóngniǔ 重紐 Reflexes in the Ménggǔ Zìyùn
6.4.4 The Loss of Final Consonant Codas
6.4.5 Diphthongization of Middle Chinese Syllables with Coda -k
6.4.6 The Loss of the Parallel Relationship between yáng and rù Syllables
6.4.7 The Contrast of jan/jɛn and jaw/jɛw
6.4.8 The High Central (Apical) Vowel
6.4.9 Vowel Raising
7 Old Mandarin: The Zhōngyuán Yīnyùn 中原音韻
7.1 The Base Dialect
7.1.1 The Influence of Southern Mandarin
7.2 The Zhèngyǔ Zuòcí Qǐlì 正語作詞起例
7.2.1 Dialect Features
7.3 The Phonetic Reconstruction
7.3.1 Initials (21)
7.3.1.1 The wēi 微 and rì 日 Initials
7.3.1.2 The Value of the Retroflex Series
7.3.2 Finals (47)
7.4 Innovative Characteristics of the Zhōngyuán Yīnyùn 中原音韻
7.4.1 Devoicing of the Middle Chinese Voiced Obstruents in the Initial Position
7.4.2 The Emergence of the Labiodental Fricative
7.4.3 The Development of the High Central (Apical) Vowels
7.4.4 The New Four-Tone System
7.4.4.1 The Loss of the Stop Coda and the Disappearance of the Middle Chinese rù Tone
7.4.5 The Loss of the Middle Chinese yí 疑 Velar Nasal Initial
7.4.6 The Disappearance of the Four-Rank System of Finals
7.4.6.1 The Complete Loss of the Contrast between Middle Chinese Rank-III and -IV Syllables
7.4.6.2 The Residual Chóngniǔ Contrast
7.4.6.3 Rank-II and Rank-III Contrast
7.4.6.4 The Emergence of the chē-zhē 車遮 Rhyme
7.4.7 The Preservation of the -m Coda
7.4.8 The jiān-tuán 尖團 Distinction
7.5 Examples of Yuan Pronunciation
7.6 Summary
8 Old Mandarin: The Persian Transcriptions
8.1 The Ancient Persian Script
8.2 History of China
8.2.1 Middle Chinese Voiced Stops and Affricates
8.2.2 The Loss of Velar Nasal Initial yí 疑 ŋ-
8.2.3 Middle Chinese rì 日 Initial
8.2.4 Middle Chinese wēi 微 Initial
8.2.5 The Coda of rù 入 Syllables
8.2.6 High Centralized (Apical) Vowels
8.2.7 Main Vowel -e
8.2.8 Summary of Features
8.3 The Mài Jué 脈訣
8.3.1 The Stop Codas of the rù 入 Syllables
8.3.2 The Inconsistency of the Stop Coda in the Transcription
8.3.3 The Dialectal Features Reflected in the Transcription
8.3.4 The Nebulous Nature of Early Guānhuà 官話
Part VI: Toward Modern Mandarin
9 The Mandarin of the Ming Dynasty
9.1 North Versus South
9.1.1 Guānhuà 官話
9.2 The Hóngwǔ Zhèngyùn 洪武正韻 (1375)
9.2.1 Initials
9.2.2 Finals
9.2.3 Tones
9.2.4 A Coherent System
9.2.5 Zhèngyīn 正音 and Súyīn 俗音
9.2.6 Revision of the Hóngwǔ Zhèngyùn
9.3 The Yùnluè Yìtōng 韻略易通 (1442)
9.3.1 Initials and the Zǎo Méi 早梅 ‘Early Plum Blossom’ Poem
9.3.2 Finals and the Twenty Rhyme Groups
9.3.3 Běnwù’s 本悟 Revision
9.3.4 Tones
9.4 The Sìshēng Tōngjiě 四聲通解 (1517), Fānyì Lǎo Qǐdà 翻譯老乞大, and Fānyì Piáo Tōngshì 翻譯朴通事
9.4.1 Initials
9.4.1.1 The Description of Retroflex Initials
9.4.1.2 Exceptional Changes
9.4.2 Finals
9.4.2.1 The Loss of -p, -t, -k
9.4.2.2 The Loss of -m, -m > -n
9.4.2.3 Middle Chinese tōng 通 Rhyme Group Syllables
9.4.2.4 The -w- of Middle Chinese kāikǒu 開口 Syllables
9.4.2.5 The Change of -jaj and -waw
9.4.2.6 The Change of -jan/m > -jɛn/m
9.4.2.7 The Change of -jɛw > -jaw
9.4.2.8 The Change of -iw > -əw
9.4.3 Tone Sandhi
9.5 The Xīrú Ěrmù Zī 西儒耳目資 (1626)
9.5.1 Initials
9.5.2 Finals
9.5.3 Tones
9.6 The Děngyùn Tújīng 等韻圖經 (1606)
9.6.1 Initials
9.6.1.1 Palatalized Initials
9.6.1.2 Initial Conditioned Change: The Vowel Centralization and the Loss of the Medial -j- after Retroflex Initials
9.6.2 Finals
9.6.3 Tones
9.6.4 Innovative Features
9.6.4.1 The Merger of Middle Chinese Division-II and Division-III Syllables with zhī 知 and zhào 照 Initials
9.6.4.2 The Loss of the Middle Chinese wēi 微 ɱ- Initial
9.6.4.3 The Loss of the Middle Chinese yí 疑 ŋ- Initial
9.6.4.4 The Rounded Central Vowel -ʉ
9.6.4.5 The Emergence of ɚ
10 The Mandarin of the Qing Dynasty and the Modern Era
10.1 The Qing Dynasty (1644–1912)
10.1.1 The Yùdìng Pèiwén Yùnfǔ 御定佩文韻府 (1711)
10.1.2 The Yīnyùn Chǎnwēi 音韻闡微 (1728)
10.1.3 The Wade Textbook (1867)
10.1.3.1 Initials
10.1.3.2 Finals
10.1.3.2.1 Further Notes on Finals
10.1.3.3 Tones
10.1.3.3.1 Tone Sandhi
10.1.3.4 Colloquial Speech: Neutral Tone and Rhotacized Finals
10.1.3.5 Alternative Readings
10.2 Post-Imperial China (1912–present)
10.2.1 The Old National Standard
10.2.2 The New National Standard
10.2.3 Modern Standard Phonology
10.2.4 Modern Dialects (Fāngyán 方言)
10.2.4.1 Dialect Classification
10.2.4.2 The Northern Dialects: Mandarin and Jin
10.2.4.3 The Central Dialects: Wu, Hui, Gan, and Xiang
10.2.4.4 The Southern Dialects: Min, Kejia, Yue, and Ping
10.2.4.5 The Nature of Dialect Formation
References
Index

Citation preview

A Phonological History of Chinese

The phonological history of Chinese can be traced back to two main traditions: one starting with the Qieyun of 601, and the other starting with the Zhongyuan Yinyun of 1324. The former marks the beginning of Middle Chinese, and the latter marks the beginning of Old Mandarin. Both of these systems, as well as reconstructed Old Chinese, should be understood as ideal phonological standards and composite in nature. Until modern times, phonological standards were never based strictly on the phonology of a single dialect. This book provides the first study written in English of the phonological history of Chinese. It provides information about the standard phonological systems for each of the language’s major historical periods, drawing on a range of historical materials such as dictionaries, rhyming tables, and poetry, and is a reference book for understanding the key developments in the Chinese sound system. professor zhongwei shen is a Full Professor of Chinese Linguistics in the Department of Languages, Literatures, and Cultures at University of Massachusetts Amherst. He has served as a co-editor of the Journal of Chinese Linguistics, and co-editor-in-chief of the Bulletin of Chinese Linguistics.

A Phonological History of Chinese Zhongwei Shen University of Massachusetts Amherst

University Printing House, Cambridge CB2 8BS, United Kingdom One Liberty Plaza, 20th Floor, New York, NY 10006, USA 477 Williamstown Road, Port Melbourne, VIC 3207, Australia 314–321, 3rd Floor, Plot 3, Splendor Forum, Jasola District Centre, New Delhi – 110025, India 79 Anson Road, #06–04/06, Singapore 079906 Cambridge University Press is part of the University of Cambridge. It furthers the University’s mission by disseminating knowledge in the pursuit of education, learning, and research at the highest international levels of excellence. www.cambridge.org Information on this title: www.cambridge.org/9781107135840 DOI: 10.1017/9781316476925 © Zhongwei Shen 2020 This publication is in copyright. Subject to statutory exception and to the provisions of relevant collective licensing agreements, no reproduction of any part may take place without the written permission of Cambridge University Press. First published 2020 Printed in the United Kingdom by TJ International Ltd, Padstow Cornwall A catalogue record for this publication is available from the British Library. Library of Congress Cataloging-in-Publication Data Names: Shen, Zhongwei, author. Title: A phonological history of Chinese / Zhongwei Shen. Description: Cambridge ; New York, NY : Cambridge University Press, 2020. | Includes bibliographical references and index. Identifiers: LCCN 2019036955 (print) | LCCN 2019036956 (ebook) | ISBN 9781107135840 (hardback) | ISBN 9781316501658 (paperback) | ISBN 9781316476925 (epub) Subjects: LCSH: Chinese language–Phonology, Historical. Classification: LCC PL1201 .S456 2020 (print) | LCC PL1201 (ebook) | DDC 495.1/15–dc23 LC record available at https://lccn.loc.gov/2019036955 LC ebook record available at https://lccn.loc.gov/2019036956 ISBN 978-1-107-13584-0 Hardback Cambridge University Press has no responsibility for the persistence or accuracy of URLs for external or third-party internet websites referred to in this publication and does not guarantee that any content on such websites is, or will remain, accurate or appropriate.

Contents

List of Figures List of Tables Foreword by William S-Y. Wang 王士元 Preface Major Periods of Chinese History Locations of Ancient Capitals of China

The Keys to Traditional Phonology

1

An Introduction to Chinese Historical Phonology

3

Part I 1

page xi xiii xxx xxxiii xxxv xxxvi

1.1 Defining Chinese 1.2 The Nature of Phonological History 1.2.1 Chinese Phonology through Written Record 1.2.2 Phonological Categories in Logographic Systems 1.2.3 Terminology of Traditional Chinese Phonology 1.2.4 Syllable Structure 1.2.5 Nonsyllabic Phonological Characteristics 1.3 Yùnshū 韻書 ‘Rhyme Dictionaries’ 1.3.1 The Qièyùn 切韻 (601) 1.3.2 Fǎnqiè 反切 1.3.3 Zhíyīn 直音 Notation 1.3.4 Shēngdiào 聲調 ‘Tones’ 1.3.5 Yùnxì 韻系 ‘Rhyme Series’ 1.4 Yùntú 韻圖 ‘Rhyme Tables (Rhyme Charts)’ 1.4.1 The Thirty-Six Initials 1.4.2 The Qīngzhuó 清濁 System 1.4.3 Yīndiào 陰調 and Yángdiào 陽調 1.4.4 Děng 等 ‘Division’ and ‘Rank’ 1.4.5 Kāihé 開合 1.4.6 Chóngniǔ 重紐 and Chóngyùn 重韻 1.4.7 Sìhū 四呼 1.4.8 Wàizhuǎn 外轉 and Nèizhuǎn 內轉 1.4.9 Shè 攝 ‘Rhyme Groups’ 1.4.10 Yīnshēng Yùn 陰聲韻, Yángshēng Yùn 陽聲韻, and Rùshēng Yùn 入聲韻

4 6 6 7 8 9 10 11 12 16 19 20 21 22 25 27 29 30 35 36 38 39 46 48

v

vi

Contents 1.5 Nonanalytical Source Materials 1.5.1 Poem Rhyming 1.5.2 Foreign Transcriptions and Sino-Xenic Pronunciation 1.5.3 Xiéshēng Characters 諧聲字 1.6 The Availability and Reliability of Source Materials

48 48 49 50 51

1.7 Periodization 1.8 Reconstruction 1.9 Transcriptions

54 54 56

Part II 2

57

Old Chinese

59

2.1 Introduction 2.1.1 The Tibeto-Burman Connection 2.1.2 Methodology and Working Principles 2.2 Source Materials 2.2.1 Shījīng 詩經 Rhyming 2.2.2 Xiéshēng Characters 諧聲字 2.2.3 Partially Reduplicated Words 2.2.4 The Qièyùn 切韻 System 2.3 Reconstruction 2.3.1 The Reconstruction of Syllable Structures 2.3.2 The Reconstruction of Onsets 2.3.3 The Reconstruction of Rhymes 2.3.4 The Reconstruction of Syllable Structures 2.3.5 Recent Advances 2.4 Examples of Old Chinese

59 61 61 62 63 63 65 65 66 67 69 78 97 99 102

Part III 3

Old Chinese

Middle Chinese

Middle Chinese: The Qièyùn 切韻

3.1 Special Notes on the Qièyùn 切韻 and Fǎnqiè 反切 3.1.1 The Fǎnqiè of Rank-III 3.1.2 The Information Contained within the Qièyùn 3.1.3 Some Main Issues of the Qièyùn 3.1.4 A Single System or a Composite System 3.1.5 The Qièyùn Rhymes and the Nature of the Yùn 韻 3.1.6 The Revisions of the Qièyùn and the Splitting of Rhymes 3.2 The Nature of the Rhymes 3.2.1 New Approaches 3.2.2 A Possible Solution 3.3 Complementary and Near-Complementary Relationships 3.4 Syllable Structure 3.4.1 Suprasegmental Tones 3.4.2 Segmentals 3.4.3 Initials (I) 3.4.4 Medials (M) 3.4.5 Main Vowels (V) 3.4.6 Codas (E)

105 107 108 108 110 112 112 116 119 120 122 123 126 128 129 130 131 132 133 134

Contents 3.5 Phonological Characteristics of Middle Chinese 3.5.1 The Tripartite Distinction of Stop and Affricate Initials 3.5.2 The Labials 3.5.3 The Retroflex Stops 3.5.4 The Four Ranks (Divisions) 3.5.5 The Kāihé 開合 Contrast 3.5.6 The Contrast of Chóngniǔ 重紐 Syllables 3.5.7 The Contrast of Chóngyùn 重韻 Rhymes 3.5.8 Final Types 3.6 Modern Evidence of the Phonological Contrasts 3.6.1 The Three-Way Contrast of Phonation Types 3.6.2 The Coda System 3.6.3 The Chóngniǔ 重紐 Contrast 3.6.4 The Chóngyùn 重韻 Contrast 3.6.5 The Contrast of Rank-III and Rank-IV 3.7 Examples of Middle Chinese Phonology

4

Middle Chinese: The Tang and Song Dynasties 4.1 The Tang Dynasty (618–907) 4.1.1 Xuánzàng’s 玄奘 Inscriptions 4.1.2 Simplification and Tóngyòng 同用 Marking 4.1.3 The New Standard 4.1.4 Devoicing of Voiced Stops and Affricates 4.1.5 Labiodentalization 4.1.6 Later Middle Chinese 4.2 The Five Dynasties and Ten Kingdoms (907–960) 4.2.1 Ěryǎ Yīntú 爾雅音圖 (tenth century) 4.2.2 Northwest Dialects 4.3 Northern Song (960–1127) 4.3.1 Guǎngyùn 廣韻 (1008) 4.3.2 Jíyùn 集韻 (1039) and Lǐbù Yùnlüè 禮部韻略 (1037) 4.3.3 Shēngyīn Chànghè Tú 聲音唱和圖 4.3.4 Summary 4.4 Southern Song (1127–1297) 4.4.1 Hangzhou Dialect 4.4.2 The Sìshēng Děngzǐ 四聲等子 4.4.3 The Qièyùn Zhǐzhǎngtú 切韻指掌圖 4.4.4 The Qièyùn Zhǐnán 切韻指南 4.5 Examples of Tang Poetry

Part IV 5

The Beginnings of Mandarin

The Chinese of the Liao, Jin, and Xixia Dynasties 5.1 The Liao Dynasty (916–1125) 5.1.1 The Khitan Materials 5.1.2 Diphthongization of Middle Chinese Syllables with Coda -k 5.1.3 Devoicing of Middle Chinese Voiced Obstruents 5.1.4 Labiodentalization 5.1.5 The Merger of the Middle Chinese zhī 知 and zhào 照 Initial Series 5.1.6 The Loss of the Middle Chinese Stop Codas -p, -t, -k

vii 134 134 135 135 137 141 141 143 143 146 146 147 148 148 150 150

152 152 152 154 155 159 160 162 163 163 165 166 166 168 169 179 180 181 183 187 189 192

195 197 197 197 200 201 202 203 204

viii

Contents 5.1.7 The Status of the Middle Chinese Velar Nasal Initial ŋ5.1.8 The Middle Chinese Bilabial Nasal Coda -m 5.1.9 Middle Chinese má-III (麻) Syllables 5.1.10 The Foundations of Mandarin 5.1.11 Examples 5.2 The Jin Dynasty (1115–1234) 5.2.1 The Gǎibìng Wǔyīn Jíyùn 改併五音集韻 (1212) 5.2.2 The Jurchen Materials 5.3 The Xixia Dynasty (1038–1227) 5.3.1 The Tangut Script 5.3.2 Phonological Characteristics and Differences of Northern Dialects

Part V 6

A New Standard

Old Mandarin: The Ménggǔ Zìyùn 蒙古字韻 6.1 The hP’ags-pa Script 6.1.1_ Adaptations between Tibetan and hP’ags-pa Script _ 6.1.2 The Syllable Block 6.1.3 The Problems of the Script 6.1.4 Phonological Contrasts That Cannot Be Reflected 6.2 The Phonological System of the Ménggǔ Zìyùn 蒙古字韻 6.2.1 The Fifteen Rhyme Groups 6.2.2 The Phonetic Values of the hP’ags-pa Letters _ 6.2.3 The Spelling of Finals 6.2.4 The Subsystems of Finals 6.2.5 Tonal Marks 6.3 Sound Changes in the Initial System 6.3.1 The Loss of Voiced Obstruents 6.3.2 Labiodentalization of Middle Chinese Bilabial Initials 6.3.3 The Loss of the Middle Chinese yí 疑 ŋ- Initial 6.4 Sound Changes in the Final System 6.4.1 The Medial 6.4.2 The Reduction of the Vowel System 6.4.3 Chóngniǔ 重紐 Reflexes in the Ménggǔ Zìyùn 6.4.4 The Loss of Final Consonant Codas 6.4.5 Diphthongization of Middle Chinese Syllables with Coda -k 6.4.6 The Loss of the Parallel Relationship between yáng and rù Syllables 6.4.7 The Contrast of jan/jɛn and jaw/jɛw 6.4.8 The High Central (Apical) Vowel 6.4.9 Vowel Raising

7

Old Mandarin: The Zhōngyuán Yīnyùn 中原音韻 7.1 The Base Dialect 7.1.1 The Influence of Southern Mandarin 7.2 The Zhèngyǔ Zuòcí Qǐlì 正語作詞起例 7.2.1 Dialect Features 7.3 The Phonetic Reconstruction 7.3.1 Initials (21) 7.3.2 Finals (47)

204 204 205 206 206 207 208 210 215 215 225

229 231 234 234 238 238 240 241 243 244 244 245 246 247 247 248 248 249 249 250 251 255 255 256 257 258 259

262 264 268 270 270 272 272 276

Contents 7.4 Innovative Characteristics of the Zhōngyuán Yīnyùn 中原音韻 7.4.1 Devoicing of the Middle Chinese Voiced Obstruents in the Initial Position 7.4.2 The Emergence of the Labiodental Fricative 7.4.3 The Development of the High Central (Apical) Vowels 7.4.4 The New Four-Tone System 7.4.5 The Loss of the Middle Chinese yí 疑 Velar Nasal Initial 7.4.6 The Disappearance of the Four-Rank System of Finals 7.4.7 The Preservation of the -m Coda 7.4.8 The jiān-tuán 尖團 Distinction 7.5 Examples of Yuan Pronunciation 7.6 Summary

8

Old Mandarin: The Persian Transcriptions 8.1 The Ancient Persian Script 8.2 History of China 8.2.1 Middle Chinese Voiced Stops and Affricates 8.2.2 The Loss of Velar Nasal Initial yí 疑 ŋ8.2.3 Middle Chinese rì 日 Initial 8.2.4 Middle Chinese wēi 微 Initial 8.2.5 The Coda of rù 入 Syllables 8.2.6 High Centralized (Apical) Vowels 8.2.7 Main Vowel -e 8.2.8 Summary of Features 8.3 The Mài Jué 脈訣 8.3.1 The Stop Codas of the rù 入 Syllables 8.3.2 The Inconsistency of the Stop Coda in the Transcription 8.3.3 The Dialectal Features Reflected in the Transcription 8.3.4 The Nebulous Nature of Early Guānhuà 官話

278 279 279 281 282 285 286 290 290 291 293

294 294 295 296 296 298 299 300 301 301 302 302 304 306 307 313

Toward Modern Mandarin

317

The Mandarin of the Ming Dynasty

319

Part VI 9

ix

9.1 North Versus South 9.1.1 Guānhuà 官話 9.2 The Hóngwǔ Zhèngyùn 洪武正韻 (1375) 9.2.1 Initials 9.2.2 Finals 9.2.3 Tones 9.2.4 A Coherent System 9.2.5 Zhèngyīn 正音 and Súyīn 俗音 9.2.6 Revision of the Hóngwǔ Zhèngyùn 9.3 The Yùnluè Yìtōng 韻略易通 (1442) 9.3.1 Initials and the Zǎ o Méi 早梅 ‘Early Plum Blossom’ Poem 9.3.2 Finals and the Twenty Rhyme Groups 9.3.3 Běnwù’s 本悟 Revision 9.3.4 Tones 9.4 The Sìshēng Tōngjiě 四聲通解 (1517), Fānyì Lǎo Qǐdà 翻譯老乞大, and Fānyì Piáo Tōngshì 翻譯朴通事 9.4.1 Initials

319 320 320 322 322 323 325 325 326 327 328 329 331 333 334 335

x

Contents 9.4.2 Finals 9.4.3 Tone Sandhi 9.5 The Xīrú Ěrmù Zī 西儒耳目資 (1626) 9.5.1 Initials 9.5.2 Finals 9.5.3 Tones 9.6 The Děngyùn Tújīng 等韻圖經 (1606) 9.6.1 Initials 9.6.2 Finals 9.6.3 Tones 9.6.4 Innovative Features

10

The Mandarin of the Qing Dynasty and the Modern Era 10.1 The Qing Dynasty (1644–1912) 10.1.1 The Yùdìng Pèiwén Yùnfŭ 御定佩文韻府 (1711) 10.1.2 The Yīnyùn Chǎnwēi 音韻闡微 (1728) 10.1.3 The Wade Textbook (1867) 10.2 Post-Imperial China (1912–present) 10.2.1 The Old National Standard 10.2.2 The New National Standard 10.2.3 Modern Standard Phonology 10.2.4 Modern Dialects (Fāngyán 方言)

References Index

336 340 342 342 345 349 350 350 355 356 356

360 360 361 366 369 377 377 378 379 379

385 395

Figures

1.1 A page from the Kānmiù Bǔquē Qièyùn 刊謬補缺切韻. page 13 1.2 The first table in the Yùnjìng 韻鏡 ‘Mirror of Rhymes’, one of the earliest rhyme tables. 24 1.3 A page from the Sìshēng Děngzǐ 四聲等子. 32 1.4 The Yùnjìng pages for the rhymes dōng 東, dōng/zhōng 冬/鍾, and jiāng 江, and their zè 仄 tone equivalents. 40 2.1 Baxter and Sagart’s representation of the Old Chinese root. 100 3.1 Middle Chinese (and Modern Mandarin) syllable structure. 128 4.1 A page from the Ze-Cun-Tang version (澤存堂本 zécúntáng běn) of the Guǎngyùn 廣韻. 167 4.2 Two pages from the Sìshēng Děngzǐ, showing the merger of the syllables of the dōng 東, dōng 冬, and zhōng 鍾 rhyme series. 184 4.3 The first table in the Qièyùn Zhǐzhǎngtú 切韻指掌圖. 188 4.4 The first page of the tables in the Qièyùn Zhǐnán 切韻指南. 190 4.5 A page from the Qièyùn Zhǐnán, showing the Ménfǎ Yùyàoshi 門法玉鑰匙. 191 5.1 A rubbing from the Dàozōng Āicè 道宗哀冊. 199 5.2 The titles of Yélǜ Gù 耶律固. 207 5.3 A page from the Zhǎng Zhōng Zhū 掌中珠. 217 6.1 A page from the Ménggǔ Zìyùn 蒙古字韻. 232 6.2 A comparison of the Tibetan and hP’ags-pa scripts. 235 _ raising in the Yuan era. 6.3 A vowel chart demonstrating vowel 260 7.1 A page from the Zhōngyuán Yīnyùn 中原音韻. 263 7.2 A page from the second part of the Zhèngyǔ Zuòcí Qǐlì 正語作詞起例. 271 8.1 A page of the Mài Jué 脈訣. 303 9.1 A page from the Hóngwǔ Zhèngyùn 洪武正韻 (Sìkù Quánshū 四庫全書 version). 321 9.2 A page from the Lǎo Qǐdà Yànjiě 老乞大諺解. 343 9.3 Two pages from the Xīrú Ěrmù Zī 西儒耳目資. 344

xi

xii

List of Figures

10.1 The first three pages of the Pèiwén Yùnfǔ 佩文韻府 (Sìkù Quánshū version). 10.2 The first page of the Yīnyùn Chǎnwēi 音韻闡微 (Sìkù Quánshū version). 10.3 The cover of 語言自邇集 Yü-yen tzǔ-êrh chi.

363 367 370

Tables

1.1 1.2 1.3 1.4 1.5 1.6 1.7 1.8 1.9 1.10 1.11 1.12 1.13 1.14 1.15 1.16 1.17 1.18 1.19 1.20

Additional syllable terms with their component segments page Tone sandhi in modern Beijing dialect Illustration of the Qièyùn structure, as excerpted from the Kānmiù Bǔquē Qièyùn The representative characters of each rhyme of the Qièyùn The structure of fǎnqiè spellers An example of the fǎnqiè system The pronunciation of zhuāng 樁 ‘stake’ throughout history More examples of fǎnqiè spellings that are obsolete in Modern Standard Mandarin The basic syllable structure with the medial broken down into its component parts Categorical combinations realized differently phonetically The four tones of Middle Chinese, with píng-zè and shū-cù tones demarcated Rhyme series with both nasal and nonnasal píng tone endings An example of allophonic variants of the columns of rhyme tables The thirty-six initials in the Qīyīn Lüè, with their reconstructions The reconstructed values of three groups of the syllables from the thirty-six initial characters A comparison of the initial characters and their categorical terminologies Definitions and categorizations of the Jīngshǐ Zhèngyīn Qièyùn Zhǐnán Categories in the Yùnjìng directly compared to their modern phonemic analogues An analysis of the four groups of qīngzhuó patterns The eight tones divided along traditional tonal categories and the yīndiào–yángdiào distinction

10 11 14 15 17 17 19 19 19 20 21 21 25 25 26 26 27 27 28 29

xiii

xiv

List of Tables

1.21 The eight register-tones as commonly labelled in Chinese phonology 1.22 The names and Middle Chinese representations of Mandarin tones 1.23 Divisions reflected in their reconstruction through similar vowels or through different medials 1.24 Divisions reflected in their reconstruction through different initials 1.25 Divisions allowed among each initial group 1.26 Subdivision of zhào 照 initials into zhào-èr 照二 initials and zhào-sān 照三 initials 1.27 The relationship between divisions and ranks 1.28 The relationship between the Qièyùn and rhyme tables (Rank-III rhymes) 1.29 Division placements of syllables of the Rank-III yáng 陽 rhyme 1.30 Division placements of syllables of the Rank-III zhī 支 rhyme 1.31 A comparison of rhyme groups in the Qièyùn and Guǎngyùn, showing a rhyme split along the kāihé contrast 1.32 A comparison of the classification of kāihé in the Yùnjìng and the Gǔjīn Zìyīn Duìzhào Shǒucè 1.33 The four kāihé labeled tables in the Yùnjìng 1.34 Examples of the sìhū 1.35 The medial systems of Middle Chinese and the sìhū 1.36 The division assignments of different rhymes within the rhyme groups (shè), divided by their zhuǎn 1.37 The vowel quality of rhymes in the nèizhuǎn and wàizhuǎn rhyme groups, according to Luó Chángpéi 1.38 The phonetic values of nèizhuǎn and wàizhuǎn rhyme groups, according to the phonetic components of the Zhuang characters 1.39 A summary of the long and short vowels in Table 1.38 1.40 A summary of the long and short vowels in Table 1.37 1.41 The sixteen shè with their basic main vowels and codas 1.42 The shè system systematically categorized into different vowel-coda pairings 1.43 The parallel structure of Rùshēng Yùn and Yángshēng Yùn seen through a comparison of their codas 1.44 The xiéshēng characters of gōng 工 1.45 A grading of different source materials by six categories in historical phonology 1.46 The availability of different source materials for the study of the major historical periods

29 30 31 33 33 34 34 34 35 35 35 36 36 38 39 43 43 45 45 46 47 47 48 51 53 53

List of Tables

1.47 A generalization of the phonological information available at each period of Chinese phonological history 1.48 An example of the reconstruction format, using the character luò 落 2.1 The periodization of Old Chinese associated with different Chinese dynasties 2.2 The four language families, and some of their constituent languages, in the Sino-Tibetan language family 2.3 Some poems from the Shījīng 2.4 Examples from the gōng 工, chéng 成, mén 門, jiān 監, and měi 每 xiéshēng series 2.5 Examples of partially reduplicated words 2.6 Two analyses of Chinese syllable structure 2.7 A common analysis of an Old Chinese syllable 2.8 Reconstructions exemplifying the varied syllable structure of Old Chinese, with as many as six segments CCCVCC 2.9 Li Fang-Kuei’s reconstruction system (thirty initials) 2.10 Baxter’s reconstruction system (thirty-seven initials) 2.11 Zhèngzhāng’s reconstruction system (twenty-five initials) 2.12 The nineteen universal initials seen throughout Old Chinese reconstructions 2.13 Some characters with component bā 巴, categorized by Middle Chinese initial 2.14 Xiéshēng characters from the gè 各 and jiān 監 components 2.15 Jaxontov’s reconstruction of several words with their cognates 2.16 Li Fang-Kuei’s reconstruction of gè 各 component xiéshēng characters 2.17 A summary of Li Fang-Kuei’s explanation of the -r- medial being responsible for the creation of the retroflex series 2.18 Jaxontov’s reconstructions of sC- clusters across different nasals and approximants 2.19 Li Fang-Kuei’s initial reconstruction of cì 賜 ‘bestow’ and xiē 楔 ‘wedge’ as alveolar and palatal stop consonant initials, respectively 2.20 Zhèngzhāng’s reconstruction of sC- and hC- clusters 2.21 Li Fang-Kuei’s reconstruction of voiceless nasals and laterals 2.22 Jaxontov’s examples of three types of syllables with l- and syllables from Rank-II with other initials 2.23 Jaxontov’s examples of syllables with l- and syllables from Rank-IIIb with other initials 2.24 Some examples of Old Chinese rhyme groups (yùnbù)

xv

54 56 60 60 64 65 66 67 68 68 69 69 70 70 71 71 72 72 73 74

74 75 75 76 77 78

xvi

List of Tables

2.25 The yùnbù of Old Chinese arranged according to three types of codas 2.26 Some of Kǒng Guǎngsēn’s paired yīn-yáng rhymes with their reconstructions 2.27 Dài Zhèn’s suggested relationship between the yīn, yáng, and rù 入 syllables 2.28 The xiéshēng components of the rhyme groups zhī bù 之部, zhí bù 職部, and zhēng bù 蒸部 2.29 The possible combinations of medials in Middle Chinese, and their respective kāihé and rank 2.30 The four rhyme patterns listed by Jaxontov 2.31 Jaxontov’s proposal 2.32 The frequency of Qièyùn rhymes (ignoring tone) by their rank 2.33 Rank-III syllables in the Taishun dialect 2.34 Go-on and kan-on readings of Rank-III characters 2.35 More examples of go-on readings lacking palatal glides otherwise seen in kan-on 2.36 Length contrast and vowel quality in Chinese dialects 2.37 The fǎnqiè spellings of Rank-III syllables 2.38 Karlgren’s fourteen reconstructed vowels 2.39 Li Fang-Kuei’s four-vowel system 2.40 Baxter’s six-vowel system 2.41 Zhèngzhāng’s six-vowel system 2.42 A comparison of -ŋ coda syllables in Li Fang-Kuei’s four-vowel system and Zhèngzhāng’s six-vowel system. 2.43 Relationship between tones in Middle Chinese and coda clusters in Old Chinese 2.44 Rhyming words in the Shījīng and Chǔcí 2.45 Zhèngzhāng’s suggested tones that coexisted with different finals 2.46 Basic words in the shǎng tone 2.47 Middle Chinese qù syllables with -s coda in Sino-Korean 2.48 Sanskrit -s and -ʂ transcribed by using qù tone syllables 2.49 Middle Chinese qù tone characters used to transcribe syllables with a stop coda in Sanskrit transcription 2.50 Xiéshēng characters with qù tone and rù tone 2.51 More examples of rhymes jì 祭, tài 泰, guài 夬, and fèi 廢 2.52 The reconstructed earlier forms of jì 祭, tài 泰, guài 夬, and fèi 廢 2.53 Reconstructed finals of jì 祭, tài 泰, guài 夬, and fèi 廢 for Old Chinese and Middle Chinese

79 79 80 80 81 82 82 83 84 85 85 85 86 87 88 89 89 90 91 91 92 93 94 94 94 94 95 95 95

List of Tables

2.54 Examples of -s functioning as a morphological particle 2.55 Further examples of the morphological function of -s with different characters 2.56 A summary of the relationship between the codas and tones 2.57 A comparison of two-way and three-way contrast for the alveolar codas 2.58 Morphologically related pronoun pairs 2.59 Some examples illustrating the devoicing regular-causative pattern in Old Chinese 2.60 Baxter and Sagart’s reconstruction of Type A and B syllables 2.61 Nonpharyngealized Old Chinese initial alveolar stops and nasals and Middle Chinese palatal affricates and nasals 2.62 The Old Chinese onsets with initial or medial *r and Middle Chinese retroflex initials 2.63 The Old Chinese nasal pre-initial and Middle Chinese voiced stops and affricates 2.64 Correspondence table between Pān’s uvular proposal (uvulars and velars) for Old Chinese, and the Middle Chinese reflexes 2.65 The reconstructions of various stages of Chinese, reflecting the original Old Chinese pronunciation of Jìng Nǚ 靜女 2.66 The reconstructions of various stages of Chinese, reflecting the original Old Chinese pronunciation of Táo Yāo 桃夭 3.1 Evidence of codas -m, -n, -ŋ and -p, -t, -k in the Qièyùn system 3.2 The fǎnqiè spellings and Middle Chinese reconstruction for Rank-III syllables, with the lower speller of jū 居 or yú 魚 3.3 Karlgren and his successors’ reconstructions 3.4 Upper spellers of Rank-III and non-Rank-III characters 3.5 The divisional placement of the Qièyùn rhymes in the rhyme tables 3.6 The divisional placement of the Rank-III syllables of the Qièyùn in the rhyme table Yùnjìng 3.7 The nature of the Qièyùn rhymes 3.8 The parallel nature of rhymes across different tones 3.9 195 rhymes from Wáng Rénxù’s version of the Qièyùn 3.10 Pān’s reconstruction organized in table style 3.11 Main vowels and medials of Pān’s reconstruction 3.12 Pān’s reconstructed main vowels organized by medial 3.13 The rhyme splits between the Qièyùn, Kānmiù Bǔquē Qièyùn, and Guǎngyùn 3.14 Rhyme splitting from the Kānmiù Bǔquē Qièyùn to the Guǎngyùn

xvii

96 96 97 97 98 98 100 101 101 101 102 103 103 108 109 109 109 110 111 115 116 117 117 118 118 119 120

xviii

List of Tables

3.15 Samples of two different approaches (Mài 1995; Pān 2000) 3.16 An example of Mài’s reconstruction of Middle Chinese rhymes 3.17 The Main Vowel Approach and the Medial Approach to yùn 韻 analysis 3.18 Rhyme categories of Yán Zhītuī’s Guān Wǒ Shēng Fù 3.19 An interpretation of Lù Fǎyán’s judgments 3.20 Examples of the rhymes that split or not along the medial difference 3.21 The zhēn 真 rhyme and the zhēn 臻 rhyme are in complementary distribution 3.22 The type-distribution of Rank-III rhymes 3.23 The relationship between the gēng-III 庚三 and qīng 清 rhyme series 3.24 The relationship between the yōu 幽 and yóu 尤 rhyme series 3.25 A basic outline of the syllable structure as analyzed from the Qièyùn 3.26 Simplification of the Chinese syllable structure 3.27 The parallelism of reconstructed pronunciations categorized by tone 3.28 The Chinese-Sanskrit transcriptions in the Yíqiè Jīng Yīnyì by Buddhist monk Huìlín 3.29 Some descriptions of tones in the Xī Tán Zàng by the Japanese monk Annen 3.30 The sporadic organization of initials in different rhymes 3.31 Pān’s reconstruction of thirty-seven initials 3.32 An example of the twelve-vowel system needed to maintain non-medial based rhyme differentiation 3.33 Three phonation types of stops and affricates for each place of articulation 3.34 The labial series before the split 3.35 Sanskrit letters for retroflex stops and nasals and corresponding zhī 知 initial series characters 3.36 The retroflex obstruent and nasal syllables in Sanskrit and the corresponding Chinese characters 3.37 The coexistence of retroflex and palatal affricates, and retroflex stops 3.38 The changes of grave and acute initials, and the -r- of Old Chinese 3.39 The four ranks of the xiào 效 rhyme group, illustrating Karlgren’s reconstruction the four ranks of the Guǎngyùn 3.40 Karlgren’s reconstruction in terms of medials and main vowels

121 121 123 123 124 125 126 127 127 127 128 129 130 130 131 132 133 134 135 136 136 137 137 138 138 138

List of Tables

3.41 Redundancy in criteria for distinguishing the syllables of the four ranks 3.42 The relationship between the four ranks as related to the medials -ɰ- and -j3.43 Simpler main vowels reconstructed phonemically 3.44 Calculations of the prevalence of the rank of upper spellers for Rank-III characters in the Qièyùn 3.45 The complementary distribution of syllables with labial initials in rhymes with -n coda 3.46 Sino-Sanskrit proof of the existence of the medial -r- in IIIb characters 3.47 Examples of chóngniǔ contrasts across all eight rhymes that contain them 3.48 The traditional chóngyùn rhymes 3.49 Some samples of the reconstructed phonetic values of the chóngyùn pairs 3.50 A series of minimal pairs, exemplifying a difference in the manner of articulation of the coda 3.51 The parallel relationship between syllables with a nasal coda and syllables with a stop coda 3.52 Rhymes jì 祭, tài 泰, guài 夬, and fèi 廢 with no equivalent píng and shǎng tone rhymes 3.53 A comparison of the Beijing, Suzhou, Wenzhou, and Shuangfeng dialects, highlighting the maintenance of a three-way phonation type contrast in the Wu and Xiang dialects 3.54 A comparison of the Beijing, Suzhou, Guangzhou, and Yangjiang dialects, highlighting different coda systems, each descended from Middle Chinese, maintaining different contrasts, and merging others 3.55 An example of chóngniǔ contrast reflexes 3.56 Some examples of the chóngyùn contrast of Rank-I rhymes hāi 咍 and tài 泰 3.57 The chóngyùn contrast between the yú 魚 and yú 虞 rhyme groups in the Wu dialects 3.58 The contrast between Rank-III and -IV in the Wu dialects 3.59 The reconstructed Middle Chinese pronunciation of Qiān Zì Wén 3.60 The reconstructed Middle Chinese pronunciation of Běi Zhǔ 4.1 Reconstructions of Xuánzàng’s transliterations

xix

139 139 140 140 141 142 142 143 144 144 145 145

147

147 148 149 149 150 151 151 153

xx

4.2 4.3 4.4 4.5 4.6 4.7 4.8 4.9 4.10 4.11 4.12 4.13 4.14 4.15 4.16 4.17 4.18 4.19 4.20 4.21 4.22 4.23 4.24

4.25 4.26 4.27

List of Tables

The rhyming words in poems composed during the imperial examinations before the year 717 The twelve -ŋ coda rhymes of the Qièyùn The seven groups according to tóngyòng labels The twelve rhymes in the Qièyùn system with -n coda The additional tóngyòng of wén 文 and yīn 殷, added in 1037 All tóngyòng rhymes, organized by coda A rhyme chart arranged according to tóngyòng markings The values of Qièyùn and post-717 Tang poetry rhymes The relationship between the phonetic, phonemic, and rhyming systems of Tang-era Middle Chinese The progression of devoicing across Tang-era Northwestern Chinese dialects The Tang era Chinese three-way contrast within labiodental sounds, reconstructed as f-, v-, and ɱ-, respectively The labiodentalization of labials preceding the medial -j- in Tang-era Chinese The labiodental fricatives realized as /f/ and /v/ in SinoVietnamese The distinction of Early and Late Middle Chinese in Pulleyblank’s reconstruction The distribution of Qièyùn initials in the Ěryǎ Yīntú The devoicing pattern in the Ěryǎ Yīntú The twelve Yīn Tú tables, with their respective reconstructed initials for each row’s gaps The twenty-one initials of the Shēngyīn Chànghè Tú Shào Yōng’s assignment of bh- initials resembles Indic phonology The relationship between voiced píng and voiced zè syllables The complementary distribution of labiodentals across tones The distribution of labiodentals in the Shēngyīn Chànghè Tú, categorized by Middle Chinese initial and tone A comparison of different texts’ reconstructions of the bilabial and labiodental series, showcasing the evolution of the series from Middle Chinese to Modern Mandarin The seven non-empty Shēng Tú tables, with their respective reconstructed finals for each row The merger of syllables with -k coda and a back vowel with syllables with -w coda The merger of syllables with -k coda and a non-back vowel with syllables with -j coda

155 155 155 155 156 156 157 158 159 159 160 161 161 162 164 164 170 170 171 171 172 172

173 174 175 175

List of Tables

4.28 The phonological conditions for the reflexes with -w and -j coda 4.29 The main vowel of shuāi 衰 and shuài 帥 in Song-era Chinese 4.30 A comparison of Middle Chinese -ŋ rhymes and those of the Shēngyīn Chànghè Tú 4.31 The simplification of rhymes with -ŋ codas, resulting in a four-way contrast 4.32 A final chart of the Shēngyīn Chànghè Tú 4.33 The tonal reflexes of Middle Chinese tone II syllables according to different types of initials 4.34 The tonal reflexes of Middle Chinese tone II syllables according to different types of initials (tones written in full) 4.35 A correspondence table between Middle Chinese, the Chinese of the Shēngyīn Chànghè Tú, and Modern Mandarin 4.36 Láng Yīng’s referenced characters 4.37 Middle Chinese shǎng tone syllables merge differently both historically and geographically 4.38 Chao’s noted features in the Hangzhou dialect 4.39 The kāihé divide in the jiāng 江 rhyme series 4.40 The reflexes of the syllables jiāng 江 rhyme in Modern Mandarin 4.41 A comparison of the ordering of tone and division in the Yùnjìng and Sìshēng Děngzǐ 4.42 The double listing of the rùshēng syllables in the Sìshēng Děngzǐ 4.43 An allophonic split in the zhī 之 and zhī 支 rhyme series 4.44 The reconstructions of Middle Chinese, reflecting the original Tang dynasty Chinese pronunciation of Lù Zhài 4.45 The reconstructions of Middle Chinese, reflecting the original Tang dynasty Chinese pronunciation of Xīn Yí Wū 5.1 Diphthongization of coda -k syllables as shown in Khitan sources 5.2 Khitan characters illustrate a Mandarin-like obstruent devoicing pattern 5.3 Devoicing as shown in Khitan materials 5.4 Labiodentalization of phu in Sino-Khitan 5.5 Initial series, zhī 知 and zhào 照, merge in Sino-Khitan 5.6 The loss of stop codas in Sino-Khitan 5.7 Velar nasals in the onset position of Sino-Khitan 5.8 Nasal codas’ three-way distinction in Sino-Khitan 5.9 Conditioned vowel raising after a palatal glide in Sino-Khitan 5.10 A multidialectal gloss of the titles of Yélǜ Gù

xxi

176 176 177 178 178 179 179 180 182 183 183 185 186 186 187 189 193 193 201 202 202 203 203 204 205 205 206 208

xxii

List of Tables

5.11 Examples of different rhymes merging in the GYJ 5.12 Rank and chóngniǔ mergers during the Jin dynasty 5.13 A comparison of mergers across three texts before, during, and after the Jin dynasty 5.14 Diphthongization of coda -k syllables as shown in Jurchen sources 5.15 Diphthongization of coda -k causes mergers between rù and non-rù syllables 5.16 Devoicing patterns in Sino-Jurchen as a comparison of píng tone characters 5.17 The labiodentalization process of Sino-Jurchen 5.18 Obstruent coda loss in Sino-Jurchen 5.19 The velar nasal in the onset position of Sino-Jurchen 5.20 Nasal codas’ three-way distinction in Sino-Jurchen 5.21 The initial system of Sino-Tangut 5.22 The Sino-Tangut vowel system 5.23 Obstruent devoicing across different tones in Sino-Tangut 5.24 Voiced and voiceless fricatives merge in Sino-Tangut 5.25 The labiodentalization process of Sino-Tangut 5.26 Initial series, zhī 知 and zhào 照, merge in Sino-Tangut 5.27 The loss of stop codas in Sino-Tangut 5.28 No velar nasal coda transcribed in Sino-Tangut 5.29 Tangut words transcribed by syllables both with and without velar nasal coda 5.30 The loss of nasal codas in Sino-Tangut 5.31 High front vowels of Middle Chinese zhī 支, zhī 脂, and zhī 之 rhyme undergo a conditioned centralization in Sino-Tangut 5.32 A comparison of five modern Fenghe dialects 5.33 A comparison of northern and northwestern dialect sound changes 6.1 The Ménggǔ Zìyùn as a fusion of previous works 6.2 A comparison of syllable creation in Tibetan and hP’ags-pa _ script 6.3 Non-consonantal alloglyphs 6.4 The realization of different FT/vowel combinations in hP’ags-pa _ script 6.5 The realization of hP’ags-pa finals in different locations _ 6.6 Types of syllable blocks for transcribing Chinese characters 6.7 Differences across the orthographic-phonological interface 6.8 Transcription of dōng 東 and gēng 庚 rhyme groups

209 210 210 211 212 213 213 214 214 215 216 218 219 219 220 221 222 223 223 223

224 226 227 233 235 236 237 237 239 240 241

List of Tables

6.9 6.10 6.11 6.12 6.13 6.14 6.15 6.16 6.17 6.18 6.19 6.20 6.21 6.22 6.23 6.24 6.25 6.26 6.27 6.28 6.29 6.30 6.31 6.32 6.33 6.34 6.35

6.36 6.37 7.1

The thirty-six initial characters with their hP’ags-pa equivalents, _ enumerated Three zìmǔ split according to their allophonic variations The fifteen rhyme groups of the Ménggǔ Zìyùn An example of the initial arrangement pattern in the Ménggǔ Zìyùn The thirty-two initial sounds of the Ménggǔ Zìyùn Phonetic values of the finals in the fifteen rhyme groups hP’ags-pa spellings of the transcriptions in Table 6.14 _ Vowels of the Ménggǔ Zìyùn The vowel-coda subsystem of the Ménggǔ Zìyùn No tonal distinction in hP’ags-pa transcriptions The voicing reversal of _stop initials Labiodentalization in the Ménggǔ Zìyùn Conditional loss of Middle Chinese initial ŋDifferences in the realization of medials depending on kāihé Changes to the Middle Chinese medial -ɰ- in different contexts Vowel Reduction in the Ménggǔ Zìyùn from the Middle Chinese final system The phonetic values of -Vŋ/-Vk finals before and after the Ménggǔ Zìyùn’s vowel reduction Type A and Type B rhyme groups Chóngniǔ contrasts maintained between Rank-III and Rank-IV More examples showing the contrast of IIIa and IIIb syllables A summary of chóngniǔ reflexes The chóngniǔ reflexes of Type A and Type B rhyme groups Diphthongization of syllables with Middle Chinese coda -k Middle Chinese -k syllable changes and their rules Change of relationship between yáng and rù syllables The contrast between the finals of Rank-II and Rank-III/IV syllables The changes between the Chinese of the Ménggǔ Zìyùn and Modern Mandarin, with the data from Table 6.34 reorganized as examples of each change Centralization of high vowels in the Ménggǔ Zìyùn A set of rules determining vowel raising in Yuan-era Mandarin A comparison of the arrangement of the Qièyùn, Ménggǔ Zìyùn, and Zhōngyuán Yīnyùn

xxiii

242 243 243 244 245 246 246 247 247 247 248 248 249 249 250 251 251 252 252 253 253 254 256 257 257 258

259 259 260 264

xxiv

7.2 7.3 7.4 7.5 7.6

7.7 7.8 7.9 7.10 7.11 7.12 7.13 7.14 7.15 7.16 7.17 7.18 7.19 7.20 7.21 7.22 7.23 7.24 7.25 7.26 7.27

List of Tables

All the rhyme groups of the Zhōngyuán Yīnyùn Middle Chinese rù tone syllables found in the xiāoháo rhyme group Some characters that maintain the -aw reflex from Dadu dialect The presence and absence of -w coda in the relevant modern dialects A comparison of rhyme-ending realizations between the Zhōngyuán Yīnyùn, and the modern Luoyang and Beijing dialects The irregular changes of modern Luoyang and Beijing dialects Contrast of -on and -wan in the Zhōngyuán Yīnyùn A comparison of rank contrast across many modern dialects Dual pronunciation of some Chinese characters Characteristics of different Mandarin subdialects during the Yuan era Some contrasts listed by Zhōu Déqīng in the Zhōngyuán Yīnyùn The initial system of the Zhōngyuán Yīnyùn Minimal triplets imply a three-way contrast A comparison of different hypotheses for the value of the retroflex series Reconstructed values of finals in the Zhōngyuán Yīnyùn, organized by rhyme group A comparison of the vowel systems of the Zhōngyuán Yīnyùn and Ménggǔ Zìyùn Summary of the vowel system in the Zhōngyuán Yīnyùn The vowel system of the Zhōngyuán Yīnyùn before null and -n codas The vowel system of the Zhōngyuán Yīnyùn before -ŋ coda Devoicing pattern of Middle Chinese voiced stops and affricates The Zhōngyuán Yīnyùn demonstrates a merger of voiced and voiceless unaspirated stops and affricates The merger of labiodental fricatives The merger of three labial stops in qù syllables The loss of nasality of the wēi ɱ- initial The contrast between the wēi ɱ- and the yǐng Ø- initials A comparison of the vowels of the Middle Chinese zhǐ 止 rhyme group within zhī 知, zhāng 章, and zhuāng 莊 group initials

264 265 266 266

267 267 268 269 269 270 273 274 274 275 276 277 277 278 278 279 280 280 280 281 281

281

List of Tables

7.28 The rhyme and tone of non-zhǐ 止 rhyme group words with vowel -i 7.29 Centralization of high vowels in the Zhōngyuán Yīnyùn 7.30 The merging and splitting conditions between Middle Chinese and the early Mandarin of the Zhōngyuán Yīnyùn 7.31 Mergers of the rù tone with three other tones 7.32 Middle Chinese codas are lost and rù tone syllables with different codas merge 7.33 Interference between two changes to the Middle Chinese velar nasal initial 7.34 Rank-II syllables merged with different finals based on rhyme and place of articulation of the initial 7.35 The Zhōngyuán Yīnyùn merge between Rank-III and Rank-IV syllables 7.36 A comparison of chóngniǔ contrasts (Middle Chinese IIIa and IIIb) present in reflexes of the Zhōngyuán Yīnyùn and modern Beijing dialect 7.37 The contrast of Rank-II and Rank-III syllables in the Zhōngyuán Yīnyùn 7.38 The syllables of the chē-zhē rhyme group 7.39 A correspondence table showing the Beijing dialect reflexes of the Zhōngyuán Yīnyùn chē-zhē rhyme 7.40 Contrasts between the -m and -n coda that remained in the Zhōngyuán Yīnyùn 7.41 Contrasts of the jiān-tuán distinction 7.42 A four-stage process resulting in the merging of velars and coronals in front of high vowels 7.43 The pronunciation of Mǎ Zhìyuǎn’s Qiūsī in the Mandarin of the Zhōngyuán Yīnyùn compared with Modern Mandarin 8.1 The Persian letters present in the Mài Jué and their IPA values 8.2 A comparison of transcriptions of voiceless aspirated and unaspirated Middle Chinese initials into Persian script in various personal names 8.3 Persian transcriptions follow the same devoicing pattern present in Mandarin 8.4 A loss of the yí 疑 initial is evident in the History of China 8.5 Examples highlighting the non-rhotic transcription of the Middle Chinese rì 日 initial in the History of China 8.6 Character èr 二 transcribed as a rhotacized vowel

xxv

282 282 283 284 285 285 286 286

287 288 289 289 290 291 291

292 295

297 297 298 298 298

xxvi

8.7 8.8 8.9 8.10 8.11 8.12 8.13 8.14 8.15 8.16 8.17 8.18

8.19 8.20

9.1 9.2 9.3 9.4 9.5 9.6 9.7 9.8 9.9 9.10 9.11 9.12 9.13 9.14 9.15

List of Tables

Different renderings of hP’ags-pa ʋ- initial _ a lack of rù syllable stop codas History of China features Rù syllables transcribed with ‹ h › and ‹ q › endings Transcription of high central vowels Zero form of vowels -e and -ɛ A series of rù syllables in the Persian Mài Jué that maintain their coda in transcription A transcription of the Rènfù Shānghán Gē Middle Chinese codas and their transcriptions in the Persian Mài Jué Chinese characters with different transcriptions in the Persian Mài Jué Phonological attributes of the PMJ-1; contrast with Table 8.17 Phonological attributes of the PMJ-2; contrast with Table 8.16 A comparison of characters across phonological attributes of the Persian Mài Jué, along with modern Beijing, modern Guangzhou, and the dialect of the Zhōngyuán Yīnyùn A compact generalization of the phonological attributes shown in Table 8.18 A comparison of Beijing and Suzhou dialects, reflecting the differences in rì 日 initial syllables and Rank-II kāikǒu syllables The thirty-one initials of the Hóngwǔ Zhèngyùn The relationship between Hangul letters, the intrinsic values of Hangul, and Ming Chinese values The phonetic values of the finals in the Hóngwǔ Zhèngyùn Yìxùn A final chart of the Hóngwǔ Zhèngyùn Yìxùn A comparison of the zhī 支 rhyme before and after the split A comparison of the modern Beijing and Hefei dialects Zǎo Méi poem with phonetic values and pinyin By reorganizing the characters of Zǎo Méi, the initial system of the Ming dynasty is illuminated Conditional variants of zhī 枝, chūn 春, and shàng 上 initials Rhyme groups of the Yùnluè Yìtōng and their respective finals Examples of the mergers in Běnwù’s revision The loss of the palatal medial -j- after retroflex initials The contrast of the rù syllables in Mandarin dialects In a retroflex initial environment, -i becomes centralized, and merges with -ɨ syllables The main vowel system of the Yùnluè Yìtōng

299 300 300 301 301 305 305 306 307 309 310

311 312

314 322 323 324 324 327 327 328 329 329 330 331 331 332 332 333

List of Tables

9.16 A final chart representing the phonological system of the Yùnluè Yìtōng 9.17 The tonal system of the Yùnluè Yìtōng 9.18 Yīnpíng and yángpíng syllables as in the Yùnluè Yìtōng 9.19 The initial system of the Sìshēng Tōngjiě 9.20 Some of the exceptional words 9.21 The phonetic transcriptions of finals in the Sìshēng Tōngjiě 9.22 A final chart of the Chinese phonology based on the Hangul spellings 9.23 Rank-III syllables of the tōng 通 rhyme group in the súyīn and the jīn súyīn 9.24 The merging of the dàng 宕 and jiāng 江 groups 9.25 The change of syllables with final -jaj 9.26 The change of syllables with final -waw 9.27 The merger of Rank-II and Rank-III/IV syllables with coda -n or -m 9.28 The merger of Rank-II and Rank-III/IV syllables with coda -w 9.29 The merger of Rank-II and Rank III/IV finals conditioned by codas 9.30 The change of words with final -iw 9.31 Tone sandhi as explained in the Sìshēng Tōngjiě 9.32 Tone sandhi for longer phrases 9.33 The dialogue as presented in Figure 9.2, with pinyin and gloss 9.34 Initials of the Xīrú Ěrmù Zī 9.35 Initial chart of the Xīrú Ěrmù Zī 9.36 Some examples of finals transcribed by Trigault 9.37 The spellings of each of the fifty final tables as written in the Xīrú Ěrmù Zī 9.38 The phonetic interpretations of the spellings in the fifty final tables of the Xīrú Ěrmù Zī 9.39 The phonetic interpretations (fifty-four finals) of the spellings in the fifty final tables of the Xīrú Ěrmù Zī, arranged by medial and rime 9.40 Phonetic features of central vowels and other relevant vowels 9.41 Medials, main vowels, and codas of Trigault’s Xīrú Ěrmù Zī 9.42 The tones of the Xīrú Ěrmù Zī 9.43 The twenty-two initials given in the Děngyùn Tújīng 9.44 The nineteen actual initials of the Děngyùn Tújīng 9.45 The merger of k- initial series and ts- initial series 9.46 Three possible contrasts between the two initial groups

xxvii

333 333 333 335 336 336 337 338 338 339 339 339 339 340 340 341 341 342 345 345 346 346 347

348 349 349 349 350 351 351 351

xxviii

9.47 9.48 9.49 9.50 9.51 9.52 9.53 9.54 9.55 9.56 9.57 9.58 9.59 10.1 10.2 10.3 10.4 10.5 10.6 10.7 10.8 10.9 10.10 10.11 10.12 10.13 10.14 10.15

List of Tables

The change order of the palatalization from Middle Chinese to Modern Mandarin The change order of palatalization from Middle Chinese to modern Zhongyuan Mandarin The centralization of main vowel i after retroflex initials The change of front vowels after retroflex initials The loss of palatal medial -j- after retroflex initials The thirteen rhyme groups of the Děngyùn Tújīng A comparison of the rhyme groups of the Děngyùn Tújīng and of the Zhōngyuán Yīnyùn Medials, main vowels, and codas of the Děngyùn Tújīng Some examples of the zhī 知 and zhào 照 initials merger Some examples of the wēi 微 and yǐng 影 initials merger Some examples of the loss of the yí 疑 initial merger The rounded central vowel of the Děngyùn Tújīng The change of the rounded central vowel Examples of the jiān-tuán héliú The merging process of jiān (ts- group) and tuán (k- group) initials The complementary distribution of the tɕ-group and other initial groups An example of the new fǎnqiè method of the Yīnyùn Chǎnwēi A schematic illustration of the new fǎnqiè method of the Yīnyùn Chǎnwēi A correspondence table showing a comparison of the Yü-yen tzu-êrh chi with modern pinyin, and accompanying IPA The orthographical difference of the alveolar and retroflex initials The retroflex serials and palatal series represented by the same letters, ch and chʻ The Yü-yen tzǔ-êrh chi finals in comparison with Modern Standard Mandarin in pinyin spelling The Yü-yen tzǔ-êrh chi multiple readings of jiǎo 角, què 卻, lüè 略 The Yü-yen tzǔ-êrh chi multiple readings of jué 爵, què 卻, lüè 略 Labial initial or initial-less syllables with êng finals The tonal pitch of the four tones described as different intonations -rh as a suffix of the preceding syllable Some examples of characters with multiple readings

352 353 353 353 354 355 355 356 357 357 357 358 358 361 361 361 368 369 372 372 373 374 374 374 375 376 376 377

List of Tables

10.16 The Modern Standard Phonology, as represented by pinyin and IPA 10.17 The píng-zè voicing mergers with unvoiced syllables, as represented by p, ph, and b 10.18 A schematic representation of the geographical distribution of the devoicing patterns of the major dialect groups 10.19 A comparison of colloquial and literary pronunciations of Min Chinese

xxix

380 381 382 383

Foreword by William S-Y. Wang 王士元

Studies in the history of languages in the West may be dated to the famous lecture in Kolkata by William Jones, when he compared Sanskrit with the classical languages of Europe. Over the ensuing century and a half, the comparative method and internal reconstruction, originating from IndoEuropean studies, have been increasingly refined and applied to a great diversity of languages across the world. These achievements have become the pride of linguistics as a science. In contrast, language study in China took a different trajectory, in large part due to the millennia of centralized power structure of the dynastic tradition, maintained through its nonalphabetic writing system. Although a pioneering study of how words differed from region to region was reported several centuries before the Common Era (Fāngyán 方言 by Yáng Xióng 揚雄), scholarly attention has always focused exclusively on the Sinitic language of the Central Plain, first recorded in the Oracle Bone Inscriptions three millennia ago. The writing system that has evolved from inscriptions on bone and bronze has remained the exclusive method to represent the language, in spite of the deep influence of Buddhism early in the Common Era, with its original sutras written alphabetically in Sanskrit. Unfortunately, the diversity of hundreds of other languages that covered the Chinese landscape aroused little scientific interest across the dynasties. The Chinese tradition of language study has always centered on its writing system, dating back some two thousand years to a dictionary of over nine thousand characters (Shuō Wén Jiě Zì 說文解字), preferably called ‘sinograms.’ The spoken language, therefore, was studied through various ways of grouping these sinograms according to their pronunciation at the time. A primary purpose of these studies was for the analysis and composition of poetry. An early rime book of some fifteen centuries ago first grouped the sinograms by their lexical tone, then by their initial consonant, and finally by the remainder of the syllable. (See Figure 1.1.) Several centuries after that, rime tables were developed wherein various phonetic parameters were introduced to describe the consonants and vowels in xxx

Foreword

xxxi

terms of their articulation and perception. (See Figure 1.2.) In this connection, one might note that the science of musical acoustics had flourished much earlier. People in China understood the relations between vibrating bodies and the pitches they produce by the middle of the first millennium before the Common Era – witness the ingenious design of the massive bronze bells that can resonate at two different frequencies according to where they are struck. By around AD 1600, a prince of the Ming dynasty (Zhū Zàiyù 朱載堉) had published the world’s first account of the twelve-tone equal temperament scale in terms of the twelfth root of two. Unfortunately, little of this acoustic knowledge carried over to language studies. The vocal tract was never analyzed as an acoustic tube, and experimental phonetics never developed indigenously. Philological studies in the Chinese tradition reached their zenith in the Qing dynasty, largely motivated by the desire to understand the pronunciations of the poems of the Shījīng 詩經, composed three millennia ago. Scholars began to distinguish types of evidence that support various types of conclusions concerning phonetic events that took place over these three millennia, as well as to invent many technical terms beyond the straightforward articulatory descriptors used in the rime tables. Some of these terms are traditional words given new phonetic meanings, such as yīn 陰 and yáng 陽, qīng 清 and zhuó 濁; others are more opaque, such as shè 攝, děng 等, zhuǎn 轉, chóngniǔ 重紐, and many others. Much information about language change that is of great interest to linguistics at large, beyond Chinese linguistics as a subfield, has not been accessible to international scholarship because of these terminological hurdles. For the first time, Professor Shen Zhongwei’s volume promises to change that. His discussion of the phonological history of Chinese draws upon the relevant philological literature in a most effective and lucid fashion, couched in the familiar framework of modern phonetics, so that the lay reader will not become disoriented by the forbidding terminology. This is an important contribution the volume makes toward integrating Chinese linguistic scholarship into international linguistics at large. By the same token, this volume additionally makes an important contribution in presenting the phonological research of many Chinese scholars, who have primarily published just in Chinese sources, and thus are not easily accessible internationally. In addition to the well-known works of Wáng Lì 王力 of Peking University, Shen also draws upon the contributions of less publicized papers of indigenous scholars such as Lù Zhìwéi 陸志韋, Mài Yún 麥耘, Pān Wùyún 潘悟雲, Yóu Rǔjié 游汝傑, and especially the insightful Zhèngzhāng Shàngfāng 鄭張尚芳. Integrated with Western publications, starting with the great Bernhard Karlgren, who introduced the comparative method to China, followed by works of Edwin Pulleyblank, Jerry Norman,

xxxii

Foreword

William Baxter, Laurent Sagart, and many others, this volume offers a more balanced account of the varied scholarship in this field. Another important feature of this volume is the attention it gives to the linguistic interactions in Northern China. Much of the discussion of language contact in China has centered on the South and Southwest, with TibetoBurman, Austro-Asiatic, and Austronesian. But such a focus neglects the important linguistic interactions in the North and Northeast. These interactions were especially prominent in the second millennium of the Common Era with the Khitans, Jurchens, Tanguts, Mongols, and Manchus. The latter two peoples overpowered and governed the Hans for many centuries in the Yuan and Qing dynasties respectively. In recent years, Shen has pioneered research in this neglected area with significant findings. Shen examines these interactions between Altaic and Sinitic in this volume, often providing source materials that greatly facilitate understanding of many unfamiliar issues. As examples, Figure 5.1 is a rubbing from the Dàozōng Āicè 道宗哀冊, an example of the Greater Script in the Khitan language. Figure 5.3 illustrates Tangut writing, both the phonetic and sinographic forms, from Zhǎng Zhōng Zhū 掌中珠. Most interestingly, Figure 6.2 is a comparison of the Tibetan script with the hP’ags-pa system that the Mongols used in the Yuan _ dynasty. One hopes that these source materials will fascinate the reader and lead them to ever deeper studies. Gathering these source materials in a succinct overview adds much to the value of the volume, not only for linguistic knowledge, but also toward understanding that complex segment of Chinese history at the beginning of the Second Millennium, at the collapse of the Song dynasty. Shen’s study of these materials has already significantly advanced our knowledge in dating the origin of Early Mandarin, which was actually earlier than the received date by more than a century. This breakthrough, which he achieved through investigations of language contact, has been amply documented in his other writings and summarized in this volume. All in all then, Professor Shen Zhongwei has produced a volume that at once presents a balanced account of what is currently known about the phonological history of Chinese and opens a window for future work in the area with an emphasis on language contact. Much fruitful fundamental research lies ahead! Very few languages in the world offer the kind of time depth and rich cultural history that is presented in this volume. It is a great pleasure for me to recommend the volume most highly, not only to readers in linguistics, but to all readers who are interested in a fascinating chapter of human history. William S-Y. Wang 王士元 Research Centre for Language, Cognition, and Neuroscience Hong Kong Polytechnic University

Preface

As a first attempt, writing a book with this title brings with it unique and varied difficulties. It is well-known that Chinese is a language that has an unbroken written history of more than three thousand years. Firstly, unlike the alphabetical spelling system of European languages, the Chinese writing system is logographic, and so it does not reflect the sound of one’s speech in any transparent way. The sounds of historical versions of Chinese thus must be reconstructed by using the knowledge of modern linguistics. Secondly, the quality of available source materials for different historical periods varies significantly. This inconsistency in quality directly relates to the reliability of various research results. It is necessary that before presenting and discussing the research done on Chinese historical phonology, I discuss in depth the quality of various source materials. I hope that a better understanding of the source materials available for different historical periods can help the reader to better understand the nature of Chinese historical phonological research work. To me this is a more objective way to judge reliability, which is often blurred by the more subjective interests and enthusiasm of researchers. Thirdly, Chinese history is long, but the importance of each period to reconstruction varies due to available materials. In this book, the phonological history is presented in a commonly accepted way, with major periods defined by their respective source materials, which show a significant difference from previous source materials. Lastly, it should also be pointed out that not only are the main source materials in Chinese, but the vast majority of studies and related terminologies are also in Chinese. This is probably one of the main reasons why Chinese scholarship is not well known to the scholars in the West, even to those who are involved in the study of phonological history. Many of the traditional terminologies represent how phonological studies were carried out throughout history when there were no available phonetically transparent tools such as an alphabet. It is quite natural that Chinese phonological research historically centered on phonological categories rather than phonetic values, which is almost a given in an alphabetical writing.

xxxiii

xxxiv

Preface

I began to work on this book in 2010 at the suggestion of Professor Alain Peyraube of the Centre de recherches linguistiques sur l’Asie orientale (CRLAO) when we met during the 18th Annual Conference of the International Association of Chinese Linguistics (IACL-18) held at Harvard University. The original plan was to jointly write a book titled A History of the Chinese Language. Due to a difference in the progress of our writings, we decided to write separate books instead. My part is the book you are currently reading, and Professor Peyraube’s part will be another book. I appreciate Professor Peyraube’s invitation, otherwise I would not have had the courage to take on such an ambitious project. During the course of writing the present book, I received help and encouragement from many colleagues and friends. Without them, finishing this book project would have been a lot more difficult. For their professional help I would like to express my special thanks to Professor William S-Y. Wang at Hong Kong Polytechnic University for his comments, Professors Sūn Bójūn and Má Xiǎofāng at the Chinese National Academy of Social Sciences for their assistance with the input of special fonts, and the anonymous reviewer for the detailed comments and suggestions, which provide an important perspective from potential readers. Finally, I would like to thank my assistant for this book project, Mr. Jack Rabinovitch, who provided valuable assistance in many aspects, including making tables and indexing, as well as reading and editing this multi-script linguistic manuscript. Of course, it is needless to say, all the views expressed and the errors that remain in this book are entirely the responsibility of the author. This book is an attempt to introduce the scholarship of Chinese researchers to the West, where English is an academic lingua franca. In writing I have tried to keep in mind both the general phonologist and specialist of Chinese studies, as well as both beginners and experts. Due to the scale of Chinese history and the difficulty of these topics, many of the research results in this book are based on other scholars’ works. However, inclusion does not necessarily mean I am totally in agreement with the viewpoints of the works presented in this book. This book should be viewed as a beginning rather than the end of the representation of the historical phonology of Chinese. As a Chinese proverb says, this book is just an effort of pāo zhuān yǐn yù 抛磚引玉 ‘casting a brick to attract jade.’ I sincerely hope that more linguistic research results achieved by Chinese scholars will be introduced in English and will enrich the general discussion and understanding of historical linguistics.

Major Periods of Chinese History

Period

Years

Capital city

Shāng dynasty Western Zhōu dynasty Eastern Zhōu dynasty Spring and Autumn period Warring States period Qín dynasty Hàn dynasty Western Hàn dynasty Eastern Hàn dynasty Three Kingdoms period Jìn dynasty Western Jìn dynasty Eastern Jìn dynasty Northern and Southern dynasties Suí dynasty

1556–1046 BC 1045–771 BC 770–255 BC 771–476 BC 476–221 BC 221–206 BC 206 BC–AD 220 206 BC–AD 9 AD 25–220 AD 220–280 AD 266–420 AD 266–316 AD 317–420 AD 420–589 AD 581–618

Yin (Anyang) Fenghao (near Xi’an) Luoyi (Luoyang)

Táng dynasty Five Dynasties and Ten Kingdoms period Sòng dynasty Northern Sòng dynasty Southern Sòng dynasty Liáo dynasty

AD 618–907 AD 907–960/979 AD 960–1279 AD 960–1127 AD 1127–1279 AD 907–1125

Jīn dynasty

AD 1115–1234

Western Xià dynasty Yuán dynasty Míng dynasty Qīng dynasty Republic of China People’s Republic of China

AD 1038–1227 AD 1279–1368 AD 1368–1644 AD 1644–1912 AD 1912–1949 AD 1949–present

a

Xianyang (near Xi’an) Chang’an Luoyang

Luoyang Jiankang (Nanjing) Daxing (Xi’an) Luoyang Chang’an (Xi’an)

Bianjing (Kaifeng) Lin’an (Hangzhou) Shangjing (Linhuang) Nanjing (Beijing)a Zhongzhou (Beijing), Kaifeng Xingqing (Yinchuan) Dadu (Beijing) Nanjing, Beijing Beijing Nanjing Beijing

Shangjing (Linhuang) was ranked the first of five capitals of Liao. The other four capitals were Nanjing (today’s Beijing), Dongjing (Liaoyang), Xijing (Datong), and Zhongjing (Dading, today’s Ningcheng).

xxxv

Locations of Ancient Capitals of China

Beijing Yinchuan Anyang Xi’an Luoyang

Kaifeng Zhengzhou Nanjing Hangzhou

xxxvi

Part I

The Keys to Traditional Phonology The following introductory chapter mainly provides two kinds of information. One is the explanation of some basic terminologies used in the traditional study of Chinese phonology. The other is various source materials that are used in the study of Chinese historical phonology. These two kinds of information are very much interrelated. Due to the logographic nature of the Chinese writing system, the study of a historical sound system is a rather difficult task. Despite such a difficulty, Chinese scholars have been diligently analyzing the phonological system of Chinese for more than fifteen hundred years. It is not difficult to imagine all the problems that arise from using such a phonetically opaque writing system to analyze its phonology. Traditional phonology is considered one of the most difficult academic disciplines and is often called juéxué 絕學 ‘unique and esoteric knowledge’ by its students. However, significant achievements have been made. As early as the third century, a special phonetic notation method called fǎnqiè 反切 was in use. This method uses two Chinese characters to provide the phonetic information of the initial and the final as well as the tone for the character in question. In the sixth century the initial consonants had been worked out. Each initial was represented by one Chinese character that shares the same initial consonant. The most well-known list of Middle Chinese initials is the thirty-six initial characters (refer to Section 1.4.1). For the purpose of poetry composition, rhyme dictionaries, such as the Qièyùn 切韻 of 601, were compiled. The rhyme dictionaries actually provide a very thorough phonological classification for all Chinese monosyllabic words. The tones, rhyming parts, and initials of the Chinese syllables are well recognized and analyzed. Chinese phonology made another significant achievement in the form of rhyme tables, such as the Yùnjìng 韻鏡 of the twelfth century (or earlier). Rhyme tables were pure phonological analyses. By using tables, the initials and rhymes are systematically organized into rows and columns. The initials are classified according to their place of articulation and manner of articulation; the finals are classified according to their medials and main vowels. The accuracy of the analyses involved still amazes modern scholars. In the long history of Chinese scholarship, the achievements of phonological studies are represented by a large number of terminologies. Since all these 1

2

The Keys to Traditional Phonology

achievements were made in the absence of an alphabetical spelling system, what has been achieved by and large is the collection of categorical information, such as how many tones, how many initials, and the different categories of finals a given dialect or standard has. There are systematic phonetic transcriptions, such as the hP’ags-pa spelling system of the thirteenth century _ Latin alphabet transcription by Nicolas Trigault (refer to Chapter 6) and the (1577–1628), a Belgian/French missionary, in the early seventeenth century (refer to Section 9.5). But such phonetic information did not become an accepted part of traditional phonology until the twentieth century. The terminologies are the keys to understanding traditional phonology. They are frequently used in the phonological studies for all periods – Old Chinese, Middle Chinese, Old Mandarin, and even Modern Mandarin. Without a good understanding of the foundational terminologies of traditional phonological studies, the study of Chinese phonology would be quite impossible. The recently published Encyclopedia of Chinese Language and Linguistics (Sybesma et al. 2016) is a good reference in addition to the introduction of this book for more information on these terminologies.

1

An Introduction to Chinese Historical Phonology

The study of the phonological history of Chinese is in essence a study of a history of phonological standards. Historically, the Chinese language is a set of not just variants in time but variants in space as well. But the purpose of this book is not to cover all the available historical variants in time and space, but to focus on the phonological standards of major historical periods. It is commonly accepted that there are two main traditions, one starting with the Qièyùn 切韻 of 601 and the other starting with the Zhōngyuán Yīnyùn 中原音韻 of 1324. According to the commonly accepted view (e.g. Wáng Lì [1957] 2004), the former marks the beginning of Middle Chinese and the latter marks the beginning of Old Mandarin. Both of these systems, as well as the reconstructed Old Chinese, represent ideal standards (“literary language”) and are more or less composite in nature. While it is accepted that these standards must be based on real phonological systems, it would be a mistake to consider them as representing any single phonological system; until modern times, phonological standards were never based strictly on a single dialect. For social, political, and cultural reasons the dialects of the capitals always become the base dialect of the standard pronunciation. Throughout history, Luoyang, Xi’an, Beijing, and Nanjing were the capitals for various dynasties. The phonological systems of standard Chinese thus have a close relationship with these capital dialects. Historical phonology is discovery-driven. Research and reconstruction of Chinese at different periods of time depends on data given during and before the time period being researched. Data is compared to see when certain phonological features emerged in the written record in order to pinpoint where and when they emerged in history, with models of successive innovative features and sound changes based on the knowledge of older materials. Because of this nature of relativity to the past, “innovative” itself becomes a relative concept. Without comparison against older data, every phonological feature of a given time period or dialect would be a “new” feature. Because Old Chinese is the furthest back that researchers have been able to reconstruct so far, the phonology of Old Chinese is a collection of phonological features that are considered the “earliest” until older data becomes available. The 3

4

The Keys to Traditional Phonology

identification of innovative features is much improved after the Qièyùn, which provides a phonological system with quite clear information concerning time and space. Because of the long and unbroken written history of Chinese having lasted for more than thirty-five hundred years, many historical documents can be used as source materials for the study of phonological history. These source materials, systematic or not, are more reliable than any reconstructed results. For the Middle Chinese of the sixth century, the phonological categories of an entire system were recorded in the rhyme dictionary Qièyùn. Only the phonetic values need to be determined based on the modern dialects and ancient loans in non-Chinese languages. For this particular purpose the comparative method is applied. If there were no such rhyme dictionary, the comparative method could not have easily reconstructed the phonological categories as we see in the Qièyùn. No complete edition of the Qièyùn has survived to the present day, and so much of the corpus of study on Chinese linguistics is connected to the more readily available Guǎngyùn 廣韻 compiled and edited by Chén Péngnián 陳彭年 and Qiū Yōng 邱雍 in 1008. The Guǎngyùn was the most common rhyme dictionary for Song poetry, and in regard to Chinese phonological reconstruction it is one of the most important texts, although it is ultimately based on and secondary to the Qièyùn. For the later period of Old Mandarin of the thirteenth century, a systematic record of both phonological categories and their phonetic values can be found in a phonological work entitled Ménggǔ Zìyùn 蒙古字韻, in which Chinese characters are listed according to phonological categories and each syllable has its phonetic transcription in the hP’ags-pa script. In this case there is no need for the comparative method. In _the study of Old Chinese, because of poetry rhyming, the phonetic information of Chinese characters, and the categorical information of the Qièyùn, the phonological categories can be worked out. The need for the comparative method is again limited to the reconstruction of phonetic values. 1.1

Defining Chinese

The territory of modern China is the result of more recent changes in history. All the people living in China today are “Chinese” by definition; among them the Han ethnic group make up an overwhelming majority, with non-Han people as many smaller minorities. These non-Han peoples have different cultures, histories, and civilizations, as well as languages, both from each other and from the Han. Historically many of them probably are indigenous to the areas they currently inhabit. So, in a more precise sense, Chinese people are not the same as the Han people, and the languages of China are not the

An Introduction to Chinese Historical Phonology

5

same as the Chinese (Han) language (in Chinese, Hànyǔ 漢語, literally ‘Han language’). Within the boundaries of modern China, there are many dozens – fifty-five according to the Chinese government – non-Han peoples. Thus, “in a world neatly divided into nations and states, these people are Chinese – and yet they are not Chinese” (Ramsey 1987: 148). Modern China is called Zhōngguó 中國 ‘Middle Kingdom’ by Chinese people themselves. Historically, many dynasties with various names existed, such as Qin (221–206 BC), Han (206 BC–AD 220), Tang (AD 618–907), Song (AD 960–1279), Yuan (AD 1279–1368), Ming (AD 1368–1644), and Qing (AD 1644–1912). These dynasties may or may not have been ruled by the Han, but their territories have significant overlapping with that of modern China, and were also often composed of Han people as the majority of their population. In the time when different states coexisted, the Han were either the majority or a significant part of the populations. The Han language is called “Chinese” in English. As I have already said, it is confusing, because Chinese is not the language of China, rather the language of the Han. Referring to Chinese, there is even a problem of nomenclature (Norman 1988) in the Chinese language itself. There are many Chinese words referring to “Chinese.” In the academic publication it is Hànyǔ 漢語, while the Chinese script is Hànzì 漢字 ‘Han characters.’ As Jerry Norman says, “China is not an island; it is now and always has been surrounded by non-Chinese or, to be more precise, non-Han peoples. A non-Han people is by definition an ethnic group which uses a non-Han language” (Norman 1988: 16). Three thousand years ago, the southern part of modern China was inhabited by various non-Han groups, referred to as bǎiyuè 百越 ‘Hundred Yue.’ The word Yue is still used as the place-name referring to the southern area of China today (although with an alternative character, yuè 粵). Throughout history, “countless groups in South China gradually gave up their original ways of life and became Chinese. . . . more and more native groups took up Chinese dress, customs, social values, and of course the Chinese language” (Ramsay 1987: 34). When these people were in contact with the Han, many transferred their languages from non-Chinese to Chinese; it was inevitable that their native language would leave traces in the Chinese they acquired. Thus, many dialects of south China have the features of the non-Han languages spoken in adjacent regions. This phenomenon could be the main force behind language change and the formation of Chinese dialects. But these linguistic influences or contributions are not well realized and are considered just the features of Chinese or Chinese dialects, largely to do with internal development. In this book, the word “Chinese” is used to refer to the Han language, or Hànyǔ. The history of Chinese is therefore, in a sense, the history of the Han language, or the history of the language used by the ancestors of the modern Han people.

6

The Keys to Traditional Phonology

1.2

The Nature of Phonological History

Language is realized in its carriers. A specific language is not inherited but acquired by a person through the contact of its users. Language change is thus a result of the language change within the language’s carriers. The histories and the cultures of the language carriers must be the bases for understanding language changes. Throughout history, people who spoke different languages were in contact, and through this contact these people learned new languages and, in some cases, switched languages. This very basic mechanism, which introduces many linguistic changes, is not the interest of the Western tradition of historical linguistics, which is instead represented by the tree model and its related methodology, including the very influential comparative method. The comparative method, if used correctly, can provide a way of tracing an ancestral language based on its descendants, usually modern languages. But this method only shows one origin of a language, modern or historical. Regardless of the number of languages, if they are related, they are traced back to one protolanguage. After all the languages are linked together, the diagram is in the form of a tree. The base or root of the tree is the protolanguage and the end nodes are the modern languages. This tree gives an account of the historical development and the relationship of related languages. But such a diagram is far from reality. Because this tree only demonstrates how closely the languages are related to each other, it ultimately does not provide ample explanation of why languages change, diversify, and split apart. A mistake often made by the students of historical linguistics is to confuse reconstructed language family trees with real history. Many equate the tree diagram to the historical development of modern languages. In this way, the comparative method is only useful to a certain extent; it is completely unable to recover the real history of a language. 1.2.1

Chinese Phonology through Written Record

Although the history of Chinese is continuous, the available materials by and large represent just a few points in its history. The phonological history of Chinese is inevitably discontinuous. Language varies not only in time but in space as well. The Chinese language must have had its geographical variations, so-called dialects, throughout its history. Because emphasis was always placed on a national standard, the phonology of dialects or nonstandard variants was not well recorded before the sixteenth century,1 when 1

During the Han dynasty, Yáng Xióng 揚雄 (53 BC–18 AD) wrote Yóuxuān Shǐzhě Juédàiyǔ Shìbié Guó Fāngyán 輶軒使者絕代語釋別國方言, often shortened as Fāngyán 方言, in which he recorded contemporary dialectal (and possibly foreign, due to the fact that no distinction between “foreign” and “dialectal” was made) words from various places throughout the Han Empire.

An Introduction to Chinese Historical Phonology

7

recording local dialects came to be within the interest of some of the literati. Limited by the available source materials, geographical variants in the early times of Chinese are therefore quite impossible to retrieve. But this limitation of materials should not be interpreted as evidence of a lack of dialectal variations throughout history. The various standards of a language throughout history must also be understood as the written standard instead of the spoken standard. All standards were established for literary purposes, mainly for the sake of rhyming in the various poetry compositions of the time. Until modern times, while there was governmental effort to establish the standard for the written language, there was no effort made by the government to promote a spoken standard. In comparison with the written language, systematic information about the spoken language is very rare. Because of its logographic nature, written Chinese can only provide categorical information and no direct information about pronunciation. Because of the phonetic nature of their writing systems, only foreign transcriptions can reveal the phonetic values of Chinese logographs. There is no question that the phonetic information about the spoken language can help us better understand various aspects of the written language, such as whether some phonological categories were artificially preserved in the rhyming dictionaries and what the variations of individuals’ pronunciation were, besides the phonetic values of the known phonological categories. As will be seen in Chapter 8, the transcriptions of Chinese in the Old Persian script can be used to show the real pronunciation of the standard Chinese of the time, which can be realized in significantly different ways. The spoken language of individuals was never required to be strictly standard until recently, when the advent of long-range communication and recording made standardization practical, and the 1950s’ new language policy within the People’s Republic of China was enforced (P. Chen 1999: 24). The main function of language is for communication, and so when this function can be achieved, the amount of detail of the standard pronunciation that is adopted by a speaker will ultimately vary depending on the linguistic circumstance of said individual. 1.2.2

Phonological Categories in Logographic Systems

From its initial creation, the Qièyùn 切韻, as well as most of its successors, existed as a record of phonological categories only. This system was never intended to provide information concerning phonetic values. Because of the logographic nature of the Chinese writing system and the numerous dialectal variations present even at the time of the Qièyùn, no phonetic information can really be attached to such a system. Because it is categorical, it can exist beyond the limitations of time and space. The status of the written standard Chinese phonological system as categorical is one that has existed and been

8

The Keys to Traditional Phonology

maintained by Chinese culture for centuries; for more than fifteen hundred years the Qièyùn system has undergone several revisions, but its categorical nature has never changed. Such a supratemporal and supradialectal categorical system in fact worked out quite ideally with the writing system and the geographical variations of the Chinese language. The result of having specific phonological categories in a poetic standard was such that a poet from Beijing could write in correspondence with a poet from Guangzhou, and they would be able to notice the intricate rhyme schemes of each other’s writing without ever hearing their mutually unintelligible dialects. Ultimately, the phonetic values were never necessary information for the Chinese people to compose and enjoy poetry in its traditional formats. On the flip side, there is no phonetic information about historical changes and geographical variations that can be learned from this tradition. Such a categorical system was a literary norm in Chinese culture at the time, but it is unsatisfactory as information regarding the phonological system for modern linguists, especially those scholars who are used to more phonetically transparent alphabetical spellings. It is quite natural that the reconstruction of the phonetic values started by Western scholars when they encountered such a categorical system – for example, the studies done by the French linguist Henri Maspéro (1883–1945) and by the Swedish linguist Bernhard Karlgren (1889–1978) – represent the effort of identifying the phonetic values of traditional categories. 1.2.3

Terminology of Traditional Chinese Phonology

In the study of Middle Chinese phonology, many traditional terminologies are used to refer to the classification of the tones, initials, and finals. These phonological terms can be found in two very commonly used Middle Chinese phonological references: the Gǔjīn Zìyīn Duìzhào Shǒucè 古今字音 對照手冊 ‘Handbook of the Ancient and Modern Pronunciations of Characters’ by Dīng Shēngshù 丁聲樹 and the Fāngyán Diàochá Zìbiǎo 方言調查字表 ‘List of Characters for the Investigation of Dialects’ by the Zhōngguó Shèhuì Kēxuéyuàn Yǔyán Yánjiūsuǒ 中國社會科學院語言硏究所 ‘Institute of Language, Chinese Academy of Social Sciences.’ In these two references, the information of Middle Chinese is given in terms of traditional terminologies. The Fāngyán Diàochá Zìbiǎo lists commonly used characters, about 3,700 of them, according to phonological categories of Middle Chinese in a format of tables. The Gǔjīn Zìyīn Duìzhào Shǒucè lists the characters according to modern pronunciation with the information of Middle Chinese, such as the fǎnqiè of the Guǎngyùn and the categorical labels of each character according to Middle Chinese phonology (refer to Section 1.4 for more details). For example, for the pinyin spelling bāng, bāng 幫 is the first character of this pronunciation, bó páng qiē 博旁切 is the fǎnqiè of the Guǎngyùn, and dàng

An Introduction to Chinese Historical Phonology

9

kāi yī píng táng bāng 宕開一平唐幫 denotes the categorical labels of this character according to Middle Chinese phonology. These terminologies are the keys to understanding Middle Chinese phonology, the phonological systems both before and after it, as well as modern Chinese dialects. Although these terminologies are often used and thought of as the categories of Middle Chinese phonology, many of them are not from the Qièyùn, the standard reference of Middle Chinese, itself, but are actually from various rhyme tables produced many centuries after the Qièyùn. The most frequently used ones include: For initials: For finals:

For tones:

The thirty-six initials (sānshíliù zìmǔ 三十六字母) (Section 1.4.1); the qīngzhuó 清濁 system (Section 1.4.2) The shè 攝 ‘rhyme group’ (Section 1.4.9); děng 等 ‘division’ (Section 1.4.4); kāihé 開合 (Section 1.4.5); yīnshēng yùn 陰聲韻, yángshēng yùn 陽聲韻, and rùshēng yùn 入聲韻 (Section 1.4.10); chóngniǔ 重紐 and chóngyùn 重韻 (Section 1.4.6); wàizhuǎn 外轉 and nèizhuǎn 內轉 (Section 1.4.8) The sìshēng 四聲 (Section 1.3.4); yīndiào 陰調, and yángdiào 陽調 (Section 1.4.3)

To understand these terms, the syllable structure of Chinese itself must first be made clear. This structure has changed over time, but just as much of Chinese historical phonology revolves around the Qièyùn, so do modern conceptions of Chinese syllable structure rely on the division and categorization of the Qièyùn and its successors. 1.2.4

Syllable Structure

From the very beginning of recorded history in China, ancient written documents and literature works overwelmingly indicate that the Chinese writing system was always monosyllabic, although it is probable that the syllable structure of earlier stages was more complicated (Baxter & Sagart 2014; see Section 2.3.1 and Section 2.3.5.1). Middle Chinese words themselves tended to be monosyllabic more often than not, although there are of course exceptions to this, one such famous one being húdié 蝴蝶 ‘butterfly.’ Traditionally a Chinese syllable is analyzed into three basic structural components: initials, finals, and tones. In recent years, some scholars who work on Old Chinese phonology have proposed that Chinese syllables can be structurally more complicated, a minor syllable with fewer structural elements can appear in front of the main syllable. (Refer to Section 2.3.1 and Section 2.3.5.1 for more information). Because of the monosyllabic nature of written Chinese, historical phonology in general consists of a phonotactic analysis of syllables, a survey of the number and types of initials, finals, and their basic elements, tones, as well

10

The Keys to Traditional Phonology

Table 1.1 Additional syllable terms with their component segments Component segments Onset Codaa Rime Final

= = = =

IM E VE MVE

“Coda” and “ending” are synonymous when discussing Middle Chinese onward. a

as the rules governing the combination of these. At the syllabic level, the analyses will be on the segmental elements, initials, medials, main vowels, and codas, and the suprasegmental tones. From Middle Chinese on, the basic structural slots of a syllable are the initial, medial, main vowel, and ending; this is sequentially abbreviated as IMVE, with T representing tone; it is suprasegmental and thus placed above the syllable, unless otherwise noted: T IMVE In the discussion of phonology, the aspects of a syllable often include certain combinations of the aforementioned segments, each with their own name (see Table 1.1). Rime, as a combination of main vowel and coda, should not be confused with rhyme. Rhyme is a term whose definition changes depending on the source material, but tends to correspond to the rime and tone, with the medial often included as well. For example, the yáng 陽 rhyme of the Qièyùn is -jɐŋ with a píng tone. By using the terms in Table 1.1, it is structurally MVE and T, or a final and a tone. The reconstructions of Old Chinese involve more complex syllable structures, and more than one syllable is sometimes used to reconstruct the phonetic value of single characters (e.g. Baxter & Sagart 2014). In some Old Chinese reconstructions, the suprasegmental tone is eliminated and reconstructed as segmental elements. This is explained alongside the discussion of minor syllables in greater depth in Section 2.3.1 and Section 2.3.5.1. 1.2.5

Nonsyllabic Phonological Characteristics

Any phonological characteristics that transcend individual morphemes or syllable structure are not in the main interests of most scholars of historical linguistics. An example of a nonsyllabic phonological characteristic is tone sandhi, a well-known phonological phenomenon that exists in many modern

An Introduction to Chinese Historical Phonology

11

Table 1.2 Tone sandhi in modern Beijing dialect

好山 好人 好酒 好看

hǎoshān hǎorén hǎojiǔ hǎokàn

‘good mountain’ ‘good person’ ‘good wine’ ‘good looking’

Underlying tone

Realized tone

3–1 3–2 3–3 3–4

3–1 3–2 2–3 3–4

Chinese dialects. Tone sandhi is a phonological rule that shifts the tone of a syllable based on the tone of the surrounding syllables. In the modern Beijing dialect, the most obvious tone sandhi is that if there are two third-tone syllables in sequence, the first one will be realized as a second tone. For example, the hǎo 好 in hǎoshān 好山, hǎorén 好人, and hǎokàn 好看 is realized as its underlying third tone because it is succeeded by a non-third (first, second, or fourth) tone syllable; whereas the hǎo 好 in hǎojiǔ 好酒 is realized as a second tone because of the tone sandhi rule in the condition of preceding a third-tone syllable, jiǔ 酒 (see Table 1.2). Tone sandhi phenomena are very complex in the southern dialects and vary from place to place (M. Chen 2000). In any given Wu speaking area, each Wu dialect may have its own unique tone sandhi system (Qián 1992). The fact that syllables change their tones in words and phrases is an across-the-board phenomenon and could show in all tonal categories for any given dialect. It is not clear how such an inter-syllabic phenomenon existed throughout Chinese history. The only information regarding tone sandhi before the late Qing dynasty is found in the Korean transcriptions of the Chinese syllables of the Ming dynasty (see Section 9.4.3) (Mei 1977). 1.3

Yùnshū 韻書‘Rhyme Dictionaries’

The rhyme dictionaries, or yùnshū 韻書 ‘rhyme book’ in Chinese, contain the most systematic information about Chinese historical phonology. There are many different forms of rhyme dictionaries, but a common feature is that these dictionaries arrange all listed characters according to the rhyming part of a syllable, the tone, the main vowel, and the coda. However, rhymes of a syllable cannot provide information about their initials.2 2

Fǎnqiè 反切 spellings use two Chinese syllables to provide the information concerning pronunciation of the initial and the rhyme parts of the syllable in question. Systematic studies of the fǎnqiè spellings (refer to Section 1.3.2, Section 3.1.1, Section 3.4.3, and Section 10.1.2) have also revealed important phonological information, which is more complete and detailed than the information about the rhyming categories. If the fǎnqiè spellings are provided, information about the initials can be retrieved.

12

The Keys to Traditional Phonology

Among all the surviving rhyme dictionaries, the Qièyùn 切韻, published in 601, is the earliest. Most of the later rhyme dictionaries are either revisions or expansions of the Qièyùn, or are imitations of the format of the Qièyùn. The revisions of the Qièyùn tend to only affect the splitting or merging of existing rhyming categories. The studies of both earlier and later periods always refer to the phonological categories of the Qièyùn, using its system to understand the phonological changes before and after it. This anachronistic approach to studying Old Chinese inevitably means that in order to understand the work behind Old Chinese, one must understand the terminology of Middle Chinese, and so it is necessary to detail these Middle Chinese terms in the introduction. The phonological history of Chinese heavily relies on the phonological standards of a few historical periods. It is thus important to understand the nature of phonological standards in the context of Chinese culture. Based on the notes in the Qièyùn, it is quite clear that a phonological standard is not a live phonological system in a certain time and at a certain location. It was or had to be based on the dialect of a cultural center, such as the national capital, but the intended phonological information cannot even be closely tied to this central dialect, because rhyme books are ultimately made to be supradialectal (so that people in many places could use them) and supratemporal (so that they could be used to reference previous works). It will be discussed in great detail in Section 3.1.4 and Section 3.1.5 that it is and was always the intention of the authors of rhyme books to include all the retrievable phonological contrast known to them. The author of the Qièyùn, Lù Fǎyán 陸法言, seemed to have consulted all the rhyming dictionaries available. The criteria for what determined syllables in the same rhyme and what separated them was clear when he was compiling the dictionary. In the category of rhyme books, there are other kinds of phonological works that are quite different from the Qièyùn. For example, the Ménggǔ Zìyùn 蒙古 字韻 of the thirteenth century is basically a comparative list of Chinese characters and their corresponding spellings in the hP’ags-pa script (refer to Chapter 6). The Zhōngyuán Yīnyùn 中原音韻 of the_ fourteenth century is a handbook for the composers of operas and poems. It contains two parts: part one is a list of Chinese characters, arranged according to the rhyming parts of the syllables, and part two is the alternative pronunciations and examples of compositions (refer to Section 7.2). Neither of these works are dictionaries; they only list the characters with no glossaries. 1.3.1

The Qièyùn 切韻 (601)

The Qièyùn is the earliest surviving record of an entire phonology; as such, it is unquestionably the centerpiece of historical Chinese linguistics as well as the reference point for phonologies both before and after it. Only fragments of the

An Introduction to Chinese Historical Phonology

13

Figure 1.1 A page from the Kānmiù Bǔquē Qièyùn 刊謬補缺切韻 (1968) by Wáng Rénxù 王仁昫 (Lóng Yǔchún’s (龍宇純) annotation version (1968)), which shows the second part of the index of píng 平 tone rhymes and the beginning part of the homophonic groups of the dōng 東 rhyme.

original Qièyùn survive, therefore the most commonly used documents for reconstruction are the revisions of the Qièyùn, namely the Kānmiù Bǔquē Qièyùn 刊謬補缺切韻 of 706 (Figure 1.1) and the official publication Guǎngyùn 廣韻 of 1008. Because of the incomparable importance of this book, its format and nature must be discussed in detail. The Qièyùn is a list of homophonic characters grouped into yùn 韻 ‘rhymes.’ The Qièyùn is first divided into four parts according to the tones: píng 平,

14

The Keys to Traditional Phonology

shǎng 上, qù 去, and rù 入. Within each tone, there are several yùn, which further categorize the characters by homophonous rimes; that is, these yùn groups are listed according to their main vowels and codas. For example, píng tone syllables start with the dōng 東 -(j)uŋ rhyme, followed by the dōng 冬 -woŋ rhyme, the zhōng 鍾 -joŋ rhyme, the jiāng 江 -ɰɔŋ rhyme, etc. The rhymes are numbered according to their order, with a number used in front of the rhyme head, e.g. yī-dōng 一東 ‘one-dōng,’ èr-dōng 二冬 ‘two-dōng,’ sānzhōng 三鍾 ‘three-zhōng,’ sì-jiāng 四江 ‘four-jiāng,’ etc. Within these groups with the same main vowel and coda, characters are then further arranged according to their initial and medial, with homophones appearing in groups. Within each rhyme, words are further divided into groups; the words in each group are homophones, thus these groups are called homophonic groups. Different homophonic groups of the same rhyme differ from each other in terms of the onset part of the syllable (the initial or the combination of initial and medial). For example, in the dōng 東 -(j)uŋ rhyme, the first homophonic group contains two homophonic words, and the second group contains twenty homophonic words. These two groups differ in their initial consonants: according to reconstructed values, the initial of the first group is t- and the initial of the second group is d-. Each character has a definition, and the first character of each homophonic group also has fǎnqiè to provide pronunciation and a note of how many homophonic characters it has (see Section 1.3.2). The examples shown in Table 1.3 are from the Kānmiù Bǔquē Qièyùn – only the characters, fǎnqiè, and number of homophones are provided. In total, there are 195 rhymes (see Table 1.4). The total number of rhymes is 54, 52, 57, and 32 for píng, shǎng, qù, and rù tones, respectively. The rù

Table 1.3 Illustration of the Qièyùn structure, as excerpted from the Kānmiù Bǔquē Qièyùn 東 . . . 德紅反二凍 同 . . . 徒紅反二十一童僮銅桐峒 硐筒 瞳 罿潼筩犝橦曈衕烔 中 . . . 陟隆反四衷忠 ... This can be further analyzed as: Group

fǎnqiè

Number (hànzì)

Numbera

Homophones

東 同

德紅 徒红

二 二十一

2 21



陟隆



4

凍 童僮銅桐峒 硐筒 瞳 罿潼筩犝橦曈衕 烔 衷忠

a

The number of homophones in the list is one less than the number reported. This is because the group name is considered in the list.

An Introduction to Chinese Historical Phonology

15

Table 1.4 The representative characters of each rhyme of the Qièyùn píng

shǎng



1 2 3 4 5 6 7 8 9 10 11 12

東 冬 鍾 江 支 脂 之 微 魚 虞 模 齊

dōng dōng zhōng jiāng zhī zhī zhī wēi yú yú mó qí

1



dǒng

2 3 4 5 6 7 8 9 10 11

腫 講 紙 旨 止 尾 語 麌 姥 薺

zhǒng jiǎng zhǐ zhǐ zhǐ wěi yǔ yǔ mǔ jì

13 14

佳 皆

jiā jiē

12 13

蟹 駭

xiè hài

15 16

灰 咍

huī hāi

14 15

賄 海

huǐ hǎi

17 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

真 臻 文 殷 元 魂 痕 寒 刪 山 先 仙 蕭 宵 肴 豪 歌 麻 覃 談 陽 唐 庚 耕 清 青

zhēn zhēn wén yīn yuán hún hén hán shān shān xiān xiān xiāo xiāo yáo háo gē má tán tán yáng táng gēng gēng qīng qīng

16



17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40

吻 隱 阮 混 佷 旱 潸 産 銑 獮 筱 小 巧 晧 哿 馬 感 敢 養 蕩 梗 耿 靜 迥



zhěn

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21

送 宋 用 絳 寘 至 志 未 御 遇 暮 霽 祭 泰 卦 怪 夬 隊 代 廢 震

sòng sòng yòng jiàng zhì zhì zhì wèi yù yù mù jì jì tài guà guài guài duì dài fèi zhèn

wěn yǐn ruǎn hùn hěn hàn shān chǎn xiǎn xiǎn xiǎo xiǎo qiǎo hào gě mǎ gǎn gǎn yǎng dàng gěng gěng jìng jiǒng

22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45

問 焮 願 慁 恨 翰 諫 襇 霰 線 嘯 笑 效 号 箇 禡 勘 闞 漾 宕 敬 諍 勁 徑

wèn xìn yuàn hùn hèn hàn jiàng jiàn xiàn xiàn xiào xiào xiào hào gè mà kàn kàn yàng dàng jìng zhèng jìng jìng

1 2 3 4

屋 沃 燭 覺

wū wò zhú jué

5 7 6 8 9 10

質 櫛 物 迄 月 沒

zhì zhì wù qì yuè mò

11 13 12 14 15

末 鎋 黠 屑 薛

mò xiá xiá xuē xuē

20 21 27 28 19 18 17 16

合 盍 藥 鐸 陌 麥 昔 錫

hé hé yào duó mò mài xī xī

16

The Keys to Traditional Phonology

Table 1.4 (cont.) píng 43 44 45 46 47 48 49 50 51 52 53 54

shǎng 尤 侯 幽 侵 鹽 添 蒸 登 咸 銜 嚴 凡

yóu hóu yōu qīn yán tiān zhēng dēng xián xián yán fán

41 42 43 44 45 46 47 48 49 50 51 52

qù 有 厚 黝 寑 琰 忝 拯 等 豏 檻 广 范

yǒu hòu yǒu qǐn yǎn tiǎn zhěng děng xiàn jiàn yǎn fàn

46 47 48 49 50 51 52 53 54 55 56 57

rù 宥 候 幼 沁 豔 㮇 證 嶝 陷 鑑 嚴 梵

yòu hòu yòu qìn yàn tiàn zhèng dèng xiàn jiàn tàn fàn

26 24 25 29 30 22 23 31 32

緝 葉 怗 職 德 洽 狎 業 乏

qī yè tiē zhí dé qià xiá yè fá

tone syllables are arranged according to the corresponding píng, shǎng, and qù tones,3 with the number indicating the original Qièyùn order. Two types of phonological information in rhyme dictionaries can be used: rhyme categories and fǎnqiè spellings. Other material such as rhyme tables, philological works, and poetry rhyming also provide important categorical information for Middle Chinese phonology, though again these source materials are all ultimately based on the Qièyùn. Having a phonological work provides a tremendous advantage as tonal categories, rhyme categories, and the relationship between different types of syllables all become obvious. In addition, by analyzing the fǎnqiè spellings, the initial system can be worked out with a high degree of certainty. The Qièyùn has irreplaceable value in the understanding of Middle Chinese phonology, not only in regards to the historical periods before and after it, but also for the entire phonological history of Chinese. Such information simply cannot be obtained from poetry rhyming and foreign transcriptions. The phonological system of the Qièyùn can also provide the change conditions for later historical periods as well as for modern dialects (Cheng & Wang 1971; Wang & Cheng 1987). 1.3.2

Fǎnqiè 反切

In the Qièyùn and its revisions, fǎnqiè is provided as a phonetic notation for every homophonic group. Being systematically used, it constitutes an 3

This arrangement of “corresponding” syllables is called yùnxì 韻系 ‘rhyme series,’ and though not a traditional term, it is useful in the analysis of traditional Chinese phonology. It is discussed in greater depth in Section 1.3.5.

An Introduction to Chinese Historical Phonology

17

independent piece of information to study the Qièyùn system. Chinese characters are logographic, and are not spelled phonetically by letter, so in order to represent the phonetics of the character, it can only be shown by comparison to other characters. The fǎnqiè method uses two Chinese characters to indicate the pronunciation of the target character. The fǎnqiè formula is designed based on a binary analysis of the Chinese syllable in which each syllable is divided into two parts, a shēngmǔ 聲母 or shēng 聲 ‘initial’ and a yùn 韻 ‘rhyme’ (the yùn also bears a tone). Two characters are selected as “spellers” to represent the initial and final parts of the character being described: the first character, shàngzì 上字 ‘upper speller,’ shares the same onset as the character in question, and the second character, xiàzì 下字 ‘lower speller,’ shares the same rhyme as the character in question (see Table 1.5). In order to use the fǎnqiè method to “spell” a character, (a) the pronunciation of a character is binarily analyzed into two parts, initial and rhyme, (b) an upper speller character that has the same initial as the character to be spelled is chosen, as well as a lower speller character that has the same rhyme as the character to be spelled, and (c) the initial of the upper speller and the rhyme of the lower speller are combined. See, for example (Table 1.6): 同, 徒紅切.4 Here the character to be spelled is tóng 同 (Middle Chinese: duŋ), the upper speller is tú 徒 (Middle Chinese: du), and the lower speller is hóng 紅 (Middle Table 1.5 The structure of fǎnqiè spellers Character to be spelled Upper speller Lower speller

Initial-a Initial-a (Initial-x)

Rhyme-a (Rhyme-x) Rhyme-a

Table 1.6 An example of the fǎnqiè system Hànzì

Initial

Final

Tone

Character to be spelled Upper speller Lower speller

同 徒 紅

ddɦ-

-uŋ -u -uŋ

píng píng píng

Upper speller Lower speller Character

徒 紅 同

d— d-

— -uŋ -uŋ

— píng píng

Note: The first set of characters are shown along with their tones, finals, and initials, whereas the second set shows only which parts of the syllable are carried over.

4

Modern Mandarin pronunciation: tóng, tú hóng qiè.

18

The Keys to Traditional Phonology

Chinese: ɦuŋ). The character fǎn 反 ‘turn over’ or qiè 切 ‘cut’ marks the fǎnqiè, and is always included in the texts, though which one (fǎn or qiè) differs between different texts. By adding the onset of the upper speller (d-) and the rhyme of the lower speller (-uŋ píng), the pronunciation of the character (duŋ píng) is “spelled out.” Although knowledge of the binary analysis of the spellers is required, fǎnqiè was the most efficient way for the Chinese writing system to provide phonetic information. Another advantage of fǎnqiè is that its nonphonetic nature also provides a supradialectal system applicable to different dialects.5 Chinese scholars have been working on the fǎnqiè system of the Qièyùn and other phonological works for centuries. Since Chén Lǐ’s 陳澧 (1810–1882) systematic analysis of the fǎnqiè of the Guǎngyùn, scholars have used his methods and improved the results to work out the initial and final systems of the Guǎngyùn. Many phonological details are distinguished in fǎnqiè spellings, including the chóngniǔ contrast. Based on Wáng Rénxù’s 王仁昫 Kānmiù Bǔquē Qièyùn, Lǐ Róng’s 李榮 book entitled Qièyùn Yīnxì 切韻音系 (1956) is a standard reference for fǎnqiè spellings. The seemingly simple fǎnqiè method can sometimes be complicated due to several factors. Firstly, these fǎnqiè spellings were not devised by one individual but rather collected from various preexisting phonological and philological works. Some spellings show clear inconsistencies and reflect different historical layers (see Table 1.7 and Table 1.8 for some examples). Secondly, the binary analysis is sometimes not easy to apply due to the structurally ambiguous medial, which can be considered a constituent of the onset (IM + VE), the rhyme (I + MVE), or even both (IM + MVE),6 although statistically the IM + MVE and I + MVE analyses are more frequently used. In the Qièyùn some fǎnqiè spellings are likely copied from previous works and therefore may not indicate an accurate pronunciation of even the standard of the time. Among the Middle Chinese syllable structural slots, all but the medial (M), can be occupied by only one tone, consonant, or vowel. The medial slot can have as many as three medial segments (see Table 1.9). It is not known if the medials went in a particular order, or if they co-occurred in some way. It is also unclear whether these three medials were uttered as one segment or a sequence of segments in actual pronunciation. For example, if medials -j- and -w- are combined as -ɥ-, the number of medials can be reduced from three to two.

5

6

The categorical information of the fǎnqiè tends to work across different dialects because the phonological system of dialects usually have quite good correspondence with the Qièyùn system. The differences usually exist in the phonetic realization rather than on a categorical level. This structure is used particularly consistently with Division-III characters, and is discussed more thoroughly in Section 3.5.4.1.

An Introduction to Chinese Historical Phonology

19

Table 1.7 The pronunciation of zhuāng 樁 ‘stake’ throughout history Old Chinese Character to be spelled Upper speller Lower speller Fǎnqiè result

樁 都 江

trooŋ t(a) (k)rooŋ t + rooŋ

Middle Chinese > > >

ʈɰɔŋ t(wo) (k)ɰɔŋ

Old Mandarin > > >

t + ɰɔŋ

Modern Standard Mandarin > > >

tʂwaŋ t(u) (k)jaŋ t + jaŋ

tʂwaŋ t(u) (tɕ)jaŋ t + jaŋ

Note: This fǎnqiè spelling works in Old Chinese, t- + -rooŋ. The changes of the initial and final in the characters zhuāng 樁 and jiāng 江 from Old Chinese to Middle Chinese (as shown in the Qièyùn) already made this fǎnqiè problematic.

Table 1.8 More examples of fǎnqiè spellings that are obsolete in Modern Standard Mandarin Middle Chinese 篇 家 方

piān jiā fāng

‘essay’ ‘home’ ‘square’

芳連切 古牙切 府良切

phjɛn kɰa pwjɐŋ

faŋ + ljɛn ku + ja fu + ljaŋ

> > >

Expected MSM

Actual MSM

fjɛn kja fjaŋ

phiɛn tɕia fɑŋ

Note: It is even more difficult to understand the fǎnqiè if only modern pronunciation is used. MSM = Modern Standard Mandarin.

Table 1.9 The basic syllable structure with the medial broken down into its component parts I M V E

M = w, ɰ, j

However, this is not the approach of many reconstructions. Instead each of the three medials is used as the labels of traditional categories. Different categorical combinations may have been realized differently phonetically; however, for the sake of reconstruction, the categorical combinations are used (see Table 1.10). 1.3.3

Zhíyīn 直音 Notation

Zhíyīn is a traditional way to provide phonetic information about the character in question (A) by using a homophonic character, A 音 B (A = B, yīn 音

20

The Keys to Traditional Phonology

Table 1.10 Categorical combinations realized differently phonetically Categorically wɰj wɰ ɰj wj

Phonetically > > > >

ɰɥ wɰ ɰj ɥ

‘sound (it sounds as)’). Such phonetic notations can reveal the information of sound change if A and B historically have different phonetic values (A 6¼ B) while a later rhyme book has them as homophones, e.g. róng yīn róng 融音容, the pronunciation of róng 融 ‘melt’ is the same as róng 容 ‘contain, allow.’ If two characters in a rhyme book are noted as zhíyīn, then the reconstruction of that rhyme book’s phonology must also have these two characters as homophones. However, this notation does not definitively state whether two characters are not homophones. That is, if two characters in a rhyme book are not noted as zhíyīn, that does not necessarily mean that those two characters are not homophonic in the phonology of that rhyme book. This method of phonetic notation is also mentioned as dúruò 讀若 ‘pronounced as’ in other ancient books.

1.3.4

Shēngdiào 聲調 ‘Tones’

Traditionally the four tones of Middle Chinese are referred to as píng 平 ‘level,’ shǎng 上 ‘rising,’ qù 去 ‘departing,’ and rù 入 ‘entering.’ These four tonal labels are from their own tonal categories and their meanings describe the tonal values. The píng is a level tone; shǎng is a rising tone; qù is a falling tone, and rù is a short tone. The rù tone is particularly distinct from the other three tones because of its historical stop consonant endings: -p, -t, -k (see Section 1.4.10), as well as the fact that the rù tone was considered a short tone. The píng tone was a long and level tone, where the other tones were not. In addition, it is possible that the frequency of píng tone characters in Middle Chinese was about equal to the other three tones combined; this half-and-half division, along with the level/non-level prosodic contrast, is quite natural for the requirement of poetry composition. Because of these special attributes of the píng and rù tones, there are terms to describe tones that are not in those categories, with non-píng tones termed zè 仄 and non-rù tones termed shū 舒 (see Table 1.11). Although ultimately synonymous with the rù tone, the term cù 促 is used to further highlight the contrast with shū tones.

An Introduction to Chinese Historical Phonology

21

Table 1.11 The four tones of Middle Chinese, with píng-zè and shū-cù tones demarcated píng-zè: shū-cù:

1.3.5

píng = píng, shū = píng, shǎng, and qù

zè = shǎng, qù, and rù cù = rù

Yùnxì 韻系 ‘Rhyme Series’

Two syllables that share the same final but not the same tone are not considered rhyming, but are rather categorized in the same yùnxì ‘rhyme series.’ For example, the dōngyùnxì 東韻系 or dōng 東 rhyme series includes the píng tone dōng 東 -uŋ rhyme, as well as the shǎng and qù rhymes with the same -uŋ final, dǒng 董 -uŋ and sòng 送 -uŋ. The rù tone endings (-p, -t, -k) do not exist in the other tones; however, their endings correspond with the shū tone (see Section 1.3.4) endings (-m, -n, -ŋ, respectively) in their place of articulation, and so rù tone rhymes are always placed in the yùnxì with the corresponding nasal ending. In the case of the dōng 東 rhyme series, this rù tone rhyme is wū 屋 -uk. Thus the dōng 東 rhyme series includes in full, the dōng 東 -uŋ, dǒng 董 -uŋ, sòng 送 -uŋ, and wū 屋 -uk rhymes. Because of this shū-cù correspondence, if the píng tone rhyme has a nasal ending, this rhyme series includes the corresponding píng, shǎng, qù, and rù rhymes. However, if the píng tone rhyme has a nonnasal ending, this rhyme series contains the corresponding píng, shǎng, and qù tones only, because there is no corresponding rù rhyme. See Table 1.12 for examples of rhyme series with both nasal and nonnasal píng tone endings, illustrating the inclusion of the rù tone in nasal píng tone rhyme series. The representative character of the píng tone rhyme is always used to represent the rhyme series; the exceptions to this are the four qù tone rhymes jì 祭, tài 泰, guài 夬, and fèi 廢, which do not have their píng, shǎng, and rù Table 1.12 Rhyme series with both nasal and nonnasal píng tone endings píng

shǎng





With nasal ending Táng 唐 rhyme series Wén 文 rhyme series

唐 -ɑŋ 文 -jun

蕩 -ɑŋ 吻 -jun

宕 -ɑŋ 問 -jun

鐸 -ɑk 物 -jut

With nonnasal ending Háo 豪 rhyme series Qí 齊 rhyme series

豪 -ɑw 齊 -ej

皓 -ɑw 薺 -ej

号 -ɑw 霽 -ej

22

The Keys to Traditional Phonology

counterparts, and their rhyme series is thus represented by their qù tone representative (refer to Section 2.3.3.5.4 and Section 3.5.8.1 for more detail). 1.4

Yùntú 韻圖 ‘Rhyme Tables (Rhyme Charts)’

Beginning at the end of the tenth century there is evidence that special efforts were made to analyze and classify syllable structures; this work resulted in the rhyme tables. These tables classify all syllables (not words) according to their initials, finals, and tones. The earliest evidence of this are the fragments of Shǒuwēn’s 守溫 (fl. sixth century) phonological works, in which some basic terminologies of later rhyme tables were mentioned, including the initial characters, divisions, and kāihé 開合. However, Shǒuwēn’s works themselves are not rhyme tables (Coblin 2006). In the Song dynasty, there was a significant development in phonological analysis, which is represented in the various rhyme tables, or yùntú 韻圖, of the time. Early examples are the Qīyīn Lüè 七音略, Yùnjìng 韻鏡, Qièyùn Zhǐzhǎngtú 切韻指掌圖, and Sìshēng Děngzǐ 四聲等子. The Qīyīn Lüè and the Yùnjìng are quite similar to each other, though they are different from the Qièyùn Zhǐzhǎngtú and the Sìshēng Děngzǐ (see Section 5.4.2 and Section 5.4.3). The format of the rhyme tables is quite straightforward and easy to read. Syllables are first separated into tables based on the similarities of their main vowels and codas. In each table the syllables are arranged according to their initials, finals, and tones. Different columns indicate different initials. Initials are classified according to the places of articulation, and then the manner of articulation. The places of articulation: chúnyīn 脣音: labial shéyīn 舌音: alveolar and retroflex (stops) yáyīn 牙音: velar chǐyīn 齒音: alveolar and postalveolar (affricates and fricatives) hóuyīn 喉音: laryngeal The manners of articulation:7 qīng 清: voiceless unaspirated stops and affricates, and voiceless fricatives cìqīng 次清: voiceless aspirated stops and affricates zhuó 濁: voiced stops and affricates, and voiced fricatives cìzhuó 次濁 (or qīngzhuó 清濁): nasals, laterals, approximants 7

This is aptly called the qīngzhuó system, and it is discussed in greater detail in Section 1.4.2.

An Introduction to Chinese Historical Phonology

23

The rows are first divided according to the four tones, píng, shǎng, qù, and rù, then further divided into the four děng 等 ‘division.’ The two leaves in Figure 1.2 show all the possible syllables of four rhyme groups. Indicated in the leftmost column are the labels of these four rhymes, dōng 東, dǒng 董, sòng 送, and wū 屋. These four rhymes are phonologically parallel, the first three, dōng 東, dǒng 董, and sòng 送, with the same final -uŋ but differing in tone. The fourth rhyme, wū 屋, has the final -uk, which has a velar stop ending instead of a velar nasal like the other three. The columns in Figure 1.2 indicate different initials. The four rows with separating lines indicate the different tones, píng, shǎng, qù, and rù. Each lineseparated row is divided into four smaller rows without separating lines. The traditional term of the division of the finals is called děng 等 ‘division’ or ‘grade.’ The four divisions are traditionally written using roman numerals I, II, III, and IV. The finals of each division are the same; for example, in the first row of Figure 1.2 dōng 東 and gōng 公 have the same final. The finals in different rows have different finals; for example, gōng 公 in the first row and gōng 弓 in the third row have different finals. The finals of different divisions belong to the same rhyme, which means that they share the same ending and the same or similar main vowels; whether they share the same or similar vowels depends on the specific rhyme. The initials in the same vertical column have the same manner of articulation and the same phonation type. The places of articulation can be either the same or just similar. For example, in the first column of Figure 1.2 all the initials have the exact same initial p-. However, for the alveolar stops, fricatives, and affricates, the places of articulation are not the same, but they are similar and historically related. Table 1.13 is an explanation of the shéyīn 舌音 (with four columns) and chǐyīn 齒音 (with five columns), showing allophonic variants conditioned by divisions. For example, in the fifth column of Figure 1.2 (from the right), the initials of the first four rows are t-, ʈ-, ʈ-, and t- (historically t-, tr-, tr-, and t-), respectively. Such an arrangement is made with phonological reasons, because these syllables are in complementary distribution. The syllables with t- are only in Divisions-I and -IV, and the syllables with ʈ- are only in Divisions-II and -III. This arrangement can also reduce the number of columns and make the tables more compact. The relationship between initials and finals with respect to complementary distribution goes further and may help to explain certain phenomena such as chóngniǔ 重紐 syllables. This relationship is discussed in more detail in Section 3.3. The well-known thirty-six initial characters (see Section 1.4.1) emerge in the early Song time. Later, the thirty-six initial characters are used as alphabets for the initials in the rhyme tables and become basic terminology in the traditional

24

Figure 1.2 The first table in the Yùnjìng 韻鏡 ‘Mirror of Rhymes,’ one of the earliest rhyme tables. The leftmost column lists four corresponding rhymes, dōng 東, dǒng 董, sòng 送, wū 屋 in píng 平, shǎng 上, qù 去, and rù 入 tones respectively. The indicates phonological gaps.

An Introduction to Chinese Historical Phonology

25

Table 1.13 An example of allophonic variants of the columns of rhyme tables (described from right to left) Groups with four columns

e.g.

Groups with five columns

e.g.

Voiceless unaspirated stop Voiceless aspirated stop Voiced stop Nasal

t-/ʈth-/ʈhd-/ɖn-/ɳ-

Voiceless unaspirated affricate Voiceless aspirated affricate Voiced affricate Voiceless affricate Voiced affricate

ts-/tʂ-/tɕtsh-/tʂh-/tɕhdz-/dʐ-/dʑs-/ʂ-/ɕz-/ʐ-/ʑ-

Note: The resulting allophone is often a result of the syllable’s division.

Table 1.14 The thirty-six initials in the Qīyīn Lüè, with their reconstructions 幫 p 非 pf 端 t 知 ʈ 見 k 精 ts 照 tʃ 影 Ø 來 l

bāng fēi duān zhī jiàn jīng zhào yǐng lái

滂 敷 透 徹 溪 清 穿 曉 日

ph pfh th ʈh kh tsh tʃh h ɻ

pāng fū tòu chè xī qīng chuān xiǎo rì

並 奉 定 澄 群 從 床 匣

b bv d ɖ ɡ dz dʒ ɦ

bìng fèng dìng chéng qún cóng chuáng xiá

明 微 泥 娘 疑 心 審 喻

m ɱ n ɳ ŋ s ʃ j

míng wēi ní niáng yí xīn shěn yù

邪 z xié 禪 ʒ shàn

phonology. Because of such highly organized tables, the phonological information provided in the rhyme tables is very systematic; every syllable can be identified in terms of its tone, initial, rhyme, and děng, all very important pieces of information. Based on these rhyme tables, all the Qièyùn/Guǎngyùn rhymes can be organized into a coherent phonological system. 1.4.1

The Thirty-Six Initials

The rhyme table Qīyīn Lüè 七音略 ‘Summary of Seven Sounds,’ is dated to some time before 1161. The author of the book is unknown, though it owes its survival to Zhèng Qiáo 鄭樵, a historian of the Song dynasty who recorded the entirety of the Qīyīn Lüè in his 1161 encyclopedia Tōngzhì 通志. Within the Qīyīn Lüè, thirty-six characters are used to represent the initial consonants of characters in Chinese (see Table 1.14). Table 1.15 provides the reconstructed values of three groups of the initial characters from the thirty-six initials. A hyphen is inserted between the initials and finals. From the reconstructed values, it is easier to see why they were chosen to represent the Middle Chinese initial consonants.

26

The Keys to Traditional Phonology

Table 1.15 The reconstructed values of three groups of the syllables from the thirty-six initial characters 幫 端 見

p-ɑŋ t-wɑn k-en

滂 透 溪

並 定 群

ph-ɑŋ th-əw kh-ej

明 泥 疑

b-eŋ d-eŋ ɡ-jun

m-ɰiaŋ n-ei ŋ-ɨ

Table 1.16 A comparison of the initial characters and their categorical terminologies 脣音

‘sound of lips’

舌音

‘sound of tongue’

牙音 齒音

‘sound of back teeth’ ‘sound of front teeth’

喉音 舌音齒

‘sound of throat’ ‘sound of tongue, teeth’

幫 非 端 知 見 精 照 影 來

p pf t ʈ k ts tʃ Ø l

滂 敷 透 徹 溪 清 穿 曉 日

ph pfh th ʈh kh tsh tʃh h ɻ

並 奉 定 澄 群 從 床 匣

b bv d ɖ ɡ dz dʒ ɦ

明 微 泥 娘 疑 心 審 喻

m ɱ n ɳ ŋ s ʃ j

邪 z 禪 ʒ

Note: The reconstructed values of the thirty-six initials are reinforced by the descriptions given in the Yùnjìng 韻鏡.

In another rhyme table, the Yùnjìng 韻鏡 (1161 or earlier), instead of using thirty-six initial characters, the categorical terminologies for the place of articulation, the manner of articulation, and the phonation type are used (see Table 1.16). Initials can be categorized into initial groups, or zǔ 組, each named after the first character in that group (in Table 1.16 distinguished as each row); for example, the bāng group (幫組) consists of bilabial sounds, that is, bāng 幫 p, pāng 滂 ph, bìng 並 b, and míng 明 m. The phonetic values of the fēi 非 group initials (fēi 非 pf, fū 敷 pfh, fèng 奉 bv, wēi 微 ɱ) are reconstructed parallel to their bilabial counterparts. The phonetic values of these initials may not have ever existed. There is no evidence in modern dialects or foreign transcription that suggests that a contrast between fēi 非 (pf ) and fū 敷 (pfh) initial syllables ever existed; rather, it is likely that the maintenance of this contrast was in order to maintain the elegant four character pattern. It should be mentioned that in the traditional phonology, the subcategories of chúnyīn 脣音, shéyīn 舌音, and chǐyīn 齒音 have their own labels as well. According to the Jīngshǐ Zhèngyīn Qièyùn Zhǐnán 經史正音切韻指南 ‘Guide to the Fǎnqiè and Rhyme for the Standard Pronunciation of the Classics,’ a rhyme table of the fourteenth century, they are zhòngchún yīn 重脣音 and

An Introduction to Chinese Historical Phonology

27

Table 1.17 Definitions and categorizations of the Jīngshǐ Zhèngyīn Qièyùn Zhǐnán 脣音

重脣音 ‘sound of heavy lips’ 輕脣音 ‘sound of light lips’ 舌音 舌頭音 ‘sound of tongue tip’ 舌上音 ‘sound of tongue top’ 齒音 齒頭音 ‘sound of teeth tip’ 正齒音 ‘sound of straight teeth’

幫 非 端 知 精 照

p pf t ʈ ts tʃ

滂 敷 透 徹 清 穿

並 奉 定 澄 從 床

ph pfh th ʈh tsh tʃh

b bv d ɖ dz dʒ

明 微 泥 娘 心 審

m ɱ n ɳ s ʃ

邪 z 禪 ʒ

Table 1.18 Categories in the Yùnjìng directly compared to their modern phonemic analogues Bilabial stops Labiodentals Alveolar stops Retroflex stops Velars stops Alveolar affricates and fricatives Palatal affricates and fricatives Laryngeals and palatal approximant Lateral and retroflex approximant

幫 非 端 知 見 精 照 影 來

p pf t ʈ k ts tʃ Ø l

滂 敷 透 徹 溪 清 穿 曉 日

ph pfh th ʈh kh tsh tʃh h ɻ

並 奉 定 澄 群 從 床 匣

b bv d ɖ ɡ dz dʒ ɦ

明 微 泥 娘 疑 心 審 喻

m ɱ n ɳ ŋ s ʃ j

邪 禪

z ʒ

qīngchún yīn 輕脣音, shétóu yīn 舌頭音 and shéshàng yīn 舌上音, and chǐtóu yīn 齒頭音 and zhèngchǐ yīn 正齒音, respectively (Table 1.17). The classification of the place of articulation can be quite perfectly explained in terms of modern phonetics. Just as the values of the thirty-six initials can be reconstructed, the categories given in the Yùnjìng can also be directly compared to their modern phonemic analogues (see Table 1.18). 1.4.2

The Qīngzhuó 清濁 System

In the Yùnjìng, initials with the same place of articulation are further classified according to their manner of articulation and phonation types. The thirty-six initial characters are divided into four natural groups according to their relationships. There are four different types of initials, qīng 清 ‘clear,’ cìqīng 次清 ‘secondary clear,’ zhuó 濁 ‘murky,’ and qīngzhuó 清濁 ‘clear-murky.’ The initials in rhyme books are often separated into four groups based on their qīngzhuó pattern. The four groups can be distinguished through an analysis of their qīngzhuó patterns; for instance, Group 1 has the pattern qīng, cìqīng, zhuó, qīngzhuó for all of its places of articulation (see Table 1.19).

28

The Keys to Traditional Phonology

Table 1.19 An analysis of the four groups of qīngzhuó patterns Group 1

Group 2

Group 3

Group 4

qīng

cìqīng

zhuó

qīngzhuó

幫p 非 pf 端t 知ʈ 見k

滂 ph 敷 pfh 透 th 徹 ʈh 溪 kh

並b 奉 bv 定d 澄ɖ 群ɡ

明m 微ɱ 泥n 娘ɳ 疑ŋ

qīng

cìqīng

zhuó

qīng

zhuó

精 ts 照 tʃ

清 tsh 穿 tʃh

從 dz 床 dʒ

心s 審ʃ

邪z 禪ʒ

qīng

qīng

zhuó

qīngzhuó

影Ø

曉h

匣ɦ

喻j

qīngzhuó

qīngzhuó

來l

日ɻ

This classification can be easily explained in modern phonetics and makes perfect phonological sense: qīng: cìqīng: zhuó: qīngzhuó:

voiceless unaspirated obstruents8 voiceless aspirated obstruents voiced obstruents sonorants (also called cìzhuó 次濁).

This classification also determines the tonal development throughout history, as seen in Section 1.4.3. The thirty-six initial system represents the initials in use at the time it was created.9 It does not represent the initial system of the Qièyùn exactly (compare with the initials of the Qièyùn system in Section 3.4.3). These four labels were used as a basic classification of the manners and phonation types in the study of traditional phonology. It should be noted that the term qīngzhuó was later replaced by cìzhuó 次濁 ‘secondary murky’ in some texts. For the purpose of contrast between qīng and cìqīng, and zhuó and cìzhuó, qīng is sometimes called quánqīng 全清, while zhuó is commonly called quánzhuó 全濁. 8 9

Obstruents are stops (p, d), affricates (ts, dʑ), and fricatives (f, s). They contrast with sonorants (m, l, r, ʋ). The earliest publication date of the Yùnjìng is 1161, so the qīngzhuó system was reflective of the phonology of 1161 or earlier.

An Introduction to Chinese Historical Phonology

1.4.3

29

Yīndiào 陰調 and Yángdiào 陽調

The four tones of Middle Chinese were further developed into eight tones (in Chinese, called diào 調, as opposed to the four tones, called shēng 聲) conditioned by the initial consonants. This distinction is called the sìshēng bādiào 四聲八調 ‘four sounds, eight tones.’ According to traditional terminology, the syllables with the qīng 清 ‘voiceless’ initials and the zhuó 濁 ‘voiced’ initials split into different tones for all four tones. As discussed in Section 1.4.2, of the four types of initials, quánqīng 全清 and cìqīng 次清 are voiceless consonants, and quánzhuó 全濁 and cìzhuó 次濁 (qīngzhuó 清濁) are voiced consonants. The syllables with voiceless initial consonants become high register tones (yīn 陰), and the syllables with voiced initial consonants become low register tones (yáng 陽); here tone is referring to the contour (level, rising, etc.) while the register refers to the overall height (high register, low register) (see Table 1.20). As dialects evolve over time, different tones and registers may split or merge, and so this system is still used to describe modern tonal systems, such as Mandarin’s yīnpíng, yángpíng, shǎng, and qù. This can also be seen in Section 4.3.3.3 in a discussion on tonal mergers in the Song dynasty. A notation consisting of numerals for the four tones and using letters “a” and “b” for the upper and lower register tones (yīn and yáng) is often used (see Table 1.21). For the sake of this book, however, the common roman numeral labelling will not be used so as to not confuse tonal categories with děng categories, which already uses a roman numeral system.

Table 1.20 The eight tones divided along traditional tonal categories and the yīndiào–yángdiào distinction

Voiceless initials Voiced initials

Píng

Shǎng





yīnpíng yángpíng

yīnshǎng yángshǎng

yīnqù yángqù

yīnrù yángrù

Table 1.21 The eight register-tones as commonly labeled in Chinese phonology

Voiceless initials Voiced initials

Píng

Shǎng





Ia Ib

IIa IIb

IIIa IIIb

IVa IVb

30

The Keys to Traditional Phonology

Table 1.22 The names and Middle Chinese representations of Mandarin tones Mandarin tone number

Tone mark

Name

Middle Chinese notation

1 2 3 4

ā á ǎ à

yīnpíng yángpíng shǎng qùa

Ia Ib II III

a

Note that no rù or IV tone appears. This is because in the development of Mandarin, the rù tone was completely lost and all characters in the rù tone were divided among the other remaining tones (see Section 7.4.4).

Another reason why the roman numeral system is not employed for tonal categories in this book is because it may become confused with Mandarin’s modern numeral tone system. Mandarin uses a tonal system that is often referred to by its numbers (first tone, second tone, etc.) The Mandarin tones were created after tonal mergers and splits, and although each Mandarin tone contains Middle Chinese characters from multiple tones each, its numbers can be compared to the Middle Chinese numeral system through Table 1.22. 1.4.4

Děng 等 ‘Division’ and ‘Rank’

Děng, or ‘division,’ is a way that different types of finals are classified in rhyme tables. Although the Qièyùn itself does not use this methodology, early rhyme tables such as the Yùnjìng and Qīyīn Lüè, which analyzed the Qièyùn, categorized syllables from the Qièyùn into one of the four děng. Although the criteria for the děng assignment is not completely clear, it is generally agreed that finals with the same ending and similar main vowels are arranged mainly according to their medials as well as the quality of the main vowels (Baxter 1992). Because the Qièyùn and its subsequent rhyme tables are separated by centuries, and because any reference to the děng system imposed on the Qièyùn is external from the Qièyùn itself, there are discrepancies in how děng are discussed as they are arranged in rhyme tables and the projection of děng on other phonological material such as rhyme books. For this book, “division” refers to the děng as described in a given rhyme table, while “rank” refers to děng projected onto works that do not originally have this definition. Based on such an arrangement as seen in Table 1.23, all rhymes of the Qièyùn are classified accordingly with a děng assignment. For example, in the Yùnjìng, all syllables of the dōng 冬 rhyme (of the Qièyùn) are listed in the first Division, so the dōng 冬 rhyme is a Rank-I rhyme. All the syllables of the zhōng 鍾 rhyme are listed in the third Division, so the zhōng 鍾 rhyme is a Rank-III rhyme. But some rhymes have syllables that belong to different

An Introduction to Chinese Historical Phonology

31

Table 1.23 Divisions reflected in their reconstruction through similar vowelsa or through different medialsb

高 交 驕 澆 a b

gāo jiāo jiāo jiāo

The děng assignment of the rhyme table

The phonetic value of the Qièyùn

Division-I Division-II Division-III Division-IV

kɑw kɰaw kjɛw kew

In this case, ɑ a ɛ e all being related. In this case, -Ø-, -ɰ-, -j-.

divisions. The syllables of the dōng 東 rhyme are listed either in the first Division or the third Division. So, the dōng 東 rhyme is a Rank-I and Rank-III combined rhyme (yī děng sān děng hé yùn 一等三等合韻). It must be made clear that the děng is a concept of rhyme tables. Although it is frequently used in the study of the phonological system of the Qièyùn and each of the rhymes of the Qièyùn/Guǎngyùn is assigned with a děng, the Qièyùn/Guǎngyùn phonology does not contain the concept of děng originally. Although the děng assignments in the rhyme tables are in general a reflection of the fǎnqiè spellings of the Qièyùn, the relationship between the děng assignment of the rhyme tables and the rhymes of the Qièyùn is quite complicated for some rhymes. This complication is due to two basic reasons. One reason is that the sound system changed from the time of the Qièyùn (601 or earlier) to the time of the early rhyme tables (1161 or earlier); the phonetic values of the Qièyùn rhymes simply were no longer accessible to the authors of the rhyme tables. The other reason for this complication is the compact format of the rhyme tables. Some different initials are listed in the same column according to their complementary distributions at the time the rhyme tables were designed. Alveolar and retroflex stop initials are listed in the same columns, because the syllables with alveolar stops and retroflex stops combine with the finals in different děng. The alveolar initials only combine with the finals of Division-I and Division-IV and the retroflex ones only combine with the finals of Division-II and Division-III, so there is no possible ambiguity between the two series when reading them from the same column. In Figure 1.3, it can be seen that there are two rows of initial characters listed in the same column, duān 端, tòu 透, dìng 定, ní 泥, and zhī 知, chè 徹, chéng 澄, niáng 娘. There are four “big rows” representing four different divisions, in the order of Division-I, -II, -III, and -IV. Within each big row there are four rows of characters representing four different tones, in the order of píng,

32

The Keys to Traditional Phonology

Figure 1.3 A page from the Sìshēng Děngzǐ 四聲等子

shǎng, qù, and rù. The syllables with píng tone (the first row) in the first big row (dāo 刀, dāo 叨, táo 桃, náo 猱) and the fourth big row (diāo 碉, tiáo 祧, tiáo 條, ráo 嬈) are with alveolar stops (t-, th-, d-, n-), but the ones in the second big row (cháo 嘲, chāo 䫸, táo 䄻, náo 鐃) and the third big row (cháo 朝, chāo 超, cháo 晁, ○) are retroflex stops (ʈ-, ʈh-, ɖ-, ɳ-). Divisional

An Introduction to Chinese Historical Phonology

33

Table 1.24 Divisions reflected in their reconstruction through different initials

東 中

dōng zhōng

The děng assignment of the rhyme table

The phonetic value of the Qièyùn

Division-I Division-III

tung ʈjung

Table 1.25 Divisions allowed among each initial group Bilabials Labiodentals Alveolars Retroflexes Velars Alveolar affricates Palatal affricates Laryngeals and palatals Laterals and retroflexes 影 匣 來

Ø ɦ l

幫 非 端 知 見 精 照 影 來

p pf t ʈ k ts tʃ Ø l

滂 敷 透 徹 溪 清 穿 曉 日

ph pfh th ʈh kh tsh tʃh h ɻ

All divisions Division-I, -II, -IV Division-I, -II (very few), -III, -IV

並 奉 定 澄 群 從 床 匣

b bv d ɖ ɡ dz dʒ ɦ

曉 h 喻 j 日 ɻ

明 微 泥 娘 疑 心 審 喻

m ɱ n ɳ ŋ s ʃ j

邪 z 禪 ʒ

Division-I, -III, -IV Division-III Division-I, -IV Division-II, -III All divisions Division-I, -IV Division-II, -III (see below) (see below)

All divisions Division-III, -IV Division-III

differences can also be reflected in their reconstruction through different initials (see Table 1.24). The distribution of divisions across different initials is shown in Table 1.25. The allophonic distribution across alveolars and retroflex/palatal initials matches the logic employed by the creators of the rhyme tables. In the study of Middle Chinese phonology, the zhào 照 group initials of the syllables placed in Division-II are referred to as zhào-èr 照二 initials, and the zhào 照 group initials of the syllables placed in Division-III are referred to as zhào-sān 照三 initials. In standard references, they are distinguished using different characters. They actually represent different places of articulation of the consonants. While zhào 照 initials are represented by a palato-alveolar series (tʃ, tʃh, dʒ, ʃ, ʒ), their subdivision into zhào-èr initials and zhào-sān initials is distinguished by their retroflex and alveopalatal reconstructions respectively (see Table 1.26). Some contrastive syllables cannot be placed in the design of the four divisions and were likely forced to be placed in one of the four divisions, such as the the so-called chóngniǔ 重紐 contrast, where the Type-B and Type-A chóngniǔ syllables are placed in Division-III and Division-IV, respectively (see Section 1.4.6). The complications between the Qièyùn and rhyme tables are related to the Rank-III rhymes only, as shown in Table 1.27. The syllables of Rank-I,

34

The Keys to Traditional Phonology

Table 1.26 Subdivision of zhào 照 initials into zhào-èr 照二 initials and zhào-sān 照三 initials zhào 照 initials zhào-èr 照二 initials zhào-sān 照三 initials

照 莊 章

tʃ tʂ tɕ

穿 初 昌

tʃh tʂh tɕh

床 崇 船

dʒ dʐ dʑ

審 生 書

ʃ ʂ ɕ

禪 俟 禪

ʒ ʐ ʑ

Table 1.27 The relationship between divisions and ranks The Qièyùn

Rhyme tables

Syllables of Rank-I rhymes Syllables of Rank-II rhymes Syllables of Rank-III rhymes Syllables of Rank-IV rhymes

Division-I Division-II Division-II, Division-III, and Division-IV Division-IV

Table 1.28 The relationship between the Qièyùn and rhyme tables (Rank-III rhymes) The Qièyùn

Condition

Rhyme tables

Syllables of Rank-III rhymes with medial -j-

Initial is a retroflex Initial is an alveolar affricate or fricative Initial is j- (not ɦj-) Syllable is Type-A chóngniǔ (IIIa) Syllable is Type-B chóngniǔ (IIIb) Otherwise

Division-II Division-IV Division-IV Division-IV Division-III Division-III

Rank-II, and Rank-IV rhymes are all placed in the corresponding divisions in the rhyme tables. But the syllables within Rank-III rhymes are mainly placed in Division-III, but also placed in Division-II and Division-IV. Within this complication, however, Rank-III rhymes still maintain a regular pattern of distribution. Because of these conditions, the Qièyùn syllables of the same rhyme will appear in different děng. For the Rank-III rhymes, the general relationship between the Qièyùn and rhyme tables is given in Table 1.28. For example, syllables within the Rank-III yáng 陽 rhyme are not necessarily Division-III (see Table 1.29), and syllables within the Rank-III zhī 支 rhyme illustrate how the condition of chóngniǔ affects syllable division placement (see Table 1.30). The history and development of the reconstruction of these divisions is discussed in more depth in Section 4.1.4, which deals with each division individually in its relation to Middle Chinese medials and medial structure.

An Introduction to Chinese Historical Phonology

35

Table 1.29 Division placements of syllables of the Rank-III yáng 陽 rhyme 莊 tʂjaŋ II

Character Reconstruction Division

將 tsjaŋ IV

羊 jaŋ IV

張 ʈjaŋ III

章 tɕjaŋ III

Table 1.30 Division placements of syllables of the Rank-III zhī 支 rhyme 卑 pjɛ A IV

Character Reconstruction Chóngniǔ type Division

碑 pɰjɛ B III

岐 ɡjɛ A IV

奇 ɡɰjɛ B III

Table 1.31 A comparison of rhyme groups in the Qièyùn and Guǎngyùn, showing a rhyme split along the kāihé contrast Qièyùn

Guǎngyùn

1.4.5

開口 合口

寒 ɑn 寒 wɑn

刪 ɰan 刪 wɰan

仙 jɛn 仙 wjɛn

先 en 先 wen

開口 合口

寒 ɑn 桓 ɔn

刪 ɰan 刪 wɰan

仙 jɛn 仙 wjɛn

先 en 先 wen

Kāihé 開合

In rhyme tables, each of the tables is labeled according to the type of final it has. The labels kāikǒu 開口 and hékǒu 合口 are used to indicate whether there is a -w- medial or whether the vowel is rounded, where hékǒu syllables have either a rounded vowel or a -w- medial, and kāikǒu have neither. In the Qièyùn, syllables within a rhyme may contrast due to the presence or absence of a medial -w-. Though not labeled as such in the Qièyùn itself, those without a -w- medial are kāikǒu and those with -w- are hékǒu. The hán 寒 rhyme of the Qièyùn was split into the hán 寒 and huán 桓 rhymes in the Guǎngyùn. This split indicates that the finals -ɑn (kāikǒu) and -wɑn (hékǒu) of the hán 寒 rhyme in the Qièyùn changed to -ɑn and -ɔn. Although the kāihé contrast was maintained, the phonetic values of these two rhymes were now different, not in the medial but in the main vowel instead, and were thus declared separate rhymes (see Table 1.31). In order to distinguish kāihé contrasts within rhymes (such as hán 寒 and huán 桓 in the Qièyùn) from kāihé contrasts between rhymes (such as hán 寒 and huán 桓 in the Guǎngyùn), some phonological studies refer to the rhymes that contain only

36

The Keys to Traditional Phonology

Table 1.32 A comparison of the classification of kāihé in the Yùnjìng and the Gǔjīn Zìyīn Duìzhào Shǒucè

Yùnjìng Shǒucè



冬 and 鍾



kāi hé

kāihé hé

kāihé kāi

Table 1.33 The four kāihé labeled tables in the Yùnjìng Table 2 Table 4

冬 and 鍾 支

Table 3 Table 12

江 模 and 虞

hékǒu syllables as zhēn hékǒu 真合口 ‘real hékǒu’ and the rhymes that contain both kāikǒu and hékǒu syllables as jiǎ hékǒu 假合口 ‘fake hékǒu.’ Although the kāihé classification was already labeled in even the earliest rhyme tables such as the Yùnjìng, the kāihé classification as is used in modern studies appeared quite late. The kāihé classification in the early rhyme tables is quite different in the terminologies used and the classification of the rhymes. In the Yùnjìng, there are three categories, kāi 開, kāihé 開合, and hé 合 (see Table 1.32). There are four tables (2, 3, 4, and 12) in the Yùnjìng that are labeled with kāihé (Table 1.33). It has been shown that the kāihé label of Table 4 is due to a mistake (Yáng Jūn 2007). If so, the other three all have a rhyme containing a non-high round vowel, 冬 woŋ, 江 ɰɔŋ, and 模 wo. It is possible that the label kāihé is related to this type of vowel. The kāihé classification of the commonly used Gǔjīn Zìyīn Duìzhào Shǒucè was by and large based on Karlgren’s Études sur la phonologie chinoise (1915–1926). Karlgren used the kāihé classification of the Děngyùn Qièyīn Zhǐnán 等韻切音指南, which is an attachment of the Kāngxī Zìdiǎn 康熙字 典. The sound system of the Děngyùn Qièyīn Zhǐnán reflected the sound system of the fourteenth century and later (Mǎ Déqiáng 2012). The standard reference of Middle Chinese phonology, the Gǔjīn Zìyīn Duìzhào Shǒucè, followed Karlgren’s practice and used the kāihé classification of the Děngyùn Qièyīn Zhǐnán with minor corrections. Thus, the kāihé classification in standard references does not reflect the phonetic values of the Qièyùn rhymes exactly. 1.4.6

Chóngniǔ 重紐 and Chóngyùn 重韻

In the traditional study of Chinese phonology there are two similar sounding terminologies that must be clearly distinguished. One is chóngniǔ 重紐 ‘duplicated button’ and the other is chóngyùn 重韻 ‘duplicated rhyme.’

An Introduction to Chinese Historical Phonology

37

In the Qièyùn and its revised versions, such as the Guǎngyùn, words are hierarchically classified first according to their tones, second to their rhyme part of the syllable, and last to their initials. The smallest unit in this classification is the tóngyīn zìzǔ 同音字組 ‘homophonic group.’ In most cases, different homophonic groups classified in the same rhyme have different initials but have the same kāihé distinction (Section 1.4.5) and the same děng classification (Section 1.4.4) according to the rhyme tables. However, for some Rank-III rhymes two separate homophonic groups can have the same kāihé, the same děng, and the same initial. These seemingly duplicated homophonic groups are traditionally termed chóngniǔ 重紐 ‘duplicated buttons,’ because each homophonic group in rhyme dictionaries is marked with an “ο” shape, or, in Chinese, niǔ 紐 ‘button.’ How to distinguish these two homophonic groups was a mystery for Chinese scholars for many centuries, because if the words have the same initial, and are classified the same in final, kāihé, and děng, they should be classified together in the same homophonic group in the same rhyme. Since this distinction was not well preserved in modern dialects, the phonetic difference was not clear until quite recently. Scholars gradually realized the phonetic difference of these chóngniǔ pairs is the medial: one homophonic group has a -ɰ- medial that is historically developed from -r-, and the other does not. For example, words yāo 邀 ‘invite’ and yāo 妖 ‘demon’ belong to two different homophonic groups of the same rhyme. Their phonetic difference in the Middle Chinese time is that yāo 邀 is Ɂjɛw but yāo 妖 is Ɂɰjɛw, with the same rhyme part -jɛw and the same initial Ɂ- but different medials, -jand -ɰj-, respectively. This is why they belong to the same rhyme: because they have the same rhyme part -jɛw, but different homophonic groups although they have the same initial Ɂ-. This is the general format for chóngniǔ syllables, with a -ɰj- syllable (I(w)ɰjVE) commonly labeled as IIIb (placed into the Division-III row in rhyme tables) and a -j- syllable (I(w)jVE) labeled as IIIa (placed into the Division-IV row in rhyme tables). Chóngniǔ is discussed in much more depth in Section 3.5.6 and Section 3.6.3, with more examples and information in Section 2.3.2.3.2, Section 3.1.2, Section 3.3, Section 3.5.4, Section 5.2.1.2, Section 6.4.3, and Section 7.4.6.2. The concept of chóngyùn was another mystery of the phonology of Middle Chinese. In the Qièyùn and the Guǎngyùn, there are some pairs of yùn 韻 ‘rhymes’ that are classified with same děng, same kāihé, and are also in the same shè 攝 ‘rhyme group’ (see Section 1.4.9), according to rhyme tables. For example, rhyme yú 魚 and rhyme yú 虞 are both classified as Division-III and hékǒu, which means that they have a -j- medial and a rounded main vowel, and have no coda (because they belong to the same shè). Because they share many phonological features, they are referred to as chóngyùn 重韻 ‘duplicated rhymes.’ This term means that they are so similar, it seems as if the rhymes

38

The Keys to Traditional Phonology

are duplicated. Through more detailed analysis of their fǎnqiè spellings and the residual distinction in modern dialects, scholars in general agreed that they were separate rhymes with different main vowels. For example, the phonetic value for rhyme yú 魚 is Middle Chinese -jɔ and for rhyme yú 虞 is Middle Chinese -jo. Chóngyùn is discussed in much more depth in Section 3.5.7 and Section 3.6.4, with more examples in Section 5.2.1.1. 1.4.7

Sìhū 四呼

Somewhat of an expansion of the kāihé 開合 analysis, the traditional term sìhū 四呼 ‘four calls’ is another classification system for finals, used often with late Mandarin phonology. This classification is based on the medials and their corresponding vowels. The four types of finals are kāi 開, qí 齊, hé 合, cuō 撮 or kāikǒu 開口, qíchǐ 齊齒, hékǒu 合口, and cuōkǒu 撮口. The finals are classified into these four categories: Kāikǒu: Qíchǐ: Hékǒu: Cuōkǒu:

finals with finals with finals with finals with

no medials, and the main vowel is neither -i, -u, nor -y. the -j- medial, or the main vowel is -i. the -w- medial, or the main vowel is -u. the -ɥ- medial, or the main vowel is -y.

Examples of the sìhū are given in Table 1.34. Each of the four types have examples, which are identified by their vowel or their medial, except for kāikǒu, which does not have a given medial. This terminology first appeared in the anonymous rhyme table Yùnfǎ Zhítú 韻法直圖 (1612) as an attachment of Méi Yīngzuò’s 梅膺祚 Zìhuì 字匯. This classification in the traditional phonology was mainly used for Mandarin dialects but is widely used in the study of modern dialects today. The sìhū system itself indicates that the phonological system of finals has changed from that of Middle Chinese, which is classified in terms of děng and kāihé. The Table 1.34 Examples of the sìhū Example Kāikǒu

No medial, no -i, -u, -y vowel

Qíchǐ

High front medial or vowel

Hékǒu

High rounded medial or vowel

Cuōkǒu

High rounded front medial or vowel

馬 紅 蛇 西 錢 虎 萬 女 泉

mǎ hóng shé xī qián hǔ wàn nǚ quán

Meaning ‘horse’ ‘red’ ‘snake’ ‘west’ ‘money’ ‘tiger’ ‘ten thousand’ ‘woman’ ‘spring’

(vowel a) (vowel o) (vowel ɤ) (vowel i) (medial j) (vowel u) (medial w) (vowel y) (medial ɥ)

An Introduction to Chinese Historical Phonology

39

Table 1.35 The medial systems of Middle Chinese and the sìhū Middle Chinese medial system kāi -Ø-ɰ-j-, -ɰj-Ø-

Rank-I Rank-II Rank-III Rank-IV Sìhū medial system kāi qí hé -Ø-j-w-

hé -w-wɰ-wj-, -wɰj-w-

cuō -ɥ-

most significant difference is related to the change in Middle Chinese Division-II syllables that had a -r- and later -ɰ- medial, the merger of Division-III and -IV syllables, and the changes of initial consonants, such as labiodentalization and retroflexization, which causes the loss of medial -j- (see Table 1.35). The rhyme tables that employed the sìhū system represent a breakthrough in the format of rhyme tables in the tradition of the Yùnjìng and Qièyùn Zhǐzhǎngtú, and a reclassification of the finals. Although this terminology emerged in the seventeenth century, the change of the final system as seen in the ancestral forms of Modern Mandarin can be traced back much earlier. As early as the Yuan time in the thirteenth century, such a final system already existed (see Section 6.4.1). 1.4.8

Wàizhuǎn 外轉 and Nèizhuǎn 內轉

In the Yùnjìng each table is also labeled with wàizhuǎn 外轉 and nèizhuǎn 內 轉 (see Figure 1.4), which is a binary classification system for all the rhymes in many rhyme tables, such as the Yùnjìng, Qīyīn Lüè, and Sìshēng Děngzǐ (Section 4.4.2). The phonological reason and phonetic basis for this classification still remain controversial. Zhuǎn 轉 is a concept that first appeared in the prefatory pages of the Yùnjìng, in which each rhyme table is also called zhuǎn, sorted as the fortythree tables of the Yùnjìng. There are two types of zhuǎn, wàizhuǎn 外轉 ‘outer turn’ and nèizhuǎn 內轉 ‘inner turn.’ A traditional way to differentiate these two types of tables is whether a rhyme group (shè, a later concept, discussed in Section 1.4.9) has dúlì èr děng yùn 獨立二等韻 ‘an independent Rank-II rhyme’ (that is, the shè contains a rhyme that only contains Division-II syllables). For a given shè, if one of its rhymes is an independent Rank-II rhyme, the tables of the rhymes of this rhyme group are labeled as wàizhuǎn, if not, then the tables are labeled as nèizhuǎn. For example, table one of the Yùnjìng is nèizhuǎn, because there are no Rank-II rhymes; dōng 東, dōng 冬, or zhōng 鍾 are either Division-I or Division-III rhymes (see Table 1.36).

40

Figure 1.4 The Yùnjìng pages for the rhymes dōng 東, dōng/zhōng 冬/鍾, and jiāng 江, and their zè 仄 tone equivalents.

41

Figure 1.4 (cont.)

42

Figure 1.4 (cont.)

An Introduction to Chinese Historical Phonology

43

Table 1.36 The division assignments of different rhymes within the rhyme groups (shè), divided by their zhuǎn Wàizhuǎn 外轉 ‘outer turn’ rhyme groups according to the Yùnjìng 江 jiāng II 蟹 xiè I, II, III, IV 山 shān I, II, III, IV 梗 gěng II, III, IV 咸 xián I, II, III, IV 臻 zhēn II,a III

效 xiào 假 jiǎ

Nèizhuǎn 內轉 ‘inner turn’ rhyme groups according to the Yùnjìng 通 tōng I, III 止 zhǐ III 遇 yù I, III 深 shēn III 曾 zēng I, III 果 guǒ I, III

流 liú I, III 宕 dàng I, III

I, II, III, IV II, III

Note: The characters represent different rhyme groups, while the roman numerals describe the different ranks of rhymes within that rhyme group. a The zhēn 臻 and zhì 櫛 rhymes of the zhēn 臻 rhyme group are independent Division-II rhymes in the Guǎngyùn, but often analyzed together with the zhēn 真 rhyme due to their complementary distribution (the syllables of the zhēn 臻 and zhì 櫛 rhymes pair with retroflex initials only, but the syllables of the zhēn 真 pair with other initials). However, this is a categorical observation with no phonetic explanation.

Table 1.37 The vowel quality of rhymes in the nèizhuǎn and wàizhuǎn rhyme groups, according to Luó Chángpéi Wàizhuǎn 外轉 ‘outer turn’ rhyme groups 江 jiāng ɔ 蟹 xiè e, ɐ, æ, a, ɑ 梗 gěng e, ɐ, æ, ɛ 咸 xián e, ɐ, æ, a, ɑ 宕 dàng a, ɑ Nèizhuǎn 內轉 ‘inner turn’ rhyme groups 通 tōng o 止 zhǐ i 深 shēn ə 曾 zēng ə

山 shān e, ɐ, æ, a, ɑ 假 jiǎ a

效 xiào e, æ, a, ɑ 果 guǒ ɑ

遇 yù u, o 臻 zhēn ə, ě

流 liú

ə

Source: Luó 2004: 125, 128–129.

To explain the phonetic nature of this opposition, Luó Chángpéi (1933, 2004) suggested that this contrast is related to vowel quality. The representative main vowels are given in Table 1.37. The phonetic values are mostly based on Karlgren’s reconstruction, except the mó 模 and yú 魚 rhymes of the yù 遇 rhyme group, and the dōng 東 rhyme of the tōng 通 rhyme group. In order to show a better correlation he changed the labels of three rhyme groups: the guǒ 果 and the dàng 宕 rhyme groups from nèizhuǎn to wàizhuǎn, and the 臻 zhēn rhyme group from wàizhuǎn to nèizhuǎn. In Luó’s new classification, there are nine wàizhuǎn but seven nèizhuǎn groups instead. After the adjustment of rhyme groups, his explanation therefore is that the nèizhuǎn groups have back vowels u, o, central vowels ə, and high-front vowels i, e, ě, and the

44

The Keys to Traditional Phonology

wàizhuǎn groups have front vowels e, ɛ, æ, a, central vowels ɐ, and back vowels ɔ, ɑ: Inner: Outer:

i, e, ě, ə, u, o e, ɛ, æ, a, ɐ, ɔ, ɑ

This classification, although it rearranges the rhymes of these two categories, can account for the conditions of a number of sound changes in the Chinese dialects, for example, in Zhào Yuánrèn et al.’s study of the Hubei dialects (Chao et al. 1948: 10–11). In many locations the change of the Middle Chinese zhuāng 莊 initial series is conditioned by nèizhuǎn and wàizhuǎn, the reflexes of nèizhuǎn rhyme syllables are alveolars, ts-, tsh-, s-, and the reflexes of wàizhuǎn syllables are retroflexes, tʂ-, tʂh, ʂ-. The other phonetic explanation is the contrast of vowel length. Some researchers (Huáng Xiàoshān 2000; Zhèng Wěi 2018) note that the nèizhuǎn and wàizhuǎn opposition seemingly has a correlation with the long and short vowel contrast in the Zhuang language as well as the Chinese Yue dialects. As early as the sixth century, by imitating Chinese characters, Zhuang characters or Sawndip (a Zhuang word that means “immature characters”) were created to write the Zhuang language. Like the Chinese characters, the phonetic–semantic compounds are also used as a principle for creating Zhuang characters. Many of the phonetic components of the Zhuang characters are Chinese characters, which are used to provide phonetic cues for the Zhuang words. For example, the phonetic component of 岜 bya ‘mountain’ is the Chinese character 巴 bā, and the phonetic component of 伝 vunz ‘person’ is the Chinese character 云 yún. Researchers in general agree that such Sino-Zhuang pronunciation reflects the pronunciation of late Middle Chinese (Huáng Xiàoshān 2000). One of their findings is that there is a certain degree of correlation between the length contrast of the vowels of Zhuang and the nèizhuǎn/wàizhuǎn contrast of Middle Chinese rhyme groups. Generally speaking, the main vowels of the wàizhuǎn rhyme groups are more related to the long vowels, and the main vowels of the nèizhuǎn rhyme groups are more related to the short vowels (see Table 1.38). However, upon close examination, the distribution is rather complicated. In the modern Wuming dialect of Zhuang, monophthongs a, i, u, e, o, ɯ are all long vowels and do not have their corresponding short vowels. Table 1.39 is a further analysis based on the information of vowel length given in Table 1.38. It can be observed that regardless of a given zhuǎn assignment, if the main vowel of a rhyme group is a ə, their reflexes in Zhuang are always a short ɐ. The rhyme groups with a main vowel ə, namely the liú 流, shēn 深, zēng 曾, and zhēn 臻 rhyme groups, consistently have short vowel reflexes in Zhuang. But liú 流, shēn 深, and zēng 曾 rhyme groups are nèizhuǎn, while the

An Introduction to Chinese Historical Phonology

45

Table 1.38 The phonetic values of nèizhuǎn and wàizhuǎn rhyme groups, according to the phonetic components of the Zhuang characters Wàizhuǎn 外轉 ‘outer turn’ rhyme groups 江 jiāng II 蟹 xiè I, II, III, IV a:ŋ, o:ŋ a:j, o:j, i, ej, aj 梗 gěng II, III, IV e:ŋ, i:ŋ, a:ŋ

咸 xián I, II, III, IV a:m, i:m, am, om, um Nèizhuǎn 內轉 ‘inner turn’ rhyme groups 通 tōng I, III 止 zhǐ III oŋ, uŋ, o:ŋ, i, ej, aj u:ŋ 深 shēn III 曾 zēng I, III am, im, aŋ om, um

山 shān I, II, III, IV a:n, u:n, o:n, i:n, e:n 臻 zhēn II*, III an, on, un, ɯn

效 xiào I, II, III, IV a:w, e:w, i:w 假 jiǎ II, III a, e

遇 yù

I, III o, u, ow, aɰ

流 liú

果 guǒ

I, III o, a, u

宕 dàng I, III o:ŋ, u:ŋ, i:ŋ

I, III aw, ow

Source: Zhèng Wěi 2018: 31.

Table 1.39 A summary of the long (L) and short (S) vowels in Table 1.38 Wàizhuǎn 外轉 ‘outer turn’ rhyme groups, according to the Yùnjìng 江 jiāng L 蟹 xiè L, S 山 shān L 梗 gěng L, S 咸 xián L, S 臻 zhēn S

效 假

xiào jiǎ

L L

Nèizhuǎn 內轉 ‘inner turn’ rhyme groups, according to the Yùnjìng 通 tōng S, L 止 zhǐ L, S 遇 yù L, S 深 shēn S 曾 zēng S 果 guǒ L

流 宕

liú dàng

S L

zhēn 臻 rhyme group is wàizhuǎn. This indicates that Zhuang, which lacks a ə in its phonology, uses a short ɐ to transcribe the phonetically similar ə of Chinese. So using short ɐ for these rhyme groups is a phenomenon of Zhuang rather than one of Chinese. In addition, there is no other independent evidence suggesting the existence of this contrast; the similar phenomenon of the Yue dialect is just an influence of Zhuang, since historically the formation of Yue is under the strong influence of Zhuang. Based on what has been analyzed so far, unlike what has been claimed, the length contrast does not show a very strong pattern of correlation with the outer/inner contrast. However, if we put the long and short vowels of Zhuang in Luó’s system, the fit is actually much better (see Table 1.40). It is quite clear that Luó’s classification, which is based on the vowel height, shows a better correlation with the long and short vowels of Zhuang.

46

The Keys to Traditional Phonology

Table 1.40 A summary of the long (L) and short (S) vowels in Table 1.37 Wàizhuǎn 外轉 ‘outer turn’ rhyme groups, according to Luó Chángpéi 江 jiāng L 蟹 xiè L, S 山 shān L 梗 gěng L, S 咸 xián L, S 假 jiǎ L 宕 dàng L

效 果

xiào guǒ

L L

Nèizhuǎn 內轉 ‘inner turn’ rhyme groups, according to Luó Chángpéi 通 tōng S, L 止 zhǐ L, S 遇 yù L, S 深 shēn S 曾 zēng S 臻 zhēn S



liú

S

In Table 1.38, which is based on the vowel height, the six rhyme groups (jiāng 江, shān 山, xiào 效, jiǎ 假, guǒ 果, dàng 宕) that are only transcribed by long vowels are all wàizhuǎn, and the four rhyme groups (liú 流, shēn 深, zēng 曾, zhēn 臻) that are only transcribed by short vowels are all nèizhuǎn. The remaining six rhyme groups are transcribed by both long and short vowels. Of these six rhyme groups, xiè 蟹, gěng 梗, and xián 咸 are wàizhuǎn and tōng 通, zhǐ 止, and yù 遇 are nèizhuǎn. Thus, the vowel quality can give a better explanation for the contrast of vowel length in the Sino-Zhuang pronunciation. The length contrast is a result of how Chinese phonology was interpreted by Zhuang speakers when the two languages were in contact. Since there is no other independent evidence suggesting that this contrast was a phonological feature of Middle Chinese, the vowel length is just a reflection of the Zhuang language and shows how Chinese phonology was interpreted by the Zhuang phonology when the two languages were in contact. The phonological and phonetic interpretation of the nèizhuǎn and wàizhuǎn are not perfect; many pieces of evidence indicate that it is likely a binary classification of rhyme groups based on vowel quality. Unlike other properties of Middle Chinese phonological analysis, wàizhuǎn and nèizhuǎn are not referred to in the main Middle Chinese reference books, such as the Gǔjīn Zìyīn Duìzhào Shǒucè and the Fāngyán Diàochá Zìbiǎo 方言調查字表, as well as in most Middle Chinese research works. 1.4.9

Shè 攝 ‘Rhyme Groups’

The concept of shè 攝 or yùnshè 韻攝 ‘rhyme groups’ as the classification of rhymes first appeared in a rhyme table called Sìshēng Děngzǐ 四聲等子, an anonymous work of the twelfth century (and discussed in depth in Section 5.4.2). In the Sìshēng Děngzǐ, all the rhymes of the Qièyùn are classified into sixteen shè, according to their main vowels and codas (see Table 1.41). This classification reflects the phonetic values of the rhymes at the time it was

An Introduction to Chinese Historical Phonology

47

Table 1.41 The sixteen shè with their basic main vowels and codas 通 蟹 果 曾

tōng xiè guǒ zēng

uŋ/uk aj ɔ əŋ/ək

江 臻 假 流

jiāng zhēn jiǎ liú

ɔŋ/ɔk ən/ət a əw

止 山 宕 深

zhǐ shān dàng shēn

i an/at aŋ/ak əm/əp

遇 效 梗 咸

yù xiào gěng xián

u aw ɛŋ/ɛk am/ap

Table 1.42 The shè system systematically categorized into different vowel-coda pairings

-Ø -j -w -m/-p -n/-t -ŋ/-k

i

ə

ɛ

a

ɔ

u

y

止 – 流 深 臻 曾

止 – 流 深 臻 曾

– 蟹 效 咸 山 梗

假 蟹 效 咸 山 宕

果 – – – – 江

遇 – – – 臻 通

遇 – – – 臻 通

produced. This system is often referred to in the study of Middle Chinese phonology, although it appeared much later than the Qièyùn. Each rhyme group can contain the rhymes listed in different divisions. These sixteen shè can be arranged according to their codas. They represent a final system (see Table 1.42). This representation is based on the following phonological considerations: 1. The natural vowel system for finals with different codas. Basic vowels -i, -a, -u must be included before other vowels, such as -ɛ and -ɔ. 2. For the shè with vowel -ə, such as liú 流, zhēn 臻, shēn 深, and zēng 曾, their Division-III syllables have a main vowel -i, e.g. for the liú shè, the Division-I syllable 狗 gǒu ‘dog’ is kəw, and the Division-III syllable jiǔ 九 ‘nine’ is kiw. 3. For the shè with vowel -u, such as yù 遇 and tōng 通, their Division-III syllables have a main vowel -y, e.g. for the tōng shè, the Division-I syllable gōng 攻 ‘attack’ is kuŋ, and the Division-III syllable gōng 宫 ‘palace’ is kyŋ (gōng 攻 and gōng 宫 become homophones in Modern Mandarin). 4. For the zhēn 臻 shè, the Division-I hékǒu syllables have a main vowel -u and the Division-III hékǒu syllables have a main vowel -y, e.g. the Division-I syllable zūn 尊 ‘respect’ is tsun, and the Division-III syllable zūn 遵 ‘obey’ is tsyn (zūn 尊 and zūn 遵 become homophones in Modern Mandarin). 5. For the zhǐ 止 shè, the central vowel (apical vowel) -ɨ is listed under -ə.

48

The Keys to Traditional Phonology

Table 1.43 The parallel structure of Rùshēng Yùn and Yángshēng Yùn seen through a comparison of their codas

Rùshēng Yùn Yángshēng Yùn

1.4.10

Labial

Coronal

Dorsal

p m

t n

k ŋ

Yīnshēng Yùn 陰聲韻, Yángshēng Yùn 陽聲韻, and Rùshēng Yùn 入聲韻

Another often used classification of the rhymes is based on their codas. Those syllables that end with a semivowel -j or -w, or have no coda, are the Yīnshēng Yùn 陰聲韻, those that end with a nasal consonant -m, -n, -ŋ are the Yángshēng Yùn 陽聲韻, and those that end with an oral stop -p, -t, -k are the Rùshēng Yùn 入聲韻. In Mandarin, the loss of the rù tone signals with it the loss of the parallel nature of Rùshēng Yùn and Yángshēng Yùn (see Table 1.43). Because rù tone syllables were necessarily rùshēng yùn syllables, all rù tone syllables in Middle Chinese had a coda and could not be parallel to any rhymes in the other three tones that had no coda (see Section 2.3.3 for a discussion on these applications to Old Chinese). 1.5

Nonanalytical Source Materials

1.5.1

Poem Rhyming

Chinese poetry employs rhyming, the requirements of which also involve tonal match. For this reason, many historical phonologists use the rhyming of poetry to identify changes related to rhymes and tones. The rhyming syllable is the last syllable of a line. Not all the lines must contain a rhyming syllable, however; which line should contain a rhyming syllable varies across different historical periods. For example, in the Tang dynasty, a poem with four lines required the second and fourth lines to have a rhyming syllable, the first line could optionally rhyme with lines two and four, but the third line was not to have a rhyming syllable (refer to Section 2.4, Section 4.5, and Section 7.5 for examples of poem rhyming). Almost all the poems of the ancient Shījīng 詩經 ‘Book of Odes’ involve rhyming. The rhyming syllables are usually at the end of a line, but not all the lines contain a rhyming syllable at the end. A prerequisite is to identify which lines contain a rhyming word and which rhyming words rhyme with which. Scholars may come up with different identifications for the rhyming

An Introduction to Chinese Historical Phonology

49

words. Even more complicated, the rhyming syllables sometimes are not at the end of a line. It is thus a difficult task for the scholar to determine the rhyming syllables of the ancient poems in the study of Old Chinese (Baxter 1992). Because of the strength of the Chinese rhyming tradition, any given poetry rhyming may not reflect the pronunciation of its respective time. Many types of poems follow the rhyming standards established in rhyme dictionaries closely. One of the most influential rhyming standards was the one contained in the Qièyùn and its revision Guǎngyùn, which started in the early seventh century. The phonological information of the time when the poems were written can only be detected by the mistakes the poets made due to the influence of their contemporary pronunciations. 1.5.2

Foreign Transcriptions and Sino-Xenic Pronunciation

Chinese was in contact with other languages throughout history, and of those that had writing systems, most writing systems were phonetic. Through comparison of transcriptions, then, the phonetic value of the Chinese loans can be retrieved. This information is very valuable because Chinese materials such as rhyme dictionaries, rhyme tables, and poetry rhyming only provide categorical information. Examples of these contact languages and writing systems included, among others, Uyghur and Tibetan as early as the Tang dynasty, and subsequently Khitan, Jurchen, Persian, Mongol, Korean, and Manchu. The earliest transcription of an entire Chinese phonology is the Ménggǔ Zìyùn 蒙古字韻 of the Yuan dynasty. From this, many important phonetic details not available before can be learned. Foreign transcriptions made in identifiable geographic areas and time periods are especially informative; for example, beginning in the tenth century the Khitan territory included the north and northeastern part of modern China. Its southern boundary was south of modern Beijing, which makes it clear that the Chinese pronunciation was from dialects within its territory, not from Chinese dialects spoken outside its territory. Throughout history, Chinese words, their written form (characters), pronunciation, and meaning, were also borrowed into the languages of neighboring countries, such as Japan, Korea, and Vietnam. Although these Chinese loans are a very significant part of these languages, Japanese, Korean, and Vietnamese are not genetically related to Chinese. Depending on the different languages, the pronunciation of these loan words is referred to as Sino-Japanese, Sino-Korean, and Sino-Vietnamese individually, and collectively as Sino-xenic pronunciation. Since they are more or less frozen pronunciations in these languages after they were borrowed, the Sino-xenic pronunciation can reveal phonological contrasts that may not be clear in the Chinese language itself.

50

The Keys to Traditional Phonology

Chinese words were borrowed from multiple periods of time in history, so the Sino-xenic pronunciation has historical layers. But the main layer should reflect the Middle Chinese pronunciation because, during that period of time, the time of the Tang dynasty (618–907), the influence of Chinese culture was at its peak. Because a large number of Chinese words were borrowed, complete phonological systems of consonants and vowels can be retrieved. In Vietnamese, the tonal information can also be retrieved because Vietnamese is tonal. These systems are referred to as “Sino-xenic dialects” (Martin 1953), because they can be used in a similar way as the Chinese dialects in the study of Middle Chinese. In his pioneering work of reconstructing Middle Chinese phonology, Karlgren (1915–1926) uses Sino-Japanese and Sino-Korean along with various Chinese dialects to decide the phonetic values of the existing Middle Chinese phonological categories. These different systems of Sino-xenic pronunciation were approximations of Middle Chinese. They are all altered by the phonological system of the language that borrowed the words. However, many important phonological contrasts of Middle Chinese may still be kept in these languages, such as the contrast of the chóngniǔ syllables (refer to Section 1.4.6) of the Qièyùn and the contrast of Division-III and Division-IV syllables (refer to Section 1.4.4) as classified in the rhyme tables. The Chinese borrowings are not limited to foreign countries. Chinese words are also extensively borrowed into many non-Chinese languages spoken in China, for example, the Zhuang language and the Miao language. The Chinese words in these languages also form various Sino-xenic pronunciation systems. But these languages are in close contact with Chinese continuously throughout history, and thus the pronunciations have multiple layers from different periods and are quite complex (Huáng Xiàoshān 2000). Many groups of people ruled China or part of China historically. Chinese words have also been borrowed into these languages, such as Khitan, Mongolian, and Manchu, forming various systems of Sino-xenic pronunciation. These systems also provide very important information about the Chinese language of a certain historical period. More research shows that the Sino-Khitan pronunciation is similar to Old Mandarin instead of Middle Chinese, although chronologically the Khitan’s rule (916–1125) is almost parallel to the Northern Song dynasty (960–1127), which still used Middle Chinese phonology (the Guǎngyùn) as its standard pronunciation (Shen 2007). 1.5.3

Xiéshēng Characters 諧聲字

Chinese characters are generated in different ways. According to the classification in the first dictionary of Chinese characters, the Shuōwén Jiězì 説文解 字, written by Xǔ Shèn 許慎 in the second century, all Chinese characters can

An Introduction to Chinese Historical Phonology

51

Table 1.44 The xiéshēng characters of gōng 工 工 gōng ‘worker’ k-

貢 gòng ‘tribute’ k-

攻 gōng ‘attack’ k-

江 jiāng ‘river’ k-

豇 jiāng ‘bean’ k-

紅 hóng ‘red’ ɦ-

虹 hóng ‘rainbow’ ɦ-

be analyzed into six categories. The six categories are called liùshū 六書 ‘the six types of characters, or six principles of creating characters.’ One of the six categories, which contains phonetic information, is xíngshēng 形聲 ‘phonetic compound.’ The principle of xíngshēng is to use the existing characters or graphs to create a new character. Such characters have two components, one radical and one phonetic. The radical indicates the semantic category and the phonetic indicates the sound category of the new character. Characters that are xíngshēng and share the same phonetic component are called xiéshēng 諧聲 characters.10 Quite often the xiéshēng characters demonstrate obvious phonetic similarities. For example, in Modern Mandarin the characters with the phonetic component gōng 工 have the same main vowel and ending -uŋ, and similar initials, either a voiceless unaspirated velar stop k-, or a voiced velar fricative ɦ- in Middle Chinese phonology (see Table 1.44). The information retrieved from the xiéshēng characters is pivotal in the reconstruction of Old Chinese phonology. The phonological information available from the rhyming of poems contained in the Book of Odes and other ancient documents is limited to the rhyming part of the syllable: from the main vowel to the end of a syllable. Information about the element in front of the main vowel must come from other materials. However, it should be noted that a methodological problem for retrieving phonological information from xiéshēng characters is that historically Chinese characters are created gradually and may also be based on phonologically different ancient dialects. There is no guarantee that the phonological relationships between the xiéshēng characters were from one single system. However, scholars often use pieces of information retrieved from the xiéshēng characters for the reconstruction of a single system. Furthermore, the phonological information from xiéshēng characters and from rhyming materials may not form the same phonological system. 1.6

The Availability and Reliability of Source Materials

Information regarding a phonological system intrinsically concerns its contrastive categories and their phonetic values; the date and location of a 10

This includes the phonetic component by itself as well, which may not itself be a xíngshēng character.

52

The Keys to Traditional Phonology

system are also crucial. Different materials vary in these qualities but are often used and discussed together in reasoning and when making conclusions. For example, poetry rhyming can provide categorical information but not phonetic information, and while it can be identified in time and roughly in space based on information about the authors, such information is far from systematic. Various source materials are very different in nature and reveal phonological information with varying degrees of quality and reliability. It is important that these source materials be explained before using them to represent a history of phonology. To better understand the quality of the various materials, we must first evaluate them with unified criteria. In the presentation and discussion of phonological information the following criteria are used as basic measures: (1) whether the phonological material contains clear information about time, (2) whether the phonological material contains clear information about space, (3) whether the phonological material provides information about phonological categories, (4) whether the phonological material provides phonetic values, and (5) whether the phonological material is systematic. For example, a rhyme dictionary usually provides systematic information about the phonological categories but does not include the phonetic values of these categories, and it could have quite clear information about when it was compiled but no indication of the specific location of the dialect on which the dictionary was based. To study any phonological system two pieces of basic information are required: its phonological categories and the phonetic values of these categories. These two pieces of information can be the two sides of the same coin for a writing system that represents speech sounds, such as English, French, or Spanish. But the Chinese writing system is logographic. Although it may have phonetic components, the basic writing unit, the character, does not represent precise phonetic information. Because of such a logographic nature, there is no easy way to retrieve the pronunciation of the characters. The phonological categories in a Chinese rhyme dictionary will not reveal any phonetic information. For example, in the Qièyùn, the first two rhymes, dōng 東 and dōng 冬, are just two different categories. As the labels of these two rhymes, dōng 東 and dōng 冬, themselves do not provide phonetic values, the phonetic values of these two rhymes must be determined through a process of reconstruction, which is based on the phonetic values of modern dialects and phonetic transcriptions in foreign languages. Therefore, the phonological categories and phonetic values of the Qièyùn are not equally reliable, because the former is given in the dictionary and the latter is reconstructed by scholars in modern times. But some other phonological works may contain both phonological and phonetic information. As another example, take the Ménggǔ Zìyùn 蒙古字韻 of the thirteenth century, which is a comparative list of

An Introduction to Chinese Historical Phonology

53

Table 1.45 A grading of different source materials by six categories in historical phonology

Systematic Categorical information Phonetic information Date Location Standard language

RD

RT

FT

PR

XSC

very good very good no very good good yes†

very good very good no good poor yes†

varies poor very good good poor yes†

poor good no good poor yes†

poor poor no poor no unclear

Note: RD = Rhyming dictionary, RT = Rhyme tables, FT = Foreign transcription (and Sino-xenic pronunciation), PR = Poetry rhyming, XSC = Xiéshēng characters. † Denotes that this is true for most cases.

Table 1.46 The availability of different source materials for the study of the major historical periods

Old Chinese Early Middle Chinese Late Middle Chinese Old Mandarin

RD

RT

FT

PR

XSC

+ + +

+ + +

+ + +

+ + + +

+ -

Note:.RD = Rhyming dictionary, RT = Rhyme tables, FT = Foreign transcription (and Sino-xenic pronunciation), PR = Poetry rhyming, XSC = Xiéshēng characters.

Chinese syllables showing both their Chinese characters and the hP’ags-pa _ script. This work contains the systematic information of both phonological categories and their phonetic values. The Chinese characters are listed according to the categories within Chinese phonology, and each of the categories has its corresponding hP’ags-pa spellings, which are phonetic. _ Different materials vary significantly in these qualities but are often used and discussed together in reasoning and conclusions. Table 1.45 and Table 1.46 roughly summarize the qualities of the source materials; they only show the quality in a general way. It should be noted that different source materials in the same category may also vary. Other types of source materials may also be used for the study of a historical period. The nature of these materials will be introduced when they are mentioned. In general, the available phonological information can be summarized as in Table 1.47. Since the availability and the quality of source materials is very different for different historical periods, different degrees of reliability are expected for

54

The Keys to Traditional Phonology

Table 1.47 A generalization of the phonological information available at each period of Chinese phonological history

Source Phonological category Phonetic value Systematic Standard Date Location

Old Chinese

Middle Chinese

Old Mandarin

reconstruction reconstructed reconstructed yes no no no

Qièyùn given reconstructed yes yes yes no

Ménggǔ Zìyùn given given yes yes yes no

different historical periods, with more recent historical periods usually having more consistent and complete information. 1.7

Periodization

The history of Chinese phonology is usually divided into four periods based on the two most important rhyme dictionaries, the Qièyùn of 601 and the Zhōngyuán Yīnyùn of 1324. Old Chinese is before the Qièyùn and is the earliest period; the Qièyùn represents the beginning of Middle Chinese; the Zhōngyuán Yīnyùn represents the beginning of Old Mandarin; and Modern Mandarin is the period leading up to the present time, with no fixed beginning date. Of course, such a linear presentation is a great deal simplified and far from the real history; the history of Chinese phonology, as well as the history of standard phonology, is not covered by a few successive periods. Combinations of vertical transmission caused by internal system developments and horizontal changes caused by external influences create complex historical layers in a system. When a significant phonological change occurs in a region, it is generally related to demographic events in the region, mainly significant migrations caused by wars and a change of dynasties. It is likely that a phonology will split into two distinct systems due to such relatively rare events, rather than one system gradually transforming into new forms. Many aspects of these complex processes are beyond the purpose of this book; therefore, mainly the standard phonology of different historical stages will be presented in the following chapters. 1.8

Reconstruction

In an effort to determine phonological systems of different historical periods, the method of reconstruction is often employed to determine the phonological

An Introduction to Chinese Historical Phonology

55

categories and the phonetic values of these categories. Phonological reconstruction is the practice of establishing an ancestor language based on its descendant languages, both modern and ancient. In the literature, reconstructed forms are commonly prefaced with an asterisk, to distinguish them from attested forms. The term “reconstruction” is often used in an unprecise and ambiguous way in the context of Chinese historical phonology. It could mean the reconstruction of both phonological categories and their phonetic values, or just the phonetic values of known phonological categories. The reconstruction of Old Chinese phonology is mainly in the sense of the former and the reconstruction of Middle Chinese phonology is the latter. There is no source material that can provide a complete phonological system for the period of Old Chinese, while the Qièyùn contains information about all the phonological categories of Middle Chinese. What needs to be determined or reconstructed is the phonetic values of these existing categories. Karlgren’s (1915–1926) pioneering work Études sur la phonologie chinoise is a good example of this. His reconstruction of the phonetic values of Middle Chinese was a half reconstruction. Such a distinction between “full reconstruction” and “half reconstruction” is necessary because the reliability of the two are clearly different. In the study of Old Chinese both phonological categories and their phonetic values must be reconstructed: in this way, reconstructions of Old Chinese and Middle Chinese are very different in nature. There is no question that the study of the period with information about only phonological categories will be more difficult than the study of the period with both, and the study of the historical period with no systematic information about phonological categories and their phonetic values is more difficult and challenging than the study of the periods with it. In comparison with Old Chinese, Middle Chinese is actually a halfreconstruction for the phonetic values of the existing phonological categories. The difficulty level of phonological studies is inversely proportional to the availability and reliability of the source materials. The reconstruction of phonetic values requires specific knowledge, and the reconstruction of both phonological categories and phonetic values requires a very broad range of very specific knowledge. However, although the difficulty involved in reconstruction work is acknowledged, the “reliability of result” of reconstruction is not the same as the historical materials that can provide direct evidence for the categories and their phonetic values. So it is thus the tradition that reconstructed forms are marked with an asterisk (*), because it is needless to explain that what is reconstructed is not the same as what is given. In this book, only the reconstructed forms of Old Chinese will be marked with an asterisk (*). The phonetic values that are directly obtained from the materials or the reconstructed phonetic values that are based on given phonological categories are not marked.

56

The Keys to Traditional Phonology

1.9

Transcriptions

In publications, various phonetic notations and symbols are used. So, it is necessary to transcribe them in a unified way. In this book, phonetic values are represented in a traditional, usually articulatory, framework; for phonetic transcriptions, the symbols of the International Phonetic Alphabet (IPA) are used. In particular: 1. Semivowels are used for medials and codas rather than vowel letters, for example: kjaw, instead of kiau. 2. The symbol h is used to indicate aspiration of consonants. These symbols all appear regular rather than superscript, for example: khjaw, instead of kʰjaw. 3. According to the common usage in Chinese linguistics IPA symbols tɕ, tɕh, dʑ, ɕ, and ʑ are used for palatal affricates and fricatives. The symbol Ø is used for the so-called “zero initial” and other zero forms. 4. For Middle Chinese, tones are not marked; if they are relevant, they will be referenced either by context or by their tone label appearing next to their transcription (e.g. luŋ qù) Note (1) can be understood by the fact that these vowels and their corresponding glides are only different in whether they are syllabic or not: -i- can be phonologically understood as -ji-, -u- as -wu-, and -y- as -ɥy-. Many classifications and conditioned changes that apply to the syllables with -j-, -w-, and ɥ- will apply to the syllables with -i-, -w-, and -y-, respectively, as well. Because of the important role of Middle Chinese phonology, all historical forms will be provided along with the phonetic values of Middle Chinese and Modern Mandarin. Phonetic information will be provided in IPA, with the Modern Mandarin in pinyin spellings. If the historical forms are Middle Chinese, then only the Modern Mandarin forms will be provided (see Table 1.48). The form in front of the symbol > only indicates the “the form of the previous period,” instead of indicating that the earlier form changed to the later form. Such a definition of notation suggests that the two forms in the different historical periods may or may not have a direct relationship in the development. Table 1.48 An example of the reconstruction format, using the character luò 落 Old Chinese Middle Chinese Old Mandarin

*ɡ·raaɡ lɑk lɑw

> lɑk > luò > luò

> luò

Part II

Old Chinese Shang dynasty (c. 1556–1046 BC), Western Zhou dynasty (1045–771 BC), Spring and Autumn period (771–476 BC), Warring States period (476–221 BC)

Old Chinese refers to the Chinese language before the Qin dynasty (Mandarin: xiānqín 先秦). Unlike later periods, there is no systematic record of an entire phonology. Thus, the phonology of Old Chinese must be completely reconstructed, based on various pieces of unrelated source materials. The reconstruction of Old Chinese represents the most difficult work in the study of Chinese historical phonology. It also should be noted that the source materials often have unclear information regarding the time and geographical locations of the particular dialect or standard that they represent. The source materials mainly include the rhyming words of ancient poetry, the Chinese characters that share phonetic components, the phonology of Middle Chinese, which is based on rhyme dictionaries with complete phonological information, and the pronunciation of modern Chinese dialects. The rhyming words mostly come from the Shījīng 詩經 ‘Book of Odes,’ which is the oldest collection of Chinese poems. Since the sixteenth century, Chinese scholars have been working on the rhyming categories and gradually a widely accepted result has been reached. The rhyming characters of the same rhyme may not rhyme according to later dictionaries. For example, gē 歌 kaal > kɑ, jiā 加 kraal > kɰa, pí 皮 bral > bɰiɛ. The reconstruction of phonetic values for rhyming categories and initials started in the twentieth century, and has been a continuation and refinement of earlier attempts to understand Old Chinese poetry. The poetry rhyming words, however, can only provide information about the rhyming parts of Chinese syllables. The information about initial consonants therefore must come from a different piece of information. Some Chinese characters contain a component that is used to indicate pronunciation. Interestingly, this character and its contained component, which is an independent character (the so-called xiéshēng characters 諧聲字, refer to Section 1.5.3), may have different initial consonants according to the dictionaries of a later period. For example, according to the earliest dictionaries, the pronunciation of character luò 洛 ‘the name of a river’ and its phonetic component gè 各 ‘each’ are lɑk and kɑk, respectively. These two different initials, l- and k-, suggest a 57

58

Old Chinese

possibility that the initials of these two characters must have been more similar at the time when the character 洛 was created. By using knowledge of modern phonetics and genetically related languages, the pronunciations of these two characters are reconstructed as luò 洛 ɡ·raaɡ (change to lɑk), gè 各 klaaɡ (change to kɑk), respectively. These xiéshēng characters can also provide more examples for the limited characters used in rhyming. The study of Old Chinese is one of the most difficult challenges – the phonological system is reconstructed by piecing different sorts of information together, and its results cannot be independently verified. The source materials used for constructing a single phonological system are not of the same time period and not from the same geographical location. As indicated in the Shījīng, the poems were collected from different states and from periods across several hundred years, from the Western Zhou dynasty to the middle of the Spring and Autumn period. Chinese characters were created before the Shījīng over a long period of time and in different geographical locations. Although Old Chinese phonology must appear first according to chronology, it would be very helpful (and is recommended) to read Part III (Middle Chinese) first before coming back to read this part, as linguists studying Old Chinese use Middle Chinese phonology as their main reference point in the reconstruction of phonology. A good understanding of Middle Chinese phonology, and especially the terminologies involved, will benefit the reading of this part.

2

Old Chinese

2.1

Introduction

The period of Old Chinese has no fixed start date. Because of this, scholars have proposed further dividing Old Chinese into smaller periods. Zhèngzhāng Shàngfāng ([2003] 2013), for example, proposed three periods: Early Old Chinese (shànggǔ qiánqī 上古前期), Middle Old Chinese (shànggǔ zhōngqī 上古中期), and Late Old Chinese (shànggǔ hòuqī 上古後期) (see Table 2.1). Because there is no systematic phonological work dating from this period, the phonological system is pieced together with information from different bodies of data. These source materials only provide certain aspects of the phonology and vary greatly in reliability. At its earliest stage, the study of Old Chinese phonology is not much more than a study of rhyming materials of that period of time. Middle Old Chinese draws largely from the Shījīng 詩經 ‘Book of Odes,’ which is believed to have been composed during the Western and early Eastern Zhou dynasty, though the texts were likely edited later. Since the evidence used in reconstruction comes from varied sources, reconstructed phonological forms are usually not clearly identified in space and time. In a more strict sense, Old Chinese is not “a language,” because Old Chinese reconstruction represents information from different centuries and different dialects (Schuessler 2009). In order to reconstruct Old Chinese with a more natural system, scholars often turn to languages related to Chinese for information, a practice that tightly connects the reconstruction of Old Chinese to research on the genetic relationship of the Chinese language with others. Reconstructed forms usually have similar forms in languages historically related to Chinese or Proto-Chinese. Scholars in China continue to insist on a genetic relationship among the Sino-Tibetan language family (Table 2.2), which includes the Sinitic (all Chinese dialects, often including Bai as well) and Tibeto-Burman languages, and depending on the scholar, also may include Tai-Kadai and Miao-Yao (Hmong-Mien) languages. Western scholars include only Tibeto-Burman, Bai, and Chinese in the Sino-Tibetan family. Arguments for and against certain genetic classifications are based on the evidence of cognates, which are a rare 59

60

Old Chinese

Table 2.1 The periodization of Old Chinese associated with different Chinese dynasties Period

Dynastya

Sources

Early Old Chinese

Shang, early Zhou

Middle Old Chinese Late Old Chinese

Zhou Qin, Han, Wei

Oracle bone and bronze inscriptions Pre-Qin texts Foreign transcriptions

a

Shang (Yin) dynasty (1600–1046 BC), Zhou dynasty (1046–256 BC), Qin dynasty (221–206 BC), Han dynasty (206 BC – AD 9, AD 25–220), Wei dynasty (AD 220–265)

Table 2.2 The four language families, and some of their constituent languages, in the Sino-Tibetan language familya Sinitic

Tibeto-Burman

Tai-Kadai

Miao-Yao

Chinese Bai

Tibetan Burmese Qiang Karen

Thai Zhuang Lao Dong (Kam) Kra

Hmong Yao

a

Only the first two are commonly accepted.

occurrence, and sometimes depend heavily on theoretical orientations as well as methodology for identifying the cognates. The limited number of cognates available for analyzing regular patterns of sound change makes the arguments for genetic relationships less convincing. The process of finding related words in different languages and thus identifying a genetic relationship should always be carefully avoided as it can become a feeding process of circular arguments. How many examples and how regular a pattern is sufficient for reconstructed phonology has not been established, though if more than one word can be found to have undergone the same sound change, it seems to be understood and used beyond doubt of accidental similarity. According to the tree model of historical linguistics, if these other languages are indeed genetically related to Chinese, there should be at least some examples of cognates that are shared between all the proposed languages, rather than one group of words showing a relationship with some language A and another group of words showing a relationship with some other language B. Such cognates have not been successfully demonstrated among the Tai-Kadai and Miao-Yao languages.

Old Chinese

2.1.1

61

The Tibeto-Burman Connection

The fact that Chinese is related to the Tibeto-Burman languages has long been recognized, with the most influential classification of Sino-Tibetan languages being the one proposed by Paul Benedict (1972). In China, scholars are still trying to show the relationship between Chinese and the Miao-Yao and TaiKadai languages; however, whether the relationships shown by phonetically and semantically similar words are genetic or the result of language contact has not as yet been determined. Membership of the Sino-Tibetan language family remains controversial, although the relationship between Chinese and Tibetan is certain. Laurent Sagart (1993a) argued that there is a connection between Chinese and Austronesian languages. Many Chinese monosyllables can be connected to the last syllable of polysyllabic words of Austronesian languages, but whether these cognates are of the same quality as cognates between Chinese and Tibeto-Burman languages has been controversial. The phonetic and semantic similarities of the cognates often require ad hoc explanations. Sergej Starostin (1995) did an interesting test of the various proposed hypotheses. He used a list of thirty-five basic words, originally proposed by Sergej Jaxontov, to compare potentially related languages, including Tibeto-Burman, North Caucasian, Austronesian, and Tai. Twenty-four of these Chinese words can be identified as cognates with Tibeto-Burman, thirteen with North Caucasian, and four with Austronesian. However, all these arguments for and against certain models of genetic relationship are based on the family tree hypothesis, which views the historical development of language as a continual splitting process, with ancestral languages splitting into a greater number of languages. Whether such a model is suitable for languages in China, and Asia in general, has not been determined. The greater time depth and complex migration historically involved in this area put the validity of the tree model based on the Indo-European languages into serious question. Language history cannot be a simple splitting process. Throughout history, various types of language contact – especially language shift – create complex linguistic relationships that present problems unresolved by the tree hypothesis. 2.1.2

Methodology and Working Principles

The celebrated comparative method of Indo-European historical linguistics is not used in the reconstruction of Old Chinese due to the following reasons: the availability of rhyming materials of the Zhou dynasty provides existing categorical information; phonological information about Chinese dialects does not allow for reconstruction past the Han dynasty; and there is a possibility that the development of Chinese dialects did not follow the same splitting pattern

62

Old Chinese

seen in Indo-European languages. We can use Middle Chinese as an example: without the Qièyùn 切韻, it is quite unlikely that any reconstruction based on modern dialects would yield phonological categories anywhere approaching the Qièyùn system. (Bernhard Karlgren only used dialect information to determine the phonetic values of existing phonological categories.) If Middle Chinese could not be accurately reconstructed with this method, the reconstruction of Old Chinese would be even more difficult. The identification of genetic relationships between Chinese and other languages is deeply rooted in methodology and hypotheses developed in the historical linguistic field of Indo-European languages. Because the reconstruction of Old Chinese is still a work in progress, it is necessary to describe the steps leading to each conclusion. In addition, many of the proposals remain controversial, and for this reason works by particular scholars must be often referenced. In this way, the reconstruction of various aspects of Old Chinese phonology continues to be adjusted and improved upon. In order to reduce possible confusion, the same reconstructed system will be used throughout in this book. If the author is not mentioned, then the Old Chinese is based on Zhèngzhāng’s recent reconstruction ([2003] 2013) and the Qièyùn system is based on Zhèngzhāng and Pān’s reconstruction (Pān 2000). If other scholars’ works are referenced, the phonetic symbols used are the same as used by the original authors. In tables, along with the referenced reconstructions, Zhèngzhāng’s reconstruction will appear in parentheses. The xiéshēng characters 諧聲字, another important source material, provide linguists with many seemingly unsolvable puzzles, with numerous pieces of evidence not making any sense from a phonological viewpoint. In such cases, morphology, especially inflectional morphology, is used to come to a satisfactory answer to the questions brought up by xiéshēng characters. In recent studies much effort has been devoted to this line of research and a great deal of progress has been made. 2.2

Source Materials

Old Chinese is mainly based on language materials from the early to mid-Zhou dynasty. Earlier materials include oracle bone inscriptions and inscriptions on bronze vessels from the Shang dynasty; however, for phonological analysis the Shang materials have very restricted usage because of the limited number of legible characters and the lack of rhyming texts. Based on identified xiéshēng characters 諧聲字, some phonological information can be extrapolated (Hé [1998] 2002; Mei 2008), although the evidence is very sporadic and can only be used to verify information derived from other materials. Language materials from the Zhou dynasty, on the other hand, are rich enough for scholars to reconstruct complete phonological systems.

Old Chinese

63

The most useful materials are xiéshēng characters (which share phonetic components) and the rhyming words of poems collected in the Shījīng 詩經. It should be noted, however, that both sources have shortcomings: the rhyming words can only reveal information concerning part of a syllable, and although xiéshēng characters can be used to study the phonological information of entire syllables, the characters were created in different time periods, with many of them earlier than the poetry in the Shījīng. 2.2.1

Shījīng 詩經 Rhyming

The Shījīng is the oldest collection of poems, or lyrics of songs, in Chinese history. There are 305 poems in total. Confucius probably participated in the editing of the book, thus it likely reflects the language of his time (551–479 BC). Rhyming is evident in all the poems of the Shījīng; for example, the poems in Table 2.3 (from the Zhōu Nán 周南 ‘Southern Zhou’ section) demonstrate a variety of rhyme schemes, and the resultant reconstructions are in accordance with the rhyme scheme (based on Zhèngzhāng’s reconstruction [[2003] 2013]). The rhyming words may or may not rhyme according to modern Chinese dialects, but because of the rhyming pattern, the syllables can be organized into rhyming groups. In the phonological analysis, rhyming materials from other pre-Qin texts can also be utilized. 2.2.2

Xiéshēng Characters 諧聲字

Xiéshēng characters provide another important source of information. One of the ways in which Chinese characters are created is to combine two components, one indicating meaning and another indicating sound. These types of characters are called xíngshēng zì 形聲字 ‘phonetic compound characters.’ For example, jiāng 江 ‘river’ has two components. The component on the left is the radical for shuǐ 水 ‘water,’ indicating the semantic category, that is, the word is something related to water; and the component on the right is gōng 工 ‘work, labor,’ indicating the phonetic similarity. The character gōng 工, along with other characters sharing the same phonetic component, form a xiéshēng series. Because a large number of Chinese characters are created this way, there are quite a few xiéshēng series in existence. See Table 2.4 for some examples. Among the xiéshēng series, some phonetic relationships are easier to see and understand than others. The assumption, which started with Duàn Yùcái 段玉 裁 (1735–1815), is that when these characters were created their phonetic component must have been similar. Since the rhyming characters in the Shījīng are limited, the shared phonetic components of xiéshēng characters can provide

64

Old Chinese

Table 2.3 Some poems from the Shījīnga Old Chinese Guānjū 關雎, stanza I (AAXA) 關關雎鳩 A guān guān jū jiū 在河之洲 A zài hé zhī zhōu 窈窕淑女 X yǎo tiǎo shū nǚ 君子好逑 A jūn zǐ hǎo qiú

Middle Chinese

Modern Mandarin

*ku *tju *naʔ *gu

> > > >

kiw tɕiw ɳjɔ ɡiw

> > > >

jiū zhōu nǚ qiú

*mrɯɯ *rɯɯ *rɯɯ *snɯ

> > > >

mɰæj ləj ləj sɨ

> > > >

mái lái lái sī

*ʔsjaa *rteeŋ

> >

tsja ʈɰæŋ

> >

jū zhēng

*pa *djeŋ

> >

pjo dʑjɛŋ

> >

fū chéng

The fishhawk sings gwan gwan on sandbars of the stream. Gentle maiden, pure and fair, fit pair for a prince. Zhōng fēng 終風, stanza II (AAAA) 終風且霾 A zhōng fēng qiě mái 惠然肯來 A huì rán kěn lái 莫往莫來 A mò wǎng mò lái 悠悠我思 A yōu yōu wǒ sī First the wind, then blowing dust, would you be kind and come? For if you do not visit me, longing lasts on and on. Tùjū 兔罝, stanza I (ABAB) 肅肅兔罝 A sù sù tù jū 椓之丁丁 B zhuó zhī zhēng zhēng 赳赳武夫 A jiū jiū wǔ fū 公侯干城 B gōng hóu gān chéng The rabbit snare has mesh so fine, with a thump we knock its pegs in. A staunch and fearless warrior, our lord duke’s shield and bastion. a

All English translations are by Owen (1996).

important additional information, including phonological information concerning the non-rhyming part of the syllables (which cannot be retrieved from Shījīng rhymes). It has become clear that the main phonetic similarity between xiéshēng characters is that they must be similar in terms of the onset and the rhyme (Zhèngzhāng [2003] 2013: 55). Shěn Jiānshì’s 沈兼士 (1945) Guǎngyùn Shēngxì 廣韻聲系 is an essential reference for the study of xiéshēng

Old Chinese

65

Table 2.4 Examples from the gōng 工, chéng 成, mén 門, jiān 監, and měi 每 xiéshēng series 工 gōng 成 chéng 門 mén 監 jiān 每 měi

攻 gōng 城 chéng 悶 mēn 檻 jiàn 梅 méi

功 gōng 誠 chéng 捫 mén 鑑 jiàn 霉 méi

紅 hóng 盛 shèng 聞 wén 藍 lán 海 hǎi

虹 hóng 晟 shèng 問 wèn 覽 lǎn 悔 huǐ

Note: Each row’s characters have the same component.

characters. However, the accuracy of the information retrieved from the rhyming and xiéshēng materials should always be questioned. Similar materials from modern language suggest that although there is a very strong tendency toward similarity in pronounciation of xiéshēng characters, their accuracy cannot be completely reliable (Chén Qíguāng 2007: 198–214, 215–233). This is to say, characters that have the same phonetic components may not have sounded similar, even historically, and so xiéshēng characters are more useful as tools of confirming hypotheses based on rhyme, rather than vice versa. 2.2.3

Partially Reduplicated Words

In Old Chinese, there are different types of reduplicated words, which have been the subject of study by a number of scholars (Baxter & Sagart 1998; Sūn Jǐngtāo 2008). The disyllabic forms of partially reduplicated words (called liánmián cí 聯綿詞) usually show a certain phonological relationship between syllables. The first three words in Table 2.5 can be categorized into one group, and the last three into a second group. The words in the first group share the same initial and a similar final. It seems that the final of the first syllable does not have a rounded feature, but the second syllable does. The words in the second group have the same final but different initials. The initial of the second syllable is always a liquid l- or r-. Clearly these reduplications are generated according to the syllable structure and phonological rules. Thus this type of morphophonological process is very informative for the reconstruction of phonological relationships and checking the results of various reconstructions. 2.2.4

The Qièyùn 切韻 System

The rhyme dictionary Qièyùn of 601 is the earliest complete record of a Chinese phonological system. In the tenth century the phonological

66

Old Chinese

Table 2.5 Examples of partially reduplicated words

輾轉 踟躕 契闊 蜉蝣 螳螂 螟蛉

zhǎnzhuǎn chíchú qìkuò fúyóu tángláng mínglíng

‘repeated’ ‘hesitate’ ‘distant’ ‘mayfly’ ‘mantis’ ‘corn earworm’

Pulleyblank

Li

Zhèngzhāng

*trànʔ trwànʔ *dràj drà *khát khwát *də̀w lə̀w *dáŋ ráŋ *máŋj ráŋj

*trjanx trjuanx *drjig drjug *khiat khwat *djəɡw rjəɡw *daŋ laŋ *miŋ liŋ

*tenʔ tonʔ *de do *kheed khood *du lu *daaŋ raaŋ *meeŋ reeŋ

Source: Pulleyblank and Li’s reconstructions are from Sūn Jǐngtāo 2008 and Zhèngzhāng’s are from Zhèngzhāng [2003] 2013.

categories were later organized into rhyme tables, where special structural terminologies were employed to classify the sound categories of the Qièyùn. Scholars commonly use these terminologies to determine the characteristics of the rhymes of the Qièyùn and also use them in their discussion of Old Chinese phonology. Some of the structural terminologies will be introduced here: shēngmǔ 聲母 yùnmǔ 韻母 yùn 韻 děng 等 kāihé 開合

shēngdiào 聲調

The initial consonant of a syllable The part of a syllable after the initial The rhyming part of a syllable, usually referring to the segments from the main vowel on Types of yùn, split into four Ranks in Middle Chinese Types of yùn, kāikǒu 開口 and hékǒu 合口 indicate whether the main vowel is unrounded or rounded and the absence or the presence of the medial -w-, with kāi 開 representing the absence of -w- and hé 合 representing the presence of -wAlso called shēng 聲, or tones, distinctive suprasegmental tonal contours.

It should be noted that these explanations are in reference to Middle Chinese. For a more detailed explanation, refer to the relevant sections on Middle Chinese (Section 3.4.3, Section 3.5.5, Section 3.5.8, Section 3.6). 2.3

Reconstruction

The reconstruction of Old Chinese is an attempt to explain all the relevant data according to the economic principle, meaning that the goal is to minimize the required syllable types while maximizing coverage of the relevant data. During this process the naturalness of the reconstruction is also often checked against genetically related languages.

Old Chinese

67

Table 2.6 Two analyses of Chinese syllable structure Term

2.3.1

Constituents

Ternary branching Syllable (S) Final (F)

> Tone (T), Initial (I), Final (F) > Medial (M), Main vowel (V), Coda (E)

Binary branching Syllable (S) Segment cluster (S*) Final (F) Rime (R)

> > > >

Tree

Tone (T), Segment cluster (S*) Initial (I), Final (F) Medial (M), Rime (R) Main vowel (V), Coda (E)

The Reconstruction of Syllable Structures

Modern Chinese is a monosyllabic language, which means that most morphemes are one syllable in pronunciation. In Old Chinese not only most morphemes but also most words are monosyllabic, which makes it natural to start the phonological analysis from single words (or single syllables). The structure of modern Chinese syllables can be analyzed in three basic units: tone, initial, and final. The final can be further divided into medial, main vowel, and coda.1 Such a syllable structure (as seen in Table 2.6) is only applicable from the sixth century on. The developing reconstruction of Old Chinese demands more and more structural slots for Old Chinese syllables. In recent proposals, monosyllabic morphemes are not enough for explaining newly discovered evidence of the possible relationships between Chinese and other languages. 1

This structure has five basic units, in comparison to the common American and European analysis of syllables as three parts: onset (O), nucleus (N), and coda (C). Initial and medial are members of the onset, while the ending is a member of the coda, and the main vowel is a member of the nucleus. The tone is analyzed as a suprasegmental and cannot belong to any of the three aforementioned parts. See Section 1.2.4 for a more thorough discussion.

68

Old Chinese

Table 2.7 A common analysis of an Old Chinese syllable Syllable

>

Onset cluster, main vowel, coda cluster Onset cluster Main vowel Coda cluster

> > >

Pre-initial segment, initial stem, post-initial segment Main vowel Coda, post-coda segment

Table 2.8 Reconstructions exemplifying the varied syllable structure of Old Chinese, with as many as six segments CCCVCC 頭 父 大 家 馬 位 車 獸 少

tóu fù dà jiā mǎ wèi chē shòu shǎo

‘head’ ‘father’ ‘big, great’ ‘home’ ‘horse’ ‘location’ ‘cart’ ‘animal’ ‘few, less’

*doo *baʔ *daals *kraa *mraaʔ *ɢʷrɯbs *klja *qhljus *hmjewʔa

CV CVC CVCC CCV CCVC CCVCC CCCV CCCVC CCCVCC

Source: Zhèngzhāng [2003] 2013. a In reconstructions where hm represents a voiceless m, this structure is actually CCVCC. See Section 2.3.2.1 for more detail.

Some reconstructions employ a half syllable in front of the main syllable (see Section 2.3.5.1). According to recent reconstructions (Baxter 1992; Pān 2000; Zhèngzhāng 1987, [2003] 2013), the syllable structure can be divided into three basic components: main vowel (V), pre-V segments, and post-V segments (see Table 2.7 and Table 2.8). The pre-V segments are the onset of a syllable, traditionally called shēngmǔ 聲母 ‘initial,’ and the post-V segments, traditionally called yùnwěi 韻尾 ‘coda,’ are the coda. Recent research suggests that both the pre-V and post-V parts of the syllable require more than one structural slot. There is also a strong tendency in reconstruction for tones to not be treated as independent features but rather as aspects of the pre-V and post-V segments, which become tones through tonogenesis and rephonologization. Whether Old Chinese words are strictly monosyllabic has been questioned (Pān 1999, 2000; Baxter & Sagart 2014). In comparison with modern Chinese and Middle Chinese, the languages related and/or geographically adjacent to Chinese (such as the Tibeto-Burman and Tai-Kadai languages) have more complex syllable structures and are not strictly monosyllabic. The increase of structural slots in and beyond a single syllable makes Old Chinese syllables

Old Chinese

69

more like other genetically related languages and allows the phonological reconstruction more possibilities and flexibility. 2.3.2

The Reconstruction of Onsets

The reconstruction of Old Chinese is based on information from Middle Chinese. The information from xiéshēng characters 諧聲字 and the Qièyùn 切韻 strongly indicates that the initial system of Old Chinese must be significantly different. 2.3.2.1 Simple Initials The initials of Old Chinese are reconstructed as much more complex than those of Middle Chinese and can consist of either a single consonant or a consonant cluster. Scholars have proposed a number of initial systems in their reconstructions of Old Chinese. It is thus necessary to list a few of them to see the similarities and differences. Tables 2.9–2.12 illustrate a few different reconstructions of simple initials, that is initials which are allowed to be alone in the onset cluster. Because these tables illustrate simple initials, it must be understood that multiple character sequences represent only one sound. The h+consonant combinations (hm, hn, hŋ, hŋw, hj, hl, hr) are voiceless counterparts to the voiced sonorant (m, n, ŋ, ŋw, j, l, r). These combinations change to their sonorant or -h- form in different contexts/dialects, so it is reasonable to treat them as clusters as well. Zhèngzhāng ([2003] 2013) views these h+consonant combinations as clusters, while Baxter and Sagart (2014)

Table 2.9 Li Fang-Kuei’s reconstruction system (thirty initials) p t ts k kw a

ph th tsh kh khw

d d dz ɡ ɡw

hm hn

m n

hŋ hŋw

ŋ ŋw

hl s h hw

l, r Ø Øwa

Because a null initial is not voiced, the resulting sound is /w/.

Table 2.10 Baxter’s reconstruction system (thirty-seven initials) p t ts k kw

ph th tsh kh khw

b d dz ɡ ɡw

hm hn

m n

hŋ hŋw

ŋ ŋw

w j s h

hw hj z ɦ

l ʔ ʔw

hl

r

hr

70

Old Chinese

Table 2.11 Zhèngzhāng’s reconstruction system (twenty-five initials)a p t s k q/ʔ a

ph th sh/tsh kh qh/h

b d z/dz ɡ ɢ/ɦ

m n

mh nh

ŋ

ŋh

l r

lh rh

The pairs of initials separated by / indicate early and late forms.

Table 2.12 The nineteen universal initials seen throughout Old Chinese reconstructions p t ts k

ph th tsh kh

b d dz

m n ŋ

l s h

ɦ

ʔ

view them as voiceless equivalents without any realized [h] present. The -wfollowing gutturals represents labialization of these sounds. Li Fang-Kuei (1980: 21) proposed a reconstruction system with thirty initials (Table 2.9). The system of thirty-seven initials that William Baxter (1992) proposed (Table 2.10) was similar to Li Fang-Kuei’s, though with more initials. And, as shown in Table 2.11, Zhèngzhāng ([2003] 2013: 70) proposed a system of twenty-five initials. He treats the -w- of kw- as a post-initial element. According to the initial system of Middle Chinese and the phonological gaps, nineteen initials can be determined as basic initials inherited from Old Chinese because they can combine with all different types of finals; as a result, in most of the reconstructions these nineteen initials (as shown in Table 2.12) are included. The reconstruction of cluster initials is much more complicated. The number and type vary greatly among proposals (refer to Baxter & Sagart 2014). 2.3.2.2 Cluster Initials The study of the initial consonant clusters is the most challenging aspect of the reconstruction of Old Chinese and as such has attracted the interest of many top scholars. Most Chinese characters with the same phonetic components are quite similar. For example, the characters with phonetic component bā 巴 all have bilabial stop initials (see Table 2.13). There are a number of xiéshēng characters with the same or similar finals but different initials, such as gè 各 gé 格 gé 閣 and luò 洛 luò 落 luò 絡, jiān 監 jiàn 檻 jiàn 鍳, and lán 藍 lán 覽 làn 濫. Their phonetic contrast in the Middle

Old Chinese

71

Table 2.13 Some characters with component bā 巴, categorized by Middle Chinese initial p-

ph-

b-

笆豝疤把靶



爬琶杷耙

Table 2.14 Xiéshēng characters from the gè 各 and jiān 監 components 各 kɑk 監 kɰam

格 kɑk 檻 ɦɰam

閣 kɑk 鍳 kɰam

洛 lɑk 藍 lɑm

落 lɑk 覽 lɑm

絡 lɑk 濫 lɑm

Chinese system is even more obvious (see Table 2.14, which illustrates a divide among initials in Middle Chinese). Such a phenomenon leads scholars to believe that these initials developed from Old Chinese consonant clusters. As early as in the nineteenth century, British sinologist Joseph Edkins (1876) suggested that the relationship between characters with l- and characters with other consonants comes from Cl- consonant clusters. Based on Edkins’ idea, Karlgren (1923) raised his theory that the xiéshēng relationship between gè 各 and luò 絡 is evidence for ancient consonant clusters. The following year, Chinese scholar Lín Yǔtáng (1924) proposed kl-, pl-, and tl- clusters with more evidence from ancient documents, including disyllabic monomorphemic words (liánmián cí 聯綿詞, as seen in Section 2.2.3), alternative readings, and xiéshēng characters. Chén Dúxiù (1937) found more evidence in Chinese dialects and genetically related languages. With the effort of several generations of scholars (F-K. Li 1971; Baxter 1992; Pān 2000; Zhèngzhāng [2003] 2013), the reconstruction of consonant clusters has advanced. In order to explain some irregular xiéshēng relationships, the types of consonant clusters became increasingly complicated, although more rules have been proposed to restrict the type and order of consonants in the clusters. It is also worth noting that research on consonant clusters over the last few decades has gradually developed from pure phonological reconstruction into morpho-phonological reconstruction, also taking into account research on the genetic relationships between Chinese and other languages. According to Zhōu Fǎgāo’s statistics (1970), xiéshēng characters showing a relationship between different initial groups (mainly in terms of place of articulation) make up less than ten percent of all Chinese characters. Dǒng Tónghé’s Shànggǔ Yīnyùn Biǎogǎo 上古音韻表稿 ([1944] 1948) provides a list of irregular xiéshēng relationships.

72

Old Chinese

2.3.2.2.1 Initial Clusters with lCl- and Cr- consonant clusters are the most common type in many languages. Karlgren (1915–1926) proposed three possible reconstructions. For a pair 各 k- and 洛 l-, there are three equally possible solutions, (1) 各 kl- vs. 洛 l-; (2) 各 k- vs. 洛 kl-/ɡl-; (3) 各 kl- vs. 洛 ɡl-. Jaxontov (1960) made a very important breakthrough in the study of xiéshēng characters involving consonant l-. The important observations he made included: (1) in Middle Chinese the l- syllables are very rare in Rank-II (there are only three: lěng 冷 luò 犖 liǎn 醶), (2) many characters having a xiéshēng relationship with characters with lbelong to Rank-II (監 k-, 柬 k-, 數 ʂ-, 灑 ʂ-), and (3) some characters have dual readings, one with l- initial and the other pronunciation belonging to Rank-II (鬲 lek/kɣɛk, 樂 lɑk/ŋɰɛu, 率 liuɪt/ʂʷit). He concluded that all Middle Chinese Rank-II syllables had a -l- after the initial consonant (e.g. kl- pl- ml-). This explains the lack of syllables with l- in Rank-II, as these Rank-II syllables would have initial l- followed by another -l-. Jaxontov also provided cognates in related languages to support his reconstruction (see Table 2.15 for some examples). Li Fang-Kuei (1971) proposed that Rank-II syllables had a -r- instead of -l-. He suggested the relationship given in Table 2.16 for xiéshēng characters gè 各, gé 格, kè 客, luò 洛, and luè 略, with Rank-II ones having an -r-. Li FangKuei’s proposal that Rank-II has an -r- is now widely accepted. His proposal was based on the retroflexization of the Middle Chinese initial zhī 知 and zhào

Table 2.15 Jaxontov’s reconstruction of several words with their cognates 八 百 甲 江

bā bǎi jiǎ jiāng

‘eight’ ‘hundred’ ‘armor’ ‘river’

*plet (*preed)a *plɑk (*praag) *klap (*kraab) *kloŋ (*krooŋ)

Tibetan: Tibetan: Tibetan: Thai:

brgyad brgya khrab khloŋb

a

Zhèngzhāng’s reconstructions ([2003] 2013) are provided in parentheses. Jaxontov cited that the meaning of the Thai word khloŋ is actually ‘canal,’ though the meanings are close enough that it is viable that the meaning would change over time.

b

Table 2.16 Li Fang-Kuei’s reconstruction of gè 各 component xiéshēng characters 各 客 洛

‘each’ ‘guest’ ‘river name’

*klak (*klaaɡ) *khrak (*khraaɡ) *ɡlak (*ɡraaɡ)



‘grid’

*krak (*kraaɡ)



‘strategy’

*ɡljak (*ɡraɡ)

Old Chinese

73

Table 2.17 A summary of Li Fang-Kuei’s explanation of the -r- medial being responsible for the creation of the retroflex series Old Chinese *tr *thr *dr *nr *tsr *tshr *dzr *sr *zr

Middle Chinese > > > > > > > > >

ʈʈhɖɳtʂtʂhdʐʂʐ-

知 徹 澄 娘 照 穿 床 審 禪

zhī chè chéng niáng zhào chuān chuáng shěn chán

照 initial series (see Table 2.17, which also includes the initial shàn 禪 ʐ-). He reasoned the following: These initials must be followed by a set of medials, which cause the retroflexization. Since we already reconstructed an initial *r-, this *r- can be readily used as the medial after these initials, so we can have the following changes. Old Chinese *tr-, *thr-, *dr-, *nr- > Middle Chinese 知 zhī ʈ-, 徹 chè ʈh-, 澄 chéng ɖ-, 娘 niáng ɳ- Old Chinese *tsr-, *tshr-, *dzr, *sr- > Middle Chinese 照 zhào tʂ-, 穿 chuān tʂh-, 床 chuáng dʐ-, 審 shěn ʂ-. (F-K. Li 1971: 15, translation mine)

This proposal was a significant improvement for the entire system, because it solved several problems and provided more convincing explanations in the initials, medials, and rhymes systematically. More evidence for -r- has been shown by Zhèngzhāng (1987) and Xǔ and Pān (1994). The increase in both quantity and quality of the evidence improves scholars’ confidence in their reconstructions. 2.3.2.2.2 Initial Clusters with sOf the xiéshēng characters, the relationships between voiceless fricatives and nasals are quite common. Scholars have proposed different types of relationships between the relevant xiéshēng characters. Jaxontov (1960) suggested that these are the results of sC- clusters (C = nasal or lateral) (Table 2.18). In his 1980 proposal, Li Fang-Kuei suggested that only Middle Chinese alveolar and palatal consonants are related to the s- prefix in Old Chinese. Only the initial parts of the syllables are provided in Li’s reconstruction (Table 2.19). Because of the sporadic nature of the data, different proposals have different sC- types and result in different Middle Chinese initials; it becomes almost impossible to judge their validity. In order to make sC- clusters result in a more

74

Old Chinese

Table 2.18 Jaxontov’s reconstructions of sC- clusters across different nasals and approximants *sm- > xʷm- > x(ʷ) 忽 *sm(*hm-) 徽 *sm(*hm-)

‘suddenly’ ‘insignia’

勿 *m微 *m-

(*m-) (*m-)

‘do not’ ‘tiny’

*sŋ- > xŋ- > x 羲 *sŋ(*hŋr-)

‘surname’

義 *ŋ-

(*ŋr-)

許 *sŋ-

‘permit’

午 *ŋ-

(*ŋ-)

‘justice, meaning’ ‘noon’

*sn- > thn- > th, ʈh 態 *sn(*nh-) 恥 *sn(*nh-)

‘attitude’ ‘shame’

能 *n耳 *n-

(*n-s) (*nj-)

‘able’ ‘ear’

*sɲ- > ʃɲ- > ʃ 饟 *sɲ(*hnj-) 恕 *sɲ(*hnj-)

‘soldier’s pay’ ‘forgive’

釀 *ɲ女 *ɲ-

(*n-) (*n-)

‘ferment’ ‘female’

*sl- > ʂl- > ʂ 使 *sl(*sr-) 森 *sl(*sr-)

‘messenger’ ‘dense forest’

吏 *l林 *l-

(*r-) (*ɡr-)

‘minor official’ ‘grove of trees’

(*hŋ-)

Table 2.19 Li Fang-Kuei’s initial reconstruction of cì 賜 ‘bestow’ and xiē 楔 ‘wedge’ as alveolar and palatal stop consonant initials, respectively 賜 *st- > s楔 *sk- > s-

(*sl-) (*sqh-)

‘bestow’ ‘wedge’

(剔 *th- > th-) (契 *kh- > kh-)

regular change, Zhèngzhāng (1990) suggested that sC- be separated from hC-, with sC- clusters resulting in Middle Chinese ts- and tʂ- type initials, and hCclusters resulting in Middle Chinese h- type initials. Zhèngzhāng’s reconstruction differentiated sC- and hC- clusters, explaining different coronal and dorsal reflexes, respectively (Table 2.20). 2.3.2.2.3 hN- Clusters or Voiceless Nasals The relationships between nasals and velar or pharyngeal fricatives have long attracted scholars’ attention. Karlgren (1933) reconstructed *xm- for hēi 黑 and huǐ 悔 (hēi 黑 h- / mò 墨 m-, huǐ 悔 h- / měi 每 m-). Dǒng ([1944] 1948)

Old Chinese

75

Table 2.20 Zhèngzhāng’s reconstruction of sC- and hC- clusters sC- > s小 xiǎo 綏 suí 穌 sū 賜 cì

*smewʔ *snul *sŋaa *sleeɡs

‘small’ ‘pacify’ ‘revive’ ‘bestow’

緲 餒 魚 易

hC- > h悔 huǐ 漢 hàn 許 xǔ

*hmɯɯʔ *hnaans *hŋaʔ

‘regret’ ‘name of a river’ ‘permit’

每 měi 難 nàn 午 wǔ

miǎo něi yú yì

*mewʔ *nuulʔ *ŋa *leeɡs

‘indistinct’ ‘hungry’ ‘fish’ ‘change’

*mɯɯʔ *naans *ŋaaʔ

‘every’ ‘disaster’ ‘noon’

Table 2.21 Li Fang-Kuei’s reconstruction of voiceless nasals and laterals 悔 忽 嘆 丑 許 羲 獺 體

*xm- > x*xm- > x*xn- > th*xɳ- > ʈh*xŋ- > x*xŋ- > x*xl- > th*xl- > th-

(*hmɯɯʔ) (*hmɯɯd) (*nhaans) (*nhuʔ) (*hŋaʔ) (*hŋral) (*rhaad) (*rhiiʔ)

‘regret’ ‘suddenly’ ‘sigh’ ‘clown’ ‘permit’ ‘surname’ ‘otter’ ‘body’

每 勿 難 扭 午 義 賴 禮

*m- > m*m- > m*n- > n*ɳ- > ɳ*ŋ- > ŋ*ŋ- > ŋ*l- > l*l- > l-

(*mɯɯʔ) (*mɯd) (*naans) (*nuʔ) (*ŋaas) (*ŋrals) (*raads) (*riiʔ)

‘every’ ‘do not’ ‘disaster’ ‘twist’ ‘noon’ ‘righteousness’ ‘rely’ ‘rite’

revised *xm- to a voiceless nasal m̥- (written mh- in Zhèngzhāng’s reconstructions). Jaxontov (1960) proposed sC- clusters that include the n- and threlationship such as nán 難 and tān 灘. Different from Jaxontov’s proposal but similar to Dǒng’s proposals, Li Fang-Kuei (1971) proposed a set of voiceless nasals xN- (hN-) and laterals xl- (hl-) to account for the xiéshēng relationships in a more systematic way. As seen in Table 2.21, while voiceless labial and velar sonorants undergo debuccalization, voiceless coronal sonorants become plosives. Li Fang-kuei’s reconstruction implies conditioned changes determined by place of articulation. Zhèngzhāng suggests that Li’s xC- clusters (which Zhèngzhāng analyzes as hC-) should only exist for syllables that become hin Middle Chinese. In his reconstruction he does not include *xn- > th- and *xl- > th-, but constructs *nh- > th- and *rh- > th- instead (see the examples tàn 嘆, chǒu 丑, tǎ 獺 and tǐ 體 in Table 2.21). The rules for change can be constructed as: hN > h Ch > th.

76

Old Chinese

Baxter (1992) stated his working principle for the relationship between xiéshēng characters: In order to be written with the same phonetic element, words must normally have identical main vowels and codas, and their initial consonants must have the same position of articulation. (Additionally, nasal and obstruent initials are generally kept separate.) Otherwise pre-initial, medial and post-coda elements, and the manner of articulation of the initial, may differ. (p. 348)

It should be noted that Baxter’s approach is totally based on phonological information and therefore has nothing to do with morphological process. This is the same approach as that of Li Fang-Kuei (1971). 2.3.2.3 Post-Stem Elements 2.3.2.3.1 The Post-Stem -r- and Middle Chinese Rank-II Syllables As shown in Section 2.3.2.2.1, the study of -l- advanced significantly when Jaxontov (1960) connected it for the first time to the Middle Chinese Rank-II syllables with very convincing evidence. There is almost no Middle Chinese Rank-II syllable with l- initial, leading scholars to believe that the special medial or post-initial element must be an l- like liquid consonant, explaining the lack of syllables with l- initial; otherwise there would be two l’s in a row or an initial l- followed by a liquid. However, most syllables having xiéshēng connection with syllables with l- initial are from Rank-II. Jaxontov provided three types of examples showing such a connection (see Table 2.22): type 1 shows that the characters of the phonetic component are syllables with l-, type 2 shows that the characters of the phonetic compounds are syllables with l-, and type 3 shows the dual pronunciation of the same characters, one of which is a syllable with l-. Jaxontov also showed that -l- in Rank-II syllables also connects etymologically to syllables with liquid -l- or -r- in other languages, which can also be seen

Table 2.22 Jaxontov’s examples of three types of syllables with l- and syllables from Rank-II with other initials Type 1 Division-II lType 2 Division-II lType 3 Division-II l-

k- 膠 l- 翏

ʂ- 數 l- 婁

pl-

剝 彔

m- 麥 l- 來

k- 柬 l- 闌

k- 監 l- 濫

m- 卯 l- 柳

kl-

降 隆

k- 鬲 l- 鬲

ŋ- 欒 l- 欒

m- 龍 l- 龍

ʂl-

率 率

Old Chinese

77

Table 2.23 Jaxontov’s examples of syllables with l- and syllables from RankIIIb with other initials Division IIIb p- 變 biàn k- 禁 jìn k- 兼 jiān k- 京 jīng ph- 品 pǐn m- 文 wén ll- 䜌 luán l- 林 lín l- 廉 lián l- 涼 liáng l臨 lín l- 吝 lìn

in Table 2.22. Scholars have now reached a consensus that there was an -r- in Old Chinese that became Middle Chinese Rank-II syllables. The weakened version of this -r- is still preserved in Middle Chinese phonology. 2.3.2.3.2 The Post-Stem -r- and Middle Chinese Rank-IIIb Syllables Such a connection as that found between l- syllables and Rank-II syllables is also found between syllables with l- and Middle Chinese Rank-IIIb syllables. Jaxontov discovered and showcased some of these Rank-IIIb and l- xiéshēng characters (see Table 2.23). This discovery provides a very important clue to understanding the so-called chóngniǔ 重紐 syllables (the contrast between Rank-IIIb and IIIa syllables, see Section 1.4.6) and the phonetic value of IIIb syllables in Middle Chinese phonology. This proposal has now been widely accepted. 2.3.2.3.3 Structure of Initial Consonant Clusters The relationships between xiéshēng characters are more complicated than a few simple patterns. A serious question that gradually becomes inevitable is how many types of consonant clusters should be reconstructed. The early proposals for Cl-, sC-, and hN-2 types are explicitly or implicitly based on the naturalness of the clusters and were commonly found in various languages; however, additional types are required to explain more complex relationships in the xiéshēng data. In recent research extra attention has been paid to genetically related and geographically adjacent languages. It has been proposed that consonant clusters must have an internal structure. A cluster must have its stem and other consonants can serve as either pre-stem or post-stem elements (Zhèngzhāng [2003] 2013). Based on genetically related languages, Zhèngzhāng proposes only four consonants for the post-stem slot: -w-, -j-, -r-, and -l-. For the pre-stem slot, five types of consonants can appear: fricatives, laryngeals, nasals, liquids, and stops (Zhèngzhāng [2003] 2013).

2

According to Li Fang-Kuei (1980), h + sonorant combinations are also clusters.

78

Old Chinese

2.3.3

The Reconstruction of Rhymes

Early studies of Old Chinese phonology focus mostly on the rhyme groups (yùnbù 韻部). The work started as a branch of traditional philology attempting to understand the rhyming in the Shījīng 詩經 and other classics. It was carried out in large part by the Qian-Jia school (qián jiā xuépài 乾嘉學派) of scholars, mostly from the end of the Ming and through the Qing dynasties. Many revisions have been proposed and the results have more or less reached an optimum stage. Among them the major revisions have been done by Gù Yánwǔ 顧炎武 (1613–1682), Jiāng Yǒng 江永 (1681–1762), Duàn Yùcái 段玉裁 (1735–1815), Dài Zhèn 戴震 (1724–1777), Kǒng Guǎngsēn 孔廣森 (1752–1786), and Wáng Niànsūn 王念孫 (1744–1832). After Wáng Lì’s division of wēi 微 from zhī 脂, modern phonologists of Old Chinese have more or less come to a consensus concerning the thirty yùnbù (for more details about various scholars’ categories, see Wáng Lì 1985 or Baxter 1992). Traditionally each rhyme group is labeled with a representative character from the rhyme group itself, usually characters of the rhymes in the Guǎngyùn 廣韻. There are three types of rhyme group, yīnshēng yùn 陰聲韻, yángshēng yùn 陽聲韻, and rùshēng yùn 入聲韻 (see Section 1.4.10; also see Table 2.24 for some examples). The rhyme groups of yīnshēng yùn and yángshēng yùn are labeled with representative píng 平 tone characters, but the rhyme groups of rùshēng yùn are labeled with representative rù 入 tone characters. 2.3.3.1 The Codas of the Rhyme Groups While the Qièyùn system of Middle Chinese is often used as the major reference for deciding phonological categories, the genetic lineage between Qièyùn and Shījīng rhymes and other materials has not been firmly established. The codas are generally accepted to not have changed much historically, although post-coda elements are responsible for the genesis of tones in Middle

Table 2.24 Some examples of Old Chinese rhyme groups (yùnbù) yīn



yáng

魚 支 侯

鐸 錫 屋

陽 耕 東

Note: The yīnshēng yùn are represented by píngshēng 平聲 characters yú 魚, zhī 支, and hóu 侯, the rùshēng yùn are represented by rùshēng 入聲 characters duó 鐸, xī 錫, and wū 屋, and the yángshēng yùn are represented by píngshēng characters yáng 陽, gēng 耕, and dōng 東.

Old Chinese

79

Table 2.25 The yùnbù of Old Chinese arranged according to three types of codas yīn yáng rù

脂 -j 談 -m 盍 -p 錫 -k

微 -j 侵 -m 緝 -p 藥 -k

宵 -w 真 -n 質 -k

之 -Ø 文 -n 物 -t

幽 -w 元 -n 月 -t

侯 -Ø 蒸 -ŋ 祭 -t

魚 -Ø 冬 -ŋ 職 -k

支 -Ø 東 -ŋ 覺 -k

歌 -j 陽 -ŋ 屋 -k

耕 -ŋ 鐸 -k

Table 2.26 Some of Kǒng Guǎngsēn’s paired yīn-yáng rhymes with their reconstructions 脂 -ii 支 -ee

真 -iiŋ 耕 -eeŋ

之 -ɯɯ 魚 -aa

蒸 -ɯɯŋ 陽 -aaŋ

幽 -uu 侯 -oo

冬 -uuŋ 東 -ooŋ

Note: All yáng rhymes in this example have -ŋ coda. The reconstruction is Zhèngzhāng’s.

Chinese (see Section 2.3.3.5.1 on post-codas). The yùnbù of Old Chinese can be arranged according to three types of codas: the yīn (-Ø, -G), yáng (-N), and rù (-C) (Baxter 1992: 562–564) (see Table 2.25). 2.3.3.2 The Relationships between Rhyme Groups Rhyme groups are phonological categories containing syllables with the same or similar main vowel and the same post-vowel elements. With the given rhyme groups, the number of required main vowels can be estimated. Based on their working principles, Qing scholars proposed that there are certain relationships between different rhyme groups. Based on categorical information, Kǒng Guǎngsēn’s 孔廣森 (1752–1786) theory of yīn-yáng duì zhuǎn 陰陽對轉 ‘the interaction between yīn (open syllables) and yáng rhymes (closed syllables with a nasal coda)’ (see Table 2.26 for examples) actually shows corresponding rhymes that have similar main vowels (see Wáng Lì 1956). With a few corrections modern phonologists reconstruct the same main vowels for corresponding rhyme groups. In his system, Dài Zhèn 戴震 (1723–1777) suggested a relationship between three types of rhymes, the yīn, yáng, and rù syllables (see Wáng 1956). In Table 2.27 you can see the yīn with no coda and rù rhymes with -ɡ paired, and their main vowels matched quite well. Clearly this type of information is very helpful to modern scholars in determining the main vowels of corresponding rhymes. However, the idea of an ideal system lying behind this reconstruction also strongly influences modern scholars. A system with fewer phonological gaps is the goal of various reconstructions, though this is not necessarily a natural representation of language.

80

Old Chinese

Table 2.27 Dài Zhèn’s suggested relationship between the yīn, yáng, and rù syllables 脂 -ii 支 -ee

質 -iiɡ 錫 -eeɡ

之 -ɯɯ 魚 –aa

職 -ɯɯɡ 鐸 -aaɡ

尤 -ua 侯 -oo

沃 -uuɡ 屋 -ooɡ

Note: The reconstruction is Zhèngzhāng’s. The character yóu 尤 has a short -u.

a

Table 2.28 The xiéshēng components of the rhyme groups zhī bù 之部, zhí bù 職部, and zhēng bù 蒸部 之部: 職部: 蒸部:

之, 里, 思, 才, 子 . . . 直, 北, 則, 革, 黑 . . . 登, 曾, 乘, 朋, 丞 . . .

The other important piece of information for rhyme reconstruction is xiéshēng characters. The working principle is that characters with the same phonetic component should belong to the same rhyme groups.3 Once the group of a phonetic component can be determined, other characters with the same xiéshēng component (especially those in non-compounding forms) will fall into the same group. For example, the rhyme groups zhī bù 之部, zhí bù 職 部, and zhēng bù 蒸部, contain the xiéshēng components listed in Table 2.28, as well as every character containing those components. This theory allows Chinese characters that do not appear in the rhyming materials to be used in the reconstruction of the rhyme categories. A problem scholars are fully aware of is that the time depth of the xiéshēng characters is not the same as the rhyming materials, with xiéshēng characters being both older and more varied in age than the rhyming materials with which they are analyzed. It is worth noting here that rhyming categories are not the same as individual rhymes or finals and must not be confused (see Zhèngzhāng [2003] 2013 and Yú Mǐn 1984a for more details). Yú Mǐn uses the rhyme system of the modern Beijing dialect as an example. There are thirty-seven finals in the Beijing dialect, but they form only thirteen rhyming groups (shísān zhé 十三轍) in the local singing performances. Finals -ən, -in, -un, -yn; -əŋ, -iŋ, -uŋ, -yuŋ; -ɤ, -o, -uo, -i, -ɿ, -ʅ, -y, ɚ belong to the same rhyming group respectively. Similarly the yùnbù of Old Chinese should be analyzed as rhyming groups instead of individual finals, and so any given yùnbù can have finals with different vowels. 3

“同聲必同部”段玉裁《說文解字注》(Duan [1815] 1981).

Old Chinese

81

Table 2.29 The possible combinations of medials in Middle Chinese, and their respective kāihé and rank

kāikǒu hékǒu

I

II

III

IV

-Ø-w-

-ɰ-wɰ-

-(ɰ)j-w(ɰ)j-

-Ø-w-

2.3.3.3 The Absence of Medials After many decades of research, scholars are gradually reaching the consensus that there is no need to have a medial slot in the syllable structure of Old Chinese. Middle Chinese has a structural slot for a medial between the initial and the main vowel. The medial system of Middle Chinese can be illustrated as in Table 2.29. The syllable structure of Middle Chinese can be simply stated as IMVE, where M = (w)(ɰ)(j). Karlgren (1915–1926) reconstructed medial -w- and -jfor Old Chinese accordingly (the medial of Middle Chinese Rank-II was not recognized in Karlgren’s time). For Old Chinese the medial is not a necessary element, although the syllable structure of Old Chinese has more segment slots and the initial can be a consonant cluster. Old Chinese syllables can be described as IVE with both I and E allowed to be consonant clusters. In the following sections the reasons for deleting Middle Chinese medials are provided. The medial is a structurally ambiguous element; it can be a part of the initial consonant, a part of the final, or both. The Chinese traditional fǎnqiè 反切 spellings show this ambiguity (refer to Section 1.3.2). In Old Chinese reconstructions, most researchers (Baxter 1992; Zhèngzhāng [2003] 2013; Baxter & Sagart 2014) prefer to treat post-stem elements (-r-, -j-, -w-) as a part of the initial mainly based on the interactions between the post-stem elements and the stem and their changes phonologically and morphologically. The medial -ɰof Middle Chinese Rank-II syllables is a weakened form of Old Chinese poststem -r- of the initial position. 2.3.3.3.1 The Deletion of Middle Chinese -w- Medial The deletion of medial -w- in the reconstruction of Old Chinese is mainly based on two pieces of information. One is that -w- syllables are basically related to syllables with grave initials, that is, labials and gutturals. The other is that there is no phonological contrast between syllables with and without -wafter the labial initials. Jaxontov (1960) made an interesting observation on the distribution of two different types of syllables, labialized (with a rounded vowel or with a w- medial) and plain (without a rounded vowel or -w- medial). He listed

82

Old Chinese

Table 2.30 The four rhyme patterns listed by Jaxontov (a) (b) (c) (d)

poŋ p(w)eŋ p(w)en p(w)en

toŋ teŋ twen ten

koŋ keŋ kwen ken

kweŋ kwen

Table 2.31 Jaxontov’s proposal (a) (b) (c) (d)

poŋ p(w)eŋ pon p(w)en

toŋ teŋ ton ten

koŋ keŋ kon ken

kweŋ kwen

four rhyme patterns (see Table 2.30). The initial t- represents acute initials, k- represents guttural initials, and p- represents labial initials. The vowel -o represents rounded vowels and vowel -e represents unrounded vowels. The (w) indicates that -w- is a non-contrastive feature because in Middle Chinese syllables that differ by having initials p- and pw- alone never make contrast. Jaxontov points out that patterns (c) and (d) cannot rhyme together, indicating that twen and ten do not differ in the medial -w- but do have different vowels. So the syllables in the pattern (c) are round vowels, and the medial -win patterns (b) and (d) should be the modification of guttural initials. After labial initials -w- is not a distinctive feature (see Table 2.31). Thus, medial -w- does not exist in Old Chinese. Jaxontov’s approach treating -w- as a modification of the guttural initials has been accepted by many scholars in their reconstructions (F-K. Li 1971; Baxter 1992; Zhèngzhāng [2003] 2013). Li Fang-Kuei (1971) added a series of labialized guttural initials in the system and eliminated the medial -w- in his reconstruction. Baxter (1992) took the same approach. In Zhèngzhāng’s system, -w- was treated as a post-stem element instead of a feature of an additional group of guttural initials. He also treats -w- as a modifying feature (labialization) of the guttural initials, so the labialized gutturals can take other post-stem elements -r-, -l-, and -j- (e.g. guā 瓜 *kwraa > kwɰa). 2.3.3.3.2 The Short Vowels and Middle Chinese Rank-III Syllables The disproportional distribution of the Middle Chinese rhymes among the Ranks (see Table 2.32) makes scholars suspect that the Middle Chinese Rank-III syllables might have a different type of vowel (Zhèngzhāng [2003] 2013: 171).

Old Chinese

83

Table 2.32 The frequency of Qièyùn rhymes (ignoring tone) by their rank Rank

I

II

III

IV

Total

Number of rhymes Frequency

14 .23

12 .20

30 .49

5 .08

61 1

As can be seen in Table 2.32, Rank-III rhymes make up almost half of the total rhymes in all ranks. In modern dialects the dominant feature of Rank-III syllables is the existence of a palatal medial -j-. However, in accordance with language universals, syllables without a medial are unmarked and the ones with a medial are marked (with more elements), therefore syllables with a palatal medial should be less common than the ones without a medial. Since this trend is reversed in Modern Mandarin, scholars suspect that it is a result of a change from a more balanced distribution in Old Chinese. Many proposals try to use vowel length to explain such unbalanced syllable types (vowel length is also often replaced with initial pharyngealization, see Section 2.3.5.2). Some conceptual ideas can be found as early as publications of the 1960s. Mǎ and Luó (1962) suggested that the difference between Middle Chinese Rank-III and Rank-IV syllables is in the length contrast of the main vowels, with Division-IV having long vowels and Rank-III having short vowels. Edwin G. Pulleyblank (1962–1963) suggested that Rank-III syllables had long vowels and developed a palatal medial in Middle Chinese. Pulleyblank’s proposal that Rank-III has a long vowel is clearly in contrast with others. Zhèngzhāng ([2003] 2013) accepts Mǎ and Luó’s proposal and rejects Pulleyblank’s with the acknowledgment that the palatal medial of Rank-III is a late development. Evidence supporting the vowel length hypothesis comes from many different sources, including Chinese dialects, genetically related and geographically adjacent languages, and transcriptional materials. The literary and colloquial readings in the Taishun dialect (see Table 2.33), a variety of Min, have a clear contrast in Middle Chinese Rank-III syllables. The literary readings have -jand the colloquial readings of the same characters do not. A similar phenomenon can be observed from the different layers of SinoJapanese, go-on 吳音 (Mandarin: wúyīn) and kan-on 漢音 (Mandarin: hànyīn) readings in Japanese (Karlgren 1915–1926). Go-on readings precede kan-on readings, and while both retain attributes of Middle Chinese, the go-on readings of Rank-III characters do not have any reflex of the palatal medial -j- (seen as -i- in kan-on readings), as shown in Table 2.34 and Table 2.35. The literary readings formed after the colloquial readings and kan-on is a later layer than go-on. Both show a development from no palatal medial to the existence of a palatal medial.

84

yāng ‘seedling’ ljɔ̃ ɔ̃

尺 chǐ ‘foot’ tɕhjɪʔ tshœ

cháng ‘long’ tɕjã tɔ̃

別 bié ‘depart’ piʔ pəʔ

Source: Pān 2000: 143, tone marks are omitted.

Literary Colloquial

Literary Colloquial





燭 zhú ‘candle’ tɕɔʔ tsœ

gòng ‘together’ kjəŋ kəŋ



Table 2.33 Rank-III syllables in the Taishun dialect

綠 lǜ ‘green’ lɔʔ lœ

píng ‘level’ pjɪŋ pã



雨 yǔ ‘rain’ y hɔ

méng ‘alliance’ mjɪŋ mẽ



晝 zhòu ‘noon’ tɕjou taw

yǐn ‘drink’ jɪŋ ã



巾 jīn ‘towel’ kyœŋ kœŋ

yǒu ‘have’ jow u



船 chuán ‘boat’ syœŋ sœŋ

jiǔ ‘nine’ kjow kaw



林 lín ‘grove’ lin lan

liú ‘flow’ ljow law



Old Chinese

85

Table 2.34 Go-on and kan-on readings of Rank-III characters

Go-on Kan-on

秧 yāng

牛niú

獵 liè

呂 lǚ

竹 zhú

綠 lǜ

‘seedling’

‘cow’

‘hunt’

surname

‘bamboo’

‘green’

o: jo:

ɡu ɡju:

ro: rjo:

ro rjo

toku tɕjoku

roku rjoku

Table 2.35 More examples of go-on readings lacking palatal glides otherwise seen in kan-on

Go-on Kan-on

疆 jiāng

良 liáng

共 gòng

今 jīn

音 yīn

九 jiǔ

曲 qǔ

‘boundary’

‘fine’

‘together’

‘present’

‘sound’

‘nine’

‘curvy’

ko: kio:

ro: rio:

gu kio:

kon kin

on in

ku kiu:

koku kioku

Table 2.36 Length contrast and vowel quality in Chinese dialects Guangzhou Rank-IV (33) Rank-III (5)

跌 tit 質 tsɐt

結 kit 吉 kɐt

切 tshit 七 tshɐt

屑 sit 失 sɐt

Linchuan Rank-IV (33) Rank-III (5)

跌 tjɛt 質 tit

結 tɕjɛt 吉 tɕit

切 tɕhjɛt 七 tshit

屑 ɕjɛt 失 sit

跌 diē 質 zhì ‘tumble’ ‘quality’

結 jié 吉 jí ‘knot’ ‘lucky’

切 qiè 七 qī ‘cut’ ‘seven’

屑 xiè 失 shī ‘scraps’ ‘loss’

Pinyin Meaning

Note: The numbers in parentheses describe the tonal values, with 5 being a short high tone, and 33 being a long mid tone.

To explain the development of a palatal medial from short vowels, Zhèngzhāng ([2003] 2013) points to Sino-Tibetan languages where long vs. short is a common vowel contrast. It is found in most of the languages of the Dong-Tai family, the Yao language of Miao-Yao, and the Dulong (Drung), Deng (Kaman), and Menba (Monpa) languages of Tibeto-Burman. In Chinese dialects some syllables show a length contrast as well as vowel quality (see Table 2.36).

86

Old Chinese

Table 2.37 The fǎnqiè spellings of Rank-III syllables Character spelled IjV(E)

Upper speller =

Ij(V(E))

Lower speller +

(I)jV(E)

However, a timeline of the Chinese language must be kept in mind when examining these pieces of information. If these dialects formed after the standard Middle Chinese system as represented by the Qièyùn, the key question that must be answered is whether such a contrast in vowel length is a feature of standard Middle Chinese phonology. According to Zhèngzhāng’s reconstruction of Middle Chinese, the contrast of vowel length had already disappeared and a medial had developed (three variants in different phonological conditions). The fǎnqiè spellings of the Qièyùn seem to suggest the existence of a medial in the Rank-III syllables. Fǎnqiè as it relates to Rank-III syllables is discussed in more depth in Section 3.1.1. Commonly there is an agreement between the upper and lower spellers, and both must be Rank-III syllables. This phenomenon can be only explained by the existence of a medial (see Table 2.37). Special Chinese characters designed to transcribe the short vowels in Sanskrit become Rank-III syllables in Middle Chinese. In Buddhist translations of the Eastern Han dynasty, syllables with a short a, namely, ka, kha, and ga, were transcribed in Sanskrit by specially invented Chinese characters jiā 迦, qiē 佉, and jiā 伽 (Yú Mǐn 1984a). Later in the Middle Chinese system, all three belong to the Rank-III gē 歌 rhyme, and have developed a palatal medial -j-.4 In notations on the Huáinánzi 淮南子5 and Lǚshì Chūnqiū 呂氏春秋,6 Gāoyòu 高誘 uses the terms jíyán 急言 ‘hurried speech,’ to refer to syllables in Rank-III and huǎnyán 緩言 ‘unhurried speech’ for syllables in Rank-IV. Jíyán syllables, then, are short, and huǎnyán syllables are long. 2.3.3.4 Main Vowels The reconstructions of the phonetic values of main vowels have benefited from the transcriptions of Chinese words written in foreign scripts as well as the existence of loanwords from Chinese in foreign languages. One of the pioneering studies done by Wāng Róngbǎo (1923) successfully identifies the phonetic values of Middle Chinese rhymes in Old Chinese by using Sanskrit 4

5 6

Tibetan syllables with -j- correspond with short vowels in the Drung language, a member of the Tibeto-Burman language family. This does not mean that the same developments happened in Chinese, but it does suggest that this kind of development is possible. A collection of articles written in Western Han times by Liú Ān 劉安. An encyclopedia compiled in 239 BC by Lǚ Bùwéi 呂不韋.

Old Chinese

87

Table 2.38 Karlgren’s fourteen reconstructed vowels

e æ a

ě ǎ

ə

u ω o ɔ ɑ

ǔ o˘

transcriptions and Japanese loanwords. He shows that the phonetic value of the gē 歌 and gē 戈 rhymes of Middle Chinese (Qièyùn/Guǎngyùn) is -a instead of -o before Tang and Song times, and the phonetic value of yú 魚, yú 虞, and mó 模 is also -a instead of -u or -y before Wei–Jin times. In Chinese-Sanskrit transcriptions, characters tuó 陀 (gē 歌 rhyme), mó 摩 (gē 戈 rhyme), and pó 婆 (gē 戈 rhyme) are used to transcribe the syllables with -a in Sanskrit words fótuó 佛陀 ‘Buddha’ (Sanskrit: buddha), móhē 摩訶 ‘big, broad’ (Sanskrit: mahat), típó 提婆 ‘sky’ (Sanskrit: diva). Japanese kana syllabary was invented during the late Tang dynasty (ninth century). The kana syllables a, ka, sa, ta, na, ha, ma, ya, ra, and wa7 are represented by the Chinese characters ā 阿, jiā 加, zuǒ 左, duō 多, nuó 那, bō 波, mò 末, yě 也, luó 羅, and huó 和. Of these ten characters, seven are from gē 歌 or gē 戈 rhymes. These pieces of evidence clearly point to the phonetic value of gē 歌 and gē 戈 rhymes being -a during the Tang dynasty or earlier. 2.3.3.4.1 The Number of Main Vowels The reconstruction of main vowels in Old Chinese is essentially a pursuit of reducing the required vowels to a minimum while ensuring that all distinctions can be maintained. With very different phonological orientations, scholars have proposed various systems of main vowels. As with the nature of maintaining distinctions there is usually a trade-off: a system with fewer main vowels is compensated with a more complex medial and/or coda system, and vice versa. The early reconstructions, which tended against using medials and allophony to account for multiple distinctions, often required a large number of main vowels. Karlgren’s reconstruction, for example, has fourteen main vowels (see Table 2.38). The imbalance is obvious: five degrees of openness are required for back vowels while the basic high front vowel -i is not even used. After the confirmed identification of the syllable types of the four Middle Chinese ranks, the recognition of medials allows the required vowel number to be significantly reduced. The proposed syllable types of Rank-II, -III, and -IV

7

あ, か, さ, た, な, は, ま, や, ら, わ.

88

Old Chinese

for Old Chinese are: (1) Rank-IV syllables with no medial (similar to Rank-I syllables), (2) Rank-II syllables with a post-stem -r-, and (3) Rank-III syllables with short vowels. Each proposal leads to a new round of reconstruction. With the successful identification of the post-initial -r-, the contrast between Middle Chinese Ranks-I and -II has been changed from a main vowel contrast to that of a post-initial -r-. Thus the required number of main vowels has been greatly reduced; a representative of this significant vowel reduction is Li FangKuei’s four-vowel system (1971). 2.3.3.4.2 Four-Vowel System In an innovative four-vowel system, Li Fang-Kuei dramatically reduced the number of contrasting main vowels. His reconstruction is based on (1) the rhyme categories of Old Chinese and (2) Middle Chinese phonology (F-K. Li 1980: 27). In his reconstruction, Li uses four monophthongs, i, u, ə, and a, and three additional diphthongs, iə, ia, and ua. Table 2.39 shows the distribution of yùnbù across different vowels. Li’s jiā 佳 and zhōng 中 are equivalent to others’ zhī 支 and dōng 冬. The purpose behind choosing different characters is to avoid confusion with their homophones, zhī 之 and dōng 東. One distinctive characteristic of Li Fang-Kuei’s reconstruction is that one group can have only one main vowel, thus diphthongs -iə, -ia, and -ua are used to distinguish all vocalic contrasts in the groups with main vowels -ə and -a. 2.3.3.4.3 Six-Vowel System Recent reconstructions seem to settle on a six-vowel system. It is quite clear that each of the contrasting rhyme groups with coda -ŋ must be distinguished with a different main vowel. There are six groups with coda -ŋ (yáng 陽 gēng 耕 dōng 東 dōng 冬 zhēn 真 zhēng 蒸), so six main vowels are required. In Baxter’s system (see Table 2.40) they are distinguished by main vowels, while

Table 2.39 Li Fang-Kuei’s four-vowel system

a ia ua u i ə iə

-ɡ/k



-ɡw/kw

-ŋw

-p/b

-m

-t/d

-n

-r

魚 魚 ○ ○ 佳 之 之

陽 陽 ○ 東 耕 蒸 蒸

宵 宵 ○ 侯 ○ 幽 幽

○ ○ ○ ○ ○ 中 ○

葉 葉 ○ ○ ○ 緝 緝

談 談 ○ ○ ○ 侵 侵

祭 祭 祭 ○ 脂 微 微

元 元 元 ○ 真 文 文

歌 歌 歌 ○ ○ ○ ○

Source: F-K. Li 1971.

Old Chinese

89

Table 2.40 Baxter’s six-vowel system

a e o u i ɨ



-k



-w

-wk

-p

-m

-j

-t

-ts

-n

魚 支 侯 幽 ○ 之

鐸 錫 屋 覺 職質 職

陽 耕 東 冬 蒸真 蒸

宵 宵 ○ ○ 幽 ○

藥 藥 ○ ○ 覺 ○

盍 盍 盍 緝 緝 緝

談 談 談 侵 侵 侵

歌 (歌) 歌 微 脂 微

月 月 月 物 質 物

祭 祭 祭 ○ ○ ○

元 元 元 文 真 文

Source: Baxter 1992.

Table 2.41 Zhèngzhāng’s six-vowel system

a e o u i ɯ



-g



-u

-ug

-b

-m

-l/-i

-d

(-s)

-n

魚 支 侯 幽 脂 之

鐸 錫 屋 覺 質 職

陽 耕 東 終 真 蒸

宵 宵 宵 ○ 幽 幽

藥 藥 藥 ○ 覺 覺

盍 盍 盍 緝 緝 緝

談 談 談 侵 侵 侵

歌 歌 歌 微 脂 微

月 月 月 物 質 物

祭 祭 祭 隊 至 隊

元 元 元 文 真 文

Source: Zhèngzhāng 1984, [2003] 2013.

in Li’s four-vowel system the six coda -ŋ groups are distinguished by main vowels and an off-glide after the coda –ŋ. Zhèngzhāng’s system (Table 2.41) has been rearranged for easy comparison to Li’s and Baxter’s systems and for a clear illustration of the relationship between the representative characters for the groups. Zhèngzhāng’s duì 隊 and zhì 至 are equivalent to others’ wù 物 and zhì 質. (-s) is actually a post-coda, which causes these syllables to change to Middle Chinese qù tone. The reconstructing of the main vowel -o can explain a number of changes, such as *oT > waT (T = acute coda, -j, -t, -n, -r), which also causes labialization *k- > kw-, for example 果 *[k]ˁo[r]ʔ > *kˁwarʔ > *kˁwajʔ > Middle Chinese kwaX (X = shǎng tone) > guǒ (Baxter & Sagart 2014: 282). One of the main differences is that in these six-vowel systems one rhyme group can have as many as three main vowels. Both Baxter’s and Zhèngzhāng’s reconstructions, for example, have three main vowels in the tán 談 rhyme: -am, -em, and -om; and three main vowels in the yuán 元 rhyme: -an, -en, and -on. It should be mentioned that the six-vowel system is also adopted by Starostin in his reconstruction of Old Chinese phonology (S. Starostin 1989).

90

Old Chinese

Table 2.42 A comparison of -ŋ coda syllables in Li Fang-Kuei’s four-vowel system and Zhèngzhāng’s six-vowel system

Six Four

yáng 陽

gēng 耕

dōng 東

dōng 冬

zhēn 真b

zhēng 蒸

aaŋ aŋ

eeŋ iŋ

ooŋ uŋ

uuŋ uŋw

iiŋ in

ɯɯŋ əŋ

In Li Fang-Kuei’s four-vowel system dōng 東 and dōng 冬 (Li’s zhōng 中) are not distinguished by main vowels but by an off-glide w after the coda -ŋ. Table 2.42 shows a comparison of Zhèngzhāng’s six-vowel system and Li’s four-vowel system, using syllables with -ŋ as examples. Li Fang-Kuei (1980) does not distinguish zhēn 真b and zhēn 真a, which are distinguished by later reconstructions (Baxter 1992; Pān 2000; Zhèngzhāng [2003] 2013). 2.3.3.5 The Codas The codas of Old Chinese are not as difficult to reconstruct as the initials and vowels because of the rhyme groups and their corresponding phonological values in Middle Chinese. The reconstructed coda system varies in different proposals, but in general it is similar to that of Middle Chinese, namely, -m, -n, -ŋ, -p, -t, -k, -j, -w, and -Ø. However, for each word the historical relationship between Middle Chinese and Old Chinese must be examined individually; the relationship between a Middle Chinese coda and an Old Chinese coda should not be applied to individual words automatically. 2.3.3.5.1 The Post-Coda Segments and Tonal Developments The coda may or may not be the last element in a syllable. The phonology of Middle Chinese has four tones: píng, shǎng, qù, and rù (the non-píng tones are traditionally referred to as zè tones). In Old Chinese, however, these suprasegmental tonal contrasts are segmental differences and the four tones differ in the coda slot. In addition to the nasal codas -m, -n, and -ŋ, Middle Chinese shǎng and qù tone syllables have -ʔ and -s, respectively (Zhèngzhāng [2003] 2013). Tones in Middle Chinese arose from the simplification of different coda clusters in Old Chinese. The ending -ds of tài 泰 *thaads indicates that the tài 泰 rhyme (as well as jì 祭, guài 夬, fèi 廢 rhymes, see Section 2.3.3.5.4 and Section 3.5.8.1 for more detail) of Middle Chinese has the ending -d(t) of rù tone syllables. This explains why it does not have its corresponding píng tone and shǎng tone rhymes in the Qièyùn (see Table 2.43). Syllables with -ʔ and -s post-codas later lost these segments and the originally redundant tonal differences became distinctive (see following sections). The degree of sonority of the post-coda elements -Ø > -ʔ > -s > -p/t/k,

Old Chinese

91

Table 2.43 Relationship between tones in Middle Chinese and coda clusters in Old Chinese Middle Chinese: Old Chinese: Example:

píng -Ø 同 *dooŋ 刀 *taaw

shǎng -ʔ 動 *dooŋʔ 島 *taawʔ

qù -s 洞 *dooŋs 到 *taaws 泰 *thaads

rù -b -d -ɡ 獨 *dooɡ

Table 2.44 Rhyming words in the Shījīng and Chǔcí Stanza I of Gōng Liú 公劉, Shījīng 康 kāng 疆 jiāng 倉 cāng 光 kuāng 張 zhāng 揚 yáng All the syllables belong to Middle Chinese píng tone.

糧 liáng 行 háng

囊 náng

Stanza V of Qī Yuè 七月, Shījīng 股 gǔ 羽 yǔ 野 yě 下 xià 鼠 shǔ 户 hù All the syllables belong to Middle Chinese shǎng tone.

宇 yǔ 子 zǐ

户 hù 處 chǔ

邁 mài

穢 huì

舄 xì 若 ruò

碩 shuò

Jiǔ Biàn 九辯, Chǔcí 帶 dài 介 jiè 慨 kài 敗 bài 昧 mèi All the syllables belong to Middle Chinese qù tone. Stanza IX of Bì Gōng 閟宫, Shījīng 柏 bǎi 度 dù 尺 chǐ 奕 yì 作 zuò 碩 shuò All the syllables belong to Middle Chinese rù tone.

eventually determines the order of the tones, píng, shǎng, qù, and rù in Middle Chinese. 2.3.3.5.2 Segmentals to Suprasegmentals In pre-Qin rhyming materials such as the Shījīng and Chǔcí 楚辭, rhyming words are commonly in the same Middle Chinese tonal category. As shown in Table 2.44, in a stanza there can be a number of rhyming words belonging to the same Middle Chinese tonal category used consecutively without any words of other tones involved. Such rhyming evidence could easily lead people to conclude that tones existed in Old Chinese. However, there is quite convincing evidence suggesting that such rhyming could be based on segmental elements as well, not necessarily only on suprasegmental tones. As will be shown in the following

92

Old Chinese

Table 2.45 Zhèngzhāng’s suggested tones that coexisted with different finals

Segments Tone

píng

shǎng





-Ø 33

-ʔ 35

-s 31

-C (-b, -d, -ɡ) 3

sections, the tonal differences in Middle Chinese developed from different segments at the end of syllables. Zhèngzhāng ([2003] 2013) suggested that in Old Chinese segments and suprasegmental tones were in coexistence (see Table 2.45). 2.3.3.5.3 The -ʔ Becomes Middle Chinese Shǎng Tone Middle Chinese shǎng tone syllables have a -ʔ coda, which causes an abrupt but high rising tonal contour (Mei 1970). This phonetic characteristic was still a basic feature of the Middle Chinese shǎng tone in the Tang dynasty. Many descriptive statements of tonal values suggest the special tonal value of the shǎng tone, for example “shǎng tone is sharp and rising.”8 In some Chinese dialects shǎng tone can have a special glottal tension, for example the Wenzhou dialect. Of the eight tones, the two shǎng tones, yīnshǎng and yángshǎng, are in very high register, short, and with similar tonal contours (Zhèngzhāng 2014). Zhèngzhāng (1994, [2003] 2013) shows a large number of basic words in the shǎng tone (see Table 2.46), although it has the fewest words overall in comparison with other tones. He thinks it could be a diminutive suffix -q, which weakens to -ʔ and later becomes a tonal feature (Zhèngzhāng [2003] 2013). The concentration of shǎng tone words cannot be satisfactorily explained simply by chance. Middle Chinese shǎng syllables are either tone 3 or 4 in Modern Mandarin. The reconstruction of coda segments to account for the Middle Chinese tones started with Haudricourt’s proposal (1954a, 1954b). He explained tonogenesis, that is the phonological change from segments to tones, in Vietnamese and numerous other East and Southeast Asian languages, providing the possibility of reconstructing a nontonal Old Chinese. (A more comprehensive account of the development and evolution of tonal systems was published by Haudricourt in 1961.) Many scholars followed the scheme of tonogenesis developed by Haudricourt, using -ʔ as the post-coda element for Middle Chinese shǎng tone, and -s for Middle Chinese qù tone. The -ʔ hypothesis has been supported and improved upon by a number of scholars

8

See Section 3.4.1.

Old Chinese

93

Table 2.46 Basic words in the shǎng tone In many pairs of adjectives with opposite meanings, the lesser (latter) are shǎng tone: 大小 dàxiǎo ‘large/small’ 遐邇 xiáěr ‘distant/near’ 多少 duōshǎo ‘many/few’ 高矮 gāoǎi ‘high/low’ 深淺 shēnqiǎn ‘deep/shallow’ 鹹淡 xiándàn ‘salty/insipid’ 奢儉 shējiǎn ‘extravagant/frugal’ 衆寡 zhòngguǎ ‘many/few (people)’ 鬆緊 sōngjǐn ‘loose/tight’ In many pairs of verbs (action and quality), the negative (latter) are shǎng 增减 zēngjiǎn ‘increase/decrease’ 勤惰 qínduò 勝負 shèngfù ‘win/lose’ 生死 shēngsǐ 成毀 chénghuǐ ‘achieve/destroy’ 真假 zhēnjiǎ 安險 ānxiǎn ‘safe/dangerous’ 褒貶 bǎobiǎn 續斷 xùduàn ‘continue/break’ Many kinship terms are shǎng tone: 祖 zǔ ‘ancestor’ 考 kǎo ‘deceased father’ 妣 bǐ ‘deceased mother’ 父 fù ‘father’ 母 mǔ ‘mother’

女 姐 弟 舅 嫂

nǚ jiě dì jiù sǎo

Many words for body parts are shǎng tone: 首 shǒu ‘head’ 顙 腦 nǎo ‘brain’ 頦 眼 yǎn ‘eye’ 頸 臉 liǎn ‘face’ 乳 口 kǒu ‘mouth’ 手 齒 chǐ ‘teeth’ 掌

sǎng kē jǐng rǔ shǒu zhǎng

‘forehead’ ‘chin’ ‘neck’ ‘breast’ ‘hand’ ‘palm’

tone: ‘diligent/lazy’ ‘alive/dead’ ‘real/fake’ ‘praise/censure’

‘daughter’ ‘elder sister’ ‘younger brother’ ‘maternal uncle’ ‘elder brother’s wife’ 指 肚 股 腿 踝 踵

zhǐ dù gǔ tuǐ huái zhǒng

‘finger’ ‘belly’ ‘thigh’ ‘leg’ ‘ankle’ ‘heel’

(Pulleyblank 1962–1963; Mei 1977; Sagart 1993a; Zhèngzhāng 1994) with various evidence. Still, the historical source of -ʔ is yet to be explained. 2.3.3.5.4 The -s and the Middle Chinese Qù Tone In comparison with the relationship between -ʔ and shǎng tone, evidence for the relationship between -s and qù tone is stronger and more convincing. Many pieces of evidence point to Middle Chinese qù syllables having an -s at the end of the syllable. Since Haudricourt (1954b), the relationship between -s and Middle Chinese qù tone has been confirmed by a number of scholars (Pulleyblank 1973; Mei 1980; Yú Mǐn 1984a; Zhèngzhāng 1994, [2003] 2013; Sagart 1999) with evidence from various foreign transcriptions. SinoKorean vocabulary has quite a few words with -s coda. These are borrowed from Old Chinese and all are Middle Chinese qù syllables (Zhèngzhāng [2003] 2013) (see Table 2.47). In early Chinese Buddhist translations Sanskrit -s and -ʂ are often transcribed by using qù tone syllables (Yú Mǐn 1984a) (see Table 2.48).

94

Old Chinese

Table 2.47 Middle Chinese qù syllables with -s coda in Sino-Korean 磨 制 界

‘grindstone’ ‘regulation’ ‘boundary’

mais tsis kas

篦 器 味

‘fine-toothed comb’ ‘utensil’ ‘taste’

pis kɯrɯs mas

Table 2.48 Sanskrit -s and -ʂ transcribed by using qù tone syllables -s -ʂ(s) _

奈 nas 替 tis _

陛 pas 膩 nis _

會 bhas 沸 pus _

衛 vas 費 pus _

會 vas 賴 ras _

奈 ras _

Table 2.49 Middle Chinese qù tone characters used to transcribe syllables with a stop coda in Sanskrit transcription 制 jet

逝 jet

世 sat

衛 pat

貝 pat

類 rod

Table 2.50 Xiéshēng characters with qù tone and rù tone qù rù

害 ɦɑj 割 kɑt

鳜 kwjᴇj 橛 ɡwjɐt

快 khɰwɛj 決 kwet

廢 pwjaj 發 pwjɐt

qù rù

奈 nɑj 捺 nɑt

例 ljᴇj 列 ljᴇt

話 ɦwɰɛj 刮 kwɰat

税 ɕwjᴇj 悦 jwjᴇt

One of the most important pieces of evidence is from Chinese itself. In the Qièyùn system there are four unusual qù rhymes that, unlike all other qù rhymes, do not have píng and shǎng counterparts, namely jì 祭, tài 泰, guài 夬, and fèi 廢. The qù tone characters without non-qù equivalents are used to transcribe syllables with a consonant stop coda in Sanskrit transcription (Zhèngzhāng [2003] 2013). The consonant codas indicate that these syllables were not open syllables (see Table 2.49). The xiéshēng relationship between qù and rù characters also shows a strong connection between these four rhymes and rù syllables in the Qièyùn system. The pairs of qù/rù syllables have the same phonetic components in their written forms (see Table 2.50 and Table 2.51). According to Middle Chinese phonetic values these qù syllables had consonant codas historically, with the palatal coda as the remaining trace of such consonant codas, -ts > -j. Zhèngzhāng ([2003] 2013) reconstructs the Old Chinese value as a -ds coda. As shown in Table 2.52, the earlier forms of jì 祭,

Old Chinese

95

Table 2.51 More examples of rhymes jì 祭, tài 泰, guài 夬, and fèi 廢

jì 祭 tài 泰 guài 夬 fèi 廢

















敝 -j 奈 -j 快 -j 廢 -j

憋 -t 捺 -t 決 -t 發 -t

祭 -j 賴 -j 話 -j 肺 -j

察 -t 獺 -t 活 -t 芾 -t

世 -j 害 -j

泄 -t 割 -t

鱖 -j 最 -j

厥 -t 撮 -t

袚 -j

拔 -t

Table 2.52 The reconstructed earlier forms of jì 祭, tài 泰, guài 夬, and fèi 廢 -ts

>

>

-js

-j, qù tone

Table 2.53 Reconstructed finals of jì 祭, tài 泰, guài 夬, and fèi 廢 for Old Chinese and Middle Chinese

Old Chinese Middle Chinese







祭A

祭B

*-aads -ɑj

*-raads -ɰaj

*-ads -jɐj

*-eds -jɛj

*-rads -ɰæj

tài 泰, guài 夬, and fèi 廢 are reconstructed as syllables with -ts or -ds codas, which eventually changed to -js. Later, the loss of the -s was responsible for the creation of the qù tone, similar to other syllables that had an -s coda (Zhèngzhāng [2003] 2013: 214). Zhèngzhāng describes the change from Old Chinese to Middle Chinese as -ds > -j, with reconstructed finals for Old Chinese and Middle Chinese reflecting this change (see Table 2.53). In Old Chinese grammar, -s has a morphological function to change a verb to a noun. Many cases of this nominalization function are shown in the studies of Pulleyblank (1973) and Mei (1980). This morphological function can also be found in Tibetan and thus is likely a feature of the common ancestral language of Chinese and Tibetan (Mei 1980). Some well-known examples of -s functioning as a morphological particle are given in Table 2.54. Note how the reflexes of the noun forms are all Mandarin tone 4, which qù tone tends to correspond to. The verb/noun pair of liáng/liàng 量 ‘measure/quantity’ even has cognates in Tibetan: ɦɡraŋ (verb) / ɡraŋs (noun). Zhèngzhāng ([2003] 2013) provides further examples of the morphological function of -s with different characters rather than separate readings of the same character (see Table 2.55). Because rù syllables have a stop coda, the rù examples actually indicate the simplification of codas from -Cs to -s (Baxter 1992: 317). Note that the rù/qù

96

Old Chinese

Table 2.54 Examples of -s functioning as a morphological particle Verb 磨 量 數 處

píng/qù shǎng/qù

mó liáng shǔ chǔ

Noun ‘grind’ ‘measure’ ‘count’ ‘reside’

磨 量 數 處

mò liàng shù chù

‘grindstone’ ‘quantity’ ‘number’ ‘place, location’

noun characters tuì 蛻, jì 髻, and huì 會 have -j endings in addition to a tonal change; it is likely that the endings underwent the change -ds > -js, since these rhymes have a -j ending in Middle Chinese, jì 祭 -jɛj, tài 泰 -ɑj, guài 夬 -ɯaj, fèi 廢 -jɐj in most reconstructions (Pān 2000: 83–88). Table 2.56 is a summary of the relationship between the codas and tones. Table 2.55 Further examples of the morphological function of -s with different characters Verb píng/qù

shǎng/qù

rù/qù

藏 陳 稱 比 坐 負 脫 結 合

cáng chén chēng bǐ zuò fù tuō jié hé

Noun ‘hide’ ‘display’ ‘weigh’ ‘be near’ ‘sit’ ‘carry on one’s back’ ‘shed’ ‘tie’ ‘be together’

臟 陣 秤 篦 座 背 脫 髻 會

zàng zhèn chèng bì zuò bèi tuì jì huì

‘internal organs’ ‘disposition of troops’ ‘steelyard’ ‘fine-toothed comb’ ‘seat’ ‘back’ ‘shedded skin’ ‘topknot’ ‘meeting’

2.3.3.5.5 The -r Coda In Baxter and Sagart’s recent reconstruction, coda -r is added in addition to -j and -n for the reconstruction of the rhyme groups (Baxter & Sagart 2014: 252–268), which were paired by the Qing scholars and termed yīnyáng duìzhuǎn 陰陽對轉 (the interaction between syllables with vocalic coda (yīn) and nasal coda (yáng)). Karlgren (1954) originally proposed the -r for the yīn rhyme to explain rhyming syllables with Middle Chinese -n and -j codas. For example, in the Shījīng the wén 文 rhyme group syllable chén 晨 dʑin can rhyme with the wēi 微 rhyme group syllables huī 煇 hwɨi, and qí 旂 ɡɨi. Also, as xiéshēng characters, the wén 文 rhyme group syllable jīn 斤 kɨn is the phonetic component of the wēi 微 rhyme group syllable qí 旂 ɡɨi. Other scholars reconstructed -d, -j, -δ ([sic] representing [-ð]), or -l, to account for

Old Chinese

97

this rhyming phenomenon. However, Sergej Starostin (1989) proposed that it is not all wéi 微 rhyme group syllables that have contacts with the wén 文 rhyme group, and vice versa. He reconstructed *ər as a third type of coda to account for the interaction. This proposal of a three-way contrast, -n, -r, -j, was adopted by Baxter and Sagart’s (2014) recent reconstruction. Parallelly, the -r coda is also used to account for the interaction between the traditional yuán 元 and the gē 歌 rhyme groups (see Table 2.57). 2.3.4

The Reconstruction of Syllable Structures

The Middle Chinese syllable, as well as that of most modern dialects, can be summarized as IMVE with a tonal distinction. For Old Chinese phonology, Table 2.56 A summary of the relationship between the codas and tones

Post-coda Nasal coda Stop coda Accompanying tone

píng

shǎng





-Ø -m, -n, -ŋ

-ʔ -mʔ, -nʔ, -ŋʔ



33

35

-s > -h -ms, -ns, -ŋs -bs, -ds, -ɡs 31

-b, -d, -ɡ 3

Source: Zhèngzhāng [2003] 2013.

Table 2.57 A comparison of two-way and three-way contrast for the alveolar codas

Rhyme group Karlgren Zhèngzhāng S. Starostin Baxter & Sagart

Rhyme group Karlgren Zhèngzhāng S. Starostin Baxter & Sagart

巾 jīn ‘kerchief’

斤 jīn ‘ax’

旂 qí ‘flag’

幾 jǐ ‘how many?’

文 *kjɛn *krɯn *krən *krən

[文?] *kjən *kɯn *kər *[k]ər

[微?] *kjər *kɯl *ɡər *C.[ɢ]ər

微 *kjər *kɯlʔ *kəjʔ *kəjʔ

反 fǎn ‘reverse’

燔 fán ‘burn’

番 bō ‘martial’

歌 gē ‘sing’

元 *pjwǎn *panʔ *panʔ *Cə.panʔ

[元?] *bʻjwǎn *ban *bar *[b]ar

[歌?] *pwâr *paal *pār *pˁar

歌 *kâ *kaal *kāj *[k]ˁaj

Source: Based on Baxter and Sagart (2014).

98

Old Chinese

Table 2.58 Morphologically related pronoun pairs 吾 汝 夫 胡

我 爾 彼 何

*ŋaa *njaʔ *pa *ɡaa

‘first person, singular’ ‘second person, singular’ ‘demonstrative’ ‘interrogative’

*ŋaalʔ *njelʔ *pralʔ *ɡaal

Table 2.59 Some examples illustrating the devoicing regular-causative pattern in Old Chinese

Regular Causative

“defeat”

“depart”

“break”

“spoil”

敗 *braads 敗 *praads

別 *bred 別 *pred

斷 *doonʔ 斷 *toonʔ

壞 *ɡruuls 壞 *kruuds

Source: Pān 2000.

every structure slot and each element for each slot must be carefully examined. It has become clear to scholars that the syllable structure of Middle Chinese is very different from that of Old Chinese, because a syllable structure of IMVE (CCVC) cannot explain the various pieces of evidence that have accumulated over the last century. The assumption that Old Chinese was monosyllabic like Middle Chinese (with segmental elements maximally a single initial, up to three medials, a main vowel, and a coda (I)(M)V(E)) has been challenged. Because the reconstruction of phonology is also related to the reconstruction of morphology, the syllable structure also has to take care of the issues related to stems and affixes (Sagart 1999). Some phonological relationships between different initial consonants would be better explained as morphophonological processes. Zhèngzhāng suggests (Pān 2000: 128) that the pairs of pronouns in Table 2.58 are morphologically related. The forms with no -l coda in Old Chinese or píng tone in Middle Chinese are regular forms. The forms with an -l coda in Old Chinese or a shǎng tone in Middle Chinese are emphasized forms. There are many words showing a relationship between regular and causative forms. In Chinese writing these morphological processes are not reflected in the choice of characters. The same characters could be used for both morphological forms. This relationship has been shown by a number of scholars (Zhōu Fǎgāo 1962; Wáng Lì 1965; and Zhōu Zǔmó 1966a). The examples in Table 2.59 are from Pān 2000. The pattern is that the regular forms have a voiced initial and the causative forms have a voiceless initial. Many nonChinese languages in China such as Old Tibetan, Luoba, Bumi, Qiang, Yi, and Lahu have similar patterns (Sūn Hóngkāi 1980, 1981, 1982). It is not

Old Chinese

99

difficult to see that voiced to voiceless change used to be a quite common morphological process of Tibeto-Burman languages. It has been the goal of Old Chinese phonologists to show that Old Chinese had more inflectional morphological processes. Some phonological phenomena are actually rooted in morphology that has become fossilized and nonproductive since then, leaving the morphological processes to appear as phonological differences. Scholars are working hard to reveal some of these lost morphological processes. 2.3.5

Recent Advances

In recent years scholars have challenged various traditional views of Old Chinese reconstruction. The new approaches are well represented in Baxter and Sagart’s Old Chinese (2014). Baxter and Sagart take a broad approach with more kinds of evidence, including modern dialects such as the Min dialect, non-Chinese languages such as the Kra-Dai, Hmong-Mien, Tibeto-Burman, and Vietic, which preserve early loanwords from Chinese. They also utilize the documents of pre-Qin periods excavated in recent decades, such as the documents on bamboo strips from the Guodian 郭店 archeological site. Because of a broad range of different pieces of evidence, a more complex system must be reconstructed. Because of the nature of the materials, as I explained in the Introduction, as well as theoretical and methodological issues, these new approaches remain controversial (G. Starostin 2015; Ho 2016; and Harbsmeier 2016). 2.3.5.1 Syllable Structures Baxter and Sagart (2014) suggested that Old Chinese words consisted of a root plus possible affixes. Word roots were either P monosyllables consisting of a full syllable ( ) or disyllables consisting of a full syllable P preceded by a minor syllable (σ), that is, having the structure σ+ . Minor syllables σ were reduced in comparison with full syllables, in terms both of the member of structural positions they allowed and of the number phonemes that contrasted in each position. (p. 50) P

As illustrated in Figure 2.1, a full syllable ( ) contains an obligatory initial (Ci), an optional medial -r- (Cm), an obligatory vowel (V), an optional coda (Cc), and an optional glottal stop -ʔ (Cpc). Baxter and Sagart do not consider a post-coda -s to be within the stem, but rather as a morphological suffix, and thus outside of the rhyme. Final -s was always a morphological suffix (e.g. 賜 *s-lek-s > sjeH > cì ‘give’) (Baxter & Sagart 2014: 51). A minor syllable contains a pre-initial consonant (Cpi) and a pre-initial vowel (Vpi). The Cpi position can have consonants p, t, k, r, s, m, and N, where N is a positional allophone of n- or ŋ-, and the Vpi is a schwa -ə. Of all these elements, only initial consonant and main vowel of the main syllable are obligatory.

100

Old Chinese

2.3.5.2 Pharyngeals For the initials it is proposed that Old Chinese words had an obligatory main initial consonant and sometimes a pre-initial. They are constructed with significantly more forms with pre-initial elements. Baxter and Sagart’s (2014) reconstruction of the elements of a preceding vowel nucleus includes: (1) presyllabic elements may contain up to two consonants and a possible vowel *ə; (2) the initial of the main syllable, and (3) medial *-r- (e.g. 落 *kə.rʕak [Zhèngzhāng’s *ɡ·raaɡ]> lɑk > luò). This reconstructed form is to account for the xiéshēng characters gè 各, gé 格, which have an initial k-, and luò 落, luò 洛, which have an initial l- in Middle Chinese. Baxter and Sagart divide Old Chinese initials into two types, the plain and pharyngealized, adopting Pulleyblank’s distinction of type A and type B, which is based on the Middle Chinese ranks. Ranks-I, -II, and -IV belong to type A, and Rank-III belongs to type B (Norman 1994). In comparison with Zhèngzhāng’s reconstruction, the difference is that Baxter and Sagart treat this as a contrast of consonants while Zhèngzhāng treats it as a contrast of the main vowels. Baxter and Sagart adopted Norman’s approach (1994), treating it as two different types of consonants, pharyngealized (Type A) vs. plain (Type B) (see Table 2.60). But Zhèngzhāng considers the contrast of the two types of syllables is the length of the main vowel, long (Type A) vs. short (Type B) (see Section 2.3.3.3.2); for example, gāng 綱 ‘guiding rope of Table 2.60 Baxter and Sagart’s reconstruction of Type A and B syllables Plain (Type A) Pharyngealized (Type B)

pˁ p

phˁ ph

bˁ b

mˁ m

m̻ˁ m̻

Note: The phonetic symbol ˁ is used to mark the pharyngealized initials.

Figure 2.1 Baxter and Sagart’s representation of the Old Chinese root as a P combination of a minor syllable (σ) and a full syllable ( ). (Source: Baxter & Sagart 2014: 50.)

Old Chinese

101

net’ (Rank-I) and jiāng 疆 ‘boundary’ (Rank-III) are reconstructed as *klaaŋ and *kaŋ by Zhèngzhāng, but as *kˁaŋ and *kaŋ by Baxter and Sagart. Palatalization, retroflexion, and secondary voicing are employed to account for various proposed changes from Old Chinese to Middle Chinese. As shown in Table 2.61, nonpharyngealized Old Chinese initial alveolar stops and nasals become palatal affricates and nasals in Middle Chinese, unless the palatalization was blocked by prevocalic *-r- (Baxter & Sagart 2014: 76). The Old Chinese consonant system had no retroflex initials; Middle Chinese retroflex initials are the result of reduction of certain Old Chinese onsets with initial or medial *r (Baxter & Sagart 2014: 80). The capital H in reconstructions represents the presence of a qù tone (see Table 2.62). Table 2.63 shows that not all the Middle Chinese voiced stops and affricates reflected voicing in the Old Chinese main initial. In some cases, the Middle Chinese voicing is secondary, reflecting a voiceless stop or affricate preceded by a rightly attached nasal preinitial (Baxter & Sagart 2014: 81). Table 2.61 Nonpharyngealized Old Chinese initial alveolar stops and nasals and Middle Chinese palatal affricates and nasals *t*na

> >

tsj-a ny-

真 入

*ti[n] *n[u]p

> >

tsyin nyip

> >

‘true, real’ ‘enter’

zhēn rù

Baxter and Sagart write this as tsy-.

Table 2.62 The Old Chinese onsets with initial or medial *r and Middle Chinese retroflex initials 鎮 沙

*ti[n]-s *sˁraj

> >

> >

trinH srae

zhèn shā

‘press down’ ‘sand’

Table 2.63 The Old Chinese nasal preinitial and Middle Chinese voiced stops and affricates 濁 淨 黃

*[N-tˁ]rok *N-tseŋ-s *N-kwˁaŋ

> > >

*N-dˁrok *N-dzeŋ-s *ŋɡwˁaŋ

>

> > *ɡwˁaŋ

>

draewk dzjengH hwang

> > >

zhuó jìng huáng

‘muddy’ ‘clear’ ‘yellow’

2.3.5.3 The Uvular Series As explained and adopted in Baxter and Sagart’s reconstruction (2014), Pān’s (1997) proposal that uvular consonants should be added to the inventory of initials

102

Old Chinese

Table 2.64 Correspondence table between Pān’s uvular proposal (uvulars and velars) for Old Chinese and the Middle Chinese reflexes Old Chinese

Condition

*q*qh*ɡ*ɡ*ɢ*ɢ-

Anywhere Anywhere Rank-III Rank-I, -II, -IV Rank-I, -II, -IV Rank-III

Middle Chinese > > > > > >

Ø-, kh-, khɡjɣɣɦj-

has been well received. Velar, uvular, and glottal consonants have close interaction with each other in many languages. For example, in the Kam languages (Kam, Sui, Maonan), the word wūyā 烏鴉 ‘crow’ (Old Chinese: wū 烏) is pronounced as follows: Kam: ʔa; Sui: qa; and Maonan: ka. Pān (1997) argued that the initial consonant of their proto-form should be a uvular stop *q-. The uvular stop can better explain some ancient place names such as dàyuān 大宛 ‘name of an ancient country Tocharia,’ Old Chinese: *dalsqon > Middle Chinese: daj.ʔwjɐn, Greek: Tahoroi, Latin: Tochari, Sanskrit: Tukhara (Pān 1997). With the uvular stops, the relationship between glottal and velar stops of some semantically related words can be explained; for example, jǐng 景 kɯiaŋ, ‘scenery, shadow’, yǐng 影 ʔɯiaŋ ‘shadow,’ gōng 公 kuŋ ‘male, old man,’ and wēng 翁 ʔuŋ ‘old man.’ The initial is *q- before it changes to k- and ʔ-, respectively. Pān’s uvular proposal is for a uvular series, which is summarized in Table 2.64. In his proposal Pān (1997) inclined to but did not exclude the *ʔ-, *h-, and *ɦ- as possible initials of Old Chinese. 2.4 Examples of Old Chinese Two stanzas of Shījīng 詩經 poems illustrate the phonetic values of Old Chinese around the sixth–fifth century BC, when these poems were recorded. Of the phonetic transcriptions in Table 2.65 and Table 2.66, the first line is the Old Chinese reconstructed value (Zhèngzhāng [2003] 2013), the second is the Middle Chinese reconstructed value (Pān 2000), the third is the phonetic value of Modern Mandarin, and the fourth is pinyin. Jìng Nǚ 靜女 ‘Gentle Girl’ (Stanza I), anonymous 靜女其姝 俟我於城隅 愛而不見 搔首踟躕

A gentle girl and fair Awaits by the crook of the wall; In shadows I don’t see her; I pace and scratch my hair. Source: English translation by Stephen Owen 1996: 44.

Old Chinese

103

Table 2.65 The reconstructions of various stages of Chinese, reflecting the original Old Chinese pronunciation of Jìng Nǚ 靜女 靜 *zleŋʔ dzjᴇŋ tɕiŋ jìng

女 *naʔ ɳjɔ ny nǚ

其 *ɡɯ ɡɨ tɕhi qí

姝 *thjo tɕhjo ʂu shū

俟 *sɢrɯʔ ʐɨ sɨ sì

我 *ŋaalʔ ŋɑ wo wǒ

於 *qa Øjɔ y yú

城 *djeŋ dʑjɛŋ tʂəŋ chéng

愛 *qɯɯds Øəj aj ài

而 *njɯ ȵɨ ɚ ěr

不 *pɯd pjut pu bú

見 *geens ɦen ɕjɛn xiàn

搔 *suu sɑw saw sāo

首 *hljuʔ ɕiw ʂəw shǒu

踟 *de ɖjɛ tʂɨ chí

躕 *do ɖjo tʂu chú

隅 *ŋo ŋjo y yú

Táo Yāo 桃夭 ‘Peach Tree Soft and Tender’ (Stanza I), anonymous 桃之夭夭 灼灼其華 之子于歸 宜其室家

Peach tree soft and tender, How your blossoms glow! The bride is going to her home, She well befits this house. Source: English translation by Stephen Owen 1996: 34.

Table 2.66 The reconstructions of various stages of Chinese, reflecting the original Old Chinese pronunciation of Táo Yāo 桃夭 桃















*l’aaw dɑw thaw táo

*tjɯ tɕɨ tʂɨ zhī

*qrow Øɰjɛw jaw yāo

*qrow Øɰjɛw jaw yāo

*pljewɢ tɕjɐk tʂwo zhuó

*pljewɢ tɕjɐk tʂwo zhuó

*ɡɯ ɡɨ tɕhi qí

*ɡʷraa hwɰa xwa huá

之 *tjɯ tɕɨ tʂɨ zhī

子 *ʔslɯʔ tsɨ tsɨ zǐ

于 *ɢwa ɦjo y yú

歸 *klul kwɨj kwej guī

宜 *ŋral ŋɰjɛ i yí

其 *ɡɯ ɡɨ tɕhi qí

室 *hliɡ ɕit ʂɨ shì

家 *kraa kɰa tɕja jiā

Part III

Middle Chinese Northern and Southern dynasties (AD 420–589), Sui dynasty (AD 581–618), Tang dynasty (AD 618–907), Five Dynasties and Ten Kingdoms period (AD 907–960), Song dynasty (AD 960–1279)

The quality of source materials for the study of Middle Chinese is much more systematic than those for Old Chinese. Middle Chinese phonology is mainly based on the information from the phonological system of the Qièyùn 切韻, a rhyme dictionary compiled in 601 by Lù Fǎyán 陸法言 of the Sui dynasty. The literary standard represented by the Qièyùn and its later revisions served as the rhyming standard for centuries, even up until modern times. The Qièyùn represents the phonology of the Northern and Southern dynasties although its publication in 601 places it in the Sui dynasty. Modifications and revisions of this system were done throughout the imperial examination era. In the Tang dynasty, Wáng Rénxù’s 王仁昫 Kānmiù Bǔquē Qièyùn 刊謬補缺切韻 (706) became popular, while in the Song dynasty, Chén Péngnián et al.’s Guǎngyùn 廣韻 (1008), Dīng Dù 丁度 et al.’s Jíyùn 集韻 (1039), and simplified versions such as the Lǐbù Yùnlüè 禮部韻略 ‘Simplified Rhyme Dictionary by the Department of Rites’ were prominent. These rhyme works have completely survived and are used as main references in the study of Middle Chinese phonology. The phonological system of the Qièyùn as a rhyming standard is closely related to the imperial examination. In a country with many local dialects, a national standard had to be established for the examination; the existing Qièyùn seemed to fit this purpose well. Once the phonological standard was established, it became almost impossible to change it in the context of the imperial examination for both the examinees and examiners. The Qièyùn system could be modified, but the basic structure had to be kept unchanged. As time went on, this system became less and less related to the real pronunciation of any dialect in any time period, becoming instead an artificial coding for the composition of poetry. Since it was the standard used in the imperial examination, however, it remained prestigious and continued to be taught nationwide. The literary standard for poetry rhyming was simplified as many of the distinctions were no longer existant in the spoken language. The simplification of the rhyming categories of the Qièyùn represents the realization of the 105

106

Middle Chinese

unpractical details of said rhyming categories. Starting in the Tang dynasty, as an effort by the government, decisions were made to merge rhyming categories (refer to Section 4.1.2). The final result can be found in the Guǎngyùn, in which each rhyme is labeled either as dúyòng 獨用 ‘independent’ or tóngyòng 同用 ‘shared.’ Shared rhymes are the ones that cannot for the most part be distinguished by those who spoke with the standard pronunciation of the time. The rhyming system, the commonly known 106 rhymes, was finalized in the thirteenth century and used as the rhyming standard from then on. It is worth noting that unlike the phonology of Old Chinese, the categorical information of Middle Chinese has been thoroughly analyzed in the rhyme tables, such as the děng 等 ‘division’ as well as kāikǒu 開口 and hékóu 合口 ‘unrounded and rounded,’ which provide information about the relationships between the rhyming categories of the Qièyùn system. The reconstruction of Middle Chinese phonology requires the reconstruction of phonetic values for the already existing phonological categories. The reconstruction of the phonetic values of the phonology of Middle Chinese started in the early twentieth century by Western scholars, spearheaded by Bernard Karlgren’s monumental work, Études sur la Phonologie Chinoise (1915–1926). In all the proposals of phonetic reconstructions, the categorical contrasts of the Qièyùn are strictly preserved and the phonological labels of the rhyme tables are consistently followed. Middle Chinese phonology also has its variants in time and space. It has been proposed that Middle Chinese should be further divided into Early Middle Chinese and Late Middle Chinese, with the former represented by the Qièyùn and the latter by early rhyme tables such as the Yùnjìng 韻鏡. The colloquial language of the Five Dynasties period and the Song dynasty reveals characteristics of a new phonological system. Based on the characters used for the phonetic notations of zhíyīn 直音, a number of phonological characteristics can be identified in the Ěryǎ Yīntú 爾雅音圖, written in the tenth century. It shows the devoicing pattern of Middle Chinese voiced obstruents as seen in Mandarin phonology. The Shēngyīn Chànghè Tú 聲音 唱和圖 by Shào Yōng 邵雍 (1011–1077) displays phonological characteristics that seem to be “Mandarin,” although it is still not clear whether the Shēngyīn Chànghè Tú is a description of Shào Yōng’s home dialect, the Luoyang dialect, or a system that includes characteristics of both.

3

Middle Chinese: The Qièyùn 切韻

The reconstruction of Middle Chinese phonology, in a more accurate sense, is the reconstruction of phonetic values for the existing categories, which are commonly assumed to be phonemic. Because of the rigorous categorical nature of the Qièyùn, its categories do not require any reconstruction. The arrangements of the rhyme tables, especially structural terminologies such as division/rank and kāikǒu 開口 and hékǒu 合口 ‘unrounded and rounded,’ are commonly used to determine the relationships between the rhyming categories of the Qièyùn system. In many proposed reconstructions, all phonological contrasts between the categories of the Qièyùn are strictly preserved and the phonological labels of the rhyme tables are closely followed. It should be noted, however, that this commonly accepted practice is actually quite problematic. The Qièyùn system is at least three centuries older than the rhyme tables, and the Qièyùn itself was based on phonological systems from different geographical locations than each of the rhyme tables. It is by no means guaranteed that categorical labels from the rhyme tables of the tenth century or later are suitable for the phonology of the Qièyùn. The phonetic values of Qièyùn categories, on the other hand, are difficult to determine since the Chinese script is logographic and does not provide direct phonetic information. For example, the phonetic values of the rhymes dōng 東 and dōng 冬 cannot be directly determined, though we know that in the Qièyùn the two rhymes were different from one another. In comparison with the Qièyùn, modern Chinese dialects are phonologically simpler and preserve only some of the phonological contrasts of the Qièyùn. In the Qièyùn there are more categories in the initial system, the medial system, and the system of nuclear vowels than in most modern dialects. To reconstruct such a complex phonology is therefore a very challenging task for historical phonologists. Since Karlgren’s initial attempt to reconstruct the phonetic values of the Middle Chinese categories, many proposals have been put forward. Most attempts made to reconstruct the phonology of the Qièyùn are based on the assumption that it is a single phonological system, be it colloquial or literary, and that all contrasts must be distinguished phonetically. It is quite ironic that while most historical phonologists now realize that the Qièyùn is not 107

108

Middle Chinese

Table 3.1 Evidence of codas -m, -n, -ŋ and -p, -t, -k in the Qièyùn system Rhyme Character Beijing Suzhou Guangzhou

唐 唐 thaŋ dɒŋ thɔŋ

曾 曾 tsəŋ tsən tʃɐŋ

庚 庚 kəŋ kaŋ tʃɐŋ

東 東 tuŋ toŋ tʊŋ

寒 寒 xan ɦø hɔn

真 真 tʂən tsən tʃɐn

覃 潭 than dø tham

侵 侵 tɕhin tshin tʃhɐm

Rhyme Character Beijing Suzhou Guangzhou

鐸 洛 luo loʔ lɔk

德 德 tɤ tɤʔ tɐk

陌 陌 mo mɒʔ mɐk

屋 屋 u uɤʔ ʊk

渴 渴 khɤ khɤʔ hɔt

質 質 tʂʅ tsɤʔ tʃɐt

合 合 xɤ ɦɤʔ hɐp

緝 急 tɕi tɕiɪʔ kɐp

based on a single dialect, in practice, all attempts at reconstruction assign a unique phonetic value to all Qièyùn categories. Thus it is inevitable that the reconstructed Qièyùn system will contain some questionable distinctions. Because of their large number, phonetically reconstructing rhyme categories is always a challenge for scholars. In practice, the reconstruction is based on information from modern dialects and Sino-xenic languages. As shown in Table 3.1, nasal and stop codas can be retrieved through a comparison of multiple Chinese dialects, with the dialectal data suggesting the existence of codas -m, -n, -ŋ and -p, -t, -k in the Qièyùn system. 3.1

Special Notes on the Qièyùn 切韻 and Fǎnqiè 反切

Before delving into the reconstruction of the phonetic values of the Qièyùn system and rhyme tables, it is important to have a grasp of the basic terminology used in both the Qièyùn and its phonetic reconstruction, as well as some of the particulars of the Qièyùn and fǎnqiè. For reference to the terminology, it would be good to first read the first chapter (Section 1.2.3 through Section 1.4.6) of this book. For the particulars, we look at the following subsections. 3.1.1

The Fǎnqiè of Rank-III

The fǎnqiè spellings for Rank-III syllables have a strong tendency for both the upper speller and lower speller to also be Rank-III syllables. Table 3.2 gives some examples of the relationships between the syllables to be spelled and the upper spellers of their fǎnqiè. The characters jū 居, xū 墟, yú 魚, and xū 虛 are all Rank-III syllables with the upper speller of their fǎnqiè also Rank-III (jǔ 舉 qù 去 yǔ 語 xǔ 許). In his Études sur la Phonologie Chinoise (1915–1926), Karlgren reasoned that the initials of Rank-III syllables are

The Qièyùn 切韻

109

Table 3.2 The fǎnqiè spellings and Middle Chinese reconstruction for Rank-III syllables, with the lower speller of jū 居 or yú 魚a 居 jū ‘reside’ kjɔ 舉魚 kjɔ + ŋjɔ

墟 xū ‘mound’ khjɔ 去魚 khjɔ + ŋjɔ

魚 yú ‘fish’ ŋjɔ 語居 ŋjɔ + kjɔ

虛 xū ‘void’ hjɔ 朽居 hiw + kjɔ

All words will rhyme with yú 魚, while the word yú 魚 itself has the lower speller jū 居 in order to not be self-referential. a

Table 3.3 Karlgren and his successors’ reconstructions Character Karlgren Chao and Lǐ

IʲjV(E) IjV(E)

Upper speller > >

IʲIj-

Lower speller + +

-jV(E) -jV(E)

Table 3.4 Upper spellers of Rank-III and non-Rank-III characters Character Upper speller

Rank-III Rank-III

Rank-I, -II, -IV non-Rank-III

palatalized (refer to Section 1.3.2 for more information); this idea was not well received. Scholars (Chao 1941; Lǐ Róng 1956) argue that this could be due to the harmonic principle involved, which indicates that the palatal medial -jshould be reflected in both the upper speller and the lower speller. Table 3.3 shows the differences between reconstructions among Karlgren and his successors. Note the palatization marker present in Karlgren’s work. For a discussion on the proof behind whether Rank-III syllables have a palatal medial -j-, see Section 3.5.4.1. Here “IjV(E)” indicates a Rank-III syllable with a palatal medial -j-, with “(E)” representing an optional coda. In the fǎnqiè spelling, medial -j- is included in both spellers. A syllable with Ij- (initial and a palatal medial) and a syllable with -jV(E) are used to spell Rank-III IjV(E) syllables. In contrast, the syllables of Divisions-I, -II and -IV rarely use Rank-III syllables as upper spellers in their fǎnqiè (see Table 3.4). The fǎnqiè spellings indicate the existence of a palatal medial in the syllables that were classified as Rank-III in the rhyme tables; this medial must have still been distinct during the creation of the fǎnqiè notation, otherwise the selection of the upper spellers cannot be easily explained.

110

Middle Chinese

3.1.2

The Information Contained within the Qièyùn

The problem in looking at these rhyme tables is whether the information they contain is consistent with the phonology of the Qièyùn, for example, in the categorization of the four Divisions and the kāi 開 and hé 合 rhymes. It is not always clear how authors of the rhyme tables made decisions concerning the phonological categories for all rhymes. How did they know which rhyme is Division-III and which rhyme is Division-IV, which is kāikǒu 開口 and which is hékǒu 合口? A possible answer is that the arrangements are made according to the fǎnqiè spellings, the possible combinations of the rhymes with types of the initials (see Section 1.4), and the real pronunciation of the date. Thus, the rhyme tables actually have two different phonological systems embedded, a phonological system of the time when the rhyme tables were made, and the phonological categories of the Qièyùn or the Guǎngyùn. The discrepancies are basically shown in the placements of the syllables in different děng, or divisions (see Section 3.5.4 for discussion on the Qièyùn medials and their reconstruction). The problem lies with the initials and the medials. In order to keep all the distinctions of the Qièyùn/Guǎngyùn system, all syllables must be placed in the phonological framework of the rhyme tables, to fit both the initial categories and the divisions of the final categories (see Table 3.5). The Qièyùn syllables with no medial (-Ø-) are arranged in either Division-I or Division-IV; these syllables with /e/ are arranged in Division-IV and the rest are in Division-I. This is due to a phonological change that happened after the Qièyùn was written, in which the syllables with the front vowel /e/ developed a -j- medial. This medial, -j-, is distinctive from the syllables placed in DivisionIII, -ɰj-. The Qièyùn rhymes which may have either -j- or -ɰj- medials (the socalled chóngniǔ syllables, refer to Section 1.4.6 and Section 3.5.6) are put into Division-III and Division-IV respectively in rhyme tables. Table 3.5 The divisional placement of the Qièyùn rhymes in the rhyme tables Qièyùn medial Qīyīn Lüè 七音略 division

-ØI 高gāo ‘high’ kɑw

-ØIV 澆 jiāo ‘water’ kew

-ɰII 交 jiāo ‘cross’ kɰaw

-ɰjIII 驕 jiāo ‘arrogant’ kɰjɛw

-jIV –

褒 bāo ‘praise’ pɑw



包 bāo ‘wrap’ pɰaw

鑣 biāo ‘bit’ pɰjɛw

飈 biāo ‘whirlwind’ pjɛw

The Qièyùn 切韻

111

Table 3.6 The divisional placement of the Rank-III syllables of the Qièyùn in the rhyme table Yùnjìng Qièyùn

Condition

Yùnjìng

Other name

medial is -ɰjmedial is -j-

Division-III Division-IV

三等b類 (IIIb) 三等a類 (IIIa)

initial is tʂ- series initial initial is ts- series initial initial is ɦinitial is zero (no initial) otherwise

Division-II Division-IV Division-III Division-IV Division-III

Rank-III rhyme which contains contrast of medial -ɰj- and -j-

Rank-III rhyme which contains only syllables with palatal medial -j-

喻母三等 (喻三) 喻母四等 (喻四)

The difficulty involved in the study of Middle Chinese phonology is that the division (děng) concept of the rhyme tables is assigned to the rhymes of the Qièyùn and is often used as if děng is a concept of the Qièyùn system rather than one that follows it by hundreds of years. Thus, in the study of Middle Chinese the concept of děng can at times be very confusing. Table 3.6 is an illustration of the relationship between the Qièyùn and the Yùnjìng; it shows how the syllables of the Rank-III rhymes or the so-called “Division-III” rhymes of the Qièyùn are listed in the Yùnjìng according to the děng system. Sometimes the divisions are named with certain additional modifications such as in the rightmost column. Such terminology is quite commonly used in the literature of Middle Chinese studies. From this correspondence table it can be seen that the divisions of the rhyme tables (Yùnjìng) have a quite complex relationship with the rhymes of the Qièyùn. This assignment sometimes divides the syllables in the same Qièyùn rhyme into different Divisions due to their differing initials and medials. But on the other hand, the Division assignments in the rhyme tables provide very important information about the syllable types of the Qièyùn rhymes. For example, rhyme tables list the syllables of rhyme dōng 東 of the Qièyùn in Division-I (with no medial) and Division-III (with medial -j-) and list the syllables of the má 麻 rhyme in Division-II (with -ɰ- medial) and DivisionIII (with medial -j-). Certain rhymes will only pair with specific divisions, and thus, in practice the assignment of division to the Qièyùn (called “rank” in this book) are commonly used as a means to describe and determine the phonetic identity of Qièyùn rhymes.

112

Middle Chinese

3.1.3

Some Main Issues of the Qièyùn

According to fǎnqiè spellings used, the phonological system of the Qièyùn is very similar to that of Lù Démíng’s 陸德明 Jīngdiǎn Shìwén 經典釋文 (583), and also very similar to that of Gù Yěwáng’s 顧野王 Yùpiān 玉篇 (548). Both were compiled before the Qièyùn, making it unquestionable that the Qièyùn truly presents the phonological standard of the sixth century. Although the Qièyùn system has been the focus of Chinese historical phonologists since Karlgren’s study (1915–1926), the nature of this rhyme work itself has since been the center of controversies, concerning fundamental aspects of the Qièyùn such as: (1) whether it is a single system or a composite system, whether geographical variants are included and whether historical variants are included; and (2) what the nature of the yùn 韻, the basic category of the Qièyùn, is. 3.1.4

A Single System or a Composite System

Worthy of discussion is the question, what does a phonological standard really mean? A better understanding of what a standard is can help us to understand the Qièyùn system in particular, as well as other rhyme dictionaries in general. Until very recently there was probably never the requirement for a standard to be strictly based on a single dialect from a certain point in time. On the contrary, in ancient times a standard was always a composite, though it may have been based on a particular dialect. The high respect given to ancient standards resulted in archaic features still being regarded as standard. This ideal phonology is what has been called the “literary phonology” or “literary pronunciation” by Chén Yínkè (1949), Zhōu Zǔmó (1963), and other scholars. This has been the tradition of the Chinese phonological standard and has been reflected in the phonological standards throughout history, even in the relatively recent Guóyīn Zìdiǎn 國音字典 ‘Dictionary of National Pronunciation’ of 1920. This national standard incorporated a number of phonological distinctions into Beijing phonology, including the contrast between alveolar and velar consonants before high front vowels and medials. Such a system is standard but is not the actual pronunciation of any place. However, it should also be noted that the different phonological systems included in the Qièyùn were closely related, namely the systems reflected in the five rhyme dictionaries mentioned by Lù Fǎyán, as well as his own understanding. Based on the Kānmiù Bǔquē Qièyùn, it seems the discrepancies among these systems were not great. It is unrealistic to imagine that the scholars mentioned in the preface of the Qièyùn knew much beyond what was recorded in the previous rhyme works or much beyond the dialects in the time and space in which they were active.

The Qièyùn 切韻

113

Such an ideal phonology continuously evolved throughout history, adjusting itself to new political, social, and linguistic conditions. This long tradition becomes a serious problem when modern phonologists start trying to “reconstruct” it and pin it down to a certain time and location. Karlgren, for example, tries to identify the Guǎngyùn/Qièyùn system as the Chang’an dialect of the Tang dynasty (Karlgren 1954: 212). Although his reconstruction implements innovative Western methodology and phonetic values, the lack of understanding of the Chinese phonological tradition is clearly a blind spot in the work itself. The evidence is clear that all retrievable distinctions are preserved. In the Kānmiù Bǔquē Qièyùn, special annotations from the original Qièyùn (Zhōu Zǔmó 1983: 863–865, 867–868, 883) are provided to indicate which distinction is made by which phonological work and are among the most valuable pieces of information from this version of the Qièyùn. In the annotations, it is clear that the decision was always made to maximize phonological contrast. In the preface of the Qièyùn,1 Lù Fǎyán mentions five phonological works: 呂靜 《韻集》 Yùnjí by Lǚ Jìng 夏侯詠《韻略》 Yùnlüè by Xiàhóu Yǒng 陽休之《韻略》 Yùnlüè by Yáng Xiūzhī 李季節《音譜》 Yīnpǔ by Lǐ Jìjié 杜臺卿《韻略》 Yùnlüè by Dù Táiqīng In the Guǎngyùn, an additional work, Zhōu Sīyán’s 周思言 Yīnyùn 音韻 is also listed; however, this work is not found in earlier prefaces of the Qièyùn or in the notations regarding discrepancies between phonological works. Zhōu’s work was likely added in by the compilers of the Guǎngyùn. If there is a discrepancy among the referential works regarding whether two phonological categories should be distinguished, the decision is always made in favor of making a distinction. For example, the notation for the zhī 脂 rhyme mentions: 呂、夏侯與之、微大亂雜,陽、李、杜別。今依陽、李、杜。 (According to) Lǚ and Xiàhóu (the zhī 脂 rhyme is) greatly mixed with the zhī 之 and wēi 微 rhymes, (but according to) Yáng, Lǐ and Dù (there is a) distinction (between them). (We) now follow Yáng, Lǐ, and Dù.

In the parallel shǎng and qù rhymes zhǐ 旨 and zhì 至, the same notation can be found. The decision is not based on whose pronunciation is more correct or standard, but rather on who can still make a distinction. In the dōng 冬 rhyme, the notation is: 1

An English translation of Lù Fǎyán’s preface of the Qièyùn can be found in Ramsey’s The Languages of China (1987: 116–117).

114

Middle Chinese

陽,與鍾、江同,呂、夏侯別。今依呂、夏侯。 (According to) Yáng, (the dōng 冬 rhyme is) the same as zhōng 鍾 and jiāng 江, (but according to) Lǚ and Xiàhóu (there is a) distinction (between them). (We) now follow Lǚ and Xiàhóu.

The situation here is reversed in terms of individuals. Because Lǚ and Xiàhóu can still make a distinction, the decision is made in their favor, even though they were ruled against in the zhī 脂 rhyme. It is clear that Lù Fǎyán decided to include all distinctions made in the referenced phonological works. Sometimes Lù Fǎyán made the decision himself, not mentioning the names of other rhyme dictionaries. For example, in the hún 魂 rhyme he noted, 呂、陽、夏侯與痕同,今別. (According to) Lǚ, Yáng, and Xiàhóu (the hún 魂 rhyme) is the same as the hén 痕 rhyme. (We) now distinguish (them).

Thus, Lù’s method of distinguishing can be summed up as follows: A: Ra = Rb, A: Rc ¼ 6 Rd,

B: Ra 6¼ Rb B: Rc = Rd

> >

Ra ¼ 6 Rb Rc ¼ 6 Rd

All phonological distinctions must be recognized, respected, and kept. Such an archaism-based approach reflects the criterion of respecting all the phonological distinctions that ever existed. Based on such a criterion of maximizing distinctions, the Qièyùn is a composition of all the phonological distinctions known to the author. Scholars call such a composite system a “literary system” because its distinctions cannot be pronounced and only exist in the practice of poetry composition. The tradition of archaism was very strong in the establishment of national standards until the modern era when the modern linguistic concept of a single live language became familiar to Chinese linguists. So, the traditional rhyme books always contain phonological contrasts that are not based on one single phonological system. Based on these notes it is unlikely that all the distinctions in the Qièyùn existed in a single phonological system at Lù Fǎyán’s time. They are, however, likely to be real distinctions historically and could be from a single earlier dialect; the dating of such a dialect would be earlier than Lù Fǎyán’s time and also earlier than the authors whose works were quoted in the Qièyùn. These authors likely compiled their rhyme dictionaries based on the same principle Lù Fǎyán adopted, which was to keep all distinguishable categories they could identify. Some scholars, such as Abraham Chan (2004), have done reconstructions based on the idea that the phonology of the standard literary language was based on the pronunciation of the dialects of Jīnlíng 金陵 (modern Nanjing), Luòyáng 洛陽, and Yèxià 鄴下 (in Henan province). But as been pointed out, the distinctions are actually based on previous rhyming books that were also composite; there is no reason to assume that these works

The Qièyùn 切韻

115

are based on a single contemporary dialect. The two dialects that Chan reconstructed, Nanjing and Luoyang, turn out to be not very different from each other. It is also possible that they are not the phonological system from these two locations at the time of Lù Fǎyán either, as Chan’s separation of the two dialects is based on notes in the Qièyùn concerning previous rhyme dictionaries (something he does not state explicitly in his writing). These rhyme dictionaries were composed during the Northern and Southern dynasties period before the Sui dynasty. As with other rhyme dictionaries, it is not very likely that these authors compiled their books based on a single dialect. The nature of the rhyme dictionaries compiled earlier is likely to be similar to the Qièyùn, which is ideal and thus composite. Thus, the two different “dialects” are actually two slightly different composite systems, a little less composite than the Qièyùn, and there should be no surprise that they are very similar. As a literary standard, the Qièyùn is not a record of a single contemporary dialect. However, no one has seriously argued that any of the distinctions are artificially created. All of the Qièyùn categories must have existed at some time in history and probably in one single dialect. The categories that the Qièyùn recognizes are either real at the time or historically real; there is no simple reason to exclude any of them. The Qièyùn was most likely more historically accurate to an ancestral language than its individual references (see Table 3.7). The author of the Qièyùn, Lù Fǎyán, actually tried to keep all the traditional categories that existed historically rather than using his own pronunciation or the pronunciation of his contemporaries in order to write these rhyme tables. At the time of the Qièyùn’s compilation, these categories were incompletely kept in different versions of rhyme dictionaries. The Qièyùn represents an effort to restore all the known historical categories. Scholars (e.g. Baxter & Sagart 2014: 12) sometimes refer to the phonological system of the Qièyùn as a literary reading system (dúshū yīn 讀書音). But it is very doubtful that all the contrasts of a phonological standard can be realized in real pronunciation by anyone who lived at the same time or after the time of the Qièyùn, including Lù Fǎyán himself. These contrasts are not based on the real pronunciation of a single phonological system. Table 3.7 The nature of the Qièyùn rhymes Time 1

Time II

Time III

Ancestral: Ra 6¼ Rb 6¼ Rc

Reference 1: Ra = Rb 6¼ Rc Reference 2: Ra 6¼ Rb = Rc

Qièyùn: Ra 6¼ Rb 6¼ Rc

Note: R is used for “Rhyme,” and Ra, Rb, and Rc represent three different Qièyùn rhymes.

116

Middle Chinese

It should be noted that the analysis of the Qièyùn phonology often refers to the categories used in the late rhyme charts. However, there is a time gap of a few hundred years between the Qièyùn and the rhyme charts. It is very likely that the phonological information of the two systems may have changed significantly. For example, which rhyme is kāi 開 and which rhyme is hé 合 as labeled in the rhyme charts may or may not be the same as in the actual phonology of the fifth or sixth century. The interpretation of the Qièyùn system is heavily based on the categorical information from the rhyme tables. But it is unlikely that the authors of tenth-century rhyme charts could accurately reconstruct the phonetic values of the phonological categories of the fifth or sixth century, and for this reason a complete dependence on the categorical information of the rhyme charts in reconstructing the Qièyùn system presents a serious problem. Therefore it must be made clear that the rhyme charts are the analyses of the Qièyùn but not the Qièyùn itself. 3.1.5

The Qièyùn Rhymes and the Nature of the Yùn 韻

As discussed in Section 1.3.5, in the Qièyùn, corresponding píng, shǎng, and qù rhymes have the same phonetic values. The rù rhymes differ from their corresponding rhymes in the coda, a stop instead of a nasal with the same places of articulation. Examples of the parallel nature of rhymes across different tones are given in Table 3.8, showing that there is no need to list all corresponding rhymes for reconstruction purposes. For the purpose of convenience there is no need to list all the corresponding rhymes. The rhymes dōng 東 (j)uŋ, dǒng 董 (j)uŋ, sòng 送 (j)uŋ, and wū 屋 (j)uk together can be represented by dōng 東 (j)uŋ/k, or just dōng 東 (j)uŋ. As shown in Table 3.9, the relationship between rù 入 and shū 舒 (non-rù; or píng, shǎng, and qù) rhymes are based on the order of the Qièyùn. This particular reconstruction strictly follows the principle that different rhymes must differ from each other in the main vowel, coda, or both (the only exception is dōng 冬 and zhōng 鍾 rhymes, which are reconstructed with the same main vowel and coda). A table-style organization by main vowel and coda based on Pān (2000) is given in Table 3.10; note that in this

Table 3.8 The parallel nature of rhymes across different tones píng

shǎng





1 東 -(j)uŋ 2 冬 -woŋ 3 鍾 -joŋ

1 董 -(j)uŋ — 2 腫 -joŋ

1 送 -(j)uŋ 2 宋 -woŋ 3 用 -joŋ

1 屋 -(j)uk 2 沃 -wok 3 燭 -jok

The Qièyùn 切韻

117

Table 3.9 195 rhymes from Wáng Rénxù’s version of the Qièyùn -ŋ

1 37 41

東 uŋ/juŋ 陽 jɐŋ 清 jɛŋ

2 冬 woŋ 38 唐 ɑŋ 42 青 eŋ

3 39 49

鍾 joŋ 庚 ɰaŋ/ɰjaŋ 蒸 ɨŋ

4 40 50

江 ɰɔŋ 耕 ɰæŋ 登 əŋ



5 9 33

支 jɛ/ɰjɛa 魚 jɔ 歌 ɑ/jɑ

6 脂 i/ɰi 10 虞 jo 34 麻 ɰa/ja

7 11

之 ɨ 模 wo

8 13

微 ɨi 佳 ɰæ

-j

14† 泰 ɑj 17† 夬 ɰaj

12 齊 ej 15 灰 woj

13† 祭 jɛj/ɰjɛj 16 咍 əj

-n

17 21 24 28

真 元 寒 仙

18 臻 in 22 魂 won 25 刪 ɰan

19 23 26

文 jun 痕 ən 山 ɰæn

20

殷 ɨn

27

先 en

-w

29 43

蕭 ew 尤 iw

30 宵 jɛw/ɰjɛw 44 侯 əw

31 45

肴 ɰaw 幽 ɨw

32

豪 ɑw

-m

35 48 54

覃 əm 添 em 凡 jɐm

36 談 ɑm 51 咸 ɰæm

46 52

侵 im/ɰim 銜 ɰam

47 53

鹽 jɛm/ɰjɛm 嚴 jɐm

in/ɰin jɐn ɑn jɛn/ɰjɛn

14 皆 ɰæj 20† 廢 jɐj

Note: As reconstructed based on Pān (2000) with rhymes ordered by coda. The numbers indicate the order of the rhymes in píng tone. Four rhymes (祭, 泰, 夬, and 廢) which are in qù tone only are marked with † (see Section 3.5.8.1). The finals with -w- medial are not included. a Pān’s reconsctruction for the zhī 支 rhyme is ie/ɰie (2000: 86). In this book jɛ/ɰjɛ will be used instead.

Table 3.10 Pān’s reconstruction (2000) organized in table style

-Ø -j -w -m/-p -n/-t -ŋ/-k

ɑ

a

æ

歌一三 泰一 豪一 談一 寒一 唐一

麻二三 夬二 肴二 銜二 刪二 庚二三

佳二 支三 皆二 祭三 宵三 咸二 鹽三 山二 仙三 耕二 清三

脂三 魚三 齊四 蕭四 添四 侵三 先四 真臻三 青四 江二

II, III

II

IV

Rank I, III

ɛ

III

e

i

III

ɔ

o

u

模一虞三 尤三 灰一

ɐ

ə

廢三

咍一 侯一 覃一 痕一 登一

嚴凡三 魂一 文三 元三 冬一鍾三 東一三 陽三

II, III I, III

I, III

III

I

ɨ 之三 微三 幽三 殷三 蒸三 III

reconstruction, the main vowel to a certain degree hints at which rank the rhyme belongs to. Fǎnqiè spellings of the Qièyùn suggest a palatal medial. More recently, scholars (Zhèngzhāng 2001; Shī 1983, etc.) have provided convincing

118

Middle Chinese

evidence to show that Rank-II syllables still had a medial in the Tang dynasty. In the Táng-Fān Huìméng Bēi 唐蕃會盟碑 ‘Stele of Forming Alliance between the Tang dynasty and Tubo Kingdom’, the transcription of jiàng 絳 in Lǐ Jiàng 李絳 ‘a personal name’ is khaŋ. The h indicates the medial -ɰ- of Rank-II in kɰɔŋ (Zhèngzhāng’s kɣʌŋ). The transcription of jiǎn 檢 in the Jiǎnxiào Shàngshū 檢校尚書 ‘Acting Adviser to the Director’ is khem. The h indicates the medial -ɰ- of Rank-IIIb in kɰjɛm (Zhèngzhāng’s kɣiᴇm) (Zhèngzhāng 2001). This piece of information was not available at the time Karlgren reconstructed the Middle Chinese system. In previous reconstructions like the one in Table 3.10, the contrast between Rank-II and Rank-I syllables can only be the differences of the main vowels. The newly discovered medial of Rank-II syllables will have a strong impact on new reconstruction because by adding the medial of Rank-II syllables to the system, the vowel contrasts between Rank-II and other divisions become redundant. For easy comparison, see Table 3.11 and Table 3.12, where only the corresponding main vowels and medials are listed (the reconstructed values are based on Pān 2000). Table 3.11 shows that main vowels (V) in the Pān (2000) reconstructions only have a limited number of medials (M) that they can pair with. Table 3.12 shows that through organizing the main vowels by medial, the redundancy of the complex vowel system can be seen. It is much more likely that contrasts between some rhymes are based on their medial than on both their medial and main vowel. There is a clear redundancy involved because there is a strong correlation between the main vowel and medial: all rhymes with main vowel -ɑ- (except rhyme 歌) basically have no medial, but all rhymes with main vowel -a- have medial -ɰ- and -j-; all rhymes with main vowel -æ- have medial -ɰ-, but all rhymes with main vowel -ɛ- have medial -j-; and so on. There are two possible causes for such redundancy: the difference of main vowels could be Table 3.11 Main vowels and medials of Pān’s reconstruction V M

ɑ Ø, j

a ɰ, j

æ ɰ

ɛ j

e Ø

ɔ ɰ

i j

o Ø, j

u Ø, j

ɐ j

ə Ø

ɨ j

Table 3.12 Pān’s reconstructed main vowels organized by medial -Ø-ɰ-ja

ɑ a ɑa

e æ a

o ɔ ɛ

The gē 歌 rhyme is the only exception.

u

ə

i

o

u

ɐ

ɨ

The Qièyùn 切韻

119

nonphonemic and rather allophonic, or the contrast between some rhymes could merely be a difference in medial. The first possibility indicates that too many phonetic details are involved in the reconstruction, and the second possibility indicates that some main vowels are unnecessarily reconstructed. Future reconstructions must take this problem into consideration. Since seven vowels is the maximum for rhymes with the same medial, it is very possible that the four degrees of vowel contrast of openness in earlier reconstructions can be reduced down to three degrees. 3.1.6

The Revisions of the Qièyùn and the Splitting of Rhymes

It should be mentioned that the Qièyùn has several revised versions. In these revisions of the Qièyùn the total number of rhymes increased from the original 193 to 195 in Wáng Rénxù’s Kānmiù Bǔquē Qièyùn, then to 206 in the Guǎngyùn. The splitting of rhymes does not mean an increase in finals, however; the different finals were grouped in the same rhyme in the earlier versions as shown by Lǐ Róng (1956: 82). Table 3.13 lists these splits. The increase in the number of rhymes from 193 to 195 was an effort to keep the rhyme categories parallel with píng, shǎng, qù, and rù, whereas the increase in the number of rhymes from 195 to 206 was an effort to differentiate kāikǒu and hékǒu syllables. The cause of these three sets of splits is commonly explained as the main vowel after medial -w- changing to a rounded vowel. The splits indicate that the main vowels became contrastive. But the cause of the splits can be simply based on medial -w-. What were once the same rhyme with different medials in the Kānmiù Bǔquē Qièyùn are considered separate rhymes in the Guǎngyùn; this Table 3.13 The rhyme splits between the Qièyùn, Kānmiù Bǔquē Qièyùn, and Guǎngyùn Qièyùn (193) yán 嚴, yè 業

>

Kānmiù Bǔquē Qièyùn (195) zhēn 真, zhěn 軫, zhèn 震, zhì 質 (kāi, hé)

>

hán 寒, hàn 旱, hàn 翰, hé 曷 (kāi, hé)

>

gē 歌, gě 哿, gè 箇 (kāi, hé)

>

Kānmiù Bǔquē Qièyùn (195) yán 嚴, yǎn 广, yàn 嚴, yè 業a Guǎngyùn (206) zhēn 真, zhěn 軫, zhèn 震, zhì 質 (kāi) zhūn 諄, zhǔn 準, zhùn 稕, shù 術 (hé) hán 寒, hàn 旱, hàn 翰, hé 曷 (kāi) huán 桓, huǎn 緩, huàn 換, mò 末 (hé) gē 歌, gě 哿, gè 箇 (kāi) gē 戈, guǒ 果, guò 過 (hé)

Note: The splits that occured between the Kānmiù Bǔquē Qièyùn and Guǎngyùn are all based along kāihé distinction. a In the Kānmiù Bǔquē Qièyùn, 嚴 has two pronunciations, píng (yán) and qù (yàn), each standing as the representative for their own separate rhymes.

120

Middle Chinese

Table 3.14 Rhyme splitting from the Kānmiù Bǔquē Qièyùn to the Guǎngyùn V(E) wV(E)

> >

V[–round](E) V[+round](E)

may have (but not necessarily) corresponded to a rounding of the main vowel in the condition of a following medial -w- (see Table 3.14). 3.2

The Nature of the Rhymes

One of the key issues that must be dealt with first is whether the main vowel of each rhyme must be phonemically different. Beginning with Karlgren, the principle has been that each rhyme must differ in the main vowel, the coda, or both. In practice scholars have allowed different degrees of freedom for this principle. It should be noted that due to its relative availability, Karlgren used the Guǎngyùn 廣韻 to reconstruct Middle Chinese. It is also worth noting that Karlgren did not like the American Structuralists2 and felt that they had become obsessed with the “phonemic principle” (Ramsey 1987: 132). So Karlgren’s reconstruction actually is phonetic instead of phonemic, therefore phonetic variations and complementary distributions were not examined. Recently scholars have begun to think that different rhymes can share the main vowel and coda but differ in medial (Mài 1995: 96). Some samples of two different approaches (Mài 1995; Pān 2000) are given in Table 3.15. The hékǒu syllables with medial -w- are not included. For easy comparison, the reconstructions are altered so that semi-vowel symbols are used for medials and codas in both reconstructions. One vowel (e) is used in Mài’s reconstruction, but more vowels (e, æ, iᴇ, ɛ, and a) are used in Pān’s system. Clearly there is a redundancy in having a different main vowel for every rhyme. For example, xiān 先 -en, shān 山 -ɰæn, and xiān 仙 -(ɰ)jɛn differ from each other not only in their main vowels but also in their medials. A crucial difference is that Mài’s approach seems to be based on the assumption that medial is part of difference in rhyming and, because of this, his reconstructed xiān 先 -en, shān 山 -ren, and xiān 仙 -(r)jen have the same main vowel and coda, differing only in the medials. These two different approaches stem from the rhyme dictionaries themselves; in some 2

The phonemic principle is the concept that any given language’s sound system was bound by a finite set of speech sounds (phonemes). All sounds within the language’s possible utterances are referable to a small set of phonemes, and that only those phonemes are significant to the given language. American Structuralists had been using the concept as a tool for simplifying hard data found in phonetics into smaller and more elegant categories of phonemics (Swadesh 1934).

The Qièyùn 切韻

121

Table 3.15 Samples of two different approaches (Mài 1995; Pān 2000) Rank-IV

Mài Pān

齊 ej ej

Rank-II

蕭 ew ew

先 en en

青 eŋ eŋ

添 em em

佳 re ɰæ

Rank-IIIA

Mài Pān a b

支 je jɛ

皆 rej ɰæj

肴 rewa ɰaw

山 ren ɰæn

耕 reŋ ɰæŋ

咸 rem ɰæm

宵 仙 rjew rjen ɰjɛw ɰjɛn

清 rjeŋ ɰjɛŋ

Rank-IIIB

祭 jej jɛj

b

宵 jew jɛw

仙 jen jɛn

清 jeŋ jɛŋ

鹽 jem jɛm

支 rje ɰjɛ

祭 rjej ɰjɛj

鹽 rjem ɰjɛm

The 肴 rhyme is also reconstructed as /raw/ in Mài’s system. 祭(齊三).

Table 3.16 An example of Mài’s (1995) reconstruction of Middle Chinese rhymes dōng 東 uŋ

juŋ

má 麻 ra

ja

gēng 庚 raŋ

jaŋ

xiān 先 en

xiān 仙 jen

jiē 皆 rej

jì 祭 jej

gēng 耕 reŋ

qīng 清 jeŋ

rhymes the medial is not considered a part of the rhyme, while in others it is. For example, in the dōng 東 rhyme there are two types of syllables, with or without medial -j-. A similar situation occurs in the má 麻 and gēng 庚 rhymes, where two types of syllables with different medials are grouped together as the same rhyme. Table 3.16 shows an example of Mài’s (1995) reconstruction of Middle Chinese rhymes. While some endings are the same regardless of their medial, such as dōng 東, má 麻, and gēng 庚, other syllables are categorized as different syllables based on their medials. According to the reconstructed values, the relationship between rhymes xiān 先 and xiān 仙, -en and -jen, is parallel to the two types of finals, -uŋ and -juŋ, of the dōng 東 rhyme (the same can be said for the relationships between má 麻 vs. jiē 皆/jì 祭 and gēng 庚 vs. gēng 耕/qīng 清). It brings up a question of inconsistency: why are -uŋ and -juŋ considered the same rhyme, while -en and -jen are different rhymes? One possible explanation offered (Mài 1995: 101) is that rhymes xiān 先 and xiān 仙 have different vowels phonetically, but phonemically they can be grouped together as variants of the same phoneme. Recently Huáng Xiàoshān (2002a, 2002b, 2004) also used the same approach to argue that the required

122

Middle Chinese

main vowels should be reduced to seven. This new approach indicates a change in a very fundamental assumption of the reconstruction, which is that different rhymes must be phonemically different either in the main vowel or in the coda, or both. Instead the phonological difference of contrastive rhymes can be the medial, and often only the medial (Yùchí 2007). To separate rhymes based on medial only was the practice of Xiè Tiǎo 謝脁 (464–499) and Shěn Yuē 沈約 (441–513), two famous literati and poets of the Northern and Southern dynasties. However, as Lù Fǎyán said in the preface,3 such a practice of rhyming was not the common practice for poetry composition (Yùchí 2007). The difference between the two standards is whether the rhyme includes the medial or distinguishes subphonemic difference of the main vowel. 3.2.1

New Approaches

The key issue is whether the main vowel of different rhymes with the same coda must be distinctive. The discussion on this issue has become more explicit with recent discourse about the nature of rhymes (Huáng Xiàoshān 2007; Yùchí 2007). The new approach suggests that the difference could be in the medial, with two different rhymes having the same main vowel and coda. If two rhymes have the same coda, the traditional approach suggests that the main vowel must be different regardless of whether there is a medial. But the modified approach adds a possibility that the main vowel could be the same if two rhymes differ in the medial position. A comparison of the Main Vowel Approach and the Medial Approach to yùn 韻 analysis is given in Table 3.17. Note that the Medial Approach results in a simpler vowel system, while the Main Vowel Approach results in rhymes not being reliant on medials for contrast. The Medial Approach actually suggests that the phonological unit, yùn 韻 ‘rhyme,’ in the rhyme dictionaries is not defined by main vowel and coda (VE) only, but also by the medial or the whole yùnmǔ 韻母 ‘final’ (MVE). This explanation is actually consistent with the poetry rhyming tradition of the Northern and Southern dynasties. Zhōu Zǔmó (1963) used the rhyming of Yán Zhītuī 顏之推, one of two scholars who made most of the decisions about rhyming categories of the Qièyùn 切韻, to show that Yán’s own rhyming practice is different from the Qièyùn rhyme categories. Rhyme categories as found in the Guān Wǒ Shēng Fù 觀我生賦, a prose poem by Yán Zhītuī 顏之 推, are given in Table 3.18. Yán Zhītuī’s rhyming method does not distinguish some of the rhymes in the Qièyùn. Yán’s rhyming indicates that the Qièyùn rhymes that he uses

3

欲賞知音,非廣文路 yù shǎng zhīyīn, fēi guǎng wén lù.

The Qièyùn 切韻

123

Table 3.17 The Main Vowel Approach and the Medial Approach to yùn 韻 analysis Main vowel approach (M)VaE jɛn ən

6¼ 6 ¼ 6¼

Medial approach (M)VbE en won

MaVaE jen on

6¼ 6 ¼ 6¼

(Mb)VaE en won

Table 3.18 Rhyme categories of Yán Zhītuī’s Guān Wǒ Shēng Fù yáng 陽 and táng 唐 (平) wén 文 (平) gēng 庚二庚三 and qīng 清 (平) xiān 先 and xiān 仙 (平) yáng 陽 and táng 唐 (去) mò 陌 (去) qīn 侵 (平) wū 屋 (入)

qí 齊 (入) xuē 薛 (入) háo 豪 (上) yuè 月 and mò 陌 (入) shān 刪 and xiān 仙 (平) yáng 陽 and táng 唐 (上) zhī 之 and zhī 脂 (上) zhēn 真 and zhēn 臻 (平)

Source: Based on Zhōu Zǔmó 1963: 462–463. Note: Tones are shown in parentheses.

together may have a distinction in the medial position only; for example, yáng 陽 and táng 唐 can be in rhyme because they are -jaŋ and -aŋ sharing the same rime (VE). Such a difference between actual rhyming practice and the rhyme categories of dictionaries is often treated as looseness in rhyme (called yùnkuān 韻寬 in Chinese discourse). However, in the Qièyùn, different finals with or without a medial can coexist in the same rhymes. The dōng 東 rhyme has Rank-I (without medial) and Rank-III (with medial -j-) syllables. The má 麻 and the gēng 庚 rhymes have Rank-II (with medial -ɰ-) and Rank-III syllables (with medial -j-). Such rhymes introduce a problem of inconsistency. It means that the finals that have a difference in the medial position sometimes belong to the same rhyme but sometimes belong to different rhymes. Neither Mài (1995) nor Huáng Xiàoshān (2002a, 2002b, 2004) offer any explanation for why different finals coexist in the same rhymes, such as in rhymes dōng 東, má 麻, and gēng 庚. 3.2.2

A Possible Solution

That various approaches cannot provide a satisfactory explanation of the Qièyùn rhymes is because they try to explain the rhyme categories with phonological reasons only. The question of this inconsistency between the

124

Middle Chinese

Qièyùn and its phonological analyses can be explained by the approach that Lù Fǎyán adopted to determine individual rhymes. It has been shown that in compiling the Qièyùn, five existing rhyme dictionaries were consulted. Lù Fǎyán and the scholars he worked with had a maximizing approach to the discrepancies found between different rhyme dictionaries; however, his work was to resolve discrepancies found only in these five dictionaries but not more. So it is important to realize that Lù Fǎyán’s approach is not really phonologically based. This is why the pure phonologically based analyses will have unsolvable problems. In this way, Qièyùn rhymes were not established based on the principle of the phonological analysis per se; rather, the traditional categories of the previous rhyme dictionaries influenced the determination of rhyme categories. It has been previously pointed out a few times that Lù Fǎyán did not compile the Qièyùn solely based on his own analysis. He consulted at least five existing rhyme dictionaries, which did not have the exact same rhyme categories. Thus, logically, for a pair of finals having the same main vowel and (optional) coda but different medials (or no medial) there could be four different scenarios across two rhyme dictionaries in terms of their categorization into separate rhymes. Let us write the different finals as VE and MVE and two rhyme dictionaries as A and B (Lù Fǎyán also may have acted according to his own belief, so for the sake of this, he himself can also be either A or B). Lù Fǎyán’s judgments can be interpreted as in Table 3.19. In all these finals, V and E are identical, with the only difference being the presence of a medial M. If dictionaries A and B have discrepancies as in Scenarios II and III, Lù Fǎyán would separate the VE and MVE into different rhymes. But if neither dictionary A nor B made a distinction (Scenario I), there would be no discrepancy for Lù Fǎyán to resolve and VE and MVE would be kept in the same rhyme. As shown in Table 3.20, for each set of finals that differ only in the medial, there are both rhymes that split along this medial difference and rhymes that do not. This scenario also explains the chóngniǔ 重紐 contrast -ɰjVE vs. -jVE (see Section 1.4.6 and Section 3.5.6). Rhymes such as gēng-III 庚三 vs. qīng 清 and yóu 尤 vs. yōu 幽 are distinguished, but other chóngniǔ syllables are in the same rhyme as zhī 支, zhēn 真, xiān 仙, xiāo 宵, qin 侵, yan 鹽, and ji 祭. The Table 3.19 An interpretation of Lù Fǎyán’s (LFY) judgments Scenario I Scenario II Scenario III Scenario IV

A: VE = MVE A: VE = MVE A: VE 6¼ MVE A: VE 6¼ MVE

B: VE = MVE B: VE 6¼ MVE B: VE = MVE B: VE 6¼ MVE

> > > >

LFY: VE = MVE LFY: VE 6¼ MVE LFY: VE 6¼ MVE LFY: VE 6¼ MVE

The Qièyùn 切韻

125

Table 3.20 Examples of the rhymes that split or not along the medial difference

VE vs. wVE jVE vs. VE ɰVE vs. jVE ɰjVE vs. jVE

Two rhymes

One rhyme

hāi 咍 vs. huī 灰 yáng 陽 vs. táng 唐 yáo 肴 vs. xiāo 宵 gēng-III 庚三 vs. qīng 清

zhēn真開 vs. zhēn真合 dōng-III 東三 vs. dōng-I 東一 gēng-II 庚二 vs. gēng-III 庚三 zhī-IIIb 支b vs. zhī-IIIa 支a

four scenarios for the decisions concerning a pair of VE and MVE finals shown in Table 3.19 can be the same for the decisions about a pair of finals with possibly different main vowels, (M)VaE vs. (M)VbE, or the same (M)VE after different initials, Ia(M)VE vs. Ib(M)VE. In this way, all theories of phonemic structure for rhymes must be second to the pure categorical information of the Qièyùn. Because the scope of Lù Fǎyán’s examination is limited to the five existing rhyme dictionaries, in the Qièyùn some finals that have the same phonological contrast are distinguished while others are not. The scenarios proposed can explain why similar pairs of finals are sometimes separated and sometimes not in the Qièyùn system, as is the case brought up by the question posed earlier in this section: -en and -jen are separate rhymes but -uŋ and -juŋ are the same rhyme. The phenomenon that dōng 東, má 麻, and gēng 庚 rhymes have finals with different medials should not be a sufficient condition to reason that different rhymes with the same coda must have different main vowels, such as yáng 陽 and táng 唐. Of all sixty-six annotations of this nature, forty-four refer to a divisional (děng 等) difference, seven refer to a -w- (kāihé 開合) difference, and twentythree refer to chóngyùn 重韻 ‘duplicated rhymes’ difference (MVaE vs. MVbE, rhymes with the same medial, same coda, and a similar main vowel, see Section 1.4.6 and Section 3.5.7). From these notations two conclusions can be made about the discrepancies among rhyme dictionaries: (1) All the notations are due to either the medial (M) or the main vowel (V) – there is no discrepancy due to the codas (E); (2) The majority of discrepancies are due to medial differences. Such differences do not require reconstructing different vowels for different rhymes. In comparison with the differences of coda in modern dialects, these dictionaries must be based on fairly similar phonological systems. When Lù Fǎyán and his associates discussed phonological differences over time and space,4 they included only the differences in these five dictionaries, 4

因論南北是非, 古今通塞 yīn lùn nánběi shìfēi, gǔjīn tōngsè ‘To discuss the correctness of North and South, and the historical distinctions of the ancient and the modern.’

126

Middle Chinese

which are more or less ideal standards. The phonological information of the Qièyùn is not much more than what is contained in these five dictionaries but still results in a more inclusive ideal standard. 3.3

Complementary and Near-Complementary Relationships

As demonstrated by the analysis of initials in rhyme tables in Section 1.4, complementary relationships between rhymes with the same coda are a large focus of reconstruction. For example, the zhēn 真 rhyme and the zhēn 臻 rhyme are in complementary distribution to each other. The rhymes zhēn 真 and zhēn 臻 never share the same initial; because of their similarities in coda, medial, and main vowel, they are likely the same final in different initial conditions (see Table 3.21). The same can be said for their corresponding rù rhymes, zhì 質 and zhì 櫛. Clearly the syllables of these two rhymes do not have any phonological contrast. In the Qièyùn system there are also similar cases, but with one or very few syllables in contrast. However, because these near-complementary rhymes often differ only in very rare characters, it is possible that the contrast is a leftover anomaly. There are three types of Rank-III rhymes: Type 1 Type 2

Rhymes with one series of grave initials (labials and gutturals). Rhymes with two series of grave initials, labials and gutturals (chóngniǔ 重 紐), and acute initials, alveolars, palatals. Rhymes with one series of grave and acute initials each.

Type 3

This type-distribution of Rank-III rhymes can be broken down and seen more clearly in Table 3.22. According to such a distribution, Type 2 rhymes look like the combination of a Type 1 and Type 3 rhyme. If we look at the relationship between some rhymes, assuming they have the same ending, it seems that some Type 1 and Type 3 rhymes could be combined to form Type 2 rhymes. The Rank-III Table 3.21 The zhēn 真 rhyme and the zhēn 臻 rhyme are in complementary distribution 真 臻

p + -

ph + -

b + -

m + -

t - -

th - -

d - -

n + -

l + -

ʈ + -

ʈh + -

ɖ + -

ts + -

tsh + -

dz + -

s + -

z + -

j + -

真 臻

tʂ - +

tʂh - -

dʐ - +

ʂ - +

ʐ - -

tʃ + -

tʃh + -

dʒ + -

ʃ + -

ʒ + -

ȵ + -

k + -

kh + -

ɡ + -

ŋ + -

h + -

ɦ + -

Ø + -

The Qièyùn 切韻

127

Table 3.22 The type-distribution of Rank-III rhymes

Number of grave initial series Number of acute initial series

Type 1

Type 2

Type 3

1 0

2 1

1 1

Note: The number of series in Type 2 rhymes is a sum of the number series in Type 1 and 3.

Table 3.23 The relationship between the gēng-III 庚三 and qīng 清 rhyme series 庚三 清

p + +

ph - +

b + +

m + +

t - -

th - -

d - -

n - -

l - +

ʈ - +

ʈh - +

ɖ - +

ts - +

tsh - +

dz - +

s - +

z - +

j - +

庚三 清

tʂ - -

tʂh - -

dʐ - -

ʂ - +

ʐ - -

tʃ - -

tʃh - -

dʒ - -

ʃ - -

ʒ - -

ȵ - -

k + +

kh + +

ɡ + +

ŋ + -

h + +

ɦ + +

Ø + +

Table 3.24 The relationship between the yōu 幽 and yóu 尤 rhyme series 幽 尤

p + +

ph - +

b + +

m + +

t - -

th - -

d - -

n - -

l + +

ʈ - +

ʈh - +

ɖ - +

ts + +

tsh - +

dz - +

s - +

z - +

j - +

幽 尤

tʂ - +

tʂh - +

dʐ - +

ʂ + +

ʐ - -

tʃ - +

tʃh - +

dʒ - -

ʃ - +

ʒ - +

ȵ - +

k + +

kh + +

ɡ + +

ŋ + +

h + +

ɦ - +

Ø + +

syllables of the rhyme series gēng 庚 (gēng-III 庚三) and qīng 清 are a good example of this (see Table 3.23). The Rank-III syllables of rhyme series gēngIII 庚三 and qīng 清 have syllables with the same initial to form minimal pairs. But because the gēng-III 庚三 series is Type 1 and the qīng 清 series is Type 3, they can be combined as a Type 2 rhyme, or a rhyme with the chóngniǔ syllables. The syllable distribution of the combined rhyme series gēng 庚 (gēng-III 庚三) and qīng 清 is very similar to the rhymes with the chóngniǔ syllables. The Rank-III syllables of the rhyme series yōu 幽 and yóu 尤 are another example, but they are only in near-complementary distribution with a few syllables in contrast (l-, ts-, ʂ-). As shown in Table 3.24, the syllables of the yōu 幽 and yóu 尤 rhyme series have syllables with the same initials to form

128

Middle Chinese

minimal pairs. But because one is a Type 3 rhyme and one is a Type 1 rhyme, they can be also combined as a Type 2 rhyme (ignoring the l-, ts-, ʂ- initials in the yōu 幽 rhyme). This relationship and other evidence has prompted some scholars to view them as having the same relationship as the rhymes with chóngniǔ syllables (Lǐ Xīnkuí 1984). More detailed analyses of the syllable distributions may provide a deeper and better understanding of the nature of the relationship between the rhymes. 3.4

Syllable Structure

It is evident in the Qièyùn 切韻 that syllables are self-contained phonological units and the natural starting point for phonological analyses. As shown in Table 3.25 and Figure 3.1, in the Qièyùn syllables are sorted first according to their tones into separate volumes, then according to their rhyming parts, (M) VE, into rhymes. Within each rhyme, the syllables are grouped together in homophonic groups based on the non-rhyming part of the syllable, I(M) (for a more in-depth explanation of syllable structure, see Section 1.2.4). The

Table 3.25 A basic outline of the syllable structure as analyzed from the Qièyùn Syllable Cluster Onset Rhyme

> > > >

Tone and cluster (segmental elements) Onset and rhyme (upper and lower spellers) Initial and (medial) (Medial), main vowel, and coda

Figure 3.1 Middle Chinese (and Modern Mandarin) syllable structure.

The Qièyùn 切韻

129

Table 3.26 Simplification of the Chinese syllable structure Tone Segmental

> >

T (I)(M)V(E)

segmental part of a syllable is analyzed into two components: rhyme and onset (the non-rhyme part). The medial of a syllable is an ambiguous element; it can be a part of the onset, rhyme, or both. Rhymes and onsets are the smallest analytical units, not phonetic segments (consonants and vowels). From the early seventh century on, the Chinese syllable structure had already become simplified so that it is recognizable as Chinese even today (see Table 3.26). In comparison with Old Chinese the syllable structure of the Qièyùn actually has quite a modern look (see Section 2.3.1 for the comparison between Middle and Old Chinese syllable structure). No consonant cluster at either end of a syllable, the same as modern Chinese dialects. While other syllable slots can be occupied by only one segment, the medial slot can have as many as three elements (-w-, -ɰ-, -j-, e.g. juǎn 卷 kwɰjɛn ‘volume’). 3.4.1

Suprasegmental Tones

As discussed in Section 1.3.4, there are four tones; their labels are píng 平 ‘level,’ shǎng 上 ‘rising,’ qù 去 ‘departing,’ and rù 入 ‘entering.’ Each syllable belongs to one of these four tonal categories. The reconstructed tonal values are level, rising, falling, and short, as their labels suggest. The phonetic values are based on Pān’s reconstruction (2000), which is a modification of the reconstruction by Zhèngzhāng (1987). The vocalic segments of the medial and coda slots are transcribed as approximants and the superscripts are not used (see Section 1.9 for greater discussion on the transcription system used in this book). As shown in Table 3.27, when Chinese characters are categorized by their Middle Chinese tone, one can see the parallelism across their reconstructed Middle Chinese pronunciations. In fact, only syllables with píng, shǎng, or qù tone are distinguished by tone alone. Syllables with rù tone are also distinguished from syllables of the other three tones by their stop consonant codas, as shown in Table 3.27. Tonal values have been of interest to modern researchers (Zhōu Fǎgāo 1956; Zhōu Zǔmó 1966b; Mei 1970; Shào 1982; Pān 1982; Yùchí 1986) who have used descriptions found in Buddhist sutras and documents to determine the values. These materials show a length contrast: the píng tone is long and the shǎng, qù, and rù tones are short, particularly the shǎng tone. The ChineseSanskrit transcriptions in the Yíqiè Jīng Yīnyì 一切經音義 ‘Sound and

130

Middle Chinese

Table 3.27 The parallelism of reconstructed pronunciations categorized by tone píng

shǎng





掂 tem diān ‘weigh in palm’ 藩 pwjɐn fān ‘fence’ 空 khuŋ kōng ‘empty’ 刀 tɑw dāo ‘knife’ 犁 lej lí ‘plough’

點 tem diǎn ‘dot’ 反 pwjɐn fǎn ‘reverse’ 孔 khuŋ kǒng ‘hole’ 島 tɑw dǎo ‘island’ 禮 lej lǐ ‘rite’

店 tem diàn ‘shop, inn’ 販 pwjɐn fàn ‘buy and sell’ 控 khuŋ kòng ‘control’ 到 tɑw dào ‘arrive’ 麗 lej lì ‘beautiful’

跌 tep diē ‘stumble’ 髮 pwjɐt fà ‘hair’ 哭 khuk kū ‘cry’ – –

Table 3.28 The Chinese-Sanskrit transcriptions in the Yíqiè Jīng Yīnyì by Buddhist monk Huìlín Sanskrit

Chinese

Reconstruction

Noted Tones

śāriputra purusāh _ _

奢利富多囉 補嚕沙 布路沙

ɕja li piw tɑ lɑ pwo lwo ʃɰa pwo lwo ʃɰa

奢 沙 沙

ɕja ʃɰa ʃɰa

píng píng píng

śarīra purusah _ _

舍梨子 補嚕灑 布路殺

ɕja li tsɨ pwo lwo ʃɰa pwo lwo ʃɰæt

舍 灑 殺

ɕja ʃɰa ʃɰæt

shǎng shǎng rù

Long (ā)

Short (a)

Meaning of All Buddhist Sutras’ by Buddhist Monk Huìlín 慧琳 (737–820) show that píng tone syllables are exclusively used to transcribe long vowels in Sanskrit, while non-píng (zè) tone syllables are exclusively used to transcribe short vowels (Pān 2000) (see Table 3.28). Some descriptions of tones also provide useful information about the tonal values, such as the one in the Xī Tán Zàng 悉曇藏 of 880 by the Japanese monk Annen 安然 (see Table 3.29). According to Annen’s description, there were five tones: the píng tone, which had two tonal contours, qīng 輕 ‘light’ and zhòng 重 ‘heavy,’ as well as the shǎng, qù, and rù tones, which were not further distinguished. 3.4.2

Segmentals

The segmental part of a syllable has four structural slots, initial (I), medial (M), main vowel (V), and coda (E), which can be filled with segmental elements

The Qièyùn 切韻

131

Table 3.29 Some descriptions of tones in the Xī Tán Zàng by the Japanese monk Annen 平聲直低,有輕有重 上聲直昂,有輕無重 去聲稍引,無重無輕 入聲徑止,無內無外a

‘píng tone is straight and low, having light and heavy distinction’ ‘shǎng tone is straight rising, having only light and no heavy’ ‘qù tone is briefly elongated, having no light or heavy distinction’ ‘rù tone has an abrupt coda, having no in or out distinction’

The terms nèi 內 and wài 外 are likely used in the same way as qīng 重 and zhòng 輕, rather than referring to any properties of wàizhuǎn 外轉 and nèizhuǎn 內轉 (Yùchí 1986).

a

(consonants and vowels). Of the four elements only the nuclear vowel is obligatory; the others are optional. A syllable can be summarized as (I)(M)V (E), with parentheses indicating optional segments. The syllable structure is not linear, which means that there is a hierarchical structure involved. A syllable is divided either into onset (IM) or rhyme (VE); or initial (I) and final or yùnmǔ 韻母, which is the combination of medial and rime (MVE). Thus different rhymes in the rhyme dictionaries may or may not be equivalent to the technical term “rime” in modern phonology. Some rhymes can contain two or more finals. For example, the syllables of the dōng 東 rhyme have the same rhyme (the same main vowel -u and coda -ŋ) but they may or may not have a medial -j-: gōng 公 kuŋ and gōng 弓 kjuŋ; lóng 籠 luŋ and lóng 隆 ljuŋ are all from the dōng 東 rhyme. The finals of the same rhymes only differ in the position of medial (e.g. the three finals of the 麻 má rhyme are -ɰa, -wɰa, and -ja; and the four finals of the gēng 庚 rhyme are -ɰaŋ, wɰaŋ, -ɰjaŋ, and -wɰjaŋ). 3.4.3

Initials (I)

Unlike rhymes, the initials are not clearly indicated in the Qièyùn. In each rhyme the homophonic groups are listed in a random way. No clear order based on the initials can be found (see Table 3.30). Different types and numbers of homophonic groups are involved within each rhyme. There is no easy way to determine the initials. The Qièyùn’s organization of initials in different rhymes is ultimately sporadic and no useful data has been extrapolated from it. The reconstruction of initials is based on information from fǎnqiè 反切 spellings in the Qièyùn and the thirty-six initial characters as seen in the later rhyme tables. Premodern study of the initials of the Qièyùn can be represented by Chén Lǐ’s 陳澧 (1810–1882) Qièyùn Kǎo 切韻考. He systematically used fǎnqiè spellers to establish the categories of initials and finals of the Guǎngyùn. Chén Lǐ’s method used in his initial study has been revised and improved by a number of scholars and applied to the study of the Qièyùn.

132

Middle Chinese

Table 3.30 The sporadic organization of initials in different rhymes Rhyme

Homophonic groups

東一 東三 冬 鍾

東 t中 ʈj冬 t鍾 tɕ-

同 d蟲 ɖj彤 d龍 l-

空 kh終 tɕj賨 dz舂 ɕ-

公 k忡 ʈhj農 n松 z-

蒙 m崇 tʂhj恭 ktɕh-

籠 l嵩 sj蜙 s容 j-

洪 ɦ戎 ȵj樅 tsh封 p-

叢 dz弓 kj攻 k胷 h-

翁 Ø融 j䃔 ɦ顒 ŋ-

... ... ... ...

Pān’s (2000) reconstruction of thirty-seven initials, like other proposals, is based on Karlgren’s (1915–1926) and the modifications and confirmations on a number of initials proposed in various subsequent studies (e.g. Li FangKuei (1971) and Shào Róngfēn (1982) on the chuán 船 and shàn 襌 initials; Lǐ Róng (1956) and Shào (1982) on the niáng 娘 initial; Edwin G. Pulleyblank (1962–1963) on the rì 日 initial; and Zhōu Zǔmó (1966c) on the yún 云 initial; Luó (1931), Pulleyblank (1962–1963) and Mài Yùn (1991) on the zhī 知 and zhuāng 莊 initial series; and Lǐ Róng (1956) on the sì 俟 initial). Scholars in general agree on the reconstruction in terms of places of articulation and manners of articulation. Pān’s (2000) reconstruction of thirty-seven initials is listed along with the traditional initial characters. Pān’s reconstruction (2000) of thirty-seven initials is given in Table 3.31. Note the differences from the traditional thirty-six initials. According to some scholars (Lù 1947c; Shào 1982), the phonetic values of initials chuán 船 and shàn 襌 should be reversed. Chuán is a voiced fricative ʑ, and shàn is a voiced affricate dʑ. 3.4.4

Medials (M)

Unlike initial consonants, the medial (M) is a rather complex structural slot. Recent proposals suggest three medials: (1) a -w- medial, required because the contrast between kāikǒu and hékǒu syllables is related to either the rounded feature of the nuclear vowel or the presence of medial -w-; (2) a -j- medial, required because it is a basic feature of Rank-III rhymes; and (3) a -ɰ- medial, required because it is the medial for Rank-II syllables and the IIIb type of chóngniǔ syllables. Medial -ɰ- is a homorganic approximant of the high back unrounded vowel ɯ, derived from -r- of Old Chinese. Thus the medial slot (M) can have up to three medials, e.g. gàn 干 kɑn, guān 冠 kwɑn, guàn 慣 kwɰan, juǎn 卷 kwɰjᴇn. Phonetic values of the medials remain notational; whether the medial clusters ɰj-, wɰ-, wj-, and wɰj- were pronounced sequentially or not is still undetermined. For the discussion on the reconstruction process for the medials, see Section 3.5.4.

The Qièyùn 切韻

133

Table 3.31 Pān’s reconstruction (2000) of thirty-seven initials Bilabials

幫p bāng

滂 ph pāng

並 b bìng

明m míng

Alveolars

端t duān

透 th tòu

定 d dìng

泥n ní

Retroflex stops

知ʈ zhī

徹 ʈh chè

澄 ɖ chéng

娘ɳ niáng

Alveolar sibilants

精 ts jīng

清 tsh qīng

從 dz cóng

心s xīn

邪z xié

Retroflex sibilants

莊 tʂ zhuāng

初 tʂh chū

崇 dʐ chóng

生ʂ shēng

俟 ʐa sì

Palatals

章 tɕ zhāng

昌 tɕh chāng

船 dʑ chuán

審ɕ shěn

禪ʑ shàn

Velar stops

見k jiàn

溪 kh xī

群 ɡ qún

疑ŋ yí

Laryngeals

影 ʔ*c yǐng

曉h xiǎo

匣 ɦ xiá

云 ɦjd yún

來l lái

日 ȵb rì

以j yǐ

The initial sì 俟 ʐ is based on Lǐ Róng’s reconstruction (1956) with a few rare characters of two homophonic groups, sì 俟 and chí 漦. b Although reconstructed as an alveolo-palatal nasal ȵ instead of retroflex approximant ɻ, the rì 日 initial is clearly a separate initial from the close palatal nasal niáng 娘 ɳ. 10 upper spellers, namely rú 如, rǔ 汝, ér 而, ěr 耳, rén 人, rì 日, rú 儒, ér 兒, ěr 爾, and réng 仍, are used to spell the syllables with rì 日 initial of the sixty-one homophonic groups (Lǐ Róng 1956). c In this book Ø is used for initial 影, because the obstruction at the onset is not as strong as plosives. d The initial of yún 云 ɦj is actually the same initial as xiá 匣 ɦ but with a -j- medial. a

3.4.5

Main Vowels (V)

Since Karlgren, the reconstruction of main vowels (V) has been changed a number of times. Efforts have been made to reduce the number of main vowels and to make the vowel system more natural in relation to linguistic universals. Excluding the tones there are fifty-eight different rhymes. Since the working principle is that the phonological contrast for these rhymes must either be on the main vowel (V) or the coda (E) and the codas are relatively certain, the focus of the reconstruction must be on the main vowel, meaning that a dozen or more vowels would be required in order to distinguish all the rhymes. Pan’s vowel system, seen in Table 3.32, is mainly based on the assumption that the contrasts of rhymes are phonemic. Beside the tone, the phonemic

134

Middle Chinese

Table 3.32 An example of the twelve-vowel system needed to maintain non-medial based rhyme differentiation i e ɛ æ a

ɨ ə ɐ

u o ɔ ɑ

Source: Based on Pān 2000.

contrast must be of the main vowel, the coda, or both. However, as discussed in Section 3.2, other scholars use the phonemic analyses and show that some main vowels are subphonemic and can be reduced. In the twelve-vowel system shown in Table 3.32, at least a four-degree contrast in height for both front vowels and back vowels is required. However, if some of the contrasts are considered subphonemic variations after different medials, as some have proposed (Mài 1995; Huáng Xiàoshān 2002a), the number of main vowels required could be reduced. A number of scholars (Hashimoto 1978–1979; Yú Nǎiyǒng 1993; Mài 1995; Huáng Xiàoshān 2002a) have proposed a sevenvowel system to account for the main vowel distinction (e.g. Huang’s sevenvowel proposal with i, u, ə, e, o, a, and ɒ). Both approaches, however, rely on the structural information of rhyme tables that appeared much later. Neither addresses the fundamental issue of whether the structural information of the rhyme tables is appropriate for the distinctions of the Qièyùn system. 3.4.6

Codas (E)

The coda (E) is either a stop consonant or an approximant. The following codas are commonly reconstructed: -m, -n, -ŋ, -p, -t, -k, -j, and -w. This coda system is quite universally accepted, except by Hashimoto Mantaro (1970), who proposed a palatal nasal coda -ɲ with the reference of Sino-Vietnamese. This proposal is not generally accepted because this coda -ɲ is (a) a feature of Vietnamese phonology, and (b) subphonemic in Sino-Vietnamese because it only appears in the rhymes that have a low-mid front vowel, such as ɛ or æ (Ruǎn 2007). 3.5

Phonological Characteristics of Middle Chinese

3.5.1

The Tripartite Distinction of Stop and Affricate Initials

The stops and affricates have three phonation types: voiceless unaspirated, voiceless aspirated, and voiced. In Table 3.33, for each place and manner of

The Qièyùn 切韻

135

Table 3.33 Three phonation types (columns) of stops and affricates for each place of articulation (rows) Labial stop píng

波 pwɑ bō ‘wave’

坡 phwɑ pō ‘slope’

婆 bwɑ pó ‘old woman’

Alveolar stop píng

低 tei dī ‘low’

梯 thei tī ‘ladder’

題 dei tí ‘title’

Retroflex stop rù

竹 ʈjuk zhú ‘bamboo’

畜 ʈhjuk chù ‘domestic animal’

逐 ɖjuk zhú ‘chase’

Velar stop píng

疆 kjɐŋ jiāng ‘border’

羌 khjɐŋ qiāng ‘Qiang nationality’

強 ɡjɐŋ qiáng ‘strong’

Alveolar affricate píng

津 tsin jīn ‘ferry’

親 tshin qīn ‘intimate’

秦 dzin qín ‘Qin dynasty’

Retroflex affricate qù

詐 tʂɰæ zhà ‘defraud’

岔 tʂhɰæ chà ‘branch off’

乍 dʐɰæ zhà ‘suddenly’

Palatal affricate píng

專 tɕwjɛn zhuān ‘specialized’

川 tɕhjɛn chuān ‘river’

船 dʑwjɛn chuán ‘boat’

articulation (rows), there are three phonation types (columns). Their Middle Chinese tonal categories are also indicated. 3.5.2

The Labials

There is only one series of labial initials in the Qièyùn 切韻 system. The labial series split (bāng 幫 series5 vs. fēi 非 series6) had not yet occurred at the time of the Qièyùn. The fēi 非 series in Table 3.34 have the same initials as their bāng 幫 series counterparts. Note, however, the -j- medial, which will eventually be the condition for the split later. The split of labials into bilabials and labiodentals is a later development. 3.5.3

The Retroflex Stops

There is a series of retroflex stops, which were initially reconstructed by Karlgren as palatals. Luó Chángpéi (1933, 2004) later used Chinese

5 6

bāng 幫 p-, pāng 滂 ph-, bìng 並 b-, míng 明 mfēi 非 pf-, fū 敷 pfh-, fèng 奉 bv-, wēi 微 ɱ-

136

Middle Chinese

Table 3.34 The labial series before the split 幫 bāng series 波 pwɑ bō ‘wave’

坡 phwɑ pō ‘slope’

婆 bwɑ pó ‘old woman’

磨 mwɑ mó ‘grind’

非 fēi series 分 pjun fēn ‘split’

芬 phjun fēn ‘fragrance’

焚 bjun fén ‘burn’

文 mjun wén ‘civil, writing’

Table 3.35 Sanskrit letters for retroflex stops and nasals and corresponding zhī 知 initial series characters Sanskrit

Pronunciation

Examples

ट ठ ड ढ ण

ta [ʈa] _ tha [ʈha] _ da [ɖa] _ [ɖha] dha _ na [ɳa] _

吒 姹 茶 荼 拏

ʈɰa ʈhɰa ɖɰa ɖɰa ɳɰa

咤 侘 荼 茶 儜

ʈɰa ʈhɰa ɖɰa ɖɰa ɳɰæŋ

transcriptions in Sanskrit material to show that the place of articulation of the zhī 知 initial series7 is retroflex instead of palatal. In the Yuánmíng Zìlún 圓明字輪 ‘The round and bright wheel of letters,’ a list of Sanskrit letters arranged in a circular way commonly found in various Buddhist documents, the corresponding Chinese characters used for the Sanskrit letters of retroflex consonants, t, th, d, dh, and n, are basically the _ 知 _ initial _ _ series _(ʈ-, ʈh-, ɖ-, ɳ-), not characters with the Middle Chinese zhī the characters with the Middle Chinese zhào 照 initial series (which consists of the Middle Chinese zhuāng 莊 series (tʂ-, tʂh-, dʐ, ʂ-, ʐ-) and zhāng 章 series (tɕ-, tɕh-, dʑ-, ɕ-, dʑ-)). These can be seen in Table 3.35 and Table 3.36. As shown in Table 3.35, Sanskrit devanāgarī letters used to write retroflex obstruents and nasals were used to transcribe Chinese characters into Sanskrit. The examples include their Middle Chinese reconstructed pronunciations. Sino-Sanskrit transliterations for proper names are given in Table 3.36. The retroflex obstruent and nasal syllables in Sanskrit and the corresponding Chinese characters are provided in bold, along with the Middle Chinese phonetic value of the Chinese character in question.

7

zhī 知 ʈ-, chè 徹 ʈh-, chéng 澄 ɖ-, niáng 娘 ɳ-.

The Qièyùn 切韻

137

Table 3.36 The retroflex obstruent and nasal syllables in Sanskrit and the corresponding Chinese characters Sanskrit

Chinese

Atali Ko_ ti Ari_ttha _ Kaus_ thila _ Candala __ Kundika _ Āsadha _ _ Virūdhaka Anu _ _ Bhiksuni _ _

阿吒釐 俱致 阿栗抽 拘絺緹 旃荼羅 軍持 阿沙荼 毗盧宅迦 阿拿 比丘尼

吒 致 抽 絺 荼 持 荼 宅 拿 尼

ʈɰa ʈɰi ʈhiw ʈhɰi ɖɰa ɖɨ ɖɰa ɖɰak ɳɰa ɳɰi

Table 3.37 The coexistence of retroflex and palatal affricates, and retroflex stops Retroflex stop

Retroflex affricate

Palatal affricate

珍 ʈin zhēn ‘treasure’ 張 ʈjɐŋ zhāng ‘stretch, open’

榛 tʂin zhēn ‘hazelnut’ 莊 tʂjɐŋ zhuāng ‘village’

真 tɕin zhēn ‘real’ 章 tɕjɐŋ zhāng ‘chapter’

The Chinese characters used to transliterate the syllables with retroflex initials are overwhelmingly the ones with the Middle Chinese zhī 知 initial series, suggesting the retroflex realization of the zhī 知 initial series. As shown in Table 3.37, both retroflex and palatal affricates are present in Middle Chinese, along with retroflex stops. Note that retroflex sounds can coexist with a palatal medial, which is not possible in Modern Mandarin. The grave and acute initials change differently. The -r- is integrated into the acute initials, changing them to retroflexes, but weakened after grave initials -r> -ɰ-. Table 3.38 shows that when preceding the medial -r-, acute initials (Old Chinese alveolars) become retroflex, while grave initials (Old Chinese dorsals and labials) do not change. The medial -r- weakened to -ɰ- while -ɰ- is lost in Rank-III syllables with acute initials. 3.5.4

The Four Ranks (Divisions)

In the rhyme tables each syllable is placed in one of the four divisions, basically according to its final. The syllables or the rhymes in the same division

138

Middle Chinese

Table 3.38 The changes of grave and acute initials, and the -r- of Old Chinese

*tsr- > tʂ*tr-

> ʈ-

*pr- > pɰ*kr- > kɰ-

Rank-II

Rank-III

窗 tʂhɰɔŋ chuāng ‘window’ 樁 ʈɰɔŋ zhuāng ‘stake’ 邦 pɰɔŋ pāng ‘state’ 江 kɰɔŋ jiāng ‘river’

莊 tʂjɐŋ zhuāng ‘village’ 張 ʈjɐŋ zhāng ‘stretch, open’ 臕 pɰjɛw biāo ‘fat’ 驕 kɰjɛw jiāo ‘arrogant’

Table 3.39 The four ranks of the xiào 效 rhyme group, illustrating Karlgren’s reconstruction of the four ranks of the Guǎngyùn Rank Example

I 豪 -ɑu

II 肴 -au

III 宵 -jiɛu

IV 蕭 -ieu

Source: Karlgren 1915–1926.

Table 3.40 Karlgren’s reconstruction in terms of medials and main vowels Rank Medial Main vowel

I -Øɑ

II -Øa

III -jiɛ

IV -ie

share certain phonological features, and the syllables and the rhymes in different divisions must be distinct from each other (see Table 3.39). The research on divisional differences and their corresponding ranking differences in the Guǎngyùn (for the terminologies “division” and “rank,” see Section 1.4.4) has made significant progress since Karlgren’s reconstruction. Table 3.40 summarizes Karlgren’s basic ideas concerning the four ranks: (1) Rank-I and -II syllables have no palatal medial; (2) the main vowel of Rank-II syllables is relatively more to the front and higher; (3) Rank-III and -IV syllables have a palatal medial; (4) the medial of Rank-IV syllables has a vowel quality but the medial of Rank-III syllables has a consonant quality; and (5) Rank-IV syllables have a higher main vowel than Rank-III. Karlgren did not distinguish Rank-IIIa and -IIIb (see Section 1.4.6). Beside Rank-III, the basic features of the other three ranks have been revised significantly. As early as in 1981, Zhèngzhāng proposed that the Rank-II syllables had medial -ɣ-, and that their historical development was

The Qièyùn 切韻

139

Cr- > Cɣ- > Cɯ- > Ci- (Xǔ & Pān 1994), a proposal which has gained support with strong evidence. It has also been proposed that Rank-III has two types of syllables, IIIa and IIIb (the so-called chóngniǔ contrast), which were not recognized in Karlgren’s reconstruction. Based on Sanskrit transcription, Yú Mǐn (1984a) proposed that Rank-IIIb and IIIa had -y- (-j-) and -r- medials, respectively. Shī Xiàngdōng (1983) showed that such distinction was apparent in the early Tang transcriptions. Zhèngzhāng (1987) reconstructed Rank-IIIb syllables with medial -ɣi- and IIIa with -i-. Like Rank-I syllables, Rank-IV syllables did not actually have a medial, but unlike Rank-I they did have a front vowel -e (Lǐ Róng 1956). These proposals have been widely accepted, although with some difference in the reconstructed phonetic values. Of the four ranks, Rank-I and Rank-IV syllables lack medials, Rank-II syllables have a medial -ɰ-, and Rank-III syllables have a -j- medial. The difference between Rank-I and Rank-IV syllables is in the main vowel. Last but not least, Type B chóngniǔ syllables of Rank-III (IIIb) require medials -ɰ- and -j-. As shown in Table 3.41, besides the contrast between Rank-I and -IV syllables, there is a clear redundancy in criteria for distinguishing the syllables of the four ranks, with differences found in both the presence or absence of medial -ɰ- and the vowel quality of the main vowel. Distinguishing by the main vowel alone results in the same outcome, with the contrast between Rank-I and -IV the only exception. Based on the relationship between the four ranks as related to the medials -ɰ- and -j- (see Table 3.42), phonemically, the main vowels can be reconstructed more simply (see Table 3.43). Four main vowels reduce to two; if such is the case, the number

Table 3.41 Redundancy in criteria for distinguishing the syllables of the four ranks Rank Medial Main vowel Example

I -Øɑ 褒 pɑw 高 kɑw

II -ɰa 包 pɰaw 交 kɰaw

IIIb -ɰjɛ 鑣 pɰjɛw 驕 kɰjɛw

Table 3.42 The relationship between the four ranks as related to the medials -ɰ- and -j-

-Ø-ɰ-

-ØI, IV II

-jIIIa IIIb

IIIa -jɛ 飇 pjɛw –

IV -Øe – 澆 kew

140

Middle Chinese

Table 3.43 Simpler main vowels reconstructed phonemically Rank Phonetic Phonemic

I ɑw aw

II ɰaw ɰaw

IIIb ɰjɛw ɰjɛw

IIIa jɛw jɛw

IV ew ɛw

Table 3.44 Calculations of the prevalence of the rank of upper spellers for Rank-III characters in the Qièyùn Upper speller Occurances with Rank-III characters Percentage of total

I 30 1.6%

II 9 0.5%

III 1825 97.5%

IV 7 0.4%

Total 1871 100%

Source: Based on Huáng (2002a).

of main vowels required for the entire system can be significantly reduced and the main vowel system would appear much more natural. 3.5.4.1 The Status of Rank-III Syllables Whether there is a palatal medial -j- in the Qièyùn system is a key phonological feature that needs to be proven. Scholars point out that foreign transcriptions do not show any clear evidence to support the existence of the palatal medial in Rank-III syllables; however, this could be due to the lack of a medial in the syllable structure of the borrowing languages. In the Qièyùn there is a very strong tendency for Rank-III syllables to be spelled with upper spellers that are Rank-III syllables in fǎnqiè (see Table 3.44).8 This tendency indicates that the distinction is in the onset of the syllables. In other words, this evidence suggests that Rank-III syllables must have a palatal medial, with fǎnqiè form CjVE > Cj + jVE (see Section 3.1.1 for more information). Another observation can be made on the use of Rank-II syllables (with medial ɰ) as an upper speller. This phenomenon indicates an additional medial ɰ in some Rank-III syllables (IIIb), with fǎnqiè form C(w)ɰjVE > C(w)ɰj + jVE, (e.g. Rank-IIIb syllable bié 別 bɰjɛt is spelled as pí 皮 bɰjɛ + liè 列 ljɛt;

8

For example, Rank-III syllable jū 居 is spelled as jiǔyú 九魚 in fǎnqiè. Both jiǔ 九 and yú 魚 are Rank-III syllables. This phenomenon indicates that the upper speller (initial + medial) and the lower speller (medial + rime) must share the same medial. This medial is commonly reconstructed as a palatal -j- (Karlgren 1915–1926; Chao 1941). So, the pronunciation of jū 居 kiɔ is spelled by jiǔ 九 kju and yú 魚 ŋjɔ, kj- + -jɔ > kjɔ.

The Qièyùn 切韻

141

bɰj- + -jɛt > bɰjɛt). See Section 1.3.2 for a summary of the fǎnqiè system and Section 3.1.1 for a discussion and examples of the fǎnqiè of Rank-III syllables. 3.5.5

The Kāihé 開合 Contrast

There are two different types of kāikǒu/hékǒu contrast. One is the presence or absence of the round feature of the main vowel, and the other is the presence or absence of medial -w-. Thus, the former is a contrast between different rhymes and the latter is a contrast within the rhyme. For example, dēng 登 təŋ is kāikǒu and dōng 東 tuŋ is hékǒu, belonging to the zēng 曾 and dōng 東 rhymes, respectively; jì 計 kei is kāikǒu and guì 桂 kwei is hékǒu, but both belong to the same jì 霽 rhyme. It is worth mentioning that kāikǒu and hékǒu syllables are not contrastive after labial initials. If a rhyme has kāikǒu and hékǒu syllables, the ones with labial initials only appear as either kāikǒu or hékǒu, never both. A kāikǒu rhyme has the syllables with labial initials, and its corresponding hékǒu rhyme will not have the syllables with labial initials, and vice versa. The examples in Table 3.45 are the rhymes with -n coda: rhyme hán 寒 is a kāikǒu rhyme and rhyme huán 桓 is a corresponding hékǒu rhyme of hán. Rhymes shān 山, shān 刪, xiān 仙, yuán 元, and xiān 先 have both kāikǒu and hékǒu syllables. Table 3.45 The complementary distribution of syllables with labial initials in rhymes with -n coda

Rank Labial kāikǒu Labial hékǒu Non-labial kāikǒu Non-labial hékǒu

寒 hán

桓 huán

山 shān

刪 shān

仙 xiān

元 yuán

先 xiān

I – – + –

I – + – +

II + – + +

II + – + +

III + – + +

III – + + +

IV + – + +

Note: “+” means presence and “–” means absence of the syllables within the given initial, kāihé, and rhyme.

3.5.6

The Contrast of Chóngniǔ 重紐 Syllables

The term ‘chóngniǔ syllable’ refers to syllables with labial or guttural initials from Rank-III rhymes. Chóngniǔ pairs belong to the same Rank-III rhyme (VE), have the same initials (I), and are the same in kāihé (either both kāikǒu or both hékǒu), yet contrast with one another. Such a seemingly impossible chóngniǔ contrast has been a focal point of the study of Middle Chinese

142

Middle Chinese

phonology for decades. The chóngniǔ syllables are found in eight rhymes, namely zhī 支, zhī 脂, zhēn 真, xiān 仙, xiāo 宵, qīn 侵, yán 鹽, and jì 祭. Their phonological contrast has been a challenge for many decades. Recently researchers have gradually come to a consensus that the contrast of the chóngniǔ syllables is due to an additional element in the medial position. Of the chóngniǔ syllables, one is referred to as IIIa and the other as IIIb. Evidence from Sino-Sanskrit transcription shows that IIIb has an extra medial, which is -r- historically (see Table 3.46). In the Qièyùn system it is probably in a weakened form as a voiced velar fricative -ɣ-, or a semivowel -ɰ-, with the same tongue position as the high back unrounded vowel ɯ. Examples of chóngniǔ contrasts across all eight rhymes that contain them are given in Table 3.47. All IIIa syllables have a -j- medial or -i- main vowel, with IIIb syllables being the same but with an additional -ɰ- medial. The chóngniǔ pairs in Table 3.47 show no distinction in Modern Mandarin. Some scholars believe that the contrasts between yóu 尤 and yōu 幽, and gēng-III 庚三 and qīng 清 are actually chóngniǔ contrast as well. Based on Table 3.46 Sino-Sanskrit proof of the existence of the medial -r- in IIIb syllables

IIIa

Non-rhotic

IIIb

Rhotic

Sanskrit

Chinese

Reconstruction

Pinyin

khya mit r(i) grān

企 蜜 乙 乾

khjɛ mit ʔɰit ɡɰjɛn

qǐ mì yǐ qián

Source: Shī 1983; Liú 1984.

Table 3.47 Examples of chóngniǔ contrasts across all eight rhymes that contain them Rhyme IIIa IIIb

Rhyme IIIa IIIb

支 zhī 岐 ɡjɛ qí ‘a mountain name’ 奇 ɡɰje qí ‘strange’

脂 zhī 比 pi bǐ ‘compare’ 鄙 pɰi bǐ ‘low, mean’

真 zhēn 頻 bin pín ‘frequent’ 貧 bɰin pín ‘poor’

仙 xiān 便 bjɛn biàn ‘convenient’ 汴 bɰjɛn biàn ‘a river name’

宵 xiāo 邀 Øjɛw yāo ‘invite’ 妖 Øɰjɛw yāo ‘demon’

侵 qīn 愔 Øim yīn ‘serene’ 音 Øɰim yīn ‘sound’

鹽 yán 魘 Øjɛm yǎn ‘nightmare’ 掩 Øɰjɛm yǎn ‘cover’

祭 jì 藝 ŋjɛj yì ‘talent’ 㓷 ŋɰjɛj yì ‘cut off nose’

The Qièyùn 切韻

143

Table 3.48 The traditional chóngyùn rhymes Rank-I Rank-II Rank-III

dōng-I 東一 vs. dōng 冬 tán 覃 vs. tán 談 jiē 皆 vs. jiā 佳 vs. guài 夬 gēng 庚二 vs. gēng 耕 dōng-III 東三 vs. zhōng 鍾 yú 魚 vs. yú 虞 zhēn 真 vs. xīn 欣 xiān 仙 vs. yuán 元 yōu 幽 vs. 尤 yóu

hāi 咍 vs. huī 灰 vs. tài 泰 shān 山 vs. shān 刪 xián 咸 vs. xián 銜 zhī 支 vs. zhī 脂 vs. zhī 之 vs. wēi 微 jì 祭 vs. fèi 廢 zhūn 諄 vs. wén 文 Gēng-III 庚三 vs. qīng 清 yán 鹽 vs. yán 嚴

their distribution patterns, some (for example, Lǐ Xīnkuí 1984) include even more Rank-III rhymes as chóngniǔ contrasts. 3.5.7

The Contrast of Chóngyùn 重韻 Rhymes

Chóngyùn ‘duplicated rhyme’ is a traditional term that refers to a pair of rhymes that have the same classification in the rhyme tables. As shown in Table 3.48, they are listed in the same rhyme group (shè 攝) and the same division (děng 等). Since the phonological classification is the same for these rhymes according to the rhyme tables, the phonological contrasts of these ‘duplicated rhymes’ are quite difficult to reconstruct. However, it should be noted that the classification system of the rhyme tables may not reflect the phonological features of the Qièyùn system, because the rhyme tables were published several centuries after the Qièyùn. Because of the time gap, the phonological features of the chóngyùn rhymes in the Qièyùn have changed, and some pairs had already merged at the time the rhyme tables were composed. In addition to this time difference, the composite nature of the Qièyùn itself makes the reconstruction of these chóngyùn rhymes a major problem in the study of Middle Chinese phonology. For example, both the dōng-I 東一 (the Rank-I part of the dōng 東 rhyme) and dōng 冬 rhymes belong to Rank-I (no medial), and the tōng 通 shè (a rounded main vowel and a velar nasal coda -ŋ). The only possible difference between these rhymes is the main vowel. The other chóngyùn rhymes have the same situation and the difference must be the main vowel. Some samples of the reconstructed phonetic values of the chóngyùn pairs are shown in Table 3.49. It is quite clear that the chóngyùn pairs have similar phonetic values in different proposals of the reconstruction. 3.5.8

Final Types

Syllables with nasal and stop codas have a very close correlation to syllable types, meaning that the permissible combinations of the phonemes in these two

144

Middle Chinese

Table 3.49 Some samples of the reconstructed phonetic values of the chóngyùn pairs BK

LR

WL

SRF

EP

ZZSF

PWY

東一 冬

dōng-I dōng

unɡ uonɡ

uŋ oŋ

uŋ uoŋ

uŋ oŋ

uŋ uoŋ

uŋ uoŋ

uŋ uoŋ

覃 談

tán tán

ậm âm

ɐ^m âm

ɒm ɑm

ɒm ɑm

əm am

ʌm ɑm

əm ɑm

魚 虞

yú yú

i̯ wo i̯ u

iå io

ĭo ĭu

iɔ io

i̯ ou i̯ o

ɨʌ ɨo

jɔ jʊ

庚三 清

gēng-III qīng

i̯ ɒnɡ i̯ änɡ

iɐŋ iäŋ

ĭɐŋ ĭɛŋ

iaŋ iæŋ

i̯ aŋ i̯ eŋ

ɣiæŋ iᴇŋ

ɯiaŋ iɛŋ

幽 尤

yōu yóu

iĕu i̯ ə̆u

iĕu iu

iəu ĭəu

ieu iəu

jiw i̯ u

iɪu ɨu

ɨu iu

Source: Based on Pān 2000: 83–88. Note: BK = Bernhard Karlgren, LR = Lǐ Róng, WL = Wáng Lì, SRF = Shào Róngfēn, EP = Edwin Pulleyblank, ZZSF = Zhèngzhāng Shàngfāng, PWY = Pān Wùyún.

Table 3.50 A series of minimal pairs, exemplifying a difference in the manner of articulation of the coda 耽 尖 干 官 江 經

təm tsjɛm kɑn kwɑn kɰɔŋ keŋ

答 接 割 括 角 擊

təp tsjɛp kɑt kwɑt kɰɔk kek

監 掂 肩 幫 弓

kɰam tem ken pɑŋ kjuŋ

甲 跌 結 博 菊

kɰap tep ket pɑk kjuk

types are parallel: a syllable with a nasal coda usually has its counterpart with a stop coda. The pairs in Table 3.50 are exactly the same except for the codas, one being a nasal and the other an oral stop. As discussed in greater depth in Section 1.4.10, there are three basic types of rhymes, termed yīnshēng yùn 陰聲韻, yángshēng yùn 陽聲韻, and rùshēng yùn 入聲韻, with the distinction found in the coda. Yīnshēng yùn rhymes have no coda or a semivowel coda (-Ø, -G), yángshēng yùn rhymes have a nasal consonant coda (-N), and rùshēng yùn rhymes have a stop consonant coda (-C). In the Qièyùn system there is a oneto-one parallel relationship between syllables with a nasal coda and syllables with a stop coda, as represented by a shūcù 舒促 contrast (see Table 3.51).

The Qièyùn 切韻

145

Table 3.51 The parallel relationship between syllables with a nasal coda and syllables with a stop coda píng

shǎng





侵 im/ɰim 寒 ɑn 唐 ɑŋ 豪 ɑw 齊 ej 模 wo

寑 im/ɰim 旱 ɑn 蕩 ɑŋ 皓 ɑw 薺 ej 姥 wo

沁 im/ɰim 翰 ɑn 宕 ɑŋ 号 ɑw 霽 ej 暮 wo

緝 ip/ɰip 曷 ɑt 鐸 ɑk

Table 3.52 Rhymes jì 祭, tài 泰, guài 夬, and fèi 廢 with no equivalent píng and shǎng tone rhymes píng

shǎng





侵 im/ɰim 寒 ɑn 唐 ɑŋ 豪 ɑw 齊 ej 模 wo

寑 im/ɰim 旱 ɑn 蕩 ɑŋ 皓 ɑw 薺 ej 姥 wo

沁 im/ɰim 翰 ɑn 宕 ɑŋ 号 ɑw 霽 ej 暮 wo 祭 jɛj/ɰjɛj 泰 ɑj 夬 ɰæj 廢 jɐj

緝 ip/ɰip 曷 ɑt 鐸 ɑk

As discussed earlier (see Section 3.2), rhymes in the Qièyùn could include the medial, which therefore makes Qièyùn rhymes different from the common understanding of the term “rhyme” or “rime,” which do not include medials in their definition. Such a difference calls for a reexamination of the Qièyùn reconstruction, an undertaking that is beyond the scope of this book. 3.5.8.1 Rhymes jì 祭, tài 泰, guài 夬, and fèi 廢 In the Qièyùn system there is a one-to-one parallel relationship between syllables with a nasal coda in píng, shǎng, and qù tones and the syllables with a stop coda in rù tone. Open syllables occur with píng, shǎng, and qù tones and have no counterparts in rù tone. However, as discussed briefly in Section 1.3.5 on yùnxì 韻系 ‘rhyme series,’ there are four rhymes that are peculiar as they occur only in the qù tone and all have a -j coda. As shown in Table 3.52, not only do the rhymes jì 祭, tài 泰, guài 夬, and fèi 廢 not have rù tone equivalents, but they also do not share any equivalency with any píng and shǎng tone rhymes.

146

Middle Chinese

This phenomenon is explained by historical developments in Old Chinese, as detailed in Section 2.3.3.5.4, and has to do with the relationship between post-coda elements of the qù and rù tone in Old Chinese. 3.6

Modern Evidence of the Phonological Contrasts

As a phonological standard, the system represented by the Qièyùn 切韻 is irreplaceably important. It serves as the main reference for the studies of the historical periods before and after it. It also serves as the framework for the study of modern dialects. With the Min dialects as an exception, all the phonological contrasts of modern dialects can be found in the Qièyùn. In other words, the phonological contrasts in most of the modern dialects are the subsets of the contrasts in the Qièyùn. Many of the phonological contrasts of the Qièyùn have been lost in Modern Mandarin or the Beijing dialect but well preserved in various modern dialects. There is no modern dialect that preserves all phonetic contrasts in the Qièyùn, but by comparing all phonetic contrasts across modern dialects, most of the phonetic contrasts in the Qièyùn can be illuminated. The phonetic values of these phonological contrasts provide valuable information for determining the phonetic values of the Middle Chinese phonological categories. This is basically what Bernhard Karlgren’s study (1915–1926) relied on. Some contrasts are systematically preserved, but some are only preserved sporadically. The phonetic values of some contrasts are easy to determine, but some are very difficult because the phonetic values have not only changed but have changed in very complex ways. 3.6.1

The Three-Way Contrast of Phonation Types

The stops and affricates with the same place of articulation have a three-way contrast in the rhyme dictionaries and rhyme tables. In modern dialects, some have a two-way contrast and some have a three-way contrast; like many other dialects, Modern Mandarin has a two-way contrast. The threeway contrast for the stops and affricates are found mainly in the modern Wu dialects and some of the Xiang dialects (see Table 3.53). The phonetic values of the dialects that still preserve the three-way contrast are the largest piece of evidence for determining the phonetic values of the Middle Chinese contrast. Also confirmed with other pieces of information, it becomes quite clear that Middle Chinese phonology had a similar contrast, which is three different phonation types: voiceless unaspirated, voiceless aspirated, and voiced. In IPA symbols, p-, ph-, b-; t-, th-, d-; and ts-, tsh-, dz-, etc., respectively.

The Qièyùn 切韻

147

Table 3.53 A comparison of the Beijing, Suzhou, Wenzhou, and Shuangfeng dialects, highlighting the maintenance of a three-way phonation type contrast in the Wu and Xiang dialects

波 bō 坡 pō 婆 pó 刀 dāo 滔 tāo 桃 táo 栽 zāi 猜 cāi 材 cái

‘wave’ ‘mountain slope’ ‘old woman’ ‘knife’ ‘torrent’ ‘peach’ ‘plant’ ‘guess’ ‘raw material’

Mandarin

Wu

Xiang

Beijing

Suzhou

Wenzhou

Shuangfeng

po/pho pho pho taw thaw thaw tsaj tshaj tshaj

pu phu bu tæ thæ dæ tse tshe ze

pu/pøy phu/pu bøy tɜ thɜ dɜ tse tshe ze

pʊ phʊ bʊ / bu tɤ thɤ dɤ tsue tsha dze

Source: This comparison is based on data found in the Běijīng Dàxué’s Hànyǔ Fāngyīn Zìhuì, 2nd ed. (2003). Note: Tones are omitted.

Table 3.54 A comparison of the Beijing, Suzhou, Guangzhou, and Yangjiang dialects, highlighting different coda systems, each descended from Middle Chinese, maintaining different contrasts, and merging others

林 鄰 零 立 栗 力

3.6.2

lín lín líng lì lì lì

‘forest’ ‘neighbor’ ‘zero’ ‘stand’ ‘chestnut’ ‘strength’

Mandarin

Wu

Cantonese

Beijing

Suzhou

Guangzhou

Yangjiang

lin lin liŋ li li li

lin lin lin liɪʔ liɪʔ liɪʔ

lɐm løn lɪŋ/lɛŋ lɐp/lap løt lɪk

lɐm lɐn lɪŋ lɐp lɐt lɪk

The Coda System

The arrangement of the rhyme dictionaries and rhyme tables indicates that Middle Chinese phonology had three nasal codas and three parallel stop codas (see Table 3.54). Some modern dialects still show such a coda system, while some show a reduced system, and some have lost contrast between the nasal and stop codas.

148

Middle Chinese

Table 3.55 An example of chóngniǔ contrast reflexes

一 乙

一 乙

yī yǐ

yī yǐ

‘one’ ‘second’

‘one’ ‘second’

MC

Mandarin

Wu

MC

BJ

SZ

Øit Øɰit

i i

iɪʔ iɪʔ

Hakka

Yue

MX

GZ

YJ

it iat

jɐt jyt

jɐt jit

Gan

Xiang

WZ

NC

CS

SF

iai/i iai

it it

i ie

i e/ia

XM

CZ

FZ

it it

ik ik

eiʔ eiʔ

Min

Note: MC = Middle Chinese; BJ = Beijing; SZ = Suzhou; WZ = Wenzhou; NC = Nanchang; CS = Changsha; SF = Shuangfeng; MX = Meixian; GZ = Guangzhou; YJ = Yangjiang; XM = Xiamen; CZ = Chaozhou; FZ = Fuzhou.

3.6.3

The Chóngniǔ 重紐 Contrast

It is hard to find evidence for the chóngniǔ contrast (refer to Sections 3.4.4 and 3.5.4.1) in the modern dialects. Only sporadic residues can be identified between some pairs of chóngniǔ words (see Table 3.55). In the Qièyùn, yī 一 ‘one’ and yǐ 乙 ‘second’ are words listed in the rù rhyme zhì 質 and both have the zero initial (yǐng 影 Ø-), so they form a pair of chóngniǔ words. But they are not homophonic, because in the rhyme tables, while yī 一 is listed in Division-IV, yǐ 乙 is listed in Division-III. They are phonologically contrastive. The contrast of these two words still remains in Changsha and Shuangfeng of the Xiang dialect, Meixian of the Kejia (Hakka) dialect, and Guangzhou and Yangjiang of the Yue dialect. One thing to note is that in Mandarin, these two characters have different tones, with yī 一 ‘one’ being first tone and yǐ 乙 ‘second’ being second tone. However, in Middle Chinese these characters were both rù tone, as seen by the -t coda. In Yue, yī 一 and yǐ 乙 also have differing tones (Yue, Guangzhou: 一 is tone 5, 乙 is tone 2 or 3),9 while in Hakka and Wu they are the same (Hakka, Miaoli: both 2; Wu, Shanghai: both 55). 3.6.4

The Chóngyùn 重韻 Contrast

The residues of chóngyùn contrasts can be found in various dialects. The character dài 戴 is a dài 代 rhyme character (the qù tone member of the hāi 咍 rhyme series), while dài 帶 is a tài 泰 rhyme character. The contrast has 9

There are two readings in Guangdong Cantonese, /jyːt2/ and /jyːt3/.

The Qièyùn 切韻

149

been lost in most dialects but is kept in the colloquial reading (noted before a “/” in Tables 3.56 and 3.57) of the Min dialect in Xiamen and Chaozhou. The Rank-III chóngyùn contrast between the yú 魚 and yú 虞 rhyme groups can be found in the Wu dialects spoken in Jiangsu and Zhejiang provinces (Chén 2003). The pairs of shǔ 鼠 vs. shū 輸 and qù 去 vs. qū 區 involve different tones (see Table 3.57). In the Wu dialects, some of the words of the yú 魚 rhyme have two readings, literary and colloquial. The literary readings are the same as the words of the yú 虞 rhyme, but the colloquial readings show the residue of the Middle Chinese contrast. Table 3.56 Some examples of the chóngyùn contrast of Rank-I rhymes hāi 咍 and tài 泰

咍: 泰:

戴 帶

‘wear’ ‘belt’

MC

Mandarin

Wu

Min

MC

Beijing

Suzhou

Xiamen

Chaozhou

Jianou

təj tɑj

taj taj

tᴇ/tɒ tᴇ/tɒ

tai/te tai/tua

tai/to tai/tua

tuɛ tuɛ

Source: Wáng Hóngjūn 1999; Běijīng Dàxué 2003. Note: MC = Middle Chinese; for two pronunciations divided by a slash, i.e. X/Y, X is the literary pronunciation, and Y is the colloquial pronunciation.

Table 3.57 The chóngyùn contrast between the yú 魚 and yú 虞 rhyme groups in the Wu dialects MC

Mandarin

Wu

MC

Beijing

Suzhou

Songjiang

Wenzhou

Lishui

tsɿ / tsy tsy

tsei tsɿ

ti / tɕy tɕy

魚: 虞:

豬 蛛

‘pig’ ‘spider’

ʈjɔ ʈjo

tʂu tʂu

tsɿ / tsʮ tsʮ

魚: 虞:

鼠 輸

‘mouse’ ‘lose’

ɕjɔ ɕjo

ʂu ʂu

tshɿ / tshʮ sʮ

tshɿ / tshy sy

tshei sɿ

tɕhi / ɕhy ɕy

魚: 虞:

鋸 句

‘saw’ ‘sentence’

kjɔ kjo

tɕy tɕy

kᴇ / tɕy tɕy

kɛ / tɕy tɕy

kei / tɕy tɕy

kəɯ / tɕy tɕy

魚: 虞:

去 區

‘go’ ‘district’

kʰiɔ kʰio

tɕhy tɕhy

tɕhi / tɕhy tɕhy

tɕhi / tɕhy tɕhy

khei / tɕhy tɕhy

khəɯ / tɕhy tɕhy

a

Note: MC = Middle Chinese; for two pronunciations divided by a slash, i.e. X/Y, X is the literary pronunciation, and Y is the colloquial pronunciation. a The symbol -ɿ is often used in Chinese and Korean phonology and corresponds to -ɨ in the IPA. The same can be said of -ʮ and -ʉ.

150

Middle Chinese

Table 3.58 The contrast between Rank-III and -IV in the Wu dialects

III IV

3.6.5

碾 年

‘grind’ ‘year’

Mandarin

Wu

Beijing

Yongkang

Pujiang

Tonglu

Qingtian

njɛn njɛn

ȵje ȵja

ȵjɪ ȵjɑ

ʔȵje ȵja

ȵje ȵjɑ

The Contrast of Rank-III and Rank-IV

The contrast of Middle Chinese Rank-III and Rank-IV rhymes of the same rhyme group can also be identified in the modern Chinese dialects, although it is scarce. It is reported that in some of the Wu dialects in Zhejiang province, the Middle Chinese Rank-III and Rank-IV words with -n or -m coda still show the phonological contrast (Jīn 1982). In Table 3.58, the contrast between RankIII and -IV in the Wu dialects is highlighted by the difference between niǎn 碾 and nián 年. 3.7

Examples of Middle Chinese Phonology

The poem Qiān Zì Wén 千字文 by Zhōu Xìngsì 周興嗣 (470–521) and the poem Běi Zhǔ 北渚 by Xiāo Gāng 蕭綱 (503–551) illustrate the Middle Chinese phonetic values that basically reflect the phonology of the sixth century. Of the phonetic transcriptions, the first line is the reconstructed value, the second is the phonetic value of Modern Mandarin and the third is pinyin. The rhyming syllables are in bold face. See Table 3.59 for the reconstructions of various stages of Chinese reflecting the original Middle Chinese pronunciation of Qiān Zì Wén 千字文. Qiān Zì Wén 千字文 ‘Thousand Character Essay’ (first four lines) by Zhōu Xìngsì 周興嗣 (470–521) 天地玄黃, Heaven and earth, dark and yellow; 宇宙洪荒。 The universe, vast and great. 日月盈昃, Sun and moon, full and setting; 辰宿列張。 Starts and lunar mansions, arranged and spread.

Source: English translation by Francis W. Paar 1963.

According to the Qièyùn system, huáng 黃 and huāng 荒 belong to the táng 唐 rhyme (Rank-I), and zhāng 張 belongs to the yáng 陽 rhyme (Rank-III). But in Qiān Zì Wén, these words rhyme together, indicating their difference could simply be presence or absence of the palatal medial -j-.

The Qièyùn 切韻

151

Table 3.59 The reconstructed Middle Chinese pronunciation of Qiān Zì Wén 天 then thjɛn tiān

地 di ti dì

玄 ɦwen ɕɥɛn xuán

黃 ɦwɑŋ xwaŋ huáng

宇 ɦjo y yǔ

宙 ɖiw tʂəw zhòu

洪 ɦuŋ xuŋ hóng

荒 hwɑŋ xwaŋ huāng

日 ȵit ɻɨ rì

月 ŋwjɐt ɥɛ yuè

盈 jɛŋ iŋ yíng

昃 tʂɨk tsɤ zè

辰 dʑin tʂhən chén

宿 siw ɕjəw xiù

列 ljɛt ljɛ liè

張 ʈjɐŋ tʂaŋ zhāng

Běi Zhǔ 北渚 ‘North Isle’ by Xiāo Gāng 蕭綱 (503–551) 岸陰垂柳葉, Shores shaded by hanging willow fronds, 平江含粉堞。 Smooth stream with white parapets within. 好值城旁人, Glad to meet others beside these walls, 多逢蕩舟妾。 Girls often encountered in boats rushing by. 綠水濺長袖, Sapphire waters splash their long sleeves, 浮苔染輕檝。 And drifting mosses dye the light paddles.

Source: English translation by Stephen Owen 1996: 327.

Table 3.60 The reconstructed Middle Chinese pronunciation of Běi Zhǔ 岸 ŋɑn an àn

陰 Øɰim in yīn

垂 dʑwjɛ tʂhwej chuí

柳 liw ljəw liǔ

葉 jɛp jɛ yè

平 bɰjaŋ phiŋ píng

江 kɰɔŋ tɕjaŋ jiāng

含 ɦəm xan hán

粉 pjun fən fěn

堞 dep tjɛ dié

好 hɑw xaw hào

值 ɖɨk tʂɨ zhí

城 dʑjɛŋ tʂhəŋ chéng

旁 bɑŋ phaŋ páng

人 ȵin ɻən rén

多 tɑ two duō

逢 bjoŋ fəŋ féng

蕩 dɑŋ taŋ dàng

舟 tɕiw tʂəw zhōu

妾 tshjɛp tɕhjɛ qiè

綠 ljok ly lǜ

水 ɕwi ʂwej shuǐ

濺 tsjɛn tɕjɛn jiàn

長 ɖjɐŋ tʂhaŋ cháng

袖 ziw ɕjəw xiù

浮 biw fu fú

苔 dəj thaj tái

染 ȵjɛm ɻan rǎn

輕 khjɛŋ tɕhiŋ qīng

檝 tsjɛp tɕi jí

See Table 3.60 for the reconstructions of various stages of Chinese reflecting the original Middle Chinese pronunciation of Běi Zhǔ. According to the Qièyùn system, yè 葉, qiè 妾, and jí 檝 belong to the yè 葉 rhyme (Rank-III), and dié 堞 belongs to the tiē 帖 rhyme (Rank-IV). But they rhyme according to Xiāo Gāng’s Běi Zhǔ. So the phonetic difference between these two rhymes may not be the difference of the main vowel, but rather of the medial.

4

Middle Chinese: The Tang and Song Dynasties

Middle Chinese phonology, as a period of the Chinese language, also has its variants in time and space. It has been proposed that Middle Chinese should be further divided into two periods: Early Middle Chinese and Late Middle Chinese. The former is represented by the Qièyùn 切韻 and the latter by early rhyme tables such as the Yùnjìng 韻鏡 (Pulleyblank 1984), which may represent the phonology of Chang’an (modern day Xi’an), the capital of the Tang dynasty. Evidence for the existence of dialectal variants is also clear; as seen in Section 4.2, the northwestern dialect of the Tang dynasty as reflected in Tibetan materials clearly shows characteristics different from the standard pronunciation (Luó 1933; Takata 1988; Coblin 1991).

4.1

The Tang Dynasty (618–907)

The Tang dynasty was one of the most prosperous times in Chinese history. Its imperial examination system demanded a national rhyming standard (Guānyùn 官韻), for which the Qièyùn was adopted. Over time a number of revisions appeared, including Wáng Rénxù’s Kānmiù Bǔquē Qièyùn 刊謬補缺切韻. Because of the Tang dynasty’s strong social and cultural influence, the Chinese language was borrowed into the neighboring countries of Korea, Japan, and Vietnam. The pronunciations of the borrowed Chinese words in these countries form what are called Sino-Korean, Sino-Japanese (Kan-on), and Sino-Vietnamese.

4.1.1

Xuánzàng’s 玄奘 Inscriptions

Xuánzàng (602–664) was a famous Buddhist monk, scholar, and translator during the early Tang dynasty. In 629 he travelled to India to study Buddhism. In 645, after more than fifteen years, he returned back to Chang’an (modern Xi’an), the capital of the Tang dynasty, with more than 600 Buddhist documents. He retired to a monastery and devoted his energy to translating Buddhist texts until his death in 664. 152

The Tang and Song Dynasties

153

Xuánzàng is known for his extensive but careful translations of Indian Buddhist texts into Chinese. He was born in Luoyang area, so the Chinese-Sanskrit transliterations of the proper words (personal names, place names, and Buddhist terminologies) contained in his translations thus reflect the sounds of the Luoyang dialect of that time. Analysis shows that Xuánzàng’s transliteration is very precise. The correspondences between Sanskrit syllables and Chinese characters are very regular and consistent (Shī 1983). Table 4.1 shows the reconstructed main vowels according to Xuánzàng’s transliterations. The píng 平 rhymes also represent their corresponding shǎng 上, qù 去, and rù 入 rhymes. It should be noted that not all the characters of each of the Qièyùn/Guǎngyùn 廣韻 rhymes are found in the transcriptions. Table 4.1 Reconstructions of Xuánzàng’s transliterations

ɒ

a

u

i

o

e



-n

-m



-w

-j

歌 gē 戈 gē

寒 hán (桓 huán) 元 yuán 凡 fán

談 tán 覃 tán 嚴 yán

唐 táng 陽 yáng

豪 háo

咍 hāi 泰 tài 灰 huī 廢 fèi

(14)

(山 shān) 刪 shān 仙 xiān

咸 xián 銜 xián 鹽 yán

皆 jiē (夬 guài) 祭 jì

(14)

麻 má 佳 jiā

模 mó 虞 yú

文 wén 魂 hún

支 zhī 脂 zhī 之 zhī 微 wēi

真 zhēn 臻 zhēn 殷 yīn (諄 zhūn)

魚 yú

痕 hén

齊 qí

先 xiān

庚 gēng 耕 gēng 清 qīng

(肴 yáo) 宵 xiāo

東 dōng (5) 侵 qīn

(8)

添 tiān

(冬 dōng) 鍾 zhōng

侯 hóu 尤 yóu

(6)

青 qīng

蕭 xiāo

(5)

ɔ

江 jiāng

ə

登 dēng 蒸 zhēng

(1) (幽 yōu)

Source: Based on Shī 1983, with the medials, -r, -w, and -j, excluded. Note: Rhymes that were not found in the transcriptions are in parentheses.

(3)

154

Middle Chinese

4.1.2

Simplification and Tóngyòng 同用 Marking

In the Guǎngyùn of 1008, some rhymes are marked as dúyòng 獨用 ‘independent’ and some are marked as tóngyòng 同用 ‘shared.’ Shared rhymes are apparently the ones that cannot for the most part be distinguished by those who spoke with the standard pronunciation of the time. It is also likely that the shared rhymes are the ones whose characters are present in fewer characters. It is not so practical to use such rhymes alone for poetry composition. Although its most famous case is in the well-known Guǎngyùn, which is a Song dynasty text, this revision of rhymes with tóngyòng marking was first mentioned in the Tang dynasty and indicates changes in the Qièyùn rhymes, although it must have started earlier in the Tang, before the Guǎngyùn’s dúyòng and tóngyòng rhymes. Fēng Yǎn’s 封演 ninth-century Wén Jiàn Jì 聞見記 ‘Records of Hearsay and Personal Experience’ recorded a very valuable piece of information regarding Xǔ Jìngzōng’s 許敬宗 (592–672) proposal to simplify the rhyming categories: Lù Fǎyán 陸法言 of Sui, together with Yán (Zhītuī) 顏之推, Wèi (Yànyuān) 魏彦淵 and others, reconciled northern and southern pronunciation and compiled the Qièyùn, altogether 12,158 characters, and made it a standard for literature. But categories such as 先 sɛn and 仙 sian, 刪 ʂarn and 山 ʂɛrn were made separate rhymes and scholars engaged in composition all found its excessive minuteness troublesome. In the early years of the present dynasty Xǔ Jìngzōng 許敬宗 and others discussed it in detail and on the grounds that its rhymes were too narrow, memorialized to the throne that [in such cases] they should be combined and used together.1 (Fēngshì Wén Jiàn Jì 封氏聞見記; English translation by Edwin G. Pulleyblank 1984: 139)

Looking at poems composed during the imperial examinations (Wáng Zhàopéng 1998, 2004), the rhyming patterns have a clear distinction before and after 717, the fifth year of the Kaiyuan reign. After 717 the rhyming of these poems is almost exactly in accordance with the dúyòng and tóngyòng categories: of 5,112 rhyming characters, only 43 (0.84 percent) are in violation of the dúyòng and tóngyòng categories, which clearly indicates that these categories actually appeared in the Tang dynasty and were incorporated into the Guǎngyùn later. Examples of the rhyming words found in the poems composed during the imperial examinations before the year 717 are shown in Table 4.2. According to the tóngyòng categories of the Guǎngyùn, words from the zhēn 真 rhyme and the wén 文 rhyme should not rhyme together.

1

隋朝陸法言與顏、魏諸公定南北音。撰為《切韻》。凡一萬二千一百五十八字,以為文楷 式。而先仙、刪山之類分為別韻,屬文之士共苦其苛細。國初,許敬宗等詳議,以其韻 窄,奏合而用之。

The Tang and Song Dynasties

155

Table 4.2 The rhyming words in poems composed during the imperial examinations before the year 717 Year

Scholar

zhēn 真 rhyme

713 714

Lǐ Méng 李蒙 Lǐ Áng 李昂

chén 臣, rén 人 chén 塵

wén 文 rhyme chún 淳 xūn 勳, wén 文, yún 雲

is rhymed with is rhymed with

This system is also closely reflected in the rhyming practice of the Tang poets in the Guanzhong area (Chǔ 2005), which includes the Tang capital of Chang’an. 4.1.3

The New Standard

A very important piece of information about the phonological system of the early eighth century, the tóngyòng label indicates the merger of Qièyùn rhymes. The Qièyùn system requires at least twelve vowels to distinguish all phonological contrasts (Féng 1998; Huáng Xiàoshān 2002a). For example, there are twelve different rhymes with -ŋ, so, until the recent incorporation of medials into the final for phonological analysis, the same number of vowels are required for each rhyme. These twelve rhymes are listed in Table 4.3. As the tóngyòng labels indicate, the twelve rhymes can be reduced to seven groups, as shown in Table 4.4. A similar observation can be made of the rhymes with -n coda (see Table 4.5). The tóngyòng of wén 文 and yīn 殷 (also shown as xīn 欣) was added in the year 1037, during the Song dynasty (see Table 4.6). Table 4.3 The twelve -ŋ coda rhymes of the Qièyùn 東























Table 4.4 The seven groups according to tóngyòng labels 東

冬 = 鍾



陽 = 唐

庚 = 耕 = 清



蒸 = 登

Table 4.5 The twelve rhymes in the Qièyùn system with -n coda 真























156

Middle Chinese

Table 4.6 The additional tóngyòng of wén 文 and yīn 殷, added in 1037 真 = 臻





元 = 魂 = 痕



刪 = 山

先 = 仙

Table 4.7 All tóngyòng rhymes, organized by coda -ŋ 東 uŋ

冬 = 鍾 oŋ

江 ɔŋ

陽 = 唐 aŋ

庚 = 耕 = 清 ɛŋ

青 eŋ

蒸 = 登 əŋ (iŋ)

-n 真 = in



文 un

殷 ɨn

元 = 魂 = ən



寒 ɑn



先 = ɛn

-m 覃 = ɑm



侵 im

鹽 = ɛm



咸 = am



嚴 = 凡 əm

-j 泰 ɑj

廢 əj

夬 = aj

皆 =



齊 = ej



灰 = 咍 ɔj

-w 豪 ɑw

肴 aw

宵 = ɛw



侯 = 尤 = əw



-Ø 支 = i

脂 = 之

微 e

魚 o



虞 = u

歌 ɑ

刪 = an



麻 a

By examining all rhymes with the same coda (the subsystem underlying the tóngyòng labeling), it becomes clear that no more than seven distinctions are needed for any of the subsystems, and therefore only seven vowels in the entire system. This is a different vowel system than that required by the Qièyùn system. All tóngyòng rhymes, organized by coda, with reconstructed vowels are shown in Table 4.7. Of all the tóngyòng rhymes in the Jíyùn 集韻, thirteen were added during the Song dynasty (píng, shǎng, qù, and rù counted separately). In an article entitled ‘Táng Sòng Yùn Tóngyòng Dúyòng Bùtóng,’2 Qián Dàxīn 錢大昕 (1728–1804) proposed that the tóngyòng between yīn 殷 (Middle Chinese: -ɨn) 2

唐宋韻同用獨用不同 ‘The differences of tóngyòng and dúyòng between the rhyme dictionaries of Tang and Song.’

The Tang and Song Dynasties

157

Table 4.8 A rhyme chart arranged according to tóngyòng markings i -Ø

支脂之 jɛ, i, ɨ

-j

e

ɛ

a

ɨ 微 ɨj

宵蕭 jɛw, ew

麻 ɰa, ja 夬皆佳 ɰaj, ɰæj, ɰæ 肴 ɰaw 咸銜 ɰæm, ɰam 刪山 ɰan, ɰæn

齊祭 ej, jɛj

-w -m

侵 im

鹽添 jɛm, em

-n

真臻 in, ɪn

先仙 en, jɛn



青 iŋ

庚耕清 ɰ(j)aŋ, ɰæŋ,jɛŋ

ə

ɑ

廢 jɐj

歌 ɑ 泰 ɑj

侯尤幽 əw, iw, ɨw 嚴凡 jɐm, jɐma

豪 ɑw 談覃 ɑm, am

殷 元魂痕 ɨn jɐn, won, ən 蒸登 ɨŋ, əŋ

ɔ

o

u

魚 jɔ

虞模 jo, wo

灰咍 woj, əj

寒 ɑn 陽唐 jɐŋ, ɑŋ

文 jun 江 ɰɔŋ

冬鍾 東 woŋ, uŋ ioŋ

Note: The rhyme patterns found in imperial examinations from 717, the fifth year of the Kāiyuán 開元 era, to 901, the fourth year of Guānghuà 光化 era, are underlined. a yán 嚴 -jɐm and fán 凡 -jɐm are not distinguished according to Pān (2000).

and wén 文 (Middle Chinese: -jun), yán 嚴 (Middle Chinese: -jɐm) and yán/ tiān 鹽/添 (Middle Chinese: -jɛm/-em), fán 凡 (Middle Chinese: -jɐm) and xián/xián 咸/銜 (Middle Chinese: -ɰæm/-ɰam), and fèi 廢 (Middle Chinese: -jɐj) and duì (duì)/dài 對 (隊)/代 (Middle Chinese: -woj/-əj)3 were proposed by Jiǎ Chāngcháo 賈昌朝 in 1037. In other words, the rest were likely inherited from the Tang dynasty. Such an idea can be confirmed in the rhyming of shī 詩 ‘poems’ and fù 賦 ‘prose-poems’ composed during imperial examinations of the Tang. Table 4.8 is a rhyme chart arranged according to tóngyòng markings. The reconstructed values of the Qièyùn show the closeness of their phonetic values. The rhymes in the same group are the ones marked as shared. This simplified system with the tóngyòng information is also an ideal standard. The tóngyòng information may come from different dialects, but many of these distinctions also could be historical instead of dialectal according to the nature of the Qièyùn. And some of the merged rhymes have a very limited number of characters that are often not enough to be used in poetry composition. Nevertheless, based on the tóngyòng information, the vowel system of the standard phonology had already been significantly

3

yīn 殷, yán 嚴, and fán 凡 include their corresponding shǎng, qù, and rù rhymes.

158

Middle Chinese

Table 4.9 The values of Qièyùn and post-717 Tang poetry rhymes Qièyùn rhyme 侵 鹽 添 咸 銜 嚴 凡 談 覃 真 臻 先 仙 刪 山 殷 元 魂 痕 寒 文 青 庚 耕 清 蒸 登 陽 唐 江 冬 鍾 東

qīn yán tiān xián xián yán fán tán tán zhēn zhēn xiān xiān shān shān yīn yuán hún hén hán wén qīng gēng gēng qīng zhēng dēng yáng táng jiāng dōng zhōng dōng

-im -jɛm -em -ɰæm -ɰam -jɐm -jɐm -ɑm -am -in -ɪn -en -jɛn -ɰan -ɰæn -ɨn -jɐn -won -ən -ɑn -jun -iŋ -ɰ(j)aŋ -ɰæŋ -jɛŋ -ɨŋ -əŋ -jɐŋ -ɑŋ -ɰɔŋ -woŋ -ioŋ -uŋ

Post-717 rhyme – -im > -ɛm > -am > -əm > -ɑm > -in > -ɛn > -an – -ɨn > -ən – – – >

-ɑn -un -eŋ -ɛŋ

> -əŋ

Qièyùn rhyme 支 脂 之 微 麻 歌 魚 虞 模 齊 祭 夬 皆 佳 廢 泰 灰 咍 宵 蕭 肴 侯 尤 幽 豪

zhī zhī zhī wēi má gē yú yú mó qí jì guài jiē jiā fèi tài huī hāi xiāo xiāo yáo hóu yóu yōu háo

-jɛ -i -ɨ -ɨj -ɰa/-ja -ɑ -jɔ -jo -wo -ej -jɛj -ɰaj -ɰæj -ɰæ -jɐj -ɑj -woj -əj -jɛw -ew -ɰaw -əw -iw -ɨw -ɑw

Post-717 rhyme > -i – – – – >

-e -a -ɑ -o -u

> -ej > -aj

– -əj – -ɑj > -ɔj > -ɛw – -aw > -əw –

-ɑw

> -aŋ – -ɔŋ > -oŋ –

-uŋ

reduced in the eighth and even the seventh century. Table 4.9 shows the values of Qièyùn and post-717 Tang poetry rhymes. The result of the mergings is the difference in rhyming schemes of Tang poetry. This system requires a four-degree contrast in vowel height. However, some of the rhymes (yīn 殷, yán-tiān 鹽添, guài-jiē-jiā 夬皆佳, and their corresponding shǎng, qù, and rù rhymes) were never used in poetry composition due to their limited number of characters (and are thus bolded in Tables 4.8 and 4.9) and there is a case of a violation between the dōng 東 and zhōng 鍾

The Tang and Song Dynasties

159

Table 4.10 The relationship between the phonetic, phonemic, and rhyming systems of Tang-era Middle Chinese Phonetic i e ɛ

Phonemic

ɨ ə

u ɔ ɒ

a

16 rhyme group system

i >

u ə

ɛ

i >

ɔ ɒ

a

u ə

ɛ

ɔ

a

rhymes. The relationship between the phonetic, phonemic, and rhyming systems of Tang-era Middle Chinese is shown in Table 4.10. This system is very close to the sixteen rhyme group system of the rhyme tables in which the two low vowels -a and -ɒ are merged. Their phonetic values can be compared to the Buddhist transcriptions made by Xuánzàng, as seen in Section 4.1.1. 4.1.4

Devoicing of Voiced Stops and Affricates

Voiced stops and affricates became voiceless aspirated (Liú Guǎnghé 1984), as seen in the Tang monk Bùkōng’s 不空 transcriptions. These Chinese-Tibetan manuscripts were discovered in the well-known Hidden Library of Dunhuang (dūnhuáng 敦煌). They were originally meant to assist Tibetans in learning Chinese. Their translations are not strictly limited to isolated terms because they are not limited to terminologies. Judging from the region where these manuscripts were recovered, they probably represent the northwestern dialect during the Tang and Five Dynasties period (Luó 1933). In the northwestern dialects the devoicing tends to proceed from fricatives to affricates, and then to stops (see Table 4.11). These manuscripts are the Āmítuójīng 阿彌陀經, the Jīngāngjīng 金剛經, the Tángfán Huìméngbēi 唐蕃會盟碑, the Qiān Zì Wén Table 4.11 The progression of devoicing across Tang-era Northwestern Chinese dialects

阿彌陀經 金剛經 唐蕃會盟碑 千字文 大乘中宗見解





















b b b b ph

b, mb b, mb b b, ph b, ph

d, nd d d d d , th

dʑ dʑ tɕ dʑ dʑ, tɕh

ɕ ɕ ɕ tɕh, ɕ ɕ

ɕ ʑ, ɕ ʑ, ɕ ɕ ɕ

dz dz z dz, dʑ tsh

– s – s s

ɡ ɡ k ɡ kh

h h h h h

160

Middle Chinese

千字文, and the Dàchéng Zhōngzōng Jiànjiě 大乘中宗見解. The first four are undated manuscripts from the Dunhuang caves, while the last is a tone monument, which was established in the year 823 during the Changqing reign of the Tang dynasty. The dates of the first four are estimated from the eighth to the ninth century, based on the historical information and phonetic information contained in the texts (Luó 1933). The translations of these Chinese-Tibetan sources (Luó 1933) are provided here: 阿彌陀經 金剛經 唐蕃會盟碑 千字文 大乘中宗見解

Fragments of a Chinese version of the Smaller Sukhāvatī Vyūha in Tibetan writing Fragments of a Chinese version of the Vajracchedika in Tibetan writing Rubbings of the Tarng-Fan Hueymeng Bei (Táng-fān Huìméng Bēi) Fragments of Chiantzyh Wen (Qiān Zì Wén) with interlinear Tibetan translations Fragments of the Mahāyāna-Mādhyamika-Darśana with interlinear Tibetan translations

The initial mb, seen in Table 4.11, is a nasalized stop, which is one of the basic phonological characteristics of the northwestern dialect of Tang (Luó 1933) (see Section 4.2 for more information). This characteristic still exists in the modern Jin dialects (Qiáo 2004). 4.1.5

Labiodentalization

Huáng Cuìbó ([1970] 2010) shows that bilabials and labiodentals are clearly distinguished in Huì Lín’s 慧琳 (736–820) Yīqiè Jīng Yīnyì 一切經音義, a collection of glosses of the Buddhist canon. Huì Lín’s fǎnqiè 反切 are based on Yuán Tíngjiān’s 元廷堅 Yùnyīng 韻英 (724–726) and Zhāng Jìngē’s 張晉 戈 Kǎoshēng Qièyùn 考聲切韻 (684–704). According to the analysis of the fǎnqiè spellings Huáng concluded that there were three labiodental initials in Huì Lín’s time, as seen in Table 4.12 (Huáng Cuìbó [1970] 2010: 18–19). This reduction of labial initials from a four-way contrast to a three-way contrast when preceding -j- indicates that labiodentalization has taken place and pj- and Table 4.12 The Tang-era Chinese three-way contrast within labiodental sounds, reconstructed as f-, v-, and ɱ-, respectively fvɱ-

夫 fū < pjo ‘man, husband’ 扶 fú < bjo ‘hold on’ 武 wǔ < mjo ‘martial’

非 fēi < pwɨj ‘not, wrong’ 肥 féi < bwɨj ‘fat’ 微 wēi < mwɨj ‘small’

敷 fū < phjo ‘apply’ 奉 fèng < bjoŋ ‘give with respect’ 文 wén < mjun ‘civil, writing’

峰 fēng < phjoŋ ‘peak’ 房 fáng < bwjɐŋ ‘house’ 芒 máng < mwjɐŋ ‘arista’

The Tang and Song Dynasties

161

phj- have changed to f-. Table 4.13 is a correspondence table showing the labiodentalization of labials preceding the medial -j- in Tang-era Chinese. In later rhyme dictionaries, these -j- conditioned labiodentals would remain recongnized as Rank-III characters. As shown in Table 4.13, the initials of syllables originally with pj- and phj- cannot be distinguished. The labiodentalization can also be proven with evidence from SinoVietnamese. According to the historical record, this sound system was transmitted to Vietnam during the middle of the Tang dynasty (Wáng Lì 1948). As shown in Table 4.14, the labiodental fricatives are realized as /f/ and /v/ in the Sino-Vietnamese (Wáng Lì 1948: 26–27). The initial characters according to the traditional thirty-six initial categories are given in parentheses, and examples follow written in Vietnamese, with IPA in brackets. The lack of distinction between initials fēi 非, fū 敷, and fèng 奉 also hints that they are labiodental fricatives, which merged in their borrowing to Vietnamese. Table 4.13 The labiodentalization of labials preceding the medial -j- in Tang-era Chinese Middle Chinese

Condition

Tang Chinese

Example

p

preceding -jotherwise

f p

夫 崩

pjo pəŋ

> >

fjo pəŋ

fū ‘man, husband’ pēng ‘collapse’

ph

preceding -jotherwise

f ph

峰 判

phjoŋ phwɑn

> >

fjoŋ phwɑn

fēng ‘peak’ pàn ‘judge’

b

preceding -jotherwise

v b

房 白

bwjɐŋ bɰak

> >

vjɐŋ bɰak

fáng ‘house’ bái ‘white’

m

preceding -jotherwise

ɱ m

文 母

mjun məu

> >

ɱjun məu

wén ‘civil, writing’ mǔ ‘mother’

Table 4.14 The labiodental fricatives realized as /f/ and /v/ in Sino-Vietnamese p-

(fēi 非)

ph-

(fū 敷)

b-

(fèng 奉)

m-

(wēi 微)

法 pháp [fap] 忿 phẫn [fəŋ] 伐 phat _ [fat] 萬 van _ [vaŋ]

發 phát [fat] 拂 phất [fət] 憤 phẫn [fəŋ] 物 vật [vət̚]

分 phân [fəŋ] 芬 phân [fəŋ] 房 phòng [fawŋ͡m] 望 vong _ [vawŋ͡ m]

弗 phất [fət] 肺 phế [fe] 扶 phù [fu] 微 vi [vi]

諷 phúng [fʊwŋ͡m] 覆 phúc [fʊwk͡p] 服 phuc _ [fʊwk͡p] 武 vũ [vu]

162

Middle Chinese

4.1.6

Later Middle Chinese

Edwin G. Pulleyblank (1984) pointed out that the phonological system of the Yùnjìng is based on the Qínyīn 秦音 (Chang’an pronunciation). The main evidence for this is from the Yīqiè Jīng Yīnyì, whose phonological system is based on the Qínyīn. Pulleyblank made a connection between the Yīqiè Jīng Yīnyì and the rhyme table Yùnjìng based on their phonological similarities. The revised version of the Yùnjìng by Zhāng Línzhī 張麟之 was written during the Southern Song (1127–1279), but the naming and classification, such as the divisions (děng) and initials (shēngmǔ), can be traced back to the tenth century. The fragments of a phonological work by Shǒu Wēn 守溫, a Tang Buddhist monk, contain structural terminologies such as divisions and wǔyīn 五音 (the five initial categories), which are similar to those employed in the rhyme tables. The influence of Tang culture was strong. According to Pulleyblank, the structural terminology employed in the Yùnjìng reflects the dialect of Chang’an, a standard different from that of the Qièyùn. Chinese words entered the languages of adjacent countries, resulting in standard Tang pronunciation being reflected in the readings of Chinese characters in the Korean, Japanese, and Vietnamese languages, and so these Sino-xenic pronunciations are most likely derived from Later Middle Chinese, as exemplified in the Yùnjìng. To distinguish Early and Late Middle Chinese, Pulleyblank examined various Sino-xenic pronunciations and reduced a number of phonological contrasts in his reconstruction (see Table 4.15). The phonological categories of the existing version of the Yùnjìng, however, are the same as the Guǎngyùn. Each of the 206 rhymes is separately listed. Of course, if all the categories were recognized in the reconstruction, the differences between Early Middle Chinese and Late Middle Chinese would only be at the phonetic level. It should be noted that Pulleyblank’s reconstruction of Early Middle Chinese is not bound by the structural terminologies of the rhyme tables. He Table 4.15 The distinction of Early and Late Middle Chinese in Pulleyblank’s reconstruction Rank Early Middle Chinese Late Middle Chinese

I 豪皓号 aw aw, uaw

Rank Early Middle Chinese Late Middle Chinese

I 侯厚候 ow əw

a

II 肴巧效 arw, ɛrw jaaw, aaw

III 宵小笑 iaw, jiaw iaw, jiaw

IV 蕭篠嘯 ɛwa iaw, jiaw

III 尤有宥 uw iw, əw

III 幽黝幼 jiw jiw

Pulleyblank wrote ɛu here instead of ɛw (1984: 234), seemingly a typo.

The Tang and Song Dynasties

163

reconstructs the yóu 尤 rhyme as uw. In the rhyme tables, the yóu 尤 rhyme is classified as a Rank-III rhyme, which requires a palatal medial or a high front vowel. In his reconstruction of Late Middle Chinese, Pulleyblank deletes the distinction between Rank-III and -IV. The phonetic values reflected in Sino-xenic materials are altered according to the borrowing languages, therefore whether the changes are due to the different phonology of the borrowing language or changes in Chinese phonology cannot be determined. In general, if the Sino-xenic materials reflect a distinction in Chinese, this proves the existence of the distinction, whereas if a distinction is not reflected, it does not necessarily prove the loss of the distinction. Clearly the available phonetic information is not enough to provide phonological values to each of the categories in the Yùnjìng, which Pulleyblank believes to be based on the Late Middle Chinese phonology. This makes the result unsatisfactory, though a very important point is that structural terminology such as divisions should not be automatically applied to the Qièyùn system without justification. 4.2

The Five Dynasties and Ten Kingdoms (907–960)

Between the end of the Tang dynasty and the beginning of the Song was a fifty-three-year period known as the Five Dynasties and Ten Kingdoms period. As the name suggests, this era was unstable and divided amongst multiple smaller kingdoms and empires, all vying for power. Because of this, there was no new standard created during this period that could ever be used or accepted across the ever-changing borders of the time. However, a phonological work called the Ěryǎ Yīntú 爾雅音圖 reveals some important phonological characteristics of the period. 4.2.1

Ěryǎ Yīntú 爾雅音圖 (tenth century)

The Ěryǎ Yīntú was likely written by Wú Zhāoyì 毋昭裔, a high-ranking officer of the Later Shu dynasty (Hòushǔ 後蜀 934–965) in the Five Dynasties period (Wǔdài 五代) (Féng 2007). Based on the characters used for the phonetic notations of zhíyīn 直音 (see Section 1.3.3), a number of phonological characteristics can be identified. For convenience of illustration, in Table 4.16 the Middle Chinese initial characters are listed after the phonetic values of the initials. For example, after initial p-, the Middle Chinese initial characters bāng 幫 and bìng (zè) 並仄 are listed: this indicates that Middle Chinese b- (並) of the zè (non-píng) tone syllables (並仄) are realized as the same as Middle Chinese p- (幫). And after initial ph, the Middle Chinese initial characters pāng 滂 and píng-tone bìng

164

Middle Chinese

Table 4.16 The distribution of Qièyùn initials in the Ěryǎ Yīntú 幫並仄 端定仄 精從仄 知二澄二仄 莊崇仄 tʃ 知三澄三仄

p t ts tʂ

章禪仄 k 見群仄

滂並平 透定平 清從平 徹二澄二平 初崇平 tʃh 徹三澄三 ph th tsh tʂh

m 明 n 泥娘

ʋ 微 l 來 r 日

f 非敷奉 s 心邪俟 ʂ 生禪 ʃ 書船



昌禪平 kh 溪群平

x 曉匣

ŋ 疑

Ø 影云以疑

Note: Some voiced initials merge differently based on the tone (píng-zè 平仄) of the characters themselves.

Table 4.17 The devoicing pattern in the Ěryǎ Yīntú Middle Chinese

Condition

Ěryǎ Yīntú

Example

[+voice]

píng tone

[-voice, +aspirated]

zè tone

[-voice, -aspirated]

床 剽 琴 才 狀 部 近 在

dʒbgdzdʒbgdz-

> > > > > > > >

tʃhphkhtshtʃpkts-

chuáng ‘bed’ piáo ‘rob’ qín ‘Chinese lute’ cái ‘talent’ zhuàng ‘shape’ bù ‘part, section’ jìn ‘near’ zài ‘be at’

Source: Phonetic values are Féng’s (2007).

並平 are listed: this indicates that Middle Chinese b- (並) of the píng tone syllables (並平) are the same as Middle Chinese ph- (滂). The same can be said for the other initial pairs of t and th, ts and tsh, tʂ and tʂh, tʃ and tʃh, and k and kh listed in the first two columns of Table 4.16. The Ěryǎ Yīntú is one of the earliest pieces of evidence for the devoicing pattern of Middle Chinese voiced obstruents as seen in the Mandarin phonology. Namely the voiced stops and affricates became voiceless aspirated and voiceless unaspirated counterparts according to their respective tone. Voiced stops and affricates became voiceless aspirated in the píng tone syllables and voiceless unaspirated in zè (non-píng) tone syllables. Table 4.17 shows the devoicing pattern in the Ěryǎ Yīntú. Middle Chinese has three phonation types for stops and affricates, voiced (such as 並 b-, 定 d-, 群 ɡ-), voiceless aspirated (such as 滂 ph-, 透 th-, 溪 kh-), and voiceless unaspirated (such as 幫 p-, 端 t-, 見 k-). In the Chinese of the Ěryǎ Yīntú, all stops and affricates lose their voicing, and voiced initials are divided into the voiceless aspirated and

The Tang and Song Dynasties

165

voiceless unaspirated categories by their tone, with píng tone voiced syllables becoming voiceless aspirated, and zè tone voiced syllables becoming voiceless unaspirated.4 In this initial system the Middle Chinese zhī 知 initial series also merged with the Middle Chinese zhuāng 莊 and zhāng 章 initial series according to the děng (see Section 1.4.4 for comparison). There are also indications of the loss of some of the velar nasal initials. 4.2.2

Northwest Dialects

During this period, Tang standards were used for poetry composition, but notably the northwest dialects of the time (called the northwest TangWudai dialects (Luó 1933)) had features that were unseen in the standards of the Tang. According to Luó Chángpéi’s (1933) study, in comparison with the Qièyùn system, the Tang-Wudai dialects had quite clear phonological characteristics in their initials and finals. For the initials, the Tang-Wudai dialects had the following characteristics: 1. Middle Chinese voiced obstruent initials are devoiced. According to Dàchéng Zhōngzōng Jiànjiě 大乘中宗見解 Mahāyāna-Mādhyamika-Darśana, the Middle Chinese initials b-, d-, g-, dz-, etc., with a few exceptions, were used to transcribe voiceless aspirated initials (e.g. píng tone syllables: pú 菩 phu [Luó’s phonetic transcriptions are changed to IPA] < bwo, tóng 同 thoŋ < duŋ; as well as zè tone syllables: bié 别 phar < bɰjɛt, dú 別 thoɡ < duk). 2. In all the materials Middle Chinese retroflex stop initials, ʈ-, ʈh-, ɖ-, are merged with retroflex and palatal affricate initials tʂ-, tʂh-, dʐ-; tɕ-, tɕh-, dʑrespectively (e.g. zhī 知 tʃi < ʈjɛ, chāo 超 tʃhe’w < ʈhjɛu, zhì 治 tʃhi < ɖi). 3. Middle Chinese voiced fricative initials, ʑ-, z-, ɦ-, changed to voiceless ones, ɕ-, s-, h-, respectively (e.g. shì 市 ʃi < ʑɨ/dʑɨ , xiè 謝 sja < zja, hé 河 ha < ɦɑ). 4. Middle Chinese nasal initials, m-, n-, ŋ-, are sometimes used to transcribe voiced stops, b-, d-, ɡ-, indicating that they were pre-nasalized stops, mb-, nd-, ŋɡ- (e.g. měi 每 bhe < mwoj, ní 泥 dhe < nej, wǔ 五 ɡho < ŋuo). For the finals, the Tang-Wudai dialect had the following characteristics: 1. In the Qiān Zì Wén 千字文 ‘Thousand Character Essay’ the velar nasal coda -ŋ was dropped after vowels -ɑ, -a, -ɛ (e.g. bàng 傍 bo < bɑŋ, chuáng 4

Fricatives were also affected, but because fricatives only had a two-way phonation type contrast (voiced and unvoiced), devoicing resulted in unvoiced fricatives without any tonal split.

166

Middle Chinese

床 tʃho < dʐiɐŋ, bīng 兵 pe < pɰiaŋ, qīng 清 tsye < tshjɛŋ, xīng 星 sje < seŋ, héng 橫 hwe’e < ɦwɰaŋ), but kept after vowels u, o, ə (e.g. dōng 東 toŋ < tuŋ, zēng 增 tseŋ < tsəŋ, zhōng 鍾 tʃuŋ < tɕioŋ). 2. Several transcriptions show the drop of stop codas, -p, -t, -k (e.g. shì 釋 ʃi < ɕjɛk, méi 沒 ma < mwot, xuē 薛 se < sjɛt). All these characteristics can be found in the modern northwest dialects. 4.3

Northern Song (960–1127)

In the early Northern Song dynasty, the rhyming standard did not change much, still following the Qièyùn 切韻 tradition in its revisions, the well-known Guǎngyùn 廣韻 of 1008 and the Jíyùn 集韻 of 1039, as well as its stripped-down versions such as the Lǐbù Yùnlüè 禮部韻略. But the more colloquial-based phonological information shows innovative features that are the basic characteristics of the Old Mandarin of the Yuan dynasty and even of Modern Mandarin. One of the clearest indications of these changes is found in Shào Yōng’s 邵雍 Shēngyīn Chànghè Tú 聲音唱和圖. 4.3.1

Guǎngyùn 廣韻 (1008)

The Guǎngyùn is one of the best-known rhyme dictionaries, compiled in the Northern Song dynasty at the order of the emperor Zhēnzōng 真宗. Published in 1008, its chief editors were Chén Péngnián 陳彭年 and Qiū Yōng 邱雍 (see Figure 4.1). Based on the Qièyùn system, the Guǎngyùn was used as the main reference for Middle Chinese phonology because of its availability; the influential reconstruction of Middle Chinese by Karlgren, for example, is based on the Guǎngyùn. Like the Qièyùn, the Guǎngyùn is also divided into five volumes: two volumes for rhymes of the píng 平 tone and three volumes for syllables of the shǎng 上, qù 去, and rù 入 tones. There are 57, 55, 60, and 34 rhymes for the four tones respectively: together 206 rhymes with 26,194 character entries. Because of its official status, the Guǎngyùn was the standard reference for poetry composition, especially for the imperial examinations. However, its value for studying the innovative characteristics in the history of Chinese phonology is quite limited because its phonological framework is merely a reduplication of the Qièyùn. It is obvious that the Guǎngyùn does not reflect the phonology of its time. Thus, after the discovery of earlier revisions of the Qièyùn, modern scholars have shifted their focus to Wáng Rénxù’s 王仁昫 Kānmiù Bǔquē Qièyùn 刊謬補缺切韻 of 706 as their main reference for Middle Chinese phonology.

The Tang and Song Dynasties

167

Figure 4.1 A page from the Ze-Cun-Tang version (澤存堂本 zécúntáng běn) of the Guǎngyùn 廣韻. It reads vertically from right to left. The character dōng 東 indicates the beginning of the first homophonic group of the dōng rhyme 東韻. At the end of the glossary of the character dōng 東, there are two pieces of important information, the fǎnqiè and the number of homophonic characters, 德紅切 十七 (dé-hóng qiè, shíqī ‘dé-hóng cut, seventeen’).

168

Middle Chinese

4.3.2

Jíyùn 集韻 (1039) and Lǐbù Yùnlüè 禮部韻略 (1037)

The Jíyùn, an expansion and revision of the Guǎngyùn published in 1039, is also an official publication as a national standard of the Song dynasty, at the order of the Rénzōng 仁宗 Emperor of Song. The Jíyùn has 53,525 character entries (32,381 without counting repeated characters in different rhymes). The number of character entries is almost twice that of the Guǎngyùn. Dīng Dù 丁 度 was the chief editor of both the Jíyùn and a simplified version entitled Lǐbù Yùnlüè, which had been published two years earlier in 1037. Because of its large volume (ten volumes), the number of copies printed and the circulation of the Jíyùn were quite limited. Like the Guǎngyùn, the Jíyùn inherited the phonological framework of the Qièyùn. Its number of rhymes is exactly the same as the Guǎngyùn, however some of the newly devised fǎnqiè 反切 spellings reflect contemporary pronunciation. For example, characters with labiodental nasals are used to replace the upper spellers of labiodental characters that once had bilabial nasals as initials. In the Yùnlì 韻例 ‘Examples of Rhymes’ in the preface of the Jíyùn, the problem of using wǔ 武 (ɱ-) for mǒu 某 (m-) and wáng 亡 (ɱ-) for máng 茫 (m-) is mentioned along with the change of fǎnqiè spellings. The Yùnlüè’s publication was mainly for the imperial examination, which included poetry composition. There are different versions of the Yùnlüè. The Yùnlüè published in the Jingde reign (1004–1007) is referred to as the Jǐngdé Yùnlüè and is a simplified version of the Guǎngyùn. About thirty years later, in 1037, at the order of the Emperor Rénzōng, the authors of the Jíyùn, Dīng Dù et al. revised the Jǐngdé Yùnlüè and published it as the Lǐbù Yùnlüè. It contains 9,590 character entries, which are the most frequently used characters at the time. The number of characters is much less than in the Guǎngyùn (26,194) and the Jíyùn (53,525/32,381). This literary standard coexisted along with the imperial examination system. Revised versions of the Yùnlüè published include: the Xīnkān Lǐbù Yùnlüè 新刊禮部韻略 ‘The new version of the Lǐbù Yùnlüè’ by Wáng Wényù 王文郁, published in 1229 in the territory of the Jin dynasty; Liú Yuān’s 劉淵 Rénzǐ Xīnkān Lǐbù Yùnlüè 壬子新刊禮部韻略 ‘The new version of the Lǐbù Yùnlüè of the year of Renzi’ of the Yuan dynasty (refer to Chapter 7); the Shīyùn Jílüè 詩韻輯略 of the Ming dynasty; and the Pèiwén Shīyùn 佩文詩韻 of the Qing dynasty. All inherited the same phonological system as the literary standard for imperial examinations. Because of this literary tradition, its phonological system, in terms of 206 rhymes, has been fixed in number since the eleventh century.

The Tang and Song Dynasties

4.3.3

169

Shēngyīn Chànghè Tú 聲音唱和圖

The Shēngyīn Chànghè Tú is the set of phonological tables in Shào Yōng’s 邵 雍 (1011–1077) Huángjí Jīngshì Shū 皇極經世書, a mysterious book with a purpose unrelated to phonology. In these tables he tried to use phonological categories to explain his philosophical ideas. Because of its innovative phonological features, the Shēngyīn Chànghè Tú has attracted the interest of a number of scholars (Zhōu Zǔmó 1942; Lù 1946b; Lǐ Róng 1956; Jaxontov 1980). Shào Yōng’s home town was Fanyang (modern Zhuozhou, Hebei, which borders Beijing in the southwest). He moved to Luoyang when he was thirty years old. Zhōu Zǔmó (1942) identifies the base dialect of the Shēngyīn Chànghè Tú as Luoyang, but Jaxontov (1980) points out that the reflexes of Middle Chinese -k syllables actually indicate that it is the Beijing dialect. Jaxontov (1980) also makes it clear that the northern dialects are not much different from each other. Although the phonological information of the Shēngyīn Chànghè Tú is incomplete, this quasi-phonological work provides many important pieces of information revealing the characteristics of northern Chinese, which later developed into Mandarin. All phonological information is provided in two sets of tables: the Shēng Tú 聲圖, which lists initials, and the Yīn Tú 音圖, which lists finals. 4.3.3.1 Initials The initials are listed in the twelve Yīn Tú tables. The syllables contained in the Yīn Tú tables are arranged into four columns. These four columns have a strong connection to the four děng commonly seen in the rhyme tables. The contrast between odd and even rows is marked as qīng and zhuó respectively (refer to Section 1.4.2). Traditionally qīng is the label for voiceless initials and zhuó for voiced initials. According to Zhào (1957) and Jaxontov (1980), the initials of row one and two on the same table are identical and the initials of row three and four are also identical, although Zhōu Zǔmó (1942) has a different interpretation. The key difference is how to understand the arrangement of the sonorants. For example, in table 2 of Table 4.18, syllables with a Middle Chinese velar nasal are listed in two rows, qīng and zhuó. Zhōu Zǔmó treats this difference as the difference of initials and Sergej Jaxontov treats the difference as tones instead. The initial system has twenty-one initials (see Table 4.19). Two characters from the Yīn Tú tables are used to represent each initial. Such an initial system is very similar to that of the Yuan dynasty’s Zhōngyuán Yīnyùn 中原音韻 and even Modern Mandarin. It is quite remarkable that this Mandarin-type initial system was already well established in the eleventh century. The following characteristics are the significant changes in comparison with the Middle Chinese system.

170

Middle Chinese

Table 4.18 The twelve Yīn Tú tables, with their respective reconstructed initials for each row’s gaps Table 1 一音 古甲九癸 k□□近揆 k坤巧丘棄弃 kh□□乾虯 kh-

Table 2 二音 黑花香血 x黃華雄賢 x五瓦仰□ ŋ吾牙月堯 ŋ-

Table 3 三音 安亞乙一 Ø□爻王寅 Ø母馬美米 m目皃眉民 m-

Table 4 四音 夫法□飛 父凡□吠 武晚□尾 文万□未

Table 5 五音 卜百丙必 p步白 鼻 p普朴品匹 ph旁排平瓶 ph-

Table 6 六音 東丹帝 兌大弟 土貪天 同覃田

Table 7 七音 乃妳女 內南年 老冷呂 鹿犖离

Table 8 八音 走哉足 自在匠 草采七 曹才全

Table 9 九音 思三星 寺□象

Table 10 十音 山手 士石 □耳 □二

Table 11 十一音 莊震 乍□ 叉赤 崇辰

Table 12 十二音 卓中 宅直 坼丑 茶呈

■ t■ t■ th■ th-

■ ■ ■ ■

■ s■ s□□□■ □□□■

■ ʂ■ ʂ■ ɻ■ ɻ-

■ ts■ ts■ tsh■ tsh-

■ n■ n■ l■ l-

■ ■ ■ ■

ffʋʋ-

■ tʂ■ tʂ■ tʂh■ tʂh-

■ ■ ■ ■

■ tʂ■ tʂ■ tʂh■ tʂh-

Note: The empty squares, □, are phonologically possible syllables that lack a character with said final, while the filled squares, , are the impossible syllables, or phonological gaps.



Table 4.19 The twenty-one initials of the Shēngyīn Chànghè Tú Labials Alveolars Retroflexes Velars a b

卜步 p東兑 t走自 ts莊乍 tʂ-a 古近 k-

普旁 ph土同 th草曹 tsh叉崇 tʂh-b 坤乾 kh-

母目 m乃內 n五吾 ŋ-

夫父 f山士 ʂ黑黃 x-

武文 ʋ老鹿 l思寺 s耳二 ɻ(安爻Ø-)

卓宅 is also used for tʂ-. 坼茶 is also used for tʂh-.

4.3.3.1.1 Devoicing of Voiced Obstruents The arrangement of Shào Yōng’s tables provides strong evidence suggesting the devoicing of Middle Chinese voiced obstruents: stops, affricates, and fricatives. In the Shēngyīn Chànghè Tú it seems that the voiced obstruents were still in existence. But further analysis shows that the voiced obstruents had lost their voicing and become voiceless. In Table 4.20, the stops and affricates of tables 1, 5, 6, 8, 11, and 12 (from Table 4.18) are arranged in four rows, potentially due to the influence of Indic

The Tang and Song Dynasties

171

Table 4.20 Shào Yōng’s assignment of bh- initials resembles Indic phonology

卜百丙必 步白 鼻 普朴品匹 旁排平瓶

Shào’s arrangement

Middle Chinese values

p b ph bh

p b ph b

> > > >

Real pronunciation

Modern Mandarin

p p ph ph

p p ph ph

Table 4.21 The relationship between voiced píng and voiced zè syllables p ph

卜百丙必 普朴品匹

= =

b>p b > ph

步白 鼻 旁排平瓶

Table 4

t th

東丹帝 土貪天

= =

d>t d > th

兌大弟 同覃田

Table 6

ts tsh

走哉足 草采七

= =

dz > ts dz > tsh

自在匠 曹才全

Table 8

tʂ tʂh

莊震/卓中 叉赤/坼丑

= =

dʐ > tʂ dʐ > tʂh

乍/宅直 崇辰/茶呈

Tables 11, 12

k kh

古甲九癸 坤巧丘弃

= =

ɡ>k ɡ > kh

近揆 乾虯

Table 1

phonology. But the Middle Chinese phonology only has a three-way distinction of stops and affricates. This four-way distinction actually reveals a twoway distinction, the result of the devoicing of Middle Chinese obstruent initials. 旁排平瓶, which Shào assigns as bh- initials are Middle Chinese píng tone syllables (並平), and 步白備鼻, which Shào assigns as b- initials, are Middle Chinese zè (non-píng) (並仄) tone syllables. This split is identical to the one seen in the Ěryǎ Yīntú (Section 5.1.7). This arrangement with bh- initials actually reveals that the voiced stops in píng tone syllables changed to voiceless aspirated stops but in zè tone changed to voiceless unaspirated stops; if there was no devoicing, there would be no aspiration distinction of the voiced stops (Lǐ Róng [1956] thinks they were still voiced, and others [Lù 1946b; Zhào 1957; Zhōu Zǔmó 1942; Jaxontov 1980] all agree they are voiceless). Table 4.21 is a more specific look at the relationship between voiced píng and voiced zè syllables as they merge with unvoiced syllables. This devoicing pattern can be firmly supported by evidence from the Khitan materials because the nature of the Khitan Lesser script is phonetic (more details on this in Chapter 5.1).

172

Middle Chinese

This devoicing pattern is unquestionably Mandarin and still used as one of the diagnostic features for Mandarin in dialect classification. 4.3.3.1.2 Labiodentals In the fourth table of Table 4.18, fū-fǎ-fēi 夫法飛 (Middle Chinese: pjo-piɐppwɨj) are listed in contrast with fù-fán-fèi 父凡吠 (Middle Chinese: bjo-bjɐmbwjɐj). According to the Middle Chinese pronunciation, fū-fǎ-fēi 夫法飛 are voiceless (p-) and fù-fán-fèi 父凡吠 are voiced (b-). Further examination shows that these two sets of characters have a complementary distribution of tones (see Table 4.22), and so the voicing feature was lost in the system; this way, there should be no voicing contrast between any pair of consonants Shào Yōng used tonal contrast for his qīng/zhuó contrast (Jaxontov 1980). The initial of both groups should be f-. The contrast of wǔ-wǎn-wěi 武晚尾 (Middle Chinese: mio-mwiɐn-mwɨj) and wén-wàn-wèi 文万未 (Middle Chinese: mjun-mwjɐn-mwɨj) in the same table is also the contrast of tones, the same tonal contrast as other sonorants. All these syllables are Middle Chinese syllables with the wēi 微 initial (m > ɱ). The initial of both groups should be ʋ-, an approximant counterpart of labiodental fricative f-, rather than its voiced counterpart. This initial is not v-, a voiced counterpart of f-, because from the system we know that Middle Chinese v-, the fèng 奉 initial, had changed to f-. It would be a mistake to reconstruct the Middle Chinese wēi 微 initial as a voiced v- in the system. Table 4.23 shows the distribution of labiodentals in the Shēngyīn Chànghè Tú, categorized by Middle Chinese initial and tone. Table 4.22 The complementary distribution of labiodentals across tones yīnping

yángpíng

shǎng





夫飛 —

— 凡

— 父

— 吠

法 —

Table 4.23 The distribution of labiodentals in the Shēngyīn Chànghè Tú, categorized by Middle Chinese initial and tone píng pphbm-

shǎng



夫1飛1 凡2 文4

rù 法1

父2 武3晚3尾3

吠2 万4未4

Note: Subscript numbers represent the row they occupy in the Shēngyīn Chànghè Tú.

The Tang and Song Dynasties

173

Table 4.24 A comparison of different texts’ reconstructions of the bilabial and labiodental series, showcasing the evolution of the series from Middle Chinese to Modern Mandarin

幫 滂 並

bāng pāng bìng

明 非 敷 奉 微

míng fēi fū fèng wēi

Qièyùn

Qīyīn Lüèa

Yíqiè Jīng Yīnyì

Shēngyīn Chànghè Tú

Modern Mandarin

p ph b píng b zè m pj phj bj mj

p ph b píng b zè m pf pfh bv ɱ

p ph b píng b zè m f f v ɱ

p ph ph p m f f f ʋ

p ph ph p m f f f w

a

The Qīyīn Lüè 七音略 was written sometime before 1161, and although considered a Song dynasty book, its conservative nature puts its phonology somewhere between the Qièyùn and Yīqiè Jīng Yīnyì.

This system is very close to the labial system of Modern Mandarin. The only difference is the identity of the wēi 微 initial, which in the system of the Shēngyīn Chànghè Tú still contrasts with the hékǒu yǐng 影 Ø- and yí 疑 ŋ- initials. In Mandarin these merge into a zero initial with -w- medial (see Section 9.6.1). In Table 4.24, the labial series of the Qièyùn (Section 3.5.2), Qīyīn Lüè (Section 1.4.1), Yīqiè Jīng Yīnyì (Section 4.1.5), Shēngyīn Chànghè Tú (Section 4.3.3), and Modern Mandarin are compared. 4.3.3.1.3 The Status of Middle Chinese zhī 知 and zhào 照 Initials The only contrast that is found in the Shēngyīn Chànghè Tú but not found in the Zhōngyuán Yīnyùn system is the contrast between Middle Chinese zhī 知 and zhào 照 initials. They are listed in tables 11 and 12 of Table 4.18, respectively, but if this indicates that the merger of zhī 知 and zhào 照 initials did not happen by the eleventh century, then the phonetic values of these two groups are difficult to reconstruct. Zhōu Zǔmó (1942) did not distinguish these two groups based on an observation that table 12 is listed after table 11, the retroflex affricates, but not after table 6, alveolar stops. His reasoning is that if the zhī 知 initials are still retroflex stops (ʈ, ʈh, ɖ) and distinct from the zhào 照 initials (tʂ, tʂh, dʐ) they should be listed after table 6 of alveolar stops (t, th, d). Since the zhī 知 initials are listed after table 11, the retroflex affricates, they have already changed to affricates.

174

Middle Chinese

4.3.3.1.4 The Loss of the Nasal Quality of the Middle Chinese rì 日 Initial In Table 4.18, table 10, rows three and four list ěr 耳 and èr 二. The initials of both syllables are Middle Chinese rì 日 initial, which is a palatal nasal ȵ-. Since both are listed in the same table as the counterpart of the retroflex fricative ʂ-, this indicates that the nasal quality of the rì 日 initial had been lost and had become a retroflex approximant ɻ-. Along with the Khitan materials (Section 5.1) it is the earliest indication of the changed phonetic value of the Middle Chinese rì 日 initial. Since this phonetic value is identical with reflexes of the rì 日 initial in Modern Mandarin, it provides critical information for reconstruction of the rì 日 initial in various stages of Mandarin history. 4.3.3.2 Finals The finals are listed in the Shēng Tú tables (see Table 4.25). There are ten tables, but the last three have no characters and are thus omitted here. In each table, two rows form a group. The first group is labeled pì 闢 and the second is labeled xī 翕. The pì and xī generally correspond to the kāi 開 and hé 合 of traditional terminology. The four columns are for the four tones, píng, shǎng, qù, and rù. According to the phonological categories of the characters used, each line could have more than one final. For the line with two finals, the arrangement in general is that the one with a palatal feature precedes its counterpart without a palatal feature. The phonetic values of the finals in tables 6 and 7 seem not to follow the general pattern in the other tables.

Table 4.25 The seven non-empty Shēng Tú tables, with their respective reconstructed finals for each row Table 1 一聲 多可个舌 -a, -ja 禾火化八 -wa 開宰愛○ -aj 回每退○ -waj

Table 2 二聲 良兩向○ 光廣況○ 丁井亘○ 兄永瑩○

Table 5 五聲 妻子四日 -i, -ɨ 衰○帥骨 -wi, -wə ○○○德 -əj 龜水貴北 -wəj

Table 6 六聲 宫孔衆○ 龍甬用○ 魚鼠去○ 烏虎兔○

-aŋ, -jaŋ -waŋ -əŋ, -iŋ -wəŋ,-ɥəŋ

Table 3 三聲 千典旦○ 元犬半○ 臣引艮○ 君允巽○

-uŋ -yŋ -y -u

Table 7 七聲 心審禁○ -im ○○○十 –i(p) 男坎欠○ -am, -jɛm ○○○妾 -jɛ(p)

Source: Based on Zhōu Zǔmó 1942, and modified.

-an, -jɛn -wan, -ɥɛn -ən, -in –uən, -yn,

Table 4 四聲 刀早孝岳 -aw, -jaw 毛寶報霍 -waw 牛斗奏六 -əw, -jəw ○○○玉 -ɥəw

The Tang and Song Dynasties

175

4.3.3.2.1 Loss of Stop Codas The Middle Chinese rù syllables show the loss of their stop codas. Syllables with -t coda are listed with monophthong syllables (Table 4.25, tables 1 and 5). Interestingly, syllables with -k coda are listed with diphthong syllables (tables 4 and 5), while syllables with -p coda are listed by themselves in the same table for the syllables with the -m coda (table 7), with the second and fourth row belonging to rùshēng yùn 入聲韻, and thus exclusive to the rù tone. The syllables that originally have -k coda, yuè 岳, huò 霍, liù 六, yù 玉, dé 德, and běi 北, are listed with diphthong syllables instead of with the corresponding syllables with -ŋ coda in tables 2 and 6 (see Table 4.25). By looking at the Middle Chinese phonetic values of the syllables in table 4, we can determine that -k coda syllables begin to merge with -w coda syllables around the Song dynasty. The endings of the rù syllables have changed from -k to -w since the syllables in the same row have the same main vowel and ending (see Table 4.26). Syllables with -k coda and a back vowel have lost their -k coda and acquired -w coda. Their finals became diphthongs. These are among the earliest examples of this phonological change in northern Chinese. Syllables with -k coda and a non-back vowel have lost their -k coda and acquired a -j coda. The endings of the rù syllables have changed from -k to -j since the syllables in the same row have the same main vowel and coda (see Table 4.27). The Middle Chinese syllables with -k actually had the same phonological change, -Vk > VG (G = w, j). The off-glide of the diphthongization is determined by the main vowel in the Middle Chinese system. If the main vowel is a back vowel, the coda is -w; but if the main vowel is a non-back vowel, the coda is -j (see Table 4.28). Table 4.26 The merger of syllables with -k coda and a back vowel with syllables with -w coda Table 4

-(j)aw -(w)aw -(j)əw -jəw

刀 tɑw 毛 mɑw 牛ŋiw ○

早 tsɑw 寶 pɑw 斗 təw ○

孝 hɰaw 報 pɑw 奏 tsəw ○

岳 霍 六 玉

ŋɰɔk hwɑk ljuk ŋjok

Table 4.27 The merger of syllables with -k coda and a non-back vowel with syllables with -j coda Table 4

-əj -uj

○ 龜 kwɰi

○ 水 ɕwi

○ 貴 kwɨj

德 tək 北 pək

176

Middle Chinese

Table 4.28 The phonological conditions for the reflexes with -w and -j coda Middle Chinese

Condition

Song Chinese

-k

following back vowel

-w

following non-back vowel

-j

Example 岳 霍 六 玉 德 北

ŋɰɔk hwɑk ljuk ŋjok tək pək

> > > > > >

-w -w -w -w -j -j

yuè ‘mountain’ huò ‘suddenly’ liù ‘six’ yù ‘jade’ dé ‘morality’ běi ‘north’

These changes are consistent with later rhyme tables such as the Qièyùn Zhǐzhǎng Tú 切韻指掌圖 and Sìshēng Děngzǐ 四聲等子, as well as rhyme books such as the Ménggǔ Zìyùn 蒙古字韻 and the Zhōngyuán Yīnyùn. According to the distribution of these syllables in the Modern Mandarin dialects, the -Vw forms are restricted in Northern Mandarin, including the Beijing dialect, but the -Vj forms can also be found further south in Zhongyuan Mandarin, including in the Luoyang dialect. The syllables with -t coda are listed with the syllables with no nasal coda in tables 1 and 5 of Table 4.25 instead of the syllables with coda -n in table 3. The first and second row of table 1 and the first row of table 5 are monophthongs. The arrangement of the first and the second row of table 5 is Middle Chinese i/it and wi/wit respectively. The characters qī 妻, zǐ 子, and sì 四 are the Middle Chinese kāikǒu syllables of the xiè 蟹 and zhǐ 止 rhyme groups, and shuāi 衰 and shuài 帥 are hékǒu syllables with retroflex initials from the zhǐ 止 rhyme group. The finals of the hékǒu syllables have not changed into the diphthong -aj as in the Ménggǔ Zìyùn and the Zhōngyuán Yīnyùn of the fourteenth century. Thus, it seems that these two syllables are listed because they had a different quality to the main vowel, which is probably lower and/or more centralized than other syllables from the same rhyme (see Table 4.29). The syllables with -p coda are listed in table 7 of Table 4.25. But the two syllables shí 十 and qiè 妾 are not listed in the same row for the syllables with -m coda. According to Middle Chinese phonology, shí 十, xīn 心, shěn 審, and Table 4.29 The main vowel of shuāi 衰 and shuài 帥 in Song-era Chinese Late Middle Chinese 衰帥 誰水

ʂɕ-

-wi -wi

Song > >

-wɨj -wi

Modern Mandarin > >

-waj -wej

shuāi ‘decline,’ shuài ‘commander’ shuí ‘who,’ shuǐ ‘water’

The Tang and Song Dynasties

177

jìn 禁 belong to the same rhyme group (even, in fact, the same rhyme), while qiè 妾, nán 男, kǎn 坎, and qiàn 欠 belong to the same rhyme group. It is not clear why shí 十 and qiè 妾 are listed in different rows alone. Nevertheless, phonologists (Zhōu Zǔmó 1942; Lù 1946b; Jaxontov 1980) treat this phenomenon as the existence of the -p coda. 4.3.3.2.2

Simplification of the Vowel System: The Merger of the Middle Chinese zēng 曾 and gěng 梗 Rhyme Groups The tables of Shēng Tú apparently do not list all the possible finals, but it can be observed that the system of the main vowel has been simplified. The syllables in rows 3 and 4 of table 2 of the Shēng Tú (Table 4.25) are from the Middle Chinese zēng 曾 and gěng 梗 rhyme groups. The mix of these syllables indicates that the gěng 梗 rhyme group has merged into the zēng 曾 group, or the loss of the front mid vowel -ɛ in front of -ŋ, (w)(j)ɛŋ > (j)əŋ. The syllables of the jiāng 江 rhyme group with -ŋ are not included in the tables, but the syllable with -k, yuè 岳, a Rank-II syllable, is listed in the same row as Rank-I syllables dāo 刀 and zǎo 早, as well as Rank-II syllable xiào 孝. Table 4.30 is a comparison of Middle Chinese -ŋ rhymes and those of the Shēngyīn Chànghè Tú. The Middle Chinese hékǒu syllables with medial -ware not included. If the changes of syllables with -ŋ coda and -k coda are parallel, it is likely that the Middle Chinese jiāng 寶 rhyme group had merged into the dàng 宕 rhyme group or lost the back mid vowel ɔ in front of -ŋ, ɰɔŋ > (w)(j)ɑŋ. Thus the merging of the two mid vowels -ɛ and -ɔ with other vowels is the result of this simplification (see Table 4.31).

Table 4.30 A comparison of Middle Chinese -ŋ rhymes and those of the Shēngyīn Chànghè Tú Middle Chinese rhyme groups with –ŋ Zēng 曾 rhyme group Gěng 梗 rhyme group Jiāng 江 rhyme group Dàng 宕 rhyme group Dōng 東 rhyme group

əŋ ɛŋ ɰɔŋ aŋ uŋ

Shēngyīn Chànghè Tú iŋ jɛŋ jaŋ juŋ

əŋ əŋ – aŋ uŋ

iŋ iŋ jaŋ(waŋ) jaŋ juŋ

4.3.3.2.3 Centralized Vowels, i > ɨ In table 5 of the Shēng Tú (Table 4.25), four syllables with high unrounded vowels (qī 妻 zǐ 子 sì 四 rì 日) are listed in the same row. The final of qī 妻 should be a high front unrounded vowel, but it is not clear whether zǐ 子, sì 四, and rì 日 have a centralized vowel or not. However, table 9 of the Yīn Tú

178

Middle Chinese

Table 4.31 The simplification of rhymes with -ŋ codas, resulting in a four-way contrast -ŋ

i ɛ

ə a

u ɔ

i >

u ə a

(Table 4.18) reveals the existence of a centralized vowel. In this table, sī 思 and sì 寺 are listed in the first row, which is for Division-I syllables. Both are Middle Chinese zhī 之 rhyme syllables belonging to Rank-III. This arrangement indicates the loss of the front feature of the vowel, which is the definition of Rank-III. A very similar phenomenon can be observed later in the Qièyùn Zhǐzhǎng Tú, a rhyme table produced in the late twelfth century. In its tables, syllables zī 兹, cí 雌, cí 慈, sī 思, and cí 詞 are listed in the positions for Division-I syllables. These syllables with a centralized vowel are later listed in the zhīsī 支思 rhyme in the Zhōngyuán Yīnyùn. Table 4.32 is a final chart of the Shēngyīn Chànghè Tú, with the inclusion of the central high vowel -ɨ. Table 4.32 A final chart of the Shēngyīn Chànghè Tú Medials

j, w, ɥ

Main vowels

-Ø: -n: -j:

Codas

-m, -n, -ŋ, -j, -w, -p

a, o, i, y, u, ɨ a (ɛ), ə, i, y, u a, ə

-m: -ŋ: -w:

a (ɛ), i a, ə, i, u, y a, ə

4.3.3.3 Tones The four tones are distinguished in the Shēng Tú (Table 4.25). Shǎng tone syllables with voiced obstruents changed to qù tone. Píng, qù, and rù tones have upper and lower registers, which are traditionally called yīn 陰 and yáng 陽, respectively. Such a system is the same as that of the modern Hangzhou dialect, which was strongly influenced by the Bian-Luo dialect at the end of the Northern Song dynasty (960–1127) and quite distinct from adjacent Wu dialects (Norman 1997). In the Wu dialects the shǎng syllables with sonorants usually have a lower register tone, whereas Jaxontov (1980) points out that the Middle Chinese shǎng tone syllables with sonorant initials were in high register (qīng) tone in the Shēngyīn Chànghè Tú like that of the modern Hangzhou dialect. The syllables arranged in the qīng row are all syllables of shǎng tone (五瓦 仰□, row 3 of table 2 (in Table 4.18); 母馬美米, row 3 of table 3; 武晚□尾,

The Tang and Song Dynasties

179

row 3 of table 4; 乃妳女■, 老冷呂■, rows 1 and 3 of table 7; and ■□耳■, row 3 of table 10) but the syllables in the zhuó row are all syllables with nonshǎng tones (吾牙月堯, row 4 of table 2; 目皃眉民, row 4 of table 3; 文 万□未, row 4 of table 4; 內南年■, 鹿犖离■, rows 2 and 4 of table 7, and ■□二■, row 4 of table 10). Thus, there must be a tonal difference. The shǎng tone was a high tonal value or a high register tone. So the sonorants of the syllables with high tone were perceived as qīng (high register tone), and in other tones perceived as zhuó (low register tone). This is clearly a very typical characteristic of Mandarin phonology. In other words, shǎng tone Middle Chinese syllables with sonorants and voiceless obstruents become Mandarin third tone, but those with voiced obstruents become fourth tone in modern Beijing Mandarin. Table 4.33 shows the tonal reflexes of Middle Chinese tone II (shǎng) syllables according to different types of initials as shown in numeral notation (see Table 4.34 or Section 1.4.3 for conversion to written tones). The bolded numbers in Table 4.33 indicate where the tones differ from the original Middle Chinese. Table 4.35 is a correspondence table of three syllables, 短 duǎn ‘short,’ 巨 jù ‘huge,’ and 滿 mǎn ‘full,’ between Middle Chinese, the Chinese of the Shēngyīn Chànghè Tú, and Modern Mandarin. 4.3.4

Summary

There is no question that the Shēngyīn Chànghè Tú displays a number of phonological characters that seem to be “Mandarin.” Although it is still not Table 4.33 The tonal reflexes of Middle Chinese tone II syllables according to different types of initials

Voiceless obstruent Voiced obstruent Sonorant

Middle Chinese

Wu

Chànghè Tú

Hangzhou

Beijing

IIa IIb IIb

IIa IIb IIb

IIa IIIb IIa

IIa IIIb IIa

II III II

Table 4.34 The tonal reflexes of Middle Chinese tone II syllables according to different types of initials (tones written in full)

Voiceless obstruent Voiced obstruent Sonorant

Middle Chinese

Wu

Chànghè Tú

Hangzhou

Beijing

yīnshǎng yángshǎng yángshǎng

yīnshǎng yángshǎng yángshǎng

yīnshǎng yángqù yīnshǎng

yīnshǎng yángqù yīnshǎng

shǎng qù shǎng

180

Middle Chinese

Table 4.35 A correspondence table between Middle Chinese, the Chinese of the Shēngyīn Chànghè Tú, and Modern Mandarin

MC

Condition

Shēngyīn Chànghè Tú

Example

shǎng tone

voiceless obstruent initial voiced obstruent initial sonorant initial

yīnshǎng yángqù yīnshǎng

短 巨 滿

twɑn (shǎng) ɡiɔ (shǎng) mwɑn (shǎng)

MC

Condition

Beijing

Example

shǎng tone

voiceless obstruent initial voiced obstruent initial sonorant initial

shǎng qù shǎng

短 巨 滿

twɑn (shǎng) ɡjɔ (shǎng) mwɑn (shǎng)

> > >

yīnshǎng yángqù yīnshǎng

> > >

duǎna jù mǎn

Note: MC = Middle Chinese. duǎn 短 and mǎn 滿 are considered “third” tone in Mandarin, and jù 巨 is considered “fourth”; this Mandarin numbering system does not line up with the one employed in Middle Chinese phonology. This confusion is one reason why this book prefers the written-out format of tones, and is explained further in Section 1.4.3.

a

clear whether the Shēngyīn Chànghè Tú is a description of Shào Yōng’s home dialect, the Luoyang dialect, or a system including the characteristics of both, it is clear that the basic phonological characteristics of Modern Mandarin had already emerged. These characteristics include the Mandarin devoicing pattern of Middle Chinese obstruents, the loss (or partial loss) of Middle Chinese consonant codas, vowel centralization (vowel apicalization), Middle Chinese shǎng tone syllables with voiced obstruents merging with qù tone syllables, as well as the most typical characteristic of Northern Mandarin: the diphthongization of the Middle Chinese -k syllables. 4.4

Southern Song (1127–1297)

Because of military pressure from the Jurchens to the north, the Song dynasty retreated to the south of the Huai River and moved its capital from Bianliang (Biànliáng 汴梁) (modern Kaifeng, in Henan province) to Lin’an (Lín’ān 臨 安) (modern Hangzhou, in Zhejiang province). Because of this migration the modern Hangzhou dialect is clearly distinct from the surrounding Wu dialects. Such linguistic difference was originally brought to Hangzhou by immigrants from the capital of Northern Song. The special phonological features of the modern Hangzhou dialect can provide us with important information about the phonology of Kaifeng, the standard of the Southern Song. Unlike their earlier counterparts, the three rhyme tables of the Sìshēng Děngzǐ 四聲等子, the Qièyùn Zhǐzhǎngtú 切韻指掌圖, and the Qièyùn

The Tang and Song Dynasties

181

Zhǐnán 切韻指南, works that were contemporaries of the Southern Song Hangzhou dialect, reflected the phonological features of the time when they were produced, although they were still very much based on the framework of the previous rhyme tables. Based on the placements of the rhymes, some innovative phonological features can be observed. In general, the Sìshēng Děngzǐ, the Qièyùn Zhǐzhǎngtú, and the Qièyùn Zhǐnán are more similar to each other in design than earlier rhyme tables such as the Yùnjìng 韻鏡 and the Qīyīn Lüè 七音略.5 A common phenomenon of these three rhyme tables is that the Middle Chinese rùshēng syllables are not only listed along with the syllables with their counterparts with nasal endings, they are also listed with the syllables with no nasal coda. Such a rùshēng syllable can appear in more than one table. This phenomenon should be interpreted as the loss of the final consonant -p, -t, -k of the rùshēng syllables, while maintaining their rù tone, segmentally becoming the same or similar to the syllables without nasal endings. For example, the Middle Chinese rùshēng syllables with the -k coda are listed along with the rhymes with -w coda. This actually indicates a change of -ak > -aw. This change is an unambiguous feature of Northern Mandarin (refer to Section 6.1.1). Due to the design of certain fixed arrangements, the rhyme tables are unable to reveal certain sound and categorical changes, such as the change of the initials, the change of divisions, and the change of tones of the syllables. 4.4.1

Hangzhou Dialect

Some of the special phonological features of the Hangzhou dialect are recorded in Láng Yīng’s 郎瑛 (1487–1566) Qīxiū Lèigǎo 七修類稿. In an entry about the Hángyīn 杭音 ‘Hang(zhou) pronunciation,’ few phonological features of the Hangzhou dialect were described. As a general comment, Láng Yīng said that, “(The pronunciation of Hangzhou) today is still quite similar to the pronunciation of Bian(liang).” Then he provided a few examples: “Yù 玉 is pronounced as yù 御, yīsā 一撒 is pronounced as yǐsā 倚撒, bǎilíngxiāng 百零 香 is pronounced as bǎilíngxiāng 擺零香; these are all the pronunciation of Bian(liang).”6 Based on these three homophonic pairs, a number of phonological features of the late Northern Song / early Southern Song can be learned. In Table 4.36, Láng Yīng’s referenced characters are provided with their

5

6

The Qièyùn Zhǐnán was actually published in the Yuan dynasty; however, because of its conservative nature, and because of its similarities to the Sìshēng Děngzǐ and Qièyùn Zhǐzhǎngtú, it is placed in the Southern Song dynasty section. “卷二十六 · 辯證類杭音: 城中語音,好于他郡,蓋初皆汴人,扈宋南渡,遂家焉。故至今 與汴音頗相似,如呼玉為玉(音御) ,呼一撒為一(音倚)撒,呼百零香為百(音擺)零 香,兹皆汴音也。”

182

Middle Chinese

Table 4.36 Láng Yīng’s referenced characters

玉 御 一 倚 百 擺

Middle Chinese

Modern Mandarin

ŋjok ŋiɔ Øit Øɰjɛ pɰak pɰæ

yù yù yī yǐ bǎi bǎi

[y51] [y51] [i55] [i214] [paj214] [paj214]

‘jade’ ‘drive’ ‘one’ ‘rely on’ ‘hundred’ ‘place, display’

Middle Chinese phonetic values, as well as Modern Mandarin pronunciations for comparison. All three Middle Chinese rù tone syllables yù 玉, yī 一, and bǎi 百 are pronounced the same as the non-rù syllables yù 御, yǐ 倚, and bǎi 擺. This indicates that (1) the consonant codas of these syllables were lost, -t, -k > -Ø, and (2) the rù syllables merged with non-rù syllables conditioned by initials, with voiceless initials > shǎng, and with sonorant initials > qù. These tonal changes are exactly the same as those recorded in the Zhōngyuán Yīnyùn 中原 音韻 (Section 7.4.4) and the Shēngyīn Chànghè Tú 聲音唱和圖 (Section 5.3.3). The population of Hangzhou increased dramatically as the result of migration when the Southern Song moved its capital there. These people also brought the language from the old capital, Bianliang, with them to Hangzhou. Among the modern Wu dialects, a number of phonological characteristics are unique to the Hangzhou dialect. These characteristics originate from the capital area of the Northern Song, therefore a comparison of the Hangzhou dialect and other Wu dialects should provide us with information about the northern dialect at the beginning of the Southern Song. Middle Chinese shǎng tone syllables with sonorant initials changed to yīnshǎng. This phenomenon was discussed in Section 4.3.3.3. Middle Chinese shǎng tone syllables with sonorants and voiced obstruents initials changed differently, a feature that can be found in the Hangzhou dialect. Middle Chinese shǎng tone syllables merge differently both historically and geographically (see Table 4.37). The similarity between the Hangzhou dialect, Mandarin, and the Shēngyīn Chànghè Tú, in contrast with other Wu dialects, shows a relationship between the dialects of the Northern and Southern Song capitals. The Hangzhou dialect lacks a dual pronunciation between báidú 白讀 (colloquial reading) and wéndú 文讀 (literary reading). Yuen Ren Chao noted that most of the characters (jiā 家, wèn 問, jiāo 交, jiāng 江, yīng 櫻, jiǎo 角, ěr 耳, etc.) that have dual pronunciation in other locations have only

The Tang and Song Dynasties

183

Table 4.37 Middle Chinese shǎng tone syllables merge differently both historically and geographically

With voiceless obstruents With sonorants With voiced obstruents

Hangzhou

Shēngyīn Chànghè Tú

Mandarin

Wu dialects

yīnshǎng yīnshǎng yángqù

yīnshǎng yīnshǎng yángqù

yīnshǎng (3) yīnshǎng (3) qù (4)

yīnshǎng yángshǎng yángqù

Table 4.38 Chao’s noted features in the Hangzhou dialect

Hangzhou Suzhou

a

b

c

d

e

江 jiāng ‘river’ tɕiaŋ kɒŋ/tɕiɒŋ

問 jiān ‘ask’ uən mən/vən

耳 ěr ‘ear’ ɚ, əl ȵi/l̩

櫻 yīng ‘cherry’ in aŋ/in

龜 kuī ‘turtle’ kuei tɕy/kuᴇ

literary readings in Hangzhou (Chao 1928: 83). The literary readings should be the phonological layer brought to Hangzhou by immigrants from the Northern Song territory, especially from its capital area of Bianliang. The characters Chao listed include the following features in its system: (a) Middle Chinese Rank-II kāikǒu syllables have a palatal glide, (b) the loss of Middle Chinese labial nasal initial ɱ-, (c) the loss of the Middle Chinese rì 日 initial, (d) the vowel raising of Middle Chinese Rank-II syllables of the gěng 梗 rhyme group, and (e) the diphthongization of Middle Chinese RankIII syllables of the xiè 蟹 and zhǐ 止 rhyme groups. Examples are given for Chao’s noted features in Table 4.38. The Suzhou dialect is used as a representative of the Wu dialects, contrasted against the Hangzhou dialect. These characteristics of Hangzhou phonology most likely come from the dialect of the capital at the end of the Northern Song dynasty. 4.4.2

The Sìshēng Děngzǐ 四聲等子

The Sìshēng Děngzǐ (see Figure 4.2) mentions 206 rhymes of the Guǎngyùn 廣 韻. It is likely that the surviving version was made after 1008, the publication date of the Guǎngyùn. However, because characters and categorical features can be replaced when revised, the original version of the Sìshēng Děngzǐ was not necessarily produced after the Guǎngyùn or the Jíyùn 集韻. The Sìshēng Děngzǐ contains twenty tables, which reflects the mergers of certain rhymes of the Qièyùn 切韻. For instance, in the Yùnjìng 韻鏡, the dōng 東 rhyme makes up one table, and the dōng 冬 and zhōng 鍾 rhymes together compose another

184

Figure 4.2 Two pages from the Sìshēng Děngzǐ, showing the merger of the syllables of the dōng 東, dōng 冬, and zhōng 鍾 rhyme series. The two leftmost columns of the second page list all the rhymes involved. The second column from the left lists the rhymes dōng 東, dǒng 董, sòng 送, wū 屋, and zhōng 鍾, zhǒng 腫, yòng 用, zhú 燭, while the leftmost column lists dōng 冬, zhǒng 腫, sòng 宋, wò 沃. In addition it also states dōng-dōng-zhōng xiāngzhù 東冬鍾相助 ‘(rhymes) dōng 東, dōng 冬, and zhōng 鍾 support each other.’ This arrangement is in contrast to the earlier rhyme tables, such as the Yùnjìng and Qīyīn Lüè, which list all three rhyme series in different tables. It is a clear indication of the merger of these rhymes.

The Tang and Song Dynasties

185

table. But in the Sìshēng Děngzǐ, dōng 東, dōng 冬, and zhōng 鍾 rhymes are combined and listed in the same table. Because these once separate groups have merged, the Sìshēng Děngzǐ is not simply an innovative arrangement of the Qièyùn or the Guǎngyùn rhymes, but rather reflects some of the phonological features of the time in addition to keeping with the Qièyùn tradition. The representative characters of the rhymes show a close relationship with the Guǎngyùn and an even closer relationship with the Jíyùn. Because it represents a new phonological analysis, the Sìshēng Děngzǐ was an influential work after its publication, and its phonological analyses were subsequently considered standard reference. Many arrangements in the Sìshēng Děngzǐ reveal sound changes. For example, the jiāng 江 rhyme (and its shǎng, qù, and rù counterparts) is listed in the same table with the táng 唐 rhyme (and its shǎng, qù, and rù counterparts). While the jiāng 江 rhyme was originally entirely kāikǒu, the syllables of the táng 唐 rhyme have a kāihé contrast, and the syllables of the jiāng 江 rhyme are listed in either the kāikǒu and hékǒu tables in the Sìshēng Děngzǐ. The syllables with the Middle Chinese zhī 知 group initials, and the zhào 照二 initial groups are listed in the hékǒu table, the syllables with other initials are listed in the kāikǒu table. Table 4.39 is a correspondence table showing the kāihé divide in the jiāng 江 rhyme series between Middle Chinese and the Chinese of the Sìshēng Děngzǐ. Syllables 幢 chuáng ‘carriage curtain,’ 雙 shuāng ‘pair,’ 龐 páng ‘huge’ are used as examples to show the kāihé divide. This arrangement provides important information about the change of the jiāng 江 rhyme syllables. First, listing the jiāng 江 rhyme with the táng 唐 rhyme indicates they had the same main vowel. The split of the jiāng 江 rhyme syllables into kāikǒu and hékǒu indicates the syllables had -aŋ and -waŋ, respectively. This kāihé difference is basically reflected by what has been observed in both other contemporary materials and in Modern Mandarin. Table 4.40 is a correspondence table showing the kāihé divide in the jiāng 江 rhyme series between Middle Chinese and Modern Mandarin. Note that chuáng 幢 and shuāng 雙 have a rounded medial -w-, while páng 龐 has no rounded medial. Another obvious difference in the arrangement is that the Sìshēng Děngzǐ lists the representative characters according to the divisions first, then Table 4.39 The kāihé divide in the jiāng 江 rhyme series Middle Chinese

Condition

Sìshēng Děngzǐ

Example

jiāng 江 rhyme series

zhī 知 group initials zhào 照二 initial groups otherwise

hékǒu hékǒu kāikǒu

幢 雙 龐

ɖɰɔŋ ʃɰɔŋ bɰɔŋ

> > >

-waŋ -waŋ -aŋ

186

Middle Chinese

Table 4.40 The reflexes of the syllables jiāng 江 rhyme in Modern Mandarin Modern Mandarin

Middle Chinese

Condition

jiāng 江 rhyme series

zhī 知 group initials zhào 照二 initial groups otherwise

hékǒu hékǒu kāikǒu

幢 雙 龐

ɖɰɔŋ ʃɰɔŋ bɰɔŋ

> > >

tʂwɑŋ ʂwɑŋ phɑŋ

Table 4.41 A comparison of the ordering of tone and division in the Yùnjìng and Sìshēng Děngzǐ Yùnjìng

Sìshēng Děngzǐ

píng I qù I I píng III píng

II

III

IV

II

III

IV

shǎng





shǎng





shǎng I rù I II píng IV píng

II

III

IV

II

III

IV

shǎng





shǎng





according to the tones. In the earlier rhyme tables, the Yùnjìng and the Qīyīn Lüè, the representative characters are listed according to the tones first, then according to the divisions. Table 4.41 is a comparison of the ordering of tone and division in the Yùnjìng and Sìshēng Děngzǐ; whereas the Yùnjìng places tone hierarchically above division, the Sìshēng Děngzǐ places division hierarchically above tone. The Qièyùn Zhǐzhǎngtú follows the pattern of the Yùnjìng, and the Qièyùn Zhǐnán follows the pattern of the Sìshēng Děngzǐ. The most striking difference is the double listing of the rùshēng syllables. For example, in the earlier rhyme tables, all the representative characters only appear once in one table. So, for example, in the Yùnjìng, the rùshēng character jiǎo 腳 -jak is listed in the same column with its píng, shǎng, and qù counterparts jiāng 姜, qiǎng 鏹, and jiāng 殭 (all are Middle Chinese: -jaŋ). But in the Sìshēng Děngzǐ, the character 腳 -jak is not only listed in the same column with its own píng, shǎng, and qù counterparts, but also appears listed in the same column along with the xiào 效 rhyme syllables. Rùshēng duplication as seen in the Sìshēng Děngzǐ is shown in Table 4.42. Note that the xiào 效 rhyme group is yīnshēng yùn (either lacks a coda or has a glide coda, see Section 1.4.10), and thus does not have a rùshēng equivalent in Middle Chinese, so jiǎo 腳 fills up a previously vacant spot in the rhyme table.

The Tang and Song Dynasties

187

Table 4.42 The double listing of the rùshēng syllables in the Sìshēng Děngzǐ The dàng 宕 rhyme group

姜 -jaŋ

鏹 -jaŋ

殭 -jaŋ

腳 -jak

The xiào 效 rhyme group

嬌 -jaw

矯 -jaw

驕 -jaw

腳 -jaw

This is a clear indication that Middle Chinese syllables -(j)ak have changed to -(j)aw, as is found in the Khitan materials and the Zhōngyuán Yīnyùn (refer to Sections 5.1.2 and 7.1). In general, the Sìshēng Děngzǐ still follows the basic arrangements of earlier rhyme tables. The initials and děng categories have not changed. The significant changes are the mergers of the rhymes and the selection of the representative characters. 4.4.3

The Qièyùn Zhǐzhǎngtú 切韻指掌圖

The publication date of the Qièyùn Zhǐzhǎngtú (see Figure 4.3) is unknown. It was indicated in the preface but not believed by the scholars to be the work of Sīmǎ Guāng 司馬光 (1019–1086), a famous scholar and politician of the Song dynasty. The dating of this work is controversial. Regardless of some differences in the order of the initials and the usage of phonological labels, the Qièyùn Zhǐzhǎngtú is more similar to the Sìshēng Děngzǐ than the earlier rhyme tables. Like the Sìshēng Děngzǐ, the Qièyùn Zhǐzhǎngtú also has twenty tables in total. In the Qièyùn Zhǐzhǎngtú the initials are listed in one row, unlike other rhyme tables. A few changes in the arrangement of the tables are significant. One is that the Qièyùn Zhǐzhǎngtú merged many Rank-III and Rank-IV rhymes together and listed them together in Rank-III as the same rhymes. This indicates that the contrast of these Rank-III and Rank-IV rhymes had been lost. That the Rank-IV syllables are listed as Rank-III indicates that the phonetic value of Rank-IV syllables became the same as Rank-III syllables, rather than vice versa. Another significant difference is that the syllables of the zhī 之 and zhī 支 rhyme series with the zhào 照 and jīng 精 group initials are listed as Division-I instead of Division-III. The zhī 之 and zhī 支 rhyme series undergo an allophonic split in the condition of the presence or absence of a zhào 照 and jīng 精 group initial (see Table 4.43). This arrangement should be interpreted as the centralization (or so-called apicalization) of the high front vowels, i > ɨ (ɿ, ʅ). This change also can be observed in the Khitan materials (refer to Section 5.1).

188

Figure 4.3 The first table in the Qièyùn Zhǐzhǎngtú 切韻指掌圖. The xiào 效 rhyme group is moved to the front as the first table. In the earlier rhyme tables, as well as in the Sìshēng Děngzǐ, the order of the tables is based on the order of the rhymes in the Qièyùn and the Guǎngyùn. The dōng 東 rhyme is the first rhyme in the rhyme books, so the syllables of the dōng 東 rhyme are listed in the first table. The change of order highlights the change that Middle Chinese syllables -(j)aŋ have changed to -(j)aw, another characteristic of Mandarin phonology.

The Tang and Song Dynasties

189

Table 4.43 An allophonic split in the zhī 之 and zhī 支 rhyme series Qièyùn Zhǐzhǎngtú Example

Middle Chinese

Condition

zhī 之 rhyme series

zhào 照 series initial -ɨ jīng 精 series initial -ɨ otherwise -i

輜 tʂhɨ > -ɨ 思 sɨ > -ɨ 其 ɡɨ > -i

zī ‘covered wagon’ sī ‘think’ qí ‘third person pronoun’

zhī 支 rhyme series

zhào 照 series initial -ɨ jīng 精 series initial -ɨ otherwise -i

紙 tɕjɛ > -ɨ 斯 sjɛ > -ɨ 陴 bjɛ > -i

zhǐ ‘paper’ sī ‘this’ pī ‘parapet on a city wall’

4.4.4

The Qièyùn Zhǐnán 切韻指南

The author of the Qièyùn Zhǐnán (full name Jīngshǐ Zhèngyīn Qièyùn Zhǐnán 經史正音切韻指南) is Liú Jiàn 劉鑑. According to the preface by the author himself, this work was finished in 1336, during the Yuan dynasty. It is introduced here because of its structural similarities with the Sìshēng Děngzǐ and the Qièyùn Zhǐzhǎngtú. The Qièyùn Zhǐnán is especially associated with the Wǔyīn Jíyùn 五音集韻, a rhyme dictionary compiled by Hán Dàozhāo 韓 道昭 (active in the late twelfth century and early thirteenth century). In the preface it is says that the Qièyùn Zhǐnán is an analysis of the Wǔyīn Jíyùn. The representative characters of the rhymes are all from the Wǔyīn Jíyùn (Jì 1994).7 In comparison with the Sìshēng Děngzǐ and the Qièyùn Zhǐzhǎngtú, the format of the tables of the Qièyùn Zhǐnán (see Figure 4.4) is more closely related to the former. For example, the characters are listed according to their division first, and then their tone. The names of the sixteen rhyme groups (shè 攝) and the names of the 160 rhymes, according to Nìng Jìfú’s (2016) study, are almost exactly the same as that in the Wǔyīn Jíyùn. Even some mistakes in the Wǔyīn Jíyùn were copied. It has twentyfour tables. The Qièyùn Zhǐnán provides explanations before and after the tables. The explanation before the tables defines certain terminologies, such as fēn wǔyīn 分五音 ‘distinguishing the five types of initials,’ biàn qīngzhuó 辨清濁 ‘distinguishing of voiceless and voiced (initials),’ and míng děngdì 明等第 ‘being clear about the divisions.’ The explanations are in the format of lines with seven characters each. The section after the tables is the Ménfǎ Yùyàoshi 門法玉鑰匙 ‘Jade keys to the principles,’ which is a detailed summary of the basic principles of the rhyme tables (see Figure 4.5). The thirteen principles 7

與韓氏《五音集韻韻》互爲體用,諸韻韻字音皆由此韻而出也。

190

Middle Chinese

Figure 4.4 The first page of the tables in the Qièyùn Zhǐnán 切韻指南. The format of the tables is similar to the Sìshēng Děngzǐ. The characters are listed according to the division first. In each division, the characters are then listed according to the tones. Like the Sìshēng Děngzǐ, the first table lists the characters of the tōng 通 rhyme group.

The Tang and Song Dynasties

191

Figure 4.5 A page from the Qièyùn Zhǐnán, showing the Ménfǎ Yùyàoshi 門法 玉鑰匙. This is the first page of the explanations attached after the tables. The first column is the title of this list Ménfǎ Yùyàoshi Mùlù 門法玉鑰匙目錄 ‘Index of the Jade keys to the principles.’ The names of these thirteen principles are listed in columns 2 to 6. From column 7 on are the details of these principles.

192

Middle Chinese

listed mainly explain the relationships between the fǎnqiè 反切 of the representative characters and the placements of the representative characters in the tables. Without modern phonological knowledge, these principles are really the “keys” to understanding the traditional phonology in the format of rhyme tables and rhyme dictionaries. The study of these principles is a special field in traditional phonology called děngyùnxué 等韻學. Modern study of historical phonology still relies on the information about the phonological categories, such as wǔyīn, qīngzhòng, děng, shè, etc., provided in the Qièyùn Zhǐnán, as well as the Sìshēng Děngzǐ and Qièyùn Zhǐzhǎngtú rhyme tables. These three rhyme tables, namely the Sìshēng Děngzǐ, the Qièyùn Zhǐzhǎngtú, and the Qièyùn Zhǐnán, are arranged differently from two earlier rhyme tables, the Yùnjìng and the Qīyīn Lüè. The employment of the concept shè indicates a new common ground of phonological standard, which is significantly simplified from the phonological system of the Qièyùn and the Guǎngyùn. 4.5

Examples of Tang Poetry

The following poem illustrates the phonetic system of the sixth century. For the poetry composition in the traditional format, the syllables are separated into two groups, píng 平 ‘level’ vs. zè 仄 ‘oblique (non-level),’ according to their Middle Chinese tones. Píng describes píng tone syllables and zè describes the non-píng tone (shǎng, qù, and rù tone) syllables. This contrast is based on the tonal values, level vs. non-level (see Section 1.3.4). To use píng and zè syllables correctly is a basic requirement for poetry composition in the traditional format (Wáng Lì [1958] 2005). Both poems below have four lines, and each line has five syllables. Of the phonetic transcriptions, the first line is the reconstructed value, the third is the phonetic value of Modern Mandarin and the fourth is pinyin. The rhyming syllables are highlighted. In addition, the phonetic values of the modern Guangzhou dialect (Běijīng Dàxué 2003) are provided in the second line, and the píngzè information is provided in the fifth line (see Tables 4.44 and 4.45). Lù Zhài 鹿柴 ‘Deer Fence’ by Wáng Wéi 王維 (701–761) 空山不見人 但聞人語響 返影入深林 復照青苔上

No one is seen in the deserted hills, only the echoes of speech are heard. Sunlight cast back comes deep in the woods and shines once again upon the green moss. Source: English translation by Stephen Owen 1996: 393.

In the poem Lù Zhài, the two rhyming syllables xiǎng 響 and shàng 上 (see Table 4.44) have shǎng tone and qù tone respectively in modern Beijing

The Tang and Song Dynasties

193

Table 4.44 The reconstructions of Middle Chinese, reflecting the original Tang dynasty Chinese pronunciation of Lù Zhài 空 khuŋ huŋ khuŋ kōng

山 ʂɰæn ʃan ʂan shān

不 pjut pɐt pu bú

見 ken kin tɕjɛn jiàn

人 ȵin jɐn ɻən rén

但 dɑn tan tan dàn

聞 mjun mɐn wən wén

人 ȵin jɐn ɻən rén

語 ŋjɔ jy y yǔ

響 hjɐŋ hœŋ ɕjaŋ xiǎng

píng

píng





píng



píng

píng





返 pwjɐn fan fan fǎn

影 Øɰjaŋ jɪŋ iŋ yǐng

入 ȵip jɐp ɻu rù

深 ɕim ʃɐm ʂən shēn

林 lim lɐm lin lín

復 bjuk fʊk fu fù

照 tɕjɛw tʃiw tʂaw zhào

青 tsheŋ tʃhiŋ tɕhiŋ qīng

苔 dəj thɔj thaj tái

上 dʑjɐŋ ʃœŋ ʂaŋ shàng







píng

píng





píng

píng



Table 4.45 The reconstructions of Middle Chinese, reflecting the original Tang dynasty Chinese pronunciation of Xīn Yí Wū 木 muk mʊk mu mù

末 mwɑt mut mo mò

芙 bjo fu fu fú

蓉 joŋ jʊŋ ɻuŋ róng

花 hwɰa fa hwa huā

山 ʂɰæn ʃan ʂan shān

中 ʈjuŋ tʃʊŋ tʂuŋ zhōng

發 pwjɐt fat fa fā

紅 ɦuŋ hʊŋ xuŋ hóng

萼 ŋɑk ŋɔk ɤ è





píng

píng

píng

píng

píng



píng



澗 kɰan kan tɕjɛn jiàn

戶 ɦwo wu xu hù

寂 dzek tʃɪk tɕi jì

無 mjo mow wu wú

人 ȵin jɐn ɻən rén

紛 phjun fɐn fən fēn

紛 phjun jɐn fən fēn

開 khəj khɔj khaj kāi

且 tshjæ tʃhɛ tɕhjɛ qiě

落 lɑk lɔk luo luò







píng

píng

píng

píng

píng





Mandarin. But both belong to the shǎng tone in the Middle Chinese system. A tonal match for rhyming syllables is required in Tang poetry, and so although they do not rhyme in Modern Mandarin or in the Hangzhou dialect, xiǎng 響 and shàng 上 do rhyme in Middle Chinese, and in the modern Guangzhou dialect.

194

Middle Chinese Xīn Yí Wū 辛夷塢 ‘Magnolia Dell’ by Wáng Wéi 王維 木末芙蓉花 On the tips of trees are lotus blossoms, 山中發紅萼 red calyces come out in the mountains. 澗戶寂無人 Silent gates by a torrent, no one there; 紛紛開且落 in tangled masses they blossom and fall. Source: English translation by Stephen Owen 1996: 393.

The phonetic value of the rhyming syllables in Xīn Yí Wū changed quite greatly from Middle Chinese to modern Beijing Mandarin, è 萼 ŋɑk > ɤ and luò 落 lɑk > luo (see Table 4.45).

Part IV

The Beginnings of Mandarin Liao dynasty (916–1125), Jin dynasty (1115–1234), Xixia dynasty (1038–1227)

The use of the term “Mandarin” did not start until the time of the Ming dynasty (1368–1644), but the phonological features of Mandarin can be traced to the time of the Liao dynasty (916–1125) and the Song dynasty (960–1127). Here we use “Mandarin” as a label for the ancestral language of Modern Mandarin. In studying the history of Mandarin phonology, one of the fundamental questions that must be asked is: what phonological features make Mandarin phonology “Mandarin”? The word “Mandarin” has a folk etymology that suggests it means Mǎn Dàrén 滿大人 ‘Manchu Mister,’ and the language therefore is named accordingly. In actuality, the word hails from the Portuguese word mandarium, which developed from Sanskrit mantrin, a word for ‘counselor.’ The noun passed into the English language in 1589, and the adjective appeared about fifteen years later. Later “Mandarin” was used in the West to refer to the official language of China or the language spoken by “Mandarins.” While the Qièyùn 切韻 phonology as a national standard was artificially kept in use for the imperial examination and poetry composition, a sound system of a colloquial language-based standard had been developing in northern China. This system, which became the direct ancestral language of Modern Mandarin, can be traced to the tenth century. From the tenth century on, China was partially or entirely ruled by non-Han people until the Ming dynasty. The different empires during the time of the development of Early Mandarin were: Han Chinese

Altaic1

Tibeto-Burman

Northern Song 960–1127 Southern Song 1127–1279

Liao (Khitan) 916–1125 Jin (Jurchen) 1115–1234 Yuan (Mongol) 1271–1368

Xixia (Tangut) 1038–1226

1

The term “Altaic language” is itself a controversial and problematic phrase. The Altaic language family is a proposed family that includes the Turkic, Mongolic, and Tungusic languages, as well as occasionally the Koreanic and Japonic languages. There is insufficient evidence to make a conclusive genetic connection between these languages, although due to their geographical, cultural, and linguistic similarities, the term is still often used to this day. The Jurchen are identified as a Tungusic people, the Mongols as Mongolic, while the Khitan language has not been identified as a member of any language family yet, though it is most likely Mongolic or Tungusic.

195

196

The Beginnings of Mandarin

Because the Chinese-speaking regions of this period were not unified under any single authority, the phonological standard was not consistent throughout the different kingdoms. While the Song dynasty was promoting the Tang standard as exemplified by the Qièyùn, there were new standards established and used in the territories of the Liao, Jin, and Xixia (also known as the Western Xia), coexisting with the Song standard. In fact, this is a very important period as the conservative Middle Chinese phonology represented by the Guǎngyùn 廣韻 and the Jíyùn 集韻 continued to be used as a literary standard while other phonological standards in the non-Han ruling areas emerged. The Chinese phonological standard of the Xixia dynasty has many basic characteristics of the northwestern dialect of the Tang dynasty. But in the Liao territory a different standard was established. Because of the consecutive regimes run by Altaic groups, namely the Khitans, the Jurchens, and the Mongols, in the northern and northeastern parts of modern-day China, along with the gradual expansion of their territory into the territory of the Song, the Chinese phonological standard in the Liao territory gradually became a new national standard for all Chinese-speaking areas. Later in the Yuan dynasty this new phonological standard was finally established and published in the systematic form of rhyme books. The transition of the national standard from the Middle Chinese of Sui–Tang time to the Old Mandarin of the Yuan time was not just a change of the phonological system in time, but more importantly it was a change of a geographical dialect. From the tenth century on, the northern dialects, represented by the dialect of the modern Beijing area, gradually gained its standard status. The new phonological standard of the Yuan dynasty was a continuation of the phonological standards of the Chinese language spoken in the territories of the Liao and Jin. It is thus very important to recognize the multi-standard linguistic situation of this period.

5

The Chinese of the Liao, Jin, and Xixia Dynasties

5.1

The Liao Dynasty (916–1125)

The Liao dynasty is not commonly included in the history of the Chinese language for a number of reasons. One reason is that not much material in Chinese can be utilized. However, in the Khitan scripts, and especially the Khitan Lesser Script (Qìdān Xiǎozì 契丹小字), many Chinese loanwords can be identified. Quite unexpectedly, drawing only on the Chinese loanwords in the Khitan language, an entire phonological system can be retrieved. The details of this phonology provide crucial information about the early history of Mandarin. The other reason that the Liao dynasty is often left out of the discourse of historical Chinese linguistics is that the Liao, ruled by non-Han people, coexisted with the Han-ruled Song dynasty. Many linguists prefer to rely on the more “legitimate” Chinese spoken by and standardized under Han rule, rather than the language spoken outside of the Han people’s selfgoverning areas. But the phonological standards of Chinese should not be limited to the territory ruled by the Chinese. The Chinese language spoken in the territories ruled by non-Chinese people deserves equal attention. According to traditional Chinese dialectology, Chinese dialects including Mandarin are classified based on their phonological characteristics. Recent research work (Shen 2011) on the ancient scripts used by Altaic peoples from the tenth to the thirteenth centuries reveals that the origin of Mandarin phonology can be traced at least to the time of the Liao dynasty. These phonological characteristics are historical developments from the phonology of Middle Chinese (as represented by the Qièyùn 切韻) to Modern Mandarin. In the Khitan materials from the Liao dynasty, a number of basic phonological characteristics of Mandarin can be identified. 5.1.1

The Khitan Materials

The Khitan scripts were invented and used by the Khitans of the Liao dynasty (916–1125). The Khitan people developed two scripts, the Khitan Greater Script and the Khitan Lesser Script. The former is a logographic script like 197

198

The Beginnings of Mandarin

Chinese, and the latter is phonetic, although both have a structural resemblance to Chinese characters. According to the Liáo Shǐ 遼史 ‘History of Liao,’ the Khitan Lesser Script was invented by Yēlǜ Diélà 耶律迭剌 in 924. The Khitan materials that have survived to this day are mainly the epitaphs of the emperors, empresses, and high officials (see Figure 5.1). These epitaphs usually have quite accurate information about their dates and locations, which are invaluable for the study of historical phonology. Since the discovery of the Khitan Lesser Script in 1922, various attempts have been made to decipher the basic graphs, the so-called yuánzì 原字 ‘basic graphs’ (YZ), of the Lesser Script. The most influential of these is Chinggeltai et al.’s Qìdān Xiǎozì Yánjiū 契丹小字研究 ‘The Studies of the Khitan Lesser Script’ (1985). Since then, more than a dozen epitaphs (complete or fragmental) written in the Khitan Lesser Script have been discovered (Chinggeltai 2002; Liú 2014). These new materials provide additional and crucial evidence concerning the phonetic values of the YZ graphs in question. With the new materials, it becomes possible to reconstruct an entire Sino-Khitan phonology (Shen 2007). The texts written in the Lesser Script remain basically unreadable, as they are in the still-undeciphered Khitan language. However, many proper nouns of Chinese origin such as official titles, place names, and personal names have been convincingly deciphered with a high degree of confidence. A SinoKhitan phonology has been reconstructed as the result of a systematic analysis of the Khitan Lesser Script involved in the transcription of Chinese words (Shen 2007). The Chinese language reflected in the Sino-Khitan phonology is by far the oldest recorded form of northern Chinese; and despite being so old in comparison to other records, Sino-Khitan phonology is not significantly different from the Zhōngyuán Yīnyùn 中原音韻 of 1324 (ZYYY) of the Yuan dynasty or from modern Northern Mandarin. Most importantly, the phonetic nature of the Lesser Script can provide direct phonetic information, an advantage impossible to derive from the logographic Chinese materials. Many questions that would otherwise remain ambiguous can now have straight answers thanks to this clear phonetic evidence. Abbreviations of Khitan sources DL = Hán Díliè Mùzhì Míng 韓敵烈墓誌銘 ‘Epitaph of Han Dilie’ (1101) GF = Sòngwèiguó Fēi Mùzhì Míng 宋魏國妃墓誌銘 ‘Epitaph of Yelü Hongben’s wife’ (1110) GS = Yēlǜ (Hán) Gāoshí Mùzhì 耶律(韓)高十墓誌 ‘Epitaph of Yelü (Han) Gaoshi’ (sometime after 1076) GY = Gù Yēlǜ Shì Míng Shí 故耶律氏銘石 ‘Epitaph of Madam Yelü’ (1115)

The Liao, Jin, and Xixia Dynasties

Figure 5.1 A rubbing from the Dàozōng Āicè 道宗哀冊 ‘Epitaph of the Dàozōng Emperor,’ dated 1101.

199

200

The Beginnings of Mandarin

HS = Xiāo Tè Měi Kuò Gē Fūrén Hán Shì Mùzhì Míng 蕭特每闊哥 夫人韓氏墓誌銘 ‘Epitaph of Han Shi’ (1078) HT = Hǎitángshān Qìdān Xiǎozì Mùzhì Cánshí 海棠山契丹小字墓 誌殘石 ‘Fragmentary Epitaph Found near Haitangshan’ (no date) JS = Xiāo Jūshì Mùzhì Mìng 蕭居士墓誌詺 ‘Epitaph of Xiao Jushi’ (1175) LJ = Lángjūn Xíngjì 郎君行記 ‘Memorial to Langjun’ (1134) RX = Yēlǜ Rénxiān Mùzhì Míng 耶律仁先墓誌銘 ‘Epitaph of Yelü Renxian’ (1072) SZ = Huáng Tàishūzǔ Āicè 皇太叔祖哀册 ‘Epitaph of Yelü Hongben’ (1110) TL = Yēlǜ Díliè Mùzhì Míng 耶律迪烈墓誌銘 ‘Epitaph of Yelü Dilie’ (1092) TS = Xiāo Tàishān Hé Yǒngqīng Gōngzhǔ Mùzhì 蕭太山和永清公 主墓誌 ‘Epitaph of Xiao Taishan and Princess Yongqing’ (1095) WN = Xiāo Fènwùnì Túgǔcí Mùzhì Míng 蕭奮勿膩圖古辭墓誌銘 ‘Epitaph of Xiao Fenwuni Tuguci’ (1068) XW = Xǔwáng Mùzhì 許王墓誌 ‘Epitaph of Xuwang’ (1105) ZJ = Yēlǜ Zōngjiào Mùzhì Míng 耶律宗教墓誌銘 ‘Epitaph of Yelü Zongjiao’ (1053) ZX = Yēlǜ Zhìxiān Mùzhì Míng 耶律智先墓誌銘 ‘Epitaph of Yelü Zhixian’ (1094) 5.1.2

Diphthongization of Middle Chinese Syllables with Coda -k

The Khitan spellings of Chinese loanwords clearly show diphthongization. The finals of Middle Chinese rù syllables with -k, luò 洛 and yào 藥, are transcribed by Khitan YZ and (see Table 5.1). Khitan YZ is also used to transcribe non-rù syllables with a -w coda, such as gāo < kɑw 高, cáo < dzɑw 曹, shǎo < ɕjɛw 少, and Khitan YZ is used to transcribe the syllables with vowel -u, such as wǔ < ŋwo 五, bù < bwo 部, nú < nwo 奴. This lack of distinction between rù -k coda characters and non-rù -w coda characters shows that Middle Chinese syllables with a -k coda had already undergone the phonological change of diphthongization. Note that the third column of Table 5.1 and subsequent tables are a romanized transcription of the Khitan script. The pronunciation of Modern Chinese in pinyin, the source text and location in the text, and the change from Middle Chinese to Modern Mandarin are also provided (Middle Chinese > Modern Mandarin) for the characters in question in Table 5.1 and the following tables.

The Liao, Jin, and Xixia Dynasties

201

Table 5.1 Diphthongization of coda -k syllables as shown in Khitan sources Khitan 洛 藥 略 國 水 洛 luò

Trans.

+ -aw.u -aw.u + -ɛw.u + -uj + -uj + < lɑk, 藥 yào < jɐk,

Context

Pinyin

Source

洛京留守 Luòjīng Liúshǒu XW1 藥師奴 Yào-shī-nú TL31 經略 Jīnglüè LJ1 越國王 Yuèguó Wáng XW50 漆水縣 Qīshuǐ Xiàn GY2 略 luè < ljɐk, 國 guó < kwək, 水 shuǐ < ɕwi a

Change -ɑk -jɐk -jɐk -wək -wi

> > > > >

-aw -aw -ɛw -uj -uja

Note: + indicates a YZ in the place either before or after its adjacent character. The number after the abbreviation indicates the number of lines in the epitaph.

a

The character of the third row in Table 5.1, lüè 略, is from the Lángjūn Xíngjì 郎君行記 ‘Memorial to Langjun’ inscription produced in the Jin dynasty of the thirteenth century. Since no example of diphthongization with a -w coda is found in the Jurchen material, lüè 略 in Jin-era Khitan materials1 can be used to show the existence of this very important and distinctive form of diphthongization with coda -w. The fourth and fifth row characters in Table 5.1, guó 國 and shuǐ 水 provide another clear example of the diphthongization. The final of guó 國, which is a rù syllable with -k coda in the Middle Chinese system, is transcribed the same as shuǐ 水, which has no stop coda in Middle Chinese. 5.1.3

Devoicing of Middle Chinese Voiced Obstruents

The Khitan material shows very clearly that Middle Chinese voiced obstruents have become devoiced. The change pattern of stops is the same as Modern Mandarin: aspirated in píng 平 tone syllables and unaspirated in zè 仄 tone syllables (see Table 5.2). In the Middle Chinese system both dù 度 and tú 徒 have a voiced stop initial d-. Since dù 度 is a zè tone syllable, it is transcribed as ‹ tu ›, the same as dū 都, which has a voiceless unaspirated stop initial t- in the Middle Chinese system. On the other hand, the initial of the píng tone syllable tú 徒 changed to

1

It should be noted that the Jin and Western Liao coexisted, and that Khitan was still used as a language by ethnic Khitans outside of the Liao. Though these scripts were created when their dynasties began, they are restricted to neither the time nor the borders of their dynasty.

202

The Beginnings of Mandarin

Table 5.2 Khitan characters illustrate a Mandarin-like obstruent devoicing pattern Khitan 都 度 徒 + 統 + 同 都 dū < two, 度

Trans.

Context

Pinyin

Source

tu 都監 Dūjiān TL19 tu 節度使 Jiédùshǐ TL1 tu 司徒 Sītú GS5 thu知院都統 Zhīyuàn Dūtǒng XW24 thu混同郡王 Hùntóng Jùnwáng XW2 tù < dwo, 徒 tú < dwo, 統 tǒng < thwoŋ, 同 tóng < duŋ

Change tddthd-

> > > > >

ttththth-

Table 5.3 Devoicing as shown in Khitan materials Context I. Píng tone II. Voiced obstruents

> > > >

yīnpíng yángpíng voiceless aspirated voiceless unaspirated

initial is voiceless initial is voiced syllable is yángpíng syllable is not yángpíng

aspirated stop th-. This change can be further supported by the transcription of tǒng 統 and tóng 同. The initial of tóng 同 has changed from d- to th-, becoming the same initial as tǒng 統, which has a voiceless aspirated initial th- in the Middle Chinese system. The Khitan material is the earliest evidence showing the result of devoicing. Its pattern is unmistakably Mandarin. In the Khitan materials there is clear evidence for the devoicing of voiced obstruent initials (see Table 5.3). This change also implies the split of the píng tone into yīn 陰 and yáng 陽, because after the devoicing the tonal split would lose the phonological condition for the split, namely the voicing of the initial consonants. Thus, the change order has to be one where the split of the píng tone precedes the devoicing of voiced obstruents. Because this order of change cannot be reversed, devoicing must have come after the píng tone split. So, based on the devoicing pattern in the Liao dynasty, the Middle Chinese píng had already split into two separate tones, yīnpíng and yángpíng. 5.1.4

Labiodentalization

YZ is used to represent the syllable fu in Chinese. But the Chinese syllable fu is likely perceived as a homorganic labial aspirated stop and vowel phu in Khitan (see Table 5.4). YZ is usually used alone to transcribe the Chinese syllable fu, and sometimes used as an initial with other YZ graphs.

The Liao, Jin, and Xixia Dynasties

203

Table 5.4 Labiodentalization of phu in Sino-Khitan Khitan 府 副 駙 府 fǔ < pjo, 副 fù

Trans.

Context

phu 率府副率 phu 副部署 phu 駙馬 < phiw, 駙 fù < bjo

Pinyin

Source

Change

Shuàifǔ fùshuài Fù Bùshǔ Fùmǎ

GY10 XW9 GF8

pphb-

> > >

*f*f*f-

Table 5.5 Initial series, zhī 知 and zhào 照, merge in Sino-Khitan Khitan

Trans.

Context

+ 中 tʃ中京 + tʃ公主 主 + tʃ富春郡王 春 + tʃ楚國王 楚 + tʃ使持節 持 tʃi 同知 知 + tʃi政事令 政 中 zhōng < ʈjuŋ, 主 zhǔ < tɕjo, 春 chūn < 知 zhī < ʈjɛ, 政 zhèng < tɕjɛŋ

Pinyin

Source

Change

Zhōngjīng WN9 ʈGōngzhǔ WN17 tɕFùchūn Jùnwáng RX6 tɕhChǔguó Wáng RX2 tɕhShǐchíjié GF2 ɖTóngzhī WN9 ʈZhèngshìlìng DL2 tɕtɕhwin, 楚 chǔ < tʂhjɔ, 持 chí < ɖɨ,

> > > > > > >

tʃtʃtʃtʃtʃtʃtʃ-

Here it should be made clear that in Sino-Khitan the pronunciation of these was most likely ph-, but the merger of three Middle Chinese initials (p-, ph-, b-) marks that the Mandarin spoken in these regions had labiodentalized these initials into f-, and thus the reconstructions of Mandarin are f- rather than ph-. 5.1.5

The Merger of the Middle Chinese zhī 知 and zhào 照 Initial Series

There is no distinction between the Middle Chinese zhī 知 and zhào 照 initial series (see Table 5.5). In the Middle Chinese system, zhī 知 initials are stops (zhī 知 ʈ-, chè 徹 ʈh-, chéng 澄 ɖ-, niáng 娘 ɳ-) and zhào 照 initials are affricates and fricatives (zhào 照 tɕ-, chuān 穿 tɕh-, chuáng 床 dʑ-, shěn 審 ɕ-, shàn 禪 ʑ-). In the Khitan material, zhī 知 initials are transcribed as affricates and show no contrast with their zhào 照 counterparts.2 It should be noted here that in the Khitan transcriptions voiceless unaspirated and aspirated affricates are not distinguished. The aspirated sounds are often transcribed by letters representing fricatives. 2

The zhī 知 initial series has a nasal and lacks fricatives, whereas the zhào 照 initial series lacks a nasal but has fricatives. The plosives of the zhī 知 series corresponds to the zhào 照 series affricates of the same phonation type.

204

The Beginnings of Mandarin

5.1.6

The Loss of the Middle Chinese Stop Codas -p, -t, -k

The stop codas of rù syllables are lost. The syllables with -k coda have been analyzed in Section 5.1.2. Table 5.6 exemplifies the change of -k coda characters as well as -p and -t coda characters. Although in most cases the consonant codas have been lost, a few words still have the stop coda, represented by the characters in rows 7 and 8 of Table 5.6, là 臘 and shí 十. These few cases could be from an old layer of Chinese loans that were preserved in the Khitan language. The preservation of -p is in agreement with Shào Yōng’s 邵雍 Shēngyīn Chànghè Tú 聲音唱和圖 (see Section 4.3.3) of the same time period. Table 5.6 The loss of stop codas in Sino-Khitan

密 漆 積 督 察 臘 臘 十

Khitan

Trans.

Context

Pinyin

Source

Change

+ + +

-i -i -i tu tʃha -a -a.p(u) -ap

樞密院 漆水縣 積慶宮 督軍 觀察使 臘月 臘月 高十

Shūmìyuàn Qīshuǐxiàn Jīqìnggōng Dūjūn Guāncháshǐ Làyuè Làyuè Gāo Shí

XW12 GF2 GS15 RX6 GY2 RX61 HS34 GS13

-Vt -Vt -Vk -Vk -Vt -Vp -Vp -Vp

+ + +

> > > > > > > >

-i -i -i -u -a -a -p -p

密 mì < mɰit, 漆 qī < tshit, 積 jī < tsjɛk, 督 dū < twok, 察 chā < tʂhɰæt 臘 là < lɑp, 十 shí < dʑip

5.1.7

The Status of the Middle Chinese Velar Nasal Initial ŋ-

The Middle Chinese initial yí 疑 ŋ- is transcribed by YZ . YZ is also used to transcribe the velar nasal coda -ŋ; for example shèng 聖, chéng 丞, chéng , yǒng 永, and péng 彭. There is no loss of the velar nasal initial in the available data. Traditionally, changes in initials would be expected to differ depending on the rank of the character, as initial changes are often triggered by their medial. However, as seen in Table 5.7, within the Khitan material, the Middle Chinese ŋ- initial does not show any sign of the change, regardless of rank. 5.1.8

The Middle Chinese Bilabial Nasal Coda -m

The three nasal codas in Chinese were clearly distinguished in the Khitan transcriptions (see Table 5.8). Distinctive YZ graphs , , and , are used to transcribe the Chinese finals -am, -an, and -aŋ, respectively. In accordance with

The Liao, Jin, and Xixia Dynasties

205

Table 5.7 Velar nasals in the onset position of Sino-Khitan Khitan

Trans.

Context

Pinyin

Source

Change

Rank-I and -III kāikǒu + 牛 ŋ千牛衛 Qiānniúwèi + ŋ銀青 Yínqīng 銀 + ŋ儀同三司 Yítóngsānsī 儀 + ŋ娥瑾 Éjǐn 娥 牛 niú < ŋiw, 銀 yín < ŋɰin, 儀 yí < ŋɰjɛ , 娥 é < ŋɑ

ZJ21 GF2 XW1 TS14

ŋŋŋŋ-

> > > >

ŋŋŋŋ-

Rank-I and -II hékǒu + ŋ左金吾衛 Zuǒjīnwúwèi 吾 + ŋ元妃 Yuánfēi 元 + ŋ防禦使 Fángyùshǐ 禦 + ŋ宋魏國妃 Sòngwèiguó Fēi 魏 吾 wú < ŋwo, 元 yuan < ŋwjɐn, 禦 yù < ŋjɔ, 魏 wèi < ŋwɨj

GF7 GY17 GF2 GF4

ŋŋŋŋ-

> > > >

ŋŋŋŋ-

Rank-II and -IV + ŋ樂節郡主+ 樂 + ŋ堯舜+ 堯 樂 yè < ŋɰɔk, 堯 yáo < ŋew

JS4 ZX3

ŋŋ-

> >

ŋŋ-

> > >

-m -ŋ -n

Yuèjiéjùn Zhǔ + Yáo Shùn +

Table 5.8 Nasal codas’ three-way distinction in Sino-Khitan Khitan

Trans.

監 + -am -aŋ 江 + -an 蘭 + 監 jiān < kɰam, 江 jiāng
>

-jɛ -jɛ -jɛ

syllables from the má 麻 rhyme actually indicate a rule change: a > ɛ / j_, a conditioned vowel raising after a palatal glide, and (2) the loss of -p and -t endings in the relevant syllables. Although the ZYYY was the first rhyme book to make syllables of this type an independent rhyme group, the evidence for the existence of syllables of this type can be traced back earlier to the Liao dynasty (see Table 5.9). 5.1.10

The Foundations of Mandarin

The evidence shown in Table 5.1 through Table 5.9 exemplifies that the northern Chinese dialect spoken in the territory of Liao had many of the basic characteristics of Modern Mandarin. These characteristics include: 1. 2. 3. 4. 5. 6.

Diphthongization of Middle Chinese syllables with coda -k Devoicing of Middle Chinese voiced obstruents Labiodentalization of Middle Chinese bilabial initials Merger of Middle Chinese zhī 知 and zhào 照 initials Loss of Middle Chinese stop codas -p, -t, -k Vowel raising of má-III (麻) syllables

The Khitan materials show that many basic phonological features of Mandarin were in existence at that time. Thus it is not difficult to understand that the origin of Mandarin can be traced to the Liao dynasty, about three centuries earlier than the ZYYY, which is commonly accepted as the earliest evidence of Mandarin phonology. 5.1.11

Examples

Figure 5.2, a sample of rubbings from the Xuānyì Āicè 宣懿哀冊 ‘Epitaph of Xuanyi Empress’ (1101) showcases some of the official titles of its author, Yélǜ Gù 耶律固. (The translation of the official titles is based on Hucker 1985.) From the image of the rubbings, (a), (b), and (c) from left to right, it can be easily identified that in rubbing (c) the initial YZ of the three Khitan characters

The Liao, Jin, and Xixia Dynasties

207

Figure 5.2 The titles of Yélǜ Gù 耶律固 in khitan characters. A sample of rubbings from the Xuānyì Āicè 宣懿哀冊 ‘Epitaph of Xuanyi Empress’ (1101).

making jiǎn jiào guó 檢校國 are the same ( , k-) and the initial YZ of the three Khitan characters making zǐ jì jiǔ 子祭酒 are the same ( , ts-). In the examples shown in Table 5.10, four phonetic values are provided under the Chinese and Khitan characters, the first line is Middle Chinese, the second line is Sino-Khitan pronunciation (based on Shen 2007), the third line is ZYYY (based on Yáng Nàisī 1981), and the fourth line is modern Beijing Mandarin. Pinyin is provided in the last line. In comparison with Middle Chinese, the phonology of Sino-Khitan shows some important innovative features, such as the loss of coda -k in lù 祿 and guó 國 as well as the devoicing of obstruents in chóng 崇. However, in comparison with modern Beijing Mandarin, Sino-Khitan shows preservation of old features, such as the presence of velar nasal initial ŋ- in yín 銀 and yù 御, the preservation of coda -m in jiǎn 檢, and the preservation of alveolar affricates and fricatives in front of -j- or -i in jìn 進, qīng 青, jiǎn 檢, jì 祭, and jiǔ 酒. 5.2

The Jin Dynasty (1115–1234)

The Jin dynasty was established in modern day Jilin and Heilongjiang in 1115 by the Jurchens. The Jurchens, like the Khitans, were an Altaic people who spoke an Altaic language; for the Jurchens this language was appropriately called Jurchen. The Jin dynasty allied themselves with the Song dynasty to drive out the Khitans of the Liao dynasty. Although the Song dynasty was defeated by the Liao, their campaign weakened the Liao enough so that the Jurchens could drive the Khitans westward, and move their dynasty into central Asia. The Liao reestablished themselves as the Western Liao dynasty, also known as Qara Khitai, but their impact on Chinese history never recovered. The Jin dynasty continued a campaign southward, occupying and annexing many Chinese-speaking areas.

208

The Beginnings of Mandarin

Table 5.10 A multidialectal gloss of the titles of Yélǜ Gù Rubbing (a) 御



ŋjɔ ɦwjɛn ŋy ɥɛn y ɥɛn y ɥɛn yù yuàn ‘Ceremonial Receptionist’

MC SKH ZYYY MM Pinyin





thuŋ thuŋ thuŋ thuŋ tōng

tsin tsin tsin tɕin jìn



祿

Rubbing (b) 大



ŋɰin tsheŋ ɖʐjuŋ luk dɑj ŋin tshiŋ tʂhuŋ lu taj in tshiŋ tʂhuŋ lu taj in tɕhiŋ tʂhuŋ lu taj yín qīng chóng lù dài ‘Grand Master of Chonglu with Silver Seal and Blue Ribbon’

pjo fu fu fu fū

銀 MC SKH ZYYY MM Pinyin



Rubbing (c) 檢 MC SKH ZYYY MM Pinyin

5.2.1







kɰjɛm kɰau kwək tsɨ kjɛm kjɛw kuj tsɨ kjɛm kjɛw kuj tsɨ tɕjɛn tɕjaw kwo tsɨ jiǎn jiào guó zǐ ‘Acting Chancellor of the Directorate of Education’





tsjɛj tsi tsi tɕi jì

tsiw tsjəw tsjəw tɕjəw jiǔ

The Gǎibìng Wǔyīn Jíyùn 改併五音集韻 (1212)

The Gǎibìng Wǔyīn Jíyùn (GWJ) by Hán Dàozhāo 韓道昭, a modified version of the Wǔyīn Jíyùn 五音集韻 of the 1140s by Jīng Pú 荆璞, was published in 1212. It is a rhyme dictionary that contains 160 rhymes and lists homophonic groups according to their initial in the order of the traditional thirty-six initial characters. Divisional information is provided for each rhyme and kāikǒu 開口 and hékǒu 合口 rhymes are separately listed. The rhymes are labeled with tóngyòng (written as either 同用 or 通用) and dúyòng 獨用. However, as the tóngyòng and dúyòng are different from those found in the Guǎngyùn 廣韻, these rhymes are likely an independent analysis from the Guǎngyùn (see Section 4.3.1). Jīng Pú, and by extension Hán Dàozhāo, seem to have made an important effort to incorporate the phonological information of the rhyme tables into their rhyme dictionaries.

The Liao, Jin, and Xixia Dynasties

209

The reduction from the 206 rhymes of the Guǎngyùn (the most prestigious rhyming standard) to 160 is also an important revision. This is the earliest rhyme dictionary that provides a revision of the 206 rhyme standard (Nìng 1992), which probably led to the newer and more simplified rhyming standard represented by the 106 rhymes of the Xīnkān Yùnlüè 新刊韻略, published in 1227 and compiled by Wáng Wényù 王文郁, and the 107 rhyme system of the Píngshuǐ Yùn 平水韻 (full name Rénzǐ Xīnkān Lǐbù Yùnlüè 壬子新刊禮部韻 略, compiled by Liú Yuān 劉淵 and published in 1252). The GWJ is a significant deviation from the tradition of the Qièyùn 切韻 and Guǎngyùn, although it still mainly follows the tradition of rhyme dictionaries and is more or less an ideal rhyming standard.

5.2.1.1 The Merger of Rhymes In the GWJ the following rhymes (many of which are marked as tóngyòng 同 用 ‘shared’ rhymes in the Guǎngyùn) are merged together (see Table 5.11). (The píng 平 rhymes listed include the mergers of corresponding shǎng 上, qù 去, and rù 入 rhymes). For example, the tán 覃 -əm and tán 談 -ɑm rhymes of the Guǎngyùn are merged as the tán 覃 rhyme. Some of these rhymes are traditionally termed as chóngyùn 重韻 ‘duplicated rhymes’ (see Section 1.4.6). According to the rhyme tables these chóngyùn belong to the same rank and the same kāihé 開合.3 In the twelfth century these chóngyùn had already merged in standard pronunciation. Some of them are the mergers of Rank-III and Rank-IV rhymes, which did not need to be listed separately even in a conservative rhyming standard. Table 5.11 Examples of different rhymes merging in the GYJ

Rank-I Rank-II

Rank-III

Rank-III = Rank-IV

3

GWJ rhymes

Qièyùn/Guǎngyùn rhymes

覃 tán 皆 jiè 山 shān 庚 gēng 咸 xián 真 zhēn 脂 zhī 尤 yóu 凡 fán 仙 xiān 宵xiāo 鹽 yán

覃tán -əm 皆jiè -ɰæi 山shān -ɰæn 庚gēng -ɰaŋ 咸xián -ɰæm 真zhēn -in 支zhī -jɛ 尤yóu -iw 嚴 yán -jɐm 仙xiān -jɛn 宵xiāo -jɛu 鹽yán -jɛm

< < < < < < < < < < <
> >

Note: + indicates the Jurchen character is not used alone and tells the relative position of the character to other characters.

-aj -aj -aj

212

The Beginnings of Mandarin

Table 5.15 Diphthongization of coda -k causes mergers between rù and nonrù syllables Jurchen

Trans.

Context



buj

河北

背 國 貴

buj kuj kuj

背 兼修國史 貴

Source

MC

Héběi

JJS16

pək

>

-uj

bèi Jiān Xiūguóshǐ guì

YY JJS8 YY

pwoj kwək kʷɨj

> > >

-uj -uj -uj

also a Rank-I syllable from the xiè 蟹 rhyme group. So the Jurchen character represents the phonetic value aj, which makes the pronunciation of the final of cè 策 a diphthong, rather than a -k coda. Two other examples can demonstrate the diphthongization process (see Table 5.15). The character běi 北 is a Middle Chinese rù syllable. It is is trantranscribed by the Jurchen character . In the YY, the character scribed by the Chinese character bèi 背, which is a syllable from the xiè 蟹 rhyme group with coda -j. A parallel hékǒu example is gúo 國, which is also a Rank-I syllable from the zēng 曾 rhyme group. The second Jurchen character in the spelling is the character , which is transcribed by the Chinese character guì 貴 in the YY. The character guì 貴 is a Rank-III syllable from the zhi 止 rhyme group, which does not have a coda. No rù syllable from the Middle Chinese dàng 宕, jiāng 江, or tōng 通 rhyme groups is found in the available material. But the evidence of diphthongization of Middle Chinese rù syllables with a -w coda can be found in the rhyming of several Jin-era zhūgōngdiào 諸宮調, a form of storytelling and singing literature. In Liú Zhīyuǎn’s 劉知遠 Liú Zhīyuǎn Zhūgōngdiào 劉知遠諸宮調 ‘The Zhūgōngdiào of Liú Zhīyuǎn’ and Dǒng Xièyuán’s 董解元 Xīxiāng Jì Zhūgōngdiào 西廂記諸宮調 ‘The Zhūgōngdiào of Xīxiāng Jì’ the Middle Chinese rù syllables with -k coda rhyme with the syllables that originally have a -w coda. This is the same change found in the Khitan material and later (in a more systematic way) in the Yuan rhyme books. In the twelfth part of the Liú Zhīyuǎn Zhūgōngdiào, the Middle Chinese -k word xué 學 (Middle Chinese: ɦɰɔk) rhymes with Middle Chinese -w coda words nào 鬧 (Middle Chinese: ɳɰaw), dào 道 (Middle Chinese: dɑw), dào 到 (Middle Chinese: tɑw), láo 醪 (Middle Chinese: lɑw), bǎo 寶 (Middle Chinese: pɑw), nǎo 惱 (Middle Chinese: nɑw), and lǎo 老 (Middle Chinese: lɑw). In the second volume of Xīxiāng Jì Zhūgōngdiào, the Middle Chinese -k words ké 殼 (Middle Chinese: khɰɔk), luò 落 (Middle Chinese: lɑk), dù 度 (Middle Chinese: dɑk), and zhù 著 (Middle Chinese: ʈjak/ɖjɐk) all rhyme with the Middle Chinese -w words dào 道 (Middle Chinese: dɑw), dào 盜 (Middle Chinese: dɑw), liǎo 了 (Middle Chinese: lew), jiào 叫 (Middle Chinese: kew),

The Liao, Jin, and Xixia Dynasties

213

yào 要 (Middle Chinese: Øjɛw), jiǎo 攪 (Middle Chinese: kɰaw), gào 告 (Middle Chinese: kɑw), táo 逃 (Middle Chinese: dɑw), jiǎo 剿 (Middle Chinese: dʐɰaw/tsjew), kào 靠 (Middle Chinese: khɑw), nǎo 惱 (Middle Chinese: nɑw), xiào 孝 (Middle Chinese: hɰaw), liǎo 了 (Middle Chinese: lew), biǎo 表 (Middle Chinese: pɰjɛw), and tiáo 條 (Middle Chinese: dew). So in the Jin dynasty the diphthongization of Middle Chinese -k syllables has the same change pattern as that found in the Khitan materials.

5.2.2.2 Devoicing of Middle Chinese Voiced Obstruents As seen in Table 5.16, a comparison of the spelling of the three Chinese characters tóng 同, tōng 通, and dōng 東 provides the best indication that the initial consonant of tóng 同 has changed to a voiceless aspirated stop. In the Middle Chinese system the initials of tóng 同, tōng 通, and dōng 東 are voiced, voiceless aspirated and voiceless unaspirated respectively (d, th, t). After the devoicing of the voiced initials, the initial of tóng 同 became voiceless and aspirated because tóng 同 is a píng tone syllable. So in the Jurchen spellings both tóng 同 and tōng 通 are transcribed by using the ‹ thu ›. In contrast, the initial of dōng 東 is transcribed Jurchen character by using the Jurchen character ‹ tu › instead.

5.2.2.3 Labiodentalization of Middle Chinese Bilabial Initials Shown in Table 5.17, the graph is used to transcribe a Middle Chinese voiceless unaspirated stop as in fù 傅 and fǔ 府, and a voiced stop as in fèng 鳳 Table 5.16 Devoicing patterns in Sino-Jurchen as a comparison of píng tone characters Jurchen

Trans.

Context

同 通

thuŋ thuŋ

同知 通奉大夫



tuŋ

東平府

Tóngzhī Tōngfèng Dàifū Dōngpíngfǔ

Source

MC

DS2 JJS8

duŋ thuŋ

> >

thth-

JJS20

tuŋ

>

t-

Table 5.17 The labiodentalization process of Sino-Jurchen Jurchen 鳳 奉 傅 府

+ +

Trans.

Context

phuphuphu phu

丹鳳門 通奉大夫 英王傅 大名府

Dānfèng Mén Tōngfèng Dàifū Yīngwángfù Dàmíngfǔ

Source

MC

JJS12 JJS8 JJS10 JJS19

bjuŋ bjoŋ pjo pjo

> > > >

ffff-

214

The Beginnings of Mandarin

and fèng 奉. There is no distinction between these initial consonants, which otherwise could be distinguished if they were not merged. All are Rank-III syllables but none indicate the existence of a palatal medial, the distinctive transcribes syllables with the labiodental feature of Rank-III syllables. So, fricative f- in Chinese. In other words, labiodental initials were in existence in the Chinese spoken in the Jin territory. Just as the Khitan transcribed the Chinese f- with characters that represented ph-, so did the Jurchen.

5.2.2.4 Loss of Middle Chinese Stop Codas -p, -t, -k Not many examples of Middle Chinese stop coda characters can be found in addition to the -k syllables discussed in Section 5.2.2.1. No example with -p or -t can be found, so there is no data to analyze. Table 5.18 gives two examples of -k syllables that are not in diphthong form. That the Middle Chinese rù syllable bǔ 卜 and non-rù syllable bù 部 are transcribed by the Jurchen character is an indication of the loss of -k. In the YY shī 失 -t is used to transcribe the Jurchen character , which itself is used to transcribe shí 石 -k. This phenomenon also suggests the loss of the stop coda.

5.2.2.5 Retention of the Middle Chinese Velar Nasal Initial ŋThe existence of the Middle Chinese velar nasal initial can be identified by comparing the Middle Chinese syllables with the ŋ- initial yí 疑 and the syllables with the zero initial yǐng 影. The velar nasal is maintained in the onset position of Sino-Jurchen (see Table 5.19). Table 5.18 Obstruent coda loss in Sino-Jurchen Jurchen 卜 部 石 失

Trans.

Context

pu pu ʃi ʃi

卜修洪 禮部侍郎 石盞 失

Bǔ Xiūhóng Lǐbù Shìláng Shí Zhǎn Shí

Source

MC

JY3 JJS10 JJS8 YY

-Vk -u -Vk -Vt

> > > >

-u -u -i -i

Table 5.19 The velar nasal in the onset position of Sino-Jurchen Jurchen

Trans.

Context

Source

MC



+

(ə)ŋ-

太原

Tàiyuán

DS17

ŋ-

>

ŋ-

元 宛

+

(ə)ŋØ-

宗元 宛平縣

Zōngyuán Wǎnpíng Xiàn

DS30 JJS17

ŋØ-

> >

ŋØ-

+

The Liao, Jin, and Xixia Dynasties

215

Since the Jurchen character is also used for the Chinese final -əŋ in the transcription, such as zhèng 政 (JJS11), chéng 丞 (JJS8), and yīng 應 (DS2), its phonetic value as a velar nasal is obvious. The characters yuán 原 and yuán 元 historically have an yí 疑 initial, which is a velar nasal, but wǎn 宛 has a yǐng 影 initial, which is a zero initial. The contrast in the spellings of these three syllables indicates that the velar nasal initial was not lost.

5.2.2.6 Loss of the Middle Chinese Bilabial Nasal Coda -m The Jin steles show that three graphs, , , and , are used for -im, -in, and -iŋ, respectively. Table 5.20 shows a few examples that the three nasal codas are very consistently transcribed by three distinct graphs. ali.in ‘mountain’ In the Jurchen language, is used for -in only (e.g. alin JF9). The consistent use of the same characters in the transcription indicates there were three different nasal codas in the system. Table 5.20 Nasal codas’ three-way distinction in Sino-Jurchen

林 進 平

5.3

Jurchen

Trans.

Context

+ + +

-im -in -iŋ

翰林 進士 宛平縣

Hànlín Jìnshì Wǎnpíng Xiàn

Source

MC

DS1 JJS0 JJS17

-m -n -ŋ

> > >

-m -n -ŋ

The Xixia Dynasty (1038–1227)

The Xixia dynasty, also known as the Western Xia dynasty or the Tangut Empire, existed from 1038 to 1227 in the northwestern part of modern China (now the northwestern Chinese provinces Ningxia, Gansu, eastern Qinghai, northern Shaanxi, northeastern Xinjiang, southwest Inner Mongolia, and southernmost Outer Mongolia, measuring about eight hundred thousand square kilometers) with its political center in modern Yinchuan, Ningxia. The Tangut language is a member of the Tibeto-Burman language family. The Tangut Empire coexisted and bordered the Northern Song, Liao, and Jin dynasties from the eleventh to the thirteenth centuries. 5.3.1

The Tangut Script

Yělì Rénróng 野利仁榮, a Tangut scholar and close aid to the Tangut emperor Lǐ Yuánhào 李元昊, invented the script for the Tangut language when the Western Xia dynasty was founded. Tangut characters, like Chinese, make up a logographic system, and so represent morphological units. The Tangut script became known to scholars at the beginning of the twentieth century. The

216

The Beginnings of Mandarin

discovery of the Zhǎng Zhōng Zhū 掌中珠 ‘Pearl in the Palm’ (its full title being Fān-Hàn Héshí Zhǎngzhōngzhū 番漢合時掌中珠 ‘Foreign-Han Contemporary Pearl in the Palm’) permitted scholars to retrieve information concerning the northwestern Chinese dialect at the end of the twelfth century (see Figure 5.3). The Pearl in the Palm was a Tangut–Chinese bilingual glossary, compiled by the Tangut scholar Gǔ-lè-mào-cái 骨勒茂才 in 1190. The work transcribes the pronunciation of Chinese words into Tangut characters and the pronunciation of Tangut characters into Chinese; from this, the phonetic values of Chinese characters can be reconstructed by checking the transcriptions both ways. This two-way transcription is a very important tool for historical linguists, and allows them to ensure that any merging and splitting of sounds was due to historical processes, rather than orthographic constraints. According to Gong Hwang-cherng’s reconstruction (1981, 1989, 1994, 2004), the phonology of the Chinese dialect as reflected in the Tangut materials shows a number of characteristics distinct from the Chinese dialects reflected in the Khitan and Jurchen materials.

5.3.1.1 Basic Phonological Characteristics In the Sino-Tangut pronunciation, the Chinese nasal initials are transcribed by both voiced stops and nasals. This phenomenon suggests that the Middle Chinese nasal initials can be pronounced either as nasals or nasalized stops, m/mb, n/nd, ɡ/ŋɡ, and ȵ/ȵdʑ. After close examination, Gong did not find clear phonological conditions for the variations, and he concluded that the best explanation should be that the nasals and nasalized stops are in free variation (Gong 2004: 263). The reconstructed initial system has twenty-two initials, as shown in Table 5.21. Table 5.21 The initial system of Sino-Tangut p t k ts tɕ

ph th kh tsh tɕh

m/mb n/nd ɡ/ŋɡ ȵ/ȵdʑ

f

w l

x s ɕ

ʑa

j

Source: Gong 2004: 280. a According to the predicted sound change rules, this sound should be an approximate (ɻ) instead of a voiced fricative.

The only problem with this system is the voiced palatal fricative ʑ. Since all Middle Chinese voiced stops, affricates, and fricatives were devoiced, there should be no voiced fricative in the Chinese phonology. The voiced

217

Figure 5.3 A page from the Zhǎng Zhōng Zhū 掌中珠 ‘The Pearl in the Palm’ (1190). Each entry has four vertical columns: (right to left) Tangut phonetic equivalents in Chinese, Tangut script, Chinese script translation of Tangut, and Chinese phonetic equivalents of the translation in Tangut.

218

The Beginnings of Mandarin

palatal fricative is used to transcribe the Middle Chinese initial rì 日, which is now reconstructed as a liquid consonant in the Middle Chinese system and is realized as a retroflex approximant in the modern Beijing dialect. The Middle Chinese rì 日 initial is often transcribed as a voiced fricative (such as ʒ- or ʑ-) in foreign scripts because of the lack of a similar consonant in their systems, but this phenomenon should not be interpreted as the Chinese dialect having such a consonant. It is more than likely that this consonant was an approximant (ɻ-) rather than a voiced fricative (Section 7.3.1.1). The vowel system contains phonologically contrastive oral and nasalized vowels. There are seven oral vowels and four nasalized vowels, as shown in Table 5.22. Table 5.22 The Sino-Tangut vowel system Oral i e

ɨ ə a

Nasal u o

ĩ

ũ ə̃ ã

Like the phonological systems seen in Shào Yōng’s 邵雍 Shēngyīn Chànghè Tú 聲音唱和圖 (Section 4.3.3) and the Khitan and Jurchen materials (Section 5.1 and Section 5.2), the Chinese vowels seen in the Tangut materials also have a three-degree openness contrast.

5.3.1.2 Devoicing of Middle Chinese Voiced Obstruents The Tangut materials clearly show the devoicing of Middle Chinese voiced obstruents. All voiced stops and affricates merged with their voiceless aspirated counterparts in all tonal environments, differing from the pattern of devoicing seen in the Khitan materials, which follows the same devoicing pattern as Modern Standard Mandarin. Chinese characters in the same line of Table 5.23 are used to transcribe the same Tangut character and are thus interpreted as homophones in Sino-Tangut. For example, in the first line the Tangut character (Gong 2004: 253) with phonetic value phu transcribes the Chinese words pǔ 普 and pù 鋪, which have ph- initial in Middle Chinese, and also transcribes the Chinese words pú 菩, bù 部, báo 薄, and pō 泊, which have b- initial in Middle Chinese and belong to píng 平, shǎng 上, qù 去, and rù 入 tone respectively. This phenomenon indicates that the Chinese words with ph- or b- initials in Middle Chinese are all devoiced and become aspirated ph-. As seen in Table 5.24, voiced fricatives are also devoiced and become the same as their voiceless counterparts.

The Liao, Jin, and Xixia Dynasties

219

Table 5.23 Obstruent devoicing across different tones in Sino-Tangut MC

píng



MC

píng

shǎng



phu >> ph普 鋪 = b菩 部 薄 >> ph拍 – = b– – – phê 普 pǔ < phwo ‘general,’ 鋪 pù < phwo ‘shop,’ 菩 pú < bwo ‘Buddha,’ 部 bù < bwo ‘part, section,’ 薄 bò < bwɑ ‘mint,’ 泊bó < bɑk ‘anchor vessel’; 拍 pāi < phɰak ‘pat,’ 白 bái < bɰak ‘white’

rù a

泊 白

>> th天 – = d田 – 電 – thı̭e >> th聽 鐵 = d– – 定 蝶 thɪn 天 tiān < then ‘sky,’ 田 tián < den ‘field,’ 電 diàn < den ‘electricity’; 聽 tīng < theŋ ‘listen,’ 鐵 tiě < thet ‘iron,’ 定 dìng < deŋ ‘fixed,’ 蝶 dié < dep ‘butterfly’ >> kh– 器 = gkhi >> kh– 慶 = gkhɪn 器 qì < khɰi ‘vessel,’ 其 qí < ɡɨ ‘third person pronoun’; 慶 qìng < khɰjaŋ ‘celebrate,’ 茄 qié < ɡjɑ ‘eggplant’

其 茄

– –

>> tsh清 – = dz情 – tshı̭e >> tsh– 七 = dz– – tshı̭ə 清 qīng < tshjɛŋ ‘clear,’ 情 qíng < dzjɛŋ ‘feeling,’ 淨 jìng < dzjɛŋ ‘clean’; 七 qī < tshit ‘seven,’ 集 jí < dzip ‘together’

– –

– –

淨 –

– 集

Source: Gong 2004: 253–255. Note: A >> B denotes that A is used to transcribe B. a As in bòhé 薄荷 ‘mint’ in the Zhǎng Zhōng Zhū.

Table 5.24 Voiced and voiceless fricatives merge in Sino-Tangut MC si >> 西 >> 歲 swi 西 xī < sej ‘west,’ 夕xī < zjɛk ‘dusk’; 歲 suì < swjej ‘year,’ 隨 suí < zwjɛ ‘follow’

ss-

>> 世 ɕɕi̭e >> 身 ɕɕên 世 shì < ɕjɛj ‘world,’ 石 shí < dʑjɛk ‘stone’; 身 shēn < ɕin ‘body,’ 辰 chén < dʑin ‘5th terrestrial branch’

MC = =

夕 隨

zz-

= =

石 辰

ʑʑ-

Source: Gong 2004: 256–257, 260.

This particular characteristic can be traced to the ninth century. In Lǐ Zhào’s 李肇 (fl. first half of the ninth century) Táng Guóshǐ Bǔ 唐國史補, he mentions “that the people of Jīng-Xiāng (荆湘) [modern Hunan and Hubei area] pronounce tí 提 (d-) as dī 堤 (t-), the people of Jìn-Jiàng (晉絳) [modern

220

The Beginnings of Mandarin

Shanxi area] pronounce suō 梭 (s-) as cuò 莝 (ts-), and the people of Guānzhōng (關中) [modern Shaanxi area]5 pronounce dào 稻 (d-) as tǎo 討 (th-), fǔ 釜 (b-) as fù 付 (p-), are all incorrect pronunciations which have been acquired.”6 dào 稻 (d-) as tǎo 討 (th-) is the characteristic of the northwestern dialect. fǔ 釜 (b-) as fù 付 (p-) indicates the change of labiodentalization to be discussed in Section 5.3.1.3.

5.3.1.3 Labiodentalization In the Tangut materials the Middle Chinese initials fēi 非, fū 敷, and fèng 奉 had merged into one, as shown in the first line of Table 5.25. It is a clear indication of labiodentalization because these three initials were distinct in Middle Chinese. However, the Tangut language did not distinguish *f from *x, and so in the Tangut transcriptions of Chinese, the labiodental and velar fricatives are not distinguished. This orthographic merging can be seen in the second and third row of Table 5.25. This is a good piece of evidence to support labiodentalization because transcriptions for *p and *ph existed in Tangut, and thus only after labiodentalization occurs, could Middle Chinese initials fēi 非, fū 敷, and fèng 奉 be interpreted and transcribed as ‹ x ›. The lack of a Tangut distinction between *f and *x raises some questions about both the Chinese spoken in the Tangut Empire and the reliability of phonetic transcriptions in foreign orthographies. One may question whether or not the Chinese-spoken Tangut also lacked a distinction of *f and *x, and whether or not it is fair to reconstruct them as two separate phonemes. While in Table 5.25 The labiodentalization process of Sino-Tangut xwên >> p- 風 = ph- 蜂 = b- 奉 風 fēng < pjuŋ ‘wind,’ 蜂 fēng < phjoŋ ‘bee,’ 奉 féng < bioŋ ‘give with respect’ >> p- 飛 = ph- 肺 = xi̭we 飛 fēi < pwɨj ‘fly,’ 肺 fèi < phwjɐj ‘lung,’ 揮 huī < hwɨj ‘brandish’

x- 揮

>> p- 方 = b- 房 = x- 香 xi̭on 方 fāng < pwjɐŋ ‘square,’ 房 fáng < bwjɐŋ ‘house,’ 香 xiāng < hiɐŋ ‘fragrant’ >> ɱ- 無 = ŋ- 吾 ·u 無 wú < mjo ‘not,’ 吾 wú < ŋuo ‘I, my,’ 烏 wū < Øuo ‘crow’

=

Ø- 烏

Source: Gong 2004: 270–271, 274–275.

5 6

Shaanxi is the romanization for Shǎnxī 陝西 province, while Shanxi is the romanization for Shānxī 山西 province. “今荊襄人呼提為堤, 晉絳人呼梭為莝, 關中人呼稻為討, 呼釜為付, 皆訛謬所習, . . .”

The Liao, Jin, and Xixia Dynasties

221

historical reconstruction it can sometimes be difficult to distinguish orthographic constraints from the phonemic reality of the language, the Zhǎng Zhōng Zhū provides not only Tangut transcriptions of Chinese words, but also the Tangut translation, and the Chinese transcription of the Tangut words. All native Tangut words with initial x- are transcribed back into Chinese with only characters that have initial x- and not any characters with Middle Chinese initials fēi 非, fū 敷, and fèng 奉. So although there is no distinction transcribing into Tangut, there is a distinction transcribing back into Chinese characters. It is much more likely that the Tangut script’s orthographic constraint (not having ‹ f ›) is what caused this “merge,” rather than an actual phonemic merging. The status of the Middle Chinese wēi 微 ɱ- can be estimated by transcriptions in the final row of Table 5.25. The same Tangut character is used to transcribe characters with different Middle Chinese initials. The only explanation is that Middle Chinese nasal initials, ɱ- and ŋ-, were lost and became the same as the zero initial Ø- or ʔ-.

5.3.1.4 The Merger of the Middle Chinese zhī 知 and zhào 照 Initials Series Just as the merger of the initial series in Sino-Khitan, Sino-Tangut also shows no distinction between the Middle Chinese zhī 知 initial series stops (zhī 知 ʈ-, chè 徹 ʈh-, chéng 澄 ɖ-) and zhào 照 initial series affricates (zhào 照 tɕ-, chuān 穿 tɕh-, chuáng 床 dʑ-). This is exemplified in Table 5.26.

5.3.1.5 The Loss of Stop Codas As is shown in Table 5.27, the loss of the Middle Chinese stop codas -p, -t, -k is evident. The Middle Chinese syllables with different stop codas are transcribed with the same Tangut characters, and both the Middle Chinese syllables with stop codas and the syllables with no consonant coda are transcribed by the same Tangut characters. And more obviously, the same Tangut character can be transcribed by Chinese syllables with Middle Chinese no coda finals Table 5.26 Initial series, zhī 知 and zhào 照, merge in Sino-Tangut tɕi̭e >> ʈ- 知 知 zhī < ʈjɛ ‘know,’ 枝 zhī < tɕjɛ ‘branch’

=

tɕ- 枝

>> ʈ- 長 tɕi̭o 長 zhǎng < ʈjɐŋ ‘grow,’ 掌 zhǎng < tɕjɐŋ ‘palm’

=

tɕ- 掌

>> ʈ- 豬竹 = tɕ- 珠粥 tɕi̭u 豬 zhū < ʈiɔ ‘pig,’ 竹 zhú < ʈjuk ‘bamboo,’ 珠 zhū < tɕjo ‘pearl,’ 粥 zhōu < tɕjuk ‘congee’ Source: Gong 2004: 250, 252.

222

The Beginnings of Mandarin

Table 5.27 The loss of stop codas in Sino-Tangut bu / >> 謀 -w > -Ø = 沐 -k > Ø / >> 西 -j > -Ø = 息 -k > Ø si 謀 móu < miw ‘plot,’ 沐 mù < muk ‘bathe’; 西 xī < sej ‘west,’ 息 xī < sɨk ‘rest’ >> 麼 -Ø = 沒 -t > Ø mı̭u >> 他 -Ø = 達 -t > Ø tha 麼 me < mwɑ ‘tiny,’ 沒 mò < mwot ‘sink’; 他 tā < thɑ ‘other,’ 達 dá < dɑt ‘reach’ >> 鷄 -j > Ø = 擊 -k > Ø = ki >> 賊 -k > Ø = 七 -t > Ø = tshɪ >> 錫 -k > Ø = 悉 -t > Ø = sɪ 鷄 jī < kej ‘chicken,’ 擊 jī < kek ‘strike,’ 急 jí < kɰip ‘anxious’; 賊 zéi < dzək ‘thief,’ 七 qī < tshit ‘seven,, 集 jí < dzip ‘together’; 錫 xī < sek ‘tin,’ 悉 xī < sit ‘know,’ 習 xí < zip ‘practice’

急 -p > Ø 集 -p > Ø 習 -p > Ø

Source: Gong 2004: 288–294. Note: -Ø means open syllable in Gong’s transcription.

(-Ø) or stop codas (-p, -t, -k), showing the result of the loss of Middle Chinese stop codas. These syllables now became homophonous with the corresponding open syllables.

5.3.1.6 The Loss of Nasal Codas The status of the Middle Chinese nasal codas is not as clear as the stop codas. Because Tangut phonology does not have syllables with -Vŋ, the use of Tangut -V to transcribe both -V and -Vŋ syllables in Chinese cannot prove the loss of -ŋ. In the same line of argument expressed in Section 5.3.1.3., if it is possible that the Tangut transcriptions lack a velar nasal ending because they lack the means to write it, it is not necessarily because the sound no longer exists in Chinese. Because Tangut phonology has only -V syllables, both of the Middle Chinese -V and -Vŋ are transcribed by the same type of syllable. The merger of Middle Chinese -V and -Vŋ cannot be proven based on such transcription alone. However, there are enough examples to show that both Middle Chinese -V and -Vŋ syllables are used to transcribe the same Tangut syllables. Based on the Chinese transcription of the Tangut syllables, the Tangut transcriptions of Chinese shown in Table 5.28 in conjunction with the Chinese transcriptions of Tangut in Table 5.29 unquestionably indicate the loss of Middle Chinese -ŋ in Xixia Chinese. If the character represents V in Tangut, it implies that Middle Chinese -Vŋ = Vŋ in Tangut, it implies that Middle has changed to -V in Chinese. If

The Liao, Jin, and Xixia Dynasties

223

Table 5.28 No velar nasal coda transcribed in Sino-Tangut Tangut

Chinese

xwâ >> -Ø 花 >> -Ø 河 xwo 花 huā < hwɰa ‘flower,’ 項 xiàng < ɦɰɔŋ ‘neck’; 河 hé < ɦɑ ‘river,’ 黃 huáng < ɦwɑŋ ‘yellow’

= =

項 黃

-ŋ -ŋ

Source: Gong 2004: 297.

Table 5.29 Tangut words transcribed by syllables both with and without velar nasal coda Chinese

Tangut

-Ø 果 = -ŋ 剛 -Ø 爐 = -ŋ 籠 果 guǒ < kwɑ ‘fruit,’ 剛 gāng < kɑŋ ‘strong’; 爐 lú < lwo ‘furnace,’ 籠 lóng < luŋ ‘cage’

>> >>

ko lu

Source: Gong 2004: 298–299.

Table 5.30 The loss of nasal codas in Sino-Tangut -Ø

-ŋ > -Ṽ

-n > -Ṽ

-m > -Ṽ

tı̭e >> 爹 頂 典 – >> – 清 前 – tshı̭e >> 家 – 間 監 kâ >> 沙 – 產 衫 ɕâ 爹 diē < ʈja ‘father,’ 頂 dǐng < teŋ ‘top,’ 典 diǎn < ten ‘classic’; 清 qīng < tshjɛŋ ‘clear,’ 前 qián < dzen ‘front’; 家 jiā < kɰa ‘home,’ 間 jiān < kɰæn ‘inbetween,’ 監 jiān < kɰam ‘supervise’; 沙shā < ʂɰa ‘sand,’ 產 chǎn < ʂɰæn ‘give birth,’ 衫 shān < ʂɰam ‘shirt’ Source: Gong 2004: 304, 309.

Chinese -V has changed to -Vŋ in Chinese. Since Middle Chinese -V is still -V and Middle Chinese -Vŋ is unknown (either -V or -Vŋ), Middle Chinese -Vŋ must therefore have changed to -V, and not the other way around. Nasal codas -n and -m were lost as well. As shown in Table 5.30, the same Tangut syllable can transcribe Chinese syllables with different Middle Chinese codas. Tangut did not have nasalized vowels. In order to transcribe

224

The Beginnings of Mandarin

Chinese syllables, special characters were created. These syllables do not represent the native sounds of Tangut but rather how Chinese was perceived by Tangut speakers. Since the Chinese syllables with -m and -n codas in Middle Chinese were not distinguished in the Tangut transcription, it is likely that the Chinese dialect in contact has lost the distinction of the place or articulation and -Vm and -Vn syllables merged together. According to modern northwestern Mandarin, the result of the merger is a nasalized vowel -Ṽ instead of a single nasal coda. In the modern Xi’an dialect, a representative northwestern Mandarin, dàn 淡 dɑm > tæ̃ = dàn 但 dɑn > tæ̃ , tián 甜 dem > tjæ̃ = tián 田 den > tjæ̃ . So, although these syllables lost their -n and -m codas, their main vowels were nasalized as concluded by Gong (1989; 2004: 321).

5.3.1.7 Centralization of High Front Vowels Table 5.31 shows that the vowel centralization, i > ɨ, only occurred after alveolar (Middle Chinese jīng 精 initial series) and retroflex sibilants (Middle Chinese zhào-II 照二 initials). They are transcribed by 1.27 and 1.28 (these are the reference numbers for Tangut characters used by Gong based on the 夏 漢字典 Xià-Hàn Zìdiǎn ‘Tangut-Han Dictionary’). But the syllables with Middle Chinese zhī-III 知三 and zhào-III 照三 initials still have the high front vowel because they are transcribed by the same Tangut character 1.10 that transcribes the syllables with other initials and main vowel -i. However, in the Tangut materials, the relationships between Tangut characters and Chinese phonological categories are rather complex and no easy oneto-one relationships can be identified. This may be due to several factors, including the influence of the Tangut language on the Chinese spoken within Xixia and variations between different locations and different people throughout Tangut controlled areas. Table 5.31 High front vowels of Middle Chinese zhī 支, zhī 脂, and zhī 之 rhyme undergo a conditioned centralization in Sino-Tangut 支 zhī

脂 zhī

之 zhī

jīng 精 此 (1.27) 自 (1.27) 寺 – – 狮 (1.28) 史 zhào-II 照二 知 (2.9) – – 持 zhī-III 知三 枝 (1.10) 指 (2.9) 之 zhào-III 照三 此 cǐ < tshjɛ ‘this,’ 自 zì < dzi ‘self,’ 寺 sì < zɨ ‘temple’ 獅 shī < ʂi ‘lion,’ 史 shǐ < ʂɨ ‘history’ 知 zhī < ʈjɛ ‘know,’ 持 chí < ɖɨ ‘hold’ 枝 zhī < tɕjɛ ‘branch,’ 指 zhǐ < tɕi ‘point,’ 之 zhī < tɕɨ ‘it’ Source: Gong 2004: 260–261, 352–354, 367.

(1.27) (1.28) (2.9) (2.9)

>> >> >> >>

ɨ ɨ i i

The Liao, Jin, and Xixia Dynasties

5.3.2

225

Phonological Characteristics and Differences of Northern Dialects

Many basic phonological characteristics of the Chinese dialect reflected in the Tangut materials can be identified as the characteristics of the Chinese dialect spoken in the territory of Xixia, rather than alterations caused by the Tangut language or orthography. Additional evidence for such characteristics can be found in modern dialects spoken in the region in or adjacent to the territory of Xixia. The main characteristics of the Chinese dialect reflected in the Tangut materials include: Middle Chinese voiced stops and affricates have all changed to voiceless aspirated stops and affricates, and the codas -p, -t, and -k have been lost. These phonological characteristics were once the common features of the northwestern dialect distributed in a geographical area of Jinnan (in south Shanxi), Yuxi (in west Henan), Guanzhong (in central Shaanxi) and Longxi (in Gansu). Such features still survive in these areas (Lǐ Rúlóng and Xīn 1999) even though the features of the northwestern dialect have, throughout history, been gradually replaced by the more dominant Zhongyuan and Northern Mandarin dialects. In northern China two quite distinctive dialects coexisted. The northeastern dialect and the northwestern dialect have a number of distinctive characteristics. The most significant difference is the devoicing pattern of Middle Chinese voiced stops and affricates. The devoicing pattern of the northeastern dialect is now one of the most basic characteristics of Mandarin. In contrast, the phonological characteristics of the northwestern dialect can only be found in colloquial readings of the dialects in the Mandarin-speaking areas in the northwest. It is reported (Lǐ Rúlóng and Xīn 1999; Wáng Línhuì 2003) that the devoicing pattern of the northwestern dialect of the eleventh century existed in what is now Shanxi, Shaanxi, and Gansu. This characteristic of voiced > voiceless aspirated can be clearly observed in the Fenghe dialects of Zhongyuan Mandarin (Wáng 2003), shown in Table 5.32. Other distinct phonological characteristics can also be found in the modern dialects in that region. For example, the loss of velar stop and nasal codas, -k and -ŋ, can be found in colloquial readings (Wáng 2003). As a result, both Middle Chinese syllables with -k and -ŋ become open syllables and have the same segmental elements as Middle Chinese syllables with -Ø coda. The examples in Table 5.32 show that the Middle Chinese types of syllables represented by láng 狼, luò 落, and luó 騾 have the same initial and final in these dialects. Because Middle Chinese rù tone syllables merged with other tones, some of the Middle Chinese syllables with -k and -ŋ could end up as homophones with the same tone. The last two characters in Table 5.32, cháng 嘗 and sháo 勺, are homophones in all the given dialects except Xiangfen; like in standard Mandarin, these characters are both yángpíng.

226

The Beginnings of Mandarin

Table 5.32 A comparison of five modern Fenghe dialects

Middle Chinese Devoiced initials 婆 ‘old woman’ b-, píng tone 棋 ‘chess’ g-, píng tone 在 ‘exist; in, at’ dz-, shǎng tone 道 ‘way’ d-, qù tone 坐 ‘sit’ dz-, qù tone 大 ‘big, large’ d-, qù tone 轎 ‘sedan chair’ g-, qù tone 白 ‘white’

b-, rù tone

Loss of velar codas 狼 ‘wolf’ -ŋ 落 ‘drop’ -k 騾 ‘mule’ -Ø 嘗 ‘taste’ -ŋ 勺 ‘spoon’ -k

臨汾

襄汾

侯馬

稷山

河津

Linfen

Xiangfen

Houma

Jishan

Hejin

phɔ tɕhi tshaj

phə/pə tɕhi tshaj

phə tɕhi tshaj

phə tɕhi tshaj

phə tɕhi tshaj

taw tsɔ/tshɔ ta/thɔ tɕjaw

taw/thaw tshwə ta/thə tɕhjaw

taw/thaw tshwə ta/thə tɕhjaw

taw/thaw tshwə ta/thə tɕhjaw

paj/phɤ

phej

phej

paj/phja

taw/thaw tshwə ta/thə tɕhjaw/ tʂhaw phjɛ

lɑŋ/lɔ lɔ lɔ tʂhɑŋ/ʂɔ ʂɔ

laŋ/lə lə lə tʂhɑŋ/ʂuə ʂə

ləŋ/lə lə lə tʂhəŋ/ʂə ʂə

laŋ/lwə lwə lwə tʂhaŋ/ʂə ʂə

ləŋ/lwə lwə lwə tʂhəŋ/ʂə ʂə

Note: For two pronunciations divided by a slash, i.e. X/Y, X is the literary pronunciation, and Y is the colloquial pronunciation.

After the fall of the Tangut Empire, this area became part of the territory of the Yuan and Ming dynasties. The prestigious Northern Mandarin of the capital persistently spread into the northwestern dialectal region, creating a literary layer at the initial stage and gradually replacing many other aspects of the local dialects. This process is still in progress today. The dialects in these areas are nowadays classified as subdialects of Mandarin (Lǐ Róng 1985; Xióng and Zhāng 2008); however, their devoicing pattern and many other characteristics are distinctively non-Mandarin. A comparison with historical data clearly reveals that these non-Mandarin characteristics are inherited, rather than new innovations or changes due to migration, as these features can be unambiguously traced to the Chinese dialect spoken in the Xixia territory.

5.3.2.1 A Comparison of Two Northern Dialects Because of the clear boundaries between the Tanguts, Khitans, and the Jurchens, a comparison of these contemporary northern dialects from the tenth to the thirteenth centuries can provide us with the basic phonological differences of

The Liao, Jin, and Xixia Dynasties

227

Table 5.33 A comparison of northern and northwestern dialect sound changes

Middle Chinese Píng tone devoicing pattern Zè tone devoicing pattern Labiodentalization in Rank-III hékǒu syllables Status of -ŋ Status of -m Loss of -p Loss of -t Loss of -k Merger of ʈ and tɕ

Northern dialects

Northwestern dialects

bbp-

> > >

phpf-

phphf-

-ŋ -m -p -t -k ʈ- 6¼ tɕ-

> > > > > >

-ŋ -m -Ø -Ø -Ø, -j, -w tʂ-

-Ø, -Ṽ -Ṽ -Ø -Ø -Ø, -j tɕ-

the northern Chinese. These characteristics can further provide us with information regarding chronological change and information about the spread of standard Mandarin geographically. The comparison of northern and northwestern dialect sound changes in Table 5.33 shows that these two phonologies differed from each other in many aspects including the most diagnostic feature of the devoicing pattern. On the other hand, similarities in the patterns of labiodentalization and the merger of Middle Chinese zhī 知 and zhào 照 initials indicate that these two changes took place early on, before the northern–northwestern split.

Part V

A New Standard Yuan dynasty (1279–1368)

After the North–South opposition between the Liao and the Northern Song, and later between the Jurchens of the Jin dynasty, the Tanguts of the Xixia dynasty, and the Han Chinese of the Southern Song, the Yuan dynasty was a period of history when those previous territories were unified, though under the rule of the Mongols. The Yuan capital was at Dadu (pinyin: dàdū 大都 ‘great capital’), located in the same geographical area as modern Beijing. This location of modern Beijing was the southern capital (one of five capitals) of the Liao and the central capital (again one of five) of the Jin. For ruling purposes, at the beginning of the Yuan dynasty a Guózì 國字 ‘National Script’ was invented at the order of the emperor. After the Yuan dynasty, this script was called the hP’ags-pa script, named for its inventor, Lama hP’ags-pa. Its spelling system_ is well preserved in the Ménggǔ Zìyùn _ ‘Mongol Script (arranged according to) Rhymes,’ which is a phono蒙古字韻 logical work produced in the second half of the thirteenth century, soon after the invention of the hP’ags-pa script. The Ménggǔ Zìyùn was an effort by the _ the new script and to provide an orthographic standard Yuan court to promote for transcribing Chinese. For modern Chinese historical phonologists, the obvious significance of this material is that, for the first time, Chinese was systematically transcribed into an alphabetic system. Thus, individual sounds and syllables are spelled out clearly in a systematic way. Since both phonological categories and phonetic values of the standard phonology are given, no reconstruction, in the traditional sense, is required. The phonetic values of many crucial categories, such as the contrast between Division-III and Division-IV syllables, as well as the contrast of the chóngniǔ 重紐 syllables and the palatalization of Division-II syllables can all be studied directly. In 1324, about fifty years after the establishment of the Yuan dynasty, a rhyming book entitled Zhōngyuán Yīnyùn 中原音韻 ‘Rhymes of the Central Plain’ was published. The phonological system seen in this book is completely different from the Qièyùn 切韻 system, but it shares many basic phonological characteristics with Modern Mandarin. Therefore, in the study of historical phonology, the Chinese of the Zhōngyuán Yīnyùn is referred to as Old Mandarin. This new standard represented an important turning point in the phonological history of Chinese. The Zhōngyuán Yīnyùn provides a complete 229

230

A New Standard

phonology of the standard language based on the Zhongyuan dialect as indicated by the title of the book; however, the well-established phonological characteristics of the Zhōngyuán Yīnyùn suggest that the origin of Mandarin phonology could be pinpointed to a much earlier time. Quite systematic transliterations of Chinese can be found in the Persian documents created in the Persian part the Mongol Empire. A fairly large number of the transcriptions of Chinese words can be found in two books. One is Rashid al Din’s History of China, which is a part of his world history (Tārīkh-i Chīn az Jāmiʻ al-Tavārīkh). The other is Mài Jué 脈訣 which is an annotated translation into Persian of a Chinese medical treatise in verse. The transcriptions were based on the colloquial Chinese pronunciation, and thus they can provide us with the phonetic information that cannot be retrieved from the Chinese categorical materials. The Persian transcriptions of Mài Jué show very interesting variations of individuals’ spoken language. Some are quite close to the features of the capital dialect, and some only show certain features of the Mandarin dialect while maintaining obvious non-Mandarin features of the southern dialects.

6

Old Mandarin: The Ménggǔ Zìyùn 蒙古字韻

The anonymous Ménggǔ Zìyùn (MGZY) of the 1260s (see Figure 6.1) is likely a stripped-down version of earlier works called the Ménggǔ Zì Yùnlüè 蒙古字 韻略 or Ménggǔ Yùnlüè 蒙古韻略 as mentioned in the fifteenth-century Korean work Hóngwǔ Zhèngyùn Yìxùn 洪武正韻譯訓 (Shen 2008a, 2015). Unlike the rhyming dictionaries of the time, the MGZY has only the characters listed with no definitions provided. A few exceptional glossaries were added by Zhū Zōngwén 朱宗文 in his revised version of 1308, which is the only version to have survived to this day. Traditional rhyme works provide only categorical information; foreign transcriptions usually provide phonetic information about individual words. The MGZY systematically shows the phonetic spellings for all the categories of an entire phonology, as well as dividing them categorically. Because of this, the MGZY has a special position not only in the study of Yuan-era phonology but also in the whole of Chinese phonological history; it is a pity that it is often studied simply as one of the many traditional rhyme dictionaries. The significance of the MGZY’s phonetic information has been largely overlooked and unquestionably deserves more specialized attention from scholars than it has received. It should be noted that although the MGZY is often referred to as a work of the Yuan dynasty, its categorical structure reflects the phonological standard of northern China before the Yuan dynasty, with transcriptions approximated by Yuan-era pronunciations. Its conservative phonological structure is very likely based on the format of rhyme tables, especially the Qī-Yīn-Yùn 七音韻 (QYY), a lost text mentioned in the Gǔjīn Yùnhuì Jǔyào 古今韻會舉要 (Nìng 1997). The phonetic values, which must have been based on the real pronunciation, were applied to the conservative phonological categories. Thus, just like the previous standards, this phonological standard is still an idealized system. That is to say, the MGZY is organized in a way that reflects an older pronunciation than its own spelling does. The evidence suggests that three steps were taken to compile the MGZY in the following order: (1) An existing phonological system was adopted. This phonological system was probably that of the QYY (Yáng Nàisī 1997; Nìng 231

232

A New Standard

Figure 6.1 A page from the Ménggǔ Zìyùn 蒙古字韻. It reads horizontally from left to right. The third and fourth hP’ags-pa spellings, , in _ the first line (leftmost) is the phonetic transcription of the Chinese characters zì yùn 字韻 below it. The first two spellings are moŋ ɣol, the Mongolian word for Mongol.

Old Mandarin: The Ménggǔ Zìyùn 蒙古字韻

233

Table 6.1 The Ménggǔ Zìyùn as a fusion of previous works The Ménggǔ Zìyùn’s

Phonological system Lexicon Spelling system

is derived from is derived from is

the Qī-Yīn-Yùn the Xīnkān Yùnlüè newly devised

1997). The QYY was a rhyme table, so each phonologically distinguishable syllable was represented by only one character; the number of characters listed is equal to the number of distinguishable syllables. (2) The spellings in the hP’ags-pa script were added to each of the homophonic groups to provide the _standard transcription of Chinese characters. Characters in the same homophonic group, regardless of tones, were transcribed with the same hP’ags-pa spelling. (3) Since the original rhyme table uses one character to _ represent each distinctive syllable, homophonous characters were added for the purpose of a full transcription of every character. The characters were basically copied from an existing rhyme dictionary, the Xīnkān Yùnlüè 新刊 韻略, which is a rhyme dictionary compiled by Wáng Wényù 王文郁 in the early thirteenth century. Thus, in a strict sense, the only really new information in the MGZY is the hP’ags-pa spellings. This relationship is summar_ ized in Table 6.1. Because of the way the MGZY was compiled, the phonetic values of all words appear in a very regular way without exceptions. Such a neat phonology gives the impression that the phonetic values of the characters are determined based on previous phonological categories, which in and of themselves are ideal standards, rather than any real dialect or idiolect, which is much more likely to have irregularities. It is quite clear that no efforts were made to introduce the system of the hP’ags-pa alphabet in the MGZY. Instead a shortcut was made to establish the _ syllabic correspondences directly. In the MGZY, a list of the initial consonants is provided, in which the hP’ags-pa initial letters were matched to the trad_ itional thirty-six initial characters of Middle Chinese. But these thirty-six initial characters, although they might be popular and easy for users to recognize, cannot provide accurate phonetic values for the hP’ags-pa letters. Even more _ problematic is that the vowel letters are listed without Chinese equivalents as pronunciation cues. Thus, the phonetic values of the vowel letters cannot be learned. Instead of individual letters, the entire syllabary had to be learned and memorized. It is clearly a disadvantage in learning this new script, because there are more than 900 graphs for syllables, but only 42 letters. The lack of an alphabetic table and syllable charts in the MGZY clearly shows a misunderstanding of the basic function of an alphabet in the process of promoting this script.

234

A New Standard

6.1

The hP’ags-pa Script _ The hP’ags-pa (also ‘Phags-pa) script was created in the thirteenth century and later _named after its inventor Lama hP’ags-pa (1235–1280), a Tibetan scholar _ and a religious leader. His birth name was Blo-gros rgyal-mtshan, while hP’ags-pa was his honorary name, meaning ‘sage’ in Tibetan. His official title _ the Mongol Empire was guóshī 國師 ‘national preceptor’ and later dìshī 帝 in 師 ‘imperial preceptor.’ This new script was intended to be a universal script for all languages. In historical context, “all languages” should be interpreted as the languages used in the countries that were conquered by the Mongol Empire; this list of languages is long, but most importantly included Mongolian, Chinese, Uyghur, Persian, Tibetan, and Tatar. This script was originally called Ménggǔ Xīnzì 蒙古新字, ‘New Mongolian Script’ in contrast with the old Uyghur script, which was the first script that the Mongols learned and used, taught to them by a Uyghur minister and scholar by the name of Tatatunga in the year 1206. Later, the name of the new script was changed to Ménggǔ Zì 蒙古字, ‘Mongolian Script’ and later further changed to Ménggǔ Guóshū 蒙古國書 or Ménggǔ Guózì 蒙古國字, ‘National Script of the Mongols,’ and Yuán Guóshū 元國書 or Yuán Guózì 元國字, ‘National Script of the Yuan’ after the establishment of the Yuan dynasty in 1271; after the collapse of the Yuan dynasty, this script was – and still is – referred to as the hP’ags-pa script. The connection between the hP’ags-pa script and_ the Tibetan script is evident and well known. One of _the obvious similarities is the order of the initial letters. The first thirty original letters are in the exact same order as in the Tibetan script. The hP’ags-pa letters are graphically modified from Tibetan, _ but the graphic resemblance is so obvious that the hP’ags-pa script can be considered just a special style of the Tibetan script that _is used for making seals today (Wáng Yáo 1979). In Figure 6.2 a comparison of the two scripts with the orders of the letters in both hP’ags-pa and Tibetan scripts is shown; the _ phonetic realization here represents how each of the Tibetan letters is pronounced, though this varies across time and space (with modern Lhasa Tibetan having undergone devoicing and initial mergers). One of the most obvious differences between the hP’ags-pa script and the Tibetan script is their writing _ direction. The Tibetan writing order is from left to right across the page. But the hP’ags-pa writing direction is from top to bottom and then in columns from left _to right. This writing order is clearly inherited from the Uyghur script that the Mongols were using at the time. 6.1.1

Adaptations between Tibetan and hP’ags-pa Script _ The Tibetan script is based on similar scripts used in India at the time, particularly the style used in Kashmir. These scripts, and thus Tibetan script

Old Mandarin: The Ménggǔ Zìyùn 蒙古字韻

235

Figure 6.2 A comparison of the Tibetan and hP’ags-pa scripts. _

as a daughter script, are abugidas. Abugidas differ from “true alphabets” in that while vowels and consonants are both letters in a true alphabet, abugidas only allow for consonants to be letters, while vowels are diacritics placed on or near the letter, modifying it to create a syllable. Abugidas are often used to represent languages that are multisyllabic and have somewhat simple syllable structure. They are found in Tibet, South and Southeast Asia, Ethiopia, and northern Canada. In Tibetan, an isolate letter is pronounced as its inherent sound, with the vowel /a/ added; any diacritics added to this letter will change the vowel, but not the consonant. Different abugidas will use different vowels as their “base” vowel. The design of the hP’ags-pa script, though strongly based on the Tibetan abugida, is actually_ a true alphabet. While it still maintains the property that an isolate letter is to be pronounced with an additional /a/, other vowels are created not by modifying the letter, but by adding an additional letter. Table 6.2 demonstrates this via a comparison of Tibetan and hP’ags-pa _ vowel script, showing their representation of the same syllable with differing sounds. The first five columns have an initial consonant k, while the

Table 6.2 A comparison of syllable creation in Tibetan and hP’ags-pa script _ Tibetan hP’ags-pa _

ka

ke

ki

ko

ku

a

e

i

o

u



ཀེ

ཀི

ཀོ

ཀུ



ཨེ

ཨི

ཨོ

ཨུ

236

A New Standard

Table 6.3 Non-consonantal alloglyphs a No initial Has initial



e

i

o

u

-w-

-j-



–a

The letters ‹ w › and ‹ j › are considered initials, or belonging to the initial slot, and are thus not represented within this table. a

subsequent five columns are the isolated vowel. While in many other Indic scripts isolate vowels will have their own letter, in Tibetan a non-consonantal letter ཨ represents a without any consonant preceding it, and added diacritics will change its value to another vowel. Because of the hP’ags-pa script’s alphabetic nature, isolated vowels are represented by their_ own letter rather than a modified ‹ a ›. Though the letters never represent initials, these five vowel letters in the final five columns of Table 6.2, along with two glide letters, representing medial -w- and -j- are collectively named the yù 喻 initial in the MGZY. These letters, oriented vertically (as they are in reality) are shown in Table 6.3, with their variations when an initial does not precede them. In the following tables and in the text, hP’ags-pa script sequences of letters _ will be layed out horizontally, oriented ninety degrees counterclockwise to their natural positions. Though the vowels presented by the hP’ags-pa script _ the Old Manare satisfactory for many languages, some languages, including darin of the Yuan dynasty, had more vowels than the hP’ags-pa alphabet could _ accommodate. The solution to this was the introduction of fronting feature 1 (FT) letters. FT letters ‹ ¨ ›, and ‹ ħ › were placed before a vowel to change its quality, notably increasing or reducing frontness (respectively) or causing centralization of the vowel (refer to Table 6.7). Table 6.4 depicts each FT/ vowel combination possible, along with their Latin transcriptions and resulting phonemic values. In order for the syllable block structure to even better reflect the Chinese syllable, it was devised to work in accordance with the Chinese traditional analysis of the syllable. In the Chinese tradition, phonological analysis separates the initials from the rest of the syllable. The remainder of the syllable is called the “final” in English, corresponding to the familiar phonological unit yùn 韻 in the traditional phonology. Unlike the initials, a final is not 1

Symbol ‹ ħ › is used to represent the vocalic h, the romanization of is ‹ kħin ›, is ‹ khin › phonetically khin, with being the aspirated k, written phonetically kən; ‹ kh ›.

Old Mandarin: The Ménggǔ Zìyùn 蒙古字韻

237

Table 6.4 The realization of different FT/vowel combinations in hP’ags-pa _ script Ø/ Ø

‹a› ‹ä› ‹ ħa ›

‹e› ‹ë›

-a -ɛ -ɒb

-e -e

‹i› ‹ï› ‹ ħi ›

-i -ia -ɨ~ə

‹o› ‹ö›

‹u› ‹ü›

-o -ø

*u *y

a

marks the palatalization of gutturals (velars, uvulars, etc.) before front vowels *i, *e. The sequence ‹ ħa › only appears before velar nasals. *ɒ contrasts with *a between words such *tʃaŋ and *tʃɒŋ. as b

Table 6.5 The realization of hP’ags-pa finals in different locations _ hP’ags-pa _ Initial Isolate Nucleus Coda

mma – -m

nna – -n

ŋŋa – -ŋ

ʋʋa – -w

jja – -j

– e e -j

necessarily a single segment, and can be further broken down into medials, main vowels, and codas. Thus, the finals cannot be easily transcribed like the initials. The seven yù-initials, while they represent the medial and main vowel section of the final, are not sufficient to represent all Chinese finals. Five initial letters, and one of the yù-initials were thus designated for transcribing the consonant codas. Table 6.5 illustrates those initial letters that are reappropriated for coda use in Old Mandarin. For nasal endings -m, -n, and -ŋ, their equivalent hP’ags-pa _ initial initials ‹ m › ‹ n › ‹ ŋ › are used. For glides -w and -j, either their and ‹ j › are used, or the vowel ‹ e › is used. It is notable forms ‹ w › that the initial used to represent -w actually has the value of ʋ- in initial position. The initials of these syllables are from the Middle Chinese initial wēi 微 ɱ-, a labiodental nasal. In the Ménggǔ Zìyùn 蒙古字韻 (MGZY) the syllables wù 務 ʋu and wū 汙 Øu, wàng 妄 ʋaŋ and wāng 汪 Øwaŋ are in contrast. These syllables still maintain the initial ʋ- in the Zhōngyuán Yīnyùn 中原音韻, and do not merge with the corresponding syllables with zero initial and a -w- medial. These two -j are different letters in the spelling but phonologically indicate the same palatal coda. They appear in different phonological environments and after in complementary distribution: coda -j is represented by the letter vowels a, ə and the letter after vowel -u (refer to Table 6.14 in Section 6.2.3).

238

A New Standard

The hP’ags-pa spellings used to transcribe all the syllable finals are thus _ combinations of the seven noninitial letters and the six coda letters. 6.1.2

The Syllable Block

The orthography of the hP’ags-pa script combines individual letters into syllabic blocks. The letters_ are written one after another vertically (though in this book syllable writings in hP’ags-pa script will be written horizontally, for _ sake of space). Such syllable-block writing is evident in Chinese as well as in Mongolian, and this tradition can be traced back to Tibetan writing. In Chinese phonology, the syllable is a perceptually salient unit in both oral and written language. Thus, it is an easy decision to make to spell the Chinese syllable or the characters in syllable blocks. A graphically equivalent writing unit in the hP’ags-pa script is used to match the Chinese words (refer to Figure 6.1). The _ letters within a syllable block must be strung together vertically and the letters within such a block must be connected. Because of the orthography of the hP’ags-pa script the correspondence between Chinese characters and the _ hP’ags-pa spelling is given at the syllabic level; the MGZY is a great example _ this. Each spelling block uses one to four letters to represent a syllable. But of the relationship between the hP’ags-pa letters and the structural slots of the _ syllables is not a simple one-to-one relationship. No more than four segments exist in (Middle Chinese and Early Mandarin) Chinese syllables, with slots for the initial, medial, main vowel, and coda of a word. Since some of the vowels are represented by two hP’ags-pa letters, the limitation of using no more than four letters becomes an_ interesting technical issue to deal with. Table 6.6 shows examples of each of the possible combinations of syllable slots for spelling. The letters have five types according to their functions. The five types of letters are: initial letters, glide letters, FT letters, nucleus letters, and coda letters. The first three columns represent the hP’ags-pa spelling, Chinese _ final column shows how character, and a Latin transcription spelling. The many hP’ags-pa letters are required to make that spelling. The _phonetic values they represent are listed in Table 6.7. Spellings 1–7 are straightforward and need no explanation. The spellings 8–12 involve the FT as a modifying letter for the vowels, and so the relationships between the spellings and the phonetic values are also shown. 6.1.3

The Problems of the Script

In some cases the contrasts between two different syllables cannot be reflected in the spellings, and in these instances, two different syllables are spelled the same. This is the case found between the dōng 東 rhyme group and the gēng 庚 rhyme group. The second final of the dōng 東 rhyme group is the same as

239

‹ ja ›

‹ wa ›

‹ gu ›

‹ gja ›

‹ gwa ›

‹ gjan ›

‹ gwan ›

‹ gen ›

‹ gän ›

‹ dʒħi ›

‹ dʒħaŋ ›

‹ gwän ›

‹ għin ›

























根 k

k





k

k

k

k

k

k

k

Ø

j

t

-

-

-

-

-

-

-

-

Glide

-

w

-

-

-

-

w

j

w

j

-

w

-

-

-

-

-

-

-

-

-

-

-

FT

Note: † represents that these syllables lack a vowel letter because their nucleus is ‹ a ›

‹ da ›



Initial

Table 6.6 Types of syllable blocks for transcribing Chinese characters

ħ

¨

ħ

ħ

¨

-

-

-

-

-

-

-

-

-

Nucleus





















i

a

a

i

a

e

a

a

a

a

u

a

a

a

-

-

-

-

-

-

-

Coda

n

n

ŋ

-

n

n

n

n

-

-

-

-

-

-

4

4

3

3

3

3

3

3

2

2

2

2

1

1

240

A New Standard

Table 6.7 Differences across the orthographic–phonological interface hP’ags-pa _

Hànzì

Transliteration

Realization

Modification

1 2

打 孤

‹ da › ‹ gu ›

ta ku

3



‹ gja ›

kja

4



‹ gwa ›

kwa

5



‹ gjan ›

kjan

6



‹ gwan ›

kwan

7



‹ gen ›

ken

8



‹ gän ›

kɛn

ä

9 10

葘 莊

‹ dʒħi › ‹ dʒħaŋ ›

tʃhɨ tʃɑŋ

ħi > ɨ ħa > ɑ

11



‹ gwän ›

kɥœn

wä > wœ > ɥœ

12



‹ għin ›

kən

ħi > ɨ



the third final of the gēng 庚 rhyme group. Both are spelled as ‹ üŋ › . These two separate finals existed separately in the MGZY, but were spelled identically. The phonetic values of these two finals, thus, differed, being -yŋ and -ɥəŋ, respectively, according to their historical developments. The reason why these two non-homophonous syllables are spelled the same way is because the phonetic value -ɥəŋ cannot be spelled according to the hP’ags-pa spelling rules. Only four letters can be used to spell a syllable. For a _ syllable with final -ɥəŋ, one letter must be used for the initial and one must be used for the coda -ŋ. After those two letters are used, only two letters are left to spell the middle of the syllable. But there is no way to use two letters to spell -wjə- or -ɥə-. The nucleus -ə alone requires a combination of two letters, ‹ ħi › , to spell. The glide -wj- or -ɥ- thus cannot be represented. . The closest spelling approximation for this phonetic value is ‹ ü › Phonetically, the two letters indicate the front feature and the high and rounded features respectively. If spelled this way, the features of the glides – high, front, and rounded – are well represented, but this spelling has two problems. One is that the phonological features of the nucleus -ə are not represented, and the other is that features of the glide become the feature of the nucleus. Table 6.8 shows that as a result -ɥəŋ was spelled exactly the same as -yŋ, whose rounded and high features are applied to the nucleus. These two groups with the same spelling should not be interpreted as a phonological merger (Shen 2008a). 6.1.4

Phonological Contrasts That Cannot Be Reflected

To provide the hP’ags-pa spellings for the homophonic groups of an existing _ phonological system is a difficult task. Several pieces of information can be

Old Mandarin: The Ménggǔ Zìyùn 蒙古字韻

241

Table 6.8 Transcription of dōng 東 and gēng 庚 rhyme groups Rhyme group

Hànzì

dōng 東 gēng 庚

縈營榮 扃傾瓊

hP’ags-pa _ spelling

Phonetic value of spelling

Phonetic value of Hànzì

‹ üŋ › ‹ üŋ ›

-yŋ -ɥəŋ

retrieved from the phonological system itself. The Qī-Yīn-Yùn 七音韻 is a rhyme table. It contains the following pieces of information: (a) the number of possible initial-final combinations (syllables); (b) the number and types of initials; (c) the rhyme groups; and (d) the division of the finals within each rhyme group. Of course, all these pieces of information are categorical and do not inherently reveal any phonetic information. In order to spell all the syllables, the phonetic information must be obtained. Because of its reliance on previous categorical rhyme tables it seems that the basic function of the hP’ags-pa spelling is not to transcribe the Chinese syllables but rather to label _ phonological categories. Since the nature of the script is phonetic, it has to the match the phonetic values of the phonological categories. To use a letter with intrinsic value p to transcribe value k would be totally unacceptable. But on the other hand, using similar sounds (such as ‹ d › transcribing t) or sequences (such as ‹ wj › transcribing ɥ)2 is a logical way to substitute for a sound that itself has no direct letter; how closely the letters can transcribe the sounds must be carefully dealt with by scholars. 6.2

The Phonological System of the Ménggǔ Zìyùn 蒙古字韻

The phonological system of the Qī-Yīn-Yùn 七音韻 (QYY) was an ideal standard for the Yuan time. This system was not only adapted in the Ménggǔ Zìyùn (MGZY) but also in the Gǔjīn Yùnhuì Jǔyào 古今韻會舉 要, compiled by Xióng Zhōng 熊忠 and published in 1297.3 It should also be no surprise that the QYY is not a record of a single dialect. As an idealized standard, it contains historical contrasts that may not have existed in the Dadu dialect. Some of the phonological contrasts may be based on dialects of the time and some may be based on previous phonological

2 3

‹ w › for labial rounding and ‹ j › for palatal approximate, making *ɥ, a rounded labial approximate. The Gǔjīn Yùnhuì Jǔyào was an important turning point in the tradition of rhyme dictionaries. It actually contains two phonological systems, one in the Qièyùn tradition and the other using the QYY as a new standard (Nìng 1997).

242

A New Standard

works. In the study of historical phonology, it is always important to make a clear distinction between the phonological standard and a live dialect in the same historical period. Following the framework of the QYY, the MGZY continued to use the fourtone and thirty-six initial character (zìmǔ 字母) systems. As attested by earlier data from the tenth to the thirteenth centuries (as shown in Section 5.1.6, Section 5.2.2.4, and Section 5.3.1.5), it is quite clear that both systems were artificially maintained. With reference to the hP’ags-pa alphabet, the phonetic values of the thirty-six characters are listed in_ Table 6.9. The list in Table 6.9, in fact, represents a new initial system. Initials zhī 知 ʈ-, chè 徹 ʈh-, and chéng 澄 ɖ- are traned the same as zhào 照 tɕ-, chuān 穿 tɕh-, and chuáng 床 dʑ- (each of the two sets as , , and , respectively), and both fēi 非 and fū 敷 are transcribed as the same letter . These distinctions were either not retrievable in Yuan dynasty dialects or were considered nonstandard at the time. Contrary to this, the traditional initials xiá 匣, yǐng 影, and yù 喻 each

Table 6.9 The thirty-six initial characters with their hP’ags-pa equivalents, _ enumerated Order in zìmǔ zìmǔ zìmǔ value hP’ags-pa _

1 見 k

2 溪 kh

3 群 g

4 疑 ŋ

5 端 t

6 透 th

7 定 d

8 泥 n

Order in zìmǔ zìmǔ zìmǔ value hP’ags-pa _

9 知 ʈ (tʃ)

10 徹 ʈh (tʃh)

11 澄 ɖ (dʒ)

12 娘 ɳ

13 幫 p

14 滂 ph

15 並 b

16 明 m

Order in zìmǔ zìmǔ zìmǔ value hP’ags-pa _

17 非 pf (f )

18 敷 pfh (f )

19 奉 bv (v)

20 微 ɱ

21 精 ts

22 清 tsh

23 從 dz

24 心 s

Order in zìmǔ zìmǔ zìmǔ value hP’ags-pa _

25 邪 z

26 照 tʃ

27 穿 tʃh

28 床 dʒ

29 審 ʃ

30 禪 ʒ

31 曉 h

Order in zìmǔ zìmǔ zìmǔ value hP’ags-pa _

32a 匣 ɦ

32b (合) (ɦj)

33a 影 Ø

33b (幺) (j)

34a 喻 (Ø)

34b (魚) j

35 來 l

36 日 ɻ

Old Mandarin: The Ménggǔ Zìyùn 蒙古字韻

243

split into two to become the pairs ɦɦj-, Øj-, and Øj-, respectively. Table 6.10 shows that each of these three pairs of initials are in compleØ- and Ø- are both zero initials; is for high mentary distribution. register tones, and for low register tones. The pair of j- and j- has the same respective contrast. The initials for lower register tones are indicated with an underline, Ø- j-. The result is that the thirty-five initial letters of the MGZY actually represent thirty-two initial sounds. Furthermore, the palatalization of the guttural initials is marked with the hP’ags-pa letter ‹ ¨ › _ these initials are before ‹ i › and ‹ e › (as seen in Table 6.4); because not individual letters, but are rather modified by FT letters, this series of palatalized guttural initials is not indicated in the list shown in Table 6.9. 6.2.1

The Fifteen Rhyme Groups

The body of the MGZY is the fifteen rhyme groups (shè 攝) based on main vowel coda combinations, and is shown in Table 6.11 and can be referred to in Table 6.14. No rhyme dictionary created before the MGZY uses an organizational method that includes shè to form rhyme groups. This is not just true for the official and traditional rhyme dictionaries like the Guǎngyùn 廣韻 (1008), the Jíyùn 集韻 (1037), or the Lǐbù Yùnlüè 禮部韻略 (1037), but it is also true of even relatively new and progressive-for-the-time dictionaries such as the revolutionary Gǎibìng Wǔyīn Jíyùn 改併五音集韻 (1208). After the publishing of the MGZY, the concept of rhyme groups was extremely well received. In post-MGZY dictionaries such as the Zhōngyuán Yīnyùn 中原音韻 (1324), Table 6.10 Three zìmǔ split according to their allophonic variations Original zìmǔ Original zìmǔ value MGZY zìmǔ MGZY zìmǔ value hP’ags-pa _

合 ɦ

匣 ɦ

影 Ø 匣 ɦj

影 Ø

喻 j 幺 j

魚 Ø

喻 j

Table 6.11 The fifteen rhyme groups of the Ménggǔ Zìyùn 1 6 11

東 dōng 佳 jiā 尤 yóu

2 7 12

庚 gēng 真 zhēn 覃 tán

3 8 13

陽 yáng 寒 hán 侵 qīn

4 9 14

支 zhī 先 xiān 歌 gē

5 10 15

魚 yú 蕭 xiāo 麻 má

244

A New Standard

the Zhōngzhōu Yuèfǔ Yīnyùn Lèibiān 中州樂府音韻類編 (1351), and the Yùnlüè Yìtōng 韻略易通 (1442), rhyme groups were consistently employed to organize finals. In this regard, the MGZY speaks to a defining moment in the association of syllables. The arrangement of fifteen rhyme groups is not just an improvement of the finals; it mirrors the new, Mandarin-based phonological standard for Chinese. Within the fifteen rhyme groups, syllables are arranged first according to finals and then by initials. Table 6.12 exemplifies this arrangement. The initials are listed in a strict order, the same as the arrangement of the rhyme tables. Different finals (some including the FT marking palatal feature of the guttural initial) are also listed in a strict order based on the vowel letter. Table 6.12 An example of the initial arrangement pattern in the Ménggǔ Zìyùn dōng 東 rhyme group uŋ k- 公 kh- 空 yŋ k- 弓 kh- 穹

t- 東 ɡ- 窮

th- 通 dʒ- 蟲

d- 同 ts- 縱

n- 農 tsh- 樅

tʃ- 中 dz- 從

tʃh- 忡 s- 嵩

dʒ- 崇 z- 松

... ...

6.2.2

The Phonetic Values of the hP’ags-pa Letters _ Although the hP’ags-pa script is an alphabet, the real phonetic values of the _ totally transparent. Difficulty in rendering the values is due to spellings are not the following reasons: (1) the phonological system is ultimately still an artificial system – the phonetic values of many phonological categories did not exist; and (2) spelling problems between the hP’ags-pa letters and Chinese phonology. Table 6.13 is a list of the thirty-two_ initials of the MGZY.

6.2.3

The Spelling of Finals

before ‹ i › and ‹ e › indicates the Because the hP’ags-pa letter ‹ ¨ › palatalization_ of the guttural initials rather than a change in the value of the following vowel, not all final letter spellings indicate a difference in the final. before ‹ i › and ‹ e › Thus, any spelling with this letter combination (‹ ¨ › ) does not represent a different final than the equivalent spelling without the FT letter; rather it represents a separate initial (see Section 6.1.1 and Section 6.2). and ‹ ħ › needs explanation. The The usage of hP’ags-pa letters ‹ ¨ › _ ‹i› before and ‹ e › actually indicates the palatalization of letter ‹ ¨ › the guttural initials rather than a change in the value of the following vowel; not all final letter spellings indicate a difference in the final. In the zhī 支 rhyme

Old Mandarin: The Ménggǔ Zìyùn 蒙古字韻

245

Table 6.13 The thirty-two initial sounds of the Ménggǔ Zìyùn v

ʋ

ʂ

ʐ

ɻ

s

z

h

ɦ

p

ph

b

m

f

t

th

d

n

l



tʂh



ɳ

ts

tsh

dz

k

kh

ɡ

ŋ

j

Øa

Source: Shen 2008a. a Two-letters mark for j and two-letters mark for Ø. These are based on whether the syllable has a high or low register in different languages (see Table 6.10).

group, the spellings ‹ ki › and ‹ kï › are listed as two different finals. But these two syllables are ki and ci phonetically. The difference is about the initial consonants, k- and c-. combination is deleted in On the other hand, the vocalic ‹ ħ › in ‹ ħi › > , > ). For example, the spellings after a guttural fricative ( and ‹ lħiw › are listed separately as different finals, spellings ‹ ħiw › but phonetically they are həw and ləw, with the same final. The finals listed in Table 6.14 are in the original order of the MGZY, and the hP’ags-pa spellings _ (see Table 6.15) that actually have the same phonetic values are listed in parentheses. 6.2.4

The Subsystems of Finals

The main vowels form a system with three degrees in vowel height (see Table 6.16). Vowel -ɛ has two conditional variations -ɛ and -e, and vowel -ɨ has two conditional variations -ɨ and -ə. The main vowels appear differently according to the codas. Such subsystems can be analyzed as in Table 6.17. There can be as many as seven vowels forming a three-way contrast in terms of height. Most contrasts are found in front of -n and -Ø codas. Such a system is very close to the vowel system of Modern Mandarin. The medial system and coda system are also the same as that of Modern Mandarin: medials -j-, -w-, -ɥ- and codas -m, -n, -ŋ, -j, -w. In general the finals are very much true to the Mandarin pattern.

246

A New Standard

Table 6.14 Phonetic values of the finals in the fifteen rhyme groups 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15

東 庚 陽 支 魚 佳 真 寒 先 蕭 尤 覃 侵 歌 麻

dōng gēng yáng zhī yú jiā zhēn hán xiān xiāo yóu tán qīn gē má

uŋ iŋ aŋ i u aj in an jɛn aw jəw am im o jɛ

yŋ əŋ jaŋ ɨ y waj un on (jɛn) jɛw wəw jɛm əm wo wa

ɥəŋ waŋ (i)

(iŋ) ɑŋ wej

wəŋ (waŋ) ɥej

(iŋ) ɥɛŋ (ɥej)

jaj yn wan ɥɛn (jɛw) əw (jɛm) (im)

əj ən jan (ɥɛn) waw (jəw) jam

(əj) (in)

win

ja

ɥɛ

(ɥɛ)

(jɛn) jaw (əw) (jɛm)

wi

ɥɛw

(jɛ)

a

Table 6.15 hP’ags-pa spellings of the transcriptions in Table 6.14 _ 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15

6.2.5

東 庚 陽 支 魚 佳 真 寒 先 蕭 尤 覃 侵 歌 麻

dōng gēng yáng zhī yú jiā zhēn hán xiān xiāo yóu tán qīn gē má

{ }

{

}

{ }

Tonal Marks

Although in transcriptions, hP’ags-pa did not mark for tones, the MGZY _ system, using píng 平, shǎng 上, qù 去, and follows the traditional four-tone rù 入 to label all syllables with the same initial and final. A few examples of this organization are illustrated in Table 6.18. The tonal categories are the same as in the Qièyùn 切韻 tradition, with no deviation in assignment of character to tone, so no tonal information can be learned from the arrangement.

Old Mandarin: The Ménggǔ Zìyùn 蒙古字韻

247

Table 6.16 Vowels of the Ménggǔ Zìyùn i•y ɛ a

ɨ

u o ɑ

Table 6.17 The vowel-coda subsystem of the Ménggǔ Zìyùn -Ø: -j: -w:

iɛyuoɨa əa ɛəa

iɛəa iɛyuoəa iɛyɑəa

-m: -n: -ŋ:

Table 6.18 No tonal distinction in hP’ags-pa transcriptions _ hP’ags-pa _

Transcription

píng

shǎng





‹ di › ‹ gü › ‹ tsaw › ‹ tsiŋ ›

低氐 居裾 糟醩 情晴

邸底 舉莒 早澡 靜靖

帝諦 據鋸 竈躁 淨穽

的適 菊鞠 作鑿 –a

a The rù tone at this point in time has already lost its codas, and so while transcriptions like ‹ di ›, ‹ kü ›, and ‹ tsaw › became valid syllables for the rù tone, the coda -ŋ was never a viable coda for the rù tone.

6.3

Sound Changes in the Initial System

The new phonological standard based on the northern dialect reveals a number of innovative features of the initials. 6.3.1

The Loss of Voiced Obstruents

In the Ménggǔ Zìyùn 蒙古字韻 (MGZY) the tripartite distinction of Middle Chinese is artificially preserved. This is due to the adaptation of a phonological framework that employed the traditional initial system of thirty-six letters, among other conservative characteristics. While Tibetan natively had and still maintains this distinction between voiced, voiceless aspirated, and voiceless unaspirated stops and affricates, Yuan-era spoken Chinese had only a two-way distinction. The voiceless unaspirated and voiced stops and affricates were undifferentiable to Chinese speakers, and so during the creation of the MGZY, the two categories were accidently switched. The traditional voiced stops and affricates of the Qièyùn 切韻 were represented by the hP’ags-pa equivalents of Tibetan voiceless unaspirated stops and affricates, and _vice versa (Shen 2008a:

248

A New Standard

112–116). Because this interesting voicing reversal is not present in any modern dialects, and is an unlikely sound change to occur, it indicates devoicing in the language and the conservativeness of the phonological framework. This can be seen in Table 6.19. 6.3.2

Labiodentalization of Middle Chinese Bilabial Initials

The labiodental fricatives are spelled with digraph letters based on letter ‹ h › and the glide letter ‹ w › . The Middle Chinese voiceless fēi 非 and fū 敷 initials are merged and spelled as ‹ f › , while the voiced fèng 奉 is spelled ‹ v › in the hP’ags-pa spellings (see Table 6.20). These two letters are graphically very similar. _ In the hP’ags-pa spellings the voiceless and voiced fricatives are not well _ and sometimes reversed (Shen 2008a: 92–96). The exact letters distinguished used for the three bilabial fricatives are easily confused in the only surviving copy of the MGZY. Though they are often reversed, their ordering in the MGZY itself is consistently correct, and so the true value of the initials can be restored using the order of initials. This phenomenon also shows that the voicing contrast did not exist in the language and was artificially made, although the phenomenon of labiodentalization is evident. 6.3.3

The Loss of the Middle Chinese yí 疑 ŋ- Initial

The Middle Chinese yí 疑 ŋ- initial shows a conditional loss. According to the MGZY, the loss of the velar nasal initial already happened in all the Table 6.19 The voicing reversal of stop initials Middle Chinese MGZY

幫 幫

pb-

滂 滂

phph-

並 並

bp-

Middle Chinese MGZY

端 端

td-

透 透

thth-

定 定

dt-

Table 6.20 Labiodentalization in the Ménggǔ Zìyùn 風 豐 逢

pphb-

>

fuŋ



vuŋ

夫 敷 符

pphb-

>

vu†



fu†

風 fēng < pjuŋ ‘wind’ 豐 fēng < phjuŋ ‘many’ 逢 féng < bjoŋ ‘meet’ 夫 fū < pjo ‘husband’ 敷 fū < phjo ‘apply’ 符 fú < bjo ‘symbol’ Note: † represents that in the hP’ags-pa spellings the letters for f- and v- are mistakenly _ reversed.

Old Mandarin: The Ménggǔ Zìyùn 蒙古字韻

249

hékǒu 合口 syllables as well as in Rank-II and Rank-IV kāikǒu 開口 syllables. The yù 喻 j- initial (and its newly created allophonic initial 魚 yú Ø-) is used to mark these newly created initial-less syllables. The new assignment of the yù 喻 j- initials to the syllables that once had the yí 疑 ŋinitial is illustrated in Table 6.21. The kāikǒu syllables of Rank-II and RankIV have yù 喻 j- initials, and the hékǒu syllables of Rank-II and Rank-III have yú 魚 Ø- initials. 6.4

Sound Changes in the Final System

The final system also shows some typical features of Mandarin phonology. 6.4.1

The Medial

Because of various spelling restrictions, Chinese medials are treated differently in different types of syllables. The Middle Chinese syllables in Table 6.22 exemplify this difference across divisions and the kāihé 開合 contrast. While in writing, medial -j- is indicated after a guttural initial and in front of the default vowel a, the medial -j- in front of the front vowels -e and -ɛ are Table 6.21 Conditional loss of Middle Chinese initial ŋMiddle Chinese Rank

I

II

III

IV

yí 疑 ŋ-

kāi ŋ

hé Ø

kāi j

hé Ø

kāi ŋ

hé j

kāi Ø















hé -

Table 6.22 Differences in the realization of medials depending on kāihé kāi Rank-I Rank-II Rank-III Rank-IV

干 間 鞬 堅

干 gān < kɑn ‘offend’ 間 jiān < kɰæn ‘inbetween’ 鞬 jiān < kjɐn ‘quiver’ 堅 jiān < ken ‘hard, strong’

hé ‹ ‹ ‹ ‹

gan › gjan › gen › gän ›

[kan] [kjan] [kjen] [kjɛn]

官 關 卷 涓

官 guān < kwɑn ‘official’ 關 guān < kwɰan ‘close’ 卷 juǎn < kwɰjɛn ‘volume’ 涓 juān < kwen ‘tiny stream’

‹ ‹ ‹ ‹

gon › gwan › gön › gwän ›

[kon] [kwan] [kɥØn] [kɥœn]

250

A New Standard

omitted. This omission causes no phonological confusion because -j- and front vowels (-e and -ɛ) always co-occur. Medial -w- is always indicated in spelling. Medial -ɥ- is indicated in two different ways: (1) the -ɥ- medial is omitted – both of the front and round features are indicated by the main vowel; and (2) a -w- medial is used – its front feature indicated by the main vowel, the same as for medial -j-. When cooccurring with -w-, -j- and -w- become a singular medial -ɥ-. This process of an innate -j- morphing in conjunction with an added -w- can be seen in juǎn 卷 and juān 涓, especially in contrast with jiān 鞬, which has an innate -j- but no -w-, and guān 關, which has an added -w- but because its vowel is not front lacks an innate -j-. Because -w- medial only occurs before non-front vowels, -w- and -ɥ- can easily be distinguished from each other. Such a medial system indicates a shift from the Middle Chinese system to a Mandarin system. Of the Rank-II syllables, the kāikǒu and hékǒu syllables differ only in the medial, -j- and -w-. Historically this should be viewed as two different processes: the Rank-II medial -ɰ- is fronted to be a palatal medial -jin the kāikǒu syllables, but lost in the hékǒu syllables. This is represented in Table 6.23. Although the distinction between Rank-II and Rank-III, -ja(-) vs. -je(-), is still present, the disappearance of the special -ɰ- medial of Rank-II implies that the Middle Chinese medial system has transferred to a three-medial system -w-, -j-, -ɥ-. Along with the finals lacking a medial, this system was later called the sìhū 四呼 ‘four sighs’ system (see Section 1.4.7 for more on this). 6.4.2

The Reduction of the Vowel System

Compared with Middle Chinese phonology, the vowel system is rather simplified. In the rhyme tables, five out of the sixteen rhyme groups are used to summarize the finals with -ŋ coda, namely dōng 東, gěng 梗, zēng 曾, jiāng 江, and dàng 宕. In the Ménggǔ Zìyùn 蒙古字韻 (MGZY) the number of rhyme groups is reduced to three, dōng 東, gēng 庚, and yáng 陽. According to the main vowels of the rhyme groups, the -Vŋ syllables are listed in Table 6.24.

Table 6.23 Changes to the Middle Chinese medial -ɰ- in different contexts Middle Chinese Condition Medial -ɰ-

Ménggǔ Zìyùn Example

in conjunction with a -w- medial Ø otherwise -j-

關 間

kwɰan > kɰan >

kwan kjan

Old Mandarin: The Ménggǔ Zìyùn 蒙古字韻

251

Table 6.24 Vowel Reduction in the Ménggǔ Zìyùn from the Middle Chinese final system 16 rhyme group system zēng gěng

i ɛ

zēng dàng

ə a

Mergers

dōng jiāng

>

u ɔ

>

MGZY system

zēng/gěng

dōng/ zēng/gěng

zēng/gěng

gēng >

zēng/gěng dàng/ jiāng dōng/ zēng/gěng

i

u

>

zēng/gěng dàng/ jiāng

dōng gēng yáng

ə a

Table 6.25 The phonetic values of -Vŋ/-Vk finals before and after the Ménggǔ Zìyùn’s vowel reduction, from Middle Chinese (MC) to the MGZY and to Modern Mandarin (MM) MC

Condition

MGZY

Example

MC -ɛŋ, -æŋ

-ɔŋ -ɨk -uk -ɛk -ək -ɔk -ɑk

after -jafter -wotherwise in all situations in all situations in all situations in all situations in all situations in all situations in all situations

-iŋ -uŋ -əŋ -aŋ -i -u -aj -əj -aw -aw

並 轟 耕 邦 極 谷 百 刻 剝 各

pjɛŋ hwɰæŋ kɰæŋ pɰɔŋ ɡɨk kuk pɰak khək pɰɔk kɑk

MGZY > > > > > > > > > >

piŋ huŋ kəŋ paŋ ɡi ku paj khəj paw kaw

MM > > > > > > > > > >

bìng hōng gēng bāng jí gǔ bǎi kè bō gè

‘combine’ ‘rumble’ ‘plow’ ‘nation’ ‘extreme’ ‘valley’ ‘hundred’ ‘carve’ ‘peel’ ‘each’

As shown in Table 6.25, the rhyme groups with mid vowels have merged into other rhyme groups. This merger, especially before the -ŋ coda, results in a vowel system that closely resembles Modern Mandarin. 6.4.3

Chóngniǔ 重紐 Reflexes in the Ménggǔ Zìyùn

The chóngniǔ contrast of Middle Chinese is still partially preserved in the MGZY. The pattern of which contrasts are preserved and which ones are not is

252

A New Standard

Table 6.26 Type A and Type B rhyme groups 東 dōng 陽 yáng

Type A Type B

庚 gēng 佳 jiā

支 zhī 寒 hán

魚 yú 先 xiān

真 zhēn 蕭 xiāo

尤 yóu 覃 tán

侵 qīn 歌 gē

麻 má

Table 6.27 Chóngniǔ contrasts maintained between Rank-III and Rank-IV Rhyme group kāikǒu zhī qí zhī zhī hékǒu zhī qí zhī zhī 棄 契 季 桂

Example

Rhyme group

Example

脂 齊 脂 脂

IIIa: IV: IIIa: IIIb:

棄 契 棄 器

khji khji khji khi

6¼ 6 ¼ = =

zhī wēi qí wēi

脂 微 齊 微

IIIb: III: IV: III

器 氣 契 氣

khi khi khji khi

脂 齊 脂 脂

IIIa: IV: IIIa: IIIb:

季 桂 季 愧

kyj kyj kyj kuj

6¼ 6 ¼ = =

zhī wēi qí wēi

脂 微 齊 微

IIIb: III: IV: III:

愧 貴 桂 貴

kuj kuj kyj kuj

qì < khi ‘abandon’ qì < khej ‘contract’ jì < kwi ‘season’ guì < kwej ‘osmanthus’

器 qì < khɰi ‘vessel’ 氣 qì < khɨj ‘air’ 愧 kuì < kwɰi ‘ashamed’ 貴 guì < kwɨj ‘precious’

based on the height of the nuclear vowel. Type A rhyme groups have the main or ‹ u › , and Type B rhyme groups have any other vowel. vowel ‹ i › Table 6.26 shows the rhyme groups assigned to each type. Based on the reflexes shown in Table 6.27 two generalizations can be made: (a) the chóngniǔ IIIa and IIIb syllables are in contrast, and (b) the chóngniǔ IIIb and IIIa syllables have the same phonetic realization as Rank-III and Rank-IV syllables respectively. These examples show that the chóngniǔ IIIa word qì 棄 (Middle Chinese: khi) is in contrast with the IIIb word qì 器 (Middle Chinese: khɰi), and the Rank-IV word qì 契 (Middle Chinese: khej) is in contrast with the Rank-III word qì 氣 (Middle Chinese: khɨj). The examples also show that the chóngniǔ IIIa word qì 棄 (Middle Chinese: khi) and the Rank-IV word qì 契 (Middle Chinese: khej) are the same, and the chóngniǔ IIIb word qì 器 (Middle Chinese: khɰi)

Old Mandarin: The Ménggǔ Zìyùn 蒙古字韻

253

Table 6.28 More examples showing the contrast of IIIa and IIIb syllables Rhyme group kāikǒu zhī 支 zhī 支 zhī 脂 hékǒu zhī 支 zhī 脂

Example

Rhyme group

Example

IIIa IIIa IIIa

卑 岐 比

pi ɡji pi

6¼ 6 ¼ 6 ¼

zhī zhī zhī

支 支 脂

IIIb IIIb IIIb

碑 奇 鄙

puj ɡi puj

IIIa IIIa

窺 葵

khyj ɡyj

6¼ 6 ¼

zhī zhī

支 脂

IIIb IIIb

虧 逵

khuj ɡuj

卑 bēi < pjɛ ‘humble’ 岐 qí < ɡjɛ ‘mountain name’ 比 bǐ < pi ‘compare’ 窺 kuī < khwjɛ ‘peep’ 葵 kuí < ɡwi ‘sunflower’

碑 bēi < pɰjɛ ‘stone tablet’ 奇 qí < ɡɰjɛ ‘strange’ 鄙 bǐ < pɰi ‘low’ 虧 kuī < khwɰjɛ ‘lose’ 逵 kuí < ɡwɰi ‘thoroughfare’

Table 6.29 A summary of chóngniǔ reflexes

IIIa IIIb a

P

G



i uj

ji i

yj, *wia uj

The reflex *wi is only found in syllables with the yǐng 影 and yù 喻 initials.

and the Rank-III word qì 氣 (Middle Chinese: khɨj) are the same. The same pattern can be found in the hékǒu syllables. More examples given in Table 6.28 show the contrast between the IIIa and IIIb syllables. The reflexes of chóngniǔ in the zhī 支 rhyme group are further summarized in Table 6.29. Based on all the examples of the chóngniǔ contrast examined, Table 6.30 lists the status of the chóngniǔ contrast in all the phonological conditions. Here P, G, and Gʷ represent labial, guttural, and labialized guttural initials in the MGZY, respectively. Gutturals include anything that has a velar place of articulation or further back in the mouth (velar, uvular, epiglottal, pharyngeal, glottal). In general, the chóngniǔ contrast is more systematically kept in Type A syllables. From the data set given in Table 6.30, the following observations can be made: For Type A rhyme groups:

1. The chóngniǔ contrast is lost after labial initials in syllables with codas -n, -ŋ, and -w.

254

A New Standard

Table 6.30 The chóngniǔ reflexes of Type A and Type B rhyme groups yīn Type A P G Gʷ Type B P G Gʷ

yáng



-Ø + + +

-j + + +

-w – + /

-m / + /

-n – + +

-ŋ – + –

-p / + /

-t + + +

-k + + /

/ / /

/ / /

– + /

/ + /

– (+) +

/ / /

/ (+) /

– / /

/ / /

Note: + means that a contrast exists; (+) means that a contrast is marginally kept; – means that the contrast is lost; and / means that there are phonological gaps, that is, there was no contrast before.

2. The chóngniǔ contrast is well kept after the plain gutturals. 3. The chóngniǔ contrast is kept after labialized gutturals, except in syllables with coda -ŋ. For Type B rhyme groups:

1. The chóngniǔ contrast is lost after labial initials. 2. The chóngniǔ contrast is kept after plain guttural initials, but only marginally in the yáng syllables with -n coda and rù syllables originally with the Middle Chinese coda -p. 3. The chóngniǔ contrast is kept after labialized gutturals as in the syllables with coda -n. In terms of hP’ags-pa spelling, Rank-IV and chóngniǔ IIIa syllables are _ chóngniǔ IIIb and Rank-III without chóngniǔ syllables are equivalent, while equivalent. These two newly created groups, however, have contrast between them. Because of this, we can conclude that there is no phonological difference existing in the MGZY between the contrast of the chóngniǔ IIIa and IIIb syllables and the contrast between Rank-IV and non-chóngniǔ Rank-III syllables, respectively. At the time of the MGZY, chóngniǔ contrasts were reduced to being equivalent to the contrast between Rank-III and Rank-IV syllables. In comparison with two earlier phonological works, the Qièyùn Zhǐzhǎngtú 切韻指掌圖 (QZ), a rhyme table of the Southern Song period (1167–1203), and the Gǎibìng Wǔyīn Jíyùn 改併五音集韻 (GWJ), a Jin rhyme dictionary from 1212, the chóngniǔ contrast shows a significant change in the MGZY. In both the QZ and the GWJ the chóngniǔ contrast exists after the labial initials in most phonological conditions (Nìng 1992; Féng 2001), but in the MGZY the contrast is lost in the syllables with labial initials, except in the syllables with

Old Mandarin: The Ménggǔ Zìyùn 蒙古字韻

255

no coda in Middle Chinese.4 For a discussion on the QZ, see Section 4.3.3, and for more on the intricacies of the GWJ itself, refer to Section 5.2.1. 6.4.4

The Loss of Final Consonant Codas

The stop codas are completely lost in the MGZY phonology. Historically, Middle Chinese syllables with -p, -t, -k coda become open syllable homophones without a stop coda. The finals of the Middle Chinese rù 入 syllables have changed to the finals of the corresponding syllables with -j, -w, or zero coda (Shen 2008a). This change should not be considered just a change of the finals as it actually necessarily indicates tonal changes as well, the loss of the rù tone. The MGZY adopts a phonological system that keeps many basic structural features of the Qièyùn. One of those features is the tonal system, different from that of northern Chinese of the time. The adjustments of the finals of the Middle Chinese rù syllables can provide evidence to prove the loss of the rù tone. In modern dialects, if a phonological system has the rù tone it also often has a special set of rù finals. In other words, the loss of the rù tone is not just a tonal phenomenon but a change of finals as well. In modern dialects the rù tone can exist without stop codas, but the stop codas always coexist with the rù tone. In the study of the Zhōngyuán Yīnyùn 中原音韻 (ZYYY) (as will be seen in Chapter 7), the existence of the rù tone has been controversial, but according to the final change of the rù syllables it is very unlikely that the rù tone existed in the base dialect of the ZYYY. The evidence from earlier Khitan materials also shows the similarity between the finals of rù and non-rù syllables. Some spellings even indicate that certain rù syllables have become qù syllables (Shen 2007). 6.4.5

Diphthongization of Middle Chinese Syllables with Coda -k

Middle Chinese -k syllables in the MGZY display a variety of diphthong and monophthong reflexes. Because of the advantage of hP’ags-pa spelling, the _ phonetic values of these reflexes can be obtained without going through reconstruction. As shown in Table 6.31, the different diphthong reflexes are determined by the nuclear vowel of the Middle Chinese -k syllables prior to this change. If the nuclear vowel was a back vowel, it is realized as a diphthong 4

In Table 6.26, -j, -t, and -k codas also maintain a contrast; however, at this point in time, -t and -k codas have disappeared, and -j coda is just beginning to form (under certain conditions, medials move to nucleus position, and displace -i nucleus to the final -j position). So for all intents and purposes, the only labial-initial syllables that maintain a chóngniǔ contrast have no coda.

256

A New Standard

Table 6.31 Diphthongization of syllables with Middle Chinese (MC) coda -k MC 各 腳 玃 摘 虢

-ɑk -jɐk -wjɐk -ɰæk -wɰak

MGZY > > > > >

MC -aw -ew -wɛw -aj -waj

各 gè < kɑk ‘each’ 腳 jiǎo < kjɐk ‘foot’ 玃 jué < kwjɐk ‘large ape’ 摘zhāi < ʈɰæk ‘pick’ 虢 guó < kwɰak ‘ancient state name’

德 覺 郭 隔 國

MGZY -ək -ɰɔk -wɑk -ɰæk -wək

> > > > >

-əj -jaw -waw -jaj -uj

德 dé < tək ‘morality’ 覺 jué < kɰɔk ‘wake up’ 郭 guō < kwɑk ‘outer wall’ 隔 gé < kɰæk ‘separate’ 國 guó < kwək ‘country’

with coda -w, and if the vowel was a non-back vowel, it realizes as a diphthong with coda -j (Shen 2005, 2008a). Middle Chinese -k syllables also have reflexes in monophthong forms. These monophthongs should be analyzed in the same way as diphthongs, although due to the surface form restriction they appear as monophthongs instead (see Rule 3 below) (Shen 2005). The changes of Middle Chinese rù syllables with -k can be summarized by the following three basic rules: 1. Coda Compensation Rule: After the loss of velar stop coda -k, a semivowel coda must compensate for it. Vk > VG, where G is an offglide or coda. 2. Frontness Assignment Rule: A front nuclei must be followed by a palatal semivowel -j and back nuclei must be followed by a labial semivowel -w. For the two central vowels -ə and -a, -ə behaves as a front vowel and -a as a back vowel (in the MGZY -ə is spelled as ‹ ħi › ) (e.g. ək > əj, ak > aw). 3. Homophonic Deletion Rule: If the nucleus and semivowel coda share the same basic phonological features, diphthongs reduce to monophthongs with the drop of the coda (e.g. ij > i, uw > u). Examples are given in Table 6.32. It should be noted that the codas -p and -t have no dipthongization, and they simply lose their coda and merge with other tones. 6.4.6

The Loss of the Parallel Relationship between yáng and rù Syllables

The syllables with nasal and stop codas in the Qièyùn system have a perfect oneto-one relationship. For each type of syllable with a nasal coda there is a stop

Old Mandarin: The Ménggǔ Zìyùn 蒙古字韻

257

Table 6.32 Middle Chinese -k syllable changes and their rules

Middle Chinese Pre-MGZY After Rule 1 After Rule 2 After Rule 3

各 gè

郭 guō

覺 jué

百 bǎi

德 dé

逼 bī

卜 bǔ

kɑk kak kaG kaw kɑw

kwɑk kwak kwaG kwaw kwaw

kɰɔk kjak kjaG kjaw kjaw

pɰæk pɛk pɛG paj paj

tək tək təG təj təj

pɨk pik piG pij pi

puk puk puG puw pu

Table 6.33 Change of relationship between yáng and rù syllables Middle Chinese -N (-m, -n, -ŋ) -C (-p, -t, -k)

MGZY > >

-N (-m, -n, -ŋ) -J (-j, -w), -Ø

– >

yáng (-m, -n, -ŋ) yīn (-j, -w, -Ø)

This can alternatively be seen as: yáng (-m, -n, -ŋ) yīn (-j, -w, -Ø) rù (-p, -t, -k)

coda counterpart. This relationship can be found between the syllables with -m, -n, -ŋ and the syllables with -p, -t, -k, respectively. This relationship is a trademark of Middle Chinese phonology but has totally disappeared in the MGZY. As seen in Table 6.33, rù syllables, after the loss of their stop coda, structurally become the same as yīn syllables with a glide coda or no coda at all. This change makes the MGZY final system very similar to the better-known ZYYY, such that the difference between the two is insignificant. 6.4.7

The Contrast of jan/jɛn and jaw/jɛw

The Chinese phonology of the Yuan period has a contrast between the syllables jiān 監 kjam and jiān 兼 kjɛm, jiàn 間 kjan and jiàn 見 kjɛn, jiāo 交 kjaw and jiāo 澆 kjɛw. The finals with -ja are the reflexes of Middle Chinese Rank-II syllables and the finals with -jɛ are the reflexes of Middle Chinese Rank-III and -IV syllables. This contrast existed in both the MGZY and ZYYY. The phonetic values are clearly represented in the hP’agspa spellings, shown in Table 6.34 and Table 6.35.

258

A New Standard

Table 6.34 The contrast between the finals of Rank-II and Rank-III/IV syllables MGZY

Modern Mandarin

Rank-II Rank-III

監 兼

kjam kjɛm

> >

tɕjɛn tɕjɛn

jiān jiān

‘supervise’ ‘combine’

Rank-II Rank-III

間 見

kjan kjɛn

> >

tɕjɛn tɕjɛn

jiàn jiàn

‘inbetween’ ‘see’

Rank-II Rank-III

交 驕

kjaw kjɛw

> >

tɕjaw tɕjaw

jiāo jiāo

‘cross’ ‘water’

The merger of the vowels -ɛ(-e) and -a is conditioned by the codas, ja > jɛ / _ m, n, while jɛ > ja /_ w. The vowel merges to match in backness with the coda. 6.4.8

The High Central (Apical) Vowel

The vowels of Modern Mandarin syllables such as zī 資, cí 雌, sī 思 and zhī 知, chī 蚩, shī 詩, rì 日 are called apical vowels. The term “apical vowel” was first seen in Bernhard Karlgren’s Études sur la phonologie chinoise (1915–1926). In this work he created the terms apico-gingivales and apicoalvéolaires and used phonetic symbols ‹ ɿ › and ‹ ʅ › to describe these sounds respectively. ‹ ɿ › represents the centralized /i/ after the alveolar series ts tsh s, and ‹ ʅ › represents the centralized /i/ after the retroflex series tʂ tʂh ʂ ɻ. These terms have been widely accepted by Chinese linguists, but whether they fit into a structural view of sound systems has not been seriously considered; phonologically they are high central vowels. Centralizaton of high vowels (apicalization) is a gradual process with different causes (Shen 2008b). In the MGZY the apical vowel is spelled as ‹ in the zhī 支 rhyme group. The changes observed are found under the ħi › following three phonological conditions: (a) syllables with the jīng-series initials (jīng 精 ts-, qīng 清 tsh-, cóng 從 dz-, xīn 心 s-, xié 邪 z-) in the zhī 支, zhī 脂, and zhī 之 rhymes, (b) syllables with initial zhuāng 莊 ʈʂ- series in the zhī 支, zhī 脂, and zhī 之 rhymes, and (c) rù syllables with initial zhuāng 莊 ʈʂ- series in the jī 緝, zhí 職, and zhì 質 rhymes. In the MGZY the syllables with zhī 知 ʈ- initial series, zhāng 章 tɕ- initial series, and rì 日 ɻ- initial had not shown any sign of change. Based on the initials involved we can term the possible changes as (1) jīng-centralization, (2) zhuāng-centralization, (3) zhāng-centralization (including the rì initial),

Old Mandarin: The Ménggǔ Zìyùn 蒙古字韻

259

Table 6.35 The changes between the Chinese of the Ménggǔ Zìyùn and Modern Mandarin (MM), with the data from Table 6.34 reorganized as examples of each change MGZY

Condition

MM

Example

-ja

before -m, -n

-jɛ

before -w

-ja

監 間 交

kjam kjan kjaw

> > >

tɕjɛn tɕjɛn tɕjaw

jiān jiàn jiāo

before -m, -n

-jɛ

before -w

-ia

兼 見 澆

kjɛm kjɛn kjɛw

> > >

tɕjɛn tɕjɛn tɕjaw

jiān jiàn jiāo

-jɛ

Table 6.36 Centralization of high vowels in the Ménggǔ Zìyùn Middle Chinese 精 jīng 知 zhī 莊 zhuāng 章 zhāng

tsʈtʂtɕ-

支 zhī

脂 zhī

之 zhī

祭 jì

緝 qī

質 zhì

櫛 zhì

職 zhí

昔 xī

1 4 2 3

1 4 2 3

1 4 2 3

(+) 4 4

(+) 4 2 4

(+) 4 4

2 -

(+) 4 2 4

(+) 4 4

Note: The presence of centralization is in italics. “(+)” indicates the syllables not involved in the centralization process.

and (4) zhī-centralization. The MGZY shows changes (1) and (2), as can be seen in Table 6.36. 6.4.9

Vowel Raising

In a comparison of the Middle Chinese and MGZY vowel systems, as shown in Table 6.37, the evidence of vowel raising is overwhelming. There are two vowel raising chains according to the frontness of the vowels. The front vowels have the changes, a > ɛ, ɛ > i; and back vowels, ɒ > ɔ, o > u. In the Qièyùn system the main vowel of the gē 歌 and má 麻 rhymes are low vowels, -ɑ and -a, respectively . These low vowels became higher. The main vowel of the gē 歌 rhyme became -ɔ, and the main vowel of the má 麻 rhyme became -ɛ after glide -j-. Although the main vowels of finals -ja and -jɛ are different, they are grouped in the same má 麻 rhyme group. Later in the ZYYY these two finals are separated into the jiā-má 家麻 and chē-zhē 車遮 rhyme

260

A New Standard

Table 6.37 A set of rules determining vowel raising in Yuan-era Mandarin MC rhyme má 麻 qí 齊 gē 歌 mó 模 yú 魚

Example 車 些 雞 低 歌 多 姑 都 居 豬

chē xiē jī dī gē duō gū dū jū zhū

MC ‘cart’ ‘some’ ‘chicken’ ‘low’ ‘song’ ‘many’ ‘aunt’ ‘capital’ ‘dwell’ ‘pig’

tɕhja sja kej tej kɑ tɑ kwo two kjɔ ʈiɔ

MGZY > > > > > > > > > >

MC tʂhɛ sɛ ki ti ko to ku tu ky tʂy

a a e e ɑ ɑ o o ɔ ɔ

MGZY > > > > > > > > > >

ɛ / j_ ɛ / j_ i / _j i /_ j o o u u y / j_ y / j_

Figure 6.3 A vowel chart demonstrating vowel raising in the Yuan era.

groups. But the vowel raising is evident in the MGZY. This vowel raising is roughly illustrated in Figure 6.3. As the result of this vowel raising in the MGZY, as well as similar changes in Modern Northern Mandarin, there are many syllables with high vowels and very few syllables with low vowels. Hundreds of syllables have final -i, but on the other hand the system lacks syllables such as ka and kha, creating significant phonological gaps. ka and kha syllables in Modern Mandarin are foreign loans, such as the kha syllable in kāfēi 咖啡 ‘coffee’ and kǎchē 卡車 ‘car, truck,’ and the ka syllable in 咖喱 gālí ‘curry.’ Vowel centralization is a part of this vowel raising, the high front vowel i in certain phonological conditions moving to the central position to become ɨ. As the Yuan standard phonology, the MGZY demonstrates many basic characteristics of Mandarin phonology, while, on the other hand its inherited conservative phonological frame and the spelling conventions of the hP’ags-pa _

Old Mandarin: The Ménggǔ Zìyùn 蒙古字韻

261

script made certain phonological characteristics less transparent. These questionable characteristics can be learned from the ZYYY, which represents the phonological system of the colloquial language of Yuan time, although it contains no phonetic information. By comparing the ZYYY, which will be analyzed in the next chapter, and the earlier Liao phonology based on the Khitan Lesser Script, many conservative characteristics of the MGZY can be determined.

7

Old Mandarin: The Zhōngyuán Yīnyùn 中原音韻

The much celebrated Zhōngyuán Yīnyùn (ZYYY) was compiled in 1324 by Zhōu Déqīng 周德清 (1277–1365), an influential writer of Yuan poetry (see Figure 7.1). This book is concise in comparison to rhyming books in the tradition of the Qièyùn 切韻 and Guǎngyùn 廣韻; the ZYYY is basically a list of homophonic characters with no glossary and no fǎnqiè 反切 provided. Like traditional rhyme books, the ZYYY provides phonological categories only. The phonetic values are “reconstructed” based largely on the phonetic values of the hP’ags-pa script in the Ménggǔ Zìyùn 蒙古字韻 (MGZY) (e.g. _ Yáng Nàisī 1981). Different from the official MGZY, this new rhyme book represents a complete reanalysis of a phonological standard without referring to the fixed phonological skeleton of Middle Chinese phonology. This new rhyming standard was meant for the writers of new style poems and operas, but not meant to replace the traditional rhyming standard used for old style poems and the imperial examination. The ZYYY established the northern dialect as a new phonological standard, representing a breakthrough in the rhyming tradition and a significant turning point in the history of the phonological standard. Because of its significance, the ZYYY has received much attention from historical phonologists, including scholars in the West (Stimson 1966; Hsueh 1975). The ZYYY documents a phonological system that had already existed in the north of China for centuries. Like the Qièyùn, it was not based on the phonology of any single dialect; the phonological distinctions could be from a variety of prestigious forms of Mandarin dialects at the time. Many basic phonological characteristics of the ZYYY can be traced back to the tenth century (Shen 2011). Taking this into account, the ZYYY should not be regarded as the beginning of Mandarin phonology but rather as the establishment of a new phonological standard and the beginning of the recognition of the dominance of Mandarin. The publication of the ZYYY should be realized as a confirmation of a pre-existing standard. Its publication also indicates the dissatisfaction with the traditional rhyming standard, including contemporary official publications such as the Ménggǔ Yùnlüè 蒙古韻略 and the MGZY. 262

Old Mandarin: The Zhōngyuán Yīnyùn 中原音韻

263

Figure 7.1 A page from the Zhōngyuán Yīnyùn 中原音韻. It reads vertically from right to left. The first line is the title of the book, the second line is information about the author, the third line indicates the first rhyme, represented by two characters dōngzhōng 東鍾, the fourth line indicates the tonal category píngshēng 平聲, the fifth line indicates the subcategory of the tone yīn 陰, and the sixth to twelfth lines are a list of the characters of this category, with a ○ between the homophonic groups.

264

A New Standard

The ZYYY consists of two parts: the first part, in the form of a rhyming book, consists of lists of homophonic groups arranged according to rhyme groups, tones, and initials; and the often ignored second part contains many notes and examples showing how to follow the phonological standard in practice. In the first part, 5,866 Chinese characters are divided into nineteen rhyme groups. Each rhyme group is further divided according to the tones. Then the characters are further divided into homophonic groups according to initials. In total, there are 1,586 homophonic groups, each of which represents a phonologically distinct syllable. This arrangement is different from the Qièyùn/Guǎngyùn tradition, which first divides the characters into different volumes according to tones, and then into rhymes and homophonic groups. It also is different from the MGZY in terms of the arrangements within rhyme groups, as illustrated in Table 7.1. Each rhyme group is labeled with two representative characters from the rhyme group itself. They are listed in accordance with the order of the ZYYY in Table 7.2. 7.1

The Base Dialect

Although the dialect of Dadu had become an influential standard in the Yuan dynasty, the standard Chinese of the previous dynasty, the Song, was still Table 7.1 A comparison of the arrangement of the Qièyùn, Ménggǔ Zìyùn, and Zhōngyuán Yīnyùn Qièyùn

MGZY

Tones

ZYYY

Rhymes Rhymes

Rhymes Onsets

Onsets

Tones Tones

Onsets

Table 7.2 All the rhyme groups of the Zhōngyuán Yīnyùn 1 2 3 4 5 6 7 8 9 10

東鍾 江陽 支思 齊微 魚模 皆來 真文 寒山 桓歡 先天

dōng-zhōng jiāng-yáng zhī-sī qí-wēi yú-mó jiē-lái zhēn-wén hán-shān huán-huān xiān-tiān

11 12 13 14 15 16 17 18 19

蕭豪 歌戈 家麻 車遮 庚青 尤侯 侵尋 監咸 廉纖

xiāo-háo gē-gē jiā-má chē-zhē gēng-qīng yóu-hóu qīn-xún jiān-xián lián-xiān

Old Mandarin: The Zhōngyuán Yīnyùn 中原音韻

265

considered prestigious. This complex linguistic situation was reflected in the phonological treatment of the Zhōngyuán Yīnyùn 中原音韻 (ZYYY). The nature of the ZYYY has been controversial with regard to two related issues: (1) whether the phonology is mainly based on one dialect, and if so, then (2) which dialect is its base. Many scholars (e.g. Lù 1946a; Yáng Nàisī 1981) prefer not to identify a single dialect as the base, stating that the ZYYY represents a common form of the northern dialect of Mandarin as indicated by the title of the book, Zhōngyuán Yīnyùn ‘Rhymes of the Central Plain’. However, other scholars try to identify a single dialect that serves as the base of the ZYYY phonology. There are two main views concerning the base dialect. One is that the ZYYY is based on the phonology of Dadu (Wáng Lì 1985; Nìng 1985; Gěng 2007), the capital of the Yuan dynasty, and the other is that the ZYYY is based on the phonology of the He-Luo area (modern Luoyang area in Henan province), the heartland of the Northern Song dynasty (Lǐ Xīnkuí 1962). Both views are supported by cultural, social, and linguistic evidence, but both sides try to minimize the obvious supporting evidence of the other side. The strongest evidence in favor of the Dadu dialect is the diphthong reflexes of Middle Chinese syllables with -k codas,1 especially those with -w codas in the ZYYY.2 In the ZYYY some Middle Chinese -k syllables are arranged in the rhyme groups that have the -w or -j codas. This is a similar phenomenon to that in the Ménggǔ Zìyùn 蒙古字韻 (MGZY). The Middle Chinese rù 入 tone syllables (the first three homophonic groups are given as examples) found in the xiāoháo 蕭豪 rhyme group are supplied in Table 7.3. The non-rù syllables are the Middle Chinese syllables with -w coda. In the ZYYY the Middle Chinese rù tone syllables are labeled as rù shēng zuò X shēng 入聲作X聲 (rù tone (syllables) are as X tone), where X can be either píng (first row), shǎng (second row), or qù tone (third row). This special type Table 7.3 Middle Chinese rù tone syllables found in the xiāoháo rhyme group 蕭豪-aw

1 2

入聲作平聲 入聲作上聲

tȿ- 濁濯鐲擢 kj- 角覺腳桷

入聲作去聲

Øj- 岳樂藥約 躍鑰瀹

t- 鐸度踱 tȿ- 捉卓琢斫酌 繳灼爍鑠㜰 n- 掿諾

p- 薄箔泊博 tshj- 鵲雀趞 m-

末幕漠寞莫 沫

... ... ...

Two of them, 末 mò and 沫 mò, are actually Middle Chinese -t syllables. They could have changed from -t to -k, then had the same -k > -w change as the syllables with the -k coda. Diphthong reflexes of -k coda characters are of particular interest to historical linguists, and are discussed in more depth in Section 4.3.2, Section 5.1.2, Section 5.2.2.1, and Section 6.4.5.

266

A New Standard

Table 7.4 Some characters that maintain the -aw reflex from Dadu dialect paw 薄 bó < bɑk ‘thin’ taw 度 dù dɑk < ‘degree’

paw 博 bó < bɑk ‘broad’ tʂaw 濁 zhuó < ɖɰɔk ‘muddy’

faw 縛 fù < bwjɐk ‘tie’ xaw 鶴 hè < ɦɑk ‘crane’

Table 7.5 The presence and absence of -w coda in the relevant modern dialects

ZYYY Beijing Shunping Changchun Luoyang Kaifeng







‘thin’

‘chisel’

± ± ± ± – –

± ± + ± – –







‘medicine’ ‘foot’

‘wake up’

‘learn’

± ± + + – –

+ ± ± ± – –

± ± ± ± – –

+ ± + + – –

Note: + means a -w coda is present, whereas – means that it is not present, and ± means that there is a dual pronunciation, one with a -w coda and one without. Beijing is in the same location as the Dadu area; Shunping and Changchun are close to the Dadu area; Luoyang and Kaifeng are in the He-Luo area.

of reflex is only found in modern dialects in Hebei, Liaoning, Jilin, Heilongjiang provinces (Liú Shúxué 2000; Gāo 2009). The -aw pronunciation still exists in the characters shown in Table 7.4. Modern dialects of these two areas, Dadu and He-Luo, show different reflexes of Middle Chinese syllables with -k coda (see Table 7.5). The ZYYY also recorded the alternative pronunciation of these syllables in the gē-gē 歌戈 rhyme group with main vowel -o. Many words that had Middle Chinese -k now have dual pronunciation in the ZYYY. This phenomenon seems to support the argument for a base dialect of either Dadu or He-Luo, but under closer examination (Nìng 1991) it turns out that only forty-four out of the list of Middle Chinese -k syllables have dual pronunciation, and the other fifty-nine have only the reflexes with -w coda. Of course, this piece of evidence seems to favor the Dadu argument because the -w forms are regular, and -o forms are incomplete and from the borrowed literary layer based on the then He-Luo dialect. However, the other pieces of evidence strongly support the argument for the He-Luo dialect. The Middle Chinese Rank-III syllables in the dōng-zhōng 東鍾 rhyme group are still pronounced with palatal medial in the Luoyang dialect.

Old Mandarin: The Zhōngyuán Yīnyùn 中原音韻

267

The pronunciation of the syllables in the qí-wēi 齊微 rhyme group is closer to the modern Luoyang dialect rather than the Beijing dialect, as seen in Table 7.6. Because some exceptions to the changes in the ZYYY cannot be analogically determined by Middle Chinese phonological categories, they must be based on a real dialect. As shown in Table 7.6, the similar exceptions between the Luoyang dialect and the ZYYY – but not with the Beijing dialect – imply that these pronunciations most likely come from the Luoyang dialect (Lǐ Xīnkuí 1962). The first three examples in Table 7.7 show the irregular changes found in both the ZYYY and the modern Luoyang dialect. The other six examples show that the irregular changes found in the modern Beijing dialect are not found in either the ZYYY or the Luoyang dialect. In both regular and irregular changes, the Luoyang dialect is closer to the ZYYY. Irregular changes are very special dialect identifiers. Such information cannot be deduced solely from other systems, they can only be learned from the system Table 7.6 A comparison of rhyme ending realizations between the Zhōngyuán Yīnyùn, and the modern Luoyang and Beijing dialects ZYYY

Luoyang

Beijing

dōng-zhōng 東鍾 rhyme group 濃龍隆癃踪蹤從鬆訟 Final 容融榮溶蓉瑢鎔 Whole syllable

-juŋ juŋ

-juŋ juŋ

-uŋ ɻuŋ

qí-wēi 齊微 rhyme group 非菲飛妃肥匪翡 餒內雷壘淚 彼卑丕筆

-i -wej -ej

-i -wej -ej

-ei -ej -i

Final Final Final

Table 7.7 The irregular changes of modern Luoyang and Beijing dialects

聯 娠 深 薛 詞 囚 液 營 頃

lián shēn shēn xuē cí qiú yè yíng qǐng

‘connect’ ‘pregnant’ ‘deep’ ‘marsh grass’ ‘word’ ‘prisoner’ ‘liquid’ ‘military camp’ ‘just’

ZYYY

Luoyang

Beijing

-ɥɛn tʂhən tʂən -jɛ sɨ siw i juŋ khjuŋ

-ɥɛn tʂhən tʂən -jɛ sɨ siw i juŋ tɕhjuŋ

-jɛn ʂən ʂən -yɛ tshɨ tɕhjəw jɛ iŋ tɕhiŋ

268

A New Standard

Table 7.8 Contrast of -on and -wan in the Zhōngyuán Yīnyùn

Rank-I

-on

Rank-II

-wan

k-

x-

Ø-

p-

官 guān ‘official’ 關 guān ‘close’

桓 huán ‘surname’ 還 huán ‘return’

豌 wān ‘peas’ 彎 wān ‘curve’

搬 bān ‘move’ 班 bān ‘surname’

itself. Thus, irregular changes provide very strong evidence for the argument that the main body of the ZYYY phonology is more likely based on the HeLuo dialect. In comparison, the Middle Chinese rù syllables with -w, the Dadu dialect feature, appear in a regular and predictable way in the ZYYY. During the Jin and Yuan dynasties the He-Luo dialect must have been strongly influenced by the dialect of the capital of the time. Thus, it is quite likely that the ZYYY reflects a version of accented Dadu dialect spoken in the He-Luo area. 7.1.1

The Influence of Southern Mandarin

It is also worth noting that the phonology of the ZYYY might also integrate some Southern Mandarin phonological distinctions into its system. The distinction between Middle Chinese hékǒu 合口 syllables with -n coda of Rank-I and Rank-II is not likely a phonological feature of either Dadu or He-Luo subdialects, but a distinction from Southern Mandarin. In the ZYYY there is a contrast of -on and -wan, a distinction also present in the MGZY. Table 7.8 illustrates this contrast. Modern Mandarin and non-Mandarin dialects indicate that this contrast is generally a geographically southern feature (Běijīng Dàxué 2003); this comparison is shown in Table 7.9. The Gǎibìng Wǔyīn Jíyùn 改併五音集韻 (1212) does have such a contrast, grouping these Middle Chinese Rank-I and Rank-II syllables together.3 After the ZYYY, rhyme dictionaries and charts have a different treatment of this distinction. Phonological works representing Southern Mandarin dialects, such as the Hóngwǔ Zhèngyùn 洪武正韻 (1375) and Yùnlüè Yìtōng 韻略易通 (1442), show this distinction (as is discussed in Section 9.2 and Section 9.3). But the phonological works reflecting northern Mandarin dialects, such as the Děngyùn Tújīng 等韻圖經 (1602) and Yùnlüè Huìtōng 韻略匯通 (1642), do not show this distinction (as is discussed in Section 9.6) (Tián 2004).

3

“寒桓” 同用。

Old Mandarin: The Zhōngyuán Yīnyùn 中原音韻

269

Table 7.9 A comparison of rank contrast across many modern dialects Northern Mandarin

官 ‘official’ 關 ‘close’

Southern Mandarin

Xiang

Wu

Gan

Yue

BJ

TY

XA

JN

CD

HF

YZ

CS

SZ

NC

GZ

kuan kuan

kuæ̃ kuæ̃

kuæ̃ kuæ̃

kuæ̃ kuæ̃

kuan kuan

kʊ kuæ̃

kuõ kuæ̃

kõ kuan

kuø kuɛ

kuɔn kuan

kun kuan

Note: BJ = Beijing; TY = Taiyuan; XA = Xi’an; JN = Jinan; CD = Chengdu; HF = Hefei; YZ = Yangzhou; CS = Changsha; SZ = Suzhou; NC = Nanchang; GZ = Guangzhou.

Table 7.10 Dual pronunciation of some Chinese characters 剝

‘to peel’

paw/po



‘to collapse’

puŋ/pəŋ

It is likely that the main characteristics of Mandarin – the split of the píng tone, the devoicing of Middle Chinese voiced obstruents, the loss of stop coda and tonal distinction of Middle Chinese rù syllables – were the shared features of a large Mandarin-speaking area including large cities such as Luoyang, Dadu, and Nanjing. If the phonological differences of these subdialects are mainly limited to the reflexes of Middle Chinese -k syllables and Rank-I and -II distinction of Middle Chinese hékǒu syllables with coda -n, the logical conclusion is that these dialects were in fact very similar Mandarin subdialects. In the phonology of the ZYYY some characters have dual pronunciation in distinct rhyming groups. For example, the character bō 剝 is listed in both the xiāo-háo 蕭豪 and gē-gē 歌戈 rhyme groups, and the character bēng 崩 in both the dōng-zhōng 東鍾 and gēng-qīng 庚青 rhyming groups (see Table 7.10). The dual pronunciation indicates competing phonological standards; however, the different standards seem to stem from different causes. The dōng-zhōng 東鍾 and gēng-qīng 庚青 contrast was definitively between spoken language and written language, whereas the xiāo-háo 蕭豪 and gē-gē 歌戈 contrast was a difference in pronunciation between two different dialects; the xiāo-háo 蕭豪 and gē-gē 歌戈 contrast, however, was also viewed as the difference between colloquial and literary readings (Wáng Lì 1985: 384). Both forms of the dual pronunciations reflect the standard pronunciation. In contrast, the phonological features of nonstandard dialects, be they Mandarin or non-Mandarin, were perceived very differently: as problematic pronunciation. In the second part of the ZYYY, there is a list called Gèfāng Yǔbìng 各方語病 ‘Problematic Pronunciation of Different Locations.’

270

A New Standard

7.2

The Zhèngyǔ Zuòcí Qǐlì 正語作詞起例

The second part of the Zhōngyuán Yīnyùn 中原音韻 (ZYYY) is labeled Zhèngyǔ Zuòcí Qǐlì ‘Examples of Standard Language and Poem Composition,’(see Figure 7.2) which are comments made by the author regarding various issues, including the arrangement of the book, the similar but distinct syllables, and the standard formats of poetry. 7.2.1

Dialect Features

The Zhèngyǔ Zuòcí Qǐlì contains very useful information about the phonological features of contemporaneous dialects. After examining the minimal phonological pairs listed, it seems that these contrasts are made for Mandarin speakers rather than for the speakers of non-Mandarin dialects; phonological features of non-Mandarin dialects are much more varied and plentiful. This list shows that Zhōu Déqīng’s 周德清 primary interests were to correct the nonstandard pronunciation in Mandarin. All the listed mispronunciations can be found in the subdialects of Modern Mandarin or the Gan dialect, which were the dialects Zhōu Déqīng was familiar with. These contrasts therefore also provide us with useful information about the different characteristics of different Mandarin subdialects during the Yuan time (see Table 7.11). Zhōu Déqīng listed 241 pairs of characters that should be distinguished according to the ZYYY. Of these pairs, 200 are for distinction of finals and 39 are for distinction of initials. One is for the character 造 zào with dual tones and another pair is 薄 báo and 箔 bó, where both words have dual pronunciations of their finals. Take, for example, Figure 7.2. On this page seven pairs of characters of dōngzhōng 東鍾 are listed. They are zōng yǒu zōng 宗有蹤 (ts-), sōng yǒu sōng 松有鬆 (s-), lóng yǒu lóng 龍有籠 (l-), nóng yǒu nóng 濃有膿 (n-), lǒng yǒu lóng 隴有櫳 (l-), sòng yǒu sòng 送有訟 (s-), zòng yǒu zōng 從有綜 (ts-). According to Zhōu Déqīng each of these seven pairs has the same initial but different finals (with or without medial -j-). It is quite evident that Zhōu emphasizes various contrasts differently. He used more than 155 pairs out of 200 to distinguish the three nasal codas. Of the 155 pairs, he used 94 to distinguish the contrast between the nasal Table 7.11 Characteristics of different Mandarin subdialects during the Yuan era Eastern Mandarin: Northern Mandarin:

ts-on

= =

tʂ-wan

l-

=

n-

-im = -in = -iŋ

-juŋ

=

-uŋ

Old Mandarin: The Zhōngyuán Yīnyùn 中原音韻

271

Figure 7.2 A page from the second part of the Zhèngyǔ Zuòcí Qǐlì 正語作詞 起例. After the heading of the dōng-zhōng 東鍾 rhyme, seven pairs of words are listed in the format X yǒu 有 Y ‘X is also pronounced as Y’ as a dialectal pronunciation.

272

A New Standard

coda -m and -n (Nìng 1985). It should be noted that if a contrast is stable and consistent, it requires only a few examples to illustrate; a scholar will assume a pattern and extrapolate from it. However, if a contrast is disappearing, and thus no set rule can determine which syllables follow a contrast and which ones do not, all the syllables in question must be distinguished. Zhōu most likely made his list according to this thought: the n-/l- contrast and the ts-/tʂ- contrast were not disappearing at the time of the ZYYY, whereas the -m coda was well on its way to a full merger with -n by the time Zhōu published the book. Table 7.12 illustrates some of the contrasts Zhōu listed. It is interesting to note that among the seven -juŋ vs. -uŋ pairs listed in Table 7.12, which final should be listed first and which final second is not consistent. Five pairs (2, 3, 4, 5, and 7) are in the order of -juŋ vs. -uŋ and two pairs (1 and 6) are in the order of -uŋ vs. -juŋ. In most cases, for each pair in the same group, the first character always represents an initial or a final of a certain type, and the second character always represents an initial or final of another certain type. The fact that this format was not followed in this case probably indicates that the -uŋ and -juŋ distinction was already not so clear to the author himself at the time; this phenomenon occurs with the -on vs. -wan contrast as well. 7.3

The Phonetic Reconstruction

Since the early studies by Japanese scholars – Ishiyama (1925), and Hattori and Tōdō (1958) – and the study done by Zhào Yìntáng (1936), the significance of the Zhōngyuán Yīnyùn 中原音韻 (ZYYY) as one of the earliest complete records of Mandarin phonology became evident to Chinese phonologists, many of whom have reconstructed the phonetic values for the phonological categories of the ZYYY. Despite some differing details in the reconstructions and different theoretical considerations, the main phonological characteristics of this system are not controversial. The phonetic values of the phonological categories are reconstructed based on the internal structure of the ZYYY’s framework and on the phonetic transcription of the contemporary rhyme book Ménggǔ Zìyùn 蒙古字韻 (MGZY) (Lù 1946a; Yáng Nàisī 1981; Lǐ Xīnkuí 1983).4 7.3.1

Initials (21)

The initial system of the ZYYY is clearly distinct from the one of Middle Chinese. The Middle Chinese three-way contrast of stops and affricates was 4

It should be noted that since the hP’ags-pa spellings of the MGZY were not very well studied at _ not very accurate or even correct. This problem thus affects the time, some interpretations were to varying degrees the assessment of phonetic values in the ZYYY.

273

胡 xu

貞 tʂiŋ

真 tʂin 灘 than 堅 kjɛn

蹤 tsjuŋ 攏 luŋ

之 tʂɨ

梅 mej

關 kwan

Ø- ¼ 6 x- (1 pair) 吳 Øu 6¼

-n 6¼ ŋ (61 pairs) 真 tsin 6¼

-m 6¼ -n (94 pairs) 針 tʂim 6¼ 貪 tham 6¼ 兼 kjɛm 6¼

-juŋ 6¼ -uŋ (7 pairs) 宗 tsuŋ 6¼ 隴 ljuŋ 6¼

-i 6¼ -ɨ (5 pairs) 知 tʂi 6¼

-i 6¼ -ej (8 pairs) 迷 mi 6¼

-on 6¼ -wan (5 pairs) 官 kon 6¼ 慢 mwan

脾 phi

恥 tʂhi

松 sjuŋ 送 suŋ

金 kim 南 nam 添 thjɛm

因 Øin

殘 tshan 災 tsaj







6¼ 6¼

6¼ 6¼ 6¼



6¼ 6¼

幔 mon

裴 phej

齒 tʂhɨ

鬆 suŋ 訟 sjuŋ

斤 kin 難 nan 天 thjɛn

英 Øiŋ

潺 tʂhan 齋 tʂaj

患 xwan

迷 mi

世 ʂi

龍 ljuŋ 從a tsjuŋ

沈 tʂhim 咸 xjam 險 xjɛm

君 kyn

蘇 su 絲 sɨ

a

Note: The numbers in parentheses are the number of total pairs relevant to a contrast. According to the dōng-zhōng 東鍾 part of the ZYYY, this character should be zòng 縱 instead.

弩 nu

l- 6¼ n-, (1 pair) 魯 lu 6¼

ts- 6¼ tʂ-, tsh- 6¼ tʂh-, s- 6¼ ʂ- (20 pairs) 珊 san 6¼ 山 ʂan 猜 tshaj 6¼ 差 tʂhaj

Table 7.12 Some contrasts listed by Zhōu Déqīng in the Zhōngyuán Yīnyùn







6¼ 6¼

6¼ 6¼ 6¼



6¼ 6¼

緩 xon

媚 mej

市 ʂɨ

籠 luŋ 綜 tsuŋ

陳 tʂhin 閒 xjan 顯 xjɛn

扃 kyŋ

踈 ʂu 師 ʂɨ

慣 kwan

閉 pi

智 tʂi

濃 njuŋ

審 ʂim 探 tham 點 tjɛm

溫 Øun

粗 tshu 死 sɨ









6¼ 6¼ 6¼



6¼ 6¼

貫 kon

避 pej

志 tʂɨ

膿 nuŋ

哂 ʂin 炭 than 典 tjɛn

泓 Øuŋ

初 tʂhu 史 ʂɨ

274

A New Standard

reduced to a two-way contrast; the Middle Chinese two-way contrast between voiced and voiceless fricatives completely disappeared. Table 7.13 demonstrates the new simplified system. 7.3.1.1 The wēi 微 and rì 日 Initials The devoicing of Middle Chinese voiced obstruents indicates that the initial system does not have voiced phonation. There should be no voiced obstruents (stops, affricates, or fricatives) reconstructed in the system. This phonological feature affects two initials that are commonly reconstructed as voiced fricatives in many proposals. One is the initial wēi 微 ɱ-, for syllables such as wáng 亡 (Middle Chinese: mwjɐŋ), wú 無 (Middle Chinese: mjo), wàn 萬 (Middle Chinese: mwjɐn), and the other is the initial rì 日 ȵ- for syllables such as rǎng 嚷 (Middle Chinese: ȵjɐŋ), rú 如 (Middle Chinese: ȵjɔ), rǎn 染 (Middle Chinese: ȵjɛm). The wēi 微 initial is reconstructed as ɱ- and the rì 日 initial is reconstructed as ȵ- in the Middle Chinese system. At this stage in the development of Chinese, syllables with the Middle Chinese wēi 微 initial are often reconstructed with initial v-, the voiced counterpart of f-. But since voiced phonation is not a feature of the system, this labiodental consonant cannot be a voiced fricative. Middle Chinese wēi 微 syllables are, however, still in contrast with the syllables with initial f- and zero initial, so it could have undergone neither devoicing (to become f-) nor elision (to become Ø-). In the ZYYY the following groups have the minimal triplets: wáng 亡 mwjɐŋ / fáng 房 bwjɐŋ / wáng 王 ɦwjɐŋ; wú 無 mjo / fú 扶 bjo / wú 吾 ŋwo; wàn 萬 mwjɐn / fàn 飯 bwjɐn / wàn 腕 Øwɑn, illustrated in Table 7.14. The second and third syllables in these triplets have f- and w- to Table 7.13 The initial system of the Zhōngyuán Yīnyùn

Labial Alveolar Retroflex Velar

Unaspirated

Aspirated

Nasal

Fricative

Approximant

p t ts tʂ k

ph th tsh tʂh kh

m n

f

ʋ l

s ʂ x

ɻ Ø

ŋ

Table 7.14 Minimal triplets imply a three-way contrast Triplet 1 Triplet 2 Triplet 3

亡 無 萬

ʋaŋ ‘death’ ʋu ‘not’ ʋan ‘ten thousand’

房 扶 飯

faŋ ‘house’ fu ‘hold on’ fan ‘cooked rice’

王 吾 腕

Øwaŋ ‘king’ Øu ‘I, my’ Øwan ‘wrist’

Old Mandarin: The Zhōngyuán Yīnyùn 中原音韻

275

start the syllables, respectively. The initial consonant of the first syllable therefore must be reconstructed as ʋ-, a labiodental approximant. This way, the wēi 微 initial is morphed into a sonorant (and thus is not an obstruent), and does not require devoicing to stay in the phonology of the ZYYY. At this stage in the development of Chinese, syllables with the Middle Chinese rì 日 initial are commonly reconstructed with initial ʒ- if the place of articulation is determined to be palato-alveolar, or as ʐ- if it is a retroflex consonant. But, again, the problem is that both ʒ- and ʐ- are voiced obstruents. The reconstruction for this initial should be a retroflex approximant ɻ-. The mechanism that causes these changes is the same. As Mandarin evolved to not allow voiced phonation types in obstruents, voiced obstruents must merge with their unvoiced counterparts. The wēi 微 ɱ- and rì 日 ȵ- initials were already sonorants, albeit denasalized from their Middle Chinese values, and so were not required to change phonation at all. 7.3.1.2 The Value of the Retroflex Series One of the main concerns of initial reconstruction is the phonetic value of the retroflex series, as seen in Table 7.13 at the beginning of Section 7.3.1. As shown in Table 7.15, some linguists reconstruct this series as palatal consonants,5 some reconstruct it as retroflex consonants, while others reconstruct it with two allophonic values, palatal and retroflex. Under this allophonic reconstruction, palatal initials appear in front of palatal medials j- and ɥ- and their vowel equivalents, the high front vowels -i and -y. Retroflex initials appear elsewhere. In the ZYYY system, the syllables in Table 7.15 are in contrast. The argument for the purely palatal and two allophonic variation reconstructions is based on the difficulty of articulation of a retroflex initial in front of a high front feature of the medial or the main vowel. However, Mài (1991) demonstrates that there is no real evidence to support the palatal argument Table 7.15 A comparison of different hypotheses for the value of the retroflex series 罩 zhào ‘cover’ Luó, Yáng tʃaw 6¼ Zhào, Lĭ tʂaw 6¼ Lù, Nìng tʂaw 6¼

趙 zhào

抄 chāo

‘surname’ tʃjɛw tʂjɛw tʃjɛw

‘copy’ tʃhaw tʂhaw tʂhaw

6 ¼ 6¼ 6 ¼

超 chāo

哨 shào

紹 shào

‘surpass’ tʃhjɛw tʂhjɛw tʃhjɛw

‘whistle’ ʃaw 6¼ ʂaw 6¼ ʂaw ¼ 6

‘carry on’ ʃjɛw ʂjɛw ʃjɛw

Source: Lù 1946a; Yáng Nàisī 1981; Mài 1991.

5

The palatal series in this reconstruction is actually represented by the palato-alveolar series tʃtʃh- ʃ- rather than the palatal series cç- cçh- ç- or alveopalatal series tɕ- tɕh- ɕ-.

276

A New Standard

based on the historical documents and articulatory phonetics; as a result, this book uses the purely retroflex reconstruction. 7.3.2

Finals (47)

All the characters in the ZYYY are arranged into the nineteen rhyme groups as discussed at the beginning of Chapter 7 and seen in Table 7.2. Although most scholars are fully aware that the phonological system of the literary language is not a single dialect, their reconstructions always appear as a coherent system of a single dialect. Every phonological contrast therefore must be distinguished from each of the others with different phonetic realizations. While it is not impossible to do so, such a reconstructed system should not be considered as a phonology that actually existed. The reconstructed values in Table 7.16 are based on Yáng Nàisī (1981) and Nìng (1985). The modifications made are as follows: (1) Non-vowel symbols are used for the elements in the medial and coda positions to make the positional difference of nucleus from non-nucleus obvious. As an example, the ambiguity of a sequence ‹ iu › can be clarified as either the sequence of medial + main vowel ju or main vowel + coda iw. (2) The apical vowels are Table 7.16 Reconstructed values of finals in the Zhōngyuán Yīnyùn, organized by rhyme group 東鍾 江陽 支思 齊微 魚模 皆來 真文 寒山 桓歡 先天 蕭豪 歌戈 家麻 車遮 庚青 尤侯 侵尋 監咸 廉纖

dōng-zhōng jiāng-yáng zhī-sī qí-wēi yú-mó jiē-lái zhēn-wén hán-shān huán-huān xiān-tiān xiāo-háo gē-gē jiā-má chē-zhē gēng-qīng yóu-hóu qīn-xún jiān-xián lián-xiān

uŋ aŋ ɨ i aj ən an on aw o a əŋ əw əm am

juŋ jaŋ ej jaj in jan jɛn jaw, jɛw jo ja jɛ iŋ jəw im jam jɛm

waŋ wi u waj un wan

y yn

ɥɛn waw wo wa wəŋ

ɥɛ yŋ

Note: The columns represent the four sìhū system that can be extrapolated from evidence discussed in this section.

Old Mandarin: The Zhōngyuán Yīnyùn 中原音韻

277

Table 7.17 A comparison of the vowel systems of the Zhōngyuán Yīnyùn and Ménggǔ Zìyùn ZYYY i•y ɛ

ɨ (ə) a

MGZY u o

i•y ɛ

ɨ a

u o ɒ

Table 7.18 Summary of the vowel system in the Zhōngyuán Yīnyùn -Ø: -j: -w: Medial system:

iɛyuoɨa ea ɛəa

-m: -n: -ŋ: -j-, -w-, -ɥ-

iɛəa iɛyuoəa iyuəa

replaced with central vowels. These values represent the phonological contrasts and should not be understood as exact phonetic realizations. The main vowels form a system with three degrees in vowel height. Vowels ɨ and ə are two conditional variations. This system is very similar to that of the MGZY (see Table 7.17). The only difference is the vowel -ɒ. In the ZYYY -ɒ in -ɒŋ has changed to -wa in -waŋ.6 Table 7.18 is a summary of the vowel system in the ZYYY. The subsystems of vowels and medial system are also analyzed. In many reconstructions -j- and -w- are used as separate medials (e.g. jūn 君 kiuən (kjwən), xuān 喧 xiuɛn (xjwɛn)). In some reconstructions two medials, -j- and -w-, are used where in other reconstructions one medial, -ɥ-, is used. The reason for this difference is as follows. Medial -ɥ- and its main vowel equivalent y should appear together in any system (Yáng Nàisī 1981: 39; Táng 1991: 174–175). Since there is no vowel y in the yú-mó 魚模 and dōng-zhōng 東鍾 rhyme groups, there should be no medial -ɥ- in the system. But this argument is not very convincing simply because there is no way to prove that the yú-mó 魚模 or dōng-zhōng 東鍾 rhyme groups had only main vowel -u, but did not have main vowel -y. In the zhēn-wén 真文 rhyme group rhymes -ən, -in, -un, and -yn can be grouped together, so why is it that -u and -y cannot be grouped together in the yú-mó 魚模 and dōng-zhōng 東鍾 rhyme groups? It 6

-ɒ is a back rounded vowel; when the vowel system reduced, the rounded -ɒ became a medial wand a vowel -a.

278

A New Standard

Table 7.19 The vowel system of the Zhōngyuán Yīnyùn before null and -n codas -Ø:

i•y ɛ

ɨ

u o

i•y ɛ

-n:

a

ə a

u o

Table 7.20 The vowel system of the Zhōngyuán Yīnyùn before -ŋ coda -ŋ:

i•y

u ə a

is likely that in the ZYYY system there is a -y in the main vowel slot as well as its counterpart -ɥ- in the medial slot. In other words, the ZYYY has developed the sìhū system already (see Section 1.4.7). As in the common syllable structure of Mandarin, there is only one medial slot. The single medial slot will not allow two medials -j- and -w- in the reconstruction. The reconstruction of medial -iu- (-jw-) is thus phonetically unrealistic. The maximum vocalic contrast exists in the syllables with -n coda or no coda. They form a three-degree height system. As shown in Table 7.19, in front of -Ø there are high vowels -i -ɨ -y -u, mid vowels -ɛ -o, and low vowel -a. In front of -n there are high vowels -i -y -u, mid vowels -ɛ -ə -o, and low vowel -a. The three-degree height of vowels is a system very typical of Mandarin phonology. It should be pointed out that ɨ and ə are phonologically in complementary distribution, -ɨ in front of -Ø and -ə in front of -n, therefore they are conditional variations of the same phoneme. In front of -ŋ there are high vowels -i -y -u, mid vowel -ə, and low vowel -a. As mentioned in the discussion of the MGZY in Section 6.4.2, the vowel system in front of the velar nasal coda -ŋ is typical of the Modern Mandarin system (see Table 7.20). 7.4

Innovative Characteristics of the Zhōngyuán Yīnyùn 中原音韻

In comparison with the Qièyùn 切韻, just as has been exhibited in Chapter 6 with the Ménggǔ Zìyùn 蒙古字韻 (MGZY), the Zhōngyuán Yīnyùn (ZYYY) shows a number of important innovative features that indicate that the basic characteristics of Mandarin phonology had been established before the time of

Old Mandarin: The Zhōngyuán Yīnyùn 中原音韻

279

the book’s publishing. The innovative features are summarized in Sections 7.4.1–7.4.8. 7.4.1

Devoicing of the Middle Chinese Voiced Obstruents in the Initial Position

The Middle Chinese voiced stops, fricatives, and affricates become voiceless. Whether the stops and affricates merge with unaspirated or aspirated counterparts depends on the total categories. As shown in Table 7.21, the Middle Chinese voiced stops and affricates change to aspirated counterparts in the syllables with píng 平 tone but change to unaspirated counterparts in the syllables with non-píng tones (zè 仄 tones). As illustrated in Table 7.22, in the syllables with qù tone, the Middle Chinese voiced and voiceless unaspirated stops and affricates become homophones. This change is one of the most significant changes from Middle Chinese to Mandarin phonology and becomes the most fundamental phonological characteristic of Mandarin phonology. 7.4.2

The Emergence of the Labiodental Fricative

The Middle Chinese bilabial stops change to labiodental fricatives under certain phonological conditions (with Middle Chinese medial -j- and a non-front nuclear vowel), this is the same process seen in Section 4.1.5 (see Table 7.23). Table 7.24 shows that in qù 去 syllables three Middle Chinese bilabial stops become the same fricative and the relevant syllables become identical. Table 7.21 Devoicing pattern of Middle Chinese voiced stops and affricates MC

Condition

ZYYY

Example

[+voice]

píng tone

[-voice, +aspirated]

zè tone

[-voice, -aspirated]

any

no change

平 錢 桃 除 病 賤 盜 箸 柄 箭 到 著

[-voice]

a

bdzdɖbdzdɖptstʈ-

> > > > > > > > > > > >

phtshthtʂh-a ptsttʂptsttʂ-

The retroflex series has also undergone affrication by this point in time.

píng qián táo chú bìng jiàn dào zhù bǐng jiàn dào zhù

‘level’ ‘money’ ‘peach’ ‘remove’ ‘disease’ ‘cheap’ ‘rob’ ‘chopsticks’ ‘handle’ ‘arrow’ ‘arrive’ ‘significant’

280

A New Standard

Table 7.22 The Zhōngyuán Yīnyùn demonstrates a merger of voiced and voiceless unaspirated stops and affricates píngshēng 平 phiŋ < b‘level’ 錢 tshjɛn < dz‘money’ qùshēng 病 piŋ ‘disease’ 賤 tsjɛn ‘cheap’

6¼ 6¼

< b-

=

< dz-

=

qùshēng

píngshēng

病 piŋ < b‘disease’ 賤 tsjɛn < dz‘cheap’

桃 thaw < d‘peach’ 除 tʂhy < ɖ‘remove’

qùshēng 柄 piŋ < p‘handle’ 箭 tsjɛn < ts‘arrow’

qùshēng 盜 taw < d‘rob’ 箸 tʂy < ɖ‘chopsticks’

qùshēng 6¼ 6¼

= =

盜 taw < d‘rob’ 箸 tʂy < ɖ‘chopsticks’ qùshēng 到 taw < t‘arrive’ 著 tʂy < ʈ‘significant’

Table 7.23 The merger of labiodental fricatives p- > fph- > fb- > f-

夫 藩 方 敷 翻 芳 符 煩 房

‘husband’ ‘fence’ ‘square’ ‘apply’ ‘turn over’ ‘fragrant’ ‘symbol’ ‘bothered’ ‘house’

pjo pwjɐn pwjɐŋ phjo phwjɐn phwjɐŋ bjo bwjɐn bwjɐŋ

> > > > > > > > >

fu fan faŋ fu fan faŋ fu fan faŋ

> > > > > > > > >

fū fān fāng fū fān fāng fú fán fáng

Table 7.24 The merger of three labial stops in qù syllables 廢 fi < p‘useless’ 富 fu < p‘rich’

= =

肺 fi < ph‘lung’ 副 fu < ph‘deputy’

= =

吠 fi < b‘bark’ 復 fu< b‘repeat’

Table 7.25 shows that the corresponding Middle Chinese nasal stop becomes a labiodental approximant and loses its nasality. Unlike Modern Standard Mandarin, the syllables with wēi 微 ɱ- initials are still distinct from the syllables with the yǐng 影 Ø- initial (see Table 7.26); this contrast is discussed more in depth in Section 7.3.1.1. The loss of nasality of the Middle Chinese wēi 微 initial is one of the major contrasts between Mandarin and southern dialects. Most southern dialects still preserve this nasal initial,

Old Mandarin: The Zhōngyuán Yīnyùn 中原音韻

281

Table 7.25 The loss of nasality of the wēi ɱ- initial 無 ʋu < ɱ‘not’

晚 ʋan < ɱ‘late’

味 ʋi < ɱ‘taste’

望 ʋaŋ < ɱ‘look far’

Table 7.26 The contrast between the wēi ɱ- and the yǐng Øinitials 萬 ʋan ‘ten thousand’ 望 ʋaŋ ‘look far’ 味 ʋi ‘taste’ 務 ʋu ‘affair’

6¼ 6 ¼ 6 ¼ 6¼

腕 wan ‘wrist’ 旺 waŋ ‘prosper’ 胃 wi ‘stomach’ 惡 u ‘hate’

while other modern dialects have had the wēi 微 initial undergo denasalization, some dialects (Xi’an) still maintain a ʋ/w distinction. 7.4.3

The Development of the High Central (Apical) Vowels

In the ZYYY the zhī-sī 支思 rhyme group was for the first time treated as an independent group. In this rhyme group, all syllables have the high central vowel, which has two allophonic variations conditioned by the initial consonant’s place of articulation. The tongue position of this central vowel is either similar to the alveolar sibilants or the retroflex sibilants. In almost all the literature on Chinese phonology, this high central vowel is analyzed as an apical vowel. In comparison with the modern Beijing dialect the centralization at the ZYYY time had not reached its final stage yet, as some of the syllables that have a central vowel in Modern Mandarin still have the high front vowel in the ZYYY. Table 7.27 gives some examples of the syllables from the Middle Chinese zhǐ 止 rhyme group with zhī 知 group initials (zhī 知 ʈ-, chè 徹 ʈh-, chéng 澄 ɖ-, niáng 娘 ɳ-), zhāng 章 group initials, and zhuāng 莊 group initials. Table 7.27 A comparison of the vowels of the Middle Chinese zhǐ 止 rhyme group within zhī 知, zhāng 章, and zhuāng 莊 group initials Middle Chinese zhī 知 zhāng 章 zhuāng 莊

i ɨ ɨ

知 支 輺

痴 嗤 差

池 匙 師

恥 齒 使

制 至 事

282

A New Standard

Table 7.28 The rhyme and tone of non-zhǐ 止 rhyme group words with vowel -i

ZYYY vowel MC rhyme MC tone

制 zhì

世 shì

直 zhí

十 shí

尺 chǐ

‘make’ -i 祭 jì 去 qù

‘world’ -i 祭 jì 去 qù

‘straight’ -i 職 zhí 入 rù

‘ten’ -i 緝 入

‘foot’ -i 昔 入

xī rù

jī rù

Table 7.29 Centralization of high vowels in the Zhōngyuán Yīnyùn Middle Chinese 精 知 莊 章

jīng zhī zhuāng zhāng

tsʈtʂtɕ-



















1 4 2 3

1 4 2 3

1 4 2 3

(+) 4 – 4

(+) 4 2 4

(+) 4 – 4

– – 2 –

(+) 4 2 4

(+) 4 – 4

Note: The presence of centralization is in italics. “(+)” indicates the syllables not involved in the centralization process.

Other syllables with vowel -i are either from the Middle Chinese jì 祭 rhyme or are rù syllables and are given in Table 7.28. In the ZYYY the syllables with zhī 知 initials still had not shown any sign of change, shown in Table 7.29. Of the four possible changes of (1) jīngcentralization (精), (2) zhuāng-centralization (莊), (3) zhāng-centralization (章) (including the rì 日 initial), and (4) zhī-centralization (知), the ZYYY shows changes (1), (2), and (3) (Shen 2008b). By the time of the MGZY, Mandarin had only undergone the first two (see Section 6.4.8). 7.4.4

The New Four-Tone System

The tonal system of the ZYYY has undergone significant changes and become very “Mandarin”; in fact, it was the first time such a typically Mandarin tonal system had been recorded. In the ZYYY four tonal groups are recognized. They are píngshēng yīn, píngshēng yáng, shǎngshēng, and qùshēng, which are equivalent to tones of the Beijing dialect, namely yīnpíng, yángpíng, shǎngshēng, and qùshēng. Table 7.30 illustrates the relationship between the Middle Chinese tonal system and that of the ZYYY. The syllables of the Middle Chinese rù tone are classified (zuò 作) into three of the four tones based on the types of the initial consonants of the syllables, divided just as shown in Table 7.31. Instead of merging the Middle Chinese rù tones with other tones, the ZYYY’s treatment is to list these syllables

Old Mandarin: The Zhōngyuán Yīnyùn 中原音韻

283

Table 7.30 The merging and splitting conditions between Middle Chinese and the early Mandarin of the Zhōngyuán Yīnyùn Middle Chinese

Middle Chinese initial

ZYYY

Examples

píng

voiceless

yīnpíng

voiced

yángpíng

sonorant or voiceless

shǎng

voiced obstruent



all contexts



ptsxbdznlptsxnbzpx-

冰 津 呼 平 存 南 良 比 姐 好 暖 抱 靜 半 漢

blb-

病 路 薄

bīng < pɨŋ ‘ice’ jīn < tsin ‘ferry’ hū < hwo ‘call’ píng < bɰjaŋ ‘level’ cún < dzwon ‘exist’ nán < nəm ‘south’ liáng < ljɐŋ ‘fine’ bǐ < pi ‘compare’ jiě < tsja ‘sister’ hǎo < hɑu ‘good’ nuǎn < nwɑn ‘warm’ bào < bɑu ‘embrace’ jìng < dzjɛŋ ‘quiet’ bàn < pwɑn ‘half’ hàn < hɑn ‘Han Chinese’ bìng < bɰjaŋ ‘disease’ lù < lwo ‘road’ bó < bɑk ‘thin’

zptsxnl-

雜 筆 則 黑 納 鹿

zá < dzəp ‘mixed’ bǐ < pɰit ‘brush pen’ zé < tsək ‘rule’ hēi < hək ‘black’ nà < nəp ‘accept’ lù < luk ‘deer’

shǎng





voiced obstruent initials

yángpíng

voiceless initials

shǎng

sonorant initialsa



a

All voiceless sonorants were lost by the time of Middle Chinese, so the contexts only have three categories: voiced obstruent, voiceless obstruent, and (voiced) sonorant.

separately after the corresponding tones. This categorization, or zuò, of the Middle Chinese rù tone in the ZYYY is ambiguous and has become a major controversy in the study of the ZYYY. The rù zuò 入作 X (X = píngshēng yáng, shǎngshēng, or qùshēng) can be found in the nine rhyme groups that lack a nasal coda. Table 7.31 demonstrates examples of characters in each of the syllables in these rù zuò categories. Some interpret the word zuò 作 as ‘merge into,’ and some interpret it as ‘classify as,’ with the former believing that the rù tone was lost and the latter believing that the rù tone is preserved. Both sides have evidence to support their arguments, but based on the merging patterns of modern dialects, it is most likely that the loss of the rù tone had already happened by the time of the Yuan dynasty. This change could have been a gradual process – it may have occurred

284

A New Standard

Table 7.31 Mergers of the rù tone with three other tones rù zuò píngshēng yáng zhī-sī qí-wēi

yú-mó jiē-lái

xiāo-háo

gē-gē jiā-má chē-zhē yóu-hóu

實直疾夕 賊劾 惑 仆復獨族 俗逐屬局 白宅 劃 薄鐸濁鶴 學 著芍 薄鐸濁鶴 著杓學 拔乏達閘 別迭捷傑 絕鐝 軸熟

rù zuò shǎngshēng 澀瑟塞 必的質七 德黑 筆北國 不福督谷 足粟出菊 百則策色 革客嚇 摔摑 剝託錯閣 角覺腳 爵雀酌謔 閣渴 八法答扎 龞鐵節結 雪拙說決 宿竹

rù zuò qùshēng 覓立日易 勒 墨密 木物祿 綠入玉 麥搦 額 末諾落惡 岳樂藥約 略弱虐 惡 略弱虐岳 抹襪納臘 滅裂熱業 劣月 六肉

in syllables with voiced (zhuó 濁) initials first, then spread into the voiceless (qīng 清) syllables. As can be seen in Modern Mandarin dialects, the change pattern of rù syllables with Middle Chinese voiced initials is more consistent and ubiquitous than that of those with voiceless initials.7 The merge pattern also indicates the vowel system of the rù syllables before the merging because the vowel system of rù and non-rù (shū) syllables must have been the same before merger could happen, and were thus the same after tonal contrast was lost. 7.4.4.1 The Loss of the Stop Coda and the Disappearance of the Middle Chinese rù Tone In the ZYYY, Middle Chinese rù syllables have lost their stop codas completely. Thus the Middle Chinese syllables with the same medial and nucleus but different -p, -t, or -k codas become homophones, as illustrated in Table 7.32. While it is clear that in the ZYYY no stop consonant coda is preserved, some scholars (Lǐ Róng 1985) continue to believe that the end of Middle 7

It should be noted that while the rù zuò pattern is completely regular in the ZYYY, Modern Mandarin dialects have much more irregular patterns. In Beijing Mandarin, Middle Chinese rù tone sonorants and voiced obstruents consistently merge with qù and yángpíng respectively, Middle Chinese rù tone syllables with voiceless initials are sporadically distributed across all four Modern Mandarin tones.

Old Mandarin: The Zhōngyuán Yīnyùn 中原音韻

285

Table 7.32 Middle Chinese codas are lost and rù tone syllables with different codas merge

Middle Chinese ZYYY

立 lì ‘stand’ lip li

栗 lì ‘chestnut’ lit li

力 lì ‘force’ lɨk li

臘 là ‘year end sacrifice’ lɑp la

辣 là ‘hot’ lat la

Middle Chinese ZYYY

集 jí ‘collect’ dzip tsi

疾 jí ‘disease’ dzit tsi

籍 jí ‘record’ dzjɛk tsi

攝 shè ‘combine’ ɕjɛp ʃjɛ

設 shè ‘display’ ɕjɛt ʃjɛ

Chinese rù syllables may still have a glottal tension. The glottal tension may have the tonal distinction to the four tones and this is the reason why the rù syllables only zuò 作 ‘classify as’ other tones instead of merging into other tones. But such an argument cannot be supported by convincing evidence. 7.4.5

The Loss of the Middle Chinese yí 疑 Velar Nasal Initial

The loss of the Middle Chinese velar nasal initial ŋV- > V- was in its final stage. It apparently was interfered with by another change, which is ŋi- > ni-. Since both changes happened in an irregular fashion, the loss of the velar nasal initial was incomplete. In some syllables the velar nasal initial was preserved and in some syllables it changed to alveolar nasals. The residuals of the Middle Chinese velar nasal initial can be found in the system. Table 7.33 compares the finals of syllables that maintain their ŋ- initials and those whose initials have merged with n-. These residual forms with n- initial are caused by competing changes at the time. But the residual forms with ŋ- initial are syllables that at the time have not yet been affected by the change of ŋ- > Ø-. Table 7.33 Interference between two changes to the Middle Chinese velar nasal initial MC

Condition

ZYYY

Examples

ŋ-

-aŋ -jaŋ -aw (some) -jɛ -jɛn -jo (some) -jɛ

ŋŋŋŋnnn-

昂 仰 傲 業 讞 虐 捏

áng < ŋɑŋ ‘raise’ yǎng < ŋjɐŋ ‘raise head’ ào < ŋɑw ‘arrogant’ yè < ŋjɐp ‘cause, establishment’ yàn < ŋɰjɛn ‘judge’ nüè < ŋjɐk ‘cruel’ niē < net ‘pinch’

286

A New Standard

7.4.6

The Disappearance of the Four-Rank System of Finals

Rank-II syllables lost their unique phonological identity and many merged into syllables deriving from other ranks. The merger was conditioned by the initial consonants’ place of articulation, but may have different patterns in different rhymes. Table 7.34 shows this comparison across different initials and different Middle Chinese rhymes. 7.4.6.1 The Complete Loss of the Contrast between Middle Chinese Rank-III and -IV Syllables The Middle Chinese Rank-III and -IV syllables completely lost their contrast and completed their merger. Table 7.35 illustrates some of the resulting homophones. Table 7.34 Rank-II syllables merged with different finals based on rhyme and place of articulation of the initial

Labial

Retroflex

Velar

麻 má 巴 -a bā ‘state name’ 查 -a chá ‘check’ 家 -ja jiā ‘home’

皆 jiē 排 -aj pái ‘row’ 齋 -aj zhāi ‘studio’ 皆 -jaj jiē ‘all’

肴 yáo 包 -aw bāo ‘wrap’ 爪 -aw zhǎo ‘claw’ 交 -jaw jiāo ‘cross’

刪 shān 班 -an bān ‘surname’ 刪 -an shān ‘delete’ 姦 -jan jiān ‘evil’

江 jiāng 邦 -aŋ bāng ‘state’ 窗 -waŋ chuāng ‘window’ 江 -jaŋ jiāng ‘river’

耕 gēng 棚 -wəŋ péng ‘shed’ 爭 -əŋ zhēng ‘strive’ 耕 -əŋ gēng ‘plough’

Table 7.35 The Zhōngyuán Yīnyùn merge between Rank-III and Rank-IV syllables Middle Chinese Rank-III 基 ‘foundation’ 建 ‘build’ 京 ‘capital’ 變 ‘change’ 名 ‘name’ 矯 ‘correct’ 璧 ‘jade with hole’ 消 ‘vanish’ 滅 ‘extinguish’ 龜 ‘turtle’ 冤 ‘injustice’ 戟 ‘halberd’

Middle Chinese Rank-IV kɨ kjɐn kɰjaŋ pɰjɛn mjɛŋ kɰjɛw pjɛk sjɛw mjɛt kwɰi Øwjɐn kɰjak

雞 ‘chicken’ 見 ‘see’ 經 ‘classic works’ 遍 ‘all over’ 銘 ‘inscribe’ 繳 ‘submit’ 壁 ‘wall’ 蕭 ‘artemisia’ 篾 ‘bamboo strips’ 閨 ‘door’ 淵 ‘deep’ 激 ‘rapid’

kej ken keŋ pen meŋ kew pek sew met kwej Øwen kek

> > > > > > > > > > > >

ZYYY

Modern Mandarin

ki kjɛn kiŋ pjɛn miŋ kjɛw pi sjɛw mjɛ kwej jwɛn ki

jī jiàn jīng bìàn míng jiǎo bì xiāo miè guī yuān jǐ /jī

Old Mandarin: The Zhōngyuán Yīnyùn 中原音韻

287

Along with the change of Middle Chinese Rank-II syllables, this change helped to replace the four-rank system of Middle Chinese phonology with a simpler medial system, one which closely resembles the modern sìhū system (see Section 1.4.7). 7.4.6.2 The Residual Chóngniǔ Contrast The merger of Rank-III and Rank-IV syllables also indicates that many Middle Chinese chóngniǔ 重紐 syllables (IIIa and IIIb) had lost their contrast. Only the chóngniǔ syllables with no coda still show some unsystematic contrast (Hirayama 1991: 29). Some Middle Chinese píng tone chóngniǔ syllables are listed as an example in Table 7.36, alongside the chóngniǔ syllables of the modern Beijing dialect, whose distribution pattern is notably different. Table 7.36 A comparison of chóngniǔ contrasts (Middle Chinese IIIa and IIIb) present in reflexes of the Zhōngyuán Yīnyùn and modern Beijing dialect

pphphm-

-i IIIa – 紕 脾毗 彌

IIIb – – 疲羆 –

-ej IIIa 卑 – – –

IIIb 碑陂悲 披丕邳 皮 糜眉麋

pphphm-

-i IIIa – 紕 脾毗 彌

IIIb 陂 披丕邳 疲羆皮 糜麋

-ej IIIa 卑 – – –

IIIb 碑悲 – – 眉

ZYYY yīnpíng yángpíng

Beijing yīnpíng yángpíng

7.4.6.3 Rank-II and Rank-III Contrast Some Rank-II and Rank-III syllables have already merged, while other Rank-II and Rank-III syllables are in contrast. As shown in Table 7.37, Rank-II and -III syllables merge at different times. The first pair are homophones by the time of the ZYYY. The second, third, and fourth pairs are differentiated in the ZYYY but become homophones in modern Beijing dialect. However, the final pair still contrasts in the modern Beijing dialect. Ranks I and II contrast in syllables with -n coda. The huán-huān 桓歡 rhyme group indicates the contrast between Middle Chinese Rank-I and RankII hékǒu syllables. This feature of Rank-I and -II contrast is common in southern China and is most likely a feature from the area. These southern features are present alongside the more prominent northern features in the ZYYY, as discussed in Section 7.1.1.

288

A New Standard

Table 7.37 The contrast of Rank-II and Rank-III syllables in the Zhōngyuán Yīnyùn Rank

Middle Chinese

ZYYY

Beijing Mandarin

II III

江 jiāng ‘river’ 姜 jiāng ‘surname’

kɰɔŋ kiɐŋ

>

kjaŋ

>

tɕiaŋ

II III

交 jiāo ‘cross’ 驕 jiāo ‘arrogant’

kɰau kɰjɛu

> >

kjaw kjɛw

>

tɕjaw

II III

間 jiān ‘inbetween’ 建 jiān ‘build’

kɰæn kjɐn

> >

kjan kjɛn

>

tɕjɛn

II III

減 jiǎn ‘subtract’ 檢 jiǎn ‘check’

kɰæm kɰjɛm

> >

kjam kjɛm

>

tɕjɛn

II III

耕 gēng ‘plough’ 京 jīng ‘capital’

kɰæŋ kɰjaŋ

> >

kəŋ kjəŋ

> >

kɤŋ tɕiŋa

The character jīng 京 is pronounced /tɕiŋ/ in the Beijing dialect, while gēng 耕 is pronounced /kəŋ/, with a rare dual pronunciation as /tɕiŋ/, homophonic to jīng 京.

a

7.4.6.4 The Emergence of the chē-zhē 車遮 Rhyme A new rhyme group, chē-zhē 車遮, emerged in the ZYYY (see Table 7.38). The shū 舒 (historically non-rù) syllables are from the má 麻 rhyme of Middle Chinese Rank-III. The Middle Chinese má 麻 rhyme has Rank-II and Rank-III syllables and its main vowel is -a. In the ZYYY, Rank-III and Rank-II syllables are in different rhyme groups. Rank-II syllables are in the jiā-má 家麻 rhyme group, but Rank-III syllables are in the chē-zhē 車遮 rhyme group. This separation indicates that Rank-III syllables have a raised main vowel. This change is conditioned by the palatal medial of Rank-III syllables, -ja > -jɛ (see Section 6.4.7 for a comparison between -ja and -jɛ reflexes with codas). In the chē-zhē 車遮 rhyme group the majority of syllables are from Middle Chinese rù rhymes, which were historically in Rank-III and -IV and had a -p or -t coda, indicating rù tone syllables whose codas had elided. This change is actually the same as what is observed in the MGZY, although in the MGZY these syllables are grouped together with syllables with main vowel -a. Figure 6.3 visualizes this front-raising, among vowel changes, as they occurred in the MGZY. In comparison with Modern Mandarin, there is a clear difference. The syllables with a retroflex initial have lost the palatal medial and acquired a back vowel. This is an adjustment to reduce the articulatory conflict between the retroflex initials and front vocalic elements. As a result, -jɛ > -ɤ (e.g. zhē 遮, chē 車, and shē 奢) and -ɥɛ > -uɔ (e.g. zhuō 拙, chuò 啜, and shuō 說), as seen in Table 7.39).

Old Mandarin: The Zhōngyuán Yīnyùn 中原音韻

289

Table 7.38 The syllables of the chē-zhē rhyme group Middle Chinese

Condition

ZYYY

Example

má 麻 rhyme

Rank-III kāikǒu

-jɛ

Rank-III hékǒu

-jwɛ

Otherwise

-a

kāikǒu

-jɛ

爹 車 些 蛇 靴 瘸 麻 瓜 別

ʈja tɕhja sja ʑja hwja ɡwjɑ mɰa kwɰa bɰjɛt

> > > > > > > > >

tjɛ tʂhjɛ sjɛ ʂjɛ xjwɛ khjwɛ ma kwa pjɛ

diē ‘father’ chē ‘cart’ xiē ‘some’ shé ‘snake’ xuē ‘boots’ qué ‘lame’ má ‘hemp’ guā ‘melon’ bié ‘leave’

協 列 絕 雪 月

ɦep ljɛt dzwjɛt swjɛt ŋwjɐt

> > > > >

xjɛ ljɛ tsjwɛ sjwɛ jwɛ

xié ‘assist’ liè ‘list’ jué ‘broken’ xuě ‘snow’ yuè ‘moon’

some rù tone rhymes

hékǒu

-jwɛ

Table 7.39 A correspondence table showing the Beijing dialect reflexes of the Zhōngyuán Yīnyùn chē-zhē rhyme

ZYYY

Condition

chē-zhē 車遮 rhyme

retroflex initial kāikǒu

retroflex initial hékǒu

non-retroflex kāikǒu

non-retroflex hékǒu

Beijing Mandarin

Example



遮 tʂjɛ

>

tʂɤ

zhē ‘cover’

-uo

車 tʂhjɛ > 奢 ʂjɛ > 拙 tʂhjwɛ >

tʂhɤ ʂɤ tʂhuɔ

chē ‘cart’ shē ‘extravagant’ zhuō ‘clumsy’

-jɛ

啜 tʂhjwɛ > 說 ʂjwɛ > 嗟 tsjɛ >

tʂhuɔ ʂuɔ tɕjɛ

chuò ‘sip’ shuō ‘say’ jiē ‘sigh’

-ɥœ

傑 kjɛ 爺 jɛ 缺 khjwɛ

> > >

tɕjɛ jɛ tɕhɥœ

jié ‘hero’ yé ‘grandfather’ quē ‘lack’

劣 ljwɛ 雪 sjwɛ

> >

lɥœ ɕɥœ

jiè ‘inferior’ xuě ‘snow’

290

A New Standard

7.4.7

The Preservation of the -m Coda

The -m coda is basically preserved except for the change -m > -n in syllables with labial initials, as shown in Table 7.40. This suggests that the dropping of the -m coda began as a dissimilation of labials in the initial and coda position. The character pǐn 品 (Middle Chinese: -m) is listed together with pìn 牝 (Middle Chinese: -n) in the zhēn-wén 真文 (-n) rhyme group instead of the qīnxún 侵尋 (-m) rhyme group, a change caused by dissimilation of the place of articulation, phɰim > phin. Table 7.40 Contrasts between the -m and -n coda that remained in the Zhōngyuán Yīnyùn

7.4.8

針 -m zhēn < tɕim ‘needle’



真 -n zhēn < tɕin ‘true’

金 -m jīn < kɰim ‘gold’



巾 -n jīn < kɰin ‘towel’

咸 -m xián < ɦɰæm ‘all’



閑 -n xián < ɦɰæn ‘barrier’

添 -m tiān < them ‘add’



天 -n tiān < then ‘sky’

淡 -m dàn < dɑm ‘light’



誕 -n dàn < dɑn ‘birth’

藍 -m lán < lɑm ‘indigo plant’



蘭 -n lán < lɑn ‘orchid’

謙 -m qiān < khem ‘humble’ but 品 -m > -n pǐn < phɰim ‘rank’



牽 -n qiān < khen ‘lead by hand’

點 -m diǎn < tem ‘dot’



典 -n diǎn < ten ‘classic’

=

牝 -n pǐn < bin ‘female animal’

The jiān-tuán 尖團 Distinction

The distinction between alveolar initials (ts-, tsh-, s-) and velar initials (k-, kh-, x) in front of medial -j- or a high front vowel, also called the jiān-tuán 尖團 distinction, is shown in Table 7.41. This indicates that some dialects were already undergoing the palatalization of ts-, tsh-, and s- in front of high front vowels, so Zhōu Déqīng had to emphasize keeping the two distinct. According to modern dialects, the k- initial group palatalized before the ts- initial group. This also indicates that the palatalization of k-, kh-, and x- had already happened and they had already become tɕ-, tɕh-, and ɕ-. So, the palatalization of the jīng 精 ts- initial group

Old Mandarin: The Zhōngyuán Yīnyùn 中原音韻

291

Table 7.41 Contrasts of the jiān-tuán distinction 濟 tsjì < tsej ‘ferry’



計 k- (tɕ-) jì < kej ‘plan’

就 tsjiù < dziw ‘achieve’



舊 k- (tɕ-) jiù < ɡiw ‘old’

千 tshqiān < tshen



牽 kh-(tɕh-) qiān < khen



‘lead by hand’

全 tshquán < dzwjɛn ‘whole’

權 kh-(tɕh-) quán < ɡwɰjɛn ‘power’

‘ten thousand’ 集 tsjí < dzip ‘collect’



及 k- (tɕ-) jí < ɡɰip ‘reach’

焦 tsjiāo < tsjɛw ‘burned’



驕 k- (tɕ-) jiāo < kɰjɛw ‘arrogant’

津 tsjīn < tsin ‘ferry’



巾 k- (tɕ-) jīn < kɰin ‘towel’

相 sxiāng < sjɐŋ ‘mutual’



香 x- (ɕ-) xiāng < hjɐŋ ‘fragrant’

Table 7.42 A four-stage process resulting in the merging of velars and coronals in front of high vowels I IIa III IV

Palatalization of velars Lenition of palatals Palatization of nonstop dentals

kctɕtɕ-

khchtɕhtɕh-

xɕ-b ɕɕ-

tstststɕ-

tshtshtshtɕh-

sssɕ-

a These changes only happen when these initials are before the medial -j- or -ɥ- or vowels -i- or -y-. When it is stated that the jiàn 見 k- initial group is in complete complementary distribution, it means that by looking at the initial k in context one can determine totally accurately if it is velar or alveopalatal, so there is no reason to create a separate initial for the new alveopalatal series. b The x- initial may well have been actualized as palatal fricative ç at stage II in development; if so, during stage III it would have become ɕ in order to economically join the new alveopalatal series.

causes a merger, and it is possible that the palatalized Middle Chinese-velar initials were not recognized as a separate group of initials since it is in completely complementary distribution with the original jiàn 見 k- initial group. Table 7.42 demonstrates this change across the six affected initials, showing their complete merging in stage IV. The ZYYY phonology was in stage II or III. 7.5

Examples of Yuan Pronunciation

The phonology of the Zhōngyuán Yīnyùn 中原音韻 (ZYYY) is very close to Modern Mandarin. This similarity can be demonstrated by the following short

292

A New Standard

poem in the format of Tiānjìngshā 天淨沙, a sample poem included in the ZYYY and one of the most famous short poems of the Yuan dynasty. Table 7.43 supplies the poem with a series of transcriptions. Of the phonetic transcriptions the first line is the reconstructed value based on the ZYYY, the second is the phonetic value of Modern Mandarin and the third is pinyin. Qiūsī 秋思 ‘Autumn Thoughts’ by Mǎ Zhìyuǎn 馬致遠 (1260–1325) 枯藤老樹昏鴉。 小橋流水人家。 古道西風瘦馬。 夕陽西下, 斷腸人在天涯。

Withered vines, old trees, twilight crows. Small bridge, flowing water, people’s homes. Ancient road, the west wind, gaunt horse. The evening sun sinks westwards. A man, broken-hearted, on a far horizon. Source: English translation by Stephen Owen 1996: 740.

Table 7.43 The pronunciation of Mǎ Zhìyuǎn’s Qiūsī in the Mandarin of the Zhōngyuán Yīnyùn compared with Modern Mandarin 枯 khu khu kū

藤 thəŋ thəŋ téng

老 law law lǎo

樹 ʂju ʂu shù

昏 xun xun hūn

鴉 ja ja yā

小 sjɛw ɕjaw xiǎo

橋 khjaw tɕhjaw qiáo

流 liəw liəw liú

水 ʂwej ʂwej shuǐ

人 ɻin ɻən rén

家 kja tɕja jiā

古 ku ku gǔ

道 taw taw dào

西 si ɕi xī

風 fuŋ fəŋ fēng

瘦 ʂəw ʂəw shòu

馬 ma ma mǎ

夕 si ɕi xī

陽 jaŋ jaŋ yáng

西 si ɕi xī

下 xja ɕja xià

斷 twan twan duàn

腸 tʂhaŋ tʂhaŋ cháng

人 ɻin ɻən rén

在 tsaj tsaj zài

天 thjɛn thjɛn tiān

涯 ja ja yá

There are twenty-eight syllables (twenty-six different characters with rén 人 and xī 西 used twice) in the five lines of this short poem. According to the reconstruction, there are mainly two major phonological differences between the Mandarin of the ZYYY and Modern Mandarin: the palatalization of the

Old Mandarin: The Zhōngyuán Yīnyùn 中原音韻

293

syllables with tsj- and kj-type of initial-medial sequences (e.g. qiáo 橋 khjaw > tɕhjaw), and the loss of the front feature of the vowels after the retroflex initials (e.g. rén 人 ɻin > ɻən). The ZYYY was published in 1324, which is about seven hundred years after the Qièyùn 切韻 and seven hundred years before Modern Mandarin. But its phonology is apparently much more similar to Modern Mandarin despite being equally as distant from each time period. Besides the -m coda, all other differences between the ZYYY and Modern Mandarin can be found in subdialects of Modern Mandarin. It is very likely that the phonology of the ZYYY and that of Modern Mandarin are close enough to be mutually intelligible. 7.6

Summary

The Zhōngyuán Yīnyùn 中原音韻 (ZYYY) is an important landmark in the history of Chinese. Before the ZYYY most of the evidence of northern Mandarin is from non-Chinese scripts used by the Khitans, the Jurchens, and the Mongols. The ZYYY was the first Chinese rhyme book that established a phonological system which, for the first time, totally freed itself from the phonological tradition rooted in the Qièyùn 切韻. From the Yuan dynasty onward, Mandarin phonology became a more prestigious standard. However, the tradition of the Qièyùn as a literary standard was still strong and influential; Mandarin phonology did not become the national standard until the beginning of the twentieth century.

8

Old Mandarin: The Persian Transcriptions

Under the ruling of the Mongol Empire, Chinese books were brought from China to Persia and translated into the Persian (Farsi) language. The translation projects were carried out due to the scholarly interests of Rashid al-Din (1247–1318), who was a statesman, historian, and physician in the Ilkhanateruled part of the Mongol empire. Two books contain a fairly large number of the transcriptions of Chinese words in Persia during this time. The first, Rashid’s History of China (HOC), which is a part of his world history (Tārīkh-i Chīn az Jāmiʻ al-Tavārīkh), was based on a version of the HOC brought to Persia by two Khitay scholars, Lbbahi and M.ksun (Wáng Yīdān 2006). The other is a Chinese medical treatise named Mài Jué 脈訣 (University of Tehran 1972). Because these transcriptions were based on the Chinese pronunciation, they can provide us with phonetic information that cannot be retrieved from the Chinese writings. It will be shown that the pronunciations, although they were at the time all considered standard Chinese, reveal quite different individual variations. 8.1

The Ancient Persian Script

Though both Rashid’s History of China (HOC) and the Mài Jué 脈訣 use the Ancient Persian script, the Persian script may have radically different conventions based on the formality of the text. The Persian script tends to make heavy use of diacritics, which denote vowel quality, when used in more formal and religious texts, and tends to be devoid of these diacritics in more secular or informal texts. The HOC was written without diacritics, and as a result, the vowels of Chinese are underrepresented. In the Persian script only long vowels, a, u, and i, are represented without diacritics. The letters, ‫› ‘‹ ﺍ‬, ‫ ‹ ﻭ‬w ›, and ‫ﻱ‬ ‹ y ›, representing the consonants ʕ, w, and j, also represent three long vowels a, u, and i respectively (Mahootian 1997: 2). There is no letter designated for other vowels, and short vowels are omitted in the spellings. As in the spelling tradition of the Persian/Arabic script, diphthongs are represented by the letters for the phonetic quality of off-glides, u for aw and i for aj. “.” indicates the omitted short vowels. 294

Old Mandarin: The Persian Transcriptions

295

Table 8.1 The Persian letters present in the Mài Jué and their IPA values Persian letters used in the Mài Jué (23) labials alveolars

Phonetic values (IPA) b d dz dʒ ɡ

palatals velars

p t ts tʃ k

m n

ŋ

f s, s _ ʃ, ʂ x, xj

v 1 ʒ

In the Mài Jué, diacritics are placed to indicate vowel quality throughout the text. In addition, ad hoc letters are added for transcribing Chinese sounds that do not have a suitable letter in the Persian script (Endō 1997, 2016; Shen 2016b). The Persian letters present in the Mài Jué and their IPA values are given in Table 8.1. In the Ménggǔ Zìyùn 蒙古字韻 (MGZY) the Chinese initial system is based on a conservative system. In the Yuan dynasty the voiced stops, affricates, and fricatives actually have lost their voicing feature and merged with their voiceless counterparts (refer to Section 6.3.1). In the examples given in this section (given with their page and line number in their respective texts), both the face value of the spellings and the corrected phonetic value of the Yuan phonology are given. A > is used between the face value and the corrected value, e.g. ban > ph-. For the corrected values, only the relevant sounds are given. The phonetic transcriptions for HOC are based on Wang Yīdān’s (2006) translation, while the phonetic transcriptions for Mài Jué are based on the original texts, as published by the University of Tehran (1972). 8.2

History of China

History of China (HOC),1 written in 1304, has recently been translated into Chinese (Wáng Yīdān 2006). Since it contains the names of the dynasties, and of kings and emperors from prehistoric legends to the Yuan dynasty, the phonetic transcriptions of these proper names are an important piece of information for us to understand the Chinese phonology of the Yuan dynasty. This material is a transcription based on colloquial pronunciation, thus it is different in nature from rhyme books, such as the Ménggǔ Zìyùn 蒙古字韻 (MGZY) of the late thirteenth and early fourteenth centuries and the Zhōngyuán Yīnyùn

1

For more information about Rashid al-Din and his History of China, please refer to Franke (1951) and Wáng Yīdān (2006).

296

A New Standard

中原音韻 (ZYYY) of 1324, which are systematic arrangements of phonological categories. As shown in this section, the pronunciation that Rashid’s transcription was based on is likely the standard pronunciation of the Yuan. Historically what was considered to be proper Mandarin phonology was loosely defined. The standard pronunciation of the Yuan capital, Dadu (modern Beijing), may have been prestigious and influential, but was never clearly defined in the way that the modern standard is. Since the HOC was based on colloquial speech, the main interests of this study are to find out the phonological features of Old Mandarin as a spoken language, and how these features relate to the more standard written phonological systems such as the MGZY and the ZYYY. The rest of this section covers the various phonological characteristics of the Chinese of this era, used as the base language of the Persian transcriptions.

8.2.1

Middle Chinese Voiced Stops and Affricates

The Chinese voiceless aspirated stops and affricates are consistently transcribed as voiceless. The voiceless unaspirated consonants are usually transcribed as voiced but sometimes as voiceless. This lack of consistency is due to the different phonological features involved in Chinese and Persian. The Chinese voiceless aspirated stops and affricates are very similar to the voiceless stops in Persian, but the voiceless unaspirated ones cannot be matched perfectly to a Persian equivalent. Since Persian has either voiceless aspirated or voiced unaspirated stops or affricates, it is either in conflict with the feature of aspiration or the feature of voicing. This conflict between different languages’ phonation systems is similar to the issue faced by those who transcribed Chinese into the hP’ags-pa script for the MGZY as discussed in Section 6.1 and Section 6.3.1._ In the transcriptions shown in Table 8.2, voiced stops and affricates show the devoicing pattern of Mandarin; that is, the transcriptions follow the Mandarin pattern of Middle Chinese voiced stops and affricates becoming voiceless aspirated in the píng 平 tone syllables but voiceless unaspirated in the zè 仄 (non-píng) tone syllables (see Table 8.3). Strictly speaking, the píng tone syllables with voiced stops and affricates are rendered as voiceless aspirated stops.

8.2.2

The Loss of Velar Nasal Initial yí 疑 ŋ-

As seen in Table 8.4, the nasal consonant of Middle Chinese yí 疑 initial ŋ- has been lost. The examples are either Rank-I or Rank-III syllables. This change is consistent with what was observed in the MGZY and the ZYYY.

Old Mandarin: The Persian Transcriptions

297

Table 8.2 A comparison of transcriptions of voiceless aspirated and unaspirated Middle Chinese initials into Persian script in various personal names Transliteration of Chinese word

Page

Middle Chinese

hP’ags-pa transcription _

Voiceless aspirated 天 t.n 天皇氏 湯 tānk 成湯

t.nkhūānkshī shīnktānk

558 567

thth-

then thaŋ

Voiceless unaspirated 濮 bū 濮王 北 bū 北鄉侯 褒 bū 褒姒 丹 tān 丹朱 河亶甲 亶 tān 東 tūnk 東方 真 jīn 真宗

būvāng būshānkkhū būsī tānjū khūtānkīh tūnkfūnk jīnzūn

588 576 571 565 568 574 591

pppttttɕ-

pu pej paw tan tan tuŋ tʃin

Table 8.3 Persian transcriptions follow the same devoicing pattern present in Mandarin Transliteration of Chinese word

Page

Middle Chinese

hP’ags-pa transcription _

Píng tone syllables 盤 pān 盤古 平 pīnk 平王 提 tī 攝提紀 陶 tāū 陶唐氏 唐 tānk 陶唐氏 丞 chīnk 曹丞相

pānkū pīnkvānk sh.tīkī tāūtānkshī tāūtānkshī sū chīnksānk

557 569 558 565 565 576

bbdddʑ-

ban biŋ di daw daŋ ʒiŋ

> > > > > >

phphthththtʃh-

Zè tone syllables 度 dū 蔡叔度 地 dī 地皇氏 大 tāī 大庭氏 悼 tāū 悼王 沌 dūn 混沌氏 定 dīn 定王 鐸 dāū 曹叔鐸 仲 jūnk 仲丁 趙 jū 趙 肇 jū 史弘肇 直 jī 帝直

sāī shūdū dīkhūānkshī tāītāīshī tāūvānk hūndūnshī dīnvānk sū shūdāū jūnkdīn jū shū khūjū dījī

573 559 561 569 561 569 573 568 582 590 562

dddddddɖɖɖɖ-

du di daj daw dun diŋ daw dʒyŋ dʒew dʒew dʒi

> > > > > > > > > > >

ttttttttʃtʃtʃtʃ-

298

A New Standard

Table 8.4 A loss of the yí 疑 initial is evident in the History of China Transliteration of Chinese word

Page

Middle Chinese

五 吳 外 元

558 573 567 570

ŋŋŋŋ-

ū wū vāī ūn

五龍紀 吳太伯 外丙 元王

ūlūnkī wū tāybāy vāībīnk ūnvānk

hP’ags-pa transcription _ u u Øwej Øwɛn

Middle Chinese rì 日 Initial

8.2.3

As illustrated in Table 8.5, the Middle Chinese rì 日 initial is rendered with voiced fricative, arbitrarily alveolar or palatal. As in Modern Mandarin, in Old Mandarin the reflex of the Middle Chinese rì 日 initial was not a fricative but a retroflex approximant (see Section 7.3.1.1). The Persian ‹ r › was not used at the initial position. The ‹ r › is a trill in Modern Persian and could be the same in the period the transcriptions were made. So it was not suitable to use a trill for an approximant, especially in the initial position; a voiced fricative has a closer phonetic quality to the Mandarin ɻ. Table 8.6 shows that in the transcription of èr 二, ‹ r › is however used after ‹ i ›. This is a very interesting transcription because it indicates the pronunciation of èr 二 is already a rhotacized vowel or has metathesized. As Herbert Franke (1951: 25) observed, it is one of the earliest records of vowel rhotacization. This Table 8.5 Examples highlighting the non-rhotic transcription of the Middle Chinese rì 日 initial in the History of China Transliteration of Chinese word

Page

Middle Chinese

人 人 仁 壬 壬 儒

558 559 591 567 568 574

ɲɲɲɲɲɲ-

z.n zh.n zhīn zh.n zh.n zhū

人皇氏 燧人氏 仁宗 中壬 外壬 孺子嬰

z.nkhūānkshī sūīzh.nshī zhīnzūn jūnzh.n vāīzh.n zhūz.yīnk

hP’ags-pa transcription _ ɻin ɻin ɻin ɻim ɻim ɻy

Table 8.6 Character èr 二 transcribed as a rhotacized vowel Transliteration of Chinese word

Page

二 īr

573

二 世胡亥 īrshī hūkhūī

Middle Chinese hP’ags-pa transcription _ ɲɻi

Old Mandarin: The Persian Transcriptions

299

phenomenon of vowel rhotacization is seen more extensively in Ming texts, and is discussed in more depth in Section 9.6.4.5. 8.2.4

Middle Chinese wēi 微 Initial

In Modern Mandarin, Middle Chinese wēi 微, once a nasal consonant ɱ, becomes a labial semivowel and functions as a medial (e.g. wén 文 wən, wǔ 武 wu, wáng 亡 waŋ). In Old Mandarin, represented by the MGZY and the ZYYY, the Middle Chinese wēi 微 initial is still an independent initial; although it was no longer a nasal consonant, it was in contrast with syllables with a zero initial and a -w- medial. Scholars are in agreement about the place of articulation but not about the manners of articulation. The place of articulation must be labiodental since its counterparts, Middle Chinese initials fēi 非 pf-, fū 敷 pfh-, and fèng 奉 bv-, had merged and become f-. So it is very possible that Middle Chinese wēi 微 had also changed to a labiodental consonant. In the HOC, the Middle Chinese wēi 微 initial is rendered as ‹ v ›, ‹ w ›,2 and ‹ f › as shown in Table 8.7. These three transcriptions, especially ‹ f ›, indicate that Middle Chinese wēi 微 initial still maintained its value as a labiodental approximant ʋ-. Since Table 8.7 Different renderings of hP’ags-pa ʋ- initial _ Transliteration of Chinese word

Middle Chinese

hP’ags-pa transcription _

Middle Chinese wēi 微 initial rendered as v萬 vān 萬 vān 555 無 vū 無懷氏 vūkhūāshī 561 微 vī 宋微子啟 sūn vīzī chī 573 文 vīn 文宗 vīnzūn 587

mmmm-

ʋan ʋu ʋi ʋun

Middle Chinese wēi 微 initial rendered as w武 wū 武乙 wūyī 568 武 wū 武丁 wūdīn 568

mm-

ʋu

Middle Chinese wēi 微 initial rendered as f文 fīn 簡文帝 kīnfīndī 580 文 fīn 文皇帝 fīn khūānkdī 584 武 fū 武帝 fūdī 580 武 fū 武宗 fūzūn 587

mmmm-

ʋun

2

Page

ʋu

In Wáng Yīdān’s transliteration (2006) v, w, and u are distinguished. But they are different transliterations of the same Persian letter ‹ u › ‫ﻭ‬. It is not clear on what basis they are transliterated differently.

300

A New Standard

ʋ- cannot be transcribed with an available letter, other letters for labial and labiodental consonants are used. The Coda of rù 入 Syllables

8.2.5

In general, for the Middle Chinese rù syllables, which had a stop coda, all the examples show that the stop codas have been lost (see Table 8.8). The Middle Chinese -k syllables duó 鐸 and bó 伯 are spelled as ‹ dāū › and ‹ bāy ›, respectively. These spellings reflect the modern diphthongs aw and aj, respectively. Based on the information from Sino-Khitan (Shen 2007) and the MGZY, these diphthong forms are characteristic of northern Mandarin as discussed in Section 5.1.2, Section 5.2.2.1, and Section 5.3.1.5. Table 8.9 shows that in some cases the spellings of rù syllables end with an ‹ h › or ‹ q ›. The implications of this are that rù tone syllables may have Table 8.8 History of China features a lack of rù syllable stop codas Transliteration of Chinese word

Page

Middle Chinese

伏 穆 石 沃 北 郭 惡 叔 洛 鐸 伯 葛 妲 桀 發 乙 甲

560 580 582 567 576 590 570 573 556 573 573 561 570 567 567 568 567

-k -k -k -k -k -k -k -k -k -k -k -t -t -t -t -t -p

fū mū shī ū bū kū ū sū lāū dāū bāy kū dā kī fā yī kīā

伏羲 穆帝 石勒 沃丁 北鄉侯 郭威 惡來 衛康叔 洛京 曹叔鐸 吳太伯 葛天氏 妲己 桀 發 祖乙 太甲

fūkī mūdī shīl.h ūdīnk būshānkkhū kū vī ūlī wī kānksū lāūkīn sū shūdāū wū tāybāy kūt.nshī dākī kī fā sūyī tāīkīā

hP’ags-pa transcription _ vu > fu ʋu ʒi > ʃi Øu pej kwaw Øaw ʃy law daw paj ko ta ge > ke fa Øi kja

Table 8.9 Rù syllables transcribed with ‹ h › and ‹ q › endings Transliteration of Chinese word

Page

Middle Chinese

合 甲 跋 勒

559 568 582 582

-p -p -t -k

khīh kīh būq l.h

合熊紀 河亶甲 馮跋 石勒

khīhkhūnkī khūtānkīh fūn būq shīl.h

hP’ags-pa transcription _ ɦo kja bwa ləj

Old Mandarin: The Persian Transcriptions

301

merged their coda through a process of debuccalization before undergoing full elision and merging with other tones. 8.2.6

High Centralized (Apical) Vowels

The Yuan phonological works clearly show the existence of centralized high vowels (which are often called apical vowels). As discussed in Section 6.1.1, in the MGZY’s hP’ags-pa spellings the finals of zī 葘, chà 差, shì 士, zī 兹, ɨ, with the letter ‹ h › indicating a cí 雌, sī 思 are _spelled as ‹ hi › phonological feature of [-front]. In the ZYYY these syllables are all included in the zhī-sī 支思 rhyme group, which is set up exclusively for the syllables with centralized high vowels. But the Persian spelling system does not distinguish the vowels i and ɨ. In most cases the syllables with centralized vowel ɨ (according to the ZYYY and MGZY) are transcribed with ‹ i ›. In Table 8.10, no vowel letter is used to spell zǐ 子. The missing vowel indicates that the main vowel is -ɨ, which does not have a suitable letter for it in the Persian script. In the example of shǐ 史 a back vowel ‹ u › is used, indicating that the vowel is not a front -i. 8.2.7

Main Vowel -e

The Persian script main vowel -e or -ɛ is not spelled out (see Table 8.11). But this zero form, if used correctly, can help to determine the vowel quality. Table 8.10 Transcription of high central vowels Transliteration of Chinese word

Page

Middle Chinese

子 思 子 史

573 570 574 590

tsɨ sɨ tsɨ ʃɨ

zī sī z. shū

宋微子啟 思王 孺子嬰 史弘肇

sūn vīzī chī sīvānk zhūz. yīnk shū khūjū

hP’ags-pa transcription _ tsɨ sɨ tsɨ ʃɨ

Table 8.11 Zero form of vowels -e and -ɛ Transliteration of Chinese word

Page

Middle Chinese

宣 元 天 攝 夜 瀉

569 570 585 558 570 567

swjɛn ŋwjɛn then ɕjɛp ja sja

sūn ūn t.n sh. yī s.h

宣王 元王 則天皇后 攝提紀 長夜宫 瀉

sūnvānk ūnvānk s.t.n khūānhū sh.tīkī jīnkyīkūn s.h

hP’ags-pa transcription _ swɛn Øwɛn then ʃɛ je sɛ

302

A New Standard

The transcription of xiè 瀉 uses the letter ‹ h ›. According to Persian orthography it is possible to use the letter ‹ h › to transcribe vowel -ɛ at the end of a syllable. 8.2.8

Summary of Features

After examining and analyzing the transcriptions in the HOC, a general impression is that many phonological distinctions of Old Mandarin are not very well represented in this material. But on the other hand there are enough phonological features that we can use to identify the underlying phonological system, which is unambiguously Mandarin. The observed phonological characteristics of Old Mandarin and some phonetic details can be summarized as: 1. Devoicing of Middle Chinese voiced stops and affricates, aspirated in píng tone syllables, and unaspirated in zè tone syllables. 2. Loss of the Middle Chinese initial yí 疑 ŋ-. 3. The Middle Chinese initial rì 日 ȵ- is a retroflex approximant. 4. The Middle Chinese initial wēi 微 ɱ- lost its nasal quality but still remains as a labiodental consonant. 5. Preservation of the Middle Chinese nasal codas, -m, -n, and -ŋ. 6. Loss of the Middle Chinese stop codas, -p, -t, -k. 7. Development of the labial medial in Middle Chinese Rank-III syllables with zhuang initials of the dàng 宕 rhyme group. 8. Development of the palatal medial in Middle Chinese Rank-II syllables with guttural initials. 9. Existence of centralized (apical) high vowels. 10. Existence of rhotacized vowels. 11. Existence of high front rounded vowels. 12. The Rank-III syllables of the má 麻 rhyme group were mid front vowels (as in the chē-zhē 車遮 rhyme group of the ZYYY). 13. Dual vowel letters suggest that shǎng tone has a dipping tonal contour. 8.3

The Mài Jué 脈訣

The History of China (HOC) is not the only document containing a Persian transcription of Yuan-era Chinese. An annotated translation into Persian of a Chinese medical treatise in verses, Mài Jué 脈訣 (see Figure 8.1), attributed to Wáng Shūhé 王叔和, was also produced in the same time period (1313). Compared to the Persian transcription of History of China, the Mài Jué (see Dragunov 1931; Endō 1997) includes additional phonological clues, including an ad hoc letter for Chinese sounds that do not have a suitable letter in the Persian script (refer to Table 8.1).

Old Mandarin: The Persian Transcriptions

Figure 8.1 A page of the Mài Jué 脈訣.

303

304

A New Standard

In the early fourteenth century, the Mài Jué 脈訣 was translated into Persian in the Arabic script. It was collected in a book compiled by Rashid al-Din entitled Tanksūqnāma-i Īl-khān dar funūn-i ‘ulūm-i Khatāyī, ‘Book of Precious Information of the Ilkhan on the Various Branches of Chinese Sciences.’ The Chinese transcriptions in this document were initially studied by Soviet linguist Alexander Dragunov based on twelve photographs taken from a handwritten copy of the manuscript (Dragunov 1931). The Japanese linguists Mantaro Hashimoto and Endō Mitsuaki also studied this manuscript. Endō’s article (Endō 1997) is an important summary report providing many useful pieces of information about the manuscript, especially the sequential order of the poems in relationship to the Chinese versions of the Mài Jué. Because of Endō’s study, the identification of the Chinese poems became much easier to carry out. In his study, Endō points out an interesting phenomenon: from page 434 some of the Middle Chinese rù syllables start showing their original stop codas -p, -t, -k (Endō 1997).3 As is clearly shown in the MGZY and the ZYYY, in the phonology of Old Mandarin the Middle Chinese rù syllables had already lost their stop codas. But based on the examples Endō provided, some of the rù syllables clearly show the stop codas. Based on this particular feature, Endō concludes that there were two Chinese speakers involved in the Persian transcription: speaker A, a Mandarin speaker from the north, and speaker B, a non-Mandarin speaker from the south. Beside the -p, -t, -k stop codas, the phonological characteristics of speaker B are similar to those of speaker A. Thus, speaker B’s pronunciation was in Mandarin as well. However, with more careful examination it is also possible that the differences are the result of the same speakers’ pronunciation with different degrees of standardization (Shen 2016b). 8.3.1

The Stop Codas of the rù 入 Syllables

After examination of all the rù syllables in the original manuscript, it is clear that the stop codas start to appear before page 434 as Endō observed (1997). The earliest appearance of stop codas can be found as early as page 220. In the title of the poem Zhěnhòu Rùshì Gē 診候入式歌 ‘The Song of the Principles of Diagnosis,’ the character rù 入, the tonal label of the rù tone itself, is a rù syllable in the Middle Chinese phonology transcribed with the letter representing voiceless aspirated bilabial stop ph. Thus, this is a clear example that the pronunciation of character rù 入 has a coda -p. Exemplified in

3

This phenomenon was not mentioned by Dragunov in his 1931 article, due to the lack of material available to him.

Old Mandarin: The Persian Transcriptions

305

Table 8.12 A series of rù syllables in the Persian Mài Jué that maintain their coda in transcription 入 220.11 克 369.15 得 303.8

rph khiɡ dkh

促 341.11

tshuk

十 363.10

ʃb

色 323.2



克 359.13

ɡiɡ

Table 8.13 A transcription of the Rènfù Shānghán Gē Page.line

IPA transcription of the Persian letters

430.1 430.2 430.3 430.4 430.5 430.6 430.7 430.8

ʃaŋ khī ʃāŋ tshāŋ ū īū lū sīū

han ɡib ʂŋ rē thū būeī tshī phen

dū tʃhuŋ bān bū bū ɡy rth bū

thuŋ sm dm tʃhɨ tʃhɨ ɡīaŋ lāī thuŋ

lin rāɡ tʃhaɡ dʒī sim nāū rē dāī

Chinese characters bī rū hɡ thaī fn thuŋ fūɡ phn

dzē hjē ʃī mē rē lē tʃhūŋ khē

傷寒頭痛連百節 氣急沖心溺如血 上生班點赤黑時 壯熱不止致胎滅 嘔吐不止心煩熱 腰背俱強腦痛裂 六七日來熱腹中 小便不通大便結

Note: Bolded characters and transcriptions are Middle Chinese rù syllables, but only some of the transcriptions feature stop codas. None of the rhyming syllables, 節, 血, 滅, 熱, 裂, 結, has a coda -t.

Table 8.12, the Mài Jué includes other cases of rù syllables with a stop coda found throughout the text. In the poems, the first stop coda is found on page 303, in line 8, where the character dé 得 is transcribed as ‹ dkh ›. In Middle Chinese phonology dé 得 has a -k coda, so the ‹ kh › in the spelling of ‹ dkh › indicates the preservation of coda -k. The character sè 色 on page 323 and character kè 克 on page 359 are also Middle Chinese rù syllables with -k coda. Both are transcribed with a ‹ ɡ ›, indicating the presence of the -k coda. The next stop coda does not appear until page 430. Page 430 contains the poem Rènfù Shānghán Gē 妊婦傷寒歌 ‘The Song of Maternity Pyrexia’ of the Mài Jué. This poem is in a traditional format with eight lines and seven syllables in each line, and can be seen along with a transcription in Table 8.13. The identification of Chinese characters is based on the Chinese version of the Mài Jué.4

4

Two versions are referred to (1) The Jiégǔ Lǎorén Zhù Wáng Shūhé Mài Jué 潔古老人註王叔 和脈訣 ‘An annotated version of Wáng Shūhé’s Mài Jué by Elder Jiégǔ’ by Zhāng Yuánsù 張元 素 and published between 1237 and 1282 (Zhèng Jīnshēng 2002), and (2) the Mài Jué Kān Wù 脈訣刊誤 ‘Corrected Mài Jué’ by Dài Tóngfù 戴同父 of Yuan published in the early fourteenth century (四庫全書本).

306

A New Standard

The Rènfù Shānghán Gē contains the rù syllables in bold face (Figure 8.1 is the original image of this page). There are twenty rù syllables in this poem. According to their Middle Chinese realizations, one syllable has coda -p, fourteen have coda -t, and four have coda -k. It is quite interesting to note that of these twenty syllables, only some of the syllables that originally had -p, -t, and -k codas still retain them. Except for one, all the syllables that originally had a -t coda do not show any sign of coda -t at all. Codas -p and -k are transcribed by the letter for voiced stops ‹ b › and ‹ ɡ ›, but coda -t is transcribed by the letter for voiceless aspirated stop ‹ th ›, rather than the voiced ‹ d ›. The rù syllables, the position of the rù syllable, and the consonant coda in the Middle Chinese system and in the Persian version of the Mài Jué (hereafter PMJ) are listed in Table 8.14. However, between pages 436 and 449, no stop codas are written on their respective rù tone syllables. Page 436 is the beginning of Yángdú Yīndú Gē 陽毒陰毒歌 ‘The Song of the Yang Toxin and the Yin Toxin’ and page 449 is the beginning of Zhū Bìng Shēngsǐ Mài Gē 諸病生死脈歌 ‘The Song of the Pause of Death Caused by Various Diseases.’ After page 449 and on to the end of the text the stop codas continue to appear in the poems. In the list the poems that do not contain stop codas are labeled with PMJ-1, and the poems that do contain stop codas are labeled with PMJ-2. In addition, the page numbers, the name of the poem, the chapter number, and the presence or absence of the stop coda are also provided in further analyses of the Mài Jué. 8.3.2

The Inconsistency of the Stop Coda in the Transcription

Although the spellings of the stop coda of rù syllables are frequently found, the rù syllables are not always spelled with a stop coda. In addition, some rù Table 8.14 Middle Chinese codas and their transcriptions in the Persian Mài Jué MC/PMJ 百 節 急 溺 血 赤 黑

1.6 1.7 2.2 2.5 2.7 2.5 2.6

-k/-Ø -t/-Ø -p/-b -k/-ɡ -t/-Ø -k/-ɡ -k/-ɡ

MC/PMJ 熱 不 滅 不 熱 裂 六

4.2 4.3 4.7 5.3 5.7 6.7 7.1

-t/-Ø -t/-Ø -t/-Ø -t/-Ø -t/-Ø -t/-Ø -k/-Ø

MC/PMJ 七 日 熱 腹 不 結

7.2 7.3 7.5 7.6 8.3 8.7

-t/-Ø -t/-th -t/-Ø -k/-ɡ -t/-Ø -t/-Ø

Note: Before the “/” is the Middle Chinese coda, and after is the Mài Jué coda. The numbers after the character indicate the line number and the position of the character in this line, e.g. 百 1.6 means百 is the sixth character on the first line (refer to Table 8.13).

Old Mandarin: The Persian Transcriptions

307

syllables are never transcribed with a coda regardless of how many times they appear. The Chinese word mài 脈 is a typical example. In the Middle Chinese system, mài 脈 has a -k coda. In the poems of the PMJ the transcription for 脈 appears forty-six times from page 430 to 518. Except for one mistake on page 474 (where it is transcribed as ‹ tʃū ›),5 mài 脈 is consistently transcribed as ‹ māī › with no exception. No coda -k is found in the transcription of all fortysix cases. Such uniformity in spelling is quite different from other rù syllables, which usually show a certain degree of variation in terms of the presence or absence of the stop codas. As shown in Table 8.15, in the transcriptions some syllables have a coda and some do not, and some are even transcribed with different codas. For example, jí 急 is transcribed as ‹ kī ›, ‹ khē ›, ‹ kīp ›, and ‹ kht ›. The character jí 急 is a Middle Chinese syllable with coda -p. The coda -t in ‹ kht › is unexpected. However, in general the stop codas in the transcription are in agreement with that in the Middle Chinese system. The character mài 脈 is one of the few characters that shows an invariant transcription. Thus, it is quite likely that as a frequently used technical term the transcription of mài 脈 ‘pulse’ is fixed rather than spontaneously transcribed according to the exact pronunciation. Table 8.15 Chinese characters with different transcriptions in the Persian Mài Jué 急

430.2 ɡip 432.6 ɡip 432.10 khip 469.12 ɡī 473.5 ɡī 473.15 ɡī 518.5 kht 519.8 khū 忽 431.5 hū 459.2 hū 468.11 hū 479.3 hū 482.5 hū 513.4 hūt 腹 430.7 fūk 448.1 fū 449.14 fū 472.2 fū 508.1 fūk 508.4 fūk 急 jí < kɰip ‘anxious,’ 忽 hū < hwot ‘suddenly,’ 腹 fù

8.3.3

ɡī khē

464.1 513.2

ɡī kīp

469.6 hū 516.9 hūt 461.6 fū 508.14 fūk < piuk ‘belly’

471.5



467.12 510.15

fū fūk

448.1 508.1

(12) (9) (10)

The Dialectal Features Reflected in the Transcription

Based on the presence or absence of the stop codas of the Middle Chinese rù syllables, Endō (1997) suggests that the part that does not show any stop coda of the rù syllable represents a pronunciation of a Mandarin dialect, and the part that shows stop codas of the rù syllables also represents Mandarin but with a southern accent. However, this impression needs to be further examined. The Chinese dialects in the north and in the south differ from each other in many 5

It is also possible that the vorlage used by the writers did not have the character mài 脈, or substituted it with another character in the poem.

308

A New Standard

different aspects phonologically. The preservation of the stop coda is just one of them. In order to better understand the dialect of the speaker(s) involved in the Persian transcription, more relevant phonological features should be examined. It has been an accepted view in Chinese dialectology that by the time of the early Southern Song dynasty (1127–1279) the major southern dialects had been formed (Yóu 1992). For the northern dialects, the formation of northern Mandarin can be firmly traced to the Liao dynasty (907–1125) (Shen 2007). Thus by the time of the Yuan dynasty (1279–1368) the major dialect groups had already developed and the main phonological features of major dialects were also well established. In the following sections these two parts of the PMJ are examined separately. The part without stop codas is abbreviated as PMJ-1 (pages 220–429, 436–448), and the part with stop codas as PMJ-2 (pages 430–434, 449–519). In order to identify the dialects, the following diagnostic features are adopted. These features can show a contrast between the northern and southern dialects. The northern dialect is represented by the Beijing dialect, the standard of Modern Mandarin, and the southern dialect by the Guangzhou dialect, the base dialect for standard Cantonese. 1 = presence or absence of a stop coda in the Middle Chinese rù syllables (e.g. bǎi 百, mài 脈, lè 樂, ruò 若, jué 覺). In the Modern Mandarin dialects, as well as early in the time of the Liao dynasty, the rù syllables have lost their stop coda. But the stop codas are systematically preserved in some southern dialects, especially in the Cantonese dialects. 2 = presence or absence of a palatal medial in Middle Chinese Rank-II syllables with guttural initials (e.g. xià 夏, jiā 家, xià 下, jué 覺, yǎo 齩, yā 鴉, yǎn 眼, jiǎ 甲, jiān 間, jiāng 江). The southern dialects usually do not have a palatal medial for Rank-II syllables. 3 = the vowel of Middle Chinese Rank-III syllables of má 麻 rhyme (e.g. yě 也, zhě 者, xié 邪, qiě 且, yè 夜). The main vowel of Middle Chinese Rank-III syllables of the má 麻 rhyme changed from a low vowel to mid vowel, ja > je, in the northern dialects and in Cantonese, but remains as a low vowel in many other southern dialects. 4 = diphthongized Middle Chinese rù syllables with -k (e.g. kè 客, gé 隔, bái 白, bǎi 百, zé 澤, zé 則, dé 得, zéi 賊, sāi 塞, kè 克, hēi 黑, mài 脈, mò 莫, lè 樂, è 惡, què 雀, què 鵲, ruò 若, jiǎo 腳, yào 藥, sè 色). As a result of sound change, diphthongs are from Middle Chinese -k syllables. Diphthongization of Middle Chinese rù syllables with -k is a decisive feature for the identity of northern Mandarin (for more information on this, see Section 5.1.2, Section 5.2.2.1, and Section 6.4.5). 5 = the -w- medial of the syllables with zhuàng group initials in the dàng 宕 and jiāng 江 rhyme groups (e.g. zhuàng 壯, shuāng 雙, chuáng 床,

Old Mandarin: The Persian Transcriptions

309

chuāng 瘡, shuāng 霜, shuǎng 爽). Unlike the Mandarin dialects, the southern dialects do not have the labial medial for these syllables in the phonology. 6 = phonetic value of the Middle Chinese rì 日 initial (e.g. rén 人, rè 熱, rù 入, ruò 若, rú 如, ròu 肉, rì 日). The reflexes of the Middle Chinese rì 日 initial are different in the northern and southern dialects. The retroflex ɻ mainly exists in the northern Mandarin-speaking area including Beijing. In other dialects the reflex is n, l, z, ʐ, or has been completely elided to Ø-. By the Yuan dynasty the major modern dialect groups had already developed, with their main phonological features distinguished, so if the informants were speakers of different dialects, their dialects should not be too difficult to

Table 8.16 Phonological attributes of the PMJ-1; contrast with Table 8.17 1. Absence of stop codas, -p, -t, -k 急 268.11 ɡī jí < kɰip ‘anxious’ 月 361.8 ūē yuè < ŋwjɐt ‘moon’ 得 278.12 dəī dé < tək ‘obtain’

入 359.11 rū rù < ȵip ‘enter’ 不 254.2 bū bù < pjut ‘not’ 赤 392.12 tʃhī chì < tɕhjɛk ‘red’

十 254.7 ʃī shí < dʑip ‘ten’ 出 352.5 tʃhū chū < tɕhwit ‘out’ 欲 289.9 īū yù < jok ‘desire’

2. Presence of palatal medial 家 264.5 ɡiā jiā < kɰa ‘home’

間 415.6 ɡīan jiān < kɰæn ‘inbetween’

下 250.7 hjā xià < ɦɰa ‘down’

3. Raised vowel 也 284.10 īē yě < ja ‘also’

邪 300.9 sē xié < zia ‘wicked’

夜 325.7 ī yè < ja ‘night’

4. Diphthong reflexes with -j, or -w 百 381.6 baī bǎi < pɰak ‘hundred’ 惡 336.15 āū è < Øɑk ‘evil’

色 388.13 ʂaī sè < ʂɨk ‘color’ 樂 262.3 laū lè < lɑk ‘happy’

得 278.12 dəī dé < tək ‘obtain’ 覺 414.13 ɡīeū jué < kɰɔk ‘wake up’

5. Presence of the -w- medial 壯 424.12 tʃhūāŋ zhuàng < tʂjɐŋ ‘strong’

床 287.3 tʃhāŋ chuáng < dʐiɐŋ ‘bed’

瘡 390.3 tʃhūāŋ chuāng < tʂʰiɐŋ ‘sore’

6. Presence of r- initial 人 284.8 rin rén < ȵin ‘person’

熱 253.9 rē rè < ȵjɛt ‘hot’

肉 424.12 reū ròu < ȵjuk ‘meat’

310

A New Standard

Table 8.17 Phonological attributes of the PMJ-2; contrast with Table 8.16 1. Presence of stop codas (the same characters used for PMJ-1) 急 513.2 ɡīb 入 480.5 reb 十 489.15 ʃib jí < kɰip ‘anxious’ rù < ȵip ‘enter’ shí < dʑip ‘ten’ 月 497.1 ūed 不 508.15 bud 出 490.7 tʃhud yuè < ŋwjɐt ‘moon’ bù < pjut ‘not’ chū < tɕhwit ‘out’ 得 434.9 deɡ 赤 430.3 tʃhaɡ 欲 509.9 īūɡ dé < tək ‘obtain’ chì < tɕhjɛk ‘red’ yù < jok ‘desire’ But 甲 434.12 khīā 急 473.5 ɡī 澀 469.14 ʃī jiǎ < kɰap ‘shell’ jí < kɰip ‘anxious’ sè < ʂip ‘astringent’ 八 488.2 bā 忽 471.5 hū 滑 469.14 hūa bā < pɰæt ‘eight’ hū < hwot ‘suddenly’ huá < ɦwɰæt ‘slippery’ 足 469.5 tshū 實 464.1 ʃī 藥 474.1 īā zú < tsjok ‘foot’ shí < dʑɨk ‘real’ yào < jɐk ‘medicine’ 2. Presence of palatal medial 夏 446.5 hjā xià < ɦɰa ‘summer’

下 510.13 hjā xià < ɦɰa ‘down’

齩 517.14 Īāū yǎo < ŋɰau ‘bite’

3. Raised vowel 也 485.11 ī yě < ja ‘also’

邪 480.14 sē xié < zia ‘wicked’

夜 509.11 ī yè < ja ‘night’

色 482.5 seɡ _ sè < ʂɨk ‘color’ 覺 497.1 ɡīaɡ jué < kɰɔk ‘wake up’

得 513.3 diɡ dé < tək ‘obtain’ 藥 519.8 īāɡ yào < jɐk ‘medicine’

莫 431.6 māū mò < mɑk ‘not’

得 464.13 deī dé < tək ‘obtain’

5. Absence of the -w- medial 壯 430.4 tshāŋ zhuàng < tʂjɐŋ ‘strong’

雙 498.15 ʃūŋ shuāng < ʂɰɔŋ ‘double’

瘡 473.4 dʒūŋ chuāng < tʂʰiɐŋ ‘sore’

6. Presence of r- initial 人 486.5 rin rén < ȵin ‘person’

熱 494.1 rē rè < ȵjɛt ‘hot’

肉492.1 rū ròu < ȵjuk ‘meat’

4. Presence of coda -k 白 481.15 bɡ bái < bɰak ‘white’ 錯 485.12 tshāɡ cuò < tsʰɑk ‘wrong’ But 白 483.9 bāī bái < bɰak ‘white’

identify based on modern dialects. Phonological attributes of the PMJ-1 and PMJ-2 are shown in Table 8.16 and Table 8.17, respectively. All these phonological features shown in Table 8.16 and Table 8.17 suggest a type of Mandarin dialect, and Feature 4 of Table 8.16 can further identify this

Old Mandarin: The Persian Transcriptions

311

Mandarin dialect as northern Mandarin, very similar to the Yuan capital dialect of Dadu 大都. According to the rhyming books of Yuan, the MGZY and the ZYYY, the Middle Chinese rù syllables with -k coda have undergone diphthongization, which is a very specific phonological characteristic of northern Mandarin (Shen 2008a, 2011). In the PMJ the diphthong reflexes with -j or -w can be found. The identification is thus different from what Endō observed and concluded (Endō 1997). (Endō did not find the rù reflexes with -w and thus made his conclusion accordingly. The examples provided clearly show the diphthong forms with -w, in the transcriptions of Chinese characters è 惡, lè 樂, jué 覺, ruò 若, jiǎo 腳, etc.) Although the phonological features fit the Mandarin phonology very well, a few transcriptions indicate that the northern Mandarin dialect was not the native dialect of the speaker. As shown in Table 8.17, phonologically PMJ-2 is similar to PMJ-1. A significant difference is the presence of the stop codas of the rù syllables (Feature 1 in Table 8.17). However, the frequency of occurrence is not very high. Such cases can also be found in PMJ-1: on page 303 dé 得 is transcribed as tkh, and on page 359 kè 克 is transcribed as khik, both with stop coda -k. But, on the other hand, diphthongization that is a feature of the Dadu dialect is also present in PMJ-2 (Feature 4 in Table 8.17). Table 8.18 A comparison of characters across phonological attributes of the Persian Mài Jué, along with modern Beijing (BJ), modern Guangzhou (GZ), and the dialect of the Zhōngyuán Yīnyùn 1. rù, -p, -t, -k

BJ ZYYY PMJ-1 PMJ-2 GZ

BJ ZYYY PMJ-1 PMJ-2 GZ

十 shí ʂɨ ʃi ʃī ʃib ʃɐp

月 yuè ɥe ɥɛ(iuɛ) ūē ūed jyt

2. Rank-II 赤 chì tʂhʅ tʃhi tʃhī tʃhiɡ tʃhɪk

下 xià ɕja xja hjā hjīā ha

眼 yǎn jɛn jan īān īān ŋan

3. má III 齩 yǎo jɑu jau – īāū ŋau

邪 xié ɕjɛ sjɛ sē sē tʃhɛ

夜 yè jɛ jɛ ī ī jɛ

4. rù, -k

5. dàng/jiāng finals

6. rì initial

白 覺 得 bái jué dé paj tɕiaw tɤ/tej paj kjaw tej bēɡīeū thī/deī bɡ/bāī ɡīāɡ diɡ/deī pak kɔk tɐk

壯 zhuàng tʂwaŋ dʒwaŋ tʃhūāŋ tshāŋ tʃɔŋ

人 肉 rén ròu ʐən ʐou ʒjən ʒjəu rin/ren rū/reū rin/ren rū/rūɡ jɐn jʊk

瘡 chuāng tʂwaŋ tʃhwaŋ tʃhuāŋ dʒūŋ tʃhɔŋ

熱 rè ʐɤ ʒjɛ rē rē jit

312

A New Standard

Table 8.19 A compact generalization of the phonological attributes shown in Table 8.18

BJ ZYYY PMJ-1 PMJ-2 GZ

1

2

3

4

5

6

-Ø -Ø -Ø -C/-Ø -C

-j-j-j-j-Ø-

e ɛ e/i e/i ɛ

-j, -w -j, -w -j, -w -j, -w/-C -C

-u-u-u-Ø-Ø-

ʐ-(ɻ-) ʒ-(ɻ-) rrj-

Table 8.18 supplies a comparison made between the modern dialects and the transcriptions of PMJ. Beijing and Guangzhou are selected as the modern dialects for comparison for the reason that these two dialects have a closer relationship to the Persian transcription according to the phonological characteristics. The ZYYY is also included to represent the Mandarin Chinese of the Yuan. The comparison in Table 8.18 can be further simplified as in Table 8.19. According to the comparison in Table 8.19, PMJ-1 and PMJ-2 differ on three phonological features (1, 4, and 5) but are in agreement on three features (2, 3, and 6). In comparison with other systems, PMJ-1 is very similar to the Beijing dialect as well as the ZYYY in all six diagnostic features examined. For PMJ-2 some features are similar to Cantonese (1, 3, 4, and 5) and some features are similar to Mandarin (1, 2, 3, 4, and 6). Since there is no such dialect that has all these features coexisting in the same system, PMJ-2 must be a mixture of different dialects. In PMJ-2 some syllables can be only understood as the result of dialect mixtures. The palatal medial and the -k coda of Middle Chinese Rank-II kāikǒu 開口 syllables can co-occur in the same syllable (e.g. jué 覺 492.13 khiāɡ, 497.1 ɡīaɡ). PMJ-2 was very possibly a version of Cantonese-accented Mandarin spoken by a Cantonese native speaker, who, because the Dadu Mandarin was the standard of the time, would have had to learn Mandarin. It has been shown that the two PMJ-2 sections show very different degrees of the occurrence of stop codas. This difference can be regarded as different degrees of approximation of Mandarin or different degrees of Cantonese accent. But is PMJ-1 really based on the Mandarin pronounced by a native speaker? Based on the evidence pointed out earlier, in the PMJ-1 part, Middle Chinese rù syllables dé 得, sè 色, and kè 克 are clearly transcribed with the stop coda -k. These three crucial transcriptions indicate that the speaker of PMJ-1 is not a native Mandarin speaker either. Due to the total loss of stop codas in the Yuan time, no Mandarin speaker would pronounce the rù

Old Mandarin: The Persian Transcriptions

313

syllables with a stop coda. Thus, the speaker of PMJ-1 must be a Cantonese native speaker as well. On page 288 the character hù 護 is transcribed as ‹ fū ›, a confusion of hwand f-. This feature also suggests the Cantonese accent. In Modern Cantonese hù 護 is pronounced wu. But fū 夫 and hū 呼 are not distinguished and both are pronounced as fū in Cantonese. Since hù 護’s pronunciation is xu in Mandarin, regardless of the tonal difference, it could be pronounced as fū according to Cantonese. 8.3.4

The Nebulous Nature of Early Guānhuà 官話

The Chinese term for Mandarin, then standard, is Guānhuà 官話, which literally means ‘the speech of officials.’ Guānhuà served as a koine, or the common dialect of Yuan. The diphthongization of Middle Chinese syllables with -k coda also unambiguously indicates that the Yuan Guānhuà is based on the dialect of its capital, Dadu. However, Guānhuà is not the capital dialect, nor even any particular dialect at all. On one hand Guānhuà must have its base dialect, and on the other hand Guānhuà is quite loosely defined. It can contain certain nonstandard features, as far as the basic features of the standard phonology are included. So Guānhuà should be considered a more nebulous standard, rather than the rigid standards of the Qièyùn and other rhyme dictionaries. Guānhuà had many different realizations among different speakers.6 In the PMJ some phonological features indicate that the underlying dialect of the speaker(s) is similar to modern southern dialects, especially Cantonese. All transcriptions of Chinese in the three parts are based on the pronunciation spoken by native speaker(s) of Cantonese, with different degrees of approximation to the standard Mandarin. The main difference is reflected in the pronunciation of the Middle Chinese rù syllables. In the transcription of all the poems of both PMJ-1 and PMJ-2, two features appear consistently, namely the palatal medial of Middle Chinese Rank-II kāikǒu syllables -j-, and the approximate reflex of Middle Chinese initials rì 日 as ‹ r ›. These two features clearly indicate that they are the most distinct and basic features of the standard Mandarin phonology. In modern dialects similar phenomena can be observed. The so-called literary layer and the colloquial layer have contrast for these two kinds of syllables. For example, in the Suzhou dialect (SZ in Table 8.20), a variety of Wu, the wéndú 文讀 6

Another version of Guānhuà in Persian transcription can be observed in Rashid al-Din’s History of China, which does not show any -p, -t, or -k coda in the transcription. The phonological features in that material indicate that the underlying phonological system is, in general, Northern Mandarin.

314

A New Standard

Table 8.20 A comparison of Beijing and Suzhou (SZ) dialects, reflecting the differences in rì 日 initial syllables and Rank-II kāikǒu syllables MC Rank-II kāikǒu syllables 家 ‘home’ jiā < kɰa Beijing tɕja SZ literary tɕjɒ SZ colloquial kɒ

間 ‘in between’ jiān < kɰæn tɕjɛn tɕjɪ kɛ

江 ‘river’ jiān < kɰɔŋ tɕjaŋ tɕjɒŋ kɒŋ

覺 ‘wake up’ jué < kɰɔk tɕye/tɕjaw tɕjoʔ koʔ

MC rì 日 initial syllables 日 ‘sun’ rì < ȵit Beijing ʐɨ SZ literary zɤʔ SZ colloquial niɪʔ

人 ‘person’ rén < ȵin ʐən zən nin

染 ‘dye’ rǎn < ȵjɛm ʐan zø niɪ

戎 ‘army’ róng < ȵjuŋ ʐuŋ zoŋ njoŋ

‘literary pronunciation’ mainly shows its contrast with the báidú 白讀 ‘colloquial pronunciation’ in these two kinds of syllables (Běijīng Dàxué 2003). In Table 8.20, the Beijing dialect is listed to represent the standard Chinese. The literary pronunciation of Suzhou reflects the phonological features of standard Chinese and shows the similarity in these two kinds of syllables. It is also worth noting that in the PMJ the appearance of the stop coda should not be totally unintentional. The presence of the stop coda suggests that retaining the pronunciation of rù tone syllables along with their stop coda was an acceptable practice among speakers who had the dialectal background and could correctly produce them. In the Yuan dynasty, the rù tone as a tonal category was preserved in the standard rhyming books, such as the MGZY and the ZYYY. The poems of the Mài Jué are written in the traditional forms. The rhythm and rhyme are based on Middle Chinese tonal categories. Thus, to produce the rù tone syllables in reading traditional poems could be an acceptable practice in the standard pronunciation even though the rù tone was lost in the northern standard dialect. Based on what has been discussed, there are two possible explanations for the accented Chinese pronunciation reflected in the PMJ. One is that different sections, PMJ-1 (pages 220–429 and 436–448), PMJ-2a (pages 430–434), and PMJ-2b (pages 449–519), were produced by different Chinese speakers with different degrees of dialect accent in their pronunciation of standard Chinese. An alternative is that the same Chinese speaker, very likely a Cantonese speaker, pronounced the standard Chinese with significantly different degrees of dialectal accent. Since the dialectal features can be traced to Cantonese, the different degrees of occurrence of the stop coda as well as other dialectal features can be considered to be produced by the same speaker. According to

Old Mandarin: The Persian Transcriptions

315

the introductory part of the Tanksūqnāma-i Īl-khān dar funūn-i ‘ulūm-i Khatāyī, “it is mentioned by its compiler Rashīd that he asked a promising young Persian scholar by the name Safīi al-Daula vaal-Dīn to follow Siusa (Siu-Seh), one of the most eminent Chinese physicians working in Persia (modern Iran) at that time” (Wáng Yīdān 2006: 35). Thus, it is very possible that all the Mandarin Chinese was pronounced by Siu-Seh, whose native dialect was Cantonese. In the process of the translation project, he progressively introduced more dialectal features into his pronunciation of Mandarin (Shen 2016b).

Part VI

Toward Modern Mandarin Ming dynasty (1368–1644), Qing dynasty (1644–1912), Republican period (1912–1949), People’s Republic of China (1949–present)

The Yuan dynasty ended in 1368. An ethnically Chinese ruler was installed as the new Emperor of China and the Ming dynasty began its rule. The location of the capital changed from Dadu (modern Beijing) to Nanjing (1368–1421), and later back to Beijing (1421–1644). Because of the different locations of the capital and other issues related to the standard pronunciation, the question of which dialect served as a foundation for the standard pronunciation of the Ming has been a point of strong interest to researchers. In 1375 the rhyme dictionary Hóngwǔ Zhèngyùn 洪武正韻 ‘The Standard Rhymes of Hongwu’ was compiled at the order of the first Ming emperor, Zhū Yuánzhāng 朱元璋, due to his dissatisfaction with the existing rhyme dictionaries. The difficulty faced in the production of this rhyme dictionary was to compile a dictionary that on the one hand followed the traditional phonology and on the other hand reflected contemporary pronunciation. For such an impossible mission, the emperor was not so satisfied with not only the first, but also the revised second version. In the fifteenth century Korean scholars led by Sin Suk-chu 申叔舟 (1417–1475) made a great effort to transcribe this prestigious phonological work into the newly invented Hangul spellings. The result was the Hóngwǔ Zhèngyùn Yìxùn 洪武正韻譯訓, finished in 1451 with a revision in 1455. The Hóngwǔ Zhèngyùn Yìxùn provides phonetic values for the phonology of the Hóngwǔ Zhèngyùn. Two Chinese textbooks with Hangul spellings, the Fānyì Lǎo Qǐdà 翻譯老乞大 ‘Lǎo Qǐdà with phonetic notations’ and the Fānyì Piáo Tōngshì 翻譯朴通事 ‘Piáo Tōngshì with phonetic notations’ authored by Ch’oe Se-jin 崔世珍 (1473–1542) provide systematic transcriptions of spoken Mandarin even with the tonal marks. The Xīrú Ěrmù Zī 西儒耳目資 ‘An Aid to the Eyes and Ears of Western Literati’ of 1626 by Nicolas Trigault (1577–1628) is one of the first instances of a work in which Chinese was transcribed into an alphabetical system without referring to Chinese rhyme works. While it is quite certain that the phonology of the Xīrú Ěrmù Zī is Mandarin, the linguistic identity of its base dialect has been controversial. This uncertainty of its base dialect suggests that Mandarin phonology is a set of phonological features that are exclusively Mandarin instead of a dialect of a certain location. It can be observed that this 317

318

Toward Modern Mandarin

spelling system, which uses the Latin alphabet, has its influence on the design of the modern romanization system for Chinese, pinyin. In 1618 the Manchus, led by Nurhaci (1559–1626), began to revolt in Liaoning, and eventually formed the Qing dynasty in 1636 and became the rulers of China in 1644. Just as their Mongol predecessors had, the Manchus of the Qing dynasty set their capital in Beijing. In the Qing dynasty there was also an effort by the government to establish a phonological standard, and in pursuit of this goal the Yùdìng Pèiwén Yùnfǔ 御定佩文韻府 was published in 1711 and the Yīnyùn Chǎnwēi 音韻闡微 was published in 1728 with an improved fǎnqiè 反切 system. It is interesting to note that these two rhyme dictionaries did not follow the Hóngwǔ Zhèngyùn of 1375, the rhyming standard established in the Ming dynasty. Instead it followed the earlier rhyming tradition of the Píngshuǐ Yùn 平水韻 of the thirteenth century. Like all previous rhyme dictionaries, however, it still represents an ideal phonology. The Imperial Era of China ended in 1911 after the Xīnhài 辛亥 revolution. The capital of the Republic of China started in Nanjing. However, it switched from Nanjing (1912) to Beijing (1912–1927), and then back to Nanjing (1927–1949). After the Chinese civil war (1927–1950), the capital of the People’s Republic of China was established in Beijing. The Beijing dialect was chosen as the base for the national standard pronunciation during both time periods. The colloquial pronunciation of the Beijing dialect was well transcribed in the Yǔyán zì ěr jí 語言自邇集 Yü-yen tzǔ-êrh chi: A Progressive Course Designed to Assist the Student of Colloquial Chinese by Thomas Francis Wade (1818–1895) (see Section 10.1.3).

9

The Mandarin of the Ming Dynasty

Although the phonological standard of the Ming dynasty was mainly a literary standard, the question of which dialect served as the foundation for the standard pronunciation of Ming has been a point of strong interest to researchers. 9.1

North Versus South

The regime change from the Yuan dynasty to the Ming dynasty caused a shift of emphasis of the phonological standard to the Southern Mandarin of the new capital, Nanjing. It is likely that the dialect of the first Ming emperor, Zhū Yuánzhāng 朱元璋, was also a reference for the new standard. It should be noted, however, that Nanjing Mandarin was simply more emphasized or prestigious than the many other local standards: the dialects of metropolises within the larger Mandarin-speaking area. Because of the limitation of the transmission of spoken language before modern times, a standard could only influence a limited geographical area where a large population of speakers of the standard pronunciation was accessible. The dialects of metropolises within the Mandarin-speaking area, such as Beijing, Luoyang, and Nanjing, served as local standards in their respective areas. Such a linguistic situation was quite suitable considering the very large Mandarin-speaking area; it would have been impossible to have a standard based on a single dialect for the whole country without modern media to promote it. Because of the great geographical distance and limited means of transportation, the influence of Southern Mandarin dialects might not have been significant in central and northern China, leaving only a few syllables influenced by Nanjing Mandarin; for example, lòng 弄 ‘lane, alley,’ which is now pronounced nòng due to a lack of distinction between initials n- and l- in Nanjing Mandarin (Yú Mǐn 1984b). Regardless of whether the northern or southern variety was more prestigious, however, Mandarin was firmly established as the national standard.

319

320

Toward Modern Mandarin

9.1.1

Guānhuà 官話

In the Ming dynasty the official language was called guānhuà ‘language of the officials,’ or “Mandarin” by Westerners. The term Mandarin first appeared in the sixteenth century when Jesuit missionaries learned it as the standard language or the supradialectal koine of China, as it was conventionalized for the purposes of communication (mainly between government officials), and contained common phonological features of the capital and metropolises in the Mandarin-speaking area, though its pronunciation varied from place to place (Coblin 2000). It is likely that as far as the purpose of communication was satisfied, the phonetic details behind the koine may not have been of the greatest interest to its users. As Jerry Norman stated, “Neither of these forms of bureaucratic Chinese was codified in any fashion; they simply developed as a natural response to the need for a practical medium to carry on the day-to-day business of the empire” (Norman 1988: 133). Guānhuà has a number of basic Mandarin characteristics, such as its devoicing pattern and the loss of stop codas. The history of this phonological lineage is referred to as the history of Mandarin, therefore the phonology of Yuan times is called Old Mandarin even though the word guānhuà did not exist then. The rhyme works of this period also reflect an ideal standard and most show the composite nature of guānhuà to varying degrees. Quite a number of phonological works are available, including two systematic transcriptions in foreign alphabets: the Sìshēng Tōngjiě 四聲通解 by Ch’oe Se-jin 崔世珍, published in the early sixteenth century, and the Xīrú Ěrmù Zī 西儒耳目資 by Nicolas Trigault, published in 1626. With better information from available materials, phonetic details and the identification of a base dialect have become a focal point for researchers. 9.2

The Hóngwǔ Zhèngyùn 洪武正韻 (1375)

The Hóngwǔ Zhèngyùn ‘The Standard Rhymes of Hongwu’ (HWZY), compiled by Yuè Sháofèng 樂韶鳳 and ten other scholars, was an officially published rhyme dictionary in sixteen volumes (see Figure 9.1). It represents the government’s effort at the beginning of the Ming dynasty to define a new national standard. Because it was an ideal standard, it included many traditional distinctions in order to add prestige, and as such was a continuation of the tradition of rhyme dictionaries. It was compiled based on the Zēngxiū Hùzhù Lǐbù Yùnlüè 增 修互註禮部韻略, a traditional rhyme dictionary of the Southern Song dynasty by Máo Huǎng 毛晃 and Máo Jūzhèng 毛居正 (Nìng 2003). Although many detailed changes were made, the HWZY is actually much more conservative than previous rhyming works such as the Zhōngyuán Yīnyùn 中原音韻 (ZYYY) and even the Ménggǔ Zìyùn 蒙古字韻 (MGZY). For example, the voicing distinction was kept for stops, affricates, and fricatives, the tonal system of píng 平, shǎng 上, qù 去, and rù 入 was preserved, and all consonant codas were

321

Figure 9.1 A page from the Hóngwǔ Zhèngyùn 洪武正韻 (Sìkù Quánshū 四庫全書 version), showing the first part of the first rhyme, the dōng 東 rhyme, as indicated in the fifth column, 一東. The homophonic groups are separated by ○. Fǎnqiè spelling of each homophonic group is provided after the group’s first character.

322

Toward Modern Mandarin

distinguished; these phonological distinctions had already changed or completely disappeared in Mandarin by the time that the HWZY was published. This composite nature was actually the purpose of the HWZY. According to the compilers, the standard should be mutually intelligible by all people. A standard based on a single dialect clearly did not fit such an ideology. Such a concept is clearly stated in the ‘Guide to Use’ of the HWZY: What is the standard pronunciation? The standard pronunciation is what can be understood by people in all places.1

9.2.1

Initials

According to Liú Wénjǐn’s (1931) study of the fǎnqiè 反切 spellings, the HWZY has thirty-one initials.2 These phonetic values had already been systematically worked out by Korean scholars in the fifteenth century, coming to the same conclusion. As shown in Table 9.1, despite the fact that voiced obstruents were lost in Mandarin by the time of the Ming dynasty, a distinction still exists within the Hóngwǔ Zhèngyùn. The initial characters in Table 9.1 are used in Liú (1931) to represent initials. In comparison with the thirty-six initials of Middle Chinese, the changes are mainly the merger of initials of the zhī 知 and zhào 照 series, namely zhī 知 ʈ- = zhào 照 tʃ- > zhì 陟 tʂ-, chè 徹 ʈh- = chuān 穿 tʃh- > chǒu 丑 tʂh-, chéng 澄 ɖ- = chuáng 床 dʒ- = shàn 禪 ʒ- (partially) > zhí 直 dʐ-, the merger of two labiodental initials, and the merger of two nasal initials, ní 泥 n- = niáng 娘 ɳ- > nú 奴 n- and fēi 非pf- = fū 敷 pfh- > fāng 方 f-. The four initials of the zhī 知 initial series lost their distinction with other initials and the two labiodental initials fēi 非 and fū 敷 merged. So the thirty-six initials were reduced to thirty-one in the HWZY. 9.2.2

Finals

The final system in the HWZY also shows a conservative approach. There are twenty-two rhyme groups for syllables with píng, shǎng, and qù tones, Table 9.1 The thirty-one initials of the Hóngwǔ Zhèngyùn k古 tʂ 陟 ts 子 t 都 p博

1 2

kh 苦 tʂh 丑 tsh 七 th 佗 ph 普

ɡ 渠 dʐ 直 dz 昨 d 徒 b 蒲

ŋ 五 n 奴 m莫

x呼 ʂ 所 s 蘇

ɣ胡 ʐ 時 z 徐

f 方

v 符

Ø烏 ɻ 而 l 盧 ʋ 武

j以

“欲知何者為正聲,五方之人皆能通解者斯為正音也。”《洪武正韻•凡例》 All the fǎnqiè 反切 spellings in the HWZY were copied from the Zēngxiū Hùzhù Lǐbù Yùnlüè (Nìng 2003).

The Ming Dynasty

323

and ten for syllables with rù tone. This rearrangement did not simply merge traditional rhymes; there is clear evidence that the compilers made judgments on every word (Wáng Lì 1956: 508). In the fifteenth century Korean scholars Sin Suk-chu 申叔舟 et al. made a great effort to transcribe this prestigious phonological work into the newly invented Hangul spellings. The result was the Hóngwǔ Zhèngyùn Yìxùn 洪武正韻譯訓, finished in 1451 with a revision in 1455. The interpretation of Hangul spellings is closely based on the intrinsic value of individual Hangul letters, thus they must be interpreted according to the well-established phonetic values of Chinese. The relationship between Hangul letters, their values, and Ming Chinese values is given in Table 9.2. In transcribing Chinese, only eight out of eleven vocalic letters and their combinations are used. The eight coda letters are also given. Table 9.2 The relationship between Hangul letters, the intrinsic values of Hangul, and Ming Chinese values Vocalic letters Hangul ㅡ Korean ɨa Chinese ɨ, ə

ㅣ i i

ㅜ u u

ㅠ ju y

ㅏ a a

ㅓ ə ɔ

ㅑ ja ja

ㅕ jə jɛ

Consonant codas Hangul ᇰ Korean -ŋ Chinese -ŋ

ㄴ -n -n

ㅁ -m -m

ㄱ -k -k

ㄷ -t -t

ㅂ -p -p

ᄝ -w -w

ㅣ -j -j

The phonetic value for ㅡ is a central vowel. It covers a phonetic value ranging from ɨ to ə, which is similar to the hP’ags-pa transcription (refer to Section 6.4.8). _

a

For the Chinese medial -w-, the letter ㅗ is used in front of finals with -a as the main vowel, and the letter ㅜ is used in front of other vowels. For the Chinese medial -ɥ-, the letter ㅠ is used. The Chinese palatal coda -j is represented by letterㅣ, and the labial coda -w is represented by a special letter ᄝ. This is the ideal system established by the Ming government (e.g. all finals with nasal coda -m, -n, or -ŋ have corresponding finals with a stop coda, -p, -t, -k). Although not representing any real dialect, it is quite internally consistent. Based on the phonetic values in Table 9.3, the phonological characteristics of this system can be summarized as is in Table 9.4. 9.2.3

Tones

The HWZY uses the traditional four-tone system, píng, shǎng, qù, and rù for all the syllables. The seventh fánlì 凡例 ‘Guide to Use’ of the dictionary

324

Toward Modern Mandarin

Table 9.3 The phonetic values of the finals in the Hóngwǔ Zhèngyùn Yìxùn Pinyin 1



dōng

2 3 4 5 6 7 8

支 齊 魚 模 皆 灰 真

zhī qí yú mó jiē huī zhēn

9



hán

10



shān

11



xiān

12 13 14 15 16 17

蕭 爻 歌 麻 遮 陽

xiāo yáo gē má zhē yáng

18



gēng

19 20

尤 侵

yóu qīn

21



tán

22



yán

Hangul intrinsic values uŋ uk ɨ

Ming Chinese values

juŋ juk

uŋ uk ɨ

i jəj

i jɛj

ju aj

jaj

ɨn ɨt ən ət an at

in it

aw ə a aŋ ak ɨŋ ɨk ɨw ɨm ɨp am ap

jan jat jən jət jəw jaw ja jə jaŋ jak iŋ ik iw im ip jam jap jəm jəp

u waj uj un ut wən wət wan wat

jun jut

y aj

jaj

ən ət ɔn ɔt an at

in it

jwən jwət aw ɔ a

wə wa jwə waŋ wak uiŋ uik

juiŋ juik

yŋ yk

aŋ ak əŋ ək əw əm əp am ap

jan jat jɛn jɛt jɛw jaw ja jɛ jaŋ jak iŋ ik iw im ip jam jap jɛm jɛp

u waj uj un ut wɔn wɔt wan wat

yn yt

ɥɛn ɥɛt

wɔ wa ɥɛ waŋ wak wəŋ wək

ɥəŋ ɥək

Source: (Hangul only) Nìng 2003: 73

Table 9.4 A final chart of the Hóngwǔ Zhèngyùn Yìxùn Medials

j, w, ɥ

Main vowels

i•y ɛ

Codas

j, w, m, n, ŋ, p, t, k

ɨ ə a

u ɔ

The Ming Dynasty

325

mentions the merging pattern of rù syllables, some of which become homophonous with yángpíng 陽平, shǎng, and qù tone syllables. 9.2.4

A Coherent System

The initials and finals of the HWZY actually form a very logical, ideal phonological system. The subsystems, initials, medials, main vowels, and codas do not violate any general rules of traditional Chinese phonology. The conservative features incorporated coexisted in harmony with other features. This type of phonology reflects a special effort made by Chinese phonologists, and its coherent internal relationships are shown in phonetic transcriptions by Korean scholars. Still, such an ideal phonology probably caused a great amount of difficulty for Korean scholars: eight years were spent in the effort to provide Hangul transcription for all the syllables in the HWZY. In the process, they also realized that the real pronunciation of any dialect was different from the HWZY. Sin Suk-chu called a more realistic set of pronunciations súyīn 俗音 ‘vulgar or common pronunciation’ and the pronunciations that kept traditional categories and served as a literary standard zhèngyīn 正音 ‘correct pronunciation.’ The zhèngyīn, in a sense, is actually an artificially “constructed” system. 9.2.5

Zhèngyīn 正音 and Súyīn 俗音

The “correct” zhèngyīn is the “standard pronunciation” according to the HWZY, and the “vulgar” súyīn indicates the actual (more or less) pronunciation of Beijing, according to the preface of the Hóngwǔ Zhèngyùn Yìxùn. The correct standard is not the real pronunciation of any dialect or, more precisely, of any individual because the literary standard is conservative and semi-reconstructed. It is considered “correct” because it is a national standard. The work of the Korean scholars was to provide the phonetic values of these categories. Of course, these two terms reflect the perception of the phonological standard at the time. Significant differences between the two systems include the following (Nìng 2003: 78–80): General changes (from zhèngyīn to súyīn) A general statement about the difference between zhèngyīn to súyīn is found at the beginning of the following rhymes. Loss of -m coda Rhyme qīn 侵: -im > -in, rhyme tán 覃: -am > -an

326

Toward Modern Mandarin

Loss of -p, -t, -k codas Rhyme wū 屋: -uk > uɁ, rhyme zhì 質: -it > -iɁ, rhyme jī 緝: -ip > -iɁ Rhyme yào 藥: -jak > -jaw (showing diphthongization) High central vowel in the zhì 寘 rhyme Rhyme zhì 寘: -i > -ɨ High front vowel in the qí 齊 rhyme Rhyme qí 齊: -jɛj > -i Final -uan and -on merger Rhyme hán 寒 syllables with non-guttural initials change to > rhyme shān 刪: -wɔn > -wan Loss of -jɛw final Rhyme xiāo 蕭 mergers into rhyme yáo 爻: -jɛw > -jaw Changes that relate to special characteristics The same changes are also found in many syllables in Modern Mandarin (MM). Loss of palatal medial in some dōng 東 rhyme syllables, including the rù syllables 從: -yŋ > -uŋ, MM: cóng ‘follow’; 蟲: -yŋ > -uŋ, MM: chóng ‘insect’ 弓: -yŋ > -uŋ, MM: gōng ‘bow’ 夙: -yk > uk, MM: sù ‘morning’; 叔: -yk > uk, MM: shū ‘uncle’ Merger of some Middle Chinese gēng 庚 rhyme syllables with dōng 東 rhyme syllables 轟: -wɛŋ > -uŋ, MM: hōng ‘rumble’; 榮: -ɥɛŋ > -yŋ, MM: róng ‘glory’ The final of Middle Chinese jiē 皆 rhyme syllables with guttural initials is -jej 皆: -jaj > -jɛj, MM: jiē ‘all’; 諧: -jaj > -jɛj, MM: xié ‘harmony’ The final of Middle Chinese yáng 陽 rhyme syllables with zhuāng 莊 group initials is -waŋ 莊: -aŋ > -uaŋ, MM: zhuāng ‘village’; 霜: -aŋ > -uaŋ, MM: shuāng ‘frost’ 9.2.6

Revision of the Hóngwǔ Zhèngyùn

The HWZY was revised at the request of the emperor in an effort to make a more consistent ideal standard: the Zhōngyuán Zhèngyīn 中原正音 ‘Standard Pronunciation of the Central Plain.’ The total number of rhymes increased from seventy-six to eighty (Nìng 2003). One significant change was the rearrangement of the zhī 支 rhyme, which was in contrast with the 齊 qí rhyme and should have had a high central vowel -ɨ for its phonetic value. In the

The Ming Dynasty

327

Table 9.5 A comparison of the zhī 支 rhyme before and after the split

76 rhyme version zhī 支 80 rhyme version zhī 支 wēi 微

ts- group

tʂ- group

k- group

p- group

t- group

+

+

+

+



+ –

+ –

– +

– +

– –

Table 9.6 A comparison of the modern Beijing and Hefei dialects

Beijing Hefei

Beijing Hefei

比 bǐ ‘compare’

披 pī ‘cover’

迷 mí ‘confused’

低 dī ‘low’

梯 tī ‘ladder’

泥 ní ‘mud’

雞 jī ‘chicken’

-i -ɨ

-i -ɨ

-i -ɨ

-i -ɨ

-i -ɨ

-i -ɨ

-i -ɨ

妻 qī ‘wife’

西 xī ‘west’

醫 yī ‘medical’

姐 jiě ‘sister’

且 qiě ‘also’

斜 xié ‘oblique’

夜 yè ‘night’

-i -ɨ

-i -ɨ

-i -ɨ

-ie -i

-ie -i

-ie -i

-ie -i

revised version shown in Table 9.5, however, the zhī 支 rhyme contains the only syllables with alveolar and retroflex initials. Syllables with other initials were placed in the newly established wēi 微 rhyme. There is a question of why these syllables were placed together in the first place. It seems that the compilers made an effort to incorporate some phonological features of the emperor’s dialect – that of Fengyang, Anhui – into the system of the original version. In Ming times, the Fengyang dialect may have been similar to some modern Anhui dialects, which are distinct in that the high central vowel exists in a broader phonological environment than in the Beijing dialect, shown in Table 9.6. This is the result of a phonological process centralizing the front high vowels as part of a vowel shift, -ie > -i, -i > ɨ (Běijīng Dàxué 2003). In the revision, this phonological distinction is changed in order to follow a more general pattern of Mandarin, rather than that of the emperor’s Anhui version, which was not satisfactory to many, including the emperor himself (Shen 2010). 9.3

The Yùnluè Yìtōng 韻略易通 (1442)

The Yùnluè Yìtōng (YLYT) is a primer for children in the form of a rhyme book, compiled by Lán Mào 蘭茂 (1397–1476) with a preface dated to 1442.

328

Toward Modern Mandarin

Its phonological system reflects some advanced features of Mandarin, especially in the initial system (Lù 1947a; Gěng 1992).

9.3.1

Initials and the Zǎo Méi 早梅 ‘Early Plum Blossom’ Poem

In the YLYT Lán Mào cleverly uses the initial characters to form a poem. The function of the initial characters is the same as the traditional thirty-six initial characters. The poem is in the format of wǔjué 五絕, having four lines, each with five syllables. The phonetic values (row 2) and pinyin (row 3) of the Chinese syllables in the Zǎo Méi poem are given in Table 9.7. Zǎo Méi 早梅 Early Plum Blossom by Lán Mào 蘭茂 東風破早梅 The eastern wind wakes the plum trees 向暖一枝開 Toward the warmth, one branch blossoms, 冰雪無人見 The ice and snow still hide from man, 春從天上來 From the heavens, spring has come. (English translation mine)

Table 9.7 Zǎo Méi poem with phonetic values and pinyin 東 tuŋ dōng

風 fəŋ fēng

破 pho pò

早 tsaw zǎo

梅 mej méi

向 ɕjaŋ xiàng

暖 nuan nuǎn

一 i yī

枝 tʂɨ zhī

開 khaj kāi

冰 piŋ bīng

雪 ɕɥɛ xuě

無 u wú

人 ɻən rén

見 tɕjɛn jiàn

春 tshun chūn

從 tsuŋ cóng

天 thjɛn tiān

上 ʂaŋ shàng

來 laj lái

The significance is not in the artistic presentation but in that this is the first time Mandarin initials are listed. As the name suggests, Chinese rhyme works mainly focus on rhymes, and thus often leave glaring holes in the data of initial systems. When this happens, only analyzing homophonic groups of the same rhyme can retrieve an initial system. As pointed out by Zhào (1957: 210), before the YLYT no one explicitly provided a complete list of the initials.

The Ming Dynasty

329

Table 9.8 By reorganizing the characters of Zǎo Méi, the initial system of the Ming dynasty is illuminated p t ts tʂ k

冰 東 早 枝 見

ph th tsh tʂh kh

破 天 從 春 開

m 梅 n 暖

f 風 s 雪 ʂ 上 x 向

ʋ 無 l 來 ɻ 人 Ø一

Table 9.9 Conditional variants of zhī 枝, chūn 春, and shàng 上 initials 枝春上

> >

tʂ tʂh ʂ / _VE tʃ tʃh ʃ /_ jVE

By reorganizing the characters of Zǎo Méi 早梅, the initial system of the Ming dynasty is illuminated in Table 9.8. The phonetic values of the initials represented by the characters zhī 枝, chūn 春, and shàng 上 require further explanation. In the phonology of the YLYT these initials can combine with finals with or without a palatal medial. Some scholars argue that the phonetic value in front of the finals without a palatal medial can be retroflex, but in front of finals with a palatal medial, the realization of these initials should be palatalized because it is natural in terms of articulation. Because these two groups of initials are conditional variants, they can be treated as the same on a phonemic level, as shown in Table 9.9. Some others argue that retroflex initials and finals with a palatal medial can combine, although this may not be the most natural. This argument is similar to that concerning retroflex initials in the Zhōngyuán Yīnyùn 中原音韻 (ZYYY); based on that discussion, there should be just one series of retroflex initials (see Chapter 7). A later revision by a Buddhist Monk named Běnwù 本悟 (1510–1599) from Yunnan shows the deletion of the palatal medial after zhī 枝, chūn 春, and shàng 上 initials, indicating that zhī 枝, chūn 春, and shàng 上 in front of palatal glides are retroflex initials. If there were palatalized initials, then there is no reason to delete the palatal medial. 9.3.2

Finals and the Twenty Rhyme Groups

The YLYT divides the finals into twenty groups, listed in Table 9.10. The phonetic values indicate the rhyme or the rhyme part of the finals (Zhào 1957). According to Lán Mào, rhyme groups with syllables of all four tones are listed first, followed by syllables without rù tone. The first ten rhyme groups are for

330

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20

東洪 江陽 真文 山寒 端桓 先全 庚晴 侵尋 緘咸 廉纎 支辭 西微 居魚 呼模 皆來 蕭豪 戈何 家麻 遮蛇 幽樓

Rhyme Group

dōng-hóng jiāng-yáng zhēn-wén shān-hán duān-huán xiān-quán gēng-qíng qīn-xún jiān-xián lián-xiān zhī-cí xī-wēi jū-yú hū-mó jiē-lái xiāo-háo gē-hé jiā-má zhē-shé yōu-lóu

公 崗 跟 干 官 庚 簪 甘 資支 雞 居 姑 該 高 歌 巴 勾

uŋ aŋ ən an on

əŋ əm am

ɨ i y u aj aw o a

əw

Finals

waj

wa

堅 京 今 監 兼

皆 澆 加 茄 鳩

jen iŋ im jam jem

jaj jaw ja je jəw

wi

wəŋ

恭 江 巾 間

juŋ jaŋ in jan

waŋ un wan









光 昆 關

ɥe

ɥen yŋ

yn



娟 扃



Table 9.10 Rhyme groups of the Yùnluè Yìtōng and their respective finals

ək əp ap

uk ak ət at ot 格 戢 答

谷 閣 犵 葛 括 jet ik ip jap jep

juk jak it jat 結 力 立 甲 頰

局 角 吉 戛

uək

wak ut wat



郭 骨 刮

ɥet ɥək

yt

厥 鶪



The Ming Dynasty

331

finals with a nasal coda, and the second ten are for finals with a glide or no coda. These twenty rhyme groups contain the finals given in Table 9.10. As a conservative treatment, the first ten groups for finals with nasal codas also contain their corresponding rù syllables according to the Middle Chinese system, although earlier records show that in these rù syllables the Middle Chinese stop codas have changed to glottal codas in Mandarin, which leads to different rhyme groups containing similar rù syllables. 9.3.3

Běnwù’s 本悟 Revision

About a century later Běnwù revised the YLYT, publishing this in 1586. An important aspect to the book is his notes about duplicated rhymes, chóng x yùn 重 x 韻 ‘same rhyme as rhyme x,’ where x = the number of one of the rhyme groups. As shown in Table 9.11, these notations indicate the merger of syllables in the original version with different nasal and stop codas as well as different main vowels. Only the píng and rù syllables are used for illustration as píng syllables also represent shǎng and qù syllables. These notations indicate the merger of nasal and stop codas, a distinctive characteristic of Southern Mandarin, likely Yunnan dialect, which Běnwù spoke. As shown in Table 9.12, the mergers of the syllables from the rhyme groups shān-hán 山寒, jiān-xián 緘咸, lián-xiān 廉纖, and xiān-quán 先全 Table 9.11 Examples of the mergers in Běnwù’s revision -uŋ/k = -wən/t = wəŋ/k (東洪 = 真文 = 庚晴) 公谷 = 昆骨 = 觥國 烘斛 = 昏忽 = 薨或 恭菊 = 君橘 = 扃鶪 崗閣 = 干葛 = 甘閣

-aŋ/k = -an/t = -am/p (江陽 = 山寒 = 緘咸) 康恪 = 看渴 = 堪磕 夯鶴 = 寒曷 = 酣合

鄰栗 = 靈力 = 林立

-in/t = -iŋ/k = -im/p (真文 = 庚晴 = 侵尋) 親七 = 青戚 = 侵咠 辛悉 = 星昔 = 心習 -ən/t = -əŋ/k = -əm/p (真文 = 庚晴 = 侵尋) 臻櫛 = 爭側 = 簪戢

Table 9.12 The loss of the palatal medial -j- after retroflex initials an/t = am/p = jɛn/t = jɛm/p (shān-hán 山寒 = jiān-xián 緘咸 = lián-xiān 廉纖 = xiānquán 先全) 㣶扎 = 詀劄 = 毡折 =占懾 山殺 = 衫霎 = 膻舌 = 苫攝

332

Toward Modern Mandarin

indicate that the palatal medial -j- after retroflex initials are lost in the lián-xiān 廉纖 and xiān-quán 先全 rhyme groups. They became homophonous with their counterparts without the medial in the shān-hán 山寒 and jiān-xián 緘咸 rhyme groups. This loss subsequently lowered the main vowel from -ɛ to -a. Whether the rù syllables of lián-xiān 廉纖 and xiān-quán 先全 rhyme groups had a parallel change, to lower the main vowel from -ɛ to -a, is questionable as these syllables usually have a contrast of the main vowel in most Mandarin dialects including that of southwestern Mandarin, represented by the Chengdu dialect (see Table 9.13). In a parallel change, as shown in Table 9.14, the front vowel -i changed to central vowel -ɨ, which is often called vowel apicalization. The characters zhī 知 and chī 癡 are syllables from the Middle Chinese zhǐ 止 rhyme group with Middle Chinese zhī 知 group initials, and shì 世 is a syllable from the Middle Chinese jì 祭 rhyme with initial shū 書 of the zhāng 章 group. These syllables belong to the qí-wēi 齊微 rhyme group of the ZYYY. These changes (an/t = am/p = jɛn/t = jɛm/p and -ɨ = -i) in Běnwù’s notation actually indicate the loss of palatal medial or the loss of the front feature of high front vowels. This is an earlier record than the Děngyùn Tújīng 等韻圖經 of 1606, in which the loss of palatal medials and the loss of the front feature of high front vowels are systematically recorded. The main vowels are given in Table 9.15. The vowels of finals with different codas can be analyzed in subsystems according to the codas (see Table 9.16). Table 9.13 The contrast of the rù syllables in Mandarin dialects

Beijing Chengdu

Beijing Chengdu

山 shān ‘mountain’

衫 shān ‘shirt’

氈zhān ‘felt’

占 zhàn ‘take up’

ʂan san

ʂan san

tʂan tsan

tʂan tsan

殺 shā ‘kill’

霎 shà ‘moment’

舌 shé ‘tongue’

攝 shè ‘take’

ʂa sa

ʂa sa

ʂɤ se

ʂɤ se

Table 9.14 In a retroflex initial environment, -i becomes centralized, and merges with -ɨ syllables -ɨ = -i (zhī-cí 支辭 = xī-wēi 西微) 支 zhī ‘support’ = 知 zhī ‘know’ 差 cī ‘stagger’ = 癡 chī ‘idiot’ 師 shī ‘master’ = 世 shì ‘world’

The Ming Dynasty

333

Table 9.15 The main vowel system of the Yùnluè Yìtōng i•y ɛ

ɨ ə a

u o

Table 9.16 A final chart representing the phonological system of the Yùnluè Yìtōng -Ø -j -m

9.3.4

i, y, ɨ, u, ɛ, ə, o, a ə, a i, ə, a

-w -n

ə, a i, e, ə, o, a

-ʔ -ŋ

i, u, e, ə, o, a i, u, ə, a

Tones

As shown in Table 9.17, yīnpíng and yángpíng syllables are distinguished (Zhāng Yùlái 1999); syllables of the following pairs are separated with a ○. In the YLYT itself, the order of yīnpíng and yángpíng syllables is not consistent. Of all 168 pairs of contrastive syllables, 122 pairs have yīnpíng syllables listed first and 46 pairs have yángpíng ones listed first. Examples are given in Table 9.18 to show the distinction between yīnpíng and yángpíng syllables.

Table 9.17 The tonal system of the Yùnluè Yìtōng píng 平 通 . . .○同. . .

shǎng 上 桶 ...

qù 去 痛. . .

rù 入 秃 ...

Table 9.18 Yīnpíng and yángpíng syllables as in the Yùnluè Yìtōng 風○馮 卿○擎 須○徐 些○斜

方○房 琛○沉 呼○胡 丘○求

分○坟 貪○覃 猜○才

番○藩 添○甜 操○曹

潘○盤 雌○詞 坡○婆

喧○玄 飛○肥 蝦○霞

Note: All the pairs are listed with yīnpíng syllables in front of the corresponding yángpíng syllables.

334

Toward Modern Mandarin

9.4

The Sìshēng Tōngjiě 四聲通解 (1517), Fānyì Lǎo Qǐdà 翻譯老乞 大, and Fānyì Piáo Tōngshì 翻譯朴通事

The Sìshēng Tōngjiě 四聲通解 ‘A Comprehensive Explanation of the Four Tones’ (SSTJ) was compiled by the great Korean scholar Ch’oe Se-jin 崔世珍 (1473–1542) in 1517. The purpose of this dictionary was to provide phonetic information about the Chinese language to Korean learners (Zhāng Xiǎomàn 2005). Because of the difference between the literary standard represented in the Hóngwǔ Zhèngyùn 洪武正韻 (HWZY) and the real pronunciation of colloquial speech, the SSTJ included phonetic transcriptions for both the zhèngyīn 正音 and the súyīn 俗音, as taken from the Sìshēng Tōngkǎo 四聲 通考 ‘A Comprehensive Annotation of the Four Tones’ by Sin Suk-chu 申叔 舟 (1417–1475). The zhèngyīn in this book is a theoretical system based on the official HWZY, and the súyīn is the pronunciation as recorded by Sin Suk-chu. The phonetic transcriptions are written in Hangul, a fairly precise alphabetic system for the Korean language invented in the middle of the fifteenth century. This is the second time that an entire Chinese phonology was transcribed into an alphabetic system. In the “Guide to Use,” Ch’oe Se-jin clearly states that he referred often to the hP’ags-pa spellings of the Ménggǔ Yùn 蒙古韻 (the _ Ménggǔ Yùnlüè 蒙古韻略) whenever he needed phonological information for his transcription. Ch’oe Se-jin took many trips to China in order to study and accurately record the pronunciation of the northern Chinese dialect. He recorded pronunciations different from Sin Suk-chu’s as jīn súyīn 今俗音; these new pronunciations quite closely reflect the colloquial pronunciation of Beijing in the early sixteenth century. According to Yùchí (1990), it is the Liaoyang dialect, the ancestral language of the Beijing dialect. The Lǎo Qǐdà 老乞大 and Piáo Tōngshì 朴通事 are two Chinese textbooks compiled in the Yuan dynasty (Hú 1963; Wāng Wéihuī 2005). Qǐdà 乞大 is a transliteration of khitan, referring to the Khitan of the Liao dynasty, but in this context more generally referring to the Chinese people as a whole, using lǎo 老 as a prefix for personal names. Piáo Tōngshì means ‘interpreter Park’; tōngshì is a Chinese word, which means ‘interpreter/translator.’ Piáo 朴 is the Chinese character representation of the common Korean surname Park. The versions of these books that included Hangul spellings – dubbed Fānyì Lǎo Qǐdà 翻譯老 乞大 ‘Lǎo Qǐdà with phonetic notations’ and Fānyì Piáo Tōngshì 翻譯朴通事 ‘Piáo Tōngshì with phonetic notations,’ respectively – were also authored by Ch’oe Se-jin. Because these two textbooks were designed for teaching colloquial and conversational Chinese, the phonetic transcriptions of Chinese characters are not limited by the framework of the Sìshēng Tōngjiě, which follows the ideal pronunciation of monosyllables; in addition, the books include descriptions of longer phrase structure and tone sandhi. In this way,

The Ming Dynasty

335

Table 9.19 The initial system of the Sìshēng Tōngjiě p t ts tʂ k

ph th tsh tʂh kh

m n

f l s ʂ x

ɻ Ø

the Fānyì Lǎo Qǐdà and the Fānyì Piáo Tōngshì reveal a number of important phonological features never before seen in Chinese historical linguistics. 9.4.1

Initials

In the SSTJ, the initial system of jīn súyīn is the same as that of the Sìshēng Tōngkǎo, which has all the voiced obstruents; Ch’oe Se-jin, however, changed the voiced obstruents to voiceless in the Fānyì Lǎo Qǐdà and Fānyì Piáo Tōngshì. According to previous studies (Hú 1963; Sūn Jiànyuán 2010) there are nineteen initials, shown in Table 9.19. The palatals tʃ-, tʃh-, ʃ-, ʒ- are changed to retroflexes tʂ-, tʂh-, ʂ-, ɻ- (see Section 7.3.1.2). 9.4.1.1 The Description of Retroflex Initials The description of retroflex initials by Korean scholars should be noted here, as there is no contrast between alveolar initials (ts- tsh-, s-) and retroflex initials (tʂ-, tʂh-, ʂ-) in Korean. In order to distinguish this, in the Sìshēng Tōngkǎo Sin Suk-chu provided a fairly accurate description of the place of articulation for these two series of initials in Chinese. He said: The chǐyīn 齒音, chǐtóu 齒頭 (alveolar) is to raise the tongue to touch the teeth, so it sounds shallow; zhèngchǐ 正齒(retroflex) is to curl the tongue to touch the palate, so it sounds deep.3

This is the first time an accurate description is provided to describe the traditional categories such as chǐtóu and zhèngchǐ. This description unambiguously indicates that the zhèngchǐ initials are retroflex. 9.4.1.2 Exceptional Changes This initial system is likely based on the Beijing dialect because a number of exceptions to regular sound changes are very similar to that of the modern Beijing dialect (Sūn Jiànyuán 2010: 44–46). Table 9.20 shows some of the exceptional words. These words are written with the irregularities that are also

3

“ 凡齒音,齒頭則舉舌點齒,故其聲淺;正齒則卷舌點腭,故其聲深。”

336

Toward Modern Mandarin

exhibited in the modern Beijing dialect, suggesting that this standard is also based on the Beijing dialect. 9.4.2

Finals

The phonetic values for SSTJ finals provided in Table 9.21 are based on the Hangul spellings with three modifications, ə > o; iə, yə > jɛ, ɥɛ; and ɯ > ɨ (ə in front of a coda). These adjustments make them more consistent with the hP’ags-pa spellings of the Méngyùn 蒙韻 (full title: Ménggǔ Zì Yùnlüè 蒙古字 _ 韻略) as it was one of Ch’oe Se-jin’s phonological references in the creation of the SSTJ. Table 9.20 Some of the exceptional words 佩 鳥 艇 賃 雀 囚 珊 匙 螢 完

pèi niǎo tǐng lìn què qiú shān shi yíng wán

‘ware’ ‘bird’ ‘boat’ ‘rant’ ‘small bird’ ‘prisoner’ ‘coral’ in shānhú ‘spoon’ ‘firefly’ ‘finish’

phnthltshtshʂtʂhØØ-

should be pshould be tshould be tshould be nshould be tsshould be sshould be sshould be ʂshould be xshould be x-

Note: “should be” refers to what the characters’ initial realization would be according to the regular sound changes between Middle Chinese and the SSTJ.

Table 9.21 The phonetic transcriptions of finals in the Sìshēng Tōngjiě ɨ a o ɛ aj ej aw əw an on ən aŋ əŋ

i ja jo jɛ jaj jaw iw jɛn in jaŋ iŋ

u wa

y

wɛ waj uj

ɥɛ

wan won un waŋ uŋ

ɥɛn

Source: Based on Hú 1963; Sūn Jiànyuán 2010.

ɥ

yn yŋ

The Ming Dynasty

337

Table 9.22 A final chart of the Chinese phonology based on the Hangul spellings Medials Main vowels

j, w, ɥ i, y e

Codas

ɨ ə a j, w, m, n, ŋ, p, t, k

-Ø: i, ɨ, u, y, o, a, e -n: i, ɛ, ə, o, a

-j: a, ɛ, u -ŋ: i, u, ə, a

Tones

yīnpíng, yángpíng, shǎng, qù

u o

-w:

ə, a

Table 9.22 is a final chart representing the Chinese phonology of the Hangul spelling system. It is worth noting that the main vowels are vastly different depending on the following coda. In the following subsections, the phonetic values before and after changes shown in the tables are based on the MGZY and the SSTJ respectively. 9.4.2.1 The Loss of -p, -t, -k The stop codas were completely lost by the time of the SSTJ, so the question becomes whether there was a glottal tension at the end of the Middle Chinese rù syllables. According to Ch’oe Se-jin’s description, the glottal stop was present, though this could be due to the fact that it was artificially produced, as the rù tonal category was artificially kept in the ideal standard pronunciation until the beginning of the twentieth century. 9.4.2.2 The Loss of -m, -m > -n In the “Guide to Use” of the SSTJ, Ch’oe Se-jin made it clear that Middle Chinese -m syllables were all pronounced as -n, although it seems that the change from -m to -n was still in progress near the beginning of the sixteenth century. According to Ch’oe, Middle Chinese -m syllables were pronounced as -n in colloquial pronunciation; however, they were still pronounced as -m by some people. Ch’oe’s observation (of the sixteenth century) was the earliest evidence of the complete loss of the Middle Chinese -m in Northern Mandarin. In the “Guide to Use,” Ch’oe also said “the codas, -n, -ŋ, and -m, of the rhymes were not merged originally. In the colloquial pronunciation Chinese people use -n to pronounce the -m coda of rhymes qīn 侵 (-m), tán 覃 (-m), and yán 鹽 (-m). Thus, the pronunciation of zhēn 真 (-n) and qīn 侵 (-m), shān 刪 (-n) and tán 覃 (-m), and xiān 先 (-n) and yán 鹽 (-m) are often merged.”4 4

“諸韻終聲 n、ŋ、m 之呼初不相混, 而直以侵、覃、鹽合口終聲漢俗皆呼為 n, 故真與 侵、刪與覃、先與鹽之音多相混矣。”

338

Toward Modern Mandarin

9.4.2.3 Middle Chinese tōng 通 Rhyme Group Syllables Middle Chinese Rank-III syllables of the tōng 通 rhyme group show an irregular pattern of change from -juŋ to -uŋ. According to súyīn many syllables lost the palatal medial, although it reappeared in some syllables in the jīn súyīn, as shown in Table 9.23. This seeming reversal suggests that Ch’oe recorded a different dialect or a subdialect than Sin Suk-chu did when writing the Hóngwǔ Zhèngyùn Yìxùn. 9.4.2.4 The -w- of Middle Chinese kāikǒu 開口 Syllables Middle Chinese kāikǒu syllables with retroflex initials from the dàng 宕 rhyme group developed the labial medial -w- and became the same as their counterparts from the jiāng 江 rhyme group (see Table 9.24). 9.4.2.5 The Change of -jaj and -waw The syllables whose finals have the same medial and coda, -jaj and -waw, underwent a change after guttural initials. This change seems to have been conditioned by different manners of articulation (see Table 9.25 and Table 9.26). 9.4.2.6 The Change of -jan/m > -jɛn/m As shown in Table 9.27, the finals of Rank-II syllables with guttural initials -jan/m changed to -jɛn/m, merging with the syllables of Rank-III/IV. The nasal coda -m changed to -n and the main vowel changed from -a to -ɛ. Table 9.23 Rank-III syllables of the tōng 通 rhyme group in the súyīn and the jīn súyīn

Súyīn Jīn súyīn

弓 gōng uŋ yŋ

穹 qióng uŋ yŋ

窮 qióng yŋ yŋ

顒 yóng juŋ yŋ

縱 zòng uŋ yŋ

從 cóng uŋ yŋ

嵩 sōng uŋ yŋ

頌 sòng uŋ yŋ

中 zhōng uŋ yŋ

Súyīn Jīn súyīn

充 chōng uŋ yŋ

蟲 chóng uŋ yŋ

舂 chōng uŋ yŋ

邕 yōng yŋ yŋ

胷 xiōng yŋ yŋ

雄 xióng yŋ yŋ

庸 yōng yŋ yŋ

龍 lóng uŋ yŋ

戎 róng uŋ yŋ

Table 9.24 The merging of the dàng 宕 and jiāng 江 groups Middle Chinese 窗 莊

chuāng ‘window’ zhuāng ‘village’

tʂhɰɔŋ > tʂjɐŋ >

MGZY >

-waŋ -ɒŋ

SSTJ > >

-waŋ -waŋ

jiāng 江 rhyme group dàng 宕 rhyme group

The Ming Dynasty

339

Table 9.25 The change of syllables with final -jaj Middle Chinese 街 鞋 揩 崖 矮

jiē ‘street’ xié ‘shoe’ kāi ‘wipe’ yá ‘cliff’ ǎi ‘short’

MGZY > > > > >

kɰæ/kɰæj ɦɰæ/ɦɰæj khɰæj ŋɰæ Øɰæ

SSTJ > > > > >

-jaj -jaj -jaj -jaj -jaj

-jɛj -jaj -aj -aj -aj

Table 9.26 The change of syllables with final -waw Middle Chinese

MGZY

SSTJ



zhuō ‘catch’

tʂhɰɔk

>

-waw

>

-o



guō ‘outer wall’

kwɑk

>

-waw

>

-o

jiāng 江 rhyme group dàng 宕 rhyme group

Table 9.27 The merger of Rank-II and Rank-III/IV syllables with coda -n or -m MGZY Rank-II Rank-II Rank-III/IV Rank-III/IV

奸 監 犍 檢

jiān ‘evil’ jiān ‘monitor jiān ‘bullock’ jiǎn ‘check’

-jan -jam -jɛn -jɛm

SSTJ > > > >

-jɛn -jɛn -jɛn -jɛn

Table 9.28 The merger of Rank-II and Rank-III/IV syllables with coda -w MGZY Rank-II Rank-III/IV

交 驕

jiāo ‘cross’ jiǎo ‘arrogant’

-jaw -jɛw

SSTJ > >

-jaw -jaw

9.4.2.7 The Change of -jɛw > -jaw The Middle Chinese Rank-III/IV syllables with final -jɛw changed to -jaw, merging with syllables of Middle Chinese Rank-II (see Table 9.28). The main vowel changed from -ɛ to -a.

340

Toward Modern Mandarin

Table 9.29 The merger of Rank-II and Rank-III/IV finals conditioned by codas Rank-II Rank-III/IV Rank-II Rank-III/IV

> > > >

-jan/m -jɛn/m -jaw -jɛw

-jɛn -jɛn -jaw -jaw

Table 9.30 The change of words with final -iw MGZY 繆 矛 眸 鍪

miù ‘fallacy’ máo ‘spear’ móu ‘pupil’ móu ‘iron pot’

-iw -iw -iw -iw

SSTJ > > > >

-iw -əw -əw -əw

This is in opposition to the change of -jan/m > -jɛn/m. As shown in Table 9.29, conditioned by the codas, Rank-II finals -jan/m merged into Rank-III/IV, but Rank-III/IV finals -jɛw merged into Rank-II. 9.4.2.8 The Change of -iw > -əw The Middle Chinese Rank-III syllables of the liú 流 rhyme group lost their palatal medial after a bilabial nasal initial. In the MGZY, the words in Table 9.30 are the homophones with final -iw. The character miù 繆 is the exception to the change and is still pronounced with a palatal medial in modern Beijing Mandarin. 9.4.3

Tone Sandhi

The main purpose of the SSTJ was to be a dictionary compiled to accompany the previously mentioned Chinese textbooks Lǎo Qǐdà and Piáo Tōngshì. Because the Lǎo Qǐdà and Piáo Tōngshì transcribe colloquial sentences for the purpose of fluency in conversational Chinese, some tonal phenomena beyond single syllables are recorded. In the “Guide to Use” of the Fānyì Lǎo Qǐdà and Fānyì Piáo Tōngshì, two rules of tone sandhi are stated: 1. If two shǎng tone syllables are in a row, the first one is pronounced as yángpíng (濁平 zhuópíng) and the second pronounced as the original shǎ ng. 2. If two shǎng tone syllables are in a row and the second syllable is a function word (虛詞 xū cí) or both are function words, the second syllable is pronounced as qù.

The Ming Dynasty

341

Rule (1) states that shǎng + shǎng > yángpíng + shǎng. Rule (2), because of its reference to function words, is probably using the qù tone as an impression of the pitch of the neutral tone. Neutral tone (輕聲 qīngshēng, also known as ‘zeroth tone’ or ‘fifth tone’), rather than a tone itself, describes the de-emphasis of certain syllables in Chinese: usually grammatical particles. Rather than having any specific tone, neutral tone characters will conform in tone to the end of the tone contour of the previous character. Neutral tone is high after shǎng, as a shǎng tonal contour ends high (Contour: 214, ending with midhigh 4). Because of this, it makes sense that these function words’ tone would be described as qù when following shǎng in the Lǎo Qǐdà and Piáo Tōngshì. Table 9.31 and Table 9.32 are some examples from Tsu-lin Mei (1977). Table 9.31 Tone sandhi as explained in the Sìshēng Tōngjiě

3+3>2+3 好酒 野狗 洗臉 雨水

Pinyin

Perceived pronunciation

Definition

hǎo jiǔ yě gǒu xǐ liǎn yǔ shuǐ

háo jiǔ yé gǒu xí liǎn yú shuǐ

‘good wine’ ‘wild dog’ ‘wash face’ ‘rain water’

3 + 3 > 3 + 4 (second tone is neutral tone) 果子 guǒ zǐ guǒ zì (guǒ zi) 走了 zǒu liǎo zǒu liào (zǒu liao) 水裡 shuǐ lǐ shuǐ lì (shuǐ li) 耳朵 ěr duǒ ěr duò (ěr duo)

‘fruit ‘have gone’ ‘in water’ ‘ear’

Source: Tsu-lin Mei (1977). Note: 2 = yángpíng, 3 = shǎng, and 4 = qù, as in Modern Mandarin.

Table 9.32 Tone sandhi for longer phrases Hanzi Pinyin Perceived pronunciation Original tone number Tone after sandhi Definition

他 少 tā shǎo tā sháo 1 3 1 2 ‘He owes me five

Hanzi Pinyin Perceived pronunciation Original tone number Tone after sandhi Definition

有 酒 有 花 yǒu jiǔ yǒu huā yóu jiǔ yǒu huā 3 3 3 1 2 3 3 1 ‘There is wine, there are also flowers’

Source: Tsu-lin Mei (1977).

我 wǒ wǒ 3 3 taels of

五 wǔ wú 32silver’

兩 liǎng liǎng 3 3

銀 yín yín 22-

子 zǐ zǐ 3 3

裡 lǐ lǐ 3 3

342

Toward Modern Mandarin

The first third tone sandhi rule remains the same in Modern Mandarin; as such, this is the first time suprasyllabic prosodic features of Mandarin were recorded. The mentioning of the function words also clearly indicates the existence of the neutral tone, or toneless syllables, in real pronunciation. Figure 9.2 is a page from the Lǎo Qǐdà Yànjiě 老乞大諺解. The dialogue, as presented on this page, is transcribed in Chinese with pinyin and gloss in Table 9.33. Table 9.33 The dialogue as presented in Figure 9.2, with pinyin and gloss 大哥 你 dàgē nǐ brother you Brother, where do you come from?

從 cóng from

那裡 nǎlǐ where

來 lái come

我 從 wǒ cóng I from ‘I come from the capital of Korea’

高麗 gāolí Korea

王京 wángjīng capital

來 lái come

9.5

The Xīrú Ěrmù Zī 西儒耳目資 (1626)

The Xīrú Ěrmù Zī ‘An Aid to the Eyes and Ears of Western Literati’ (XEZ) by Nicolas Trigault (1577–1628) (Chinese: Jīn Nígé 金尼閣), a Belgian/French missionary, was published in 1626 (see Figure 9.3). Historically speaking, this work is the third effort – after the Ménggǔ Zìyùn 蒙古字韻 (MGZY) and the Sìshēng Tōngjiě 四聲通解 (SSTJ) – to transcribe an entire Chinese phonology into alphabetic letters, as well as one of the earliest attempts to transcribe the whole Chinese phonology into the Latin alphabet. But the significance of the XEZ is that unlike the two previous transcriptions, it was not based on a preexisting rhyme work. In other words, the XEZ provided information about phonetic values and information about phonological categories independently for the first time. There are two main opinions about its base dialect: one represented by Lù Zhìwěi (1947c), who identifies it as Shanxi dialect; the other by Lǔ Guóyáo (2007), who identifies it as Nanjing dialect. Both use the existence of the rù tone as their evidence. However, a piece of convincing evidence is that the reflexes of the Middle Chinese zhī 知 and zhào 照 initial groups (Sūn Yízhì 2010) are basically the Nanjing type. Although there are some non-Nanjing features, it is likely that the main body phonology was based on the Nanjing dialect. 9.5.1

Initials

There are twenty-one initials (called Tóngmíng Zìfù 同鳴字父 ‘The Father of the Characters with the Same Sound’ by Trigault) in the XEZ. According to his

The Ming Dynasty

343

Figure 9.2 A page from the Lǎo Qǐdà Yànjiě 老乞大諺解. The term yànjiě 諺 解 indicates that the phonetic and semantic notes written in the Korean script Hangul are provided along with the characters. Two phonetic notations are provided underneath each Chinese character. The one on the left is zhèngyīn and the one on the right is súyīn. Korean students studying from this textbook would use the súyīn readings in order to learn the (then contemporary, and thus practical for trade or travel) colloquial pronunciation of Northern Mandarin in the fifteenth century.

344

Figure 9.3 Two pages from the Xīrú Ěrmù Zī 西儒耳目資. Note that while it contains the spellings of characters, it is organized as a rhyme table, so finding the spelling of a given character requires one to read the row (initial) and column (final) together.

The Ming Dynasty

345

tables, all syllables used to represent initials seem to have a vowel -ə or -ɨ, and carry rù tone, information that offers a clue to the phonetic values of some finals. Table 9.34 shows the initial letters and characters of the Xīrú Ěrmù Zī (refer to the first two columns on the right side of Figure 9.3) and their phonetic transcriptions. The null initial Ø- was not indicated with a letter initial, and characters with null initials were placed in the same box as the information that transcribed the final (see top of Figure 9.3). In Table 9.35, the initials of the Xīrú Ěrmù Zī are arranged according to their place and manner of articulation. Table 9.34 Initials of the Xīrú Ěrmù Zī Transliteration Character Phonetic value

ç 則 ts

‘ç 側 tsh

ch 者 tʂ

‘ch 扯 tʂh

k 格 k

‘k 克 kh

p 百 p

‘p 魄 ph

t 德 t

‘t 忒 th

Transliteration Character Phonetic value

j 日 ɻ

v 物 ʋ

f 弗 f

g 額 ŋ

l 勒 l

m 麥 m

n 搦 n

s 色 s

x 石 ʂ

h 黑 x

Table 9.35 Initial chart of the Xīrú Ěrmù Zī pttstʂk-

百 德 則 者 格

9.5.2

phthtshtʂhkh-

魄 忒 側 扯 克

m- 麥 n- 搦 ŋ- 額

f- 弗 s- 色 ʂ- 石 x- 黑

ʋ- 物 l- 勒 ɻ- 日 Ø- (自鳴)

Finals

There are fifty-seven finals arranged in sixteen groups in the XEZ. Trigault did not apply the modern phonemic principle thoroughly to his transcriptions. Some different finals (e.g. ‹ ua › vs. ‹ oa ›, ‹ eam › vs. ‹ iam ›) are not phonemic but allophonic variations. There are such contrasts in the medial position in Mandarin phonology. Since the Latin letters used in the transcription were not suitable for Chinese, the intrinsic values of the letters cannot serve as hints for the Chinese sounds. For example, he used the letter ‹ g › for the velar nasal initial, but the letter ‹ m › for the velar nasal coda. He used ‹ e › for the schwa in the final əw as well as for ɛ in the final -jɛn. His ‹ ul › represents a dialectal variation of the rhotacized schwa -ər or possibly -əl as in many modern dialects. As shown in Table 9.36, many finals are in complementary distribution conditioned by different initials. The letters ‹ u › and ‹ o › are used to transcribe the same final after different initials, and ‹ i › and ‹ e › are also used to transcribe the same

346

Toward Modern Mandarin

Table 9.36 Some examples of finals transcribed by Trigault XEZ wa: wan: y:

XEZ

ua, oa uan, oan iu

waj: un: ɥɛn:

XEZ

uai, oai uen, oen, un iuen

wej: jaŋ: yn:

uei, oei iam, eam iun

Table 9.37 The spellings of each of the fifty final tables as written in the Xīrú Ěrmù Zī 1 6 11 16 21 26 31 36 41 46

a ai em iu ua um iao ium oan uan

2 7 12 17 22 27 32 37 42 47

e, ė ao en im ue un iam iun oen uem

3 8 13 18 23 28 33 38 43 48

i am ia in ui eao ieu oai uai uen

4 9 14 19 24 29 34 39 44 49

o, o˙ an ie, iė oa uo, uo˙ eam ien oei uei uon

5 10 15 20 25 30 35 40 45 50

u, u̇, u _ eu io, io˙ oe ul iai iue oam uam iuen

final. The final -y and medial -ɥ- are transcribed using the sequence ‹ iu ›. Table 9.37 shows the original spellings in the fifty final tables. Some tables contain more than one final; the different finals within the same table are marked with diacritics ‹ ˑ › either on top or beneath the letter. A couple of points must be made clear in the interpretation of these spellings. (1) In the same final table, the letter with a diacritic is a different final. The ‹ e › and ‹ ė › of table 2 indicate the vowels -ɛ and -ɨ/-ə respectively. (2) On the other hand, different spellings from different final tables could be the same final if they are in complementary distribution. For example, the spelling ‹ oa › in table 19 and the spelling ‹ ua › in table 21 are the same final wa. The spelling ‹ oa › combines initials ʂ-, x-, but the spelling ‹ ua › combines with initials tsh-, k-, kh-. The spellings ‹ u › and ‹ o › as well as ‹ i › and ‹ e › are often used in the medial position for allophonic spellings. These spelling variations indicate the uncertain phonetic nature of the medials. The spellings ‹ oam › and ‹ uam › therefore should be considered allophonic of the same final -waŋ as well, even though the syllables with retroflex initials in tables 41 and 45 form minimal pairs. (3) Some are likely misplaced: table 48 includes characters with values -wan and -un. In table 5, the rù syllables of ‹ u › should _

The Ming Dynasty

347

Table 9.38 The phonetic interpretations of the spellings in the fifty final tables of the Xīrú Ěrmù Zī 1 6 11 16 21 26 31 36 41 46

a aj əŋ y wa uŋ jaw yŋ wan wan

2 7 12 17 22 27 32 37 42 47

ɛ, ɨ1 aw ən iŋ wə un jaŋ yn un uŋ

3 8 13 18 23 28 33 38 43 48

i aŋ ja in wej jaw jəw waj waj won, un

4 9 14 19 24 29 34 39 44 49

o, u an jɛ, jə wa wo, wə jaŋ jɛn wej wej won

5 10 15 20 25 30 35 40 45 50

u/y, ɨ2, ʉ əw jo, ɥo wə əl jaj ɥɛ waŋ waŋ ɥɛn

be grouped with the rù syllables of 16. The syllables of tables 20 and 22 (these two tables have the rù syllables only) are in complementary distribution and should be combined. The two rù syllables with final ‹ uȯ › of table 24 also should be combined with the syllables of tables 20 and 22. Table 9.38 is the phonetic interpretation of the spellings in the fifty final tables of the Xīrú Ěrmù Zī. In Table 9.39, these finals are rearranged according to their coda types (-Ø without coda; -G with -j, -w; -N with -n, -ŋ) and medials (-Ø-, -j-, -w-, -ɥ-). The finals with codas do not appear in rù syllables. The finals of the rù tone syllables are marked with a glottal stop, which also indicates the possible short length of the rù tone. This analysis assumes that the main vowel -y and its counterpart medial -ɥexisted by the Ming dynasty. Their existence would be considered evidence that the Chinese phonological system changed from the Middle Chinese fourrank system with kāihé 開合 to the sìhū 四呼 system of Modern Mandarin. There is no evidence to show a medial cluster such as -jw- in Mandarin phonology from the Yuan dynasty on, including Modern Mandarin dialects. Foreign transcriptions use two letters (such as Trigault’s ‹ iu ›) to transcribe a single medial (main vowel -y and medial -ɥ-) in Chinese, a practice that could cause scholars to reason back that the original Chinese had a cluster medial. The proposals for the existence of such a medial cluster, then, are confused by the usage of letters in the foreign transcriptions or else guided by unreliable poetic rhyming. Trigault transcribes -ɨ as “-u 次”5 after alveolar affricates and fricatives, and ‹ e › after retroflex initials. Clearly he did not have the suitable letters for these vowels. However, both transcriptions do describe some of the phonetic 5

In Trigault’s vowel system the descriptive word cì 次 ‘secondary’ is used after a vowel. This is a supplementary method to transcribe the vowels that cannot be transcribed by regular vowel letters.

348

Toward Modern Mandarin

Table 9.39 The phonetic interpretations (fifty-four finals) of the spellings in the fifty final tables of the Xīrú Ěrmù Zī, arranged by medial and rime Shū syllables, 39 finals -Ø (15)

-G (8)

-N (16)

-Øa 1 ɛ 2 o 4 i 3

-jja 13 jɛ 14

-wwa 19, 21

u 4

wo 24 y 16

əl 25

ɨ1, ɨ2 2, 5

ʉ 5

aj 6

jaj 30

waj 38, 43 wej 23, 39, 44

aw 7 əw 10

jaw 28,31 jəw 33

an 9

jɛn 34

ən 12

in 18

aŋ 8 əŋ 11

jaŋ 29, 32 iŋ 17

Rù syllables, 15 finals -ɥ(15) ɥɛ 35

wan 41, 46 won 48, 49 un 27, 42 48 waŋ 40, 45 uŋ 26, 47

-Øaʔ 1 ɛʔ 2 oʔ 4

-j-

jɛʔ 14 joʔ 15 uʔ 4,5 jəʔ 14 ɨʔ 2

-wwaʔ 19,21

woʔ 24 yʔ 5,16 wəʔ 20, 22, 24 ʉʔ 5

-ɥ-

yɛʔ 35 yoʔ 15

ɥɛn 50

yn 37

yŋ 36

features of these two central vowels (see Table 9.40). The unrounded high central vowel after the alveolar affricates and fricatives is not front and not back, so it is “-u 次.” The unrounded high central vowel after retroflex affricates and fricatives is realized as a lower vowel than its counterpart after alveolar consonants, making the use of different symbols justifiable

The Ming Dynasty

349

Table 9.40 Phonetic features of central vowels and other relevant vowels

High Front Back Rounded

i

ɨ

ʉ

u

e

+ + -

+ -

+ +

+ + +

+ -

Table 9.41 Medials, main vowels, and codas of Trigault’s Xīrú Ěrmù Zī Medials Codas

j, w, ɥ -n, -ŋ, -j, -w, -ʔ

-Ø: a, ɛ, o, u, i, y, ɨ, ɚ -n: a, ɛ, o, ə

-j: a, e -ŋ: a, ɛ, o, ə

-w: a, ə -ʔ: a, ɛ, o, u, y, ɨ, ə

Table 9.42 The tones of the Xīrú Ěrmù Zī

pph m-

清 qīng

濁 zhuó

上 shǎng

去 qù

入 rù

ā 巴 pā 葩 phā –

â – 琶 phâ 麻 mâ

à 把 pà – 馬 mà

á 霸 pá 帊 phá 禡 má

ǎ 八 pǎ 汃 phǎ 帓 mǎ

(Wú 1986). Table 9.41 lists the medials, main vowels, and codas of Trigault’s Xīrú Ěrmù Zī. 9.5.3

Tones

Five tones are distinguished in the XEZ. The first four are more or less equivalent to Modern Mandarin, namely yīnpíng (called qīng 清 in the XEZ), yángpíng (called zhuó 濁 in the XEZ),6 shǎngshēng, and qùshēng. The fifth is rù tone, shown in Table 9.42. The tone marks are diacritics on the vowel letters ā, â, à, á, and ǎ, respectively, which seemingly had influence on the design of the pinyin system of Modern Chinese. In the XEZ tonal values 6

Most likely referring to the qīngzhuó system from which the split of the píng tone is determined (see Section 1.4.2).

350

Toward Modern Mandarin

are described in terms of height. From the highest to the lowest, they are qù, rù, qīng píng (yīnpíng), shǎng, and zhuó píng (yángpíng). These simple descriptions are not enough to interpret them in tonal values. 9.6

The Děngyùn Tújīng 等韻圖經 (1606)

The Děngyùn Tújīng (DYTJ) (full title: Chóngdìng Sīmǎ Wēn Gōng Děngyùn Tújīng 重訂司馬溫公等韻圖經) is an attachment to the Hébìng Zìxué Piānyùn Biànlǎn 合併字學篇韻便覽 edited by Zhāng Yuánshàn 張元善 and published in 1606. The DYTJ, authored by Xú Xiào 徐孝 (1573–1619), is in the format of rhyme tables. The rhyme tables show a number of innovative features that are specific to the modern Beijing dialect and for this reason the dialect recorded in the DYTJ is regarded as the first direct ancestral language of the modern Beijing dialect (Lù 1947b; Wáng Lì 1985). The DYTJ divides the final system into thirteen rhyme groups. Except for the zhù 祝 rhyme group, each of the rhyme groups is divided into kāi 開 and hé 合 tables. Each table is further divided into three sections. The first section is syllables with neither palatal medial nor retroflex initials, the second section lists syllables with retroflex initials, and the third section lists syllables with a palatal medial. 9.6.1

Initials

As shown in Table 9.43, there are twenty-two initial characters listed in the tables of the DYTJ. The traditional initial characters are used to represent the initials (see Section 1.4.1). Only nineteen of the twenty-two initials of the DYTJ can actually combine with finals to represent characters. Characters fū 敷, wēi 微 and (duplicated initial character xīn 心) initials do not combine with any finals (see Table 9.44). It is worth noting some significant changes in the initial system. One is the deletion of the Middle Chinese wēi 微 initial, ɱ- > ʋ- > Ø-, which is one of the initials in the Zhōngyuán Yīnyùn 中原音韻 (ZYYY), Yùnlüè Yìtōng 韻略 易通, and Xīrú Ěrmù Zī 西儒耳目資 (see Section 7.3.1.1, Section 7.4.2, Table 9.43 The twenty-two initials given in the Děngyùn Tújīng 見 幫

溪 滂

端 明

透 精

泥 清

非 影

敷 曉

微 來



穿

a

There is a duplicated initial character xīn 心 s-.

a



心 審

The Ming Dynasty

351

Table 9.44 The nineteen actual initials of the Děngyùn Tújīng 幫 端 見 精 照

p t k ts tʂ

滂 透 溪 清 穿

明m 泥n

ph th kh tsh tʂh

非f

來l 影Ø

曉x 心s 審ʂ

稔ɻ

Table 9.45 The merger of k- initial series and ts- initial series k-, kh-, hts-, tsh-, s-

> >

tɕ-, tɕh-, ɕ- / i, y, j, ɥ tɕ-, tɕh-, ɕ- / i, y, j, ɥ

Table 9.46 Three possible contrasts between the two initial groups Possibility I kjtsj-

Possibility II kj-

> tsj-

Possibility III tɕjtsj-

kj>

tɕj-

Section 9.3, and Section 9.5). The other is the deletion of the Middle Chinese yí 疑 initial, ŋ - > Ø-. 9.6.1.1 Palatalized Initials As shown in Table 9.45, in Modern Mandarin, two groups of initials, namely k-, kh-, h- and ts-, tsh-, s-, merge, becoming palatals before high front vowels and medials (-i, -y, -j-, -ɥ-). The question about this merger is at which historical stage it started taking place. A related question is whether or not these two groups of initials were palatalized simultaneously or sequentially. If sequentially, an additional question is which of the two groups became palatal first (see Section 10.1 for more information). But these questions cannot be easily answered based on the DYTJ, which only indicates two groups of initials. However, there are three different possibilities for the identity of these two separate groups within the DYTJ. Possibility I as seen in Table 9.46 is the easiest solution because neither initial has palatalized as indicated by the initial characters, jiàn xī xiǎo 見溪曉 for velar initials k-, kh- and x-, and jīng qīng xīn 精清心 for alveolar initials ts-, tsh- and s-. Possibility II can also be true if k- and tɕ- are conditioned variants of the same initial; that is, if they were in complementary distribution, there would be no need to use different initial characters, if the two different

352

Toward Modern Mandarin

sounds are recognized as the same phoneme by speakers of the language. The DYTJ only provides us with categorical information, not phonetic details. Possibility III is not likely because according to Modern Mandarin dialects, if there is one type of palatalization it is always that of velar consonants. Thus historically speaking, the palatalization of velar consonants should precede the palatalization of alveolar consonants. Such an initial system can indicate that alveolar initials are not palatalized but cannot prove that velar initials are not palatalized. Although Possibility I seems easiest to interpret, if we examine the other aspects of the system, then Possibility II becomes most likely to be the real situation. In the DYTJ, the palatal medial after retroflex initials has been eliminated and, as a result, Middle Chinese Division-III syllables with zhī 知 and zhào 照 initials merged with their Division-II counterparts (tʂj- > tʂ-). This change was triggered by the palatalization of velar initials (kj- > cj), moving a grave consonant to the articulatory space of acute consonants. As a result, the original acute consonants have to adjust their place of articulation (tɕj- > tʂ(j)-). In other words, the changes of acute initials are caused by the changes of grave initials. Table 9.47 shows the six stages of palatal initials from Middle Chinese to Modern Mandarin. Without stage IV, velars (kj-) and palatals (tɕj-) would merge. The DYTJ is already in Stage V, so the palatalization of velar initials must have happened. The velar initials in the system should have two phonetic realizations: palatals in front of palatal medials and velars elsewhere. But since they are allophonic they are treated as variations of one phoneme. According to the traditional thirty-six initial characters, both allophones belong to the jian 見 k- initial. There is another pattern of change in the Zhōngyuán Guanhua 中原官話 ‘Zhongyuan Mandarin,’ a subdialect of Mandarin spoken in Henan province including Luoyang (Xióng 1990). As shown in Table 9.48, Middle Chinese kjand tsj- types of syllables are still distinct (tɕj- vs. tsj-), due to different changes and different change orders. Table 9.47 The change order of the palatalization from Middle Chinese to Modern Mandarin

Stage I Stage I Stage II Stage III Stage IV Stage V Stage VI

tʂjkjtɕjcjtsj-

> > > > >

tʂcjtʂ(j)tɕjtɕj-

照三/知三 tɕjtɕjtɕjtɕjtʂ(j)tʂtʂ-

照二/知二 tʂjtʂjtʂtʂtʂtʂtʂ-

精 tsjtsjtsjtsjtsjtsjtɕj-

見 kjkjkjcjcjtɕjtɕj-

The Ming Dynasty

353

Table 9.48 The change order of palatalization from Middle Chinese to modern Zhongyuan Mandarin

Stage I Stage II Stage III Stage IV Stage V’ Stage VI

tʂjtɕjkjtʂtʂj-

> > > > >

照三/知三 tɕjtɕjtʂjtʂjtʂjtʂ-

tʂtʂjtɕjtstʂ-

照二/知二 tʂjtʂtʂtʂtsts-

精 tsjtsjtsjtsjtsjtsj-

見 kjkjkjtɕjtɕjtɕj-

9.6.1.2 Initial Conditioned Change: The Vowel Centralization and the Loss of the Medial -j- after Retroflex Initials These changes are related to syllables with retroflex initials; non-retroflex initials do not participate in these changes. This condition is demonstrated in Table 9.49. All the syllables listed belong to the Middle Chinese 梗 gěng rhyme group and are spelled as -iŋ (iM) in the hP’ags-pa script. Their Modern _ Mandarin pronunciation shows that only the syllables with initial series ʈ- (j) have changed to schwa. The centralization of a front high vowel only occurs in the zhēng-zhěng 征整 (retroflex) initial context. As shown in Table 9.50, this widespread change finalizes the phonological pattern of Modern Mandarin. The relevant nuclear vowels and their changes are i > ə, i > ɨ, u > y, ɛ/e > ɤ, and ɛ/e > a. All the changes indicate a loss of Table 9.49 The centralization of main vowel i after retroflex initials

MGZY MM

京警

丁頂

征整

兵丙

精井

嬰影

velar iŋ iŋ

dental iŋ iŋ

retroflex iŋ əŋ

bilabial iŋ iŋ

alveolar affricate iŋ iŋ

no initial iŋ iŋ

Source: Shen 2008b.

Table 9.50 The change of front vowels after retroflex initials Rule MGZY MM a

i > ɨ, ə 知 tʂi tʂɨa

珍 tʂin tʂən

y>u 朱 tʂy tʂu

諄 tʂyn tʂun

ɛ/e > ɤ, a 遮 展 tʂɛ tʂɛn tʂɤ tʂan

In the hP’ags-pa spellings the contrast between high central (apical) vowels and high _ front vowels is in the presence or absence of the letter ħ, which indicates the change of the front feature of the following vowel i. i /i/ [+high, +front, -round], ħ+i /ɨ/ [+high, -front, -round].

354

Toward Modern Mandarin

Table 9.51 The loss of palatal medial -j- after retroflex initials MGZY 珍 周 貞 煮 知 纏 超

DYTJ tʂin tʂiw tʂiŋ tʂy tʂi dʐɛn tʂɛw

6¼ 6 ¼ 6¼ 6 ¼ 6 ¼ 6 ¼ 6¼

臻 鄒 爭 阻 輜 潺 抄

tʂən tʂəw tʂəŋ tʂu tʂɨ dʐan tʂaw

> > > > > > >

珍 周 貞 煮 知 纏 超

tʂən tʂəw tʂəŋ tʂu tʂɨ dʐan tʂaw

= = = = = = =

臻 鄒 爭 阻 輜 潺 抄

tʂən tʂəw tʂəŋ tʂu tʂɨ dʐan tʂaw

front feature. The so-called apicalization, i > ɨ, is a part of this general change. It is worth noting that there is no place for apical vowels in any phonological system as they are actually central vowels: the change is simply i > ɨ. If the central vowel occurs after retroflex sibilants, the tongue position is approximately the same as for the consonant articulation. The same can be said for alveolar sibilants. There is a contrast between these front and non-front vowel syllables as seen in the Ménggǔ Zìyùn 蒙古字韻 (MGZY) (Section 6.4.8). As shown in Table 9.51, when the main vowels lost their front feature the palatal medial -j- (not indicated in the hP’ags-pa spellings) was also lost after retroflex _ initials. These two distinct syllables of the MGZY became homophones in the DYTJ and were listed in the same homophonic group just as they are in the modern Beijing dialect. Shen (2008b) shows that high vowel centralization and other related changes after retroflex initials are the results of the same phonological change. The result of this change is that finals lose their front feature after retroflex initials. Thus phonologically apical vowels must be central vowels so that they are defined as lacking the front feature (as well as lacking the back feature). This is why apical vowels are analyzed as Rank-I syllables in rhyme tables such as the Qièyùn Zhǐzhǎng Tú 切韻指掌圖; they lack the front feature that Rank-III finals should have. The term “apical vowel” itself is misleading and causes problems in phonological analysis. Because of their special articulation, apical vowels cannot be described as front, central, or back, and cannot be included in a regular vowel chart. This is probably why Karlgren could not include them in his vowel diagram (1915–1926: 217). The nature of the socalled apical vowels and their phonological relationship with other vowels is clearly indicated in the hP’ags-pa spellings in the MGZY, but it is unfortunate that the significance of _the phonological analysis indicated by the hP’ags-pa _ spellings has not been fully recognized. In conclusion, the following three points should be emphasized: (1) the historical cause of the emergence of the high central vowel is unrelated to the

The Ming Dynasty

355

emergence of its retroflex counterpart; (2) both retroflex and alveolar apical vowels are phonologically central vowels; and (3) centralization (apicalization) and the deletion of the front feature of front vowels are actually the same change, which occurs to resolve the articulatory conflict between the retroflex initials and the front feature contained in the finals. 9.6.2

Finals

The DYTJ has thirteen rhyme groups, a reduction from the nineteen groups of the ZYYY. Except for the zhù 祝 rhyme group, each of the groups is divided into kāi and hé tables (see Table 9.52). The reduction of rhyme groups indicates a change in finals as well as a different phonological analysis of the relationship between finals. Table 9.53 is Table 9.52 The thirteen rhyme groups of the Děngyùn Tújīn kāi 通 止 祝 蟹 壘 效 果 假 拙 臻 山 宕 流

tōng zhǐ zhù xiè lěi xiào guǒ jiǎ zhuō zhēn shān dàng liú



登 資

əŋ ɨ

iŋ i

咍 盃 蒿 訶 他 遮 根 干 當 齁

aj ej aw o a ɛ ən an aŋ əw

jaj

東 居 都 乖 灰 包 多 誇 靴 昏 湍 光 捋

ɚ

jaw jo ja jɛ in jɛn jaŋ jəw

uŋ ʉ u waj wej waw wo wa wɛ un wan waŋ wəw

yŋ y

ɥɛ yn ɥɛn

Table 9.53 A comparison of the rhyme groups of the Děngyùn Tújīng and of the Zhōngyuán Yīnyùn DYTJ

ZYYY

通 止 祝 蟹 壘 效 果

東鍾 支思 魚模 皆來 齊微 蕭豪 歌戈

庚青 魚模

DYTJ

ZYYY

假 拙 臻 山 宕 流

家麻 車遮 真文 寒山 江陽 尤侯

侵尋 先天

桓歡

監咸

廉纖

356

Toward Modern Mandarin

Table 9.54 Medials, main vowels, and codas of the Děngyùn Tújīng Medials

j, w, ɥ

Vowels

i•y e (ɛ) a

Codas -Ø: i (ɨ) u (ʉ) y o a ɛ -n: i ə u y a ɛ

-j, -w, -n, -ŋ -j: e a -ŋ: i ə u y a

ɨ ə

u o

-w: ə a -

a comparison of the rhyme groups of the DYTJ and of the ZYYY (see Section 7.3.2); the medials, main vowels, and codas of the finals are given in Table 9.54. 9.6.3

Tones

There are four tones featured in the DYTJ, called píng 平, shǎng 上, qù 去, and rú 如,7 corresponding to the yínpíng 陰平, shǎngshēng 上聲, qùshēng 去聲, and yángpíng 陽平 of the modern Beijing dialect, respectively. It is also worth noting the merging pattern of Middle Chinese rù syllables. Middle Chinese rù syllables with voiced initials merged with other tone groups according to the types of initials, with obstruents and sonorants merging into rú 如 (yángpíng) and qù tone respectively. But the merging pattern of the Middle Chinese rù syllables with voiceless initials does not match that of modern Chinese exactly, particularly many syllables which merged into qù tone and thus differ from the modern Beijing dialect. 9.6.4

Innovative Features

9.6.4.1 The Merger of Middle Chinese Division-II and Division-III Syllables with zhī 知 and zhào 照 Initials Middle Chinese syllables with zhī 知 and zhào 照 initials are listed in DivisionII and Division-III in rhyme tables, respectively. They are referred to as zhī-èr 知二/zhī-sān 知三 and zhào-èr 照二/zhào-sān 照三. In the DYTJ these two initial series merged. This is a very significant aspect of the historical development of Beijing phonology. The DYTJ was the first phonological work to systematically show such a change (see Table 9.55). The distinction between

7

Not the traditional rù 入.

The Ming Dynasty

357

Table 9.55 Some examples of the zhī 知 and zhào 照 initials merger II = III

MGZY 生 ʂəŋ 森 ʂəm 鄒 tʂəw

6¼ 6¼ 6¼

升 深 舟

ʂiŋ ʂim tʂiw

DYTJ 生 ʂəŋ 森 ʂəm 鄒 tʂəw

升 深 舟

= = =

ʂəŋ ʂəm tʂəw

Note: Division-II syllables are on the left of the associated pair, and the corresponding Division-III syllables are on the right. Merged characters are listed together in the same homophonic groups.

Table 9.56 Some examples of the wēi 微 and yǐng 影 initials merger wēi = yǐng ʋ- > Ø-/w-/u-

MGZY 味 ʋi 文 ʋən 望 ʋaŋ

6 ¼ 6¼ 6 ¼

威 溫 汪

uj wən waŋ

DYTJ 味 wej 文 un 望 waŋ

= = =

威 溫 汪

wej un waŋ

Note: wēi 微 initial syllables are on the left of the associated pair, and the corresponding yǐng 影 initial syllables are on the right.

Table 9.57 Some examples of the loss of the yí 疑 initial merger ŋ- > Ø-

MGZY 敖 ŋ- 我 ŋ- 昂 ŋ-

DYTJ 敖 Ø- 我 Ø- 昂 Ø-

zhī-èr 知二/zhī-sān 知三 and zhào-èr 照二/zhào-sān 照三 initials is one that has existed historically, and it is discussed in more depth in Section 1.4.1. 9.6.4.2 The Loss of the Middle Chinese wēi 微 ɱ- Initial Syllables with the Middle Chinese wēi 微 ɱ- initial lost their consonant initial and merged with Middle Chinese syllables with the yǐng 影 (Ø-) initial (see Table 9.56). 9.6.4.3 The Loss of the Middle Chinese yí 疑 ŋ- Initial Most Middle Chinese syllables with the yí 疑 ŋ- initial lost their consonant initial and merged with Middle Chinese syllables with the yǐng 影 initial (see Table 9.57). 9.6.4.4 The Rounded Central Vowel -ʉ One of the interesting characteristics of the DYTJ vowel system, as derived by comparison with phonological information found earlier and later than the DTYJ, is the rounded high central vowel (see Table 9.58). There are three groups of hékǒu 合口 syllables in contrast: zhū 珠 -ʉ, jū 居 -y, and gū 孤 -u. In

358

Toward Modern Mandarin

Table 9.58 The rounded central vowel of the Děngyùn Tújīng

止攝 止攝 祝攝

kāi



ɨ: 支止至直 i: 雞及

ʉ: 珠主住逐 y: 居局 u: 孤督母租阻畜梳辱

Table 9.59 The change of the rounded central vowel

支 珠

MGZY

DYTJ

Modern Beijing Mandarin

-i -y

-ɨ -ʉ

-ɨ -u

the DYTJ the rounded high central vowel -ʉ (the rounded counterpart of zhī 支 -ɨ) occurs in syllables with retroflex initials. Between the MGZY and DYTJ there was a process of vowel centralization after the retroflex initials. The high front vowels, unrounded and rounded, lost their front feature to become central vowels. The rounded central vowel later moved further back to merge with the high back vowel u, but the unrounded central vowel remained unchanged (see Table 9.59). While it was possible for -ʉ to continue its trend of vowel backing to merge with -u, -ɨ had no advantage, as the -ɯ vowel did not exist. 9.6.4.5 The Emergence of ɚ The phonological process changing pinyin ri to er (as in modern Beijing readings of 兒耳二而) involved a few intermediate steps. As discussed in Section 9.6.1.2, the front vowel i became a centralized vowel after the retroflex approximant ɻ-. Because of the retroflex approximant initial, the centralized vowel is also rhotacized, and metathesized, as recorded in the Persian transcriptions (see Section 8.2.3). The vowel quality should be similar to that of rì 日 in Modern Mandarin. Then the tongue position is lowered to a position close to schwa and even lower while the r-quality is still maintained. This process can be written as ɻi > ɻɨ > ɨr > ər (ɚ/ɐɚ) (for semantic reasons, rì 日 only participated in the first transition of this change to ɻɨ). It should be noted that the r in this transcription ɨr and ər indicates the rhotacization of the main vowel (R-colored vowel) rather than an alveolar trill taking the place of the coda. This process of lowering the vowel from the retroflex apical position to the schwa (directly underneath in a vowel chart) is another indication that the so-called apical vowel is actually a central vowel.

The Ming Dynasty

359

The recognition of ɚ as an independent final is first found in rhyme dictionaries of the late Ming, although there is evidence to suggest that this final was already in the phonological system in the fourteenth century. In the Liáoshǐ 遼史 (compiled at the end of the Yuan dynasty, in 1344 under Toqto’a [Chinese: Tuōtuō 脫脫]) there are a number of transcriptions indicating the existence of ɚ. The name of the Iranian city ‘Kerman’ is transcribed using Chinese characters 起兒漫 qǐ er màn, and an ethnic group inhabiting the area that is now Xinjiang province of China, ‘Uyghurs,’ is transcribed as 畏兀兒城 wèi wù er chéng. The syllabic final use of er 兒 indicates the loss of its initial and rhotacized vowel.

10

The Mandarin of the Qing Dynasty and the Modern Era

10.1

The Qing Dynasty (1644–1912)

The Mandarin recorded during the Qing dynasty has a very close resemblance to Modern Mandarin as represented by the Beijing dialect, thus the phonetic values of most of the phonological categories can be accurately estimated. The significant change of this period was the merger of velar initials and alveolar initials in front of palatal medials and vowels. In traditional Chinese phonology the result of this change is called jiān-tuán héliú 尖團合流 ‘the merger of jiān (尖) and tuán (團) initials.’ The jiān and tuán refer to two initial groups, jiānyīn 尖音 and túanyīn 團音. Jiānyīn initials are originally from the Middle Chinese jīng 精 initial group (ts-, tsh-, dz-, s-, and z-) and tuányīn initials are originally from the Middle Chinese jiàn 見 initial group (k-, kh-, and ɡ-) and initials xiǎo 曉 (h-) and xiá 匣 (ɦ-). This merger is clearly mentioned in the Yuányīn Zhèngkǎo 圓音正 考 of 1743 or earlier, which is actually an effort to reinstate the distinction between these two groups of initials (Yáng Yìmíng and Wáng 2003). In front of palatal medials -j- and -ɥ- and high front vowels i and y, syllables with earlier distinctive initials k-, kh-, x- and ts-, tsh-, s- change to tɕ- tɕh-, ɕ-.1 Table 10.1 shows the so-called jiān-tuán héliú 尖團合流, pairs of syllables with velar initials and alveolar initials became homophones. It should be noted that the two groups of initials involved in this merger did not change at the same time. As discussed in Section 9.6.1.1, many pieces of evidence indicate that the palatalization of velar initials occurred before the alveolar ones; if these two groups of syllables are distinct, then the contrast is usually palatal vs. alveolar, which means that this merger involved at least two steps: first the palatalization of the k- group initials, and then the palatalization of the ts- group. Evidence for the palatalization of velar initials is difficult to identify because the palatalization of velar initials just creates conditional variants. Palatal initials and velar initials can be perceived as the same initials in the phonological system. Table 10.2 shows the three-stage process of the 1

Not necessarily from Middle Chinese k-, kh-, x-, as the Middle Chinese initials xiǎo 曉 (h-) and xiá 匣 (ɦ-) had by this point already merged with the velar series.

360

The Qing Dynasty and the Modern Era

361

Table 10.1 Examples of the jiān-tuán héliú k- and ts- > tɕkh- and tsh- > tɕhx- and s- > ɕɥɛ 掘=絕

jɛ 結=節 怯=妾

血=雪

i 計=濟 契=砌 系=細

y 舉=咀 區=趨 吁=需

jaw 驕=焦 橋=樵 囂=消

jəw 九=酒 丘=秋 休=修

ɛn 肩=箋 牽=千

ɥɛn

in 巾=津 欽=侵 釁=信

jaŋ 姜=將 羌=槍 香=相

拳=泉 喧=宣

Note: The first character of each pair has a Middle Chinese velar initial and the second one has a Middle Chinese alveolar initial.

Table 10.2 The merging process of jiān (ts- group) and tuán (k- group) initials kkk-

Stage I Stage II Stage III

kjtɕjtɕj-

tststs-

tsjtsjtɕj-

only k- and tstɕj- as conditional variation of ktɕj- as conditional variation of both k- and ts-

Table 10.3 The complementary distribution of the tɕ-group and other initial groups

ktstʂtɕ-

Non-high front

High front

+ + + –

– – – +

merger of jiān (ts- group) and tuán (k- group) initials, as it affected the two initials k- and ts- and their palatal medial counterparts kj- and tsj-. Although the palatal initial group is in complementary distribution with the velar k- and alveolar ts- initial groups, the palatal group is treated as an independent initial group in Beijing phonology. The technical as well as theoretical problem is that the palatal initial group is in complementary distribution with both the velar and alveolar group and even with the retroflex group (see Table 10.3), as pointed out by Yuen Ren Chao (1934) in his well-known article titled “The non-uniqueness of phonemic solutions of phonetic systems.” 10.1.1

The Yùdìng Pèiwén Yùnfǔ 御定佩文韻府 (1711)

The Yùdìng Pèiwén Yùnfǔ ‘The Imperial Approved Rhyme Storehouse of Peiwen’ (often referred to as the Pèiwén Yùnfǔ 佩文韻府, without the phrase Yùdìng

362

Toward Modern Mandarin

‘imperial approved’)2 is a rhyming dictionary compiled on imperial order by Emperor Kangxi 康熙 (Kāngxī, 1654–1722) and under the guidance of Zhāng Yùshū 張玉書 (1642–1711) (see Figure 10.1). It was published in 1711 and has 106 main volumes and 106 supplementary volumes. It was expanded to a total of 444 volumes later. The preface was written by Emperor Kangxi, who himself was one of the compilers (Xiào 1990). The preface provides a simple statement about the reason for compiling this dictionary and its main structure. It was the largest rhyming dictionary of the time, representing the highest achievement of the traditional rhyming dictionary. It is called the “Kangxi Treasures” by James Legge (1893) in his well-known Chinese Classics. Although the phonological standard of the written Chinese as shown in the Pèiwén Yùnfǔ (PWYF) does not reflect the contemporary pronunciation, it had immense significance in Imperial China. The phonological system of the official rhyme dictionaries provided a national standard for the composition of traditional poetry, which requires rhyming. The PWYF, as a continuation of this phonological standard, helped to ensure that the standards that made learning and composing traditional poetry possible for Chinese people for more than a thousand years would still do that into the Qing dynasty. Without such a supratemporal (accessible across different eras) and supradialectal (accessible across different dialects) system, it would have been extremely difficult for traditional poetry composition to continue to exist in a way that allowed poetry from the Tang dynasty to be read alongside poetry from a later time. The PWYF is arranged according to the system of 106 rhymes, or the so-called Píngshuǐ Yùn 平水韻, established in the thirteenth century (refer to Section 5.4.1). The pronunciation notations are the fǎnqiè 反切 spellings copied from the Guǎngyùn 廣韻 and the Jíyùn 集韻. This phonological system, mainly the rhyming system, was based on the phonological system of Middle Chinese rather than the contemporary pronunciation. For more than a thousand years such a phonological system of literary language was still very much actively used in Chinese society until the beginning of the twentieth century – when the imperial examination ended and there was no need for a rhyme standard any more. In imperial times, however, such a phonological tradition had a very practical purpose for anyone who wanted to be literate and be successful passing imperial examinations. The content of the dictionary is designed to provide convenience for poetry composition. In a homophonic group, characters are listed according to their usage frequencies, from most frequently used to the least used rare characters. Each character has an entry that provides a concise explanation about its

2

Pèiwén is derived from the Kangxi emperor’s study, the Pèiwén Shūzhāi 佩文書齋 ‘Peiwen Studio.’

363

Figure 10.1 The first three pages of the Pèiwén Yùnfǔ 佩文韻府 (Sìkù Quánshū 四庫全書 version): (a) The second column contains the title of the rhyme dictionary. The fourth column is the first entry under the character dōng 東, the phonetic and semantic notations. The fifth column starts with the title, yùnzǎo 韻藻. The rest are the yùnzǎo phrases with the character dōng 東 as the last character, as well as their allusions. (b) The fifth column from the left is the beginning of the duìyǔ 對語. (c) The first column is the beginning of the zhāijù 摘句.

364

Figure 10.1 (cont.)

365

Figure 10.1 (cont.)

366

Toward Modern Mandarin

pronunciation and its meaning. The next part is the so-called yùnzǎo 韻藻, which is a list of disyllabic or polysyllabic phrases and their allusions. The character that determines the position of the entry in the book is always the last character of the phrase. So, in a sense, this is a kind of reverse dictionary. The first part of the yùnzǎo was copied from two previous dictionaries, the Yùnfǔ Qúnyīn 韻府群音 of the Yuan dynasty (1279–1368) and the Wǔchē Yùnruì 五車韻瑞 of the Ming dynasty (1368–1644), and the second part was newly added, starting with the character zēng 增 ‘adding.’ After the yùnzǎo are new additions, which contain duìyǔ 對語 ‘parallel phrases,’ and zhāijù 摘句 ‘quotations,’ all of which contain the rhyme character. These additional parts clearly indicate the purpose of the PWYF, which is for the composition of traditional poetry. About the same time, the Kāngxī Zìdiǎn 康熙字典 ‘Kangxi Dictionary’ was published. Unlike the PWYF, the Kāngxī Zìdiǎn, published in 1716, is a character dictionary, in which all fortyseven thousand characters are arranged according to the radicals of the characters instead of the rhymes. These two dictionaries have different purposes, the PWYF is for poetry composition and the Kāngxī Zìdiǎn is for general purposes. Due to its huge volume, a stripped down version called the Pèiwén Shīyùn 佩文詩韻 ‘Poetry Rhyming of Peiwen’ was also published. It has the characters, their pronunciation, and their meanings only. The rest, yùnzǎo, duìyǔ, and zhāijù, were all deleted. Such a concise version is convenient for carrying around and looking up the rhyme of characters. 10.1.2

The Yīnyùn Chǎnwēi 音韻闡微 (1728)

In the Qing dynasty there was also an effort by the government to establish a phonological standard, and in pursuit of this goal the Yīnyùn Chǎnwēi (YYCW) was compiled by Lǐ Guāngdì 李光地 (1642–1718), Wáng Lánshēng 王蘭生 (1679—1736), and Xú Yuánmèng 徐元夢 (1655-1741), and published in the sixth year of the Yongzheng reign (1728) (see Figure 10.2). For linguistic as well as political reasons, this rhyme dictionary did not follow the Hóngwǔ Zhèngyùn 洪武正韻 of 1375, the rhyming standard established in the Ming dynasty. Instead it followed the earlier rhyming tradition of the Song and the Yuan, such as the Píngshuǐ Yùn 平水韻 (~1223) and the Yùnhuì Jǔyào 韻會舉要 (1297). Similar to all previous rhyme dictionaries, however, it still represents an ideal phonology, the so-called zhèngyīn 正音 ‘standard pronunciation.’ This standard is not based on any dialect, including that of the capital. The concept of zhèngyīn is clearly defined in the dialogue of the Zhèngyīn Jǔhuá 正音咀華 of 1836 by Shā Yízūn 莎彝尊: What is zhèngyīn 正音 ‘standard pronunciation’? Answer: Following the emperor’s approved pronunciation of the characters in the Zìdiǎn (Kāngxī Zìdiǎn 康熙字典) and the Yīnyùn Chǎnwēi 音韻闡微 is the zhèngyīn.

367

Figure 10.2 The first page of the Yīnyùn Chǎnwēi 音韻闡微 (Sìkù Quánshū 四庫全書 version). It shows some of the homophonic groups of the dōng 東 rhyme. Different from previous rhyme dictionaries, each homophonic group starts with a new column with a traditional initial character. The fourth column starts with the initial character jiàn 見 to indicate the first homophonic group, and the eighth column starts with the initial character xī 溪 to indicate the second homophonic group, etc.

368

Toward Modern Mandarin

What is nányīn 南音 ‘southern pronunciation’? Answer: In the ancient time the capital was set up in Jiangnan province, so the speech of Jiangnan province is the nányīn. What is běiyīn 北音 ‘northern pronunciation’? Answer: Now the capital is set up in Beiyan (north of the Yan area), so the speech of Beijing city is the běiyīn.3

Around sixteen thousand Chinese characters are arranged into 106 rhymes according to the Píngshuǐ Yùn. Three types of fǎnqiè spellings are provided for each homophonic group, the fǎnqiè of the Guǎngyùn, Jíyùn, and a new fǎnqiè. In this newly designed fǎnqiè, the upper spellers are syllables with an open syllable and main vowels i, y, e, a, o, u; the lower spellers are syllables with zero initial. For example, in the Guǎngyùn the spellers for gōng 公 are gǔ 古 and hóng 紅; in the YYCW the new fǎnqiè spellers are gū 姑 and wēng 翁, which have no coda and no initial respectively. As with the old method, the upper speller has the same initial (I) as the character in question, the difference being that the final of the upper speller gū 姑, u (V+E), also matches the main vowel of the character to be spelled, gōng 公, -uŋ. The lower speller has the same final, -uŋ (V+E), but zero initial. Also, tonally, both upper and lower spellers must match the character in question (see Table 10.4). Therefore, by pronouncing the upper and lower spellers together, the pronunciation of the character in question can be learned in a more natural way. Of the upper spellers, they can be divided into four types of main vowel, a, i, u, and y, each corresponding to the four types of medial, -Ø-, -j-, -w-, and -ɥ-, respectively. Table 10.5 is a more thorough look at the creation of the new fǎnqiè. Four types of main vowel, a, i, u, and y, of the upper speller also represent corresponding four types of medial, -Ø-, -j-. -w-. and -ɥ-, respectively. If a medial is present, it must be matched with both spellers. For the same initials and same finals, the same spellers are used. This is a significant improvement from the fǎnqiè spellings of the Qièyùn and Guǎngyùn and in the phonological analysis. The phonological categories are very similar to Modern Mandarin, so much so that a reconstruction is not required for this book. Table 10.4 An example of the new fǎnqiè method of the Yīnyùn Chǎnwēi

Character Upper speller Lower speller

3

公 姑 翁

YYCW

Initial

Main vowel

Ending

kuŋ ku uŋ

k k Ø

u u u

ŋ Ø ŋ

“何為正音?答曰:遵依欽定《字典》《音韻闡微》之字音即正音也。何為南音?答曰: 古在江南省建都,即以江南省話為南音。何為北音?答曰:今在北燕建都,即以北京城話 為北音。”

The Qing Dynasty and the Modern Era

369

Table 10.5 A schematic illustration of the new fǎnqiè method of the Yīnyùn Chǎnwēi Without medial

Character Upper speller Lower speller

10.1.3

With medial

Initial

Medial

Main vowel

Final

Initial

Medial

Main vowel

Final

A A Ø

– – –

C C C

D – D

A A Ø

B B B

C – C

D – D

The Wade Textbook (1867)

語言自邇集 Yü-yen tzǔ-êrh chi:4 A Progressive Course Designed to Assist the Student of Colloquial Chinese by Thomas Francis Wade (1818–1895) (see Figure 10.3) is a good historical record of colloquial Mandarin (Wade 1867). This textbook was published in London in 1867. Wade was a British diplomat who worked for over forty years at the British embassy in China. He returned to England in 1883. In 1888, he was appointed the first Professor of Chinese at the University of Cambridge. He served as president of the Royal Asiatic Society from 1887 to 1890. Wade is the sole author of the first edition (1867), but Walter Caine Hiller was added as a co-author of the second edition (1886). According to the preface written by Wade, it is likely that the pronunciation was based on the Peking (Beijing) dialect spoken by his language teacher Ying Lung-tʻien. Regarding this textbook Wade wrote: The work is in two principal divisions, respectively denominated Colloquial and Documentary Series. The words TZǓ ÊRH CHI which recur in the title of both may be fairly translated Progressive Course. To go far, says a Chinese classic, we must start tzǔ êrh, from what is near. The two courses are chi, collections of matter, of which that distinguished by the prefix yü-yen, words and phrases, is the colloquial, the other, being a collection of wen-chien, written papers, the documentary course. (1867, Preface, iii)

The colloquial series is based on the spoken Peking (Beijing) dialects, as “in the oral language of the metropolitan department, styled for brevity the Peking dialect” (Wade 1867, Preface, iv). In this book, not only are the phonetic transcriptions provided but a description of each consonant and vowel is also detailed. Such information provides the phonetic qualities of the sounds, but also, more interestingly, provides knowledge on how these sounds were perceived by Wade, a native English

4

This is Wade’s romanization system, the pinyin for this is: Yǔyán Zì’ěr Jí.

370

Toward Modern Mandarin

Figure 10.3 The cover of the 語言自邇集 Yü-yen tzǔ-êrh chi.

speaker. The phonological information is basically contained in the first part, which is devoted to pronunciation. The romanization system used in the Yü-yen tzǔ-êrh chi later developed as a standard spelling, as shown in Herbert A. Giles’s Chinese–English Dictionary of 1892, and is commonly referred to as Wade–Giles. This system was used in the English-speaking world until 1979, when it was replaced by the pinyin

The Qing Dynasty and the Modern Era

371

(Hànyǔ Pīnyīn 漢語拼音) system. From the comparisons shown in Table 10.6 it can be seen that the selection of the letters in Wade–Giles has clear influence on the design of pinyin. The sound table of the Yü-yen tzǔ-êrh chi is a very important piece of information. Although the syllable types are listed without tones, this is probably one of the first lists of this kind for Mandarin phonology. It should be noted that at the time, Wade wrote: My difficulty, when I first tried to form a list of syllables, was this, that no native work contained a syllabic system at all to be relied on . . . It was not till 1855, when I had been making and re-making orthographies for some eight years, that a native author brought out a fair approximation to a Peking sound-table. (1867, Preface, vii)

The phonological information is not given in the format of Chinese tradition in terms of initials, finals, and tones. The phonological information of initials, finals, and tones can be retrieved from the spelling system that Wade devised for the transcription and the sound table (a syllable list). In his Part 1. Pronunciation, Wade (1867) says, “In order to correctness of Pronunciation of Chinese, three conditions must be satisfied; there must be accuracy of Sound, of Tone, and of Rhythm (intonation)” (p. 3). In the section on Sound, the alphabet employed is given along with descriptions of the values. The letters of the alphabet are listed in two parts: Vowels and Diphthongal Sounds, and Consonantal Sounds. From these two lists different types of initials and finals can be retrieved in comparison with the pinyin system. 10.1.3.1 Initials In the Consonantal Sounds, twenty-five letters are listed in the following order, which does not follow the traditional order from labials to the velars, but rather the order of English letters. The unaspirated and aspirated consonants of the same place of articulation are listed in pairs. The ‹ʻ›, called “breathings” in the book, are used to mark the aspirated sounds. ch, chʻ, f, ʻh, hs, j, k, kʻ, l, m, n,’ng, p, pʻ, s, sh, ss, t, tʻ, ts, tsʻ, tz, tzʻ, w, y

These twenty-five initials have correspondences with pinyin as shown in Table 10.6. Based on the comparison shown in Table 10.6, the system described by Wade is very much like that of Modern Mandarin. Besides the different choices of letters, the phonological differences are explained in the following list. 1. tz, tzʻ, and ss only represent initials in the syllables of tsɨ (zi), tshɨ (ci), sɨ (si). As described by Wade (1867), “tz is employed to mark the peculiarity of the final ǔ, but is hardly of greater power than ts” (p. 5); tz, tzʻ, and ss are conditional variations of ts, tsʻ, s. It shows that these syllables, tzǔ, tzʻǔ, and ssǔ, seem to catch Wade’s special attention. However, in pinyin the initials in the syllables tʂɨ (zhi), tʂhɨ (chi), ʂɨ (shi), and ɻɨ (ri) are not separately listed when before a high central vowel.

372

Toward Modern Mandarin

Table 10.6 A correspondence table showing a comparison of the Yü-yen tzuêrh chi with modern pinyin, and accompanying IPA tzu-êrh chi p t ts tz ch ch k y a

pʻ tʻ tsʻ tzʻ chʻ chʻ kʻ w

Pinyin [IPA] m n

f l s ss sh hs ‘h

‘ng

j

b [p] d [t] z [ts] z [ts] zh [tʂ] j [tɕ] g [k] -

p [ph] t [th] c [tsh] c [tsh] ch [tʂh] q [tɕh] k [kh] -

m [m] n [n]

a

f [f] l [l] s [s] s [s] sh [ʂ] x [ɕ] h [x]

r [ɻ]

See item 6 in the list of phonological differences for an explanation of ng’s absence in pinyin initials.

Table 10.7 The orthographical difference of the alveolar and retroflex initials Pinyin Alveolars Retroflexes

z zi zh zhi

tzǔ-êrh chi c ci ch chi

s si sh shi

ts tzǔ ch chih

tsʻ tzʻǔ chʻ chʻih

s ssǔ hs hsih

2. ch, chʻ are used to represent two series of affricates, tʂ, tʂh and tɕ, tɕh (see Table 10.7), because these two series of affricates, retroflexes, and palatals are in complementary distribution. In Modern Mandarin the palatal series only combines with the finals with high front vowels or medials, while the retroflexes cannot combine with them. These two series of consonants do not occur in the same condition and are not contrastive. Thus, these two phonetically different consonants can be represented by the same set of letters. In Mandarin phonology, palatal affricates are however in complementary distribution with three series of consonants, the alveolars and the velars, as well as the retroflexes. To group together the retroflexes with the palatals indicates that the closeness of different consonants was perceived by Wade. The retroflex and palatal fricatives are not treated the same way, however; they are represented by sh and hs respectively (see Table 10.8). 3. j stands for the retroflex approximant ɻ. Wade (1867) describes it as “most nearly the French j in jaune; our s, in fusion, or z in brazier, are the nearest imitation of which our alphabet admits” (p. 5). This transcription indicates that r was perceived as a voiced fricative. 4. The medials y and w are listed as initials. However, Wade (1867) was not so certain about the status of their pronunciation. He says “y; as in English, but very faint before i, or ü.” And “w; as in English; but very faint before u, if

The Qing Dynasty and the Modern Era

373

Table 10.8 The retroflex serials and palatal series represented by the same letters, ch and chʻ

Palatals

Retroflexes

家 恰 夏 乍 茶 殺

‘home’ ‘just’ ‘summer’ ‘suddenly’ ‘tea’ ‘kill’

Pinyin

tzǔ-êrh chi

jiā qià xià zhà chá shā

chia chʻia hsia cha chʻa sha

indeed it exist[s] at all” (p. 5). However, it is worth noting that to list these two medials as initials indicates that they have more friction when they start a syllable. 5. The merger of Middle Chinese alveolar and velar consonants in front of high front vowels. hs is used to spell both. 6. ng is not listed, but Wade (1867) noted: “The following sounds: a, ê, ai, ên, an, êng, ang, o, ao, où, are as often pronounced nga, ngai, ngan, and so on” (p. 11). ng is likely a variation of individual speakers. Based on these six points, the initials system of the Yü-yen tzǔ-êrh chi is in fact the same as Modern Mandarin. 10.1.3.2 Finals Wade did not list all the finals. In the list of Vowels or Diphthongal Sounds, twenty-nine types of sounds are provided: a, ai, ao, e, ei, ê, i (as in syllable mêi), êrh, i (as a single syllable), ia, iai, iao, ie, io, iu, o, ou, ü, üa, üe, üo, u, ua, uai, uei, uê, ui, uo, ǔ

According to the syllable list, the following forty-four finals can be identified. a, ai, an, ang, ao (5) e, ei (2) ê, ên, êng, êrh (4) i, o, ou (3) i, ia, iai, iang, iao, ie, ien, ih, in, ing, io, iu, iung (13) ü, üan, üeh, ün, üo (5) u, ua, uai, uan, uang, uei, uê, ui, un, ung, uo, ǔ (12) (44 in total) The seven sounds listed in Table 10.9 are explained here: (1) e: “in eh, en, as in yet, lens” (Wade 1867: 3). It is not a final by itself. (2) iai: used to represent the residual pronunciation of the Middle Chinese Rank-II words. kǎi 楷, is chʻiai, and yá 涯, is yai. This final was common in Old Mandarin but is not in the pinyin system of Modern Mandarin.

374

Toward Modern Mandarin

Table 10.9 The Yü-yen tzǔ-êrh chi finals in comparison with Modern Standard Mandarin in pinyin spelling Pinyin (37)

tzǔ-êrh chi (44)

ɑ, o, e, (ê), i, u, ü ɑi, ei, ɑo, ou ia, ie, iɑo, iou uɑ, uo, uɑi, uei üe ɑn, en, ang, eng, ong ian, in, iang, ing, iong uan, uen, uang, ueng üan, ün -i, er

a, o, ê, e(1), i, u, ü ai/iai(2), ei/êi(3), ao, ou ia/iai(2), ie, iao/io(4), iu ua, uo, uai, ui/uei(5) üeh/io(4)/üo(6) an, ên, ang, êng, ung ien, in, iang, ing, iung uan, un/uê(n)(7), uang, (ung) üan, ün ǔ/ih, êrh

(3) i: there are two i’s in the list of Vowels and Diphthongal Sounds, one is a main vowel, or a final. The other i is used only after ê. Wade (1867) explains, “the foregoing ê followed enclitically by y. Strike out the n from the word money, and you have the syllable mêi” (p. 3). (4) io: used for the alternative pronunciation of Middle Chinese rù syllables (e.g. 角 chio, 卻 ch‘io, 略 lio). In the Table of Characters Subject to Changes of Sound and Tone, these syllables have multiples readings (see Table 10.10; tones are omitted). (5) uei: not used as a final in the sound table. (6) üo: is used for the alternative pronunciation of Middle Chinese rù syllables (e.g. 爵 chüo, 卻 chʻüo, 略 lüo). The other readings are shown in Table 10.11 (with tones omitted). Table 10.10 The Yü-yen tzǔ-êrh chi multiple readings of jiǎo 角, què 卻, lüè 略 角 jiǎo ‘corner’ 卻 què ‘but’ 略 luè ‘brief’

chio, also chiao, chüeh; ch‘io, also ch‘i, ch‘üeh lio, also liao, lüo, lüeh

Table 10.11 The Yü-yen tzǔ-êrh chi multiple readings of jué 爵, què 卻, lüè 略 爵 jué ‘the rank of nobility’ 卻 què ‘but’ 略 luè ‘brief’

chüo, also chiao, chio, chʻio, chüeh ch‘üo, also ch‘io, ch‘i, ch‘üeh lüo, also liao, lio, lüeh

The Qing Dynasty and the Modern Era

375

(7) uê: only found in the final uên, frequently wên, or wun. It is difficult to distinguish uên from un in many cases. It is also found that lêi 累, with final êi, is listed along with lei as an alternative pronunciation. 10.1.3.2.1 Further Notes on Finals When describing the pronunciation of ǔ, Wade (1867) says, “Between the i in bit and the u in shut; only found with the initials ss, tz, tzʻ which it follows from the throat, almost as if the speaker were guilty of a slight eructation. We have no vowel sound that fairly represents it” (p. 4). This vowel was perceived as a high central vowel (ɨ in IPA) by Wade. This is actually quite close to the description in many earlier transcriptions, for example in the hP’ags-pa script. _ labial initials The final êng instead of ung is used to transcribe syllables with and with no initial (see Table 10.12). This is quite a recent change; in the earlier Manchu spellings these finals were still ung. The letter ê is used to transcribe the finals e, en, and eng in pinyin. Wade (1867) describes it as “nearest approached in English by the vowel sound in earth, in perch, or in any word where e is followed by r, and a consonant not r, as in lurk” (p. 3) (here, of course, Wade must refer to the schwa ə). To group the main vowel of these three finals together is quite an accurate observation. 10.1.3.3 Tones Four tones are listed. The Middle Chinese rù tone is not distinguished. “The number of the shêng differs in different dialects. Books recognise five. In the Peking dialect there are now four; . . . The ju, or entering, an abrupt tone still recognized in studying the writing language . . . is now ignored in the practice of the spoken language of Peking” (Wade 1867: 6). Wade called the four tones shangpʻing 上平, hsiapʻing 下平, shang 上, and chʻü 去, and used the numbers 1, 2, 3, and 4 at the upper right corner of a syllable to indicate the tonal categories respectively. He also provided his description of the tonal pitch in a dialogue by using different intonations. A, B, C, and D are the four persons engaged (see Table 10.13). “In the four lines below [Table 10.13], I have supposed A to assert his death in the 1st tone, B, to express his apprehension that he has been killed, in the 2nd, C, to scout this suspicion, in the 3rd, and D, to confirm it sorrowfully, in the 4th” (Wade 1867: 7). The tonal values cannot be interpreted exactly. How close they are to the tonal values of Modern Mandarin is unclear. Table 10.12 Labial initial or initial-less syllables with êng finals 風 fêng ‘wind’

夢 mêng ‘dream’

朋 pʻêng ‘friend’

翁 wêng ‘old man’

376

Toward Modern Mandarin

Table 10.13 The tonal pitch of the four tones described as different intonations pa1 pa2 pa3 pa4

1 2 3 4

shangpʻing hsiapʻing shang chʻü

A. Dead. B. Killed? C. No! D. Yes!

10.1.3.3.1 Tone Sandhi The tone sandhi of the third tone is noted in Wade’s book. Two Chinese characters both with the third tones but in different order, xiǎo mǎ 小馬 and mǎ xiǎo 馬小, are used to illustrate the sandhi rule, which is that, as is the same in both Modern Mandarin and the Chinese found in Section 9.4, a third tone changes to a second tone in front of another third tone. In the discussion Wade was quite sure that the third tone changes to a second tone. But the tonal value of the third tone in front of the other three tones was not mentioned. In Modern Mandarin, the tonal value of the third tone in front of a non-third tone is described as 21 or a low tone, which is different from the value of the citation form 214. The 21 realization is also considered as the result of tone sandhi. If Wade does not distinguish the tonal value of the citation form and the sandhi form in front of a non-third tone, he must treat the citation form as a low tone. This observation is actually quite correct, because the full tonal curve 214 only appears when the third tone syllable is emphasized. In normal speech a low tone is often the tonal value. 10.1.3.4 Colloquial Speech: Neutral Tone and Rhotacized Finals In the Yü-yen tzǔ-êrh chi two phonological phenomena in the colloquial speech were paid attention to. The first is the neutral tone. In the spellings, the syllables with neutral tone are not marked with the tonal numbers. Table 10.14 shows that the syllables with rhotacized finals are written as if two separate syllables, the normal final plus -rh, which is not the final êrh. 10.1.3.5 Alternative Readings The Table of Characters Subject to Changes of Sound and Tone is a very valuable piece of information. It lists multiple readings of 1,525 characters (Zhāng Wèidōng 2002). The majority of multiple readings are about the Table 10.14 -rh as a suffix of the preceding syllable 把 pa3 BA ‘To cover it up with a

蓋儿 kai4-rh cover (noun) cover.’

蓋 kai4 cover (verb)

上 shang4 ‘up’

The Qing Dynasty and the Modern Era

377

Table 10.15 Some examples of characters with multiple readings 暫 襯 森

zàn chèn sēn

‘temporary’ ‘lining’ ‘grove’

chan4, tsan4 chʻên4, tsʻên4 sên1, shên1, shêng1

阮 容 銳

ruǎn róng rui

‘surname’ ‘contain’ ‘sharp’

juan3, yüan2, yüan3 jung2, yung2 jui4, tui4, wei4

挨 楷 街

āi kǎi jiē

‘suffer’ ‘standard’ ‘street’

ai1, ai2, yai2 chʻiai3, kʻai3 chieh1, kai1

白 北 擇 色

bái běi zé, zhái sè, shǎi

‘white’ ‘north’ ‘select’ ‘color’

po1, pai2, po2, po4 pei3, po4, pu3 chai2, tsê2 shai3, shê4, sê4

並 盡 判 窘

bīng, bìng jìn pàn jiǒng

pin4, ping1, ping4 chin4, ching4 pʻan4, pang4 chiong3, chün3

different pronunciations of the literary reading and the colloquial reading. The list shows how the standardization process selected the competing readings. Some of them still exist in the Beijing dialect and other Mandarin dialects of the modern time. The ones that still exist in the Beijing dialect, and are the same as the pinyin spellings, are in bold in Table 10.15. The selections are mainly based on the regular changes from the Middle Chinese phonological categories reflected in the Beijing dialects. The other readings are basically the influence from the Mandarin dialects spoken in other cultural and political centers, similar to the system of the Zhōngyuán Yīnyùn 中 原音韻 (see Section 7.1). 10.2

Post-Imperial China (1912–present)

10.2.1

The Old National Standard

In 1921 Zhào Yuánrèn 趙元任 (1892–1982), better known in non-Chinese circles as Yuen Ren Chao, or Y. R. Chao, was asked to make an audio recording of this ‘national pronunciation’ to be used in schools. In 1913 the Committee of Unified Pronunciation (Guóyīn Tǒngyī Huì 國音統一會) of the Ministry of Education approved the national pronunciation of 6,500 characters and also

378

Toward Modern Mandarin

decided on letters for their phonetic notation. As a result, the Guóyīn Zìdiǎn 國 音字典 was published in 1919, which was referred to as lǎo guóyīn 老國音 ‘old national pronunciation.’ Because it was not based on a real dialect, for many years Chao was the only person who could pronounce the national standard. The old national pronunciation, which is a mixed system using the traditional category system and the pronunciation of different dialects, was ultimately another idealized standard. It kept the traditional distinctions, such as the jiantuan 尖團 contrast (tsj- vs. tɕj, tshj- vs. tɕhj-, and sj- vs. ɕj-), the velar nasal initial (ŋ-), the labiodental approximant (ʋ-/v-), and the rù 入 tone. For the purpose of the time, in an age when mass media lacked the ability to propagate information across the entire country, and in a newly founded republic where communication and education needed to be more reliable and consistent, the old national standard was a failure. 10.2.2

The New National Standard

In 1932 the Guóyīn Chángyòng Zìhuì 國音常用字匯 ‘A Glossary of Frequently Used Characters in National Pronunciation’ was published using the pronunciation of the Beijing dialect as its phonetic standard. This standard was referred to as Xīn Guóyīn 新國音 ‘new national pronunciation.’ Since then, the Beijing pronunciation has become the norm of pronunciation and the controversy over the standard has been settled for the time being. As one of the scholars to be involved in the process and witness this historical transition, in his article entitled “What is correct Chinese?” Chao ([1961] 1976) wrote with a good sense of humor to recapture this change: For thirteen years I was the sole speaker of this idiolect, meant to be the national language of 4, 5, or 600 million speakers. By 1932, without publicly announcing any radical change, the Gwoin Tzyhdean (Guóyīn Zìdiǎn) was quietly revised in the form of Gwoin Charngyong Tzyhhuey (Guóyīn Chángyòng Zìhuì), in which the actual content was based on the speech of Peiping [Beijing]. Thus, at one stroke, were created more than one million potential teachers instead of only one. (pp. 79–80)

The differences between this national standard and the then Beijing dialect were listed in Wáng Pú’s 王璞 Guóyīn Jīngyīn Duìzhàobiǎo 國音京音對照表 ‘A Comparative List of the National Pronunciation and the Beijing Pronunciation,’ published in 1921. This list, however, also shows that the “new national standard” is not exactly the same as the colloquial speech of Beijing. The pronunciation of many syllables of the new standard actually was artificially rendered by the author based on his understanding of traditional categories (Chin 2011). There are 1,250 characters included in the Guóyīn Chángyòng Zìhuì. Since it represented the new standard and was based on the Beijing dialect, it is interesting to see how closely the new standard represented the Beijing dialect recorded by Wáng Pú. The 1,250 characters have 1,490 pronunciations

The Qing Dynasty and the Modern Era

379

because some characters have multiple readings. Of these, 233 readings are the same in all three references: the Beijing dialect, the new standard, and the old standard. Among the rest of the 1,257 readings, 80 percent of the syllables are the same as the Beijing pronunciation, 5 percent are the same as the old standard, and 15 percent are the same as neither the Beijing pronunciation nor the old standard (Chin 2011). From the statistics it is clear that from the beginning the national standard, while being based on the Beijing dialect, was not the Beijing dialect. Nevertheless, for the first time in the history of China, the standard phonology departed from the literary standard for the written language, and became a pronunciation standard of the spoken language. The national standard was no longer supradialectal. This change also reminds us that if even by the twentieth century there was no real dialect-based standard pronunciation, it is likely that such was also the situation throughout history, with idealized standards never based on any real dialect. 10.2.3

Modern Standard Phonology

In the People’s Republic of China, the official spelling system of modern standard Chinese is a version of the Latin alphabet called Hànyǔ Pīnyīn 漢語 拼音 ‘Chinese spelling system,’ or simply romanized as pinyin (see Table 10.16). The first edition of Hànyǔ Pīnyīn was approved and adopted at the Fifth Session of the First National People’s Congress on February 11, 1958. In 1982 Hànyǔ Pīnyīn was adopted by the International Organization for Standardization as the standard romanization for modern Chinese, and in 1986 it was adopted by the United Nations. The final system has three medials: -j-, -w-, and -ɥ-, which in pinyin correspond to -i- (y- initially), -u- (w- initially), and -ü- (yu- initially); and four codas: -n, -ŋ, -j, -w (pinyin: -n, -ng, -i, -o/u). An ‹ i › written after zh, ch, sh, or r (tʂ-, tʂh-, ʂ-, ɻ-) is read as ɨ. 10.2.4

Modern Dialects (Fāngyán 方言)

Traditionally, the purpose of phonological study was either philological or for providing a standard for poetry. Dialects were never the main interest of scholars before the modern era, and very little phonological information was recorded regarding dialects, although it is unquestionable that they must have existed throughout history. Since the introduction of the discipline of modern linguistics in the early twentieth century, the study of dialects has become a very important field in modern Chinese linguistics. The complexities of dialects, their historical development, and their classification have been a major attraction to generations of Chinese linguists.

380

Toward Modern Mandarin

Table 10.16 The Modern Standard Phonology, as represented by pinyin and IPA Initials (21)

Labial Alveolar Palatal Retroflex Velar

Pinyin b d z j zh g

p t c q ch k

m n s x sh h

IPA p t ts tɕ tʂ k

ph th tsh tɕh tʂh kh

m n s ɕ ʂ x

f l

r

f l

ɻ

Finals (35) Pinyin a i ia u ua ü

o

e ie

uo

ai

ei

uai

uei

aj

ej

waj

wej

ao iao

ou iou

an ian uan üan

en in uen ün

ang iang uang

eng ing ueng

ong iong

aw jaw

ow jəw

an jɛn wan ɥɛn

ən in wɑn yn

ɑŋ jɑŋ wɑŋ

əŋ iŋ wəŋ

uŋ juŋ

üe

IPA i/ɨ u y

a ja wa

o

ɤ jɛ

wo ɥɛ

Tones (4)

Numerical notation

ā 55

á 35

ǎ 214

à 51

Systematic surveying of dialects started with the prominent Chinese linguist Y. R. Chao. He studied thirty-three sites in Wu-speaking areas in the 1920s and the Hubei and Yunnan dialects in the 1930s and 1940s, and designed a highly practical survey form known as Fāngyán Diàochá Zìbiǎo 方言調查字 表 ‘List of Characters for the Investigation of Dialects,’ which is still used by Chinese dialectologists today. The journal Fangyan, the Language Atlas of

The Qing Dynasty and the Modern Era

381

China of 1987 (Australian Academy of the Humanities, Zhōngguó Shèhuì Kēxuéyuàn, et al. 1987) and the Hànyǔ Fāngyán Dìtú Jí 漢語方言地圖集 ‘Atlas of Chinese Dialects’ of 2008 are all showcases of Chinese dialectology (Cáo, Zhìyún et al. 2008). Based on available information, however, the historical development of Chinese dialects still cannot be determined with confidence, due to the lack of both historical data and theories concerning language change. It is unlikely that historical linguistics based on Indo-European languages will provide adequate explanation of the language changes observed in Chinese dialects. The fertile field of Chinese dialectology has provided a test ground for various existing theories in traditional historical linguistics, such as the regularity of sound change, the tree model, the wave model, etc. It can also provide materials for the formation of new theories such as lexical diffusion (Wang 1967, 1979). It has become well known that because of the lack of mutual intelligibility, different Chinese dialects are actually different languages. If mutual intelligibility is used as the criterion to judge the diverse Chinese dialects, then Chinese is a language family comprising of what must be hundreds of “languages.” 10.2.4.1 Dialect Classification The practice of dialect classification closely reflects the knowledge of the field. Many proposals have been raised, most influential among them being the seven-dialect classification and the most recent ten-dialect classification (Lǐ Róng 1989). These seven or ten dialects, however, must be understood as dialect groups, each of which may contain dozens of different mutually unintelligible dialects. Although the classification is a synchronic grouping, the criteria employed are historical. Phonological innovations are used as the main features: the devoicing pattern of Middle Chinese voiced obstruents (see Table 10.17), the change of Middle Chinese syllables with stop codas, and the change of Middle Table 10.17 The píng-zè voicing mergers with unvoiced syllables, as represented by p, ph, and b

b- píng 平 b- zè 仄 Pattern a

Mandarin

Wu

Gan

Xiang

Min

Kejiaa

Yue

ph p A

b b B

ph ph C

b b B

p p D

ph ph C

ph ph/p E

Kejia people are a sub-ethnic group of Han Chinese, the name for the people in Chinese is kèjiā 客家, which is pronounced hak-kâ by their linguistic neighbor, the speakers of the Yue dialect or the Cantonese dialect. So the Kejia dialect is often called Hakka in Western literature.

382

Toward Modern Mandarin

Table 10.18 A schematic representation of the geographical distribution of the devoicing patterns of the major dialect groups Northern Central Southern

Xiang Yue

Gan Kejia

Mandarin Wu Min

B E

C C

A B D

Chinese syllables with nasal codas are used to determine and distinguish the major dialectal groups. The p/ph split of Yue is conditioned by tones as well but is different from the Mandarin pattern. Instead the Middle Chinese voiced stops and affricates become voiceless aspirated in the píng and shǎng tones, but voiceless unaspirated in the qù and rù tones.The presentation of the devoicing pattern in Table 10.18 shows the relationship between the Wu and Xiang dialects, and between the Gan and Kejia dialects. The Ping, Jin, and Hui dialects are divided from Yue, Mandarin, and Wu, respectively, mainly based on major phonological features, such as the devoicing of Middle Chinese voiced obstruents and the change of the Middle Chinese rù tone. The Jin dialects preserve the rù tone, while the surrounding Mandarin dialects have lost this tonal distinction. The distinctive feature of the Wu dialects is that they preserve voiced obstruents, so there is a threeway contrast of stops and affricates: voiceless unaspirated, voiceless aspirated, and voiced. But the Hui dialects show a two-way contrast instead. According to Lǐ Róng (1989), Ping has a different devoicing pattern from Yue. In Ping, all the Middle Chinese voiced stops and affricates changed to voiceless unaspirated, unlike the Yue dialects, in which certain tonal conditions result in the Middle Chinese voiced stops and affricates becoming aspirated. 10.2.4.2 The Northern Dialects: Mandarin and Jin Some features of the Mandarin and Jin dialects can be traced to the post-Tang time. The phonological characteristics of Jin and some of the Mandarin dialects in the Northwest actually suggest different lineages from the Northern Mandarin represented by Beijing. Many of the similarities to Northern Mandarin were the result of strong influence from the standard phonology throughout history rather than a close genetic relationship. 10.2.4.3 The Central Dialects: Wu, Hui, Gan, and Xiang A Wu–Xiang continuum must have existed in pre-Tang times, before the massive migration that created a Gan–Kejia corridor. Certain linguistic features, such as syllabic nasals, occur in both dialect groups, and indicate an

The Qing Dynasty and the Modern Era

383

ancient dialect, which includes the geographical area of modern Wu, Gan, and Xiang. This old layer also shows typological similarity with the Miao language, such as the lack of coda, tone sandhi, syllabic nasals, and phonation types of initial consonants. In the seven-dialect system, the Hui dialect is considered a variety of Wu, with Hui having a different reflection of Middle Chinese phonation types. 10.2.4.4 The Southern Dialects: Min, Kejia, Yue, and Ping In the Southern Min dialect there are multiple phonological registers, which give insight into the historical phonology of Min. The colloquial and literary readings of the same characters have quite distinct forms. One interesting phenomenon is that the colloquial readings represent the local original pronunciation and the literary readings represent later externally introduced pronunciations. In a comparison of the coda systems of the two readings, colloquial readings do not make distinctions of place of articulation with regard to stops (-p, -t, -k) and nasals (-m, -n, -ŋ) in the coda position, but the literary system makes a clear distinction between them (based on Zhōu Chángjí and Ōuyáng 1998). Table 10.19 is a comparison of colloquial and literary pronunciations of Min Chinese, represented here by Xiamen pronunciations, revealing a difference in distinctions of the coda. Chronologically the colloquial system must have formed before the introduction of the literary system. But the sound change must be from -VC/-VN to -Vʔ/-Ṽ. This seems to be a paradox. The later change appears in the earlier historical layer and the earlier stage before undergoing change appears in the later historical stage. These seemingly reversed phonological layers in the Southern Min dialect provide an important clue to understanding the relationship between various historical layers of dialects. A number of phonological features of the Min dialect cannot be explained by the Middle Chinese phonology. This phenomenon indicates that the earliest layer of the Min pronunciation could have formed during the Han dynasty Table 10.19 A comparison of colloquial and literary pronunciations of Min Chinese

Middle Chinese Colloquial Min Literary Min Mandarin

-p

-t

-k

-m

-n



答 ‘answer’ təp taʔ tap ta

辣 ‘hot’ lɑt lwaʔ lat La

學 ‘study’ ɦɰɔk oʔ hak ɕɥɛ

焰 ‘flame’ jɛm jã iam jɛn

搬 ‘move’ pwɑn pwã pwan pan

平 ‘level’ bɰjaŋ pjã piŋ phiŋ

384

Toward Modern Mandarin

(Norman 1979), earlier than the time of the Middle Chinese represented by the Qièyùn 切韻. Kejia, Yue, and Ping represent different historical layers brought about by the major migrations of various time periods during the end of the Tang dynasty after the ninth century (Mài 2009). 10.2.4.5 The Nature of Dialect Formation Chinese dialects are historically formed through language contact. When China extended its territory, the Chinese language entered areas of nonChinese language, where the people gradually adopted the Chinese language and lost their own. Through such language shift, non-Chinese features were left in the Chinese they adopted. For example, similarities can be found between the Yue dialect and the Tai languages, including the existence of more than two rù tones, -p, -t, -k and -m, -n, -ŋ codas, and a contrast in vowel length. Such similarities result from the influence of Tai languages such as Zhuang on Chinese. After the language shift, non-Chinese people became Han Chinese ethnically and linguistically, but because this shift was commonly incomplete and imperfect, some linguistic material will remain in the Chinese they learned. This happens between Chinese and non-Chinese as well as between standard Chinese and Chinese dialects. However, Chinese historical phonologists and dialectologists have not really started their research from the perspective of language shift. It is to be hoped that future research into relevant linguistic phenomena from this new perspective will yield a better understanding of the formation of modern dialects, as well as historical phonological change. The history of Chinese is not only the history of the standard language: it extends vastly beyond what has been covered in this book. It is not difficult to understand that each dialect has its own history. Because of the lack of historical records, the formation and change of Chinese dialects are very difficult, sometimes impossible, to study. However, recent theories of language contact can provide a much better explanation of the formation of Chinese dialects. As pointed out by Sarah Thomason (2001), “language contact is everywhere: there is no evidence that any languages have developed in total isolation from other languages” (p. 8). She concludes that, “language contact is the norm, not the exception” (p. 10).

References

Australian Academy of the Humanities and the Chinese Academy of Social Sciences in collaboration with, and assisted by, the Department of Linguistics, the Research School of Pacific Studies, the Australian National University 1987. Language Atlas of China. Hong Kong: Longman. Baxter, William 1992. A Handbook of Old Chinese Phonology. Berlin: Mouton de Gruyter. Baxter, William and Laurent Sagart 1998. Word formation in Old Chinese, in Jerome L. Pachard (ed.) New Approaches to Chinese Word Formation, 35–76. (Trends in Linguistics: Studies and Monographs 105.) Berlin: Mouton de Gruyter. 2014. Old Chinese: A New Reconstruction. Oxford: Oxford University Press. Běijīng Dàxué 2003. Hànyǔ Fāngyīn Zìhuì. 2nd ed. Beijing: Yuwen Chubanshe. Benedict, Paul 1972. Sino-Tibetan: A Conspectus (James A. Matisoff, Contributing Editor). Cambridge: Cambridge University Press. Branner, David Prager (ed.) 2006. The Chinese Rime Tables: Linguistic Philosophy and Historical-Comparative Phonology. Amsterdam: John Benjamins. Cáo, Zhìyún (ed.) 2008. Hànyǔ Fāngyán Dìtú Jí. Beijing: Shangwu Yinshuguan. Chan, Abraham 2004. Early Middle Chinese: Towards a new paradigm, T’oung Pao 90: 122–162, 193–194. Chao, Yuen Ren 1928. Xiàndài Wúyǔ de Yánjiū. Beijing: Qinghua University. 1934. The non-uniqueness of phonemic solutions of phonetic systems, Bulletin of the Institute of History and Philology, Academia Sinica 4: 363–398. 1941. Distinctions within Ancient Chinese, Harvard Journal of Asiatic Studies 5: 203–233. 1948. Húběi Fāngyán Diàochá Bàogào. (Special publication.) Taipei: Institute of History and Philology, Academia Sinica. (1961) 1976. What is correct Chinese?, in Anwar S. Dil (ed.) Aspects of Chinese Sociolinguistics: Essays by Yuen Ren Chao, 72–83. Stanford: Stanford University Press. Chen, Matthew Y. 2000. Tone Sandhi: Patterns across Chinese Dialects. Cambridge: Cambridge University Press. Chen, Ping 1999. Modern Chinese: History and Sociolinguistics. New York: Cambridge University Press. Chén, Dúxiù 1937. Zhōngguó gǔdài yǔyīn yǒu fùshēngmǔ shuō, Dongfang Zazhi 34: 21–22. Chén, Qíguāng 1993. Lùn xiéshēng, in Mínzú Yǔwén Lùnwénjí. Beijing: Zhongyang Minzu Xueyuan Chubanshe. Also in Chén, Qíguāng 2007, 215–233. 2007. Lùnyǔ Shuōwén Jí. Beijing: Minzu Chubanshe. 385

386

References

Chén, Yínkè 1949. Cóng shǐshí lùn Qièyùn, Lingnan Xuebao 9: 1–18. Chén, Zhōngmǐn 2003. Wúyǔ jíqí línjìn fāngyán yúyùn de dúyīn céngcì: Jiānlùn Qièyùn yúyùn de yīnzhí, Yuyanxue Luncong 27: 11–55. Cheng, Chin Chuan and William S.-Y. Wang 1971. Phonological change of Middle Chinese initials, Tsing Hua Journal of Chinese Studies 9: 216–270. Chin, Hui-fang 2011. Lǎoguóyīn Yánjiū: Yǐ Wáng Pú Guójīngyīn Duìzhàobiǎo wéi Zhōngxīn. Master’s Thesis, National Pingtung University, Taiwan. Chinggeltai 2002. Qìdān Xiǎozì Shìdú Wèntí. Tokyo: Institute for the Study of Language and Culture of Asia and Africa. Chinggeltai, Liú, Fèngzhù, Chén Nǎixióng, Yú Bǎolín, and Xíng Fùlǐ 1985. Qìdān Xiǎozì Yánjiū. Beijing: Zhongguo Shehui Kexue Chubanshe. Chǔ, Sōngtài 2005. Táng-Wǔdài Guāngzhōng Fāngyīn Yánjiū. Hefei: Anhui Daxue Chubanshe. Coblin, W. South 1991. Studies in Old Northwest Chinese. (Journal of Chinese Linguistics, Monograph Series No. 4.) Berkeley: University of California Press. 2000. A brief history of Mandarin, Journal of American Oriental Society 120: 537–552. 2006. Reflections on the Shǒuwēn fragments, in David Prager Branner (ed.) The Chinese Rime Tables: Linguistic Philosophy and Historical-Comparative Phonology, 99–122. Amsterdam: John Benjamins. Dīng, Shēngshù and Lǐ Róng 1981. Gǔjīn Zìyīn Duìzhào Shǒucè. Beijing: Zhonghua Shuju. (Corrected version of an earlier edition from Beijing: Kēxué Chubanshe, 1960.) Dǒng, Tónghé (1944) 1948. Shànggǔ yīnyùn biǎogǎo, Bulletin of the Institute of History and Philology, Institute of Academia Sinica 18: 1–249. (Corrected reprint of original 1944 lithograph edition.) Dragunov, Alexander A. 1931. A Persian transcription of Ancient Mandarin, Bulletin de l’ Académie des Sciences de l’ URSS (Classe des sciences sociales) 3: 359–375. Duàn, Yùcái (1815) 1981. Shuōwén Jiězì Zhù. Shanghai: Shanghai Guji Chubanshe. Edkins, Joseph 1876. Introduction to the Study of the Chinese Characters. Hertford: Trübner & Co. Endō, Mitsuaki 1997. Ōshukuwa Myaku Ketsu perusha go yaku ni han-eishita 14 seiki sho Chūgoku on, in Anne O. Yue and Endō Mitsuaki (eds.) Hashimoto Mantarō Kinen Chūgoku Gogaku Ronshū, 61–77. Tokyo: Uchiyama Shoten. 2016. Gendai no Kenkyū: Myaku Ketsu Perusha Go Yaku ni Yoru. (Aoyama Gakuin Daigaku Keizai Kenkyūjo, Kenkyū Sōsho 8.) Tokyo: Kyūko Shoin. Franke, Herbert 1951. Some Sinological remarks on Rašid ad-Din’s History of China, Oriens 4: 21–26. Féng, Zhēng 1998. Lùn Qièyùn de fēnyùn yuánzé, àn zhǔyào yuányīn fēnyùn, bù àn jièyīn fēnyùn – Qièyùn yī-sān děng yùn, sān-sì děng yùn bùtóng yuányīn shuō, Yuyan Yanjiu (additional volume): 164–185. 2001. Lùn Qièyùn Zhǐzhǎngtú sān-sì děng duìlì zhōng de chóngniǔ hé zhǔn-chóngniǔ, Yuyan 2: 103–178. 2007. Lùn Ěryǎ Yīntú de yīnxì jīchǔ, in Gěng Zhènshēng (ed.) Jìndài Guānhuà Yǔyīn Yánjiū, 284–292. Beijing: Yuwen Chubanshe. Gāo, Xiǎohóng 2009. Běijīnghuà Rùshēngzì de Lìshǐ Céngcì. Beijing: Beijing Language and Culture University Press.

References

387

Gěng, Zhènshēng 1992. Míng-Qīng Děngyùnxué Tōnglùn. Beijing: Yuwen Chubanshe. 2007. Zài tán jìndài Guānhuà de ‘biāozhǔnyīn,’ Gu Hanyu Yanjiu 1: 16–22. Gong, Hwang-cherng 1981. Shí-èr shìjì mò Hànyǔ xīběi fāngyīn (shēngmǔ wèntí), Bulletin of the Institute of History and Philology, Academia Sinica 52: 37–78. Also in Gong 2004: 243–281. 1989. Shí-èr shìjì mò Hànyǔ xīběi fāngyīn (yùnwěi wèntí), in Dì-èr Jiè Guójì Hànxué Huìiyì Lùnwén Jí (Yǔyán yú wénzì zǔ), 145–190. Also in Gong 2004: 283–330. 1994. Shí-èr shìjì mò Hànyǔ xīběi fāngyīn yùnmǔ xìtǒng de gòunǐ. Paper read at the joint meeting of the 4th IACL and NACCL, 1995. Also in Gong 2004: 331–377. 2004. Hàn-Zàng Yǔ Yánjiū Lùnwén Jí. Beijing: Beijing Daxue Chubanshe. Harbsmeier, Christoph 2016. Irrefutable conjectures: A review of William H. Baxter and Laurent Sagart, Old Chinese: A New Reconstruction, Monumenta Serica: Journal of Oriental Studies 64: 445–504. Hashimoto, Mantaro 1970. Internal evidence for Ancient palatal endings in Mandarin, Language 46: 336–365. 1978–1979. Phonology of Ancient Chinese. 2 vols. (Study of Language and Cultures of Asia and Africa Monograph Series.) Tokyo: Japan Print Kamiya. Hattori, Shirō and Tōdō Akiyasu 1958. Chūgen On’in no Kenkyū. Tokyo: Kōnan Shoin. Haudricourt, André-Georges 1954a. Comment reconstruire le chinois archaïque, Word 10: 351–364. English translation by Guillaume Jacques: How to reconstruct Old Chinese, Non-final version (November 9, 2017). In preparation for: Haudricourt, André-Georges, Studies in the Evolution of Languages and Techniques. Edited by Martine Mazaudon, Boyd Michailovsky, and Alexis Michaud. (Trends in Linguistics: Studies and Monographs 270.) Berlin: Mouton de Gruyter. 1954b. De l’origine des tons en vietnamien, Journal Asiatique 242: 69–82. English translation by Marc Brunelle: The origin of tones in Vietnamese, Non-final version (January 8, 2018). In preparation for: Haudricourt, André-Georges, Studies in the Evolution of Languages and Techniques. Edited by Martine Mazaudon, Boyd Michailovsky, and Alexis Michaud. (Trends in Linguistics: Studies and Monographs 270.) Berlin: Mouton de Gruyter. 1961. Bipartition et tripartition des systèmes de tons dans quelques langues d’Extrême-Orient, Bulletin de La Société de Linguistique de Paris 56: 163–180. English translation by Christopher Court: Two-way and three-way splitting of tonal systems in some Far Eastern languages, in Jimmy G. Harris and Richard B. Noss (eds.) Tai Phonetics and Phonology, 58–86. Bangkok: Mahidol University, 1972. Hé, Jiǔyíng (1998) 2002. Shāngdài fùfǔyīn shēngmǔ, in Yīnyùn Cónggǎo, 1–15. Beijing: Shangwu Yinshuguan. Hirayama, Hisao 1991. Zhōnggǔ chúnyīn chóngniǔ zài Zhōngyuán Yīnyùn qī-wēi yùn lǐ de fǎnyìng, in Zhōngyuán Yīnyùn Xīn Lùn, 28–34. Beijing: Beijing Daxue Chubanshe. Ho, Dah-an 2016. Such errors could have been avoided, Review of Old Chinese: A New Reconstruction by William H. Baxter and Laurent Sagart, Journal of Chinese Linguistics 44: 175–230. Hsueh, F. S. 1975. Phonology of Old Mandarin. The Hague: Mouton.

388

References

Hú, Míngyáng 1963. Lǎo Qìdà Yánjiě he Piáo Tōngshì Yànjiě zhōng suǒjiàn de Hànyǔ, Cháoxiǎnyǔ duìyīn, Zhongguo Yuwen 3: 185–192. Huáng, Cuìbó (1970) 2010. Tángdài Guānzhōng Fāngyán Yīnxì. Beijing: Zhonghua Shuju. Huáng, Xiàoshān 2000. Fāngkuài Zhuàngzì de shēngpáng hé hànyǔ zhōnggǔ yùnmǔ, Zhōnggǔ Jìndài Hànyǔ Yánjiū 1: 22–46. 2002a. Qièyùn yuányīn fēnyùn de jiǎshè he yīnwèihuà gòunǐ, Gu Hanyu Yanjiu 3: 10–16. 2002b. Zhōnggǔ èrděng yùn jièyīn hé Qièyùn yuányīn shùliàng, Zhejiang Daxue Xuebao, Renwen Shehui Kexue Ban 32: 30–38. 2004. Qièyùn hé Zhōngtáng Wǔdài Yīnwèi Xìtǒng. Taipei: Wenjin Chubanshe. 2007. Qièyùn yùnmǔ xiǎozhù ‘tóng’ yú ‘bié,’ Yuyan Yanjiu Jikan 4: 143–160. Hucker, Charles O. 1985. A Dictionary of Official Titles in Imperial China. Stanford: Stanford University Press. Ishiyama, Fukuji 1925. Kōtei Chūgen On’in. Tokyo: Tōyō Bunko. Jaxontov, Sergej 1960. Consonant combinations in Archaic Chinese. (Papers presented by the USSR delegation at the 25th International Congress of Orientalists, Moscow.) Moscow: Oriental Literature Publishing House. Chinese translation in Táng and Hú 1986: 42–52. 1980. Shí-yī shìjì de Běijīng yīn. In Táng and Hú 1986: 187–196. Jì, Fú 1994. Wǔyīn Jíyùn yu děngyùnxué, Yinyunxue Yanjiu 3: 80–88. Jīn, Yǒujǐng 1982. Guānyú Zhèjiāng fāngyán zhōng xiánshè sānsìděng zì de fēnbié, Yuyan Yanjiu 1: 148–162. Kane, Daniel A. 2009. The Kitan Language and Script. Leiden: Brill. Karlgren, Bernhard 1915–1926. Études sur la phonologie chinoise. 4 vols. Uppsala: K. W. Appelberg. 1923. Analytic Dictionary of Chinese and Sino-Japanese. Paris: Paul Geuthner. 1933. Word families in Chinese, Bulletin of the Museum of Far Eastern Antiquities 5: 9–120. 1954. Compendium of phonetics in Ancient and Archaic Chinese, Bulletin of the Museum of Far Eastern Antiquities 26: 211–367. Legge, James 1893. The Chinese Classics: Vol. 1, Confucian Analects, 2nd ed. Oxford: Clarendon Press, p. 135. Li, Fang-Kuei 1971. Shànggǔ yīn yánjiū, Tsing Hua Journal of Chinese Studies 9: 1–61. Also in F-K. Li 1980: 1–83. 1980. Shànggǔ Yīn Yánjiū. Beijing: Shangwu Yinshuguan. Lǐ, Róng 1956. Qièyùn Yīnxì. Beijing: Kexue Chubanshe. 1985. Guānhuà fāngyán de fēnqū, Fangyan 1: 2–5. 1989. Hànyǔ fāngyán de fēnqū, Fangyan 4: 241–259. Lǐ, Rúlóng and Xīn Shìbiāo 1999. Jìnnán, Guānzhōng de ‘quánzuó sòngqì’ yu Táng-Sòng xīběi fāngyīn, Zhongguo Yuwen 3: 197–203. Lǐ, Xīnkuí 1962. Zhōngyuán Yīnyùn de xìngzhì jíqí dàibiǎo de yīnxì, Jianghua Xuebao 8: 39–43. 1983. Zhōngyuán Yīnyùn Yīnxì Yánjiū. Zhengzhou: Zhongzhou Shuhuashe. 1984. Chóngniǔ yánjiū, Yuyan Yanjiu 2: 73–104. Lín, Yǔtáng 1924. Gǔ yǒu fùfǔyīn shuō, Chen Bao, Liù Zhōunián Jìniàn Zēngkān, 206–216.

References

389

Liú, Fèngzhù 2014. Qìdān Wénzì Lèibiān. Beijing: Zhonghua Shuju. Liú, Guǎnghé 1984. Tángdài bā-shìjì Cháng’ān yīn shēngniǔ, Yuwen Yanjiu 3: 45–50. Liú, Shúxué 2000. Zhōnggǔ Rùshēngzì zài Héběi Fāngyán zhōng de Dúyīn Yánjiū. Baoding: Hebei University Press. Liú, Wénjǐn 1931. Hóngwǔ Zhèngyùn shēnglèi kǎo, Bulletin of the Institute of History and Philology, Academia Sinica 3: 237–247. Lóng, Yǔchún 1968. Táng Xiěquánběn Wáng Rénxù Kānmiù Bǔquē Qièyùn Jiàojiān. Hong Kong: Chinese University Press. Lǔ, Guóyáo 2007. Yánjiū Míngmò Qīngchū guānhuà jīchǔ fāngyán de niàn-èr nián lìchéng, in Gěng Zhènshēng (ed.) Jìndài Guānhuà Yǔyīn Yánjiū, 122–142. Beijing: Yuwen Chubanshe. Lù, Zhìwěi 1946a. Shì Zhōngyuán Yīnyùn, Yanjing Xuebao 31: 1–36. Also in Lù 1988: 1–34. 1946b. Jì Shào Yōng Huángjí Jīngshì de Tiānshēng Dìyīn, Yanjing Xuebao 31: 71–80. Also in Lù 1988: 35–44. 1947a. Jì Lán Mào Yùnluè Yìtōng. Yanjing Xuebao 32: 161–168. Also in Lù 1988: 45–53. 1947b. Jì Xú Xiào Chóngdìng Sīmǎ Wēngōng Děngyùn Tújīng, Yanjing Xuebao 32: 169–196. Also in Lù 1988: 54–84. 1947c. Jīn Nígé Xīrǔ Ěrmù Zī suǒ jì de yīn, Yanjing Xuebao 33: 115–128. Also in Lù 1988: 94–108. 1988. Jìndài Hànyǔ Yīnyùn Lùnjí. Beijing: Shangwu Yinshuguan. Luó, Chángpéi 1931. Zhī, chè, chéng, niáng yīnzhí kǎo, Bulletin of the Institute of History and Philology, Academia Sinica 3: 121–157. 1933. Táng-Wǔdài Xīběi Fāngyīn, Bulletin of the Institute of History and Philology, Academia Sinica, Monograph Series A 12. 2004. Luó Chángpéi Yǔyánxué Lùnwén Jí. Beijing: Shangwu Yinshuguan. Mǎ, Déqiáng 2012. Dāngqián tōngxíng de děngyùn tǐxì zhōng kāihékǒu de láiyuán kǎolùn, Yangzhou Daxue Xuebao, Renwen Shehui Kexue Ban 16(4): 116–128. Mǎ, Xuéliáng and Luó Jìguāng 1962. Wǒ guó Hàn-Zàng yǔxì yǔyán de yuányīn chángduǎn, Zhongguo Yuwen 4: 533– 539. Mahootian, Shahrzad 1997. Persian. London: Routledge. Mài, Yún 1991. Zhōngyuán Yīnyùn de shéjiān hòuyīn shēngmǔ bǔzhèng, in Zhōngyuán Yīnyùn Xīnlùn, 145–155. Beijing: Beijing Daxue Chubanshe. 1995. Qièyùn yuányīn xìtǒng shì nǐ, in Yīnyùn yú Fāngyán Yánjiū, 96–118. Guangzhou: Guangdong Renmin Chubanshe. 2009. Cóng yuèyǔ de chǎnshēng hé fāzhǎn kàn Hànyǔ fāngyán de xíngchéng de móshì, Fangyan 3: 219–232. Martin, Samuel E. 1953. The Phonemes of Ancient Chinese, Journal of the American Oriental Society 16 (supp.). Mei, Tsu-lin (Méi Zǔlín) 1970. Tones and prosody in Middle Chinese and the origin of the rising tone, Harvard Journal of Asiatic Studies 30: 86–110. 1977. Tones and tone sandhi in 16th century Mandarin, Journal of Chinese Linguistics 5: 237–260. 1980. Sìshēng biéyì zhōng de shíjiān céngcì, Zhongguo Yuwen 6: 427–433. 2008. Jiǎgǔwén lǐ de jǐgè fùfǔyīn shēngmǔ, Zhongguo Yuwen 3: 196–207. Nìng, Jìfú 1985. Zhōngyuán Yīnyùn Biǎogǎo. Jinlin: Jilin Wenshi Chubanshe.

390

References

1991. Shísì shìjì Dàdǔ fāngyán de wénbái yìdú, In Zhōngyuán Yīnyùn Xīnlùn, 35–43. Beijing: Beijing Daxue Chubanshe. 1992. Jiàodìng Wǔyīn Jíyùn. Beijing: Zhonghua Shuju. 1997. Gǔjīn Yùnhuì Jǔyào jí Xiāngguān Yùnshū. Beijing: Zhonghua Shuju. 2003. Hóngwǔ Zhèngyùn Yánjiū. Shanghai: Shanghai Guji Chubanshe. 2016. Hànyǔ Yùnshū Shǐ, Jīn-Yuán Piān, 34–35, 195–197. Shanghai: Shanghai Renmin Chubanshe. Norman, Jerry 1979. Chronological strata in the Min dialects, Fangyan 4: 268–274. 1988. Chinese. Cambridge: Cambridge University Press. 1994. Pharyngealization in Early Chinese, Journal of the American Oriental Society 114: 397–408. 1997. Some thoughts on the early development of Mandarin, in Anne O. Yue and Mistuaki Endo (eds.) In Memory of Mantaro Hashimoto, 21–28. Tokyo: Uchiyama Bookstore. Owen, Stephen 1996. An Anthology of Chinese Literature: Beginnings to 1911. New York: Norton. Paar, Francis W. (ed.) 1963. Qian Zi Wen: The Thousand Character Classic: A Chinese Primer. (Calligraphy in five scripts by Fong-chih Lui.) New York: Frederick Ungar. Pān, Wùyún 1982. Guānyú Hànyǔ shēngdiào fāzhǎn de jǐgè wèntí, Journal of Chinese Linguistics 10: 359–385. 1997. Hóuyīn kǎo, Minzu Yuwen, 5: 10–24. 1999. Hàn-zàng yǔ zhōng de cìyào yīnjié, in Pān Wùyún and Shí Fēng (eds.) Zhōngguó Yǔyánxué de Xīn Tuōzhǎn, 125–147. Hong Kong: City University of Hong Kong Press. 2000. Hànyǔ Lìshǐ Yīnyùnxué. Shanghai: Shanghai Jiaoyu Chubanshe. Pulleyblank, Edwin G. 1962–1963. The consonant system of Old Chinese, Asia Major 9: 58–144, 206–256. 1973. Some further evidence regarding Old Chinese –s and its time of disappearance, Bulletin of the School of Oriental and African Studies 36: 368–373. 1984. Middle Chinese: A Study in Historical Phonology. Vancouver: University of British Columbia Press. Qián, Nǎiróng 1992. Dāngdài Wúyǔ Yánjiū. Shanghai: Shanghai Jiaoyu Chubanshe. Qiáo, Quánshēng 2004. Xiàndài Jìn fāngyán yǔ Táng Wǔdai xīběi fāngyán de guanxi, Zhongguo Yuwen 3: 262–266. Ramsey, S. Robert 1987. The Languages of China. Princeton, NJ: Princeton University Press. Ruǎn, Tíngxián 2007. Cóng Hàn-Yuè-yǔ yánjiū zhìyí Hànyǔ Zhōnggǔ yǒu shémiàn yùnwěi, Zhongguo Yuwen 6: 554–557. Sagart, Laurent 1993a. Chinese and Austronesian: Evidence for a genetic relationship, Journal of Chinese Linguistics 21: 1–62. 1993b. Austronesian final consonants and the origin of Chinese tones, in Jerold A. Edmondson and Kenneth J. Gregerson (eds.) Tonality in Austronesian Languages, 47–59. (Oceanic Linguistics Special Publication No 24.) Honolulu: University of Hawaii Press. 1999. The Roots of Old Chinese. (Amsterdam Studies in the Theory and History of Linguistic Science. Series IV, Current Issues in Linguistic Theory 184.) Amsterdam: John Benjamins.

References

391

Schuessler, Axel 2009. Minimal Old Chinese and Later Han Chinese: A Companion to Grammata Serica Recensa. (ABC Chinese Dictionary Series.) Honolulu: University of Hawaii Press. Shào, Róngfēn 1982. Qièyùn Yánjiū. Beijing: Zhonghua Shuju. Shen, Zhongwei 2005. Cóng Ménggǔ Zìyùn lùn rùshēng yīnjié de fùyuányīn huà, in Yīnshǐ Xīnlùn, 310–324. Beijing: Xueyuan Chubanshe. 2006. Liáodài běifāng fāngyán de yǔyīn tèzhēng, Zhongguo Yuwen, 6: 483–498. 2007. Sino-Khitan phonology, Bulletin of Chinese Linguistics 1(2): 147–211. 2008a. Studies on the Menggu Ziyun. (Language and Linguistics, Monograph Series No. A-16.) Taipei: Institute of Linguistics, Academia Sinica. 2008b. Apicalization and its related changes before and after the Menggu Ziyun, Bulletin of Chinese Linguistics 2(2): 15–34. 2010. Tànsuǒ lìshǐ cáiliào suǒ fǎnyìng de zhèngzài fāshēng de yīn biàn, in Pan Wuyun and Shen Zhongwei (eds.) Yánjiū zhī Lè: Qìngzhù Wángshìyuán Xiānshēng Qīshí-wǔ Shòuchén Xuéshù Lùnwén Jí. Shanghai: Shanghai Jiaoyu Chubanshe. 2011. The origin of Mandarin, Journal of Chinese Linguistics 39: 1–31. 2012. The phonological characteristics of Northern Chinese of the Jin dynasty. Bulletin of Chinese Linguistics 6(2): 95–120. 2015. Ménggǔ Zìyùn Jíjiào. Beijing: Zhonghua Shuju. 2016a. Héngxiàng chuándì hé fāngyán xíngchéng, Journal of Chinese Linguistics, Monograph Series 26, 21–54. 2016b. Accented Mandarin of the early 14th century as seen in the Persian transcription, in Pang-hsin Ting, Samuel Hung-nin Cheung, Sze-Wing Tang and Andy Chin (eds.) New Horizons in the Study of Chinese: Dialectology, Grammar, and Philology, 599–620. Hong Kong: Institute of Chinese Studies, Chinese University of Hong Kong. Shěn, Jiānshì 1945. Guǎngyùn Shēngxì. Beijing: Furen Daxue. (Also Beijing: Wenzi Gaige Chubanshe, 1960.) Shī, Xiàngdōng 1983. Xuánzàng yīzhù zhōng de Fán-Hàn duìyīn hé Táng chū zhōngyuán fāngyīn, Yuyan Yanjiu 1: 27-48. Also in Shī 2009: 1–79 (full version). 2009. Yīn Shǐ Xún Yōu. Tianjin: Nankai Daxue Chubanshe. Starostin, George 2015. Book review: William H. Baxter, Laurent Sagart, Old Chinese: A New Reconstruction, Journal of Language Relationship 13: 383–389. Starostin, Sergej Anatol’evič (Sergai) 1989. Rekonstrukcija drevnekitajskoj fonologičeskoj sistemy, Moscow: “Nauka,” Glavnaja redakcija vostočnoj literatury. Chinese translation by Lín Hǎiyīng and Wáng Chōng, Gǔdài Hànyǔ Yīnxì de Gòunǐ. Shanghai: Shanghai Jiaoyu Chubanshe, 2010. 1995. Book review: Old Chinese Vocabulary: A Historical Perspective, Journal of Chinese Linguistics Monograph Series 8: 226–251. Stimson, Hugh M. 1966. The Jongyuan in Yunn: A Guide to Old Mandarin Pronunciation. New Haven, CT: Far Eastern Publications, Yale University. Sūn, Hóngkāi 1980. Ménbā, Luòbā, Dèngrén de Yǔyán. Beijing: Shehui Kexue Chubanshe. 1981. Qiāngyǔ Jiǎnzhì. Beijing: Minzu Chubanshe. 1982. Dúlóngyǔ Jiǎnzhì, Beijing: Minzu Chubanshe. Sūn, Jiànyuán 2010. Sìshēng Tōngjiě Jīnsúyīn Yánjiū. Beijing: Zhonghua Shuju.

392

References

Sūn, Jǐngtāo 2008. Gǔ Hànyǔ Chóngdié Gòucífǎ Yánjiū. Shanghai: Shanghai Jiaoyu Chubanshe. Sūn, Yízhì 2010. Cóng zhī, zhuāng, zhāng de fēnhé kàn Xīrǔ Ěrmù Zī yīnxì de xìngzhì, Zhongguo Yuwen 5: 438–450. Swadesh, Morris 1934. The phonemic principle, Language 10: 117–129. Sybesma, R. P. E., Wolfgang Behr, Yueguo Gu, Zev J. Handel, Cheng-Teh James Huang, and James Myers 2016. Encyclopedia of Chinese Language and Linguistics. 5 vols. Leiden: Brill. Takata, Tokio 1988. Tonkō Shiryō ni yoru Chūgokugo shi no kenkyū. Tokyo: Soubunsha. Táng, Zuòfān 1991. Zhōngyuán Yīnyùn de kāihé kǒu, in Zhōngyuán Yīnyùn Xīnlùn, 167–179. Beijing: Beijing Daxue Chubanshe. Táng Zuòfán and Hú Shuāngbǎo (eds.) 1986. Hànyǔ Shǐ Lùn Jí. Beijing: Beijing Daxue Chubanshe. Thomason, Sarah 2001. Language Contact. Washington, D.C.: Georgetown University Press. Tián, Yèzhèng 2004. Cóng Huánhuān yùn kàn Zhōngyuán Yīnyùn de jīchǔ fāng yán, Ankang Shizhuan Xuebao 61: 32–35, 45. University of Tehran 1972. Tahksūq-Nameh, Collected Works of Rashīd-al-Dīn Fadlallāh, Vol. 2 (with an introduction by Mojatabā Minovi). Tehran: University of Tehran. Wade, Thomas Francis 1867. Yü-yen Tzŭ-erh Chi, A Progressive Course Designed to Assist the Student of Colloquial Chinese, As Spoken in the Capital and the Metropolitan Department. London: Trübner. Wang, William S-Y. 1967. Competing changes as a cause of residue, Language 45: 9–25. 1979. Language change: A lexical perspective, Annual Review of Anthropology 9: 353–371. Wang, William S-Y. and Chin Chuan Cheng 1987. Middle Chinese tones in modern dialects, in R. Channon and L. Shockey (eds.) In Honor of Ilse Lehiste, 514–523. Dordrecht: Foris Publishers. Wāng, Róngbǎo 1923. Gē, gē, yú, yú, mó gǔdú kǎo, Guoxue Jikan 1: 241–263. Wāng, Wéihuī 2005. Cháoxiǎn Shídài Hànyǔ Jiàokéshū Cóngkān. Beijing: Zhonghua Shuju. Wáng, Hóngjūn 1999. Cóng kāikǒu yīděng chóngyùn de xiàndài fǎnyìng xíngshì kàn Hànyǔ fāngyán de lìshǐ guānxi, Yuyan Yanjiu 1: 61–75. Wáng, Lì 1948. Hàn-yuè yǔ yánjiū, Lingnan Xuebao 9: 1–96. 1956. Hànyǔ Yīnyùnxué. Beijing: Zhonghua Shuju. (1957) 2004. Hànyǔ Shǐ Gǎo, Vol. 1. Beijing: Zhonghua Shuju. (1958) 2005. Hànyǔ Shīlǜxué. Shanghai: Shanghai Jiaoyu Chubanshe. 1965. Gǔ Hànyǔ zìdòngcí he shǐdòngcí de pèiduì, Zhonghua Wenshi Luncong 6: 121–142. 1985. Hànyǔ Yīnyùn Shǐ. Beijing: Zhongguo Shehui Kexue Chubanshe. Wáng, Línhuì 2003. Fénhé Liúyù Fāngyán de Yǔyīn Tèdiǎn jíqí Liúbiàn. Beijing: Zhongguo Shehui Kexue Chubanshe. Wáng, Pú 1921. Guóyīn Jīngyīn Duìzhàobiǎo. Shanghai: Shangwu Yinshuguan. Wáng, Yáo 1979. Zàngwén. Minzu Yuwen 1: 71–75. Wáng, Yīdān 2006. Bōsī Lāshītè Shǐjí Zhōngguó Shǐ Yánjiū yú Wénběn Fānyì. Beijing: Kunlun Chubanshe.

References

393

Wáng, Zhàopéng 1998. Guǎngyùn ‘dúyòng,’ ‘tóngyòng’ shǐyòng niándài kǎo: Yǐ Tángdài kējǔ kǎoshì shīfù yòngyùn wéi lì, Zhongguo Yuwen 2: 144–147. 2004. Tángdài Kējǔ Kǎoshì Shīfù Yòngyùn Yánjiū. Jinan: Qilu Shushe. Wú, Zōngjì 1986. Hànyǔ Pǔtōnghuà Dān Yīnjié Yǔtú Cè. Beijing: Zhongguo Shehui Kexue Chubanshe. Xiào, Dōngfā 1990. Pèiwén Yùnfǔ, in Zhōngguó Dà Bǎikē Quánshū, Xīnwén Chūbǎn, 228–229. Beijing: Zhongguo Da Baike Quanshu Chubanshe. Xióng, Zhènghuī 1990. Guānhuà fāngyán fēn ts tʂ de lèixíng, Fangyan 1: 1–10. Xióng, Zhènghuī and Zhāng Zhènxīng 2008. Hànyǔ fāngyán de fēnqū, Fangyan 2: 97–108. Xǔ, Bǎohuá and Pān Wùyún 1994. Shì èrděng, Yinyunxue Yanjiu 3: 119–135. Yáng, Jūn 2007. Yùnjìng Jiàojiǎn. Hangzhou: Zhejiang Daxue Chubanshe. Yáng, Nàisī 1981. Zhōngyuán Yīnyùn Yīnxì. Beijing: Zhongguo Shehui Kexue Chubanshe. 1997. Yùnhuì, Qīyīn yú Ménggǔ Zìyùn, in Jìndài Hànyǔ Yīn Lùn, 129–145. Beijing: Shangwu Yinshuguan. (Originally published, 1989). Yáng, Yìmíng and Wáng Wèimín 2003. Yuányīn Zhèngkǎo yǔ Yīnyùn Féngyuán suǒ jì jiān-tuán-yīn fēnhé zhī bǐjiào yánjiū. Zhongguo Yuwen 2: 131–146. Yóu, Rǔjié 1992. Hànyǔ Fāngyán Xué Dǎolùn. Shanghai: Shanghai Jiaoyu Chubanshe. Yú, Mǐn 1984a. Hòu-Hàn Sānguó Fán-Hàn duìyīn pǔ, in Zhōngguó Yǔyánxué Lùnwén Xuǎn, 269–319. Tokyo: Guansheng Guan. 1984b. Běijīng yīnxì de chéngzhǎng hé tā shòu de zhōuwéi yíngxiǎng, Fangyan 4: 272–277. Yú, Nǎiyǒng 1993. Zài lùn Qièyùn yīn: shì nèi-wài zhuǎn xīn shuō, Yuyan Yanjiu 2: 33–48. Yùchí, Zhìpíng 1986. Rìběn xītánjiā suǒ chuán gǔ Hànyǔ shēngdiào, Yuyan Yanjiu 2: 17–35. 1990. Lǎo Qìdà Piáo Tōngshì Yànjiě Hànzì yīn de yǔyīn jīchǔ, Yuyan Yanjiu 1: 11–24. 2007. Yùshǎng zhīyīn, fēiguǎng wénlù, Yuyan Yanjiu Jikan 4: 161–186. Zhāng, Yùshū, Chén Tíngjìng, Zhā Shìshēng et al. (1711) 1983. Yùdìng Pèiwén Yùnfǔ. Shanghai: Shanghai Guji Shudian. Zhāng, Wèidōng 2002. Cóng Yǔyán Zìěr Jí kàn bǎiniánlái Běijīngyīn de yǎnbiàn, Guangdong Waiyu Waimao Daxue Xuebao 4: 15–23. Zhāng, Xiǎomàn 2005. Sìshēng Tōngjiě Yánjiū. Jinan: Qilu Shushe. Zhāng, Yùlái 1999. Yùnluè Yìtōng Yánjiū. Tianjin: Tianjin Guji Chubanshe. Zhào, Yìntáng 1936. Zhōngyuán Yīnyùn Yánjiū. Shanghai: Shangwu Yinshuguan. 1957. Děngyùn Yuánliú. Beijing: Shangwu Yinshuguan. Zhèng, Jīnshēng. 2002. Hǎiwài Huíguī Zhōngyī Shànběn Gǔjí Cóngshū, Vol. 1. Beijing: Renmin Weisheng Chubanshe. Zhèng, Wěi 2018. Fāngkuài Zhuàngzì de hànzì jièyīn shēngpáng yǔ zhōnggǔ yùntú de nèiwàizhuǎn, Gu Hanyu Yanjiu 1: 25–34. Zhèngzhāng, Shàngfāng 1981. Shànggǔ yīnxì biǎojiě. Read at the First Annual Conference of the Linguistics Society of Zhejiang. Also in Yuyan 4: 17–46, 2003. 1984. Shànggǔ yīn gòunǐ xiǎoyì, Yuyanxue Luncong 14: 36–49. 1987. Shànggǔ yùnmǔ xìtǒng hé sìděng, jièyīn, shēngdiào de fāyuán wèntí, Wenzhou Shiyuan Xuebao 4: 67–90.

394

References

1990. Shànggǔ Hànyǔ de s-tóu, Wenzhou Shiyuan Xuebao 4: 10–19. 1994. Hànyǔ shēngdiào píngzè fēn yù shǎngshēng, qùshēng de qǐyuán, Yuyan Yanjiu (additional volume). 2001. Táng-Fán Huìméng Bēi Zàng-Hàn duìyīn lǐ xiàfù xiǎo a (ɦ) de yǔyīn yìyì, Minzu Yuwen 1: 19–21. (2003) 2013. Shànggǔ Yīnxì. Shanghai: Shanghai Jiaoyu Chubanshe. 2014. Zhèjiāng Wēnzhōu fāngyán de sìshēng bādiào lèixíng jí liándiào, qīngshēng hé yǔfǎ biàndiào, Fangyan 3: 215–220. Zhōngguó Shèhuì Kēxuéyuàn Yǔyán Yánjiūsuǒ 1981. Fāngyán Diàochá Zìbiǎo. Beijing: Shangwu Yinshuguan (Corrected version of an earlier edition from Beijing: Shehui Kexue Chubanshe, 1955.) Zhōngyuán Yīnyùn Xīnlùn Biānjí Zǔ 1991. Zhōngyuán Yīnyùn Xīnlùn. Beijing: Beijing Daxue Chubanshe. Zhōu, Chángjí and Ōuyáng Yìyún 1998. Xiàmén Fāngyán Yánjiū. Fuzhou: Fujian Renmin Chubanshe. Zhōu, Fǎgāo (Chou Fa-Kao) 1956. Fójiào dōngchuán duì Zhōngguó yīnyùnxué zhī yǐngxiǎng, in Zhōngguó Yǔwén Lùncóng, 21–51. Taipei: Zhengzhong Shuju. 1962. Zhōngguó Gǔdài Yǔfǎ: Gòucífǎ. (Special Publication 39.) Taipei: Institute of History and Philology, Academia Sinica. 1970. Lùn Shànggǔ yīn hé Qièyùn, Journal of the Institute of Chinese Studies of the Chinese University of Hong Kong 3: 321–457. Zhōu, Zǔmó 1942. Sòngdài Biànluò yǔyīn kǎo, Furen Xuezhi 12: 221–285. Also in Zhōu 1966c: 581–655. 1963. Qièyùn de xìngzhì hé tāde yīnxì jīchǔ, Yuyanxue Luncong 5: 39–70. Also in Zhōu 1966c: 434–473. 1966a. Sìshēng biéyì shìlì, in Wèn Xué Jí, 81–119. Beijing: Zhonghua Shuju. (Originally published, 1945). 1966b. Guānyú Tángdài fāngyán zhōng sìshēng dúfǎ de yīxiē zīliào, in Wèn Xué Jí, 494–500. Beijing: Zhonghua Shuju. (Originally published, 1958). 1966c. Wèn Xué Jí. 2 vols. Beijing: Zhonghua Shuju. 1983. Táng Wǔdài Yùnshū Jícún. Beijing: Zhonghua Shuju.

Index

abugida, 235 allophone, 99, 346, 352 alphabetic system, 229, 334 Altaic language, 195, 207 alternative readings, 71 Ancient Persian script, 294 apical vowels, 47, 187, 258, 354 apicalization, 258, 332, 354–355 jīng-apicalization, 258, 282 zhāng-apicalization, 258, 282 zhī-apicalization, 259, 282 zhuāng-apicalization, 258, 282 apico-alvéolarires, 258 apico-gingivales, 258 Austronesian, 61 Bai language, 59 bǎiyuè 百越, 5 base dialect, 3 Qièyùn. See single system or composite system Shēngyīn Chànghè Tú 聲音唱和圖, 169 Xīrú Ěrmù Zī, 317, 320, 342 Zhōngyuán Yīnyùn 中原音韻, 255 Baxter and Sagart’s reconstruction of Old Chinese, 69, 96, 99–100 Baxter, William, 76, 82, 89 Beijing, 3, 49, 196, 229, 296, 309, 317 dialect, 169, 308, 318, 334, 350, 360, 369, 378 Mandarin, 179, 207 běiyīn 北音, 368 Běnwù 本悟, 329, 331–332 biàn qīngzhuó 辨清濁, 189 Bùkōng 不空, 159 causative form, 98 centralized vowel, 177, 301, 358 Ch’oe Se-jin 崔世珍, 317, 320, 334 Chang’an, 152, 155, 162 Chang’an dialect, 113

Changsha dialect, 148 Chao, Yuen Ren, 109, 183, 378, 380 Chaozhou dialect, 149 Chén Lǐ 陳澧, 18, 131 Chén Péngnián 陳彭年, 4, 105, 166 Chén Yínkè 陳寅恪, 112 Chinese loanword, 197, 200 chǐtóu yīn 齒頭音 ‘alveolar affricates and fricatives’, 27 chǐyīn 齒音 ‘alveolars,’ 22 chóngyùn 重韻, 9, 36, 209 chóngniǔ 重紐, 9, 36–37 chóngniǔ contrast, 141, 254, 287 Ménggǔ Zìyùn, 251, 253 modern dialects, 148 chóngniǔ syllables, 23, 33, 37, 50, 77, 124 Gǎibìng Wǔyīn Jíyùn, 210 Ménggǔ Zìyùn, 229 Middle Chinese, 127, 132, 142 Zhōngyuán Yīnyùn, 287 chóngyùn contrast, 148 Chǔcí 楚辭, 91 chúnyīn 脣音 ‘labials’, 22 cìqīng 次清, 22, 29 cìzhuó 次濁, 22, 29 Cl- and Cr- consonant clusters, 72 cluster initial, 70 coda, 68 Coda Compensation Rule, 256 codas, 10 Hóngwǔ Zhèngyùn, 320 Ménggǔ Zìyùn, 255, 284 Middle Chinese, 134, 147 of rùshēng yùn, 48 of yángshēng yùn, 48 of yīnshēng yùn, 48 Old Chinese, 78, 90 rhyme groups, 46 colloquial language, 106, 195, 261 colloquial reading, 83, 149, 182, 225, 377, 383 comparative method, 4, 6, 61

395

396

Index

complementary distribution, 31 Ěryǎ Yīntú, 172 finals in Xīrú Ěrmù Zī, 345 hP’ags-pa spellings, 237, 243, 278 _ initials in rhyme tables, 23 initials in Yü-yen tzǔ-êrh chi, 372 initials in Zhōngyuán Yīnyùn, 291 phonetic vs. phonemic, 120 rhymes of Middle Chinese, 126 composite system, 112, 114–115 contour. See tonal contour contrast of Rank-III and Rank-IV, 150 cù 促, 20 cuōkǒu 撮口, 38 Dàchéng Zhōngzōng Jiànjiě 大乘中宗見解, 165 Dadu, 229, 317 Dadu dialect, 241, 264–265, 268 Dài Zhèn 戴震, 78 děng 等. See division division and rank, 30 Děngyùn Qièyīn Zhǐnán 等韻切音指南, 36 Děngyùn Tújīng 等韻圖經, 268, 332, 350 děngyùnxué 等韻學, 192 devoicing of Middle Chinese voiced obstruents dialect classification, 382 Jurchen transcriptions, 213 Khitan transcriptions, 201 Shēngyīn Chànghè Tú, 170 Tangut transcriptions, 218 Zhōngyuán Yīnyùn, 269, 279 devoicing pattern comparison of northern and northwestern dialect, 227 Ěryǎ Yīntú, 106, 164 Guānhuà, 320 Khitan transcriptions, 202 modern dialects, 382 modern northwestern dialects, 226 northwestern dialect, 225 Persian transcriptions, 296 Ping dialect, 382 Shēngyīn Chànghè Tú, 171, 180 Tangut transcriptions, 218, 225 dialect classification, 381 historical criteria, 381 dialectal feature, 314 Dīng Dù 丁度, 105, 168 Dīng Shēngshù 丁聲樹, 8 diphthongization Guānhuà, 313 Hóngwǔ Zhèngyùn Yìxùn, 326 Jurchen script, 211–212 Khitan script, 200–201, 206

Mài Jué, 308, 311 Ménggǔ Zìyùn, 255 Shēngyīn Chànghè Tú, 175 xiè 蟹 and zhǐ 止 rhyme groups, 183 Zhūgōngdiào, 213 division, 9, 23, 30–31, 34, 66 Dǒng Xièyuán 董解元, 212 Dong-Tai language family, 85 Dragunov, Alexander A, 304 Duàn Yùcái 段玉裁, 63, 78 duìyǔ 對語, 366 dúlì èr děng yùn 獨立二等韻, 39 Dunhuang, 159 dúruò 讀若, 20 dúyòng 獨用, 106, 154, 208 Early Middle Chinese, 106, 152, 162 Eastern Han, 86 Eastern Zhou, 59 Edkins, Joseph, 71 emergence of ɚ, 358 Emperor Kangxi 康熙, 362 Endō, Mitsuaki, 304, 307, 311 Ěryǎ Yīntú 爾雅音圖, 106, 163, 171 Études sur la phonologie chinoise, 36, 55, 258 family tree hypothesis,