Prosodic Phonology of the Fuzhou Dialect: Domains and Rule Application [1 ed.] 0367199483, 9780367199487

Citation preview

Prosodic Phonology of the Fuzhou Dialect

Prosodic Phonology of the Fuzhou Dialect: Domains and Rule Application is the first attempt to conduct a comprehensive analysis of the Fuzhou phonological system from the perspective of prosodic phonology. It addresses the following issues: what prosodic constituents exist in the Fuzhou dialect and what kinds of roles they play in the Fuzhou phonological system; how to define the domain formation of these prosodic constituents in the Fuzhou dialect; what kinds of Fuzhou phonological phenomena make crucial reference to these prosodic constituents as the domain of application; and what implications does the study of the Fuzhou phonological system have for the prosodic phonology theory. This book is a valuable text for students and scholars in the field of Chinese dialectology, Min dialects, prosodic phonology, and phonology-morphosyntax interface. Shuxiang You is Research Assistant Professor in the Department of Linguistics and Modern Languages at the Chinese University of Hong Kong. He received his M.A. and Ph.D. in Chinese Linguistics from the University of ­Wisconsin–Madison in 2013 and 2017 respectively. His research interests include phonology, phonology-syntax interface, Chinese dialectology, and teaching Chinese as a second language.

Routledge Studies in Chinese Linguistics Series editor: Hongming Zhang

Mandarin Loanwords Tae Eun Kim Intensification and Modal Necessity in Mandarin Chinese Jiun-Shiung Wu The Architecture of Periphery in Chinese Cartography and Minimalism Victor Pan Focus Manifestation in Mandarin Chinese and Cantonese A Comparative Perspective Peppina Po-lun Lee Prominence and Locality in Grammar The Syntax and Semantics of Wh-Questions and Reflexives Jianhua Hu Prosodic Studies Challenges and Prospects Edited by Hongming Zhang and Youyong Qian Diachronic Perspectives and Synchronic Variation in Southern Min Edited by Chinfa Lien and Alain Peyraube Prosodic Phonology of the Fuzhou Dialect Domains and Rule Application Shuxiang You For more information about this series, please visit: www.routledge.com/ languages/series/RSICL

Prosodic Phonology of the Fuzhou Dialect Domains and Rule Application Shuxiang You

First published 2020 by Routledge 2 Park Square, Milton Park, Abingdon, Oxon OX14 4RN and by Routledge 52 Vanderbilt Avenue, New York, NY 10017 Routledge is an imprint of the Taylor & Francis Group, an informa business © 2020 Shuxiang You The right of Shuxiang You to be identified as author of this work has been asserted by him in accordance with sections 77 and 78 of the Copyright, Designs and Patents Act 1988. All rights reserved. No part of this book may be reprinted or reproduced or utilised in any form or by any electronic, mechanical, or other means, now known or hereafter invented, including photocopying and recording, or in any information storage or retrieval system, without permission in writing from the publishers. Trademark notice: Product or corporate names may be trademarks or registered trademarks, and are used only for identification and explanation without intent to infringe. British Library Cataloguing-in-Publication Data A catalogue record for this book is available from the British Library Library of Congress Cataloging-in-Publication Data Names: You, Shuxiang, author. Title: Prosodic phonology of the Fuzhou dialect : domains and rule application / Shuxiang You. Description: New York : Routledge, 2020. | Series: Routledge studies in Chinese linguistics | Includes bibliographical references and index. | Summary: “Prosodic Phonology of the Fuzhou Dialect: Domains and Rule Application is the first attempt to conduct a comprehensive analysis of the Fuzhou phonological system from the perspective of prosodic phonology. It addresses the following issues: What prosodic constituents exist in the Fuzhou dialect and what kinds of roles they play in the Fuzhou phonological system; how to define the domain formation of these prosodic constituents in the Fuzhou dialect; what kinds of Fuzhou phonological phenomena make crucial reference to these prosodic constituents as the domain of application; and what implications does the study of the Fuzhou phonological system have for the prosodic phonology theory. This book is a valuable text for students and scholars in the field of Chinese dialectology, Min dialects, prosodic phonology, and phonology-morphosyntax interface”— Provided by publisher. Identifiers: LCCN 2019051724 (print) | LCCN 2019051725 (ebook) | ISBN 9780367199487 (hardback) | ISBN 9780429244292 (ebook) Subjects: LCSH: Chinese language—Dialects—China—Fuzhou Shi (Fujian Sheng)—Phonology. | Northern Min dialects—China—Fuzhou Shi (Fujian Sheng)—Phonology. | Chinese language—Phonology. | Chinese language—Versification. Classification: LCC PL1690.F8 Y68 2020 (print) | LCC PL1690.F8 (ebook) | DDC 495.17—dc23 LC record available at https://lccn.loc.gov/2019051724 LC ebook record available at https://lccn.loc.gov/2019051725 ISBN: 978-0-367-19948-7 (hbk) ISBN: 978-0-429-24429-2 (ebk) Typeset in Times New Roman by Apex CoVantage, LLC

To My Beloved Yanwen

Contents

List of illustrations Acknowledgements Abbreviations

xi xii xiv

1 Introduction 1.1 Statement of purpose  1 1.2 Research data  2 1.2.1 Published materials on the Fuzhou dialect  2 1.2.2 Data collection 3 1.3 Descriptive background on Fuzhou phonology  3 1.3.1 Tones in the Fuzhou dialect  3 1.3.2 Finals in the Fuzhou dialect  5 1.3.3 Initials in the Fuzhou dialect  7 1.3.4 Structure of Fuzhou syllables  8 1.3.5 Phonological rules in the Fuzhou dialect  9 1.3.6 Summary 16 1.4 Theoretical background: the prosodic phonology theory 17 1.4.1 The development of the prosodic phonology theory 17 1.4.2 Basic tenets of the prosodic phonology theory  19 1.4.3 Outstanding issues in the prosodic phonology theory 23 1.5 Research structure and organization  34

1

2 Syllable and foot in the Fuzhou dialect 2.1 Syllable: the domain of FA in the Fuzhou dialect  38 2.2 Foot as a domain of rule application in Fuzhou: a good choice or not?  43

38

viii  Contents 2.2.1 Introduction: the foot as a prosodic domain across languages 43 2.2.2 Previous studies on the foot in the Fuzhou dialect  46 2.2.3 Additional arguments against the foot as a domain of rule application in the Fuzhou dialect  57 2.2.4 Summary 61 2.3 Summary 62 3 Prosodic word: domain of lexical-level rule application in the Fuzhou dialect 3.1 Introduction 63 3.1.1 Definition of the prosodic word and its domain formation 63 3.1.2 The prosodic word as a prosodic domain across languages 65 3.2 Morphosyntactic word formation in the Fuzhou dialect  67 3.2.1 Monomorphemic words 67 3.2.2 Derived words 68 3.2.3 Compounds 71 3.2.4 Reduplication: “diminutive” nouns  72 3.2.5 Reduplication: sound-splitting words  73 3.2.6 Reduplication: reduplicated adjectives  74 3.2.7 Summary 75 3.3 Fuzhou morphosyntactic words and lexical-level phonological rule application  75 3.3.1 Application of TS in Fuzhou morphosyntactic words  75 3.3.2 Application of MTS in Fuzhou morphosyntactic words 80 3.3.3 Application of FC in Fuzhou morphosyntactic words 82 3.3.4 Application of CL in Fuzhou morphosyntactic words 85 3.3.5 Summary 88 3.4 Lexical-level phonological rule application and the prosodic word domain in the Fuzhou dialect  89 4 Enclitics and the Type A clitic group in the Fuzhou dialect 4.1 Introduction 93 4.1.1 Definition of clitics and common properties of clitics  93 4.1.2 Definition of the clitic group and evidence for the clitic group domain across languages  95

63

93

Contents  ix 4.2 Enclitics in the Fuzhou dialect and their morphosyntactic functions 98 4.2.1 Possessive/modificational/nominalization marker 其 [ki0] 99 4.2.2 Adjective reduplication markers 势 [siɛ213], 式 [seiʔ 23], and 喏 [luoʔ 23] 100 4.2.3 Aspect markers 100 4.2.4 Post-verbal particles 104 4.2.5 Locative marker 𠲥 [lɛ0] 106 4.2.6 Recursive clitic group with enclitics  106 4.2.7 Summary 107 4.3 Phonological phenomena within the Type A clitic group in the Fuzhou dialect  107 4.3.1 TS within the Type A clitic group  107 4.3.2 FC within the Type A clitic group  113 4.3.3 CL within the Type A clitic group  116 4.3.4 Summary 120 4.4 Violation of SLH within the Type A clitic group  121 4.5 Summary 123 5 Phonological phrase: domain of phrasal-level rule application in the Fuzhou dialect 5.1 Application of TS at the phrasal level  125 5.1.1 The nature of the problem  125 5.1.2 Previous analyses 128 5.1.3 Which is a better solution: RBA or EBA?  147 5.1.4 An alternative approach to phonological phrasing in the Fuzhou dialect  154 5.1.5 Summary 177 5.2 Application of FC at the phrasal level  177 5.3 Application of CL at the phrasal level  180 5.4 Summary 185 6 Proclitics and the Type B clitic group in the Fuzhou dialect 6.1 Proclitics in the Fuzhou dialect and their morphosyntactic functions 188 6.1.1 Progressive aspect marker 𠲥 [lɛ31] 189 6.1.2 Auxiliary verbs 是 [sei242] and 有 [ʔou242] 189 6.1.3 Prepositions 将 [tsuoŋ44] and 跟 [kyŋ44] 190 6.1.4 Recursive clitic group with proclitics  191 6.1.5 Summary 191

125

188

x  Contents 6.2 Phonological phenomena within the Type B clitic group in the Fuzhou dialect  192 6.2.1 TS within the Type B clitic group  192 6.2.2 FC within the Type B clitic group  194 6.2.3 CL within the Type B clitic group  196 6.2.4 Summary and discussion  199 6.3 Violation of SLH within the Type B clitic group  201 6.4 Summary 204 7 Intonational phrase in the Fuzhou dialect: variability in domain formation and rule application 7.1 Definition and restructuring of the intonational phrase in the Fuzhou dialect  206 7.1.1 Definition of the intonational phrase in the Fuzhou dialect 206 7.1.2 Restructuring of the intonational phrase in the Fuzhou dialect  209 7.1.3 Summary 213 7.2 Rule application within the intonational phrase in the Fuzhou dialect  213 7.2.1 Introduction: variability in ι-domain rule application across languages  213 7.2.2 Blocking of phonological rules across ι boundaries in the Fuzhou dialect  216 7.2.3 Optional rule application within intonational phrase in the Fuzhou dialect  220 7.2.4 Summary 223 7.3 Summary 224 8 Concluding remarks 8.1 Prosodic domains and rule application in the Fuzhou dialect  225 8.2 Theoretical issues regarding prosodic phonology  228 I The weakened Strict Layer Hypothesis  228 II The status of the clitic group  229 III Approaches to phonological phrasing  229 IV The Domain Impenetrability Condition  230 8.3 Future research  231 References Index

206

225

232 244

Illustrations

Figures 1.1 1.2 1.3 1.4 3.1 8.1

Prosodic hierarchy Prosodic hierarchy (expanded) Trisected model for prosodic hierarchy Prosodic hierarchy in Mandarin Chinese Lexical phonology in the Fuzhou dialect Prosodic hierarchy in the Fuzhou dialect

21 21 22 25 82 226

Tables 1.1 1.2 1.3 1.4 1.5 1.6 1.7 1.8 2.1 3.1 3.2 4.1 5.1 6.1 6.2 8.1

Fuzhou citation tones Fuzhou finals Fuzhou alternating finals Fuzhou initials Disyllabic TS in the Fuzhou dialect Trisyllabic TS in the Fuzhou dialect MTS in the Fuzhou dialect CL in the Fuzhou dialect Fuzhou alternating finals MTS in the Fuzhou dialect (refined) Phonological rules and morphosyntactic words in the Fuzhou dialect Phonological rules and the Type A clitic group in the Fuzhou dialect Phonological rules and phonological phrase in the Fuzhou dialect Phonological rules and the Type B clitic group in the Fuzhou dialect Phonological rules and morphosyntax-based domains in the Fuzhou dialect Prosodic domains and rule application in the Fuzhou dialect

4 6 7 8 10 11 13 15 40 81 89 121 186 199 200 227

Acknowledgements

This book is a revised and extended version of my Ph.D. dissertation submitted to the University of Wisconsin–Madison in 2017. It would not have been possible to give birth to this book without all the guidance, support, and encouragement from many people. Most immediately, I would like to express my greatest gratitude to my Ph.D. advisor, Professor Hongming Zhang, who has given me constant support, guidance, and inspiration over the years. He offered me various opportunities to get professional training in Chinese linguistics, and there would never be any way for me to become a linguist and complete this book without his valuable expertise, insightful suggestions, and devoted encouragement. I would like to extend my sincere appreciation to my Ph.D. dissertation committee members, Professor Yafei Li, Professor Eric Raimy, Professor William Nienhauser, and Professor Weihua Zhu, for their intellectual help and compassionate support throughout the dissertation project. Many of their constructive comments and invaluable suggestions I received at the defense of my dissertation have been integrated in this book. I am also obliged to Professor Marlys Macken, who retired before my defense. She not only helped me build up my phonological knowledge, but also taught me how to conduct linguistic fieldwork, which serves as the foundation of my dissertation. Since the completion of my dissertation, the contents of some chapters have been presented at several conferences. My gratitude goes to the audiences for their questions and feedback. I am very much grateful to the Routledge press for granting permission to reproduce material in my paper “Enclitics and the clitic group consisting of ‘host + enclitic’ in the Fuzhou dialect”, which was included in the book Prosodic Studies: Challenges and Prospects edited by Professor Hongming Zhang and Professor Youyong Qian in 2019. I  am also deeply grateful to John Benjamins Publishing Company for permission to reprint material in my Chinese language paper “论福州方言的粘附成分与粘附组 [Clitics and the clitic group in the Fuzhou dialect]”, which was published in the International Journal of Chinese Linguistics (Volume 5, Number 1) in 2018. Many thanks go to my colleagues in the Department of Linguistics and Modern Languages at the Chinese University of Hong Kong, for their kind encouragement and support. I wish to give my special thanks to Professor Haihua Pan, Professor Wai-Lan Gladys Tang, and colleagues of the General Office, who helped me

Acknowledgements  xiii balance my research and teaching responsibilities. I  also owe a special debt to Professor Victor Junnan Pan, who helped me with the analysis of some syntactic data in this book. I would like to take this opportunity to thank my three informants of the Fuzhou dialect, who generously helped me with the elicitation of the Fuzhou data. I thank them for their patience and support during my fieldwork in 2016 and their suggestions and comments since 2015. During the writing of this book, I  have also benefitted enormously from my friends. I  am particularly grateful to Professor Youyong Qian for sharing his own experience in turning a dissertation into a book. I would also like to express exceptional gratitude to Mengying Li, Mengyang Wei, and Miaocai Yan for reading the manuscript and offering helpful comments. Thanks also go to Professor Tae-Eum Kim, Dr. Xiang Lü, Professor Chenqing Song, Professor Yuxia Yin, and Professor Hui Yu, for their continuous support and encouragement. I am deeply indebted to the editors at Routledge, Ellie Auton, Andrea Hartill, Claire Margerison, and Geraldine Martin, for their professional advice and kind help. It was a great pleasure working with them. I am also grateful to Marie Roberts at Apex CoVantage for her kind assistance. I would also like to give my sincere thanks to the two anonymous reviewers of this book, whose constructive suggestions have helped improve the quality of this research. Lastly or rather firstly, I would like to express my deepest love to my family members, especially my wife and my parents. From the dissertation stage through to the completion of this book, they have offered me unending love, encouragement, and support, which was crucial in making the publication of this book possible.

Abbreviations

A AP AdjR Adv, AdvP C CG Cl, ClP CL CRS DIC DLM D, DP DUR DM EBA EXP FA FC FFR G HDC I/INFL, IP IC IPh LOC m M MOD MSR MTS N, NP NOM NOP

adjective/argument adjectival phrase adjective reduplication marker adverb, adverbial phrase consonant/clitic clitic group classifier, classifier phrase Initial Consonant Lenition currently relevant state Domain Impenetrability Condition delimitative aspect marker determiner, determiner phrase durative aspect marker duple meter Edge/End-Based Approach experiential aspect marker Final Alternation Final Change Foot Formation Rule glide Revised Head Dominance Condition inflection, inflectional phrase immediate constituent intonational phrase locative marker mora modifier modificational marker Main Stress Rule Morphological Tone Sandhi noun, noun phrase nominalization marker null operator

Abbreviations  xv Num, NumP O OT P, PP PASS PERF POSS PPh PRO PROG PVP PW Q, QP RBA s S SLH SR SSA t tg T T’ TP TS UR UTAH Utt v V VP w WFR X/X0, XP X’ ι μ σ Σ Σ’ υ φ ω

numeral, number phrase object Optimality Theory preposition, prepositional phrase passive morpheme perfective aspect marker possessive marker phonological phrase empty pronominal element progressive aspect marker post-verbal particle prosodic word/phonological word quantifier, quantifier phrase Relation-Based Approach strong sentence/clause Strict Layer Hypothesis surface representation Stray Syllable Adjunction trace of moved element tone group citation tone/tense sandhi tone tense phrase Phonological Tone Sandhi underlying representation Uniformity of Theta Assignment Hypothesis utterance light verb vowel/verb verb phrase weak word formation rule syntactic head of type X, maximal projection of type X intermediate syntactic phrase of type X intonational phrase mora syllable foot super-foot utterance phonological phrase prosodic word/phonological word

1 Introduction

1.1  Statement of purpose As the title suggests, this book deals with the phonological system of the Fuzhou dialect within the theoretical framework of prosodic phonology. The Fuzhou dialect, also known as the Foochow dialect or Fuzhounese, is the representative dialect of the Mindong (Eastern Min) dialect group of Chinese. Like many other southern dialects of Chinese, Fuzhou has a complex phonological system. The complexity of Fuzhou phonology partly lies in the fact that sound changes may occur to the initials, finals, and tones, of all the participating syllables in a string of sounds. Fuzhou exhibits one of the most intricate tone sandhi systems among the Chinese dialects. It has a large number of tone sandhi patterns, and a non-terminal tone in a given domain is usually changed into a sandhi tone depending on its original tonal value and that of the following tone. Along with the change in the tone, the final of a syllable may change to another final, which is affected by the tonal value the syllable assumes. Besides tones and finals, initial consonants are also often changed in Fuzhou according to the finals of their preceding syllables. In addition to the sound changes, complicated interactions between Fuzhou phonology and other components of the grammar, such as morphology, syntax, pragmatics, and semantics, make the phonological features of the Fuzhou dialect even more elusive. It has been noticed that there are two major types of tone sandhi rules in Fuzhou and that these two types target different groups of morphosyntactic words. It has also been recognized that sound changes occurring to the initials, finals, and tones may take place within domains of different sizes or levels—these sound changes not only have an effect on syllables within a lexical item, but also operate across word boundaries. Moreover, performance and semantic factors, e.g., length, rate of speech, style, and contrastive prominence, may affect the sound changes as well. The particular features of Fuzhou phonology outlined above are worthy of exploration on their own merits, which have, in addition, broader significance in helping to bring about a deeper understanding of human languages in general. Ever since the Fuzhou dialect was first systematically recorded in the 16th century, a considerable amount of previous studies, in both Chinese scholarship and English scholarship, have been published on its phonological system. After several centuries’ study, however, we have yet to obtain a full picture of Fuzhou phonology,

2  Introduction which makes it still a great challenge to linguists. Therefore, a more comprehensive work conducted within the framework of modern phonological theories is needed, and this book is such an attempt to fill this void. In this book, the aforementioned Fuzhou phonological phenomena are studied using the theory of prosodic phonology, which deals with the domains of phonological phenomena as well as the interactions between phonology and other components of the grammar. The purpose of this book, therefore, is two-fold. On the one hand, it is an attempt to provide a formal analysis of the phonological rule system of the Fuzhou dialect. On the basis of published materials and my own data collection, I explore major phonological rules in Fuzhou with respect to their domains of application. By providing an analysis of Fuzhou phonological rules from the perspective of prosodic phonology, I hope this study can capture the defining Fuzhou phonological characteristics, especially the application of different rules in different domains as well as the interactions between Fuzhou phonology and other components of the grammar. On the other hand, I hope to enhance our understanding of prosodic phonology in general through this study. Since it was developed in the 1980s, a great deal of research has been conducted from many perspectives to improve this theory. Based on the description of Fuzhou phonological phenomena, I define different prosodic domains and discuss the roles they play in the application of phonological rules in this dialect. Phonological, morphological, syntactic, semantic, and pragmatic information in Fuzhou will all be involved in the discussion of prosodic domains. This book is thus expected to provide empirical evidence for the prosodic phonology theory from the Fuzhou dialect, deepening our understanding of phonology and the relationship between phonology and other components of the grammar. This book addresses to two major groups of readers, those in the field of Chinese dialectology with varying degrees of knowledge about modern phonological theories and those in theoretical linguistics with varying degrees of knowledge about Chinese dialects. With its focus on the domains of rule application in the Fuzhou dialect and the interactions between Fuzhou phonology and other components of the grammar, this book by no means exhausts the intriguing topics related to the phonological system of this dialect, nor does it try to resolve all the controversies surrounding the theory of prosodic phonology. Other phonological phenomena in Fuzhou, such as mechanisms of tone sandhi, historical development of the entire phonological system, and so forth, would equally catch the attention of Chinese dialectologists and general linguists. These phenomena, as well as theoretical issues beyond the compass of this book, thus await future investigation.

1.2  Research data 1.2.1  Published materials on the Fuzhou dialect Data used in this book partly come from published materials, including textbooks, dictionaries, local records, and monographs on the Fuzhou dialect. It should be noted that early sources of the Fuzhou dialect, such as Maclay & Baldwin (1870, 1898, 1929); Baldwin (1871, 1909); Parker (1879a, 1879b, 1881a, 1881b); Adam

Introduction  3 (1891, 1905); Tao (1930); Gao (1947); and Lan (1953), among others, do not provide IPA transcription of sounds and thus are not consulted in this study. More­ over, these early sources only present the polysyllabic forms in the Fuzhou dialect in the way that each syllable in the polysyllabic forms is given in its citation form, with no sound changes in initials, finals, or tones indicated. Therefore, in order to obtain a full picture of Fuzhou phonology, more attention is paid to materials that present the data in the sandhi form in various phonological environments, for example, the dictionaries compiled by Chen & Norman (1965b), Li et al. (1994), Feng (1998), and Li (1998).1 In addition to these dictionaries, I  also consult Chen & Norman’s (1965a) textbook, Chen’s (1998) and Li’s (2002) monographs, and Li & Liang’s (2001) local record of the Fuzhou dialect, for more data, which not only provide data in phonological environments, but also provide those in syntactic environments. Apart from the aforementioned main sources, additional published materials containing raw data from the Fuzhou dialect, which are used in this study, include but are not limited to, Chen & Li’s (1991) monograph; Li et al. (1979) article; Liang’s (1982, 1983a, 1983b, 1986) articles; Zheng’s (1983, 1985, 1988a, 1988b, 1995) articles; Peking University’s cross-dialectal syllabary (1962, 1989) and vocabulary (1964, 1995); and Lin’s (2002) book on the morphology and syntax in Eastern Min dialects, as well as Wright’s (1983) and Chan’s (1985) dissertations. 1.2.2  Data collection Besides the published materials mentioned above, additional information was collected from three native speakers of the Fuzhou dialect. These three informants are aged 50–85 and were all born and raised in the Gulou District of the city of Fuzhou. They can only speak Mandarin Chinese and the Fuzhou dialect, and they mainly speak the dialect at home as their daily language. Moreover, none of them has experience of living or studying in other places for an extended period. All of them are literate, and their speech is quite similar. The first-hand data from these three informants was collected during my fieldwork in Fuzhou in the summer and the winter of 2016.

1.3  Descriptive background on Fuzhou phonology Before we move on to the investigation into the Fuzhou phonological system within the framework of prosodic phonology, it is necessary to present a brief description of Fuzhou phonology. The Fuzhou phonological system will be introduced in the following order: tones, finals, initials, syllable structure, and phonological rules. 1.3.1  Tones in the Fuzhou dialect There are seven citation tones in the Fuzhou dialect. Arranged according to the traditional tonal categorization of ping 平 ‘level’, shang 上 ‘rising’,

4  Introduction Table 1.1  Fuzhou citation tones ping

shang

qu

ru

yin

44

31

213

23

yang

51

242

5

Source: Cf. Li et al. 1994; Yuan 2001, among others. Note: Following Chinese tradition, here tones are represented as points along a five-point scale, a notation based on that of Chao (1930). Digits indicate the pitch value, 5 being the highest and 1 the lowest.

qu 去 ‘falling’, and ru 入 ‘entering’, these seven citation tones are presented in Table 1.1. The subsequent tonal division of yin 阴 and yang 阳 based on the historical voiceless-voiced distinction of the initial consonants is also presented. Tonal values of these seven citation tones assumed in this book are presented in the shaded boxes. As can be seen from Table 1.1, yinping is a high-level tone. Its tonal value is reported as 44 in some published materials (see Chen & Li 1991; Li et al. 1994; Li 2002; Lin 2002, among others), while as 55 in some other materials (see Chen & Norman 1965a; Chen 1998; Feng 1998, among others). Data from my informants suggest that the yinping tone in the Fuzhou dialect sounds lower than its counterpart in Mandarin Chinese (with the tonal value usually marked as 55). This is why I assume that the tonal value of the yinping tone in Fuzhou is 44 instead of 55. Unlike its counterpart in Mandarin Chinese, which is a mid-rising tone, yangping in Fuzhou is a high falling tone. It is presented as 51 or 52 in some works (see Chen  & Norman 1965a; Chen  & Li 1991; Li 2002, among others) but is recorded as 53 in others (see Li et al. 1994; Chen 1998; Feng 1998; Lin 2002, among others). According to data from my informants, this tone is quite similar to the fourth tone in Mandarin Chinese (usually marked as 51). Hence, I mark the Fuzhou yangping tone as 51. The tonal value of the shang tone in Fuzhou is also a matter of controversy in the literature. It is usually recorded as a mid-falling tone with the tonal value 31 (see Chen & Li 1991; Li et al. 1994; Lin 2002, among others) or a mid-level tone with the tonal value 33 (see Chen 1998; Feng 1998; Li 2002, among others).2 It is also presented as 22 in some works (e.g., Chen & Norman 1965a). Data elicited from my informants suggest that this is a mid-falling tone 31 since the ending of this tone is as low as the ending of yangping in pitch. Yinqu is a dipping tone in the Fuzhou dialect. The tonal value of yinqu fluctuates. In some cases, it is 213 with a tonal contour similar to the third tone in Mandarin Chinese. Yet in other cases, the ending is apparently lower than 3, and thus the tonal value seems to be 212. In this book, I mark yinqu as 213, following the transcription in most published sources (see Chen & Li 1991; Li et al. 1994; Chen 1998, among others) as well as my informants’ judgment. Another contour tone in Fuzhou is yangqu, a rising-falling tone. According to the transcription in most published materials (see Chen & Li 1991; Li et al. 1994; Chen 1998; Feng 1998; Li 2002, among others) as well as data from my informants, the tonal value of yangqu is recorded as 242 in this book.

Introduction 5 There are two ru tones in the Fuzhou dialect, namely yinru and yangru. Both yinru and yangru are accompanied by a glottal stop consonant coda [-ʔ] in modern Fuzhou, making the sound short and abrupt. Yinru has a slight rise and thus is usually recorded as 23 (see Chen & Li 1991; Li et al. 1994; Li 2002, among others) or 24 (see Chen 1998; Feng 1998, among others) in the literature. In order to distinguish yinru from the half yangqu (usually marked as 24), which occurs only in a sandhi context, I mark yinru as 23 in this book. It should be noted that, although a syllable with the yinru tone is usually presented with a glottal stop coda [-ʔ] in the literature, words with a yinru tone have two different historical sources. In history, these words end with two kinds of stop codas, namely *[-ʔ] and *[-k], which are completely neutralized in isolation forms in modern Fuzhou dialect. In a sandhi context, however, these two historical sources result in two types of sandhi forms of the yinru tone (see Section 1.3.5.1) and have different effects on the initial consonant of the following syllable (see Section  1.3.5.5). For the sake of brevity, in this book, syllables with the yinru tone are all marked with the glottal stop [-ʔ]. Only when necessary, the ending [-k] will be marked. Compared with yinru, the yangru tone is even shorter in duration. Since it has a high pitch, it is marked as 5 in this study, as recorded in most of the literature (see Li et al. 1994; Chen 1998; Feng 1998, among others). Examples of these seven citation tones in the Fuzhou dialect are given in (1). (1)

yinping yangping shang yinqu yangqu yinru yangru

单 谈 胆 旦 淡 答 达

taŋ44 taŋ51 taŋ31 taŋ213 taŋ242 taʔ23 taʔ5

‘single; alone’ ‘to discuss’ ‘gall; guts’ ‘dawn’ ‘light’ ‘to answer’ ‘to reach’

菇 糊 鼓 故 旧 骨 掘

ku44 ku51 ku31 kou213 kou242 kouʔ23 kuʔ5

‘mushroom’ ‘to paste’ ‘drum’ ‘still’ ‘old’ ‘bone’ ‘to dig’

In addition to these seven citation tones, there are two sandhi tones in the Fuzhou dialect, whose tonal values are marked as 21 and 24. Since the tonal values of these two sandhi tones look like a half of the yinqu and yangqu tones respectively, they are called half yinqu and half yangqu. These two tones only occur in the sandhi context (to be discussed in more detail in Section 1.3.5.1). 1.3.2  Finals in the Fuzhou dialect The Fuzhou dialect has 47 finals, as presented in Table 1.2. It should be pointed out that some scholars claim that there is a syllabic nasal in the Fuzhou dialect, namely [n̩] (and its variants [m̩] and [ŋ̩]) and that it should be treated as another final (Feng 1998; Yuan 1960, 2001). This syllabic nasal, however, is not considered as a final in this study for two reasons. First, in the Fuzhou dialect, there are very few words that contain this final. Even in Feng’s (1998) dictionary, there is only one word consisting of this final, namely the

6  Introduction Table 1.2 Fuzhou finals a

ɛ

ia ua

iɛ

o

œ

iau eu iu

uo yo

aŋ iaŋ uaŋ

au

ai 3

uai ui

iɛŋ

aʔ iaʔ uaʔ

uoŋ yoŋ oʔ

ɛʔ iɛʔ

i/ei u/ou y/øy

øy/oy

iŋ/eiŋ uŋ/ouŋ yŋ/øyŋ

eiŋ/aiŋ ouŋ/auŋ øyŋ/oyŋ

iʔ/eiʔ uʔ/ouʔ yʔ/øyʔ

eiʔ/aiʔ ouʔ/auʔ øyʔ/oyʔ

œʔ

uoʔ yoʔ

Source: Cf. Chan 1985; Chen 1998; Li 2002, among others. Note: Variants of an alternating final are separated by a slash ‘/’ and are treated as only one final.

negative particle 伓 ‘not’. Second, in fact, this negative particle cannot be used alone, and its pronunciation is always conditioned by the initial of the following syllable. The syllabic nasals [n], ̩ [m], ̩ and [ŋ] ̩ are all its sandhi forms, while the citation form of this negative particle is assumed to be [iŋ242] (see Chen 1998; Li & Liang 2001, among others). Therefore, it is not treated as an independent final in this study. It is also noteworthy that there are two types of finals in the Fuzhou dialect, namely non-alternating finals (listed on the left of the thick line in Table 1.2) and alternating finals (listed on the right of the thick line). Non-alternating finals refer to finals whose citation forms are the same for all tones, as exemplified in (2a), while alternating finals refer to those whose two variants alternate depending on the tone the syllable assumes, as exemplified in (2b). (2)

a.

单 谈 胆 旦 淡 答 达

taŋ44 taŋ51 taŋ31 taŋ213 taŋ242 taʔ23 taʔ5

‘single; alone’ ‘to discuss’ ‘gall; guts’ ‘dawn’ ‘light’ ‘to answer’ ‘to reach’

b.

菇 糊 鼓 故 旧 骨 掘

ku44 ku51 ku31 kou213 kou242 kouʔ23 kuʔ5

‘mushroom’ ‘to paste’ ‘drum’ ‘still’ ‘old’ ‘bone’ ‘to dig’

As can be seen from (2b), in isolation forms, the variant [u] of the final [u/ou] can only occur with tones 44, 51, 31, and 5, while the other variant [ou] can only occur with tones 213, 242, and 23. Variants of Fuzhou alternating finals are thus divided into two groups according to the tones they are accompanied by in the citation environment. The first group is called jinyun 紧韵 (lit. ‘tense finals’) or benyun 本韵 (lit. ‘original finals’) in the literature, which occurs with tones 44,

Introduction  7 Table 1.3  Fuzhou alternating finals4

Alternating Finals

Tones

Group A

Group B

i iŋ iʔ u uŋ uʔ y yŋ yʔ 44, 51, 31, 5

ei eiŋ eiʔ ou ouŋ ouʔ øy øyŋ øyʔ 213, 242, 23

Group A eiŋ eiʔ ouŋ ouʔ øy øyŋ øyʔ 44, 51, 31, 5

Group B aiŋ aiʔ auŋ auʔ oy oyŋ oyʔ 213, 242, 23

Source: Cf. Chan 1985; Li et al. 1994; Chen 1998; Feng 1998; Li 2002; Lin 2002, among others.

51, 31, and 5, while the second group is called songyun 松韵 (lit. ‘lax finals’) or bianyun 变韵 (lit. ‘changed finals’), which occurs with tones 213, 242, and 23 (cf. Chan 1985; Li et al. 1994; Chen 1998; Feng 1998; Li & Liang 2001, among others). For the sake of brevity, these two groups will be referred to as Group A variants and Group B variants respectively in this book. Thus, Fuzhou alternating finals can be presented as in Table 1.3. 1.3.3  Initials in the Fuzhou dialect It is generally accepted that the Fuzhou dialect has 15 initials, including the socalled “zero initial” (see Li et al. 1994; Feng 1998; Li & Liang 2001, among others), as presented in Table 1.4, in which voiceless initials are listed on the left in the cells, while voiced ones are listed on the right. It is noteworthy that the initials [l] and [n] are considered as only one initial in some works (e.g., Chen 1998; Li 2002). It has been reported that some Fuzhou native speakers are not able to distinguish [l] from [n] (see Chen 1998; Yuan 2001; Li 2002, among others). In the initial inventory assumed in this study, I still treat these two as different initials since they have different sandhi forms in a sandhi context (see Section 1.3.5.5) and my informants can distinguish them in citation forms. It should be pointed out that the zero initial is generally marked as [0] or [Ø] in the literature of the Fuzhou dialect (e.g., Li et al. 1994; Chen 1998). It has also been noticed that a syllable with the zero initial is actually pronounced with a slight glottal stop [ʔ-] at the beginning of the syllable (see Feng 1998; Li 2002, among others). Also, the zero initial is usually realized as the glottal stop [ʔ-] in actual speech, especially when preceded by a syllable ending with a glottal stop. Therefore, the zero initial is indicated as [ʔ-] in most examples in this book, especially in examples in the sandhi context.

8  Introduction Table 1.4  Fuzhou initials Labial Stop Affricate

unaspirated aspirated unaspirated aspirated

p ph

Nasal Fricative Lateral approximant

Alveolar t th ts tsh

m

Velar

Glottal

k kh

n s

ʔ/0

ŋ x

l

Source: Cf. Li et al. 1994; Feng 1998; Li & Liang 2001, among others.

Examples of Fuzhou initials are listed as in (3). (3)

[p]: [ph]: [t]: [th]: [k]: [kh]: [ts]: [tsh]: [s]: [m]: [n]: [ŋ]: [l]: [x]: [ʔ/0]:

边 皮 单 贪 家 看 真 出 船 问 日 眼 柳 湖 黄

peiŋ44 phui51 taŋ44 thaŋ44 ka44 khaŋ213 tsiŋ44 tshouʔ23 suŋ51 muoŋ213 niʔ5 ŋaŋ31 liu31 xu51 ʔuoŋ51/uoŋ51

‘edge’ ‘skin’ ‘single; alone’ ‘greedy’ ‘family’ ‘to look’ ‘true’ ‘to go out’ ‘ship’ ‘to ask’ ‘day’ ‘eye’ ‘willow’ ‘lake’ ‘yellow’

饭 波 答 听 鸡 气 书 七 成 蜜 难 语 蓝 喜 雨

puoŋ242 pho44 taʔ23 thiaŋ213 kiɛ44 khei213 tsy44 tsheiʔ23 siaŋ51 miʔ5 naŋ242 ŋy31 laŋ51 xi31 ʔy31/y31

‘rice’ ‘wave’ ‘to answer’ ‘to listen’ ‘chicken’ ‘air’ ‘book’ ‘seven’ ‘to succeed’ ‘honey’ ‘difficult’ ‘language’ ‘blue’ ‘happy’ ‘rain’

In addition to these 15 initials, there are two sandhi initials in the Fuzhou dialect, [β] and [ʒ]. These two initials only occur in the sandhi context (see Section 1.3.5.5). 1.3.4  Structure of Fuzhou syllables The structure of a Fuzhou syllable is (C)(G)V(V/C/VC) (C  =  consonant; V = vowel; G = glide; ‘/’ = ‘or’), in which only the nuclear vowel V is required while the other segments are all optional. The major difference between the structure of Fuzhou syllables and that of Mandarin syllables lies in the final, and more specifically, in the syllable ending position. The syllable ending in Mandarin Chinese can be either a V (i.e., [i] or [u]) or a nasal C (i.e., [n] or [ŋ]), but the combination of V + C in the ending position is not allowed. In contrast, Fuzhou allows for a sequence of two segments after the nuclear vowel V. The syllable ending in Fuzhou thus can be a V ([i], [u], or [y]) such as CVV in (4i) and GVV in (4l),

Introduction  9 a C ([ʔ] or [ŋ]) such as VC in (4e) and GVC in (4k), or a V + C combination such as VVC in (4g) and CVVC in (4j). It is noteworthy that characters with the GVVC and CGVVC syllable structure do not exist in this dialect. All the possible types of Fuzhou syllables are listed in (4).5 (4)

Syllable types a. V b. CV c. GV d. CGV e. VC f. VV g. VVC h. CVC i. CVV j. CVVC k. GVC l. GVV m. CGVC n. CGVV

Examples 雨 ʔy31/y31 菇 ku44 窝 ʔuo31/uo31 鸡 kiɛ44 安 ʔaŋ44/aŋ44 有 ʔou242/ou242 红 ʔøyŋ51/øyŋ51 单 taŋ44 故 kou213 七 tsheiʔ23 黄 ʔuoŋ51/uoŋ51 歪 ʔuai44/uai44 成 siaŋ51 我 ŋuai31

‘rain’ ‘mushroom’ ‘to curl up’ ‘chicken’ ‘safe, peaceful’ ‘to have’ ‘red’ ‘single; alone’ ‘still’ ‘seven’ ‘yellow’ ‘to recline’ ‘to succeed’ ‘I’

1.3.5  Phonological rules in the Fuzhou dialect Now, let us have an overview of the most outstanding phonological rules in the Fuzhou dialect that are relevant to the topics under discussion in this book. A brief description of each rule will be presented with a number of examples. Since the primary concern in this book is to identify the domains of application for various phonological rules in Fuzhou, as well as to evaluate the role of each prosodic domain in the application of these rules, no special attention is given to the best formalism of each rule. 1.3.5.1  Phonological Tone Sandhi (TS) It has long been noticed that tones in the Fuzhou dialect, in general, maintain their citation tonal values only when occurring on monosyllabic syllables or on the final syllable in a given domain (see Chen & Norman 1965a; Chan 1980; Li et al. 1994; Donohue 2013, among others). When a tone is followed by another tone in a domain that contains more than one syllable, the non-terminal tone usually undergoes tone sandhi. There are two types of tone sandhi rules in the Fuzhou dialect: the first type applies in most cases as long as the phonological environment is satisfied, that is, there are two or more syllables within the domain, while the second type only applies to some disyllabic words formed by the morphological process of reduplication. The first type of tone sandhi rule is referred to as Phonological Tone Sandhi (henceforth TS) in this study, while the second type is referred to as Morphological Tone Sandhi (henceforth MTS) (see Section 1.3.5.2).

10  Introduction Table 1.5  Disyllabic TS in the Fuzhou dialect Sandhi form of the 1st tone 1st tone 44

2nd tone

44

213 242 23(-ʔ) 31 23 (-k) 51

5

51

5

44 21 44

31

213

242

23

51 24 31

44 21

Source: Cf. Li et al. 1994; Feng 1998; Lin 2002, among others.

As noted by almost all the published materials on the Fuzhou dialect, the TS rule is a context-sensitive rule, according to which a citation tone will change into a sandhi tone depending on its original tonal value as well as that of the following tone (see Chen & Norman 1965a; Chan 1980; Li et al. 1994; Donohue 2013, among others). Phonological Tone Sandhi patterns for a disyllabic domain are given in Table 1.5. Sandhi tones are presented in shaded boxes. From Table 1.5, we can find that two sandhi tones, namely 21 and 24, which do not occur in isolation, appear in the sandhi context. It is also noteworthy that the yinru tone 23 has two types of sandhi patterns. As mentioned in Section 1.3.1, words bearing the yinru tone have two different historical sources, ending with *[-ʔ] and *[-k] respectively. In a sandhi context, these two different endings give rise to different tone sandhi patterns of the yinru tone. In a tone sandhi environment, the yinru tone with the historical ending *[-ʔ] behaves like the yinping, yinqu, and yangqu tones, while the yinru tone with the historical ending *[-k] behaves like the shang tone. Disyllabic tone sandhi examples of these two types of yinru tone are given in (5a) and (5b) respectively. In the examples in (5) and all the following examples in this book, citation/underlying forms are presented before the symbol “→”, while sandhi/surface forms are presented after “→”. Tones in question are marked in bold in (5). Other phonological processes are not presented here. (5) a. Tone sandhi of yinru (-ʔ) 桌骹 toʔ23 kha44 → 桌头 toʔ23 thau51 → 壁里 piaʔ23 liɛ31 → 脚色 kyoʔ23 saiʔ23 → b. Tone sandhi of yinru (-k) 福州 xuʔ23 tsiu44 → 骨头 kouʔ23 thau51 → 出口 tshouʔ23 khau31 → 拾刷 khaʔ23 souʔ23 →

toʔ44 kha44 toʔ44 thau51 piaʔ51 liɛ31 kyoʔ51 saiʔ23

‘table leg’ ‘top of the table’ ‘the against-the-wall side of bed’ ‘ability’

xuʔ21 tsiu44 kouʔ21 thau51 tshouʔ24 khau31 khaʔ44 souʔ23

‘Fuzhou’ ‘bone’ ‘exit’ ‘to tidy up; to fix; to punish’

Introduction  11 Table 1.6  Trisyllabic TS in the Fuzhou dialect T1 + T2 + T3

Tone sandhi of T1 + T2 + T3

Type 1: T2 = 51/5

T1 + (T2 + T3)→T1 + T2’ + T3→(T1 + T2’) + T3→T1’ + T2’ + T3 T1 + (T2 + T3)→T1 + T2’ + T3→21 + T2’ + T3

Type 2: T2 = 44/213/242/23/31

Source: Cf. Chan 1985; Li et al. 1994; Hung 1987, among others.

In a domain composed of three syllables, as reported in most literature (e.g., Chan 1985; Li et al. 1994; Hung 1987, among others), the pivotal point for TS is the citation tone of the penultimate syllable. If the citation tone of the penultimate syllable is yangping 51 or yangru 5, this tone will undergo TS according to the disyllabic TS patterns in Table 1.5. And then the antepenultimate tone will undergo TS depending on the sandhi tone (rather than the citation tone) of the penultimate syllable and its own citation tone. By contrast, if the citation tone of the penultimate syllable is neither yangping 51 nor yangru 5, this tone will undergo TS, and then the tone of the antepenultimate syllable will become the half yinqu tone 21 by default. Phonological Tone Sandhi patterns for a trisyllabic domain are given in Table  1.6. Sandhi patterns are presented in shaded boxes, where T’ indicates the sandhi form of T. Tones in the parentheses undergo the disyllabic TS. If T2’ is 21/24, the tone sandhi of (T1 + T2’) follows the disyllabic TS patterns of (T1 + 213/242). These two types of trisyllabic TS patterns can be illustrated in (6). (6) Type 1: 媒侬妈 番石榴 Type 2: 图书馆 开斗裤

mui51 nøyŋ51 ma31 → mui31 nøyŋ31 ma31 ‘female matchmaker’ xuaŋ44 suoʔ5 liu51 → xuaŋ51 suoʔ31 liu51 ‘pomegranate’ ‘library’ tu51 tsy44 kuaŋ31 → tu21 tsy51 kuaŋ31 khui44 tau31 khou213 → khui21 tau44 khou213 ‘open-seat pants (for babies)’

In a sandhi domain consisting of more than three syllables, the tones of syllables preceding the antepenultimate are uniformly changed into 21, and the rest of the syllables follow the trisyllabic TS patterns in Table 1.6 (see Chen & Norman 1965a; Chan 1985, 1998, among others). It has been noticed that the TS rule not only applies to lexical items like those listed in (5) and (6), but also applies to a string of sounds at the phrasal level (Chen & Norman 1965a; Chan 1985; Shih 1986; Zhang 1992, among others), as exemplified in (7), where the application of TS in question is indicated by “=” between syllables.

12  Introduction (7)

a. → c. →

饭 食 5 siɛʔ puoŋ242 21 siɛʔ = puoŋ242 eat rice ‘to eat food’ 煎 鱼尾 tsiɛŋ44 ŋy51 mui31 51 tsiɛŋ = ŋy31 mui31 fry fish tail ‘to fry fish tails’

b. → d. →

野 俊 ʔia31 tsouŋ213 44 ʔia = tsouŋ213 very beautiful ‘very beautiful’ 好 队员 xo31 tui242 ʔuoŋ51 21 xo = tui44 ʔuoŋ51 good team member ‘good team member’

In the phrasal-level examples in (8), however, the TS rule is somehow blocked. Sandhi tones in question are marked in bold. The symbol “#” denotes the position where the blocking of the TS rule occurs. (8)

a. →

悬 跳 野 thiu213 ʔia31 keiŋ51 h 213 21 t iu # ʔia keiŋ51 h 21 21 *t iu ʔia keiŋ213 jump very high ‘(somebody) jumps very high’

b. →

研究 ŋiɛŋ31 kiu213 ŋiɛŋ44 kiu213 # *ŋiɛŋ21 kiu44 study ‘to study insects’

虫 thøyŋ51 thøyŋ51 thøyŋ51 insect

Thus, we can find that, although TS is a prevalent phonological rule in the Fuzhou dialect and may apply at both the lexical level and the phrasal level, it does not apply within all domains formed by two or more syllables. 1.3.5.2  Morphological Tone Sandhi (MTS) Besides the TS rule, there is another tone sandhi rule in the Fuzhou dialect, namely Morphological Tone Sandhi (MTS). Unlike the TS rule, which occurs in most domains containing two or more syllables as long as the phonological environment is provided, the MTS rule can only apply to some disyllabic words formed through the morphological process of reduplication. In other words, MTS applies to some words that are composed of two identical syllables. As discussed in Section  1.3.5.1, when TS occurs, the sandhi tone of a non-­ terminal syllable is determined by its own citation tone as well as the following tone. By contrast, when MTS applies, the sandhi tone of the non-terminal syllable (the first syllable in this case) is conditioned only by its citation tone. If the citation tone of the first syllable is yangping, shang, or yangru, its tonal value becomes 31 when MTS applies; if the citation tone of the first syllable is yinqu, yangqu, or yinru, its tonal value becomes 21. The sandhi form of yinping within the domain of MTS is a matter of controversy in the literature. It has been argued by some linguists that yinping behaves the same as yangping, shang, and yangru and has the sandhi tone 31 within the domain of MTS (e.g., Liang 1982; Lin 2002, among others). Other linguists, however, argue that yinping remains unchanged within the domain of MTS (e.g., Chen & Norman 1965a; Chen 1998).

Introduction  13 Table 1.7  MTS in the Fuzhou dialect T (monosyllabic word)

T’ + T (disyllabic reduplicated words)

44

44 + 44 (cf. Chen & Norman 1965a; Chen 1998)

51/31/5

31 + 44/51/31/5 (cf. Liang 1982; Lin 2002)

213/242/23

21 + 213/242/23

Tone sandhi patterns of MTS can thus be presented as in Table 1.7. Examples of MTS are given in (9), where TS is blocked. (9)

a. b. c. d.

碗碗 ʔuaŋ31 舀舀 ʔiu31 袋袋 toy242 拍拍 phaʔ23

‘bowl’ → ‘ladle’ → ‘bag’ → ‘bat’ →

ʔuaŋ31 ʔuaŋ31 (MTS)

*ʔuaŋ24 ʔuaŋ31 (TS)

ʔiu31 ʔiu31 (MTS)

*ʔiu24 ʔiu31 (TS)

toy21 toy242 (MTS)

*toy51 toy242 (TS)

phaʔ21 phaʔ23 (MTS)

*phaʔ51 phaʔ23 (TS)

1.3.5.3  Final Alternation (FA) As mentioned in Section 1.3.2, there is a very special phonological phenomenon in the Fuzhou dialect, namely the tonally conditioned alternation between two variants of an alternating final. To recapitulate, in isolation forms, Group A and Group B variants of Fuzhou alternating finals are in complementary distribution in terms of the tones: Group A variants occur with tones 44, 51, 31, and 5, while Group B variants occur with tones 213, 242, and 23 (see Table 1.3 for alternating finals in Fuzhou). Final Alternation (henceforth FA) can be further illustrated by the examples in (10), in which Group A variants [uŋ], [uʔ], [i], and [iʔ] occur with tones 44, 51, 31, and 5, while Group B variants [ouŋ], [ouʔ], [ei], and [eiʔ] occur with tones 213, 242, and 23. (10) a. uŋ (uʔ) vs. ouŋ (ouʔ) 风 xuŋ44 ‘wind’ 云 xuŋ51 ‘cloud’ 粉 xuŋ31 ‘powder’ 训 xouŋ213 ‘to teach, to train’ 混 xouŋ242 ‘to get by’ 拂 xouʔ23 ‘to flick’ 佛 xuʔ5 ‘Buddha’

b. i (iʔ) vs. ei (eiʔ) 机 ki44 ‘machine’ 旗 ki51 ‘flag’ 己 ki31 ‘shrewish’ 记 kei213 ‘to remember’ 忌 kei242 ‘death anniversary’ 急 keiʔ23 ‘rapid; impatient’ 扱 kiʔ5 ‘to stab, to prick’

1.3.5.4  Final Change (FC) As can be seen from Table 1.5 and Table 1.7, yinqu 213, yangqu 242, and yinru 23 tones never occur in the non-terminal position in a sandhi context: they always undergo tone sandhi processes (either TS or MTS) in the sandhi position. Thus,

14  Introduction along with the change in tones, in the sandhi position, syllables with Group B variants of alternating finals must undergo another phonological process, namely Final Change (henceforth FC). In a sandhi environment, where non-terminal syllables can only have yinping, yangping, shang, yangru, half yinqu, or half yangqu tones, all the Group B variants in the sandhi position are “replaced” by their Group A counterparts; by contrast, Group A variants in the sandhi position can retain their citation forms even though the tones may undergo TS or MTS (cf. Chen & Norman 1965a; Chan 1985; Chen 1998, among others). FC is illustrated by the examples in (11). In each pair of the following examples, the first example exhibits the FC rule, while the second does not. Sandhi forms of finals are marked in bold, and sandhi tones are presented in the examples as well. (11) a.

[u/ou] 旧书 乌暗 b. [uŋ/ouŋ] 动作 同居 c. [uʔ/ouʔ] 熨斗 物理 d. [øy/oy] 爱食 螺丝 e. [øyŋ/oyŋ] 粽箬 棕树 f. [øyʔ/oyʔ] 触眼 毒蠓

kou242 tsy44 ʔu44 ʔaŋ213

→ ku44 tsy44 → ʔu51 ʔaŋ213

‘old book’ ‘dark’

touŋ242 tsouʔ23 tuŋ51 ky44

→ tuŋ51 tsouʔ23 → tuŋ44 ky44

‘action’ ‘neighbors; to co-inhabit’

ʔouʔ23 tau31 ʔuʔ5 li31

→ ʔuʔ24 tau31 → ʔuʔ31 li31

‘iron (for clothes)’ ‘physics’

ʔoy213 siɛʔ5 løy51 si44

→ ʔøy44 siɛʔ5 → løy44 si44

‘to like to eat’ ‘screw’

tsoyŋ213 nuoʔ5 → tsøyŋ44 nuoʔ5 ‘rice dumpling wrappings’ tsøyŋ44 tshiu213 → tsøyŋ51 tshiu213 ‘palm’ toyʔ23 ŋaŋ31 tøyʔ5 møyŋ31

→ tøyʔ24 ŋaŋ31 ‘cross eyes’ → tøyʔ31 møyŋ31 ‘midge’

As can be seen from the examples in (11), FC can not only apply to lexical items like 动作 ‘action’ and 熨斗 ‘iron (for clothes)’, but also to phrases like 旧书 ‘old book’ and 爱食 ‘to like to eat’. Therefore, like TS, FC may also apply within domains of different sizes or levels. Moreover, since FC is a tonally conditioned phonological process, in phrasal-level cases where tone sandhi rules are blocked, FC also fails to apply, as exemplified in (12), in which the finals and tones in question are all marked in bold. The positions where the rules are blocked are denoted by “#”. (12)

a. →

服侍 爸奶 xuʔ5 søy242 pa242 nɛ31 xuʔ21 søy242 # pa51 nɛ31 *xuʔ21 sy21 pa51 nɛ31 take care parents ‘to attend to one’s parents’

b. →

耳 真 ŋei242 tsiŋ44 ŋei242 # tsiŋ51 *ŋi21 tsiŋ51 ear truly ‘hard of hearing’

塞 seiʔ23 seiʔ23 seiʔ23 blocked

Introduction  15 1.3.5.5  Initial Consonant Lenition (CL) In the Fuzhou dialect, the initial of a non-first syllable within a given domain containing two or more syllables is usually changed according to the final of the preceding syllable (see Chen & Norman 1965a; Chan 1985; Li et al. 1994; Chen 1998, among others). Initial Consonant Lenition (henceforth CL) is presented in Table 1.8, in which sandhi forms of initials are given in shaded boxes. Table 1.8  CL in the Fuzhou dialect Sandhi form of the following initial

Following initial

p ph t th s l k kh x ʔ/0 ts tsh

m n ŋ

Preceding final open syllable/vowel coda/-ʔ

β

l

-ŋ

m

n

-k

ʔ/0

ʒ

no change

ŋ

ʒ

no change

no change

Source: Cf. Li et al. 1994; Chen 1998, among others.

As we can see from Table  1.8, CL gives rise to two sandhi initials, namely β and ʒ, which do not occur in the underlying Fuzhou initial inventory shown in Table 1.4. These two sandhi initials occur only in a sandhi environment where CL applies. Also, it is noteworthy that the two historical endings of syllables with the yinru tone have different effects on the initial consonant of the following syllable. If the historical ending of a syllable with the yinru tone is *[-k], it prevents the following initials from undergoing CL. By contrast, if the historical ending of a syllable with the yinru tone is *[-ʔ], it triggers CL on most following initials. In addition, if a non-first syllable is initiated by a nasal sound [m], [n], or [ŋ], its initial never undergoes CL. The application of CL is illustrated by examples in (13). Sandhi forms of initials are marked in bold. Sound changes to tones and finals (if any) are presented as well. (13)

车票 旧底 客车 暝晡 颜色 电话 野好 旧书

tshia44 phiu213 kou242 tɛ31 khaʔ23(-ʔ) tshia44 maŋ51 puo44 ŋaŋ51 saiʔ23 tiɛŋ242 ʔua242 ʔia31 xo31 kou242 tsy44

→ → → → → → → →

tshia51 βiu213 ku51 lɛ31 khaʔ44 ʒia44 maŋ44 muo44 ŋaŋ21 naiʔ23 tiɛŋ51 ŋua242 ʔia24 ʔo31 ku44 ʒy44

‘(bus or train) ticket’ ‘before, the old times’ ‘passenger car, bus’ ‘evening’ ‘color’ ‘telephone’ ‘very good’ ‘old book’

It can be seen from (13) that CL can apply to both lexical items, such as 暝晡 ‘evening’ and 颜色 ‘color’, and phrases, such as 野好 ‘very good’ and 旧书 ‘old book’, indicating that CL is another rule that can be triggered within domains of different sizes or levels in Fuzhou.

16  Introduction As shown in the examples in (14), however, the CL rule is blocked in some cases at the phrasal level for some reason. Initials in question are marked in bold and the symbol “#” denotes the positions where CL is blocked. (14)

a. →

饭 食 5 puoŋ242 siɛʔ 21 siɛʔ # puoŋ242 21 *siɛʔ βuoŋ242 eat rice ‘to eat food’

b. →

喜欢 xi31 xuaŋ44 xi21 ʔuaŋ44 # *xi21 ʔuaŋ44 like ‘to like books’

书 tsy44 tsy44 ny44 book

It has also been noticed by some linguists (e.g., Li et al. 1994; Li & Liang 2001; Li 2002, among others) that, in a given string of two or more syllables, the presence and absence of CL can be used to denote different meanings and/or syntactic relationships between members, as exemplified in (15). In (15), the first example involves the application of CL, and the sandhi form of the initial is marked in bold, while the second example does not undergo CL. (15)

a. b.

做细 做细

tso213 sa213 → tso51 la213 tso213 sa213 → tso51 sa213

‘(in) childhood’ ‘to be a concubine’

1.3.6 Summary I have so far presented a general picture of the phonological system of the Fuzhou dialect, including its tonal and segmental inventories, syllable types, and major phonological rules. Based on the discussion above, we can find that, although sound changes can occur to tones, finals, and initials in this dialect, the phonological rules we have just introduced are in actual fact not always triggered in all strings of sounds. Among the five rules, we have seen that: (a) FA occurs only within but not across syllables; (b) MTS applies only to some disyllabic words formed by reduplication; (c) TS, FC, and CL, despite the fact that they all apply within more than one domain (in both lexical items and phrases), may also be blocked in some cases at the phrasal level. Syllables, lexical items (words), and phrases are clearly different types of linguistic units forming different domains of rule application for these phonological rules. In other words, two scenarios can be seen regarding the application of phonological rules in the Fuzhou dialect. First, a given rule may apply within a particular domain but be blocked within another. Second, a given rule may apply within more than one domain and may exhibit different degrees of application in different domains. Although these two scenarios have been, to varying extents, noticed in some previous studies (e.g., Chen & Norman 1965a; Chan 1985; Shih 1986; Hung 1987; Zhang 1992; Li et al. 1994; Li & Liang 2001; Li 2002, among others), no comprehensive and satisfactory analysis has been offered. To account for the two abovementioned scenarios of rule application in the Fuzhou dialect, we need to give special attention to the following questions: (a) How to define the “domains” of

Introduction  17 rule application (both in general and in the Fuzhou dialect)? (b) How do different Fuzhou phonological rules behave in different domains? Are they triggered in these domains obligatorily or optionally? Are they blocked in these domains? (c) From the investigation into the relationship between Fuzhou phonological rules and the domains, what theoretical implications can be drawn? In order to answer these questions, we need to first learn about the theoretical framework assumed in this book, namely the theory of prosodic phonology, which is discussed in Section 1.4.

1.4  Theoretical background: the prosodic phonology theory We proceed now to a survey of the theoretical framework assumed in this study, namely the prosodic phonology theory, which in particular is the theory of phonological domains and interactions between phonology and other components of the grammar. The following subsections present the development of the prosodic phonology theory (Section 1.4.1), its basic tenets (Section 1.4.2), and some outstanding issues concerning this theory that are relevant to the topics under discussion in this book (Section 1.4.3). 1.4.1  The development of the prosodic phonology theory Prosodic phonology was developed over 30 years ago and still stands as a representative phonological theory of the interactions between phonology and other components of the grammar. The earliest precursor of the prosodic phonology theory is commonly taken to be Selkirk (1978/1981).6 Selkirk developed the first model of prosodic phonology, based on the central idea of Liberman (1975) and Liberman & Prince (1977), that segments are dominated by a multi-layered structure that expresses rhythmic properties of the linear string and assigns relative prominence to individual chunks. In Selkirk’s (1978/1981) model, a suprasegmental, hierarchically arranged organization (namely prosodic structure) of phonological representation is proposed. A six-layer prosodic hierarchy was introduced in this article, containing only six prosodic constituents, namely the syllable (σ), the foot (Σ), the phonological word/prosodic word (ω),7 the phonological phrase (φ), the intonational phrase (ι), and the phonological utterance (υ). After introducing the idea of prosodic structure into the field of the syntaxphonology interface in her early works (Selkirk 1978/1981, 1980a, 1980b), Selkirk (1984), nevertheless, abandons prosodic constituency under the influence of Prince’s (1983) grid-only approach. She proposes a version of prosodic phonology without prosodic constituency and holds that the function of all prosodic categories except the intonational phrase is taken over by the metrical grid (Selkirk 1984: 29ff). The famous condition on prosodic constituency, the Strict Layer Hypothesis (SLH), is also formally proposed in Selkirk (1984: 26ff). However, two years later, Selkirk (1986) returns to prosodic constituency, arguing that there is a “peaceful coexistence” of prosodic constituency and the metrical grid. She claims that the metrical grid is defined not with respect to syntactic domains but with respect to prosodic structure instead (Selkirk 1986: 376). This system thus

18  Introduction acknowledges two distinct mapping mechanisms that are serially ordered: (a) prosodic constituency is the output of regular mapping rules, with morphosyntactic structure as its input, and (b) the metrical grid construction is performed on prosodic structure domains by a second set of mapping rules. Following Chen (1985, 1987), Selkirk (1986) also argues for a single-edge-based approach, namely the Edge/EndBased Approach (henceforth EBA; see Section 1.4.3.3), claiming that, between the foot and the intonational phrase, the relation between syntactic structure and prosodic structure is defined in terms of the ends/edges of syntactic constituents of designated types (Selkirk 1986: 385). Specifically, it calls for the right or left edges of designated syntactic constituents to match up with edges of prosodic constituents. Built on Selkirk’s work, Nespor & Vogel’s (1986) book (re-published in 2007 with a new foreword) is another fundamental pioneering and authoritative work on prosodic phonology, in which they concentrate the insights they obtained based on the data elicited from a number of languages such as Italian and English. Nespor & Vogel (1986) added and inserted the prosodic constituent clitic group (CG) between the prosodic word and the phonological phrase and thus proposed a prosodic hierarchy composed of seven prosodic constituents. Different from the Edge/End-Based Approach advanced by Selkirk, the Relation-Based Approach (henceforth RBA; see Section  1.4.3.3) proposed in Nespor  & Vogel (1986) makes reference to X-bar notions of phrase structure, such as head-complement, ­modifier-head, and specifier-head relations, as well as syntactic branching. Other pioneering works on the theory of prosodic phonology in the early period include Hayes (1984/1989)8; Booij (1983, 1985a, 1985b, 1986); Neijt (1985); Itô (1986); and Chen (1985, 1987), among others. After the advent of Optimality Theory (OT) in early 1990s (Prince  & Smolensky 1993, among others), prosodic phonology has been translated into the constraint-based environment. Selkirk (1996) adopts the OT framework of ranked constraints and extends it to the area of the morphosyntax-phonology interface. After the articulation of the Generalized Alignment Theory in McCarthy & Prince (1993), Selkirk (1996) recast the EBA in terms of a family of alignment constraints, namely ALIGN-XP, which defines edges of prosodic constituents in terms of edges of surface syntactic structure. Moreover, Selkirk (1996) factors out the monolithic Strict Layer Hypothesis into four more primitive component constraints that can be manipulated independently, namely Layeredness, Headedness, Exhaustivity, and Nonrecursivity (see Section 1.4.3.1). Truckenbrodt (1995, 1999) introduces other constraint families into the prosodic phonology theory, including WRAP-XP, which requires that each XP be contained in the same phonological phrase, and ALIGN-FOC, which demands that each focused constituent is right-aligned or leftaligned with a phonological phrase boundary. In addition to ALIGN-XP, WRAP-XP, ALIGN-FOC, and the four constraints derived from the SLH, other constraints have also been proposed for prosodic phonology within the OT framework, such as BinMin (φ, ω) (a phonological phrase minimally consists of two prosodic words; see Selkirk 2000); BinMax (φ, ω) (a phonological phrase maximally consists of two prosodic words; see Selkirk 2000); *P-PHRASE (avoid phonological phrases; see Truckenbrodt 1999); UNIFORMITY (a string is ideally parsed into units of the same length; see Ghini 1993; Sandalo & Truckenbrodt

Introduction  19 2002; Prieto 2005, 2006); and INCREASING UNITS (phonological phrases on the recursive side are heavier than those on the non-recursive side; see Ghini 1993). Like the Align/Wrap Theory mentioned above, the Match Theory proposed by Selkirk (2005, 2009, 2011) and further elaborated by Elfner (2012, 2013) is another theory of universal constraints on the correspondence between syntactic and prosodic constituency in grammar. The Match Theory is the most recent development in the study of the morphosyntax-phonology interface. In the prosodic phonology theory before the era of OT, there is no inherent relationship assumed between prosodic and syntactic categories, while Selkirk (2005, 2009, 2011) advances the idea that the hierarchical relationship among the so-called “interface categories” (Itô & Mester 2012: 281), namely the prosodic word, the phonological phrase, and the intonational phrase, is syntactically grounded. Thus, in the Match Theory, morphosyntactic notions, including clause, phrase, and word, are made use of with respect to syntax-phonology interaction and have their correspondents (the intonational phrase, the phonological phrase, and the prosodic word respectively) in phonology. The Match Theory is in actual fact developed from the EBA. Within the framework of OT (or more specifically, the Correspondence Theory, see McCarthy & Prince 1995), the Match Theory is formally refined as two sets of violable correspondence constraints (Selkirk 2009, 2011). One set of constraints (S-P faithfulness constraints) requires that the edges of a syntactic constituent of type α (α = a variable over syntactic constituent types) correspond to the edges of a phonological constituent of type π (π  =  a variable over prosodic constituent types), while the other set (P-S faithfulness constraints) requires that the edges of a phonological constituent π correspond to the edges of a syntactic constituent α. In the OT formulation of the Match Theory, the phonological phrasing is obtained through the interaction of a syntax-prosody correspondence constraint Match (Phrase, φ) with a prosodic markedness constraint BinMin (φ, ω), and the ranking between these constraints makes for a typological difference between languages (see more detailed discussion on different approaches to phonological phrasing in Section 1.4.3.3). From the discussion above, we can find that there have been a number of developments in the prosodic phonology theory over the past three decades or so. Some early assumptions have been revised, discarded, or replaced by new ones. However, basic tenets of the prosodic phonology theory, such as the prosodic structure, the prosodic constituents as domains of rule application, and the notion of the prosodic hierarchy, have actually not been modified and thus still remain almost unchallenged, as to be discussed in more detail in Section 1.4.2. 1.4.2  Basic tenets of the prosodic phonology theory 1.4.2.1  Indirect reference: the prosodic structure There is a wealth of literature concerning the issue of interactions between phonology and other components of the grammar, or in other words, the issue of how phonological processes make reference to morphosyntactic information as domains for their application. Theories of the interface can be divided into two

20  Introduction major groups according to the role they attribute to morphosyntactic information in the process of the creation of phonological domains. The first group of interface theories, the Direct Reference Theory, insists that phonological processes should be allowed to make direct reference to morphosyntactic information (see Manzini 1983; Kaisse 1985; Odden 1987, 1990, 1996; Rizzi & Savoia 1993, among others). By contrast, the second group, the Indirect Reference Theory, represented by the prosodic phonology theory, holds that phonological processes are not directly sensitive to syntactic information (see Selkirk 1978/1981, 1980a, 1980b, 1984, 1986; Nespor & Vogel 1986, 2007; Hayes 1984/1989, among others). The main claim of prosodic phonology is that syntactic and phonological representations are not isomorphic and that morphosyntactic constituents do not determine the domains of the application for phonological rules and phonetic processes. Within the model of the prosodic phonology theory, there exists a hierarchically arranged organization called Prosodic Structure between the morphosyntactic and phonological components. A given string of sounds is thus organized into a series of hierarchically arranged prosodic constituents, with each prosodic constituent serving as the domain of application for specific phonological rules and phonetic processes. Thus, phonological operations themselves do not refer to syntactic constituents in a direct way, but instead to the already created prosodic constituents of the prosodic structure. Hence, the existence of phonological rules and phonetic processes that make crucial reference to a particular prosodic constituent is viewed as one significant motivation of the establishment of the prosodic constituent itself in a given language. The domain of each prosodic constituent is built with reference to different types of phonological and non-phonological information, but the resulting prosodic constituents that make reference to non-­phonological notions are not necessarily isomorphic to any morphosyntactic structures (cf. Nespor & Vogel 1986, 2007; Zhang 1992, 2017, among others). The two major approaches to the morphosyntax-phonology interface proposed in prosodic phonology before the era of OT, namely the EBA and the RBA, as well as the Align/Wrap Theory derived from the EBA within the framework of OT, are all indirect-reference theories. They all need to establish prosodic domains within which relevant phonological rules apply and thus phonological rules do not refer to morphosyntax directly. It is also noteworthy that the Match Theory, the latest development of the study of the morphosyntax-phonology interface, is essentially an indirect-reference theory. Although the Match Theory calls for a kind of direct correspondence between the input syntactic representation and the output phonological representation, the prosodic structure is still required. Hence, we still need to construct prosodic domains, such as the prosodic word, the phonological phrase, and the intonational phrase, for the application of phonological rules, which makes the Match Theory distinct from direct-reference approaches. 1.4.2.2  Prosodic hierarchy Prosodic constituents of the prosodic structure are arranged in hierarchical order and thus form the prosodic hierarchy. Within the hierarchy, prosodic constituents

Introduction  21 at a given level are composed of constituents on the next lower level. The prosodic hierarchy was first proposed in Selkirk (1978/1981), as presented in Figure 1.1.9 Hayes (1984/1989) and Nespor & Vogel (1986) added and inserted the prosodic constituent clitic group (CG) between the prosodic word and the phonological phrase. Zec (1988) proposed another domain, the mora (μ), which is located as the lowest constituent in the prosodic hierarchy. Hence, an expanded eight-layer universal prosodic hierarchy can be shown in Figure 1.2. Additional constituents have been proposed by proponents of prosodic phonology, such as the focal phrase, located between the phonological phrase and the intonational phrase (Kanerva 1990), and the Small Word, which comprises part of a word (Rice 1993), but they have not received universal consideration. Utt/υ

(Utterance)

IPh/ι PPh/φ PW/ω Σ

(IPh/ι) (Intonational Phrase)

(PPh/φ)

(Phonological Phrase)

(PW/ω)

(Prosodic Word)

(Σ)

σ

(Foot)

(σ)

(Syllable)

Figure 1.1  Prosodic hierarchy Source: Adapted from Selkirk 1978/1981.

Utt/υ IPh/ι PPh/φ CG PW/ω Σ σ

(IPh/ι)

(PPh/φ) (CG)

(PW/ω)

(Σ) (σ)

( ) Figure 1.2  Prosodic hierarchy (expanded)

(Utterance) (Intonational Phrase) (Phonological Phrase) (Clitic Group) (Prosodic Word) (Foot) (Syllable) (Mora)

22  Introduction Semantic & pragmatic information

Utt/υ IPh/ι

(Discourse/focus-based) Morphosyntactic information (Morphosyntax-based)

PPh/φ CG

PW/ω Phonological information Σ (Rhythm-based)

σ

(IPh/ι) (Intonational Phrase)

(PPh/φ) (CG)

(PW/ω) (Σ)

(σ) ( )

(Utterance)

(Phonological Phrase) (Clitic Group) (Prosodic Word) (Foot) (Syllable) (Mora)

Figure 1.3  Trisected model for prosodic hierarchy Source: Adapted from Zhang 1992, 2017.

It has been mentioned in Section 1.4.2.1 that prosodic constituents are defined by making use of different types of phonological and non-phonological information. It is of crucial importance that whether a given constituent makes use of a specific type of information in the definition of its domain is not a free choice (Nespor & Vogel 1986, 2007). According to the types of information to which different prosodic constituents in the prosodic hierarchy are sensitive, Zhang (1992, 2017) proposed a trisected model for prosodic hierarchy, as given in Figure 1.3. As we can see in Figure 1.3, the prosodic hierarchy is divided into three parts by two dividing lines. According to Zhang (1992, 2017), the dividing line for the first part is drawn between the prosodic word and the foot, while the other is drawn between the intonational phrase and the phonological phrase. The mora, the syllable, and the foot belong to the rhythm-based hierarchy because they share the same properties and are all sensitive to sonority features such as strong and weak. The prosodic domains between the foot and the intonational phrase constitute the morphosyntax-based hierarchy, in that they are all sensitive to morphosyntactic information in terms of their domain formation. The two prosodic constituents above the phonological phrase, namely the intonational phrase and the utterance, form the discourse/focus-based hierarchy, since their domain formation is sensitive to semantic and pragmatic information (cf. Selkirk 1984; Nespor & Vogel 1986; Zec 1988, among others). The prosodic hierarchy is assumed to be universal by proponents of the prosodic phonology theory. Nevertheless, the inventory of prosodic constituents and the definition of prosodic domains in a particular language are an empirical issue.10 Hence, it is natural that some controversy surrounding the number and the definition of prosodic domains is found in the literature, and it has been observed that some prosodic domains seem to be absent or play no roles in some languages.

Introduction  23 1.4.3  Outstanding issues in the prosodic phonology theory In this section, I  will discuss three outstanding theoretical issues in the prosodic phonology theory, which also constitute important topics I would like to explore in this book on the basis of the investigation into the Fuzhou phonological system. 1.4.3.1  The weakened Strict Layer Hypothesis The first outstanding issue regarding prosodic phonology is the only well-­ formedness condition on prosodic constituency, namely the Strict Layer Hypothesis (SLH), formulated in Selkirk (1984: 26ff). The SLH stipulates that, in the prosodic hierarchy, a prosodic constituent of a given level n immediately dominates only constituents of the lower level n − 1 and is exhaustively contained in a constituent of the immediately higher level n + 1. This is reformulated in Nespor & Vogel (1986: 7), as in (16). (16)  a. A given non-terminal unit of the prosodic hierarchy, Xp, is composed of one or more units of the immediately lower category, Xp-1.   b. A unit of a given level of the hierarchy is exhaustively contained in the superordinate unit of which it is a part. Thus, prosodic constituents must be organized to accord with the SLH, which prohibits recursive structures (which are different from syntax) and levelskipping in these hierarchically ordered prosodic constituents. A structure like (17a) would be a good example of a well-formed prosodic tree, whereas (17b) would not. (17) (17) a.

ι φ

φ

CG

φ

CG

ω

ω

CG ω

b.

ω

CG ω

ω

ι ι φ φ ω

ω

ω

ω

ω

ω

24  Introduction We can find that (17b) violates the SLH in two respects: ι and φ are recursive (namely ι dominates ι and φ dominates φ), and ι dominates ω, skipping φ. Evidence and criticisms have challenged the SLH (e.g., Ladd 1986; Odden 1987; Hyman et al. 1987; Inkelas 1989/1990; Itô & Mester 1992/2003; Prince & Smolensky 1993; Vogel 2009; Zhang 1992, 2014, 2017; Truckenbrodt 1995, 1999, among others). In responding to these evidence and criticisms, Selkirk (1996: 189ff) has factored out the SLH into four more primitive constraints on prosodic domination within the framework of OT, as mentioned in Section 1.4.1, which are presented as follows. (18) Constraints on prosodic domination (adapted from Selkirk 1996: 189ff) (where Cn = some prosodic category) a. Layeredness: No Ci dominates a Cj, j > i, e.g., “No syllable dominates a foot.” b. Headedness: Any Ci must dominate a Ci−1 (except if Ci = syllable), e.g., “A phonological word must dominate a foot.” c. Exhaustivity: No Ci immediately dominates a constituent Cj, j < i – 1, e.g., “No phonological word immediately dominates a syllable.” d. Nonrecursivity: No Ci dominates Cj, j = i, e.g., “No foot dominates a foot.” Selkirk holds that the constraints of Layeredness and Headedness are inviolable and should not be dominated in the constraint ranking universally. By contrast, Exhaustivity and Nonrecursivity are suggested not to be observed by all languages. Itô  & Mester (2012) also lay out a general model for prosodic structure, ­Recursion-based Subcategories, and suggest that the prosodic word, the phonological phrase, and the intonational phrase all can be recursive. These new developments of the prosodic phonology theory, including Recursion-based Subcategories and the violability of Exhaustivity and Nonrecursivity, call for a new mapping algorithm for matching syntax with the recursive prosodic structure. The Match Theory we mentioned in Section 1.4.1 is such a theory, which argues that level-skipping (i.e., the violation of Exhaustivity) and recursion (i.e., the violation of Nonrecursivity) in the prosodic structure actually mirror the recursion in syntax (Selkirk 2009, 2011). Many examples of the violation of Exhaustivity have been found across languages (Inkelas 1989/1990; Kanerva 1989; Itô & Mester 1992/2003; Prince & Smolensky 1993; Mester 1994; Hayes 1995; Vogel 2009; Zhang 2014, 2017, among others). Mandarin Chinese can serve as a typical example among these languages. Based on evidence found in Mandarin Chinese, Zhang (2014, 2017) proposes a prosodic hierarchy as in Figure 1.4, which entails four possible types of organization of prosodic constituents in Mandarin Chinese. We can find that level-skipping, namely the violation of Exhaustivity, is clearly allowed in this language.11 Examples of the violation of Nonrecursivity have also been found in different languages (Ladd 1986; Odden 1987; Hyman et al. 1987; Inkelas 1989/1990;

Introduction  25 Utterance Intonational Phrase Syllable

Phonological Phrase

Phonological Phrase

Prosodic Word

Mora

Clitic Group

Prosodic Word

Syllable

Prosodic Word

Syllable

Mora

Syllable

Mora

Mora Figure 1.4  Prosodic hierarchy in Mandarin Chinese Source: Adapted from Zhang 2014, 2017.

Zhang 1992, 2017; Truckenbrodt 1995, 1999, among others). According to Zhang (1992, 2017), recursivity is allowed in some Chinese dialects. For example, a clitic group may dominate another clitic group in old Chongming, and a prosodic word may dominate another prosodic word in Pingyao. In addition to Nonrecursivity and Exhaustivity, it has been noticed that the Layeredness constraint is not inviolable either. As Zhang (1992, 2017) points out, counterevidence against the inviolability of Layeredness can be found in the Pingyao dialect of Chinese. Based on the tone sandhi data from the Pingyao dialect, Zhang argues that, in this dialect, a prosodic word may dominate a prosodic constituent at the higher level, e.g., the phonological phrase. Like Selkirk (2009, 2011), Zhang (1992, 2017) also suggests that prosodic recursivity actually reflects syntactic recursivity. Different from other previous studies such as Itô & Mester (2012) and Selkirk (2009, 2011), however, Zhang (1992, 2017) argues that prosodic recursivity (including the violation of the constraints of Nonrecursivity, Exhaustivity, and Layeredness) occurs only between prosodic constituents in the morphosyntax-based hierarchy in Figure 1.3 (i.e., the prosodic word, the clitic group, and the phonological phrase). From the discussion above, we can see that, although the SLH was originally thought to be one of the most important characteristics of the prosodic structure as compared to syntactic structure, it has been demonstrated that the constraints entailed in this hypothesis are not absolutely inviolable, according to evidence across languages. As a result, the SLH has been made much less strict along with the development of prosodic phonology. Although scholars have different

26  Introduction opinions about which prosodic categories can violate the constraints entailed in the SLH, a weakened SLH has become undoubtedly necessary in today’s theory on both empirical and theoretical bases. In this book, I will show that the necessity of a weakened SLH is well supported by the evidence from the Fuzhou dialect, in which the violation of Exhaustivity, Nonrecursivity, and even Layeredness should all be allowed. 1.4.3.2  The status of the clitic group The second outstanding issue in the theory of prosodic phonology is the status of the clitic group. The existence of the clitic group as a prosodic constituent is proposed based on the observation that there are phonological rules and phonetic processes that make crucial reference only to the group consisting of a lexical word and the clitic that attaches to it (see Hayes 1984/1989; Nespor  & Vogel 1986; Cohn 1989, among others). The notion of the clitic group was first proposed by Hayes (1984/1989) and then adopted in Nespor & Vogel (1986) and subsequent work (e.g., Vogel 1990, 1991; Hannahs 1995a; Nespor 1999, among others). The clitic group is seen to be located in a position above the prosodic word and below the phonological phrase, as we have seen in the prosodic hierarchy presented in Figure 1.2 and Figure 1.3. The domain of the clitic group is defined in Nespor & Vogel (1986: 154) essentially as in (19). (19) Clitic group (CG) formation The domain of the CG consists of a ω containing an independent (i.e., nonclitic) word plus any adjacent phonological words containing: a. a directional clitic, or b. a plain clitic/nondirectional clitic such that there is no possible host with which it shares more category memberships.12 The definition of the clitic group domain is illustrated by examples of stress assignment in Latin in Nespor  & Vogel (1986: 146ff). According to Nespor  & Vogel, the clitic group is a domain for stress assignment in Latin. Specifically, when an enclitic is attached to a word, the primary stress is shifted from its original position within the word to the syllable that immediately precedes the clitic, as exemplified in (20), in which -que ‘and’, interrogative -ne, and -cum ‘with’ are all enclitics. (20)

a. b. c.

vírum ‘the man (acc.)’ virúmque vídēs ‘you see’ vidḗsne? cum vóbis ‘with you (pl.)’ vobíscum

‘and the man (acc.)’ ‘do you see? ‘with you (pl.)’

There are many other phonological phenomena applying within clitic groups, but not across their boundaries or in other prosodic domains cross-linguistically, such as v-Deletion and s, z-Palatalization in English, Stress Readjustment, Nasal Deletion, Nasal Assimilation, and Stop Voicing in Greek, Stress Assignment and Vowel

Introduction  27 Harmony in Turkish, Penultimate Schwa Specification and Post-lexical Stress Assignment in French, as well as t-Deletion in Catalan (see Hayes 1984/1989; Nespor & Vogel 1986; Hannahs 1995a; Kabak & Vogel 2001, among others). However, there are some arguments against the existence of the clitic group as a prosodic domain, e.g., (a) clitics may attach to constituents higher than the prosodic word, e.g., the phonological phrase and the intonational phrase; (b) there is a lack of evidence for the clitic group as a domain in particular languages; (c) enclitics and proclitics often show asymmetries in languages; and (d) clitics have to be given the prosodic word status according to the definition presented in (19) to satisfy the SLH. For example, Inkelas (1989/1990) points out that clitics have an irregular distribution across syntactic categories, and consequently the formation of the clitic group cannot be governed by a simple syntax-to-phonological mapping algorithm. According to Inkelas, the clitic group in languages such as Serbo-Croatian and English is dominated by the phrase, while the clitic group in languages like Hausa and Kivunjo Chaga seems to dominate the phrase. Therefore, adding the clitic group to the prosodic hierarchy as a constituent between the prosodic word and the phonological phrase fails to accommodate the distinction between word and phrasal clitics. Booij (1996) holds the same position as Inkelas (1989/1990). He claims that there are two objections to assuming the clitic group. By examining the relevant data of Latin Stress Rule and facts of main stress assignment in other languages, Booij claims that it is possible to give an account of the relevant data without the clitic group. The second objection raised by Booij is that the definition in (19) implies that clitics always form a prosodic word on their own, which is a problematic assumption since many clitics do not meet the requirements for canonical prosodic words. Through the discussion of the phonological behavior of Dutch clitics, Booij argues that the phonological behavior of clitics in Dutch can be accounted for by the hypothesis of prosodic integration, which claims that clitics are incorporated into an adjacent prosodic word. Based on these discussions, he concludes that the assumption of the clitic group is not necessary. The same position is adopted by Zec (1988, 1993); Inkelas & Zec (1995); Selkirk (1996); and Peperkamp (1997), among others. Moreover, it has been reported that asymmetries exist between proclitics and enclitics in languages such as Dutch (Booij 1996), Italian (Peperkamp 1997), and German (Kleinhenz 1996), in terms of their coherence to the host they attach to. It has been pointed out that proclitics usually present a phonological behavior independent of the host, while enclitics usually show a stronger degree of connection with the host. Thus, if the clitic group is treated as an independent prosodic constituent, the distinction between proclitics and enclitics is not easily accounted for. Despite the objections mentioned above, some researchers have continued to employ the clitic group in an insightful manner to account for prosodic phenomena in different languages (e.g., Hayes 1984/1989 for English; Zhang 1992, 2014, 2017 for Mandarin Chinese and Chinese dialects; Hannahs 1995a for French; and Kabak & Vogel 2001 for Turkish, among others). Vogel (2009) discusses the main

28  Introduction objections to the clitic group and provides arguments for retaining the clitic group in the prosodic hierarchy. She insists that the absence of evidence for the clitic group in a particular language does not necessarily mean that the clitic group is not present in that language. By examining the problems with the original clitic group, she argues that the problem is not due to the clitic group itself but rather results from the original SLH. Based on the review of relevant solutions, Vogel proposes that the SLH should be slightly weakened and revised so that a prosodic constituent may dominate a constituent more than one level lower in the hierarchy (i.e., the violation of Exhaustivity is allowed), and thus the problem with the original clitic group can be resolved. In this book, I  will show that there are two types of clitics and consequently two types of clitic groups in the Fuzhou dialect. As we will see in Chapter 4 and ­Chapter 6, the first type of clitic group (Type A clitic group) in Fuzhou is composed of the host plus the enclitic, while the second type (Type B clitic group) is composed of the proclitic plus the host. Although enclitics and proclitics, as well as the Type A clitic group and the Type B clitic group, show asymmetries in terms of their phonological behavior, the clitic group in Fuzhou as a whole has very peculiar behavior as compared to other prosodic domains, which provides motivation and evidence for the establishment of the clitic group domain in this dialect. Moreover, Fuzhou clitics may attach to constituents higher than the prosodic word, which constitutes a great challenge to the original SLH. Instead of excluding the clitic group from the prosodic hierarchy, nonetheless, this challenge can be handled by resorting to a weakened SLH,13 which will be discussed in Chapter 4 and Chapter 6. 1.4.3.3  Approaches to phonological phrasing: RBA vs. EBA The third outstanding issue I would like to discuss here is the approaches to phonological phrasing. Two main approaches have been developed within prosodic phonology, namely the Relation-Based Approach (RBA) (Nespor & Vogel 1986; Hayes 1989) and the Edge/End-Based Approach (EBA) (Selkirk 1986; Chen 1985, 1987). These two approaches differ from each other in terms of the types and amount of syntactic information they require access to in the construction of prosodic domains, especially the domain of the phonological phrase. The RBA makes reference to X-bar notions of phrase structure, such as head-­ complement, modifier-head, and specifier-head relations, as well as syntactic branching. According to Nespor & Vogel (1986), in terms of the formation of the phonological phrase, reference should be made to the recursive and the non-recursive side of a head. The recursive side is the direction of branching in a language while the non-recursive side is the side where specifiers are located. Nespor & Vogel’s definition of the phonological phrase is given in (21) (Nespor & Vogel 1986: 168). (21) Phonological phrase (φ) formation The domain of φ consists of a clitic group14 that contains a lexical head (X)15 and all clitic groups on its non-recursive side up to the clitic group that contains another head outside of the maximal projection of X.

Introduction  29 In addition, an optional rule for restructuring φ that has the effect of eliminating non-branching φs is proposed in Nespor & Vogel (1986: 173), given as follows. (22) φ-restructuring (optional) A non-branching φ, which is the first complement of X on its recursive side, is joined into the φ that contains X. The mapping algorithm in (21) and (22) is illustrated in Nespor & Vogel (1986) by the rule of Raddoppiamento Sintattico in Italian, which lengthens the initial consonant of a word when preceded by a word ending in a stressed vowel and followed by a sonorant (specifically a vowel or other non-nasal sonorants). This rule is analyzed as applying across two prosodic words within a single phonological phrase. Examples of the contexts in which this rule applies (marked with “=”) and fails (marked with “#”) are adapted from Nespor & Vogel (1986) as in (23a) and (23b) respectively, where the phonological phrasing is also presented. (23)

a.

Avrá = trovato il pescecane. [Avrá = trovato]φ [il pescecane]φ he-will-have found the shark ‘He must have found the shark.’ b. Devi comprare delle mappe [Devi comprare]φ [delle mappe]φ you-must buy some maps di cittá # molto vecchie. [di cittá]φ # [molto vecchie]φ of city very old ‘You must buy some very old city maps.’

The φ-restructuring rule is optional in Italian, and thus, the first non-branching complement of a head on its recursive side is optionally joined into the φ that contains the head, as illustrated by the optional application of Raddoppiamento Sintattico (marked with ‘_’) in (24). (24)

[I caribú]φ _ [nani]φ [sono [I caribú _ nani]φ [sono the caribou dwarf are ‘Dwarf caribous are extinct.’

estinti]φ estinti]φ extinct

In addition to Raddoppiamento Sintattico in Italian, Nespor  & Vogel (1986) also examine Iambic Reversal and the Monosyllable Rule in English, Liaison in French, Vowel Shortening in ChiMwiini, Tone Assignment and the Star Shift Rule in Japanese, as well as Word Initial Voicing Assimilation and the Reduction Rule in Quechua. They argue that all these rules are sensitive to the phonological phrasing, and φ-restructuring is optional in some languages (e.g., English and ­Italian); forbidden in some (e.g., French); and obligatory in others (e.g., ChiMwiini). Besides Nespor & Vogel (1986), more explorations have been done following the RBA,

30  Introduction and more examples of phonological rules applying within the phonological phrase domain have been found (Cho 1990; Condoravdi 1990; Kidima 1990; Hayes & Lahiri 1991; Frota 2000, among others). Different from the RBA, the EBA attempts to reduce the syntactic sensitivity of the mapping algorithm to a single property of syntactic phrase structure, namely left or right ends of heads or maximal projections. Elaborating on Chen’s (1985, 1987) study of Xiamen, Selkirk (1986) proposes the EBA and holds that “the syntax-phonology mapping can be defined simply by reference to the ends of syntactic constituents” (Selkirk 1986: 386). She suggests the following parameters for the mapping of syntactic structure onto prosodic structure. (25) End parameter settings (I) a. ]Word b. Word[ (II) a. ]Xmax b. Xmax[ (III) a. ]Xhead b. Xhead[ Three prosodic constituents are thus recognized: the prosodic word corresponding to ]Word or Word[, the phonological phrase corresponding to ]Xmax or Xmax[, and an intermediate item Selkirk calls a “small phonological phrase” corresponding to ]Xhead or Xhead[. Thus, there are four possible types of the creation of the phonological phrase domain according to (25II) and (25III). Selkirk (1986) provides French Liaison as an example of ]Xhead, tone sandhi in Xiamen and stress assignment in ChiMwiini as examples of ]Xmax, and tone sandhi in Ewe (cf. Clements 1978) as an example of Xmax[. No examples are provided by Selkirk for Xhead[ (see Chen 1990 for a supporting case in the Wenzhou dialect of Chinese). Let us take ChiMwiini as an example and see how the EBA works in the phonological phrasing. In ChiMwiini, stress assignment applies within the domain of the phonological phrase, which is delimited by the right edge of maximal projection ]Xmax, as exemplified in (26) (adapted from Selkirk 1986: 390). (26)

VP VP V NP pa(:)nzize cho:mbo ..........................................]Xmax [ ]φ ran vessel ‘he ran the vessel on to the rock’

NP

mwa:mba ................]Xmax [ ]φ rock

In (26), the verb pa(:)nzize and its complement cho:mbo are grouped into one phonological phrase domain, and the adjunct NP mwa:mba forms an independent phonological phrase on its own. The end-setting ]Xmax identifies two right ends of maximal projections and thus marks the right edges of two phonological phrases.

Introduction  31 Other works on phonological phrasing following the EBA include Hale & Selkirk (1987) for Papago, Selkirk & Tateishi (1988) for Japanese, Selkirk & Shen (1990) for Shanghai Chinese, and Cho (1990) and Kenstowicz & Sohn (1997) for Korean, among others. It is noteworthy that, although branchingness is an important parameter in the RBA, it plays no direct role in the definition of the phonological phrase in the original algorithms of the EBA, which is sensitive only to edges. This creates difficulties for languages like Italian, English, and Kinyambo where branchingness is relevant to phonological phrasing (see Bickmore 1990). Cowper & Rice (1987) and Bickmore (1990) have thus suggested adding the parameter [+/− branchingness] to the EBA. Within the framework of OT, Selkirk (1996, 2000) and Truckenbrodt (1995, 1999, 2002), among others, translate the EBA into a set of violable constraints. Thus, the formation of the phonological phrase domain is treated as the result of the evaluation of several candidate phonological phrases, which are formed on the basis of the syntactic structure and then evaluated by a ranked set of violable constraints. Selkirk (1996) integrates the EBA into the Generalized Alignment Theory (McCarthy  & Prince 1993) and recasts the parametric formulation of the EBA phonological phrasing in terms of the following two constraints in (27). (27) a. Align-R (XP, φ): the right edge of a syntactic XP is aligned with the right edge of a phonological phrase φ. b. Align-L (XP, φ): the left edge of a syntactic XP is aligned with the left edge of a phonological phrase φ. Constraints in (27) were later referred to as ALIGN-XP,L and ALIGN-XP,R or ALIGNXP (when both are jointly discussed) in Truckenbrodt (1995, 1999). In addition to the class of alignment constraints, four general constraints on prosodic domination entailed by the SLH are also proposed in Selkirk (1996), as we have seen in (18) in Section  1.4.3.1. Since the universal constraints of Exhaustivity, Nonrecursivity, and even Layeredness are taken to be violable, recursive phonological phrasing is allowed, which is strictly prohibited by the original SLH. Truckenbrodt (1995, 1999) shows that this is the case in Kimatuumbi, where a recursive phonological phrase has to be introduced. In order to account for the phonological phrasing in Kimatuumbi, Truckenbrodt (1995, 1999) proposes WRAP-XP as another constraint family of the syntax-phonology interface, as defined in (28). (28) WRAP-XP: each lexically headed XP must be contained inside a phonological phrase. WRAP-XP = For every XP, XP a projection of a lexical category, there is a phonological phrase φ, such that all terminal elements that are dominated by XP are also dominated by φ.

32  Introduction Truckenbrodt (1995, 1999) argues that ALIGN-XP, WRAP-XP, and the general violable constraint Nonrecursivity (henceforth NONREC) interact in the phonological phrasing. Consequently, the ranking of these constraints plays a key role in dealing with the differences in the phonological phrasing across languages. In Kimatuumbi, for instance, NONREC is ranked lower than WRAP-XP and ALIGNXP,R, since a recursive phonological phrase is allowed in this language. Thus, the ditransitive VP structure in (29a) would give the phonological phrasing in (29b) with the ranking ALIGN-XP,R, WRAP-XP >> NONREC in Kimatuumbi. By contrast, in Chicheŵa, WRAP-XP and NONREC should be ranked higher than ALIGN-XP (i.e., WRAP-XP, NONREC >> ALIGN-XP,R) in order not to create recursive phonological phrasing. Thus, (29a) would be mapped to the phonological phrase in (29c) in Chicheŵa. (29) a. [V NP NP]VP b. Kimatuumbi: [[V NP]φ NP]φ c. Chicheŵa: [V NP NP]φ Truckenbrodt (1995, 1999) also notices that a focused constituent in Chicheŵa is followed by a phonological boundary. To deal with this fact, he introduces another constraint ALIGN-FOC, as in (30), and argues that in Chicheŵa ALIGN-FOC has to override WRAP-XP since it enforces violations of WRAP-XP. (30) ALIGN-FOC = ALIGN (Foc, R; P, R) Each focused constituent is right-aligned with a phonological phrase φ boundary. Truckenbrodt (1999) claims that the constraint ALIGN (Foc, R; P, R) is active in Bengali (cf. Hayes & Lahiri 1991) as well. In Japanese and in dialects of Korean (cf. Jun 1993; Kenstowicz  & Sohn 1997; Kim 1997), according to Truckenbrodt, a phonological boundary should be inserted to the left of a focused constituent according to the “mirror image” of ALIGN (Foc, R; P, R), namely ALIGN (Foc, L; P, L). Besides the constraints discussed above, there are other constraints that have been proposed for prosodic phonology within the OT framework, such as BinMin (φ, ω); BinMax (φ, ω); *P-PHRASE; UNIFORMITY; and INCREASING UNITS, as mentioned in Section 1.4.1. The most recent development in the study of the syntax-phonology interface, namely the Match Theory advanced by Selkirk (2005, 2009, 2011), is actually developed from the EBA, similar to the Align/Wrap Theory discussed above. Sharing with the EBA the very limited appeal to syntactic information, the Match Theory requires that a syntactic constituent of type α and a phonological constituent of type π correspond to each other on both edges, which is formally recast as two sets of violable Match constraints within the framework of the Correspondence Theory (McCarthy & Prince 1995), as given in (31).

Introduction  33 (31) a. Match (α, π) [= S-P faithfulness] The left and right edges of a constituent of type α in the input syntactic representation must correspond to the left and right edges of a constituent of type π in the output phonological representation. b. Match (π, α) [= P-S faithfulness] The left and right edges of a constituent of type π in the output phonological representation must correspond to the left and right edges of a constituent of type α in the input syntactic representation. We can find that the Match Theory is a theory of constituent faithfulness, which expresses the Match constraints as correspondence constraints. According to the Match Theory, morphosyntactic notions, such as clause, phrase, and word, all have their correspondents in phonology, namely the intonational phrase, the phonological phrase, and the prosodic word respectively. Thus, in terms of the phonological phrasing, a syntactic-prosodic structure correspondence constraint Match (Phrase, φ) that calls for any syntactic phrase to have a corresponding phonological phrase is needed. Besides the faithfulness constraint Match (Phrase, φ), a prosodic markedness constraint, BinMin (φ, ω), which requires a phonological phrase to be minimally composed of two prosodic words (Selkirk 2000), is also necessary. The phonological phrasing in languages thus is obtained through the interaction of Match (Phrase, φ) with BinMin (φ, ω). For instance, Selkirk (2011) shows that in Xitsonga, BinMin (φ, ω) is ranked higher than Match (Phrase, φ), mapping the VP structure in (32a) to the phonological phrase in (32b) but not (32c), and the VP structure in (33a) to the phonological phrase in (33b) but not (33c) (examples are adapted from Selkirk 2011). (32)

a. b. c.

[V [N]NP] VP [V N]φ *[V [N]φ] φ

(33)

a. b. c.

[V [N A]NP] VP [V [N A]φ] φ *[[V N]φ A]φ

In (32), where the object NP is not branching, (32b) satisfies the higher-ranked BinMin (φ, ω) while (32c) violates BinMin (φ, ω). This is why (32b) is the optimal phonological phrasing of (32a), even though (32b) violates Match (Phrase, φ). In (33), where the object NP is branching, (33b) satisfies both BinMin (φ, ω) and Match (Phrase, φ) while (33c) violates Match (Phrase, φ), which is why (33b) is optimal. In some other languages, by contrast, the ranking of Match (Phrase, φ) and BinMin (φ, ω) might be reversed. For example, Selkirk (2011) notes that the opposite ranking Match (Phrase, φ) >> BinMin (φ, ω) accounts for the phonological phrasing in languages such as ChiMwiini and German, in which branchingness does not play a role in the phonological phrasing. Thus, we can find that the typological variation in phonological phrasing can be predicted by the ranking of Match (Phrase, φ) with respect to BinMin (φ, ω) within the Match Theory. In Section  1.4.3.3, we have reviewed the main approaches to phonological phrasing developed in the prosodic phonology theory. Both the EBA and the RBA

34  Introduction (including its original algorithms and recent developments as in the Align/Wrap Theory and the Match Theory) have been discussed in this section. We have seen that each of these approaches has been successful in predicting the phonological phrasing in some languages and thus accounts for a number of phonological phenomena relevant to the phonological phrase domain. In actual fact, however, there is great cross-linguistic variation of the phonological phrasing, and every language seems to have its own idiosyncratic way of constructing the phonological phrase domain, which is why there is still little agreement as to which approach is superior to the others in terms of the phonological phrasing (cf. Bickmore 1990; Chen 1990; Cho 1990). In this book, different approaches will be examined and compared with regard to the formation of the phonological phrase in the Fuzhou dialect, as to be discussed in Chapter 5. We will see that the Align/Wrap Theory and the Match Theory do not exhibit stronger explanatory power as compared to the RBA and the original parameterized algorithms of the EBA in accounting for the phonological phrasing in Fuzhou, which is why an alternative analysis is required.

1.5  Research structure and organization This book consists of eight chapters. Chapter 1, as we have seen above, discusses the purpose, methodology, data sources, and background information of this study. It presents a brief description of Fuzhou phonology, including its tonal and segmental inventories, syllable types, and major phonological rules that are relevant to this study. It also provides a critical survey of the prosodic phonology theory, introducing its history and recent development, its basic tenets, and outstanding issues that will be further explored in this book. The rest of this book, in which I develop an account of the Fuzhou phonological system within the framework of prosodic phonology, is organized as follows. From Chapter 2 to Chapter 7, I probe deeply into the definition and the formation of different prosodic domains in the Fuzhou dialect and phonological phenomena relevant to each domain. Major prosodic domains discussed in this book include the syllable, the foot, the prosodic word, the clitic group, the phonological phrase, and the intonational phrase. I will not particularly discuss the mora, since the Fuzhou dialect, like Mandarin Chinese, is not a quantity-sensitive language (cf. Chan 1985; Zhang 2014, 2017, among others). The mora, therefore, plays no direct role in the application of any phonological rules in Fuzhou. Although some previous studies claim that the mora is the tone bearing unit in this dialect (e.g., Wright 1983; Chan 1998), it will be shown in Chapter 2 that these analyses are problematic. I am not going to investigate the utterance in this book either, since no relevant phonological phenomena have been reported in the literature or observed by my informants or myself. In Chapter  2, I  examine the roles of the syllable and the foot in the Fuzhou dialect. I demonstrate in this chapter that the distribution of two groups of Fuzhou alternating finals (the FA rule) is sensitive to the domain of the syllable, and hence the syllable should be treated as an indispensable constituent in the prosodic

Introduction  35 hierarchy of this dialect. In addition, the discussion in Chapter  2 shows that, although the foot is a universal prosodic unit across languages, no independent evidence and valid arguments can be provided to demonstrate its existence and its role as a domain of rule application in the Fuzhou dialect. Chapter 3 focuses on the prosodic word. Major types of morphosyntactic words in Fuzhou are examined with respect to the application of various phonological rules. What I try to do in this chapter is to prove that it is the prosodic word, rather than the foot, that constitutes the domain of application for TS, MTS, FC, and CL at the lexical level in the Fuzhou dialect. This chapter also shows that the application of these rules within the domain of the prosodic word is conditioned by the morphological processes of morphosyntactic word formation. Chapter  4 is a formal analysis of enclitics and the clitic group composed of the host plus the enclitic (Type A clitic group) in the Fuzhou dialect. It provides a detailed analysis of the morphosyntactic properties and phonological behavior of Fuzhou enclitics and shows that the TS rule and the FC rule are obligatorily blocked between the host and the enclitic within the Type A clitic group domain, whereas the application of CL between the host and the enclitic is mandatory. This chapter also discusses the violation of the SLH within the Type A clitic group in the Fuzhou dialect and argues that a weakened SLH may be required. Chapter 5 concerns itself with the phonological phrase, with its focus on the long-standing issue regarding the application of the TS rule at the phrasal level in Fuzhou. I will demonstrate that the complicated tone sandhi phenomena at the phrasal level in Fuzhou can be accounted for with the notion of the phonological phrase, and the phonological phrase domain in the Fuzhou dialect should be defined by employing an alternative approach that combines both the EBA and the RBA and allowing the SLH to be violated. This chapter also shows that there are some “no look-back” phenomena within the phonological phrase domain in the Fuzhou dialect, which can be accounted for with the Domain Impenetrability Condition (DIC). The application of FC and CL at the phrasal level is also discussed in this chapter on the basis of the definition of the phonological phrase in this dialect and the notion of DIC, and it will be shown that CL exhibits a different degree of application within the phonological phrase domain as compared to TS and FC. In Chapter 6, I discuss proclitics and the clitic group composed of the proclitic plus the host (Type B clitic group) in the Fuzhou dialect. I will show that the Type B clitic group in Fuzhou exhibits different phonological behavior from the Type A clitic group, but the clitic group in Fuzhou as a whole can be distinguished from other prosodic domains in terms of rule application. This chapter also discusses the violation of the SLH within the Type B clitic group, based on which I will again argue for a slight weakening of the SLH in the theory of prosodic phonology. Chapter 7 investigates the formation and the restructuring of the intonational phrase in Fuzhou and analyzes Fuzhou phonological phenomena that are pertinent to the intonational phrase domain. This chapter shows how semantic and pragmatic factors, such as length, rate of speech, style, and contrastive prominence,

36  Introduction would influence the formation and the restructuring of the intonational phrase domain in Fuzhou and affect the application of phonological rules in this dialect. Chapter 8 is the concluding chapter of this book. This chapter presents a complete model of prosodic hierarchy in the Fuzhou dialect, summarizes the discoveries and conclusions drawn in this study regarding the Fuzhou dialect and the prosodic phonology theory, and outlines tasks for future research. This book is an extended and revised version of my Ph.D. dissertation, Prosodic phonology of the Fuzhou dialect, submitted to the University of Wisconsin– Madison in 2017. The discussion in Chapter  4 and Chapter  6 is a further development of my 2018 Chinese language paper (You 2018b) “Lun Fuzhou fangyan de nianfu chengfen yu nianfuzu 论福州方言的粘附成分与粘附组 [Clitics and the clitic group in the Fuzhou dialect]”, published in the International Journal of Chinese Linguistics (Volume 5, Number 1, pp. 124–165), and my 2019 paper (You 2019) “Enclitics and the clitic group consisting of ‘host + enclitic’ in the Fuzhou dialect”, included in Zhang & Qian (eds.) Prosodic Studies: Challenges and Prospects (Routledge, pp. 80–107).

Notes 1 Li’s (1998) dictionary is a Fuzhou-English dictionary based on the dictionary compiled by Li et al. (1994). 2 Li (1998) suggests that the discrepancy between 33 and 31 may be due to generational differences among the authors. 3 The final iau is not reported in some published sources (e.g., Li et al. 1994; Lin 2002) since only a couple of syllables contain this final, e.g., 喵 [miau51] ‘meow’. 4 Other alternating finals, such as [a/ɑ], [o/ɔ], and [ie/iɛ], are also reported in some published materials (e.g., Li et al. 1994; Feng 1998; Lin 2002). 5 The glides in the Fuzhou dialect should be more accurately transcribed as [j], [w], and [ɥ]. In this book, for the sake of brevity, I use [i], [u], and [y] to stand for these three glides respectively. 6 This article is commonly quoted as Selkirk (1978) or Selkirk (1981). The content of this article was first presented at an Amherst conference in 1978, while the first published version is the text that appeared in the proceedings of the Nordic Prosody conference in 1981. 7 In this book, I  use the term “prosodic word”, instead of the “phonological word”. These two are synonymous and interchangeable in the theory of prosodic phonology. It should be pointed out that the term “phonological word” actually had been used before the theory of prosodic phonology was developed (e.g., Dixon 1977a). 8 Hayes (1984) is an unpublished paper, which was later published in 1989. 9 Notice that in Selkirk’s (1978/1981) initial prosodic hierarchy, there is only one level of phonological phrase. This is held in Nespor & Vogel’s (1986) version as well. But some subsequent works indicate that a larger variety of phrase types may be motivated. Selkirk (1986) and Selkirk & Tateishi (1988), distinguish a Major Phrase and a Minor Phrase. Beckman & Pierrehumbert (1986) and Pierrehumbert & Beckman (1988) distinguish two as well, referring to them as the accentual phrase and the intermediate phrase respectively. 10 Thanks to Prof. Hongming Zhang (pers.comm.) for pointing out this to me. 11 It is noteworthy that the mora is placed in shaded boxes in Figure 1.4. As a unit that determines whether a syllable is “light” or “heavy”, mora plays a role in the phonology of some quantity-sensitive languages such as Japanese and Sanskrit. Mandarin

Introduction  37

12

13 14

15

Chinese, however, is not a quantity-sensitive language (see Chan 1985; Zhang 2014, 2017, among others). Hence, the mora is not considered as an active prosodic constituent in Mandarin Chinese. In addition, notice that the prosodic constituent foot is not presented in Zhang’s (2014, 2017) hierarchy in Mandarin Chinese. According to Zhang (2014, 2017), the foot does not exist in Mandarin Chinese due to the lack of binary metrical contrast between syllables in this language. In Nespor & Vogel’s (1986) terminology, a directional clitic refers to a clitic, which is phonologically dependent on an element to the left or right according to its own inherent property. A plain/nondirectional clitic, in contrast, refers to a clitic that finds its host either to the right or to the left. Thanks to Prof. Hongming Zhang (pers.comm.) for discussion. In accordance to the original SLH, Nespor & Vogel (1986) have to assign the clitic group status to every prosodic word in the definition of the phonological phrase, since the clitic group, rather than the prosodic word, is immediately dominated by the phonological phrase in the prosodic hierarchy. Since I assume the weakened SLH in this book, which allows the violation of Exhaustivity, the “clitic group” in (21) is actually better to be understood as “clitic group or prosodic word”. In Nespor & Vogel (1986), only V, N, and A are considered lexical heads.

2 Syllable and foot in the Fuzhou dialect

Neither the syllable nor the foot is a new concept in phonological theories, and there has been a very large body of research related to these two prosodic constituents. In this chapter, I  will restrict my attention to the roles of these two constituents in the prosodic phonology of the Fuzhou dialect. Section 2.1 concentrates on the syllable as the prosodic domain of rule application, and we will see that the final alternation rule (FA) in Fuzhou operates within the syllable but not in other contexts, which is why the syllable is an indispensable constituent in the prosodic hierarchy of this dialect. Section 2.2 focuses on the foot. It begins with a brief introduction to the foot as the domain of rule application across languages and then discusses some previous studies on the foot in the Fuzhou dialect. Based on the review of previous studies and the analysis of relevant Fuzhou data, I will argue that the foot should not be treated as a prosodic constituent in this dialect. Section 2.3 closes this chapter with some concluding remarks.

2.1  Syllable: the domain of FA in the Fuzhou dialect The syllable, as we have seen in Figure 1.3, is the intermediate unit of the rhythmbased hierarchy in the prosodic hierarchy, which is located above the mora and below the foot. It has long been reported in the literature that the syllable serves as the domain of application for phonological rules across languages (see Kahn 1976; Kiparsky 1979; Booij 1981; Hulst 1984; Nespor  & Vogel 1986; Jensen 1993, among others). Let us consider, for example, Glottalization in American English. Glottalization is a rule that glottalizes a voiceless stop when this stop is preceded by a [−consonantal] segment, which is illustrated by the examples of the voiceless alveolar stop [t] in (1). As we can see in (1), for this rule to apply, [t] must be: (a) in absolute final position, as in (1a–c); (b) followed by a consonant other than [r] within the same word, as in (1d–f); or (c) followed by a word that begins with a consonant or a glide, as in (1g–i) (cf. Kahn 1976; Nespor & Vogel 1986; Jensen 1993, among others). Segments in question are marked in bold in (1).

Syllable and foot in the Fuzhou dialect  39 (1)

a. b. c. d. e. f. g. h. i.

wait great giant butler witness atlas wait patiently wait reluctantly wait wearily

[weitʔ]σ [ɡreitʔ]σ [ʤai]σ [əntʔ]σ [bətʔ]σ [lər]σ [wɪtʔ]σ [nɪs]σ [ætʔ]σ [ləs]σ [weitʔ]σ [phei]σ . . . [weitʔ]σ [rə]σ . . . [weitʔ]σ [wɪə]σ . . .

We can find that what the above-mentioned three types of environments have in common is that the voiceless stop [t] occurs in syllable-final position. Glottalization can thus be formulated as a syllable-domain rule, as presented in (2). (2)

Glottalization in American English (Nespor & Vogel 1986) t → tʔ/ [ . . . [-cons] ___] σ

Another typical syllable-domain rule is Schwa Insertion in certain varieties of Dutch, which inserts a schwa between a liquid and a following non-coronal obstruent (cf. Booij 1981; Trommelen 1983; Hulst 1984; Nespor & Vogel 1986, among others). This rule occurs when the liquid and the non-coronal obstruent are in the same syllable, as exemplified in (3). The rule can be formulated as in (4), in which “L” represents a liquid. (3)

a. b. c.

park: [park]σ → par[ə]k ‘park’ helpster: [help]σ [ster]σ → hel[ə]pster ‘helper (fem.)’ melkachtig: [melk]σ [ach]σ [tig]σ → mel[ə]kachtig ‘milklike’

(4)

Schwa Insertion in Dutch (Nespor & Vogel 1986) Ø → ə/[ . . . L ___ [-cor] C0]σ

Another syllable-domain rule worthy of being mentioned here is Emphasis in Arabic. This rule stipulates that all segments within a syllable receive emphasis if there is an emphatic consonant in this syllable (cf. Hulst & Smith 1982; Nespor & Vogel 1986). This can be illustrated by the examples in (5), where emphasis is indicated by a dot under the emphatic segments. (5)

a. b. c. d. e.

ṛab: ṭiin: ṛaagil: bukra: ṛaaʕiḍ:

[ṛab]σ [ṭiin]σ [ṛaa]σ [gil]σ [buk]σ [ṛa]σ [ṛaa]σ [ʕiḍ]σ

→ → → → →

[ṛạḅ] [ṭịịṇ]σ [ṛạạgil] [bukṛạ] [ṛạạʕịḍ] ̣

‘lord’ ‘mad’ ‘man’ ‘tomorrow’ ‘military rank’

It is noteworthy that this rule is an autosegmental spreading rule applying within the syllable domain, which indicates that both segmental and suprasegmental phonological rules can refer crucially to the syllable as the domain of application.

40  Syllable and foot in the Fuzhou dialect Besides the rules discussed above, other well-known phonological rules that make reference to the syllable as their domain of application include Alveopalatalization in English, Velarization and r-Strengthening in Spanish, Consonant Gradation in Finnish, and so forth (see Kahn 1976; Harris 1983; Prince 1984; Keyser & Kiparsky 1984; Nespor & Vogel 1986; Jensen 1993, among others). Similar to these syllable-domain rules across languages, Final Alternation (FA) in the Fuzhou dialect also applies within the syllable domain. In the following, I will argue that this rule is only sensitive to the syllable domain, which makes it different from other phonological rules in Fuzhou and which shows that the syllable is an indispensable prosodic unit in this dialect. As mentioned in Chapter  1, there are two types of finals in the Fuzhou dialect, namely non-alternating finals that can co-occur with all tones and alternating finals that show alternations depending on the tone. Alternating finals can be further divided into two groups, Group A and Group B, according to the tones they co-occur with, as presented in Table 2.1, which reprints Table 1.3. When occurring in the non-terminal position in a sandhi environment (i.e., sandhi position), Group A  variants of alternating finals retain their citation forms, whereas Group B variants must be changed into corresponding Group A variants, since tones that co-occur with Group B variants, namely tones 213, 242, and 23, always undergo tone sandhi processes and thus never appear in a sandhi position. The aforementioned phenomenon of Fuzhou alternating finals is in fact a twofold problem, since the alternations of finals arise in both citation-tone syllables and the sandhi context. Comparing these two cases, we can find that alternations of finals in citation forms and those in the sandhi environment actually refer to different prosodic domains. To distinguish these two types of phonological phenomena in the Fuzhou dialect, the first type is referred to as Final Alternation (FA), and the second is referred to as Final Change (FC) in this book, as we have seen in Chapter 1. The domain of application for FC is by no means the syllable, since it occurs only in the context that is composed of two or more syllables, similar to TS, MTS, and CL. By contrast, the FA rule can be viewed as a well-formedness condition or phonological constraint on Fuzhou syllables with alternating finals, or more specifically, on the combination of the alternating final and the tone the syllable Table 2.1  Fuzhou alternating finals

Alternating finals

Tones

Group A

Group B

Group A

Group B

i

ei

iŋ iʔ

eiŋ

eiŋ

aiŋ

eiʔ

eiʔ

aiʔ

u

ou

uŋ uʔ

ouŋ

ouŋ

auŋ

ouʔ

ouʔ

auʔ

y

øy

øy

oy

yŋ

øyŋ

øyŋ

oyŋ

yʔ

øyʔ

øyʔ

oyʔ

44, 51, 31, 5

213, 242, 23

44, 51, 31, 5

213, 242, 23

Syllable and foot in the Fuzhou dialect  41 assumes. We have seen some examples of FA in Chapter 1, reproduced as in (6a– b). Additional examples are provided in (6c–d). (6)

a. uŋ (uʔ) vs. ouŋ (ouʔ) b. i (iʔ) vs. ei (eiʔ) 风 xuŋ44 ‘wind’ 机 ki44 ‘machine’ 51 云 xuŋ ‘cloud’ 旗 ki51 ‘flag’ 粉 xuŋ31 ‘powder’ 己 ki31 ‘shrewish’ 训 xouŋ213 ‘to teach, to train’ 记 kei213 ‘to remember’ 忌 kei242 ‘death anniversary’ 混 xouŋ242 ‘to get by’ 拂 xouʔ23 ‘to flick’ 急 keiʔ23 ‘rapid; impatient’ 佛 xuʔ5 ‘Buddha’ 扱 kiʔ5 ‘to stab, to prick’ c. eiŋ (eiʔ) vs. aiŋ (aiʔ) d. øyŋ (øyʔ) vs. oyŋ (oyʔ) 针 tseiŋ44 ‘needle’ 朦 møyŋ44 ‘fluffy’ 51 层 tseiŋ ‘floor’ 芒 møyŋ51 ‘awn’ 31 剪 tseiŋ ‘to cut’ 蠓 møyŋ31 ‘midge’ 213 荐 tsaiŋ ‘straw mat’ 梦 moyŋ213 ‘dream’ 242 赠 tsaiŋ ‘to give as a present’ 网 moyŋ242 ‘fishing net’ 23 汁 tsaiʔ ‘juice’ 抹 moyʔ23 ‘to swipe’ 5 截 tseiʔ ‘to cut off’ 目 møyʔ5 ‘eye’

From Table 2.1 and the examples in (6), we can find that the FA rule actually imposes a constraint on the combination of alternating final variants and tones within a syllable. Specifically, within the syllable domain, although both Group A  and Group B variants of alternating finals can occupy the syllable final position, their distribution is conditioned by FA in such a way that Group A variants only occur in syllables bearing tones 44, 51, 31, and 5, while Group B variants only occur in those assuming tones 213, 242, and 23. A syllable with an alternating final can be contained in a larger prosodic constituent, and there are three scenarios of the occurrence of alternating finals beyond the syllable boundary. First of all, in non-terminal positions within certain strings that are composed of more than one syllable, such as those in (7), Group B variants are never allowed in the surface form. For the sake of brevity, only TS and FC are presented in (7) and all the following examples in this chapter. Tones and finals in question are marked in bold. (7)

a. b. c. d. e. f.

动作 touŋ242 tsouʔ23 裤头 khou213 thau51 护士长 xou242 søy242 tuoŋ31 旧书 kou242 tsy44 七块 tsheiʔ23 toy213 办护照 paiŋ242 xou242 tsiu213

‘action’ → tuŋ51 tsouʔ23 ‘trouser waist’ → khu44 thau51 ‘head nurse’ → xu21 sy51 tuoŋ31 ‘old book’ → ku44 tsy44 ‘seven pieces’ → tshiʔ44 toy213 ‘to apply for a passport’ → peiŋ21 xu51 tsiu213

*touŋ51 tsouʔ23 *khou44 thau51 *xou21 søy51 tuoŋ31 *kou44 tsy44 *tsheiʔ44 toy213 *paiŋ21 xou51 tsiu213

42  Syllable and foot in the Fuzhou dialect From (7), we can find that, in the underlying form of these examples, Group B variants of alternating finals occur together with tones 213, 242, or 23, within syllables in non-terminal positions, which is required by the FA rule. By contrast, in the surface form of these examples, tones 213, 242, or 23 assumed by nonterminal syllables all undergo tone sandhi, along with which Group B variants of alternating finals in these syllables are all switched into corresponding Group A  variants. Clearly, the TS rule and the FC rule both apply within the domain that is composed of more than one syllable in (7), thus causing Group B variants in non-terminal syllables to be replaced by their Group A counterparts in the surface form. In other words, in cases like (7), it is the rule application of TS and FC that prevents the occurrence of Group B variants in the surface form of non-terminal syllables within a domain larger than the syllable. This is not decided by the FA rule. The effects of FA are limited to the combination of Group B variants and tones 213, 242, or 23 within the syllable boundary, which serves as the underlying form when the derivation has moved to the larger domain in the examples in (7). The occurrence of Group B variants of alternating finals in non-terminal syllables is not always disallowed in strings consisting of more than one syllable, as can be seen in (8). We can find that in the surface form of these examples, both Group A and Group B variants of alternating finals can occur in non-terminal positions. (8)

a. b. c. d.

糊其 旧其 灯侈 店侈

ku51 ki0 kou242 ki0 teiŋ44 sɛ242 taiŋ213 sɛ242

→ → → →

ku51 ki0 kou242 ki0 teiŋ44 sɛ242 taiŋ213 sɛ242

‘glutinous (things)’ ‘old (things)’ ‘lights are many’ ‘stores are many’

(Group A) (Group B) (Group A) (Group B)

It is noteworthy that neither TS nor FC applies to non-terminal syllables in the domain formed by the strings in (8). Hence, it is the non-application of TS and FC that results in the occurrence of Group B variants of alternating finals in the non-terminal position within the larger domain. In these cases, the FA rule does play a role, but its application is only triggered within the syllable boundary. When the derivation in these examples has moved outside the syllable domain, no phonological rules apply to non-terminal syllables containing alternating finals in the larger domain, so both Group A and Group B variants are allowed to surface. The third scenario of the occurrence of alternating finals beyond the syllable boundary can be seen in (9). We can find that in the surface form of the examples in (9), both Group A and Group B variants of alternating finals can appear in the terminal syllables within a domain larger than the syllable. (9)

a. b. c. d.

实习 样式 开灯 开店

siʔ5 siʔ5 yoŋ242 seiʔ23 khui44 taiŋ44 khui44 teiŋ213

→ → → →

siʔ31 siʔ5 yoŋ51 seiʔ23 khui44 taiŋ44 khui51 teiŋ213

‘intern’ ‘style’ ‘to turn on the light’ ‘to open the store’

(Group A) (Group B) (Group A) (Group B)

As we have seen in Chapter  1, neither TS nor FC is triggered on the terminal syllables within the domain containing two or more syllables, which is why Group B variants of alternating finals in (9) are retained in the terminal position in the surface form, similar to the Group A variants. The FA rule, again, determines

Syllable and foot in the Fuzhou dialect  43 the distribution of alternating finals with respect to the tones within the syllable domain, but it plays no role in the domain larger than the syllable. To sum up, although syllables with alternating finals in the Fuzhou dialect can be contained in a larger prosodic domain, Group A  variants are different from Group B variants in terms of their occurrence beyond the syllable boundary. When contained in a larger domain, Group A  variants are always allowed, in both terminal and non-terminal positions. By contrast, although Group B variants are allowed in the terminal position, their occurrence in the non-terminal position depends on whether the application of TS and FC is triggered in the larger domain. Specifically, as we have seen, Group B variants in the non-terminal position are all replaced by their Group A counterparts when TS and FC apply, while when TS and FC do not apply they are retained in the non-terminal position. In any case, however, the occurrence or non-occurrence of Group B variants in a domain larger than the syllable is not determined by the FA rule. The FA rule, as a well-formedness condition on the combination of alternating finals and tones, confines its application within but not across the syllable boundary and only applies to syllables with alternating finals (conditioned by syllable structure). No matter whether alternating finals are changed in a larger domain, FA applies first within the syllable. After the alternating finals and their corresponding tones have been combined within the syllable as required by FA and the derivation has moved to the larger domain, the combination of alternating finals and their corresponding tones becomes the underlying form of rule application, and FA no longer has any effect. Therefore, FA in the Fuzhou dialect is simply a well-formedness condition or phonological constraint bound to the syllable domain, and it is not sensitive to any larger prosodic domains, which makes it distinct from other phonological rules in Fuzhou, such as TS and FC. As the only domain of rule application for FA, the syllable should be treated as a prosodic constituent in the prosodic phonology of this dialect.

2.2 Foot as a domain of rule application in Fuzhou: a good choice or not? 2.2.1  Introduction: the foot as a prosodic domain across languages Now, let us discuss the role of the foot in the prosodic phonology of the Fuzhou dialect. As we have seen in Chapter 1, in the universal prosodic hierarchy, syllables are grouped into feet. The foot, in general terms, is defined on the basis of stress, an abstract relation of prominence between syllables (cf. Liberman 1975; Liberman  & Prince 1977; Hayes 1980; Kenstowicz 1994; Kager 1996, 2007; Gussenhoven & Jacobs 1998; Gordon 2011; Zhang 2014, 2017, among others). Nonetheless, stress is not the only phonological phenomenon that is related to the foot. It has long been noticed that it is the foot that forms the domain of application for a number of phonological rules in the world’s languages (Kiparsky 1979; Selkirk 1980b; Hulst & Smith 1982; Nespor & Vogel 1986; Jensen 1993, among others).

44  Syllable and foot in the Fuzhou dialect Take the Aspiration of voiceless stops in English as an example. From the examples in (10) and (11), where our attention is limited to the voiceless stop t, we can find that t in English is aspirated if and only if it occurs in the footinitial position, but not in other positions within the foot (cf. Nespor & Vogel 1986; Jensen 1993). The aspiration of t can thus be formulated as in (12), which indicates that this rule applies only when t is placed at the left end of the foot domain. (10) a. b. c. d. e.

time tuna toucan detain tree toad

(11) a. b. c. d. e.

sting after abstain austere night owl

→ → → → → → → → → →

[th]ime [th]una [th]oucan de[th]ain [th]ree [th]oad *s[th]ing *af[th]er *abs[th]ain *aus[th]ere *nigh[th] owl

[time]Σ [tuna]Σ [tou]Σ [can]Σ [de]Σ [tain]Σ [tree]Σ [toad]Σ [sting]Σ [after]Σ [ab]Σ [stain]Σ [au]Σ [stere]Σ [night]Σ [owl]Σ

(12) Aspiration t→[+spread glottis]/ [___ . . .] Σ Another typical foot-bounded rule in English is n-Velarization. This rule obligatorily velarizes an n to a following velar stop, as can be seen in words like ink in (13a) and increment in (13b), where the nasal n and the velar stop are adjacent in the same foot. By contrast, this rule is optional in cases like ìncréase (V) in (13c) and íncrèase (N) in (13d), where n and the following velar stop appear in separate feet (cf. Kiparsky 1979; Nespor & Vogel 1986; Jensen 1993). (13) a. b. c. d.

ink increment ìncréase (V) íncrèase (N)

→ → → →

i[ŋ]k i[ŋ]crement i[n]crease/i[ŋ]crease i[n]crease/i[ŋ]crease

[ink]Σ [increment]Σ [in]Σ [crease]Σ [in]Σ [crease]Σ

Nasalization in Applecross Gaelic is another rule that can be stated within the foot domain. This rule applies within the foot domain from a stressed nasal vowel forward until the end of the domain is reached and backward to and including the consonantal onset of the stressed syllable, as can be seen in (14a). Within the domain, it is blocked if and only if it reaches a stop, as in (14b), or one of the vowels /e/, /o/, or /ə/, as in (14c). This rule fails to extend beyond the domain boundary, namely the end of the foot, as can be seen in (14d) (cf. Ternes 1973; Hulst  & Smith 1982; Nespor  & Vogel 1986, among others).

Syllable and foot in the Fuzhou dialect  45 (14)

a. b. c. d.

/ʃɛnɛ.var/ ́̃ ́̃ /strai.γ/ ́̃ /ma.riçən/ /khɔ + via.t/ ́̃

→ → → →

̃̃ ̃ [ʃɛñɛ.ṽãr] ̃Σ [strãi.γ̃ ̃ ́ ̃ ]Σ [mã.rĩçən] ́ ̃ ̃ Σ [khɔ]Σ [ṽiã.t] ́̃ Σ

‘grandmother’ ‘string’ ‘mothers’ *[khɔ]̃ Σ [ṽiã.t] ́̃ Σ

(throughout the Σ) (blocked by a stop) (blocked by /ə/) ‘how much’

The foot not only serves as the domain of application for phonological rules, but also provides the context for phonotactic restrictions in some languages. For example, in Žuǀ’hõasi, a Namibian language, there is a strong phonotactic restriction within the foot, which stipulates that if a consonant occurs in intervocalic position within the foot, this consonant may only be one of the following four: b, m, r, or n (Smith 1986; Nespor & Vogel 1986). This restriction can be exemplified in (15), where consonants in question are marked in bold. (15) a. c.

[pabu] [ǀoara]

‘pumpkin’ b. ‘complete’ d.

[tama] [nǂoana]

‘kindly’ ‘tell’

When a consonant occurs intervocalically at a juncture of two feet, it is not necessarily one of the four consonants mentioned above. In this case, other consonants in the consonant inventory of this language can be found between vowels, as shown in (16). (16) a. c.

[guma]Σ [guma]Σ [kx’ũ]Σ [kx’uni]Σ

‘whisper’ b. ‘move’ d.

[ǀxana]Σ [ǀxana]Σ [ǂ’ãa]Σ [ǂ’ama]Σ

‘drill’ ‘gather’

On the basis of the phonological phenomena we have seen above, we can find that the foot can serve as the domain of application for various phonological phenomena (including phonological rules and phonotactic constraints) across languages. Hence, the foot should be established as an indispensable prosodic constituent in the universal prosodic hierarchy. However, despite the fact that the foot is part of the universal prosodic hierarchy, it is not necessarily the case that the foot must exist in all human languages. As Zhang (2017) points out, any human language can have all of its prosodic constituents found in the universal prosodic hierarchy, but not all of the human languages need necessarily have all of the prosodic constituents in the universal hierarchy. The foot serves as the domain of rule application in languages like English, Applecross Gaelic, and Žuǀ’hõasi, but this does not mean that it must be an essential prosodic unit in other languages such as Mandarin Chinese and the Fuzhou dialect. Whether a given prosodic unit exists and plays a role in a particular language is an empirical issue, which is why evidence from the Fuzhou dialect is required if we want to establish the foot as a domain of rule application in this dialect. In the following subsections, I will examine the foot in the prosodic phonology of the Fuzhou dialect. I will address the issue whether the foot functions as the domain of application for phonological rules in Fuzhou and whether the foot should thus be established as a prosodic constituent in this dialect. Section 2.2.2 presents a critical review of previous studies on the foot in the Fuzhou dialect,

46  Syllable and foot in the Fuzhou dialect based on which I will argue that all of these studies contain some theoretical and/or empirical problems. Section 2.2.3 provides additional evidence against the foot as the prosodic domain in the Fuzhou dialect. Section 2.2.4 offers the conclusion that to establish the foot as a domain of rule application in Fuzhou is not a valid choice. 2.2.2  Previous studies on the foot in the Fuzhou dialect Some scholars contend that it is the foot in the Fuzhou dialect that forms the domain of application for phonological rules such as TS and FC (Wright 1983; Chan 1985, 1998; Shih 1986; Hung 1987, among others). In this section, I will review and discuss some of the most outstanding previous studies on the foot in the Fuzhou dialect, and we will see that they are all faced with some theoretical and/or empirical problems. 2.2.2.1  Wright (1983) Spectrographic studies in Wright (1983) show that there is a significant difference in syllable duration between syllables in the non-terminal position (sandhi position) of a given sandhi domain and syllables in the terminal position (nonsandhi position) in the same environment. The former, namely those bearing sandhi tones, are regularly reduced greatly in duration as compared to their isolation duration while the latter are only somewhat reduced in length. This observation leads Wright to relate stress to duration and consider the Fuzhou disyllabic sandhi domain to be a w-s (weak-strong) iambic foot. Specifically, she refers to the sandhi position of the domain as the weak position, and the non-sandhi position as the strong position. In the trisyllabic sandhi span, Wright constructs binary w-s feet with the possibility of a single extrametrical element at the boundary of the domain, i.e., w-w-s super-feet. She limits the length of the foot in the Fuzhou dialect to a maximum of three syllables. Wright (1983) also assumes that the tone bearing unit in the Fuzhou dialect is mora and that syllables are associated with two moras in strong or isolation position (according to Wright, the only exception is syllables bearing the yangru tone, which is associated with one mora since they end with a glottal stop). She further proposes that in a w-s foot, the first mora in the weak-position syllable is deleted, which gives rise to the structure for disyllabic domains in (17) (m = mora, w = weak, s = strong). (17)

Foot σw m ↓ Ø

σs m

m

m

Syllable and foot in the Fuzhou dialect  47 Wright (1983) argues that, once the mora is deleted, the tone and the syllable in the weak position will be affected, and she thus bases her analyses of the TS rule and the FC rule in the Fuzhou dialect on the w-s/right-strong stress foot formation as well as the moraic timing device. Wright’s analysis has two theoretical drawbacks. First of all, she confuses the relationship between the application of Fuzhou phonological rules (i.e., TS and FC) and the foot formation. It is true that, in a given sandhi domain, the syllable in the sandhi position is shorter in duration than its own isolation duration as well as the duration of the syllable in the non-sandhi position. A key question is what causes the shortening of syllable duration in the sandhi domain? Now that the duration of a given syllable is greatly reduced only when it is in the sandhi position of a sandhi context, it is very natural to assume that the duration shortening is triggered only by virtue of the application of phonological rules such as TS and FC. Therefore, it would be more reasonable to argue that it is the application of phonological rules that leads to the contrast in duration, and thus the contrast in strength (weak vs. strong) between syllables in Wright’s analysis, but not vice versa. Hence, Wright’s definition of the w-s/right-strong stress foot in the Fuzhou dialect is actually based on the application of phonological rules. Beginning with Liberman & Prince (1977), it has been realized that stress is not a phonological feature that is given some content by the phonetic implementation rules (Gussenhoven & Jacobs 1998). Instead, stress should be considered as an abstract relational property between syllables, which is presented by metrical prominence relations between constituents in hierarchical structures (cf. Liberman 1975; Liberman & Prince 1977; Hayes 1980; Kenstowicz 1994; Kager 1996, 2007; Gussenhoven & Jacobs 1998; Gordon 2011; Zhang 2014, 2017, among others). Due to the fact that there is no unique/unambiguous phonetic correlate corresponding to stress, it has been argued that the evidence of primary importance that can determine the stress properties of a given syllable is native speakers’ perceptions and judgments (cf. Kenstowicz 1994; Kager 1996, 2007; Gordon 2011, among others). Although the duration contrast between syllables in the sandhi environment in the Fuzhou dialect can be perceived by native speakers, the contrast is arguably the byproduct of the application of the TS rule and/or the FC rule. In contexts without the application of phonological rules, however, no such contrast or metrical prominence relations between syllables can be perceived. In other words, there is no independent evidence for the existence of stress in this dialect. Wright’s “stress” and “foot” in the Fuzhou dialect thus exist only when Fuzhou phonological rules apply. She analyzes the application of Fuzhou phonological rules in terms of “stress” and “foot” defined without independent evidence, which is typically an example of circular reasoning. The second theoretical drawback of Wright’s analysis lies in her assignment of moras to syllables. In her analysis, syllables in the terminal position of a sandhi domain or in their isolation forms are composed of two moras, while those in the non-terminal position have only one mora. However, the term “mora” is used to count the syllable weight instead of syllable duration—heavy syllables consist of two moras, while light syllables have one mora. The distinction between heavy syllables

48  Syllable and foot in the Fuzhou dialect and light syllables resides in the internal syllable structure. Long vowels and vocalic diphthongs are always bimoraic, while short vowels are underlyingly monomoraic; coda consonants are mora-bearing on a language-specific basis (cf. Hayes 1989; Davis 2011, among others). Following Wright’s analysis, syllables that have two moras (i.e., heavy syllables) are supposed not to occur in the non-terminal position in the sandhi context. However, data from the Fuzhou dialect show that heavy syllables can occur in the non-terminal position in the sandhi environment, whether they contain alternating finals or not. From the examples in (18) and (19), where finals in question are marked in bold, we can find that syllables in the non-terminal position are heavy in weight and thus should have two moras, although they may be reduced in duration. For the sake of brevity, only TS and FC are presented here. (18) Heavy non-terminal syllables with non-alternating finals a. 桌骹 toʔ23 kha44 → toʔ44 kha44 ‘table leg’ b. 看见 khaŋ213 kiɛŋ213 → khaŋ51 kiɛŋ213 ‘to see’ c. 派头弟 phuai213 thau51 tiɛ242 → phuai51 thau21 tiɛ242 ‘stylish young man’ (19) Heavy non-terminal syllables with alternating finals a. 动作 touŋ242 tsouʔ23 → tuŋ51 tsouʔ23 ‘action’ b. 粽箬 tsoyŋ213 nuoʔ5 → tsøyŋ44 nuoʔ5 ‘rice dumpling wrappings’ c. 虱母 saiʔ23 mo31 → seiʔ24 mo31 ‘(big) louse’ Wright also wrongly excludes syllables bearing the yangru tone from having two moras. If we assume the glottal stop coda to be mora-bearing in the Fuzhou dialect, syllables with the yangru tone must all have two moras. Even if the glottal stop coda is non-moraic in this dialect, there are still a number of examples of heavy syllables with the yangru tone, e.g., 密 [meiʔ5] ‘dense’, 学 [xouʔ5] ‘to learn’, and 目 [møyʔ5] ‘eye’. Therefore, Wright’s moraic analysis is based on a false understanding of the term “mora” and is thus untenable. Moreover, Wright’s analysis has an empirical problem. In Wright’s analysis, the upper limit of foot length is three syllables. Hence, even if we give Wright the benefit of the doubt and assume that the foot is the domain of application for TS and FC in Fuzhou, there are still a lot of data that Wright’s analysis fails to cover, as exemplified by quadrisyllabic words in (20), where tones and finals in question are marked in bold. (20) a.

拨浪鼓锤

b.

天光白日

c.

死侬骹尾

d.

五落透后

pa31 laŋ31 ku31 thui242 → pa21 laŋ21 ku44 thui242 ‘tadpole’ thiɛŋ44 kuoŋ44 paʔ5 niʔ5 → thiɛŋ21 kuoŋ21 paʔ31 niʔ5 ‘daytime’ si31 nøyŋ51 kha44 mui31 → si21 nøyŋ21 kha51 mui31 ‘out-of-the-way place’ ŋou242 loʔ5 tau213 ʔau242 → ŋu21 loʔ21 tau51 ʔau242 ‘large one-story house with five courtyards’

Syllable and foot in the Fuzhou dialect  49 Clearly, there is only one sandhi domain in each example in (20) since all the tones/syllables in non-terminal positions undergo TS and/or FC and only the last remains unchanged. Wright’s “foot” composed of up to three syllables thus fails to serve as the domain of rule application in these cases. 2.2.2.2  Chan (1985) Chan (1985) argues that phonological changes in Fuzhou including the changes in tones, initials, and finals are all at least partially stress-related. Chan’s instrumental studies show that the sandhi forms of syllables are greatly reduced in duration as compared to the citation forms in the terminal position of a tone sandhi domain. Hence, similar to Wright (1983), Chan also treats syllable duration as an important criterion for determining stress in the Fuzhou dialect and considers disyllabic words with sandhi forms on the first syllable to be words with iambic stress. Chan first distinguishes three types of stress in the Fuzhou dialect, namely a main stress falling on the terminal syllable bearing the citation tone in the sandhi context, a secondary stress falling on the non-terminal syllable(s) bearing the sandhi tone, as well as a weak stress falling on unstressed, toneless suffixes. Chan argues that there is in fact only a two-way phonological contrast with respect to the presence or absence of main stress on the syllable and treats stress in Fuzhou as involving a binary contrast, between [+stress] for the main stress and [−stress] for the non-main stress (the secondary and weak stresses). She further claims that stress has an effect on tone in Fuzhou, such that syllables in pre-stress position undergo tone sandhi, while those in post-stress position generally undergo tone loss. According to Chan, the foot in the Fuzhou dialect can be analyzed as being (a) tone-sensitive, so that feet are only built on full-toned syllables, and not on syllables with neutral tone; (b) right-dominant, i.e., weak-strong; and (c) unbounded, so that structures such w-s, w-w-s, w-w-w-s, and so forth are possible. Chan proposes two rules for foot formation in Fuzhou, as shown in (21) and (22) respectively. In Chan’s analysis, the foot is the domain of the application for TS and other rules in the Fuzhou dialect. Two examples are adapted from Chan (1985), as in (23), with the word-formation process omitted. (21) Main Stress Rule (MSR) Assign an unbounded, right-dominant foot starting at the rightmost fulltoned syllable of the word. (22) Stray Syllable Adjunction (SSA) Adjoin a stray syllable as a recessive node of an adjacent foot.

50  Syllable and foot in the Fuzhou dialect (23) a.

尺 tshuoʔ23

寸 tshouŋ213

tshuoʔ23

tshouŋ213

w

‘measure, size’ UR

s

MSR

F tshuoʔ51

w

ʒouŋ213

s

TS and other rules

tshuoʔ51

ʒouŋ213

SR

我 ŋuai31

其 ki0

ŋuai31

ki0

F

b.

s

w

‘mine’ UR SSA (MSR does not apply)

F ŋuai31

s

ʔi0

w

TS and other rules

ʔi0

SR

F ŋuai31

There are some problems with Chan’s analysis. First of all, in Chan’s analysis, there is a binary contrast between syllables in the Fuzhou dialect, namely [+stress] vs. [−stress]. However, ever since Liberman & Prince (1977), stress has been viewed as a prominence relation between syllables and thus is no longer represented by means of a feature [±stress] that is analogous to other distinctive features such as [±nasal] and [±coronal] (cf. Kenstowicz 1994; Gussenhoven & Jacobs 1998, among others). Second, similar to Wright (1983), Chan’s analysis also falls into the trap of circular reasoning. Based on the results from her instrumental studies as well as evidence from Wright’s spectrographic measurements, Chan believes that syllable duration should be treated as one of the most important criteria for determining stress and claims that there is a correlation between stress and tone. Specifically, Chan argues that main stress falls on the syllable bearing the citation tone, secondary stress falls on syllables bearing the sandhi tone, and weak stress falls on syllables bearing the

Syllable and foot in the Fuzhou dialect  51 neutral tone. In such an analysis, therefore, it is the subtypes of the tone (i.e., citation tone, sandhi tone, or neutral tone) that decide the subtypes of the stress (i.e., main stress, secondary stress, or weak stress) on the syllables, but not vice versa. More­ over, it is clear that the contrast between main stress and secondary stress should be attributed to the application of tone sandhi in Chan’s analysis, since the distinction between the sandhi tone and the citation tone must not exist if TS does not apply. However, Chan claims that stress has an effect on tone, making syllables in sandhi position undergo tone sandhi, while those in post-main-stress position undergo tone loss. This obviously reverses cause and effect and thus is not convincing. Chan’s examples in (23) can serve as a good illustration of the circular reasoning in her analysis. In (23a), we can see that the application of MSR precedes the application of TS. Hence, at the step of MSR, tone sandhi has not yet occurred. According to Chan’s analysis, however, the foot in the Fuzhou dialect is constructed on the basis of the weak-strong contrast (iambic stress) resulting from the distinction in syllable duration in the sandhi context. Since tone sandhi is triggered after MSR in (23a), there is no such weak-strong contrast at the step of MSR. Therefore, Chan has to answer the following questions: how can we construct a foot with the foot formation rule MSR when there is no weak-strong contrast? How can we know that the syllable of 尺 [tshuoʔ23] should be labeled as weak while the syllable of 寸 [tshouŋ213] should be labeled as strong? Without any independent evidence for the existence of stress in the Fuzhou dialect besides the distinction in syllable duration in the sandhi environment, Chan’s analysis fails to handle these questions. The derivation of (23b) also has serious problems. First, there is in fact no tone sandhi in this example. Therefore, the “TS” here must refer to the application of tone loss in Chan’s analysis. As Chan claims, weak stress falls on syllables bearing the neutral tone. Thus, at the step of SSA, a weak stress is assigned to 其 [ki0] simply because it has a neutral tone. Then, the foot constructed on the strong-weak contrast in (23b) serves as the domain of application for tone loss on the toneless syllable 其 [ki0], which is another typical example of circular reasoning. In addition, Chan’s analysis of so-called “toneless suffixes” cannot account for the fact that some Fuzhou elements bearing full tones have the same phonological behavior with these toneless suffixes. Compare (24a, b) from Chan (1985) with (24c–e). (24) a. b. c. d. e.

我其 大学𠲥 去过 坐囇 收遘

ŋuai31 ki0 tuai242 xouʔ5 le0 kho213 kuo213 soy242 la242 siu44 kau213

→ ŋuai31 ʔi0 → tuai44 ʔouʔ5 le0 → kho213 ʔuo213 *kho51 ʔuo213 → soy242 la242 *soy51 la242 → siu44 ʔau213 *siu51 ʔau213

‘mine’ ‘at the university’ ‘to have been to’ ‘to sit awhile’ ‘to receive’

We can find that 过 [kuo213], 囇 [la242], and 遘 [kau213] all have a full tone and thus must have a main stress in Chan’s analysis. According to MSR, w-s feet should be constructed in (24c–e), and then TS should be triggered. However, as we can see from (24c–e), it is clear that TS is blocked, similar to (24a, b). Chan’s analysis thus fails to deal with the similar phonological behavior exhibited by examples in (24a, b) and (24c–e), since it is impossible to assign a weak stress to

52  Syllable and foot in the Fuzhou dialect elements bearing a full tone like 过 [kuo213], 囇 [la242], and 遘 [kau213] according to her analysis. I will demonstrate in Chapter 4 that elements such as 过 [kuo213], 囇 [la242], 遘 [kau213], 其 [ki0], and 𠲥 [le0] are all enclitics in the Fuzhou dialect, which is why they exhibit similar phonological behavior. 2.2.2.3  Shih (1986) Shih (1986) is another analysis that heavily relies on the construction of the foot domain in the Fuzhou dialect. Based on Chan (1980) and Wright (1983), Shih posits the Revised Head Dominance Condition and the Foot Formation Rule in the Fuzhou dialect, as in (25) and (26). (25) Revised Head Dominance Condition (HDC) Mark the right edge of every X0, except where XP is an adjunct. (26) Foot Formation Rule (FFR)1 Foot Construction a. Immediate Constituency (IC): link immediate constituents into disyllabic feet. b. Duple Meter (DM): scanning from left to right, string together unpaired syllables into binary feet, unless they branch to the opposite direction. Super-foot Construction Join any leftover monosyllable to a neighboring binary foot according to the direction of syntactic branching. Shih argues that the sandhi domain in Fuzhou is thus constructed in two steps— the Revised Head Dominance Condition marks the boundary of the phonological phrase (or “tone group” in Shih’s formulation), and then the Foot Formation Rule operates within each phonological phrase to construct prosodic feet and superfeet. According to Shih, all phonological rules of Fuzhou, including TS, FC, and CL, refer to the super-foot (or the foot) as the domain of application. Shih’s analysis is problematic in terms of her definition of the foot/super-foot as the prosodic domain of rule application in the Fuzhou dialect. It can be seen that Shih’s foot and super-foot are built with reference to syntactic information such as immediate constituents and the direction of syntactic branching. However, as mentioned in Chapter  1, in the theory of prosodic phonology, it is of crucial importance that whether a given prosodic constituent makes use of a specific type of phonological and/or non-phonological information in the definition of its domain is not a free choice (Nespor & Vogel 1986, 2007; Zhang 1992, 2017, among others). Specifically, the definition of the domain of the foot is only sensitive to phonological information (i.e., stress), but not sensitive to syntactic information, as we have seen in Figure 1.3. In order to account for the application of Fuzhou phonological rules at the phrasal level, Shih has to build

Syllable and foot in the Fuzhou dialect  53 the foot domain by making use of syntactic information, which puts her analysis on the wrong track. Moreover, Shih does not discuss anything regarding the stress pattern in the Fuzhou dialect, which further makes her definition of the foot groundless. 2.2.2.4  Hung (1987) Hung (1987) also argues for the existence of the foot in the Fuzhou dialect. His Foot Formation Rule is presented in (27). (27) Fuzhou Foot Formation Rules a. Link the syllables in polysyllabic lexical items into freely structured feet. b. Scanning from left to right, link heads to their arguments to form disyllabic or right-branching feet. c. Scanning from left to right, link modifiers to their heads to form disyllabic or right-branching feet. We can find that Hung’s definition of the foot has the same problem as we have seen in Shih’s (1986) analysis. By making use of notions such as head, argument, and modifier, the construction of the foot domain in Hung’s analysis wrongly refers to syntactic information. Similar to Shih (1986), another problem with Hung’s analysis is that no independent evidence is provided to show that the foot domain can be constructed on the basis of phonological information in the Fuzhou dialect. 2.2.2.5  Chan (1998) Similar to all the previous studies discussed above, Chan (1998) also suggests that the Fuzhou tone sandhi domain corresponds to the foot. She argues that the construction of the foot in the Fuzhou dialect should refer to different information at the lexical level and the phrasal level. At the lexical level, Chan argues that the foot should be constructed on the basis of the quality and position of a syllable. Specifically, Chan suggests that there exists a stress tier at which a right-dominant (w-s) metrical foot is built and that the distinction between weak syllables and strong syllables relies on the stability of a tone bearing unit in holding its citation tone. She further claims that a prosodically strong/prominent syllable provides a more stable place to anchor the citation tone, while a prosodically weak syllable does not have the strength to hold the citation tone, and thus, the tone can be deleted or changed. By assuming that mora is the tone bearing unit and that a syllable is bimoraic, Chan proposes the following foot structure for disyllabic compounds in the Fuzhou dialect. According to Chan’s analysis, in (28) the strength in holding the citation tone decreases from the rightmost mora to the leftmost mora, and thus, the TS rule is triggered within the foot domain.

54  Syllable and foot in the Fuzhou dialect (28)

Foot σw m

σs m

m

m

←

Chan’s analysis of trisyllabic compounds and quadrisyllabic compounds are based on her analysis of disyllabic compounds. Similar to Wright (1983), Chan also suggests that the foot is at most trisyllabic. Hence, in Chan’s analysis, within a quadrisyllabic compound that forms one tone sandhi domain, only the last three syllables form a foot, while the initial syllable of the compound is not included in the foot and thus receives a low default tone. At the phrasal level, Chan argues that the foot is constructed according to the length and position of a word. In her analysis, a phonological phrase must be constructed first by making use of syntactic information, and then feet defined with respect to the length and/or the position of the morphosyntactic word are built within the phonological phrase. To construct the foot at the phrasal level, Chan proposes a Prominence Assignment Rule, which is presented as follows. (29) Prominence Assignment Rule Assign prominence (indicated by the letter s) to a syntactic word that is located at the final position of a phonological phrase or that is a polysyllabic compound. There are several theoretical concerns about Chan’s analysis. First, Chan’s definition of the foot at both the lexical level and the phrasal level is on the wrong track. On the one hand, the foot is not defined on the basis of the stability of a tone bearing unit in holding its citation tone. As a prosodic constituent that organizes the syllables into higher-order units built around stressed syllables (Hammond 2011), the foot is defined on the basis of stress whose phonetic correlates may be pitch levels, duration, and/or loudness in different languages (cf. Kenstowicz 1994; Kager 1996, 2007; Gordon 2011; Zhang 2017, among others). Even if we give Chan the benefit of the doubt and assume that the foot could be defined based on the contrast in the stability of holding the citation tone, Chan’s analysis is still not defendable. Chan argues that a w-s foot should be constructed in which the distinction between the weak syllable and the strong syllable relies on the stability of holding the citation tone, which means that one must have already known which syllable can hold the citation tone and which cannot before a w-s foot is constructed. Based on such a premise, Chan claims that the weak syllable in the foot undergoes TS while the strong syllable does not because the weak syllable is not strong enough to hold the citation tone, which is clearly another example of circular reasoning. On the other hand, the foot is not defined by referring to the length of the morphosyntactic word or the position of the morphosyntactic word in a phonological

Syllable and foot in the Fuzhou dialect 55 phrase either. The foot is defined on the basis of stress, and the stress refers to the prominence relation between syllables—it simply has nothing to do with the length or the position of the morphosyntactic word. The Prominence Assignment Rule proposed by Chan is quite bizarre since the rule actually says that a foot can be composed of a couple of morphosyntactic words and that one of the morphosyntactic words is labeled strong simply due to its length or position. The second concern regarding Chan’s analysis lies in her analysis of mora in the Fuzhou dialect. As mentioned in Section 2.2.2.1, mora is used to measure the syllable weight such that a heavy syllable contains two moras while a light syllable has one mora. Therefore, the assumption that all syllables in the Fuzhou dialect have two moras is groundless. Third, it should be noticed that, in Chan (1998), she follows Bickmore (1990) and assumes that there are five prosodic constituents in the prosodic hierarchy, as shown in (30). (30) The prosodic hierarchy in Chan (1998) a. Utterance b. Intonational Phrase c. Phonological Phrase d. Clitic Group e. Word There are three problems with Chan’s employment of this prosodic hierarchy with respect to her analysis of the foot. To begin with, according to Chan, the foot is built within the domain of the phonological phrase. Thus, one would expect the foot to be a constituent located lower than the phonological phrase. However, the foot is not listed as a constituent in Chan’s hierarchy. Now that the foot is not treated as a prosodic constituent in the hierarchy, it is unexplainable why it is treated as the domain of application for tone sandhi rules. In addition, in (30), there is another prosodic constituent, namely the clitic group, between the phonological phrase and the word. Chan does not discuss this constituent and simply ignores it in her analysis. Moreover, the “word” in the theory of prosodic phonology usually refers to the phonological word/prosodic word. Nevertheless, in Chan’s analysis, the word refers to the morphosyntactic word, and her analysis does not address the relationship between the morphosyntactic word and the phonological/prosodic word in the Fuzhou dialect. The last but not the least, Chan’s analysis of quadrisyllabic compounds is debatable. Since Chan sets the upper limit of foot length to three syllables, she has to assume that a quadrisyllabic compound that forms a single domain must be divided into two parts: the last three syllables form a foot domain while the first syllable is excluded from the foot domain. Chan thus has to answer some followup questions: does the first syllable form a prosodic domain together with the other syllables in the quadrisyllabic compound? If yes, which domain do these syllables form? If not, why and how to treat this single syllable in terms of its prosodic status? Chan’s analysis provides no answer to these questions. In Section 2.2.3 and

56  Syllable and foot in the Fuzhou dialect Chapter 3, I will show that there is in fact a prosodic domain that can contain all the syllables in a quadrisyllabic compound in Fuzhou, namely the prosodic word, which constitutes a good reason to abandon the foot domain proposed by Chan. 2.2.2.6  Reflection on previous studies So far, we have reviewed some of the most important previous studies that treat the foot as a prosodic domain of rule application in the Fuzhou dialect. Based on the discussion in the preceding subsections in Section  2.2.2, we can find some common problems in these previous studies. First of all, no independent evidence is provided for the existence of the stress and the foot in the Fuzhou dialect. The construction of the foot is based on the contrast in metrical prominence between syllables, namely the stress, and the most important evidence that can confirm the existence of the stress in a given language is native speakers’ perceptions and judgments. In previous studies that argue for the existence of the stress in Fuzhou, however, we have not seen any independent evidence from native speakers’ perceptions and judgments. We have seen that different types of contrast between syllables have been proposed as fundamental in determining the positions of stressed vs. unstressed syllables (or strong vs. weak syllables) in this dialect, such as the contrast in syllable duration (Wright 1983; Chan 1985) and the contrast in the stability of a tone bearing unit in holding its citation tone (Chan 1998). Nevertheless, as I have argued in the preceding subsections, such contrasts can only be perceived in the sandhi context but never appear when no phonological rules occur. Therefore, such contrasts in the Fuzhou dialect had better be considered as the byproduct of the application of phonological rules and hence cannot serve as independent evidence for the existence of the stress. Similar to Mandarin Chinese, in which Zhang (2014, 2017) argues for the lack of binary metrical contrast between syllables and thus the lack of the foot, the construction of the foot domain in the Fuzhou dialect is also groundless without the evidence for the stress. In addition, some previous studies are trapped in circular reasoning. We have seen such a problem in Wright (1983), Chan (1985), and Chan (1998). They all contend that a weak-strong foot should be built on the basis of the contrast between stressed vs. unstressed syllables (or strong vs. weak syllables), and then phonological rules such as TS and FC are triggered within this right dominant foot domain. Since no independent evidence for the stress is found in the Fuzhou dialect, they have to take other phonological phenomena as the evidence. Both Wright (1983) and Chan (1985) attribute the stress vs. non-stress contrast to the contrast in syllable duration, but the contrast in syllable duration is actually caused by the application of TS and/or FC. Chan (1998) claims that the distinction between strong and weak syllables lies in the stability of holding the citation tone, but the definition of this “stability” implies that one should be able to identify the “strong” syllable within a word, which is only possible in the sandhi context where TS has been triggered. Therefore, in the analyses of Wright (1983), Chan (1985), and Chan (1998), the application of the TS rule and/or the FC rule is both the starting point and the goal of their reasoning, which makes their analyses superficially plausible, but actually wrong.

Syllable and foot in the Fuzhou dialect  57 Moreover, the definitions of the foot proposed in some previous studies are problematic. The definition of the foot should be based on the contrast in metrical prominence between syllables, and hence the domain of the foot should be built by referring only to phonological information (i.e., stress), but not to nonphonological information. We have seen different types of syntactic information employed in the definition of the foot domain in Fuzhou—Shih’s (1986) definition of the foot refers to the notions of immediate constituents and syntactic branching; Hung’s (1987) definition relies on the notions of head, argument, and modifier; and Chan’s (1998) definition of the foot at the phrasal level is sensitive to the length of the morphosyntactic word or the position of the morphosyntactic word in a phonological phrase. These definitions of the foot are all theoretically erroneous. The other problem regarding the definition of the Fuzhou foot domain in these previous studies is that all these studies consider the foot to be the domain of application for phonological rules at both the lexical level and the phrasal level. Since the domain of the prosodic word may be either of the same size or smaller than the terminal node of the syntactic tree, namely the morphosyntactic word/ lexical item (Nespor & Vogel 1986, 2007), the prosodic domain of rule application at the phrasal level must be larger than the prosodic word. According to the discussion in Chapter 1, prosodic constituents above the foot in the prosodic hierarchy are defined by making use of non-phonological information. Since the foot in the prosodic phonology theory is only constructed on the basis of phonological information, it is inappropriate to treat the foot as the domain of application for phonological rules at the phrasal level. Finally, the foot analyses advanced by previous studies all face empirical problems. Since it is inappropriate to see the foot as the domain of rule application at the phrasal level, the foot domain is not able to account for phrasal-level data in the Fuzhou dialect. That aside, the foot defined in previous studies is arguably not a good choice for the rule application domain at the lexical level either. To be specific, the application of phonological rules in some lexical items in the Fuzhou dialect is not accounted for with the notion of the foot, as briefly discussed in Section  2.2.2.1, whereas all phonological rules that are claimed to be footdomain rules in previous studies can be reformulated as prosodic word-domain rules. These additional arguments against the foot analyses will be discussed in detail in Section 2.2.3. 2.2.3 Additional arguments against the foot as a domain of rule application in the Fuzhou dialect One significant motivation for the establishment of a particular prosodic domain in a given language is the existence of phonological phenomena that make reference to that domain. Hence, if we would like to establish the foot as a prosodic domain in the Fuzhou dialect, we need to find evidence for Fuzhou phonological rules referring crucially to the foot as the domain of application. As we have seen in Chapter 1, Fuzhou phonological rules, such as TS, FC, and CL, can apply at

58  Syllable and foot in the Fuzhou dialect both the lexical level and the phrasal level. The application of these rules at the phrasal level, on the one hand, does not refer to the domain of the foot, since this is theoretically incorrect, as argued in Section  2.2.2.6. On the other hand, the application of these rules at the lexical level, as I will discuss in this section, is not sensitive to the foot domain either. In the following, I am going to present two additional arguments against the foot as the domain of application for Fuzhou phonological rules at the lexical level: (a) Fuzhou data at the lexical level that have been used in favor of the foot domain in previous studies can also be viewed as motivation for another prosodic constituent, namely the prosodic word, and (b) the application of Fuzhou phonological rules in some data can only be accounted for with the notion of the prosodic word, but not with the notion of the foot. Let us first revisit some relevant data provided in previous studies as the supporting evidence for the foot as the domain of rule application in the Fuzhou dialect. Examples in (31) are adapted from Wright (1983), Chan (1985), and Chan (1998). For the sake of brevity, only TS and FC are presented and marked in bold. (31)

a. b. c. d. e. f.

对比 顽固 讨厌 狐狸猫 政治家 共产主义

toy213 pi31 → tøy51 pi31 ŋuaŋ51 kou213 → ŋuaŋ21 kou213 tho31 ʔiɛŋ213 → tho44 ʔiɛŋ213 xu51 li51 ma51 → xu44 li44 ma51 tseiŋ213 teiʔ23 ka44 → tsiŋ21 tiʔ44 ka44 køyŋ242 saŋ31 tsuo31 ŋiɛ242 → kyŋ21 saŋ21 tsuo44 ŋiɛ242

‘to contrast’ ‘obstinate’ ‘to hate’ ‘weasel’ ‘statesman’ ‘Communism’

Previous studies reviewed in Section  2.2.2 would argue that the foot is the domain of application for TS and FC in the above examples. According to the foot analyses, the best way to describe the TS rule and the FC rule in these examples is to say that each example in (31) constitutes a foot and then a syllable undergoes TS and FC when it precedes another syllable within the same foot. This statement can be roughly formulated as in (32), in which T stands for a citation tone, T’ stands for a sandhi tone, and Group A and Group B refer to Group A and Group B alternating finals respectively. (32) a. b.

TS: FC:

Tn → Tn’ / [____ Tn + 1]Σ (n ≥ 1) Group B → Group A / [[C0 ____] σn [. . .] σn + 1] Σ (n ≥ 1)

The formulations in (32) seems to account for the application of TS and FC in (31). The problem with such an analysis, nevertheless, is that the application of the same rules can also be treated as the motivation for the establishment of the prosodic word domain in Fuzhou prosodic phonology. We can find that examples in (31) are all morphosyntactic words, and they all occupy the terminal node of the syntactic tree, regardless of the number of syllables they contain or the lexical category/part of speech they belong to. It will be demonstrated in the next chapter that the prosodic word in the Fuzhou dialect is coextensive with the terminal node of the syntactic tree. In other words, the size of the prosodic word domain in Fuzhou is equal to the morphosyntactic word. Therefore, the application of TS and FC in (31) can also be viewed as triggered in the domain of the prosodic word, as formulated in (33).

Syllable and foot in the Fuzhou dialect  59 (33) a. b.

TS: FC:

Tn → Tn’ / [____ Tn + 1]ω (n ≥ 1) Group B → Group A / [[C0 ____] σn [. . .] σn + 1] ω (n ≥ 1)

From the examples in (31) and the formulations in (33), we don’t see any reason to exclude the possibility of treating the prosodic word as the domain of application for TS and FC at the lexical level in Fuzhou. One can even go one step further and claim that the application of any Fuzhou phonological rules at the lexical level, which has the form of changing segment/tone A to segment/tone B within the domain of the foot proposed in previous studies, can be formulated in another way by referring to the prosodic word with no loss of generalization. The second additional argument against the foot as the domain of rule application in Fuzhou is that the foot analyses proposed in previous studies might face difficulties when dealing with examples like (31f). We have seen some similar examples in (20), repeated here as in (34a–d). Some more examples are listed as in (34e–g). (34) a.

拨浪鼓锤

b.

天光白日

c.

死侬骹尾

d.

五落透后

e.

加三古怪

f.

看命先生

g.

花狸猫糟

pa31 laŋ31 ku31 thui242 → pa21 laŋ21 ku44 thui242 ‘tadpole’ thiɛŋ44 kuoŋ44 paʔ5 niʔ5 → thiɛŋ21 kuoŋ21 paʔ31 niʔ5 ‘daytime’ si31 nøyŋ51 kha44 mui31 → si21 nøyŋ21 kha51 mui31 ‘out-of-the-way place’ ŋou242 loʔ5 tau213 ʔau242 → ŋu21 loʔ21 tau51 ʔau242 ‘large one-story house with five courtyards’ ka44 saŋ44 ku31 kuai213 → ka21 saŋ21 ku44 kuai213 ‘crafty’ khaŋ213 miaŋ242 siŋ44 saŋ44 → khaŋ21 miaŋ21 siŋ44 saŋ44 ‘fortune-teller’ xua44 li51 ma51 tsau44 → xua21 li44 ma44 tsau44 ‘gaudy’

These examples are all morphosyntactic words in the Fuzhou dialect. As argued in Section 2.2.2.1, since Wright (1983) suggests that the foot is at most trisyllabic, each example in (31f) and (34) must form a domain larger than the foot according to her analysis, which makes her analysis fail to account for the application of TS and FC in these examples. By the same token, the foot domain proposed in Chan (1998) cannot deal with these examples either. Although Chan (1998) assumes that syllables preceding the antepenultimate are not included in the foot domain, this repair operation is problematic, as discussed in Section 2.2.2.5. It seems that the foot advanced by Chan (1985), Shih (1986), and Hung (1987) can serve as the domain of rule application in these examples, since their analyses do not require the upper limit of foot length to be three syllables. However, quadrisyllabic feet are too big from a universal viewpoint. According to the typology of Hayes (1995), trisyllabic feet are possible in languages that have “weak local parsing”, while quadrisyllabic feet are universally disallowed (also cf. Liberman 1975; Prince 1983; Selkirk 1984, among others). Take English, a typical stress

60  Syllable and foot in the Fuzhou dialect language, as an example. It has been argued that the quadrisyllabic foot forms a problem in English, due to the fact that potential quadrisyllabic feet are often interpreted as two disyllabic feet (Hogg & McCully 1987). One may argue that the Fuzhou dialect can have unbounded feet that consist of any number of syllables. This is, nevertheless, invalid too. An unbounded foot must be composed of a string of one relatively strong syllable and several relatively weak syllables. With no independent evidence for the contrast in metrical prominence between syllables, we are not able to define “strong” and “weak” syllables in the Fuzhou dialect. Therefore, as in the case of binary feet, it is not possible to prove the existence of unbounded feet in the Fuzhou dialect either. Some scholars claim that quadrisyllabic feet are often tolerated in various Chinese dialects since it is normal for the scope of tone sandhi to exceed three syllables (e.g., Yip 1980; Duanmu 1993; Chen 2000, among others). However, it is noteworthy that the fact that the scope of tone sandhi can exceed three syllables does not mean that the domain of the application of tone sandhi is the foot, at least not in the Fuzhou dialect. Consider the Fuzhou examples in (35). (35) a.

tshui44 xuaŋ44 ny51 puoŋ242 → tshui21 xuaŋ51 ny21 puoŋ242 ‘to steam rice with sweet potato’ b. 食鼎边糊 siɛʔ5 tiaŋ31 piɛŋ44 ku51 → siɛʔ21 tiaŋ21 piɛŋ44 ku51 ‘to eat rice flour cake soup (a local Fuzhou delicacy)’ c. 信共产主义 seiŋ213 køyŋ242 saŋ31 tsuo31 ŋiɛ242 → siŋ21 kyŋ21 saŋ21 tsuo44 ŋiɛ242 ‘to believe in Communism’ d. 买拨浪鼓锤 mɛ31 pa31 laŋ31 ku31 thui242 → mɛ21 pa21 laŋ21 ku44 thui242 ‘to buy tadpoles’ 炊番薯饭

TS (and FC) applies in the above examples and the string in each example forms one large tone sandhi scope. However, the domain of application for TS and FC in (35) cannot be the foot, since these quadrisyllabic and pentasyllabic examples are all clearly verb-object phrases. As argued in Section  2.2.2.6, the prosodic domain of rule application at the phrasal level must be larger than the prosodic word, and hence the foot is by no means a possible choice in these cases. Therefore, the application of TS and FC in both (34) and (35) only indicates that TS and FC can not only apply to lexical items composed of more than three syllables, but also apply to quadrisyllabic and pentasyllabic units at the phrasal level. This is convincing evidence that TS and FC in Fuzhou can apply within different prosodic domains, but it does not show that quadrisyllabic or pentasyllabic feet are allowed or that the foot can play a role as the domain of rule application in this dialect. We have seen that foot analyses that posit a trisyllabic upper limit of foot length fail to cover the data of morphosyntactic words containing more than three syllables in the Fuzhou dialect, while analyses without the upper limit of foot length are quite suspicious from the perspective of language universals and also theoretically and empirically problematic. Therefore, the application of Fuzhou phonological rules in morphosyntactic words containing more than three syllables poses an unexplainable problem to all the foot analyses we have seen in previous

Syllable and foot in the Fuzhou dialect  61 studies. By contrast, these examples are not likely to be a problem if we consider the prosodic word to be the domain of rule application at the lexical level. In the next chapter, we will see evidence that the prosodic word in the Fuzhou dialect is exactly of the same size with the domain formed by morphosyntactic words. Since languages generally do not impose restrictions on the maximal size of morphosyntactic words and the prosodic word, the prosodic word is more capable of functioning as the domain of rule application in morphosyntactic words in the Fuzhou dialect as compared to the foot. In other words, rules operating in Fuzhou morphosyntactic words can be formulated as prosodic word-domain rules, but not as foot-domain rules. So far, I have discussed two further arguments against the foot analyses in the Fuzhou dialect. On the one hand, I have shown that the application of phonological rules such as TS and FC can be treated as prosodic word bounded rules with no loss of generalization. On the other hand, the application of these phonological rules in quadrisyllabic examples at the lexical level can only be captured by stating them as rules operating within the prosodic word domain, instead of the foot domain. To sum up, at the lexical level in the Fuzhou dialect, there are certain prosodic word-domain rules that cannot be reformulated in terms of the foot domain, while all foot-domain rules proposed in previous studies can be reformulated as prosodic word-domain rules. According to Occam’s Razor, it is reasonable to simply use the prosodic word domain to replace the foot domain, rather than maintain both as prosodic domains in the prosodic phonology of the Fuzhou dialect. 2.2.4 Summary In Section 2.2, I have examined another constituent in the universal prosodic hierarchy, namely the foot. We have seen that, while the foot functions as the domain of phonological phenomena in a number of languages, whether it plays a role in the Fuzhou dialect is a debatable question. On the basis of the review of previous studies and the analysis of relevant data, I  have identified the following arguments against the establishment of the foot domain in this dialect: (a) there is no independent evidence for the existence of the stress and the foot; (b) phonological rules at the lexical level that are claimed to operate within the foot domain according to previous studies can be reformulated as prosodic word domain rules, without reference to the foot; and (c) rule application in some lexical-level data can only be handled by referring to the prosodic word but not the foot. Therefore, although the foot is a universal prosodic unit across languages, no independent evidence and valid arguments can be supplied to demonstrate its existence and its role as a domain of rule application in Fuzhou. Based on these arguments, I propose to resolve the problem of whether rules such as TS and FC in the Fuzhou dialect should be expressed as foot domain rules or as prosodic word domain rules by assuming that the foot domain should be excluded from the prosodic hierarchy of this dialect. Thus, between two domains that overlap each other in terms of the function as serving as the rule application domain, I choose the one that can deal with more data while abandon the one that

62  Syllable and foot in the Fuzhou dialect lacks independent evidence and has less explanatory power. By so doing, a more constrained, and thus more highly valued, prosodic phonological system would be achieved in this dialect.

2.3 Summary In this chapter, I  have examined the existence and roles of two prosodic constituents of the universal prosodic hierarchy, namely the syllable and the foot, in the Fuzhou dialect. I have shown that the syllable is indispensable in this dialect while the foot should be excluded. On the one hand, I have argued that the syllable serves as a domain of rule application in the Fuzhou dialect, since the FA rule, as a well-formedness condition on the combination of alternating finals and tones, applies within but not across the syllable boundary. On the other hand, the foot should not be treated as a prosodic domain in this dialect, because of its lack of independent evidence and its inadequacy of explanatory power as compared to the prosodic word domain in terms of the application of Fuzhou phonological rules at the lexical level. We have seen that Fuzhou phonological rules such as TS and FC can apply at both the lexical level and the phrasal level. At the lexical level, I  have suggested that rules applying at the juncture of more than one syllables actually operate within the domain immediately higher than the foot in the universal prosodic hierarchy, namely the prosodic word. A detailed discussion on this domain will be presented in the chapter we now turn to. At the phrasal level, more syntactic information must be involved in defining the domain of rule application than what is required in the definition of the prosodic word domain, which will be discussed in later chapters.

Note 1 The Foot Formation Rule was first proposed in Chen (1984) to relate the syntactic structure of a line to the metrical template of classical Chinese poetry.

3 Prosodic word Domain of lexical-level rule application in the Fuzhou dialect

This chapter investigates the role of the prosodic word in the Fuzhou dialect and the application of various Fuzhou phonological rules (TS, MTS, FC, and CL) at the lexical level, and it is organized as follows. Section 3.1 introduces the definition of the prosodic word in general and its function as a domain of rule application across languages. Given that the prosodic word is roughly the size of a morphosyntactic word, Section 3.2 offers a survey of major types of morphosyntactic word formation in the Fuzhou dialect. Section  3.3 examines the application of TS, MTS, FC, and CL within the domain formed by Fuzhou morphosyntactic words, showing that it is the morphosyntactic words that form the domain of lexical-level application for these rules in this dialect. Section 3.4 deals with the definition of the prosodic word domain in the Fuzhou dialect, arguing that the domain of the prosodic word in Fuzhou is of the same size as the terminal node of the syntactic tree, and provides some concluding remarks for this chapter.

3.1 Introduction 3.1.1  Definition of the prosodic word and its domain formation Located higher than the syllable and the foot but lower than the clitic group and the phonological phrase in the prosodic hierarchy, the prosodic word, or alternatively the phonological word, is defined as the lowest constituent constructed on the basis of mapping rules that make substantial use of non-phonological notions (cf. Nespor & Vogel 1986, 2007; Booij & Rubach 1984; Booij & Lieber 1993; Hannahs 1995a, 1995b, among others). Roughly the size of a morphosyntactic word, the prosodic word is a prosodic constituent that represents the interaction between the phonological and the morphological components of the grammar. Logically speaking, there are three possibilities for the domain of the prosodic word—it could be larger, smaller, or equal to the morphosyntactic word, that is, the terminal node of the syntactic tree (Booij 1983, 1996, among others). Taking into consideration different types of languages they observed, however, Nespor & Vogel (1986, 2007: ch. 4) claim that there are only two possibilities—the domain

64  Prosodic word in the Fuzhou dialect of the prosodic word may be either of the same size or smaller than the terminal node of the syntactic tree. They thus propose the definition for the prosodic word domain in (1). (1)

Prosodic word (ω) domain (adapted from Nespor & Vogel 1986, 2007) A. The domain of ω is the terminal node of the syntactic tree. or B.  I. The domain of ω consists of a. a stem; b. any element identified by specific phonological and/or morphological criteria; c. any element marked with the diacritic [+W]. II. Any unattached elements within the terminal node of the syntactic tree form part of the adjacent ω closest to the stem; if no such ω exists, they form a ω on their own.

In Nespor & Vogel’s definition, Type A refers to the prosodic word that includes a stem plus all adjacent affixes, or both members of a compound, in languages such as Greek and Latin (Nespor & Vogel 1986: 110–116; Nespor & Ralli 1996). By contrast, Type B refers to the prosodic word smaller than the terminal node of the syntactic tree, which includes several subtypes. Type BIa refers to cases in languages such as Turkish and Sanskrit, in which each member of a compound forms a prosodic word domain (Nespor &Vogel 1986: 117–122; Nespor & Ralli 1996). Type BIb is exemplified by cases in languages such as Hungarian and Italian, in which prefixes form a prosodic word on their own while the stem plus suffixes form one prosodic word (Booij 1984; Nespor & Vogel 1986: 122–134)1 and also exemplified by cases in languages such as Yidiɲ, in which affixes satisfying minimal word requirements (disyllabicity in the case of Yidiɲ) form their own prosodic words (Dixon 1977a; Nespor & Vogel 1986: 134–136). Type BIc refers to the prosodic word composed of affixes idiosyncratically marked with a diacritic feature [+W] in languages such as Dutch (Nespor & Vogel 1986: 136–140; Hulst 1984: 66ff). Type BII requires elements that do not include stems to always attach to a prosodic word within the terminal node of the syntactic tree or form a prosodic word by themselves, responding to the requirement of the Strict Layer Hypothesis. From Nespor & Vogel’s definition discussed above, we can find that, although it is generally accepted that in the prosodic hierarchy there is a word-sized prosodic domain named the prosodic word, languages may vary in the way certain morphosyntactic information (i.e., stem, prefix, suffix, members of a compound, etc.) is integrated into the formation of the prosodic word domain, making the domain formation of the prosodic word a language-specific issue. Some aspects of Nespor & Vogel’s definition have been questioned. For example, the second part of Type BII allows elements like clitics to form their own prosodic words. This kind of over-assignment of the ω status to elements like clitics has raised objections. To linguists such as Itô & Mester (1992/2003) and Selkirk

Prosodic word in the Fuzhou dialect  65 (1996), among others, elements like clitics should not be analyzed as prosodic words, and hence it is inappropriate to include this information in the definition of the prosodic word domain. Under such an analysis, Booij (1996), among others, argues that the sequence of a lexical host plus a clitic forms one independent prosodic word. The domain of this type of prosodic words is thus larger than the terminal node of the syntactic tree, since a clitic constitutes a syntactic terminal node on its own (issues related to clitics will be discussed in Chapter 4). Despite the controversy surrounding the details in the formation of the prosodic word domain, however, a number of studies have demonstrated that the prosodic word is an indispensable prosodic constituent in the prosodic hierarchy since it is the domain for the application of various phonological generalizations across languages, as to be seen in Section 3.1.2. 3.1.2  The prosodic word as a prosodic domain across languages It has long been recognized that a lot of phonological phenomena across languages, including phonological rules (both segmental and prosodic/suprasegmental), phonotactic constraints, and minimal word requirements, are sensitive to the prosodic word domain. Let us take a brief look at some typical examples in this section. One typical example of segmental rules operating in the prosodic word domain is Vowel Harmony in Hungarian. In Hungarian, all vowels other than [i], [í], and [é] participate in the process of Vowel Harmony, and the harmonizing feature is [back]. According to Booij (1984) and Nespor & Vogel (1986, 2007), the domain of application for this rule is the prosodic word, and this rule applies only when its trigger and target belong to the same prosodic word, which can be presented as in (2). (2)

Hungarian Vowel Harmony: V → [αback]/ [. . .[αback] C0 ___ . . .] ω

Booij (1984) and Nespor  & Vogel (1986, 2007) show that the sequence of the stem plus suffix(es) forms an independent prosodic word in Hungarian, as exemplified in (3a), in which nek/nak is the dative singular suffix and all vowels have the same value for the feature [back]. In contrast, the two members of a compound form two different prosodic words in this language, as in (3b), and prefixes also form prosodic words on their own, as in (3c). We can find that, in (3b) and (3c), vowels in different prosodic words are different in terms of their backness. (3)

a. b.

stem + suffix: stem + suffix: stem + stem:

c.

prefix + stem:

[ölelés-nek]ω [hɑjó-nɑk]ω [könyv]ω [tɑr] ́ ω book collection ‘library ’ [oda]ω [menni]ω there go ‘to go there’

‘embracement + (dat. sg.)’ ‘ship + (dat. sg.)’

66  Prosodic word in the Fuzhou dialect Besides Vowel Harmony in Hungarian, segmental rules that refer to the prosodic word domain include phonological processes such as Penultimate Lengthening in Yidiɲ (Dixon 1977a, 1977b); Final Voicing in Sanskrit (Selkirk 1980a); Nasal Assimilation and Stop Voicing in Greek (Nespor  & Vogel 1986); Intervocalic s-Voicing, Vowel Raising, and Vowel Lengthening in Italian (Nespor & Vogel 1986); and so forth (see also Booij & Rubach 1987; Kang 1992; Raffelsiefen 1993; Hannahs 1995a, 1995b, among others). In addition to segmental rules, it has been demonstrated that some prosodic or suprasegmental rules also refer to the prosodic word as their domain of application. For example, Nespor & Vogel (1986, 2007) argue that the Main Stress Rule in Turkish assigns stress to the last syllable within, but not across, the prosodic word domain. They observe that both monomorphemic words and derived words can form independent prosodic words in Turkish, as in (4a). By contrast, each of the members of a compound forms its own prosodic word and thus has its own primary stress on the last syllable, as in (4b) and (4c).2 (4)

a. b.

[odá]ω ‘room’ [odadá]ω ‘in the room’ [odadakí]ω ‘that which is in the room’ [düǧǘn]ω [çiçegì]ω wedding flower ‘buttercup’

c.

[çáy]ω [evì]ω tea house ‘tea house’

Other examples of stress assignment rules that have been argued to make reference to the prosodic word domain have been recorded by Dixon (1977a, 1977b) for Yidiɲ; Nespor  & Vogel (1986, 2007) for Latin; Russell (1999) for Cree; ­Raffelsiefen (1999) for English; and Vigário (2003) for European Portuguese. The prosodic word not only serves as the domain of application for various phonological rules, but also defines the scope for phonotactic constraints in some languages. For instance, Peperkamp (1997) reports that there is a phonotactic constraint in Italian barring the sound [ʎ] at the beginning of a prosodic word; Booij (1999) posits a Syllable Contact Law for Dutch, which says that the first consonant should be more sonorous than the second in a sequence of adjacent consonants within the prosodic word domain, i.e., *[VC1.C2V]ω if C2 is more sonorous than C1; Hall (1999) argues that short lax, non-low vowels in German, i.e., [ɪ], [ʏ], [ɛ], [œ], [ʊ], [ɔ], are not allowed at the right edge of the prosodic word domain; and Raffelsiefen (1999) points out that there are more consonantal clusters ω-internally than ω-initially or ω-finally in English. This shows that phonotactic constraints can also cue the prosodic word since phonotactic constraints in these languages only hold at the edges of the prosodic word domain or prosodic word internally, but not in other contexts. Moreover, the prosodic word is necessary for minimal word requirements in a number of languages that impose restrictions on the minimal size of the prosodic word (usually being at least disyllabic or bimoraic). In some languages, sequences of sounds that do not form morphosyntactic words, e.g., prefixes and suffixes, can

Prosodic word in the Fuzhou dialect  67 constitute prosodic words on their own if they exhibit minimal word requirements. For example, disyllabic affixes in Yidiɲ form independent prosodic words, since this language imposes a disyllabic minimum for its prosodic words (Dixon 1977a; Nespor & Vogel 1986: 134–136). In addition, in many languages that exhibit the minimal word syndrome, there are rules conspiring to maintain the minimality restriction. For instance, SiSwati has a disyllabic minimality requirement on the formation of imperatives, and hence the imperative form of a verb must contain the suffix -ni if the original verb stem is monosyllabic (Downing 1999). From the above discussion, we have seen that there are several types of phonological phenomena across languages that refer crucially to the prosodic word domain. These phonological phenomena have been taken as evidence and diagnostics for the prosodic word. Notice, nevertheless, that not all languages exhibit all these diagnostics for the prosodic word—it is possible that only some types of the phonological phenomena discussed above may demonstrate the existence of the prosodic word in a given language. In the following sections, I will show that the existence and the roles of the prosodic word in the Fuzhou dialect can be evidenced by several phonological phenomena in this dialect, similar to what we have seen in many other languages in the world.

3.2  Morphosyntactic word formation in the Fuzhou dialect We have seen that there are two major issues with respect to the definition and the domain of the prosodic word: (a) languages vary in the type of morphosyntactic information integrated into the prosodic word domain formation, and (b) languages differ in the type of phonological phenomena that serve as the diagnostics for the prosodic word domain. In the present section and the following sections, we will investigate the prosodic word in the Fuzhou dialect from these two aspects. Since the prosodic word is roughly the size of a morphosyntactic word, and the domain formation of the prosodic word must involve morphosyntactic information, it is necessary to begin with a survey of major types of morphosyntactic word formation in the Fuzhou dialect. For the sake of brevity, only the citation/underlying segmental structure of syllables is presented in this section. 3.2.1  Monomorphemic words Like morphosyntactic words in most languages in the world, morphosyntactic words in the Fuzhou dialect can be divided into several types according to the morphological process of their formation. It is generally accepted that a morphosyntactic word is composed of morphemes (free or bound). A word that consists of a single morpheme is called a monomorphemic word. Fuzhou has a large number of monomorphemic words, most of which contain only one syllable. There are a few disyllabic monomorphemic words in Fuzhou, while monomorphemic words containing three or more syllables are very rare. Examples of Fuzhou monomorphemic words are presented as follows in (5).

68  Prosodic word in the Fuzhou dialect (5)

a. Monosyllabic 书 tsy 红 ʔøyŋ 水 tsy 好 xo 慢 maiŋ 福 xouʔ 食 siɛʔ b. Disyllabic 葡萄 pu to 橄榄 ka laŋ 支离 tsiɛ liɛ c. Trisyllabic 新加坡 siŋ ka pho 加拿大 ka na tai

‘book’ ‘red’ ‘water’ ‘good’ ‘slow’ ‘fortune’ ‘to eat’

山 肥 讲 看 笨 雪 白

‘grape’ 枇杷 ‘olive’ 沙发 ‘miserable, tragic’ 印尼 ‘Singapore’ ‘Canada’

saŋ pui kouŋ khaŋ pouŋ suoʔ paʔ

‘mountain’ ‘fat’ ‘to say, to speak’ ‘to look’ ‘stupid’ ‘snow’ ‘white’

pi pa sa xuaʔ ʔeiŋ nɛ

‘loquat’ ‘sofa’ ‘Indonesia’

马腊加 ma laʔ ka ‘Malacca’

3.2.2  Derived words The second major type of morphosyntactic words in Fuzhou is derived words. A derived word is a new word created from one existing word through the application of some morphological process, usually affixation. There are several affixes in Fuzhou, which mostly attach to the stem to derive new nouns, as exemplified in the following subsections (cf. Li et al 1994; Li & Liang 2001; Lin 2002, among others). 3.2.2.1  Prefix 依- [ʔi44] 依- [ʔi44] is one of the most frequently used prefixes in the Fuzhou dialect. It usually attaches to a noun (usually a monosyllabic noun). Derived words composed of 依- [ʔi44] and the stem are generally used as terms of address. Most of these terms of address are kinship terms, as in (6a), and some of them are social terms, as in (6b). (6)

a.

b.

依伯 依姆 依弟 依妹 依板 依志

ʔi paʔ ʔi mu ʔi tiɛ ʔi mui ʔi peiŋ ʔi tsei

‘uncle (older than one’s father)’ ‘aunt’ ‘younger brother; little boy’ ‘younger sister; little girl’ ‘boss’ ‘comrade’

3.2.2.2  Prefix 老- [lau242]/[lo31] When used as a free morpheme, 老 means ‘old’. As a prefix, nevertheless, 老- in the Fuzhou dialect does not necessarily have the meaning ‘old’. It can be attached to a name (usually a family name) to indicate a sense of familiarity, as exemplified in (7a), which is similar to lao 老- in Mandarin Chinese. It is also used in front

Prosodic word in the Fuzhou dialect  69 of other nouns to refer to people, as in (7b), with most cases indicating a sense of familiarity. Such a sense of familiarity, however, is not present when 老- is used in animal terms, as in (7c). 老- in Fuzhou has two pronunciations, [lau242] and [lo31], which is a typical example of the distinction between literary and colloquial readings of characters (wenbaiyidu 文白异读) in this dialect (cf. Li & Liang 2001, among others). (7)

a. b. c.

老郑 老公 老爹 老蛇

lo taŋ lau kuŋ lo tia lau siɛ

‘Old Zheng’ ‘Old Lin’ 老林 lo liŋ ‘husband’ 老妈 lau ma ‘wife’ ‘government official (in the old times)’ ‘snake’ 老鼠 lo tshy ‘mouse’

3.2.2.3  Suffix -囝 [kiaŋ31] As a free morpheme, 囝 [kiaŋ31] in the Fuzhou dialect means ‘son’ or ‘child’. When it is used as a suffix, it attaches to nouns to create diminutive forms, as in (8a, b), or attaches to adjectives to refer to a kind of people, as in (8c, d). (8)

a. b. c. d.

厝囝 刀囝 戆囝 矮囝

tshuo kiaŋ to kiaŋ ŋouŋ kiaŋ ʔɛ kiaŋ

‘small house’ ‘small knife’ ‘fool’ ‘short person’

3.2.2.4  Suffix -头 [thau51] 头 [thau51] as a free morpheme in Fuzhou refers to the head of a human or an animal body, the root of some vegetables, or the top of a mountain. Like its counterpart in Mandarin Chinese, it can attach to nouns as a suffix with no easily definable meaning, as exemplified in (9). (9)

a. b. c. d. e. f.

骨头 拳头 灶头 日头 裤头 心肝头

kouʔ thau kuŋ thau tsau thau niʔ thau khou thau siŋ kaŋ thau

‘bone’ ‘fist’ ‘stove’ ‘sun’ ‘trouser waist’ ‘chest’

3.2.2.5  Suffixes -哥 [ko44] and -弟 [tiɛ242] As free morphemes 哥 [ko44] and 弟 [tiɛ242] mean ‘elder brother’ and ‘younger brother’ respectively. As suffixes, both of them can attach to nouns, adjectives, and sometimes verbs (or verb phrases). -哥 [ko44] is more often used than -弟 [tiɛ242] in the Fuzhou dialect. A derived word that contains -哥 [ko44] or -弟 [tiɛ242] usually refers to a kind of people (not necessarily male), with a derogatory sense in most cases, as in (10).

70  Prosodic word in the Fuzhou dialect (10) a. b. c. d. e. f.

兵哥 瘌哥 单身哥 白面哥 流鼻哥 派头弟

piŋ ko laʔ ko taŋ siŋ ko paʔ meiŋ ko lau phei ko phuai thau tiɛ

‘soldier’ ‘baldheaded person’ ‘bachelor (a man who is not married)’ ‘prostitute’ ‘running-nose child’ ‘stylish young man’

3.2.2.6  Suffixes -婆 [po51] and -妈 [ma31] In the Fuzhou dialect, -婆 [po51] and -妈 [ma31] are suffixes that attach to nouns, adjectives, and sometimes verbs (or verb phrases). The function of these two suffixes is similar to that of -哥 [ko44] and -弟 [tiɛ242]: derived words containing -婆 [po51] or -妈 [ma31] also usually refer to a kind of people (not necessarily female), with a derogatory sense in most cases, as exemplified in (11). (11) a. b. c. d.

菜婆 贪食婆 恶妈 见事妈

tshai po thaŋ siɛʔ po ʔouʔ ma kiɛŋ tai ma

‘Buddhist nun’ ‘voracious eater’ ‘evil person’ ‘nosy person’

3.2.2.7  Suffix -角 [koyʔ23] The suffix -角 [koyʔ23] in Fuzhou denotes masculine animals, as exemplified in (12). (12) a.

鸡角

kiɛ koyʔ

‘rooster’

b.

猪角

ty koyʔ

‘boar’

3.2.2.8  Suffix -母 [mo31] The suffix -母 [mo31] in the Fuzhou dialect denotes feminine animals, as in (13a). In some cases, a noun with the suffix -母 [mo31] stands for an object or animal with a large size, as in (13b). (13) a. b.

Feminine animals 鸡母 kiɛ mo ‘hen’ Objects/animals with a large size 缸母 kouŋ mo ‘large bowl’ 蚨蝇母 pu siŋ mo ‘big fly’

猪母

ty mo

‘sow’

3.2.2.9  Derived words containing more than one affixes A derived word in Fuzhou can be composed of the stem plus more than one affixes, as exemplified in (14).

Prosodic word in the Fuzhou dialect  71 (14) I.

II.

Stem + two suffixes a. 鸡母囝 kiɛ mo kiaŋ b. 儿囝哥 niɛ kiaŋ ko c. 拳头母 kuŋ thau mo d. 石头囝 suoʔ thau kiaŋ Prefix + stem + suffix a. 依解哥 ʔi kai ko b. 老蛇囝 lau siɛ kiaŋ c. 老虎母 lau xu mo

‘pullet’ ‘child’ ‘(big) fist’ ‘small stone’ ‘PLA soldier’ ‘small snake’ ‘tigress’

3.2.3 Compounds A compound is created through the combination of two or more free morphemes (stems). According to the syntactic and/or semantic relationship between the members, compounds in the Fuzhou dialect can be subdivided into at least four major types: (a) coordination, (b) modifier-head, (c) verb-object, and (d) verb-complement.3 3.2.3.1  Coordination compounds A coordination compound in Fuzhou is created through the combination of stems that have identical, similar, related, or opposite semantic meanings. As we can see from the examples in (15), the resulting coordination compounds may have a new meaning that is not simply the combination of the meanings of the stems, and in some cases, the resulting coordination compounds obtain the meaning from only one of the stems. (15) I. II. III. IV.

Stems with identical or similar meanings a. 暝晡 maŋ puo ‘evening’ + ‘evening’ → ‘evening’ b. 思想 sy suoŋ ‘to think’ + ‘to think’ → ‘thoughts; to miss’ Stems with related meanings a. 尺寸 tshuoʔ tshouŋ ‘foot’ + ‘inch’ → ‘length’ b. 骹手 kha tshiu ‘leg’ + ‘hand’ → ‘subordinates’ Stems with related meanings a. 事计 tai kiɛ ‘affair, job’ + ‘idea, tactics’ → ‘affair, job’ b. 国家 kuoʔ ka ‘country’ + ‘family’ → ‘country’ Stems with opposite meanings a. 开关 khai kuaŋ ‘to turn on’ + ‘to turn off’ → ‘switch’ b. 长短 touŋ tøy ‘long’ + ‘short’ → ‘discrepancy’

3.2.3.2  Modifier-head compounds According to the syntactic and/or semantic relationship between the modifier and the head, modifier-head compounds in the Fuzhou dialect can be further divided

72  Prosodic word in the Fuzhou dialect into two subtypes. The first type, attribute-noun compounds, are exemplified in (16I), and the second type, adverbial-verb/adjective compounds, are exemplified in (16II). (16)

I.

Attribute-noun compounds a. 绿豆 luoʔ tau ‘green’ + ‘bean’ → ‘mung bean’ b. 书店 tsy taiŋ ‘book’ + ‘store’ → ‘bookstore’ c. 番薯饭 xuaŋ ny puoŋ ‘sweet potato’ + ‘rice’ → ‘rice with sweet potatoes’ d. 衣裳师父 ʔi suoŋ sa ʔau ‘clothes’ + ‘master’ → ‘tailor’ e. 看命先生 khaŋ miaŋ siŋ saŋ ‘fortune-telling’ + ‘teacher’ → ‘fortune-teller’ II. Adverbial-verb/adjective compounds a. 对分 toy puoŋ ‘opposite’ + ‘to distribute’ → ‘to halve’ b. 桃红 tho ʔøyŋ ‘peach’ + ‘red’ → ‘peach pink’

3.2.3.3  Verb-object compounds Similar to Mandarin Chinese, a verb-object compound in Fuzhou is composed of two stems in such a way that the stem on the left takes the stem on the right as its object. We can see from (17) that the resulting verb-object compounds may have a new meaning that is different from the combination of the meanings of the two members. (17) a. b. c. d.

乞食 有心 食力 拍掌

khyʔ siɛʔ ʔou siŋ siɛʔ liʔ phaʔ tsuoŋ

‘to beg’ + ‘to eat’ → ‘beggar’ ‘to have’ + ‘heart’ → ‘considerate’ ‘to eat’ + ‘power’ → ‘tired; laborious’ ‘to pat, to hit’ + ‘palm’ → ‘to applaud’

3.2.3.4  Verb-complement compounds Verb-complement compounds in Fuzhou is composed of a verb plus a following complement.4 The complement stem on the right in a verb-complement compound usually denotes the result of the action indicated by the verb stem on the left. Examples of verb-complement compounds in Fuzhou are presented as in (18). (18) a. b. c. d.

看见 抓紧 改善 拍平

khaŋ kiɛŋ tsua kiŋ kai siɛŋ phaʔ paŋ

‘to look’ + ‘to see’ → ‘to see’ ‘to grab’ + ‘tight, firm’ → ‘to firmly grasp’ ‘to change’ + ‘good, kind’ → ‘to improve’ ‘to pat, to hit’ + ‘flat, level’ → ‘to equalize’

3.2.4  Reduplication: “diminutive” nouns Reduplication is a morphological process that repeats all or part of a given stem. In the Fuzhou dialect, reduplication can be subdivided into several types according to the resulting forms of reduplication. The first type of reduplication creates disyllabic “diminutive” nouns. I put the word “diminutive” in quotation marks since this group of nouns in the Fuzhou dialect does not necessarily convey the smallness of the object or quality, or a sense of

Prosodic word in the Fuzhou dialect  73 intimacy or endearment, although these nouns refer to small things in many cases. Moreover, these “diminutive” nouns not only can be used when speaking to children, but also can be used when speaking to an adult without expressing tenderness or intimacy (cf. Chen 1998). A monosyllabic noun in Fuzhou may have several different meanings, but a “diminutive” noun created by reduplicating its monosyllabic counterpart usually has only one specific nominal meaning, as exemplified in (19). (19) a. b.

骹 骹骹 耳 耳耳

kha kha kha ŋei ŋei ŋei

‘leg (of human being/animals/furniture); lower part’ ‘leg (of furniture)’ ‘ear; handle’ ‘handle’

Other examples of “diminutive” nouns are presented in (20). (20) a. b. c. d. e.

袋 → 袋袋 碗 → 碗碗 拍 → 拍拍 舀 → 舀舀 碎 → 碎碎

toy → toy toy ʔuaŋ → ʔuaŋ ʔuaŋ phaʔ → phaʔ phaʔ ʔiu → ʔiu ʔiu tshoy → tshoy tshoy

‘bag’ ‘bowl’ ‘bat’ ‘ladle’ ‘fragment’

3.2.5  Reduplication: sound-splitting words The second type of reduplication in the Fuzhou dialect gives rise to sound-­ splitting words. This group of morphosyntactic words is referred to as qiejiaoci 切脚词 (lit. ‘foot-cutting word’; see Liang 1982; Li et al. 1994; Li & Liang 2001), fenyinci 分音词 (lit. ‘sound-dividing word’; see Feng 1998), or dongci fenyinshi 动词分音式 (lit. ‘sound-dividing form of verbs’; see Chen 1998), in the literature. It is also called “word derivation by the principle of fanqie 反切”5 (Chan 1985) or “VL pattern” (Li 2002). In the Fuzhou dialect, a sound-splitting word is formed through a morphological process of partial reduplication and initial consonant substitution. Specifically, the first syllable of the resulting sound-splitting word contains the initial (if any), the glide (if any), and the nuclear vowel of the original monosyllable, but the original syllable ending (if any) is not copied. The second syllable, by contrast, contains an initial consonant [l] and the entire final including the glide (if any), the nuclear vowel, and the ending (if any), as well as the tone of the original monosyllable. For example, a monosyllable with the structure CVV is partially reduplicated as CV.CVV, in which only part of the original syllable is copied. Then, a consonant [l] replaces the original initial of the second syllable, resulting in CV.lVV. As discussed in Chapter 1, the syllable structure in the Fuzhou dialect is (C)(G) V(V/C/VC), where only the nuclear vowel V is required while all the other segments are optional. Thus, the formation process of sound-splitting words can be presented as in (21). (21) (C)(G)V(V/C/VC) → (C)(G)V. l (G)V(V/C/VC)

74  Prosodic word in the Fuzhou dialect With respect to the formation and internal structure of sound-splitting words in Fuzhou, there are several subtypes, as exemplified in (22). (22)

I. II. III. IV.

(C)(G)V → (C)(G)V. l (G)V a. 糊 ku → ku lu b. 窝 ʔuo → ʔuo luo (C)(G)VC → (C)(G)V. l (G)VC a. 塌 thaʔ → tha laʔ b. 碾 niɛŋ → niɛ liɛŋ (C)(G)VV → (C)(G)V. l (G)VV a. 吊 tau → ta lau b. 歪 ʔuai → ʔua luai (C)VVC → (C)V. l VVC a. 揬 thouʔ → tho louʔ b. 夹 keiʔ → ke leiʔ

‘to paste’ ‘to curl up’ ‘to collapse’ ‘to grind’ ‘to hang’ ‘to recline’ ‘to poke’ ‘to bind’

3.2.6 Reduplication: reduplicated adjectives The third type of reduplication in Fuzhou is the reduplication of adjectives, usually indicating a higher degree of the quality denoted by the original adjective (cf. Chen 1998; Li & Liang 2001, among others). Using A to stand for a monosyllabic adjective and AB for a disyllabic adjective, the reduplication form of a monosyllabic adjective is AA, and a disyllabic adjective can be reduplicated as AABB, AAB, and ABB. AABB is the most common reduplicated form for a disyllabic adjective while ABB is the least. Examples of these major forms of reduplicated adjectives are given in (23). (23) I.

II.

A → AA a. 碎 → 碎碎 b. 热 → 热热 c. 慢 → 慢慢 d. 凊 → 凊凊

tshoy → tshoy tshoy ʔiɛʔ → ʔiɛʔ ʔiɛʔ maiŋ → maiŋ maiŋ tsheiŋ → tsheiŋ tsheiŋ

AB → AABB a. 四角 → 四四角角 b. 舒畅 → 舒舒畅畅 c. 妥当 → 妥妥当当 d. 平安 → 平平安安

III. AB → AAB a. 四角 → 四四角 b. 舒畅 → 舒舒畅

‘shattered’ → ‘(very) shattered’6 ‘hot’ → ‘(very) hot’ ‘slow’ → ‘(very) slow’ ‘cold’ → ‘(very) cold’

sei koyʔ → sei sei koyʔ koyʔ ‘square’ → ‘(perfectly) square’ tshy thuoŋ → tshy tshy thuoŋ thuoŋ ‘comfortable’ → ‘(very) comfortable’ tho touŋ → tho tho touŋ touŋ ‘appropriate’ → ‘(very) appropriate’ piŋ ʔaŋ → piŋ piŋ ʔaŋ ʔaŋ ‘safe, peaceful’ → ‘(very) safe, peaceful’ sei koyʔ → sei sei koyʔ ‘square’ → ‘(perfectly) square’ tshy thuoŋ → tshy tshy thuoŋ ‘comfortable’ → ‘(very) comfortable’

Prosodic word in the Fuzhou dialect  75 c. 光鲜 → 光光鲜 IV. AB → ABB a. 舒畅 → 舒畅畅 b. 妥当 → 妥当当 c. 平安 → 平安安

kuoŋ tshiɛŋ → kuoŋ kuoŋ tshiɛŋ ‘neat and clean’ → ‘(very) neat and clean’ tshy thuoŋ → tshy thuoŋ thuoŋ ‘comfortable’ → ‘(very) comfortable’ tho touŋ → tho touŋ touŋ ‘appropriate’ → ‘(very) appropriate’ piŋ ʔaŋ → piŋ ʔaŋ ʔaŋ ‘safe, peaceful’ → ‘(very) safe, peaceful’

It is noteworthy that reduplicated adjectives in the Fuzhou dialect generally cannot be used as the predicate or attributive on their own. They are bound on the right side and thus need to take elements such as 势 [siɛ213], 式 [seiʔ23], 喏 [luoʔ23], or 其 [ki0] (cf. Chen 1998; Li & Liang 2001; Li 2002, among others). These elements are enclitics in the Fuzhou dialect, as will be discussed in Chapter 4. 3.2.7 Summary So far, I have given an overview of the major morphological processes of morphosyntactic word formation in the Fuzhou dialect as well as subtypes of each morphological process. In Section 3.3, I will investigate the phonological behavior of these morphosyntactic words in Fuzhou by examining the application of various Fuzhou phonological rules, including TS, MTS, FC, and CL.

3.3 Fuzhou morphosyntactic words and lexical-level phonological rule application As we have seen in Chapter  1, most Fuzhou phonological rules, including TS, MTS, FC, and CL, can apply at the lexical level. It has been argued in some previous studies that the domain of application for rules such TS and FC is the foot, which, as I have pointed out in Chapter 2, is invalid. In this section, I am going to examine the lexical-level application of these rules with respect to the morphosyntactic word formation in the Fuzhou dialect. Each following subsection presents the observation and analysis of the application of one phonological rule in different types of morphosyntactic words. This investigation will finally enable us to find out the domain of application for these Fuzhou phonological rules. 3.3.1  Application of TS in Fuzhou morphosyntactic words 3.3.1.1  Application of TS in monomorphemic words Let us now begin with the application of the TS rule at the lexical level. It is generally accepted that the TS rule can apply to lexical items, namely morphosyntactic words in the Fuzhou dialect. Among the major types of Fuzhou morphosyntactic words discussed in Section 3.2, monomorphemic words composed of two or more

76  Prosodic word in the Fuzhou dialect syllables constitute a domain of application for TS, as can be seen in (24), where sandhi tones are marked in bold. (24) Application of TS in monomorphemic words I. Disyllabic a. 沙发 sa44 xuaʔ23 → sa51 ʔuaʔ23 b. 枇杷 pi51 pa51 → pi31 βa51 II. Trisyllabic a. 新加坡 siŋ44 ka44 pho44 → siŋ21 ŋa44 βo44 b. 加拿大 ka44 na31 tai242 → ka21 na44 lai242

‘sofa’ ‘loquat’ ‘Singapore’ ‘Canada’

3.3.1.2  Application of TS in derived words The TS rule in Fuzhou can also apply in the domain formed by various subtypes of derived words, as exemplified in (25). (25)

Application of TS in derived words I. Prefix + stem a. 依伯 ʔi44 paʔ23 → ʔi51 βaʔ23 b. 老鼠 lo31 tshy31 → lo24 ʒy31

‘uncle (older than one’s father)’ ‘mouse’

II.

Stem + suffix a. 厝囝 b. 裤头 c. 瘌哥 d. 派头弟 e. 贪食婆 f. 见事妈 g. 鸡角 h. 蚨蝇母

III.

Stem + two suffixes a. 鸡母囝 kiɛ44 mo31 kiaŋ31 → kiɛ21 mo24 ʔiaŋ31 b. 拳头母 kuŋ51 thau51 mo31 → kuŋ31 nau31 mo31

IV.

Prefix + stem + suffix a. 老蛇囝 lau242 siɛ51 kiaŋ31 → lau51 liɛ31 ʔiaŋ31 b. 依解哥 ʔi44 kai31 ko44 → ʔi21 ʔai21 ʔo44

tshuo213 kiaŋ31 → tshuo51 ʔiaŋ31 khou213 thau51 → khu44 lau51 laʔ23 ko44 → laʔ21 ʔo44 phuai213 thau51 tiɛ242 → phuai51 lau21 liɛ242 thaŋ44 siɛʔ5 po51 → taŋ51 niɛʔ31 βo51 kiɛŋ213 tai242 ma31 → kiɛŋ21 nai51 ma31 kiɛ44 koyʔ23 → kiɛ51 ʔoyʔ23 pu51 siŋ51 mo31 → pu31 liŋ31 mo31

‘small house’ ‘trouser waist’ ‘baldheaded person’ ‘stylish young man’ ‘voracious eater’ ‘nosy person’ ‘rooster’ ‘big fly’ ‘pullet’ ‘(big) fist’ ‘small snake’ ‘PLA soldier’

3.3.1.3  Application of TS in compounds In addition to the domain formed by polysyllabic monomorphemic words and derived words, the TS rule also applies within the domain formed by different subtypes of compounds in the Fuzhou dialect, as we can see from the examples in (26).

Prosodic word in the Fuzhou dialect  77 (26) Application of TS in compounds I. Coordination a. 暝晡 maŋ51 puo44 b. 事计 tai242 kiɛ213 II.

Modifier-head a. 绿豆 b. 共产主义 c.

对分

→ maŋ44 muo44 → tai51 ʔiɛ213

‘evening’ ‘affair, job’

luoʔ5 tau242 → luoʔ21 lau242 ‘mung bean’ køyŋ242 saŋ31 tsuo31 ŋiɛ242 → kyŋ21 naŋ21 ʒuo44 ŋiɛ242 ‘Communism’ toy213 puoŋ44 → tøy44 muoŋ44 ‘to halve’

III.

Verb-object a. 食力 b. 有心

IV.

Verb-complement a. 看见 khaŋ213 kiɛŋ213 b. 改善 kai31 siɛŋ242

siɛʔ5 liʔ5 ʔou242 siŋ44

→ siɛʔ31 liʔ5 → ʔu44 liŋ44

‘tired; laborious’ ‘considerate’

→ khaŋ51 ŋiɛŋ213 → kai44 liɛŋ242

‘to see’ ‘to improve’

3.3.1.4  Application of TS in “diminutive” nouns In the domain formed by “diminutive” nouns, however, the TS rule is blocked, as exemplified in (27). The tones in question are marked in bold. (27) Application of TS in “diminutive” nouns a. 袋袋 toy242 toy242 → *tøy51 toy242 b. 碗碗 ʔuaŋ31 ʔuaŋ31 → *ʔuaŋ24 ʔuaŋ31 c. 拍拍 phaʔ23 phaʔ23 → *phaʔ51 phaʔ23 d. 舀舀 ʔiu31 ʔiu31 → *ʔiu24 ʔiu31 e. 碎碎 tshoy213 tshoy213 → *tshøy51 tshoy213

‘bag’ ‘bowl’ ‘bat’ ‘ladle’ ‘fragment’

The sandhi tones of the syllables in “diminutive” nouns in (27) are presented in (28), which in fact result from the application of another type of tone sandhi rule in the Fuzhou dialect, namely the MTS rule discussed in Chapter 1. (28) Tonal patterns of “diminutive” nouns in (27) a. 袋袋 toy242 toy242 → tøy21 toy242 b. 碗碗 ʔuaŋ31 ʔuaŋ31 → ʔuaŋ31 ʔuaŋ31 c. 拍拍 phaʔ23 phaʔ23 → phaʔ21 phaʔ23 d. 舀舀 ʔiu31 ʔiu31 → ʔiu31 ʔiu31 e. 碎碎 tshoy213 tshoy213 → tshøy21 tshoy213

‘bag’ ‘bowl’ ‘bat’ ‘ladle’ ‘fragment’

78  Prosodic word in the Fuzhou dialect 3.3.1.5  Application of TS in sound-splitting words Since sound-splitting words in Fuzhou are created through the reduplication of monosyllabic words, I  assume that the two syllables in a sound-splitting word have the same citation tone. TS is blocked within the domain formed by soundsplitting words, as exemplified in (29). (29) Application of TS in sound-splitting words I. (C)(G)V → (C)(G)V. l (G)V a. 跳 thiu213 → thiu213 liu213 → *thiu51 liu213 b. 窝 ʔuo31 → ʔuo31 luo31 → *ʔuo24 luo31

‘to jump’ ‘to curl up’

II.

(C)(G)VC → (C)(G)V. l (G)VC a. 塌 thaʔ23 → tha23 laʔ23 → *tha51 laʔ23 b. 碾 niɛŋ31 → niɛ31 liɛŋ31 → *niɛ24 liɛŋ31

‘to collapse’ ‘to grind’

III.

(C)(G)VV → (C)(G)V. l (G)VV a. 吊 tau213 → ta213 lau213 → *ta51 lau213 b. 皱 tsau213 → tsa213 lau213 → *tsa51 lau213

‘to hang’ ‘to wrinkle’

IV.

(C)VVC → (C)V. l VVC a. 揬 thouʔ23 → tho23 louʔ23 → *tho51 louʔ23 b. 扳 paiŋ213 → pa213 laiŋ213 → *pa51 laiŋ213

‘to poke’ ‘to turn over’

The sandhi tones of the syllables in sound-splitting words in (29) are presented in (30), which is also the result of the application of the MTS rule. (30) Tonal patterns of sound-splitting words in (29) I. (C)(G)V → (C)(G)V. l (G)V a. 跳 thiu213 → thiu213 liu213 → thiu31 liu213 b. 窝 ʔuo31 → ʔuo31 luo31 → ʔuo31 luo31

‘to jump’ ‘to curl up’

II.

(C)(G)VC → (C)(G)V. l (G)VC a. 塌 thaʔ23 → tha23 laʔ23 → tha21 laʔ23 b. 碾 niɛŋ31 → niɛ31 liɛŋ31 → niɛ31 liɛŋ31

‘to collapse’ ‘to grind’

III.

(C)(G)VV → (C)(G)V. l (G)VV a. 吊 tau213 → ta213 lau213 → ta21 lau213 b. 皱 tsau213 → tsa213 lau213 → tsa21 lau213

‘to hang’ ‘to wrinkle’

IV.

(C)VVC → (C)V. l VVC a. 揬 thouʔ23 → tho23 louʔ23 → tho21 louʔ23 b. 扳 paiŋ213 → pa213 laiŋ213 → pa21 laiŋ213

‘to poke’ ‘to turn over’

Prosodic word in the Fuzhou dialect  79 3.3.1.6  Application of TS in reduplicated adjectives Unlike the domain formed by “diminutive” nouns and sound-splitting words, the domain formed by reduplicated adjectives in Fuzhou can serve as the domain for the application of TS, as exemplified in (31). (31)

Application of TS in reduplicated adjectives I. AA a. 热热 ʔiɛʔ5 ʔiɛʔ5 → ʔiɛʔ31 ʔiɛʔ5 b. 碎碎 tshoy213 tshoy213 → tshøy51 tshoy213 c. 慢慢 maiŋ242 maiŋ242 → meiŋ51 maiŋ242 d. 凊凊 tsheiŋ213 tsheiŋ213 → tshiŋ51 tsheiŋ213

‘(very) hot’ ‘(very) shattered’7 ‘(very) slow’ ‘(very) cold’

II. AABB a. 四四角角

sei213 sei213 koyʔ23 koyʔ23→si21 li21 ʔøyʔ44 koyʔ23 ‘(perfectly) square’ b. 舒舒畅畅 tshy44 tshy44 thuoŋ213 thuoŋ213 → tshy21 ʒy21 luoŋ51 nuoŋ213 ‘(very) comfortable’ c. 妥妥当当 tho31 tho31 touŋ213 touŋ213 → tho21 lo21 louŋ51 nouŋ213 ‘(very) appropriate’ d. 平平安安 piŋ51 piŋ51 ʔaŋ44 ʔaŋ44 → piŋ21 miŋ21 ŋaŋ44 ŋaŋ44 ‘(very) safe, peaceful’

III. AAB a. 四四角

sei213 sei213 koyʔ23 → si21 li21 ʔoyʔ44 ‘(perfectly) square’ b. 舒舒畅 tshy44 tshy44 thuoŋ213 → tshy21 ʒy21 luoŋ51 ‘(very) comfortable’ c. 光光鲜 kuoŋ44 kuoŋ44 tshiɛŋ44 → kuoŋ21 ŋuoŋ44 ʒiɛŋ44 ‘(very) neat and clean’

IV. ABB a. 舒畅畅

tshy44 thuoŋ213 thuoŋ213 → tshy21 luoŋ51 nuoŋ213 ‘(very) comfortable’ b. 妥当当 tho31 touŋ213 touŋ213 → tho21 louŋ51 nouŋ213 ‘(very) appropriate’ c. 平安安 piŋ51 ʔaŋ44 ʔaŋ44 → piŋ21 ŋaŋ44 ŋaŋ44 ‘(very) safe, peaceful’

3.3.1.7 Summary Based on the discussion above, we can find that TS applies in the domain formed by polysyllabic monomorphemic words, derived words, compounds, and reduplicated adjectives in the Fuzhou dialect. It does not apply to monosyllabic monomorphemic words (due to the lack of appropriate phonological environment),

80  Prosodic word in the Fuzhou dialect “diminutive” nouns, and sound-splitting words. Tonal patterns of “diminutive” nouns and sound-splitting words, in fact, are derived by the MTS rule in this dialect. 3.3.2  Application of MTS in Fuzhou morphosyntactic words As discussed in Chapter 1, there is another type of tone sandhi rule in addition to the TS rule in the Fuzhou dialect, namely the MTS rule, whose application is restricted to some particular disyllabic morphosyntactic words formed through reduplication. From the discussion in Section 3.3.1, we can find that MTS applies in the domain formed by “diminutive” nouns and sound-splitting words in the Fuzhou dialect. As we have seen in (28) and (30), the sandhi tone of the first syllable within the domain of MTS is only conditioned by its own citation tone. Examples in (28) and (30) are reproduced as in (32) and (33) respectively, in which the TS rule is blocked.

(32) Application of MTS in “diminutive” nouns a. 袋袋 toy242 toy242 → tøy21 toy242 (MTS) *tøy51 toy242 (TS) 31 31 b. 碗碗 ʔuaŋ ʔuaŋ → ʔuaŋ31 ʔuaŋ31 (MTS) *ʔuaŋ24 ʔuaŋ31 (TS) h 23 h 23 c. 拍拍 p aʔ p aʔ → phaʔ21 phaʔ23 (MTS) *phaʔ51 phaʔ23 (TS) 31 31 d. 舀舀 ʔiu ʔiu → ʔiu31 ʔiu31 (MTS) *ʔiu24 ʔiu31 (TS) h 213 h 213 e. 碎碎 ts oy ts oy → tshøy21 tshoy213 (MTS) *tshøy51 tshoy213 (TS) (33) Application of MTS in sound-splitting words I. (C)(G)V → (C)(G)V. l (G)V a. 跳 thiu213 → thiu213 liu213 → thiu31 liu213 (MTS) *thiu51 liu213 (TS) b. 窝 ʔuo31 → ʔuo31 luo31 → ʔuo31 luo31 (MTS) *ʔuo24 luo31 (TS) II.

(C)(G)VC → (C)(G)V. l (G)VC a. 塌 thaʔ23 → tha23 laʔ23 → b. 碾

III.

niɛŋ31 → niɛ31 liɛŋ31 →

(C)(G)VV → (C)(G)V. l (G)VV a. 吊 tau213 → ta213 lau213 → b. 皱

tsau213 → tsa213 lau213 →

tha21 laʔ23 (MTS) *tha51 laʔ23 (TS) niɛ31 liɛŋ31 (MTS) *niɛ24 liɛŋ31 (TS) ta21 lau213 (MTS) *ta51 lau213 (TS) tsa21 lau213 (MTS) *tsa51 lau213 (TS)

‘bag’ ‘bowl’ ‘bat’ ‘ladle’ ‘fragment’

‘to jump’ ‘to curl up’

‘to collapse’ ‘to grind’

‘to hang’ ‘to wrinkle’

Prosodic word in the Fuzhou dialect  81 IV.

(C)VVC → (C)V. l VVC a. 揬 thouʔ23 → tho23 louʔ23 → b. 扳

paiŋ213 → pa213 laiŋ213 →

tho21 louʔ23 (MTS) *tho51 louʔ23 (TS) pa21 laiŋ213 (MTS) *pa51 laiŋ213 (TS)

‘to poke’ ‘to turn over’

Thus, tone sandhi patterns of the MTS rule can be reproduced as in Table 3.1. Table 3.1  MTS in the Fuzhou dialect (refined) T (monosyllabic word) 44 51/31/5 213/242/23

T’ + T (“diminutive” nouns and sound-splitting words) 44 + 44 (cf. Chen & Norman 1965a; Chen 1998) 31 + 44/51/31/5 (cf. Liang 1982; Lin 2002) 21 + 213/242/23

The MTS rule does not apply in the domain formed by polysyllabic monomorphemic words, derived words, compounds, or reduplicated adjectives, in which only TS applies. Clearly, the TS rule and the MTS rule are in complementary distribution in terms of their application in Fuzhou morphosyntactic words— they can both apply in the domain formed by morphosyntactic words, while their application/blocking within such a domain is determined by the subtypes of morphosyntactic words in this dialect. Inspired by Zhang & Yu (2009) and Yu & Yin (2014), You (2017, 2018c) proposes an account for the relationship between TS and MTS in the Fuzhou dialect under the framework of lexical phonology (Kiparsky 1982b, 1985; Mohanan 1982; Pulleyblank 1986, among others). According to lexical phonology, there are two classes of phonological rules, namely lexical rules and post-lexical rules. The lexical phonological rules and morphological processes, or word formation rules (WFRs), can be partitioned into a series of levels in the lexicon, and each level has the lexical phonological rules distinctive of that level. By resorting to the theory of lexical phonology, You (2017, 2018c) argues that TS and MTS are placed at different levels within the Fuzhou lexicon presented in Figure 3.1, so are the morphological processes such as affixation, compounding, and different types of reduplication.8 From Figure 3.1, we can find that “diminutive” nouns and sound-splitting words are formed by reduplication 1 at Level 1 in the lexicon, where MTS applies. By contrast, other morphosyntactic words such as derived words, compounds, and reduplicated adjectives, are formed at Level 2, where TS applies. This model thus explains why MTS and TS apply to different types of morphosyntactic words in the Fuzhou dialect, which can be illustrated by a pair of typical examples in (34). The two morphosyntactic words (34a) and (34b) have the same characters and citation tones/underlying forms but have different surface forms since the former is a “diminutive” noun while the latter is a reduplicated adjective. According to Figure 3.1, these two words are formed at Level 1 and Level 2 respectively. Hence, MTS applies to (34a) while TS applies to (34b), yielding different sandhi tones.

82  Prosodic word in the Fuzhou dialect

Underived lexical entries WFRs (reduplication 1: “diminutive” nouns and sound-splitting words; ...)

MTS

Level 1

WFRs (affixation; compounding; reduplication 2: reduplicated adjectives...)

TS

Level 2

WFRs

Syntax

Level n

... ...

... ...

Post-lexical phonology

Figure 3.1  Lexical phonology in the Fuzhou dialect Source: You 2017, 2018c.

(34) a. 碎碎 b. 碎碎

tshoy213 → tshøy21 tshoy213 tshoy213 → tshøy51 tshoy213

(MTS) (TS)

‘fragment’ ‘(very) shattered’

Since the primary concern in this chapter is to identify the domain of a­ pplication for phonological rules such as TS and MTS, the readers are referred to You (2017) for detailed discussion on the relationship between TS and MTS. 3.3.3  Application of FC in Fuzhou morphosyntactic words 3.3.3.1  Application of FC in monomorphemic words Let us now move on to the application of the FC rule at the lexical level. As discussed in Chapter 1, FC is a tonally conditioned phonological process, which specifies that in a particular domain where non-terminal syllables can only have yinping, yangping, shang, yangru, half yinqu, or half yangqu tones, if the final of a non-terminal syllable belongs to Group B variants of alternating finals, it will be replaced by its Group A counterpart on the surface. Data from the Fuzhou dialect show that FC applies in the domain formed by polysyllabic monomorphemic words, as exemplified in (35), in which sandhi finals are marked in bold.

Prosodic word in the Fuzhou dialect  83 (35) Application of FC in monomorphemic words9 a. 蟋蟀 seiʔ23 souʔ23 → siʔ44 souʔ23 ‘cricket’ b. 鹭鸶 lou242 si44 → lu44 si44 ‘egret’ 3.3.3.2  Application of FC in derived words Derived words in the Fuzhou dialect can also constitute the domain of application for the FC rule, as can be seen in (36). (36)

Application of FC in derived words10 I． Stem + suffix a. 竹囝 tøyʔ23 kiaŋ31 → tyʔ24 kiaŋ31 b. 裤头 khou213 thau51 → khu44 lau51 c. 恶妈 ʔouʔ23 ma31 → ʔuʔ24 ma31 d. 虱母 saiʔ23 mo31 → seiʔ24 mo31 e. 豆腐囝 ta242 xou242 kiaŋ31 → ta21 ʔu51 ʔiaŋ31 f. 流鼻哥 lau51 phei213 ko44 → lau21 βi44 ʔo44 II. Stem + two suffixes a. 鸡角囝 kiɛ44 koyʔ23 kiaŋ31 → kiɛ21 ʔøyʔ51 ʔiaŋ31

‘small bamboo’ ‘trouser waist’ ‘bad person’ ‘(big) louse’ ‘jellied bean curd’ ‘running-nose child’ ‘pullet’

3.3.3.3  Application of FC in compounds The FC rule also applies within the domain formed by various subtypes of compounds in the Fuzhou dialect, as we can see from the examples in (37). (37) Application of FC in compounds I. Coordination a. 备办 pei242 paiŋ242 → pi51 βaiŋ242 b. 诉念 sou213 naiŋ242 → su51 naiŋ242 II.

Modifier-head a. 醋心 b. 共产主义 c.

对分

‘to prepare’ ‘to narrate, to recount’

‘jealous mind’ tshou213 siŋ44 → tshu44 liŋ44 køyŋ242 saŋ31 tsuo31 ŋiɛ242 → kyŋ21 naŋ21 ʒuo44 ŋiɛ242 ‘Communism’ toy213 puoŋ44 → tøy44 muoŋ44 ‘to halve’

III.

Verb-object a. 咒嘴 b. 有心

IV.

Verb-complement a. 注紧 tsoy213 kiŋ31 → tsøy51 kiŋ31 b. 睏着 khouŋ213 tuoʔ5 → khuŋ44 tuoʔ5

tsou213 tshui213 → tsu51 ʒui213 ʔou242 siŋ44 → ʔu44 liŋ44

‘to swear’ ‘considerate’ ‘to pay attention to’ ‘to fall asleep’

84  Prosodic word in the Fuzhou dialect 3.3.3.4  Application of FC in “diminutive” nouns Besides the domain formed by polysyllabic monomorphemic words, derived words, and compounds, the FC rule also applies in the domain formed by “diminutive” nouns, where MTS applies, as exemplified in (38). (38) Application of FC in “diminutive” nouns a. 袋袋 toy242 toy242 → tøy21 toy242 b. 锯锯 køy213 køy213 → ky21 køy213 c. 碎碎 tshoy213 tshoy213 → tshøy21 tshoy213

‘bag’ ‘saw’ ‘fragment’

3.3.3.5  Application of FC in sound-splitting words If we take a look back at Table 1.3 in Chapter 1, we can find that Group B variants of Fuzhou alternating finals all contain a diphthong, or two vocalic elements. Hence, it seems that FC is not able to apply in the domain formed by soundsplitting words, because the first syllable in a sound-splitting word (C)(G)V. l (G) V(V/C/VC) always has only one vocalic element. However, evidence from the Fuzhou dialect shows that FC may apply to sound-splitting words as well. For example, it has been reported in Liang (1982) that 揬 [thouʔ23 → tho23 louʔ23 → tho21 louʔ23] ‘to poke’ in (33IVa) can also be pronounced as [thu21 louʔ23], in which the Group B variant [ouʔ] is first changed into the correspondent Group A variant [uʔ] by FC, and then drops its consonant coda. Some similar examples presented in Liang (1982) are cited as in (39), where finals in question are marked in bold. According to Liang, these sound-splitting words have two types of pronunciations. One is derived on the basis of the application of FC, as presented on the left of the slash, while the other is derived simply by deleting the segment(s) following the nuclear vowel, as presented on the right of the slash. As can be seen from the examples in (39), FC is an optional rule within the domain formed by sound-splitting words. (39) a. b. c. d. e.

拂 戽 撙 唧 歙

xouʔ23 xou213 tshouŋ242 tseiʔ23 xeiʔ23

→ → → → →

xu21 louʔ23 / xo21 louʔ23 xu21 lou213 / xo21 lou213 tshu21 louŋ242 / tsho21 louŋ242 tsi21 leiʔ23 / tse21 leiʔ23 xi21 leiʔ23 / xe21 leiʔ23

‘to flit, to sweep’ ‘to sprinkle’ ‘to wring’ ‘to squeeze’ ‘to sniff’

3.3.3.6  Application of FC in reduplicated adjectives The FC rule can also apply to all the subtypes of reduplicated adjectives, as exemplified in (40).

Prosodic word in the Fuzhou dialect  85 (40) Application of FC in reduplicated adjectives I. AA a. 碎碎 tshoy213 tshoy213 → tshøy51 tshoy213 ‘(very) shattered’ b. 慢慢 maiŋ242 maiŋ242 → meiŋ51 maiŋ242 ‘(very) slow’ c. 凊凊 tsheiŋ213 tsheiŋ213 → tshiŋ51 tsheiŋ213 ‘(very) cold’ II. AABB a. 四四角角 sei213 sei213 koyʔ23 koyʔ23 → si21 li21 ʔøyʔ44 koyʔ23 ‘(perfectly) square’ III. AAB a. 四四角 sei213 sei213 koyʔ23 → si21 li21 ʔoyʔ44 ‘(perfectly) square’ IV. ABB a. 四角角 sei213 koyʔ23 koyʔ23 → si21 ʔøyʔ44 koyʔ23 ‘(perfectly) square’ 3.3.3.7 Summary On the basis of the discussion above, we can find that FC can apply to all major types of morphosyntactic words in the Fuzhou dialect (with the only exception of monosyllabic monomorphemic words since one single syllable never undergoes tone sandhi and final change). The lexical-level application of FC is thus different from TS and MTS, in that TS and MTS only apply within the domain formed by certain types of morphosyntactic words. Another noteworthy point is that the application of FC is not always obligatory in the domain formed by Fuzhou morphosyntactic words—it is optional in the domain formed by sound-splitting words. 3.3.4  Application of CL in Fuzhou morphosyntactic words 3.3.4.1  Application of CL in monomorphemic words Similar to TS and FC, the CL rule in the Fuzhou dialect can also apply within the domain formed by polysyllabic monomorphemic words, as shown in (41), in which sandhi initial consonants are marked in bold. (41) Application of CL in monomorphemic words I. Disyllabic a. 沙发 sa44 xuaʔ23 → sa51 ʔuaʔ23 b. 枇杷 pi51 pa51 → pi31 βa51 II. Trisyllabic a. 新加坡 siŋ44 ka44 pho44 → siŋ21 ŋa44 βo44 b. 加拿大 ka44 na31 tai242 → ka21 na44 lai242

‘sofa’ ‘loquat’ ‘Singapore’ ‘Canada’

86  Prosodic word in the Fuzhou dialect 3.3.4.2  Application of CL in derived words The CL rule also refers to the domain constituted by derived words in the Fuzhou dialect as the domain of application, as exemplified in (42). (42)

Application of CL in derived words I. Prefix + stem a. 依伯 ʔi44 paʔ23 → ʔi51 βaʔ23 b. 老鼠 lo31 tshy31 → lo24 ʒy31

‘uncle (older than one’s father)’ ‘mouse’

II．

Stem + suffix11 a. 厝囝 b. 裤头 c. 兵哥 d. 派头弟 e. 贪食婆 f. 鸡角

III.

Stem + two suffixes ‘child’ a. 儿囝哥 niɛ51 kiaŋ31 ko44 → niɛ21 ʔiaŋ21 ŋo44 b. 石头囝 suoʔ5 thau51 kiaŋ31 → suoʔ31 lau31 ʔiaŋ31 ‘small stone’

IV.

Prefix + stem + suffix a. 老蛇囝 lau242 siɛ51 kiaŋ31 → lau51 liɛ31 ʔiaŋ31 b. 依解哥 ʔi44 kai31 ko44 → ʔi21 ʔai21 ʔo44

tshuo213 kiaŋ31 → tshuo51 ʔiaŋ31 khou213 thau51 → khu44 lau51 piŋ44 ko44 → piŋ44 ŋo44 phuai213 thau51 tiɛ242 → phuai51 lau21 liɛ242 thaŋ44 siɛʔ5 po51 → taŋ51 niɛʔ31 βo51 kiɛ44 koyʔ23 → kiɛ51 ʔoyʔ23

‘small house’ ‘trouser waist’ ‘soldier’ ‘stylish young man’ ‘voracious eater’ ‘rooster’

‘small snake’ ‘PLA soldier’

3.3.4.3  Application of CL in compounds Similar to polysyllabic monomorphemic words and derived words, compounds in Fuzhou can also form the domain of application for CL, as can be seen in (43). (43) Application of CL in compounds I. II.

III. IV.

Coordination a. 暝晡 b. 事计 Modifier-head a. 绿豆 b. 共产主义

maŋ51 puo44 → maŋ44 muo44 tai242 kiɛ213 → tai51 ʔiɛ213

‘evening’ ‘affair, job’

luoʔ5 tau242 → luoʔ21 lau242 ‘mung bean’ køyŋ242 saŋ31 tsuo31 ŋiɛ242 → kyŋ21 ‘Communism’ naŋ21 ʒuo44 ŋiɛ242 toy213 puoŋ44 → tøy44 muoŋ44 ‘to halve’

c. 对分 Verb-object a. 咒嘴 tsou213 tshui213 → tsu51 ʒui213 b. 有心 ʔou242 siŋ44 → ʔu44 liŋ44 Verb-complement a. 看见 khaŋ213 kiɛŋ213 → khaŋ51 ŋiɛŋ213 b. 改善 kai31 siɛŋ242 → kai44 liɛŋ242

‘to swear’ ‘considerate’ ‘to see’ ‘to improve’

Prosodic word in the Fuzhou dialect  87 3.3.4.4  Application of CL in “diminutive” nouns Controversy exists in the literature as to whether CL applies in the domain formed by “diminutive” nouns. Li (2002) and Lin (2002) both claim that the original initial of the second syllable of a “diminutive” noun remains unchanged. Data collected in Li et al. (1994), Chen (1998), Feng (1998), and Li (1998) provide evidence to support this claim. Li & Liang (2001), however, argue that the initial of the second syllable in a “diminutive” noun undergoes CL. Examples listed in Li & Liang (2001) are given in (44), with sandhi forms of initial consonants in question marked in bold. (44) Examples of CL in “diminutive” nouns from Li & Liang (2001: 195). a. 杯杯 pui44 → pui31 βui44 ‘cup’ b. 桶桶 thøyŋ31 → thøyŋ31 nøyŋ31 ‘bucket’ Data elicited from my informants, however, show that the examples in (44) are quite suspicious. According to my informants, it is only when they produce “diminutive” nouns such as 杯杯 [pui31 pui44] and 桶桶 [thøyŋ31 thøyŋ31] very quickly in the speech that the initial of the second syllable might be changed; otherwise, the pronunciation of “diminutive” nouns in (44) sounds odd to them. Data from my informants as well as those in other published sources therefore suggest that the CL rule should be blocked in the domain formed by “diminutive” nouns, although the domain formed by “diminutive” nouns provides the phonological environment for the application of CL. 3.3.4.5  Application of CL in sound-splitting words We have seen in Table 1.8 in Chapter 1 that, if the initial of a non-first syllable in a domain is [l], it becomes [n] when preceded by the nasal coda [ŋ]. Otherwise, [l] in the syllable initial position remains unchanged. Notice that the first syllable of a sound-splitting word (C)(G)V. l (G)V(V/C/VC) never contains the nasal coda [ŋ]. Due to such a lack of phonological environment, the CL rule never applies within the domain formed by sound-splitting words. 3.3.4.6  Application of CL in reduplicated adjectives Data from the Fuzhou dialect show that CL fails to apply within the domain formed by the reduplication of monosyllabic adjectives, which has been noticed by some previous studies (e.g., Chen 1998; Li & Liang 2001; Li 2002, among others). This can be exemplified in (45). (45) Blocking of CL in reduplicated forms of monosyllabic adjectives a. 白白 paʔ5 paʔ5 → paʔ31 paʔ5 *paʔ31 βaʔ5 ‘(very) white’ h 213 h 213 h 51 h 213 b. 凊凊 ts eiŋ ts eiŋ → ts iŋ ts eiŋ *tshiŋ51 ʒeiŋ213 ‘(very) cold’

88  Prosodic word in the Fuzhou dialect By contrast, CL applies within the domain formed by the reduplicated forms of disyllabic adjectives, as exemplified in (46). (46)

Application of CL in reduplicated forms of disyllabic adjectives I. AABB a. 四四角角 sei213 sei213 koyʔ23 koyʔ23 → si21 li21 ʔøyʔ44 koyʔ23 ‘(perfectly) square’ b. 舒舒畅畅 tshy44 tshy44 thuoŋ213 thuoŋ213 → tshy21 ʒy21 luoŋ51 nuoŋ213 ‘(very) comfortable’ c. 妥妥当当 tho31 tho31 touŋ213 touŋ213 → tho21 lo21 louŋ51 nouŋ213 ‘(very) appropriate’ d. 平平安安 piŋ51 piŋ51 ʔaŋ44 ʔaŋ44 → piŋ21 miŋ21 ŋaŋ44 ŋaŋ44 ‘(very) safe, peaceful’ II. AAB a. 四四角 sei213 sei213 koyʔ23 → si21 li21 ʔoyʔ44 ‘(perfectly) square’ b. 舒舒畅 tshy44 tshy44 thuoŋ213 → tshy21 ʒy21 luoŋ51 ‘(very) comfortable’ c. 光光鲜 kuoŋ44 kuoŋ44 tshiɛŋ44 → kuoŋ21 ŋuoŋ44 ʒiɛŋ44 ‘(very) neat and clean’ III. ABB a. 舒畅畅 tshy44 thuoŋ213 thuoŋ213 → tshy21 luoŋ51 nuoŋ213 ‘(very) comfortable’ b. 妥当当 tho31 touŋ213 touŋ213 → tho21 louŋ51 nouŋ213 ‘(very) appropriate’ c. 平安安 piŋ51 ʔaŋ44 ʔaŋ44 → piŋ21 ŋaŋ44 ŋaŋ44 ‘(very) safe, peaceful’

3.3.4.7 Summary We have seen from the above discussion that the CL rule in the Fuzhou dialect applies within the domain formed by polysyllabic monomorphemic words, derived words, and compounds, as well as reduplicated forms of disyllabic adjectives. It is blocked in the domain formed by “diminutive” nouns and reduplicated monosyllabic adjectives, although the domain formed by these morphosyntactic words can serve as an appropriate environment for its application. CL also fails to apply to monosyllabic monomorphemic words and sound-splitting words, since these morphosyntactic words cannot provide an applicable phonological context for CL. 3.3.5 Summary Based on the above discussion, the application of TS, MTS, FC, and CL within the domain formed by the major types of morphosyntactic words in the Fuzhou dialect can be summarized as in Table 3.2.

Prosodic word in the Fuzhou dialect  89 Table 3.2  Phonological rules and morphosyntactic words in the Fuzhou dialect Application

Rules

Morphosyntactic words Monomorphemic words

monosyllabic disyllabic trisyllabic Derived words prefix + stem stem + suffix(es) prefix + stem + suffix Compounds coordination modifier-head verb-object verb-complement Reduplication: “diminutive” nouns Reduplication: sound-splitting words Reduplication: AA reduplicated adjectives ABAB/AAB/ABB

TS

MTS

FC

CL

N/A

N/A

N/A

N/A

√

×

√

√

√

×

N/A √ N/A

√

√

×

√

√

× × √ √

√ √ × ×

√ √ (opt.) √ √

× N/A × √

Note: “√” denotes that the application of the rule is obligatory as long as there is an applicable phonological context; “N/A” indicates that there is no phonological environment for the rule to apply; “×” indicates that the rule is blocked although there is an appropriate environment; and “opt.” suggests that the rule is optional in a given domain.

3.4 Lexical-level phonological rule application and the prosodic word domain in the Fuzhou dialect From Table 3.2, we can find that it is the domain formed by Fuzhou morphosyntactic words that acts as the domain of application for TS, MTS, FC, and CL at the lexical level. Then, the follow-up question would be: which prosodic domain in the prosodic hierarchy does the domain formed by Fuzhou morphosyntactic words correspond to? Two important characteristics of the domain formed by Fuzhou morphosyntactic words have been identified from the discussion in Section 3.2 and Section 3.3. First, morphosyntactic words in the Fuzhou dialect can be composed of (a) a single stem (e.g., monomorphemic words); (b) the stem plus any adjacent affixes (e.g., derived words); or (c) all members of a compound (e.g., compounds), so the size of the domain they form corresponds to the terminal element of the syntactic tree. Second, the domain formed by Fuzhou morphosyntactic words can ideally contain an unlimited number of syllables, as long as the syllables it contains belong to the same morphosyntactic word, as can be evidenced by quadrisyllabic examples such as 共产主义 [køyŋ242 saŋ31 tsuo31 ŋiɛ242 → kyŋ21 naŋ21 ʒuo44 ŋiɛ242]

90  Prosodic word in the Fuzhou dialect ‘Communism’ and 四四角角 [sei213 sei213 koyʔ23 koyʔ23 → si21 li21 ʔøyʔ44 koyʔ23] ‘(perfectly) square’. Given these two characteristics, if we need to establish a prosodic domain in the Fuzhou dialect to correspond to the domain formed by morphosyntactic words, the only choice would be the prosodic word, since the prosodic word can be coextensive with the terminal node of the syntactic tree and can theoretically contain an unlimited number of syllables in a language without minimal word requirements. As we have seen in Section 3.1, the prosodic word is treated as an indispensable prosodic constituent in a number of languages since there are various phonological generalizations that make crucial reference to this constituent. Clearly, the fact that the application of TS, MTS, FC, and CL at the lexical level in Fuzhou is sensitive to the domain formed by morphosyntactic words can also be taken as supporting evidence for the establishment of the prosodic word domain. I can now proceed to give the definition of the prosodic word domain in the Fuzhou dialect, as presented in (47). (47) Prosodic word (ω) domain in the Fuzhou dialect The domain of ω in the Fuzhou dialect is the terminal node of the syntactic tree. The size of the prosodic word domain in the Fuzhou dialect, therefore, is equal to that of morphosyntactic words and includes a stem plus all adjacent affixes (if any), or both members of a compound, which is similar to the prosodic word in languages such as Greek and Latin. With the definition in (47), phonological rules that are sensitive to the prosodic word domain can be roughly formulated as follows. (48) ω-domain phonological rules in the Fuzhou dialect a. TS: Tn → Tn’/ [____ Tn + 1]ω (n ≥ 1) b. MTS: T1 → 21 or 31(44)/ [____ T2]ω (in “diminutive” nouns and sound-splitting words) c. FC: Group B → Group A/ [[C0 ____] σn [. . .] σn + 1] ω d. CL: C → C’/ [ [. . .] σn [____. . .] σn + 1] ω (n ≥ 1)

(n ≥ 1)

It is noteworthy that, although TS, MTS, FC, and CL all take the prosodic word as their domain of application at the lexical level, the degree of application of these rules is conditioned by the subtypes of morphosyntactic words that form the prosodic word domain. Specifically, as we have seen in Section 3.3, TS obligatorily applies to polysyllabic monomorphemic words, derived words, compounds, and reduplicated adjectives but does not apply to “diminutive” nouns and sound-splitting words; MTS obligatorily applies to “diminutive” nouns and sound-splitting words but does not apply to other morphosyntactic words; FC obligatorily applies to polysyllabic monomorphemic words, derived words, compounds, “diminutive” nouns, and reduplicated adjectives, while optionally applies to sound-splitting words; and CL obligatorily applies to polysyllabic monomorphemic words, derived words, compounds, and reduplicated disyllabic adjectives

Prosodic word in the Fuzhou dialect  91 but does not apply to “diminutive” nouns and reduplicated monosyllabic adjectives. Thus, we can see that not only is the domain of the prosodic word in Fuzhou constructed on the basis of morphological information, but the degree of application of these phonological rules within this domain is also conditioned by the types of morphological processes. The case of the Fuzhou prosodic word domain thus serves as a very good example for the interactions between the phonological and the morphological components of the grammar. Recall that some previous studies (e.g., Wright 1983; Chan 1985, among others) argue that the domain of application for TS and FC at the lexical level in the Fuzhou dialect is the foot, as reviewed in Chapter 2. I have argued in Chapter 2 that the foot domain is not a good choice for lexical-level phonological rule application in this dialect since phonological rules such as TS and FC at the lexical level can be treated as prosodic word-domain rules with no loss of generalization, and the application of these rules in morphosyntactic words containing more than three syllables can only be accounted for by referring to the prosodic word domain. The rule formulations in (48) as well as the discussion in this chapter have once again confirmed that assuming the prosodic word to be the domain of application for these rules at the lexical level (i.e., in morphosyntactic words) is more reasonable. It should also be noted that a monosyllabic monomorphemic word in Fuzhou is required to form an independent prosodic word domain according to the definition in (47) because of the fact that it occupies the terminal node of the syntactic tree. This does not undermine the prosodic word analysis for the application of TS, MTS, FC, and CL at the lexical level, since the non-application of these rules within the prosodic word domain formed by monosyllabic monomorphemic words is simply due to the lack of phonological environment. The assignment of the ω status to monosyllabic words not only complies with the definition of the prosodic word domain, but also plays an important role in the formation of prosodic domains larger than the prosodic word (e.g., the clitic group and the phonological phrase), as to be seen in the following chapters.

Notes 1 According to Nespor & Vogel (1986: 124–134), in Italian, only prefixes that end in a vowel form independent prosodic words, while those ending in a consonant are joined into a prosodic word with the following stem. 2 According to Nespor & Vogel (1986: 120) and Lees (1961), only the first primary stress remains after compounding, while the main stress of the second member is reduced to secondary stress, which is a rule sensitive to the phonological phrase domain. 3 It is generally recognized that in Mandarin Chinese there is another subtype of compounds, namely subject-predicate compounds, in addition to these four types. In a subject-predicate compound, the stem on the right serves as the predicate of the subject symbolized by the stem on the left, such as 心疼 [ɕin55 thəŋ35] ‘heart’ + ‘painful’ → ‘to love dearly; to grudge’ and 年轻 [niɛn35 tɕhiŋ44] ‘year’ + ‘light’ → ‘young’. This type of compound is not discussed in this chapter, since the examples of this subtype are rare in the Fuzhou dialect and almost all the subject-predicate compounds in Fuzhou are borrowed from Mandarin Chinese.

92  Prosodic word in the Fuzhou dialect 4 It should be noted that the “complement” here is different from the term “complement” used under X-bar framework of syntax. In syntactic theory, the term “complement” often refers to the sister node of the head, and hence in the case of verb phrases, the “complement” is actually the object of the head verb. The “complement” in the discussion here, by contrast, refers to complementary elements describing the result of the action indicated by the verb stem on the left. 5 In traditional Chinese lexicography, fanqie is a method used to indicate the pronunciation of a monosyllabic character by combining syllables of two other characters. The pronunciation of the target character is composed of the initial consonant of one syllable and the final of the other. 6 Notice that there is another example of 碎碎 [tshoy tshoy] in (20), which means ‘fragment’. These two 碎碎 should be considered as different entries in the lexicon. The one in (20) is derived by reduplicating the monosyllabic noun 碎 [tshoy] ‘fragment’, while the one in this example is derived by reduplicating the monosyllabic adjective 碎 [tshoy] ‘shattered’. 7 Notice that the sandhi form of tones of 碎碎 [tshoy51 tshoy213] ‘(very) shattered’ in (31Ib) is different from that of 碎碎 [tshoy21 tshoy213] ‘fragment’ in (28e). 8 Following Xu (2001), Yu & Yin (2014) distinguish two types of reduplication in the lexicon of Mandarin Chinese. The first type of reduplication, “reduplication 1”, refers to the formation of reduplicated forms such as 奶奶 [nai214 nai0] ‘grandmother’ and 宝宝 [pau214 pau0] ‘baby’ (’0’ denotes the neutral tone). In these words, the first syllable bears the third tone in Mandarin while the second syllable bears a neutral tone. By contrast, the second type of reduplication, “reduplication 2”, refers to the formation of reduplicated forms such as 走走 [tsou35 tsou0] ‘to walk a while’ and 跑跑 [phau35 phau0] ‘to run a while’ in which the first syllable bears the sandhi form of the third tone while the second syllable bears a neutral tone. You (2017, 2018c) follows Yu & Yin (2014) and also distinguishes two types of reduplication in the Fuzhou dialect, among which “reduplication 1” refers to the formation of “diminutive” nouns and sound-splitting words, while “reduplication 2” refers to the formation of reduplicated adjectives. 9 Due to the small number of trisyllabic monomorphemic words in the Fuzhou dialect, no examples of the application of FC in trisyllabic monomorphemic words are provided here. 10 FC applies only if the non-terminal syllable in a certain domain has a Group B variant of alternating finals. Thus, no examples can be found for the application of FC in derived words containing the prefix, since neither 依- [ʔi44] nor 老- [lo31]/[lau242] has a Group B variant. 11 Suffixes -母 [mo31] and -妈 [ma31] never undergo CL since they both begin with [m].

4 Enclitics and the Type A clitic group in the Fuzhou dialect

The prosodic constituent located immediately above the prosodic word in the prosodic hierarchy is referred to as the clitic group, whose existence is advanced based on the observation that there are phonological rules and phonetic processes making crucial reference only to the group composed of a lexical word host and the adjacent clitic (see Hayes 1984/1989; Nespor & Vogel 1986; Cohn 1989, among others). As mentioned in Chapter 1, Fuzhou has two types of clitics, namely enclitics and proclitics, and there are correspondingly two types of clitic groups in this dialect. This chapter deals with the morphosyntactic functions and phonological behavior of enclitics and the clitic group formed by “host + enclitic” (henceforth the Type A clitic group) in the Fuzhou dialect. The properties of Fuzhou proclitics and the clitic group formed by “proclitic + host” (henceforth the Type B clitic group) will be explored in Chapter  6, after the discussion on Fuzhou phonological phrase. This chapter is organized as follows. Section 4.1 contains a basic introduction to the definition and common properties of clitics. It also briefly reviews the definition of the clitic group in prosodic phonology and the evidence for the clitic group as the domain of rule application in various languages. Section 4.2 investigates enclitics in the Fuzhou dialect, with the focus on their morphosyntactic functions. Section 4.3 examines the phonological properties of the Type A clitic group with respect to the application of Fuzhou phonological rules including TS, FC, and CL. Based on the discussion in Section 4.2 and Section 4.3, Section 4.4 discusses the violation of the Strict Layer Hypothesis (SLH) in the case of the Type A clitic group, arguing that a weakened SLH is required in the prosodic phonology of the Fuzhou dialect. Section 4.5 closes this chapter with a brief summary.

4.1 Introduction 4.1.1  Definition of clitics and common properties of clitics Many languages contain a specific type of elements, often referred to as clitics, which come from the Greek verb κλίνω ‘to lean’. The definitions of the term “clitic” given in the literature are presented in (1) and (2) (also cf. Zwicky 1977; Klavans 1982; Matthews 1991; Halpern 1998; Riemsdijk 1999, among others).

94  Fuzhou enclitics and the Type A clitic group (1)

Definition of clitics (Crystal 2008) A term used in grammar to refer to a form that resembles a word but that cannot stand on its own as a normal utterance, being phonologically dependent upon a neighboring word (its host) in a construction.

(2)

Definition of clitics (Russi 2008) Clitics refer to intermediate linguistic units, which grammatically behave like words in that they combine with other words or phrases to make phrases but are phonologically bound to an adjacent word traditionally referred to as the “host”.

According to the above definitions, clitics are phonologically dependent elements that can never occur alone and have to “lean on” the adjacent host. Depending on their position in relation to the host they attach to, clitics are mainly divided into proclitics and enclitics.1 A clitic preceding its host is called a proclitic, e.g., the French clitic pronoun le in Jean le vois ‘Jean sees it’, which takes the verb vois ‘see’ as its host. In contrast, a clitic appearing after its host is called an enclitic, e.g., English reduced auxiliaries ’s in He’s tall and ’ll in I’ll have to go now, both of which attach to their hosts on the left. The Fuzhou clitics examined in this chapter all attach to the right of their hosts and are thus enclitics, while those examined in Chapter 6 are all proclitics. It has long been noticed that clitics “exhibit some of the properties of the word and some of the properties of the affix” (Klavans 1982). As early as in Sapir (1930: 70), it has been pointed out that “enclisis is neither true suffixation nor juxtaposition of independent elements. It has the external characteristics of the former, the inner feeling of the latter”. This mixed behavior and the unclear linguistic status of clitics have posed problems for linguists. Starting with Zwicky’s (1977) pioneering study, a vast amount of research has been devoted to identifying the morphosyntactic and phonological properties of clitics across languages. Some linguists have argued that clitics should be incorporated into one of the two uncontroversial categories, namely words (Crysmann 1997, 2000, among others) or affixes (Miller 1992; Monachesi 1999; Cocchi 2000, among others). However, it has also been argued by many linguists that clitics represent an independent category due to their special properties and thus should be distinguished from both words and affixes (Hayes 1984/1989; Nespor & Vogel 1986; Haspelmath & Sims 2010, among others). On the one hand, clitics should be differed from independent words in several ways. First of all, clitics exhibit a type of phonological dependency, while independent words are free in terms of their occurrence. Due to the phonological dependency of clitics, it is impossible to (a) pause between a clitic and its host, (b) assign stress to clitics in stress languages, (c) assign contrastive stress to clitics, and (d) freely move clitics in an utterance (Haspelmath & Sims 2010). The second property of clitics, which distinguishes them from independent words, is that clitics commonly belong to some functional and considerably closed categories, such as auxiliaries, pronouns, determiners, prepositions, postpositions, conjunctions, and

Fuzhou enclitics and the Type A clitic group  95 functional particles like negatives and interrogative particles (Zwicky 1977; Klavans 1982, 1985, among others). By contrast, independent words typically come from open categories such as nouns, verbs (excluding auxiliaries), and adjectives. On the other hand, several criteria have been proposed to distinguish clitics and affixes (Zwicky & Pullum 1983; Haspelmath & Sims 2010, among others), as presented in (3). (3)

Clitics vs. affixes a. Host selection: clitics can attach to words of virtually any category, e.g., English auxiliary ’s, which can attach to nouns, prepositions, verbs, adjectives, and adverbs, while affixes are quite specific in their selections of stems, e.g., English plural suffix -s, which attaches only to noun stems. b. Arbitrary gaps: clitics do not exhibit arbitrary gaps while affixes do; e.g., English plural suffix -s does not attach to “child”. c. Morpho-phonological idiosyncrasies: clitics do not exhibit morphophonological idiosyncrasies while irregular forms are quite common in groupings of stems and affixes, e.g., forms like feet for the plural affix, went for the past affix, and best for the superlative affix. d. Semantic idiosyncrasies: the meaning of the string of the host plus the clitic(s) is predictable from the meaning of the host and that of the clitic(s), while affix-stem combinations may have an idiosyncratic meaning; e.g., the Dutch-inflected form ouder is a comparative (‘older’) in its literal meaning but has also been extended to use as a noun, which means ‘parent’. e. Syntactic operations: an affixed word is regularly treated as one unit by syntactic operations, while a string of the host plus the clitic(s) is usually treated as separated units by syntactic operations. f. Ordering: clitics can attach to material already containing clitics or affixes, but affixes cannot attach to a host containing clitics.

Furthermore, as argued by a number of linguists (e.g., Hayes 1984/1989; Nespor & Vogel 1986, among others), the phonological behavior of clitics is often different from that of both independent words and affixes. Specifically, in a given language, some phonological phenomena apply only in relation to a constituent consisting of a host plus the clitic, namely the clitic group. Hence, the role played by the clitic group as the domain of application for various phonological generalizations can serve as another important criterion to distinguish clitics from both independent words and affixes (to be discussed in more detail in Section 4.1.2). 4.1.2  Definition of the clitic group and evidence for the clitic group domain across languages It has been noticed that certain phonological rules only apply in the domain consisting of a word host and the clitic(s) but not in any other context in languages, which is why such a domain, namely the clitic group, is treated as a unique prosodic constituent in

96  Fuzhou enclitics and the Type A clitic group the prosodic hierarchy. We have seen the definition of the clitic group domain given by Nespor & Vogel (1986: 154) in Chapter 1, which is reproduced here in (4). (4)

Clitic group (CG) formation The domain of the CG consists of a ω containing an independent (i.e., nonclitic) word plus any adjacent phonological words containing a. a directional clitic, or b. a plain clitic/nondirectional clitic such that there is no possible host with which it shares more category memberships.

The clitic group, according to Nespor & Vogel (1986), is the first level of the prosodic hierarchy that represents the mapping between the syntactic and the phonological components since certain clitics choose their direction of attachment according to the syntactic constituent structure. As Nespor & Vogel (1986) suggest, the clitic group formed by the mapping from the syntactic structure is nevertheless not necessarily isomorphic to a syntactic constituent. Like other prosodic domains, the clitic group has been reported to form the domain of application for many phonological phenomena cross-linguistically, which constitutes the most substantial evidence for the existence of this constituent. In addition to the often-cited examples of stress assignment in Latin, which have been discussed in Chapter 1, there are many other phonological rules that apply within clitic groups but not across their boundaries or in other prosodic domains. For example, it has been reported that in English there is a v-Deletion rule, which deletes a word-final [v] before a [−syllabic] segment within a certain domain (cf. Selkirk 1972; Hayes 1984/1989; Nespor & Vogel 1986, among others). According to Hayes (1984/1989) and Nespor & Vogel (1986), this rule applies in fast speech within the domain of the clitic group, as exemplified in (5), but fails across the boundary of clitic groups, as exemplified in (6). Clitics in question are marked in bold. (5)

Application of v-Deletion in English (adapted from Hayes 1984/1989) a. [Please]CG [leave them]CG [alone]CG2 ↓ [Ø] b.

(6)

[Will you save me]CG [a seat]CG ↓ [Ø]

Blocking of v-Deletion in English (adapted from Hayes 1984/1989) a. [Give]CG [Maureen]CG [some]CG ↓ *[Ø] b.

[We’ll save]CG [those people]CG [a seat]CG ↓ *[Ø]

Fuzhou enclitics and the Type A clitic group  97 Another typical example that has been recognized as evidence for the clitic group domain is Vowel Harmony in Turkish. According to Nespor  & Vogel (1986), in Turkish, the harmonizing features are [back] for all vowels and [round] for high vowels. Within the clitic group domain, this rule applies from left to right, and thus the first vowel within the domain determines the value for [back] and [round] of all the following vowels, as illustrated in (7), in which -mu is an interrogative clitic. (7)

Application of Vowel Harmony in Turkish (adapted from Nespor & Vogel 1986) ‘true’ b. [bügün]ω ‘today’ a. [doğrú]ω [doğrú mu]CG ‘true?’ [bügün mü]CG ‘today?’

Other phonological rules referring crucially to the clitic group as their domain of application include s, z-Palatalization in English, Stress Readjustment, Nasal Deletion, Nasal Assimilation, and Stop Voicing in Greek, Stress Assignment in Turkish, Penultimate Schwa Specification and Post-lexical Stress Assignment in French, t-Deletion in Catalan, and so forth (see Hayes 1984/1989; Nespor & Vogel 1986; Hannahs 1995a; Kabak & Vogel 2001, among others). As discussed in Chapter 1, arguments against the existence of the clitic group as a prosodic domain have been advanced (e.g., Zec 1988, 1993; Inkelas 1989/1990; Inkelas  & Zec 1995; Booij 1996; Kleinhenz 1996; Selkirk 1996; Peperkamp 1997, among others). Despite the controversy surrounding the clitic group, some researchers still embrace a position that favors the existence of the clitic group as an indispensable prosodic constituent (e.g., Hayes 1984/1989; Zhang 1992, 2014, 2017; Hannahs 1995a; Kabak & Vogel 2001, among others). In responding to the objections and problems with the original clitic group, Vogel (2009) argues that the problems are not due to the clitic group itself but to restrictions posed by the original SLH, which can be resolved by assuming a slight weakening of this hypothesis. In the following sections and Chapter 6, I will show that Fuzhou clitics may attach to constituents higher than the prosodic word, including the clitic group and the phonological phrase. Moreover, enclitics and proclitics in this dialect exhibit asymmetries in terms of their phonological behavior, and so do the Type A  clitic group and the Type B clitic group. These would be viewed as counterevidence against the existence of the clitic group according to some previous studies mentioned in the last paragraph. Nonetheless, following previous studies that argue for the clitic group, I assume in this study that the clitic group is still part of the prosodic hierarchy since the clitic group in Fuzhou as a whole can be distinguished from other prosodic constituents. Moreover, I  adopt the weakened SLH and hold that the constraints of Nonrecursivity and Layeredness are both violable in the case of the clitic group in the Fuzhou dialect, which is well supported by the evidence from this dialect. Thus, the definition of the clitic group domain in the Fuzhou dialect can be tentatively formulated by making slight revision on Nespor & Vogel’s definition in (4), as presented in (8), in which the

98  Fuzhou enclitics and the Type A clitic group over-assignment of the prosodic word status to clitics is also avoided. I will show in the following sections and in Chapter 6 that such a definition is supported by language facts in the Fuzhou dialect. (8)

Clitic group (CG) formation in the Fuzhou dialect The domain of the CG in the Fuzhou dialect consists of one independent (i.e., non-clitic) prosodic constituent (ω, CG, or φ), plus any adjacent a. directional clitic, or b. plain clitic/nondirectional clitic such that there is no possible host with which they share more category memberships.

4.2 Enclitics in the Fuzhou dialect and their morphosyntactic functions It has long been noticed that there are a number of clitic-like elements that have very distinctive morphosyntactic and phonological properties in the Fuzhou dialect (cf. Chen  & Norman 1965a; Wright 1983; Chan 1985; Chen 1998; Li 2002, among others). However, a closer look to the relevant literature can reveal a significant absence of systematic studies on clitics in this dialect— most of the relevant literature simply treats these elements as either affixes or words. The present section and the following sections in this chapter as well as Chapter 6 thus address this gap by offering a comprehensive description and analysis of the morphosyntactic functions and the phonological behavior of these elements. It will be shown that these elements share common properties of clitics across languages and hence should be regarded as clitics. Like clitics in other languages, clitics in Fuzhou are mostly bound morphemes, which cannot be used on their own and have to attach to adjacent prosodic units as the host. According to their position in relation to the host they attach to, Fuzhou clitics can be divided into enclitics and proclitics. We will see that most clitics in the Fuzhou dialect belong to closed functional categories, including possessive/modificational/nominalization marker, adjective reduplication markers, aspect markers, auxiliaries, and prepositions, as well as other functional particles, together forming a considerably heterogeneous category. In this section and the following sections in this chapter, I focus on enclitics in the Fuzhou dialect, while the discussion of Fuzhou proclitics will be left to Chapter 6. The following subsections in Section 4.2 discuss the morphosyntactic functions of Fuzhou enclitics with a number of examples. In the examples, the enclitic is labeled with “C”, the group of “host + enclitic” with “CG”, and the prosodic word host with “ω”. In some examples, we will see that the clitic attaches to a phrasal-level unit that is composed of more than one prosodic word. As will be demonstrated in Chapter 5, these phrasal-level units are actually phonological phrases in the Fuzhou dialect, which are labeled with “φ” in the examples. For the sake of brevity, examples in this section present only the citation/underlying segmental structure.

Fuzhou enclitics and the Type A clitic group  99 4.2.1  Possessive/modificational/nominalization marker 其 [ki0] Let us begin with the most commonly used enclitic in the Fuzhou dialect, the possessive/modificational/nominalization marker 其 [ki0], which always bears a neutral tone. 其 is transcribed as [ki51]/[ki53] in some published materials (e.g., Li et al. 1994; Li 2002). This 其 [ki51]/[ki53] is pronounced with a full tone and is only used in words like 蒋其 [tsuoŋ31 ki51] ‘how, why’ and 其利 [ki51 lei213] ‘to shrilly cry’, which is actually different from the clitic form 其 [ki0]. The clitic 其 [ki0] is sometimes presented as [i0] in the literature (e.g., Chen & Norman 1965a; Chan 1985). However, within a sandhi environment, when preceded by a syllable ending with the historical *[-k] coda, the clitic 其 is pronounced with a stop initial [k-]. According to the CL rule, citation initials following the *[-k] coda remain unchanged in a sandhi environment. If the citation form of the clitic 其 has a zero initial, it should not have the stop initial [k-] when following the *[-k] coda in a sandhi context. Therefore, the initial of the citation form of 其 must be the stop consonant [k-] instead of the zero initial. Thus, 其 is transcribed as [ki0] in this book. Like its counterpart de 的 in Mandarin Chinese, 其 [ki0] in the Fuzhou dialect can be: (a) attached to the right of a noun/pronoun to indicate possession, as in (9); (b) attached to the modifier and thus connecting the modifier and the nominal expression modified by the modifier, as in (10); and (c) used to make nouns out of verbs/verb phrases, adjectives (including reduplicated adjectives), nouns/noun phrases, or pronouns, as in (11) (cf. Chen & Norman 1965a; Li 2002, among others). 其 [ki0] is presented as POSS (= possessive marker), MOD (= modificational marker), or NOM (= nominalization marker) in the gloss. (9)

Possessive marker a. [[我]ω 其C]CG [[ŋuai] ki] I POSS ‘my book’

书3 tsy book

(10) Modificational marker a. [[旧]ω 其C]CG 书 [[kou] ki] tsy old MOD book ‘old book’ (11) Nominalization marker a. [[食]ω 其C]CG [[siɛʔ] ki] eat NOM ‘food’ c.

[[红红]ω 其C]CG [[ʔøyŋ ʔøyŋ] ki] (very) red NOM ‘(very) red thing(s)’

b.

[[依妹]ω 其C]CG 衣裳4 [[ʔi mui] ki] ʔi suoŋ younger sister POSS clothes ‘younger sister’s clothes’

b.

[[旧旧]ω 其C]CG [[kou kou] ki] (very) old MOD ‘(very) old book’

书 tsy book

b.

其C]CG [[红]ω [[ʔøyŋ] ki] red NOM ‘red thing(s)’

d.

[[我]ω [[ŋuai] I ‘mine’

其C]CG ki] NOM

100  Fuzhou enclitics and the Type A clitic group 4.2.2  Adjective reduplication markers 势 [siɛ213], 式 [seiʔ23], and 喏 [luoʔ23] Chapter 3 has mentioned that reduplicated adjectives in the Fuzhou dialect generally cannot be used as the predicate on their own. When used as the predicate, reduplicated adjectives are bound on the right side and thus need to take enclitics 势 [siɛ213], 式 [seiʔ23], or 喏 [luoʔ23]5 (cf. Chen 1998; Li & Liang 2001; Li 2002, among others). In the Fuzhou dialect, 势 [siɛ213] can be used as a noun that means ‘situation’ or ‘force’, and 式 [seiʔ23] can also be used as a noun that means ‘style’. 喏 [luoʔ23] is only used as an enclitic. As enclitics, 势 [siɛ213], 式 [seiʔ23], and 喏 [luoʔ23] are freely interchangeable when attached to reduplicated adjectives. Examples of 势 [siɛ213], 式 [seiʔ23], and 喏 [luoʔ23] are given in (12), where these three enclitics are presented as AdjR (= adjective reduplication marker) in the gloss. (12) a.

伊 [[悬悬]ω 势/式/喏C]CG ʔi [[keiŋ keiŋ] siɛ/seiʔ/luoʔ] he (very) tall AdjR ‘He is rather tall.’ b. 面 [[白白]ω 势/式/喏C]CG meiŋ [[paʔ paʔ] siɛ/seiʔ/luoʔ] face (very) white AdjR ‘The face is rather white.’ c. 工作 [[闲闲落落]ω 势/式/喏C]CG køyŋ tsouʔ [[ʔeiŋ ʔeiŋ loʔ loʔ] siɛ/seiʔ/luoʔ] job (very) easy/relaxing AdjR ‘This job is very easy.’

4.2.3  Aspect markers It has long been recognized that the Fuzhou dialect has a number of aspect markers that occur after the verb or verb phrase (cf. Chen  & Norman 1965a; Chan 1985; Chen 1998; Li & Liang 2001; Li 2002, among others). These post-verbal aspect markers in the Fuzhou dialect are enclitics, which attach to the host on their left to indicate the developmental status of the event or situation. 4.2.3.1  Durative aspect marker 𠲥 [lɛ0] The first aspect marker I would like to examine is 𠲥 [lɛ0], which is a versatile enclitic in the Fuzhou dialect. 𠲥 [lɛ0] can not only serve as the durative aspect marker, but also be used as the perfective aspect marker, the post-verbal particle, and the locative marker (to be discussed in more detail in relevant following subsections). As the durative aspect marker, 𠲥 [lɛ0] behaves like the durative aspect marker zhe 着 in Mandarin Chinese, occurring in the post-verbal position to indicate a continuing state or situation denoted by the verb or verb phrase. Verbs preceding 𠲥 [lɛ0] are usually those denoting states or actions that can last for a certain amount of time, as exemplified in (13).

Fuzhou enclitics and the Type A clitic group  101 (13) a.

c.

b. 门 [[关]ω 𠲥C]CG mouŋ [[kuoŋ] lɛ] door close DUR ‘The door is closed.’ 车 里 [[坐]ω 𠲥C]CG 两 tiɛ [[soy] lɛ] laŋ tshia car inside sit DUR two ‘There are two people sitting in the car.’

伊 [[徛]ω 𠲥C]CG ʔi [[khiɛ] lɛ] he stand DUR ‘He is standing.’ 隻 侬 tsiɛʔ nøyŋ Cl people

In addition, similar to zhe 着 in Mandarin Chinese, the durative aspect marker 𠲥 [lɛ0] can occur between two verbs. In the “V1 𠲥 V2” construction in Fuzhou, 𠲥 [lɛ0] attaches to the preceding verb (V1) and indicates that the event denoted by the following verb (V2) happens in the state of “V1-ing”, as illustrated in (14). (14) a.

c.

[[坐]ω 𠲥C]CG 讲 [[soy] lɛ] kouŋ sit DUR speak ‘to speak sitting down’ 目珠 [[擘]ω 𠲥C]CG møyʔ tsiu [[paʔ] lɛ] eye open DUR ‘to sleep with eyes open’

b.

[[徛]ω 𠲥C]CG 等 tiŋ [[khiɛ] lɛ] stand DUR wait ‘to wait standing’

睏 khouŋ sleep

The durative aspect marker 𠲥 [lɛ0] can also be used in an imperative sentence, as in (15). (15) a.

[[记]ω 𠲥C]CG 通知 [[kei] lɛ] thuŋ ti remember DUR inform ‘Remember to inform him.’

伊 ʔi he

b.

[[徛]ω 𠲥C]CG 汝 ny [[khiɛ] lɛ] you stand DUR ‘You, stand there!’

4.2.3.2  Experiential aspect markers 过 [kuo213] and 着 [tuoʔ5] Enclitics 过 [kuo213] and 着 [tuoʔ5] in Fuzhou are both experiential aspect markers, which occur following the verb on their left. In the Fuzhou dialect, 过 [kuo213] means ‘to pass, to go through’ as a verb and means ‘excessively’ as an adverb. 着 [tuoʔ5] is more versatile and can be used as a verb meaning ‘to be in’, as an auxiliary verb meaning ‘have to’, and as a preposition meaning ‘in, at’. In contrast, as enclitics, the morphosyntactic function of 过 [kuo213] and 着 [tuoʔ5] is similar to that of the Mandarin experiential aspect marker guo 过, which indicates the past experience of the event or action denoted by the preceding verb. According to Li & Liang (2001), experiential aspect markers 过 [kuo213] and 着 [tuoʔ5] are interchangeable in the Fuzhou dialect, and the only difference between them is that 着 [tuoʔ5] is more often used by the older generations while 过 [kuo213] is more often used by the younger generations. Examples of 过 [kuo213] and 着 [tuoʔ5] are presented in (16).

102  Fuzhou enclitics and the Type A clitic group (16) a.

[[去]ω 过/着C]CG 天津 [[kho] kuo/tuoʔ] thiɛŋ kiŋ go EXP Tianjin ‘to have been to Tianjin before’

b.

[[食]ω 过/着C]CG 鳗鱼 [[siɛʔ] kuo/tuoʔ] muaŋ ŋy eat EXP eel ‘to have eaten eels before’

4.2.3.3  Perfective aspect marker 𠲥 [lɛ0] There are two perfective aspect markers in the Fuzhou dialect, 𠲥 [lɛ0] and 去 [kho0] (cf. Chen 1998; Feng 1998, among others). As the perfective aspect marker, 𠲥 [lɛ0] is recorded as 嘞 [lɛ0] in Chen (1998) and Lin (2002) while as 𠲥 [lɛ0] in most literature (cf. Zheng 1988b; Li et al. 1994; Feng 1998; Li 1998, among others). Here, I  follow most literature and mark this perfective aspect marker as 𠲥 [lɛ0]. Similar to the perfective aspect marker le 了 in Mandarin Chinese, both 𠲥 [lɛ0] and 去 [kho0] attach to the preceding verb or verb phrase and indicate the completion of actions. Their morphosyntactic distributions, nevertheless, are not the same. Generally speaking, 𠲥 [lɛ0] occurs after the verb and is followed by other elements such as the object or the complement. In contrast, 去 [kho0] appears after a bare verb or a verb-complement structure (cf. Chen 1998; see Section 4.2.3.4 for more details). Examples of the perfective aspect marker 𠲥 [lɛ0] in the Fuzhou dialect are presented in (17) and (18). (17) V1 𠲥 (O1) V2 (O2) a. 伊 [[食]ω 𠲥C]CG 暝 马上 去 睏 ʔi [[siɛʔ] lɛ] maŋ ma suoŋ kho khouŋ he eat PERF dinner immediately go sleep ‘He went to sleep immediately after eating the dinner.’ b. 我 [[买]ω 𠲥C]CG 两 瓶 酒 送 丈侬 ŋuai [[mɛ] lɛ] laŋ piŋ tsiu soyŋ tuoŋ nøyŋ I buy PERF two Cl wine give father-in-law ‘I bought two bottles of wine to give my father-in-law.’ c. 汝 [[看]ω 𠲥C]CG 固 想 看 kou suoŋ khaŋ ny [[khaŋ] lɛ] you watch PERF still think watch ‘You will want to watch it again after you watch it.’ (18) V + 𠲥+ frequency phrase/duration phrase a. 故事 [[讲]ω 十 𠲥C]CG 几 kou søy [[kouŋ] lɛ] kui seiʔ story speak PERF several ten ‘The story was told dozens of times.’ b. [[睏]ω 半 日 𠲥C]CG 大 tuai puaŋ niʔ [[khouŋ] lɛ] sleep PERF big half day ‘slept for most of the day’

回 xui times

4.2.3.4  Perfective aspect marker 去 [kho0] The other perfective aspect marker 去 [kho0] in the Fuzhou dialect is recorded as 咯 in Chen (1998), but as 去 in most literature (e.g., Feng 1998; Li & Liang 2001;

Fuzhou enclitics and the Type A clitic group  103 Li 2002, among others). I adopt the character used in most literature in this study. When 去 in the Fuzhou dialect is produced with a yinqu tone as [kho213], it means ‘to go’. The enclitic 去 [kho0], which bears a neutral tone, occurs right after a bare verb or a verb-complement structure, indicating the completion of the action. If the verb originally takes an object and the object has to be mentioned in the sentence, the object should be advanced to the topic position (Chen 1998). According to Li (2002), this perfective aspect marker usually indicates an unfavorable result of the action denoted by the verb or the verb-complement structure, as exemplified in (19). (19) a.

c.

伊 [[病]ω 去C]CG ʔi [[paŋ] kho] he sick PERF ‘He is sick.’ 药 食 [[绽]ω yoʔ siɛʔ [[taŋ] medicine eat wrong ‘took the wrong medicine’

b.

去C]CG kho] PERF

水缸 [[必]ω 去C]CG tsui kouŋ [[peiʔ] kho] water jar crack PERF ‘The water jar is cracked.’

4.2.3.5  Sentence final particle 了 [lau31] The aspect marker 了 [lau31] in the Fuzhou dialect is a sentence final particle that occurs at the end of a sentence or a clause and indicates a change in the state or situation. Thus, it by and large corresponds to the Mandarin sentence final particle le 了, which is considered as a perfect aspect marker indicating a change of state or a currently relevant state (CRS) (Li & Thompson 1981; Sun 2006). 了 [lau31] can be used as the only aspect marker in a sentence/clause, as in (20). It can also coexist with other aspects markers discussed above, as in (21) (Chen 1998). It is noteworthy that the enclitic 了 [lau31] in (20b) attaches to a unit that is larger than the prosodic word and that the violation of Nonrecursivity is allowed in the examples in (21). (20) a.

暝 [[好]ω 了C]CG maŋ [[xo] lau] dinner good CRS ‘The dinner is ready.’

(21) a.

门 [[[开]ω 𠲥C]CG 了C]CG mouŋ [[[khui] lɛ] lau] door open DUR CRS ‘The door is already open.’ 只 本 书 我 tsi puoŋ tsy ŋuai this Cl book I ‘I have read this book.’ 天 [[[暗]ω 去C]CG 了C]CG thiɛŋ [[[ʔaŋ] kho] lau] sky dark PERF CRS ‘The sky has become dark.’

b.

c.

b.

[[[食]ω [饭]ω]φ [[[siɛʔ] [puoŋ]] eat rice ‘It’s time to eat.’

[[[看]ω [[[khaŋ] read

过C]CG kuo] EXP

了C]CG lau] CRS

了C]CG lau] CRS

104  Fuzhou enclitics and the Type A clitic group 4.2.3.6  Delimitative aspect marker 囇 [la242] The enclitic 囇 [la242] in Fuzhou is used as the delimitative aspect marker, which occurs on the right of the verb and indicates that a situation or event lasts only a short time (Chen 1998; Li  & Liang 2001; Li 2002, among others). It is suggested in the literature that 囇 [la242] is derived from the fusion of 蜀 [suoʔ5] ‘one’ and 下 [ʔa242] ‘a classifier for instances/occasions’ (Chen 1998; Li & Liang 2001; Li 2002, among others). Taking into consideration the pronunciation of 囇 [la242], nonetheless, I assume that it would be more reasonable to consider 囇 [la242] as the fusion of the sandhi form of 蜀 [suoʔ5] ‘one’ and 下 [ʔa242], since the initial [s] of 蜀 is usually changed into [l] in a sandhi context following the CL rule (unless it is preceded by the codas [-ŋ] and [-k]). Otherwise, it would be difficult to account for the origin of the initial [l] of 囇 [la242]. Examples of 囇 [la242] are given in (22). (22) a.

c.

[[听]ω 囇C]CG thiaŋ la listen DLM ‘to listen awhile’ [[歇]ω 囇C]CG 介 试 [[xyoʔ] la] kai tshei rest DLM again try ‘Take a rest and then try it again.’

b.

汝 [[坐]ω 囇C]CG ny [[soy] la] you sit DLM ‘You sit awhile.’

4.2.4  Post-verbal particles In addition to post-verbal aspect markers discussed in Section  4.2.3, there are other three enclitics that occur right after the verb in the Fuzhou dialect: 敆 [kaʔ0], 遘 [kau213], and 𠲥 [lɛ0]. These three enclitics do not indicate the developmental status of the event. In order to distinguish them from the post-verbal aspect markers, they are named post-verbal particles (PVP) in this study. 4.2.4.1  Post-verbal particle 敆 [kaʔ0] Although they occur right after the verb, 敆 [kaʔ0] and 遘 [kau213] are different from typical post-verbal resultative complements in Fuzhou such as 完 [ʔuoŋ51] ‘finish’ in 试完 [tshei213 ʔuoŋ51] ‘to finish trying’ and 饱 [pa31] ‘full’ in 食饱 [siɛʔ5 pa31] ‘to eat one’s fill’. Resultative complements like 完 [ʔuoŋ51] and 饱 [pa31] never undergo the CL rule when occurring right after the verb (see Chapter 5 for more details). By contrast, when enclitics 敆 [kaʔ0] and 遘 [kau213] attach to the verb, they usually undergo CL and have their sandhi initials (to be discussed in more detail in Section 4.3). Hence, enclitics 敆 [kaʔ0] and 遘 [kau213] should be distinguished from typical resultative complements in the Fuzhou dialect. If 敆 takes a high entering tone, namely [kaʔ5], it is used as a preposition that means ‘in, at’ in the Fuzhou dialect and introduces time and location before the verb or verb phrase. In contrast, the enclitic 敆 [kaʔ0] has a neutral tone. It also introduces time and location of the action or event but only occurs after the verb or verb

Fuzhou enclitics and the Type A clitic group  105 phrase. The morphosyntactic functions of 敆 [kaʔ0] and 敆 [kaʔ5] are thus similar to those of zai 在 in Mandarin Chinese (cf. Chen 1998; Li & Liang 2001, among others). Examples of the post-verbal particle 敆 [kaʔ0] are presented as in (23). (23) a. [[定]ω 敆C]CG 今旦 b. [[tiaŋ] kaʔ] kiŋ taŋ set PVP today ‘to be scheduled today’ c. 眠床 [[排]ω 敆C]CG 厅中 miŋ tshouŋ [[pɛ] kaʔ] thiaŋ touŋ bed put PVP drawing room ‘The bed is put in the drawing room.’ d. 通知 [[贴]ω 敆C]CG 外斗 thuŋ ti [[thaiʔ] kaʔ] ŋiɛ tau notice paste PVP outside ‘The notice is posted outside.’

[[安排]ω 敆C]CG 明旦 [[ʔaŋ pɛ] kaʔ] miŋ taŋ arrange PVP tomorrow ‘to be scheduled tomorrow’

4.2.4.2  Post-verbal particle 遘 [kau213] 遘 [kau213] in the Fuzhou dialect corresponds to the Mandarin dao 到 (cf. Chen 1998, among others). It can be used as a verb, which means ‘to arrive’. When used as a post-verbal particle, in contrast, 遘 [kau213] has multiple functions. “V-遘” can be followed by object nouns/noun phrases, place words, time words, and even sentences or clauses indicating the result or the degree, as exemplified in (24). (24) a.

c.

d.

b. [[行]ω [[收]ω 遘C]CG 批 遘C]CG 厝 [[siu] kau] phiɛ [[kiaŋ] kau] tshuo receive PVP letter walk PVP home ‘to receive a letter’ ‘to arrive home by walking’ [[等]ω 遘C]CG 三 点 [[tiŋ] kau] saŋ teiŋ wait PVP three o’clock ‘to wait until three o’clock’ [[做]ω 遘C]CG 逢侬 都 满意 [[tso] kau] xuŋ nøyŋ tu muaŋ ʔei do PVP everyone all satisfied ‘to do (something) and make everyone satisfied’

4.2.4.3  Post-verbal particle 𠲥 [lɛ0] Unlike the durative aspect marker 𠲥 [lɛ0] and the perfective aspect marker 𠲥 [lɛ0], the post-verbal particle 𠲥 [lɛ0] does not signify the aspect. Instead, similar to the Mandarin descriptive complement marker de 得, it is used to connect the verb and the descriptive complement that indicates the result or manner of the action. The complement following the post-verbal particle 𠲥 [lɛ0] could be the reduplicated form of adjectives, a phrase, or a sentence/clause, as exemplified in (25).

106  Fuzhou enclitics and the Type A clitic group (25) a.

b.

c.

衣裳 [[颂]ω 式 𠲥C]CG 破破 seiʔ ʔi suoŋ [[søyŋ] lɛ] phuai phuai clothes wear PVP (very) ragged AdjR ‘The clothes are worn out.’ [[看]ω 𠲥C]CG 野 清楚 [[khaŋ] lɛ] ʔia tshiŋ tshu look PVP very clear ‘saw (something) very clearly’ 伊 [[跳]ω 𠲥C]CG 蜀 身 都 是 汗 suoʔ siŋ tu sei kaŋ ʔi [[thiu] lɛ] he jump PVP one body all be sweat ‘He jumped and was covered in sweat.’

4.2.5  Locative marker 𠲥 [lɛ0] The enclitic 𠲥 [lɛ0] can also serve as the locative marker, changing a regular noun into a place word. Unlike localizers in the Fuzhou dialect such as 里 [tiɛ31], 边 [piɛŋ44], and 斗 [tau213], which always cause the preceding syllable to undergo TS, the locative marker 𠲥 [lɛ0] never triggers the application of the TS rule on the preceding syllable (to be discussed in more detail in Section 4.3). Examples of the locative marker 𠲥 [lɛ0] are given in (26), in which 𠲥 [lɛ0] may have slightly different meanings in different examples. It is noteworthy that the host of the enclitic 𠲥 [lɛ0] in (26d) is a phrase, which contains more than one prosodic word. (26) a.

c.

[[面]ω [[meiŋ] face ‘on the face’ [[碗囝]ω [[ʔuaŋ kiaŋ] small bowl ‘in the bowl’

𠲥C]CG lɛ] LOC

b.

𠲥C]CG lɛ] LOC

d.

[[车]ω 𠲥C]CG [[tshia] lɛ] car LOC ‘in the car’ [[[旧]ω [书]ω]φ 𠲥C]CG [[[kou] [tsy]] lɛ] old book LOC ‘in/on the old book’

4.2.6  Recursive clitic group with enclitics In addition to the examples in (21), there are other cases in which prosodic recursivity is allowed, as can be seen in (27). (27) a.

b.

[[[看]ω 过C]CG 其C]CG 书 [[[khaŋ] kuo] ki] tsy read EXP MOD book ‘the book that has been read’ [[[收]ω 遘C]CG 其C]CG 批 [[[siu] kau] ki] phiɛ receive PVP MOD letter ‘the letter that was received’

Fuzhou enclitics and the Type A clitic group  107 4.2.7 Summary So far, we have seen the brief picture of Fuzhou enclitics and their morphosyntactic functions. From the data presented in this section, we can find that these elements in the Fuzhou dialect share some of the most common morphosyntactic properties of clitics across languages: (a) they all belong to functional categories; (b) they never occur as the only element of an utterance and must attach to the adjacent prosodic unit on the left; (c) the meaning of the string of the host plus the enclitic is predictable from the meaning of the host and that of the enclitic; and (d) they can attach to material already containing the affix, as in (9b) and (26c), or the enclitic, as in (21) and (27). Therefore, according to the discussion on general properties of clitics in Section 4.1.1, it is reasonable to consider these elements in the Fuzhou dialect as enclitics. In accordance with the definition of the Fuzhou clitic group domain in (8), the group of the host plus the enclitic thus forms a type of clitic group (the Type A clitic group) in this dialect. In Section 4.3, we will see that there are phonological phenomena characteristic only of such a type of clitic group in Fuzhou, which provides further evidence for the existence of enclitics and the clitic group consisting of “host + enclitic” in this dialect.

4.3 Phonological phenomena within the Type A clitic group in the Fuzhou dialect This section investigates the phonological behavior of the clitic group composed of “host + enclitic”, namely Type A clitic group in the Fuzhou dialect, with respect to the application of three Fuzhou phonological rules, TS, FC, and CL. I will show that there are phonological phenomena referring crucially to the domain of the Type A clitic group, but not to any other context. 4.3.1  TS within the Type A clitic group Chapter 3 has shown that the TS rule can apply within the prosodic word domain formed by some subtypes of morphosyntactic words, as exemplified in (28I). This rule can also apply to certain strings of sounds at the phrasal level, as exemplified in (28II).6 Sandhi tones are marked in bold in the following examples in (28). (28) I.

II.

Application of TS within the prosodic word domain a. 沙发 sa44 xuaʔ23 → sa51 ʔuaʔ23 ‘sofa’ 31 h 31 b. 老鼠 lo ts y → lo24 ʒy31 ‘mouse’ c. 暝晡 maŋ51 puo44 → maŋ44 muo44 ‘evening’ Application of TS at the phrasal level a. 食 饭 b. 野 ʔia31 siɛʔ5 puoŋ242 21 242 → siɛʔ puoŋ → ʔia24 eat rice very ‘to eat food’ ‘very good’

好 xo31 ʔo31 good

108  Fuzhou enclitics and the Type A clitic group By contrast, it has been noticed that some elements in Fuzhou never cause the tone of the preceding syllable to undergo TS (Wright 1983; Chan 1985; Chen 1998; Li 2002, among others). Compare the two examples in (29), in which tones in question are marked in bold. The position where the rule fails to apply is indicated with “#”. (29) a. →

b.

旧 书 kou242 tsy44 ku44 ʒy44 old book ‘old book’

→

旧 其 kou242 ki0 242 kou # ʔi0 old MOD ‘old book’

书 tsy44 tsy44 book

We can find that (29a) and (29b) have different phonological behavior in terms of the application of TS. In (29a), TS applies between 旧 ‘old’ and 书 ‘book’ and changes the tone of 旧 ‘old’, while TS is blocked in (29b), although these two examples have similar morphosyntactic structure, namely the modifier-head structure. Some linguists suggest that the blocking of TS in cases like (29b) can be ascribed to the neutral tone carried by elements like 其 [ki0] (e.g., Chan 1985; Li 2002, among others). According to these scholars, there is no applicable phonological environment for TS if the non-first tone is a neutral tone. Nevertheless, notice that even elements bearing a non-neutral tone may consistently cause the blocking of TS, as exemplified in (30), which shows that the blocking of TS in cases like (29b) and (30) cannot be simply ascribed to the tonal value. (30) a. →

c. →

悬悬 keiŋ51 keiŋ51 keiŋ31 keiŋ51 *keiŋ31 keiŋ21 (very) tall ‘rather tall’ 坐 soy242 soy242 # *soy51 sit ‘to sit awhile’

#

势 siɛ213 niɛ213 niɛ213 AdjR 囇 la242 la242 la242 DLM

b. →

d. →

白白 势 paʔ5 paʔ5 siɛ213 paʔ31 paʔ5 # liɛ213 31 21 *paʔ paʔ liɛ213 (very) white AdjR ‘rather white’ 歇 囇 xyoʔ23 la242 xyoʔ23 # la242 44 *xyoʔ la242 rest DLM ‘to take a rest’

According to the discussion in Section 4.2, elements like 其 [ki0], 势 [siɛ213], and 囇 [la242] that can trigger the blocking of TS are actually enclitics in the Fuzhou dialect. Hence, I assume that the Type A clitic group composed of “host + enclitic” in this dialect cannot form the domain of application for TS. Specifically, TS is blocked between the host and the enclitic. This assumption is well supported by data from the Fuzhou dialect, as illustrated in (31)–(36), where tones in question are marked in bold.

Fuzhou enclitics and the Type A clitic group  109 (31) Blocking of TS in “host + possessive/modificational/nominalization marker 其 [ki0]” a. [[依妹]ω 其C]CG 衣裳 [[ʔi44 mui213] ki0] ʔi44 suoŋ51 51 213 0 → [[ʔi mui ] # ʔi ] ʔi44 luoŋ51 younger sister POSS clothes ‘younger sister’s clothes’ b. [[旧]ω 其C]CG 书 c. [[我]ω 其C]CG [[kou242] ki0] tsy44 [[ŋuai31] ki0] 242 0 44 31 → [[kou ] # ʔi ] tsy → [[ŋuai ] # ʔi0] old MOD book I NOM ‘old book’ ‘mine’ (32) Blocking of TS in “host + adjective reduplication marker” a. 伊 [[悬 悬]ω 势/式/喏C]CG ʔi44 [[keiŋ51 keiŋ51] siɛ213/seiʔ23/luoʔ23] → ʔi44 [[keiŋ31 keiŋ51] # niɛ213/neiʔ23/nuoʔ23] 44 *ʔi [[keiŋ31 keiŋ21] niɛ213/neiʔ23/nuoʔ23] he (very) tall AdjR ‘He is rather tall.’ b. 面 [[白白]ω 势/式/喏C]CG meiŋ213 [[paʔ5 paʔ5] siɛ213/seiʔ23/luoʔ23] 213 31 5 → meiŋ [[paʔ paʔ ] # liɛ213/leiʔ23/luoʔ23] 213 31 21 *meiŋ [[paʔ paʔ ] liɛ213/leiʔ23/luoʔ23] face (very) white AdjR ‘The face is rather white.’ c. 工作 [[闲闲落落]ω 势/式/喏C]CG køyŋ44 tsouʔ23 [[ʔeiŋ51 ʔeiŋ51 loʔ5 loʔ5] siɛ213/seiʔ23/luoʔ23] → køyŋ51 tsouʔ23 [[ʔeiŋ21 ŋeiŋ21 noʔ31 loʔ5] # liɛ213/leiʔ23/luoʔ23] 51 23 *køyŋ tsouʔ [[ʔeiŋ21 ŋeiŋ21 noʔ31 loʔ21] liɛ213/leiʔ23/luoʔ23] job (very) easy/relaxing AdjR ‘The job is very easy.’ (33) Blocking of TS in “host + aspect marker” I. Host + durative aspect marker 𠲥 [lɛ0] a. 门 [[关]ω 𠲥C]CG mouŋ51 [[kuoŋ44] lɛ0] → mouŋ51 [[kuoŋ44] # nɛ0] door close DUR ‘The door is closed.’ c. 汝 [[徛]ω 𠲥C]CG ny31 [[khiɛ242] lɛ0] → ny31 [[khiɛ242] # lɛ0] you stand DUR ‘You, stand there!’

b. →

[[坐]ω 𠲥C]CG 讲 [[soy242] lɛ0] kouŋ31 [[soy242] # lɛ0] kouŋ31 sit DUR speak ‘to speak sitting down’

110  Fuzhou enclitics and the Type A clitic group II.

Host + experiential aspect marker 过 [kuo213] /着 [tuoʔ5] a. [[去]ω 过C]CG 天津 [[kho213] kuo213] thiɛŋ44 kiŋ44 h 213 → [[k o ] # ʔuo213] thiɛŋ44 kiŋ44 h 51 *[[k o ] ʔuo213] thiɛŋ44 kiŋ44 go EXP Tianjin ‘to have been to Tianjin before’ b. [[食]ω 着C]CG 鳗鱼 [[siɛʔ5] tuoʔ5] muaŋ51 ŋy51 → [[siɛʔ5] # luoʔ5] muaŋ31 ŋy51 31 *[[siɛʔ ] luoʔ5] muaŋ31 ŋy51 eat EXP eel ‘to have eaten eels before’

III.

Host + perfective aspect marker 𠲥 [lɛ0] a. [[食]ω 𠲥C]CG 暝 [[siɛʔ5] lɛ0] maŋ51 5 0 → [[siɛʔ ] # lɛ ] maŋ51 eat PERF dinner ‘to sleep after eating the dinner’ b. [[睏]ω 𠲥C]CG 大 [[khouŋ213] lɛ0] tuai242 → [[khouŋ213] # nɛ0] tuai21 sleep PERF big ‘slept more than half of the day’

去 kho213 kho51 go

睏 khouŋ213 khouŋ213 sleep

半 puaŋ213 βuaŋ44 half

日 niʔ5 niʔ5 day

IV.

Host + perfective aspect marker 去 [kho0] a. 伊 [[病]ω 去C]CG ʔi44 [[paŋ242] kho0] 44 242 → ʔi [[paŋ ] # ŋo0] he sick PERF ‘He is sick.’ 去C]CG b. 水缸 [[必]ω tsui31 kouŋ44 [[peiʔ23] kho0] → tsui21 ʔouŋ44 [[peiʔ23] # ʔo0] water jar crack PERF ‘The water jar is cracked.’

V.

Host + sentence final particle 了 [lau31] a. [[[食]ω [饭]ω]φ 了C]CG [[[siɛʔ5] [puoŋ242]] lau31] 21 242 → [[[siɛʔ ] [puoŋ ]] # nau31] 21 *[[[siɛʔ ] [puoŋ51]] nau31] eat rice CRS ‘It’s time to eat.’

Fuzhou enclitics and the Type A clitic group  111 b. →

VI.

暝 [[好]ω maŋ51 [[xo31] 51 maŋ [[xo31] 51 *maŋ [[xo24] dinner good ‘The dinner is ready.’

#

了C]CG lau31] lau31] lau31] CRS

Host + delimitative aspect marker 囇 [la242] a. [[坐]ω 囇C]CG b. [[歇]ω 囇C]CG [[soy242] la242] [[xyoʔ23] la242] 242 242 23 → [[soy ] # la ] → [[xyoʔ ] # la242] 51 242 *[[soy ] la ] *[[xyoʔ44] la242] sit DLM rest DLM ‘to sit awhile’ ‘to take a rest’

(34) Blocking of TS in “host + post-verbal particle” I. Host + post-verbal particle 敆 [kaʔ0] a. [[定]ω 敆C]CG 今旦 [[tiaŋ242] kaʔ0] kiŋ44 taŋ213 242 0 → [[tiaŋ ] # ŋaʔ ] kiŋ51 naŋ213 set PVP today ‘to be scheduled today’ b. [[排]ω 敆C]CG 厅中 [[pɛ51] kaʔ0] thiaŋ44 touŋ44 → [[pɛ51] # ʔaʔ0] thiaŋ44 nouŋ44 put PVP drawing room ‘to be put in the drawing room’ II.

Host + post-verbal particle 遘 [kau213] a. [[收]ω 遘C]CG 批 [[siu44] kau213] phiɛ44 → [[siu44] # ʔau213] phiɛ44 51 *[[siu ] ʔau213] phiɛ44 receive PVP letter ‘to receive a letter’ b. [[等]ω 遘C]CG 三 点 [[tiŋ31] kau213] saŋ44 teiŋ31 → [[tiŋ31] # ŋau213] saŋ51 neiŋ31 44 *[[tiŋ ] ŋau213] saŋ51 neiŋ31 wait PVP three o’clock ‘to wait until three o’clock’

112  Fuzhou enclitics and the Type A clitic group III.

Host + post-verbal particle 𠲥 [lɛ0] a. [[看]ω 𠲥C]CG 野 清楚 [[khaŋ213] lɛ0] ʔia31 tshiŋ44 tshu31 → [[khaŋ213] # nɛ0] ʔia21 tshiŋ51 ʒu31 look PVP very clear ‘saw (something) very clearly’ b. 伊 [[跳]ω 𠲥C]CG 蜀 身 都 ʔi44 [[thiu213] lɛ0] suoʔ5 siŋ44 tu44 → ʔi44 [[thiu213] # lɛ0] suoʔ44 siŋ44 tu21 he jump PVP one body all ‘He jumped and was covered in sweat.’

是 sei242 li51 be

(35) Blocking of TS in “host + locative marker 𠲥 [lɛ0]” a. b. [[车]ω [[面]ω 𠲥C]CG [[meiŋ213] lɛ0] [[tshia44] → [[meiŋ213] # nɛ0] → [[tshia44] # face LOC car ‘on the face’ ‘in the car’

汗 kaŋ242 kaŋ242 sweat

𠲥C]CG lɛ0] lɛ0] LOC

(36) Blocking of TS in recursive clitic group with enclitics a. 了C]CG 门 [[[开]ω 𠲥C]CG mouŋ51 [[[khui44] lɛ0] lau31] → mouŋ51 [[[khui44] # lɛ0] # lau31] door open DUR CRS ‘The door is already open.’ b. 我 [[[看]ω 过C]CG 了C]CG ŋuai31 [[[khaŋ213] kuo213] lau31] → ŋuai31 [[[khaŋ213] # ŋuo213] # lau31] 31 h 21 *ŋuai [[[k aŋ ] ŋuo51] lau31] I see EXP CRS ‘I have seen (that).’ c. [[[看]ω 过C]CG 其C]CG 书 [[[khaŋ213] kuo213] ki0] tsy44 h 213 213 0 → [[[k aŋ ] # ŋuo ] # ʔi ] tsy44 h 51 213 0 *[[[k aŋ ] ŋuo ] ʔi ] tsy44 read EXP MOD book ‘the book that has been read’ d. [[[收]ω 遘C]CG 其C]CG 批 [[[siu44] kau213] ki0] phiɛ44 44 213 0 → [[[siu ] # ʔau ] # ʔi ] phiɛ44 51 213 0 *[[[siu ] ʔau ] ʔi ] phiɛ44 receive PVP MOD letter ‘the letter that was received’ From the examples in (31)–(36), we can find that TS consistently fails to apply within the domain formed by the host plus the enclitic, namely the Type A clitic

Fuzhou enclitics and the Type A clitic group  113 group in the Fuzhou dialect. Therefore, one distinctive phonological property of this type of clitic group in Fuzhou is the obligatory blocking of the TS rule between the host and the enclitic CG-internally, which is distinct from the prosodic word and certain strings of sounds at the phrasal level (i.e., the phonological phrase). 4.3.2  FC within the Type A clitic group In Chapter 1 and Chapter 3, we have seen that the FC rule can not only apply within the prosodic word domain, as in (37I), but also apply within the domain formed by certain types of phrasal-level strings, as in (37II).7 Sandhi finals are marked in bold. (37) I. II.

Application of FC within the prosodic word domain a. 熨斗 ʔouʔ23 tau31 → ʔuʔ24 tau31 ‘iron (for clothes)’ b. 裤头 khou213 thau51 → khu44 lau51 ‘trouser waist’ Application of FC at the phrasal level a. 旧 书 b. 爱 食 kou242 tsy44 ʔoy213 siɛʔ5 → ku44 ʒy44 → ʔøy44 liɛʔ5 old book love eat ‘old book’ ‘to like to eat’

Nevertheless, since FC is a tonally conditioned rule that occurs along with tone sandhi rules, it must apply only within a domain where tone sandhi can occur. Since the Type A clitic group in Fuzhou cannot serve as the domain of application for TS (it is clearly not the domain of application for MTS either since MTS only applies to some prosodic words formed through morphological reduplication), as we have seen in Section 4.3.1, it is reasonable to assume that the Type A  clitic group is not the domain of application for FC. This assumption is well supported by empirical evidence from the Fuzhou dialect, as shown in the examples in (38)–(43). (38) Blocking of FC in “host + possessive/modificational/nominalization marker 其 [ki0]” b. [[旧]ω a. [[店]ω 其C]CG 门 其C]CG 书 [[taiŋ213] ki0] muoŋ51 [[kou242] ki0] tsy44 213 0 51 242 0 → [[taiŋ ] # ŋi ] muoŋ → [[kou ] # ʔi ] tsy44 213 0 51 242 0 *[[teiŋ ] ŋi ] muoŋ *[[ku ] ʔi ] tsy44 store POSS door old MOD book ‘the door of the store’ ‘old book’ c. [[竹]ω 其C]CG [[tøyʔ23] ki0] 23 → [[tøyʔ ] # ʔi0] 23 *[[tyʔ ] ʔi0] bamboo NOM ‘things made of bamboo’

114  Fuzhou enclitics and the Type A clitic group (39) Blocking of FC in “host + adjective reduplication marker” a. 水杯 [[碎碎]ω 势/式/喏C]CG tsui31 pui44 [[tshoy213 tshoy213] siɛ213/seiʔ23/luoʔ23] 21 44 h 51 h 213 → tsui βui [[ts øy ts oy ] # liɛ213/leiʔ23/luoʔ23] 21 44 h 51 h 51 *tsui βui [[ts øy ts øy ] liɛ213/leiʔ23/luoʔ23] cup (very) shattered AdjR ‘The cup is very shattered.’ b. 水 [凊凊]ω 势/式/喏C]CG tsui31 [[tsheiŋ213 tsheiŋ213] siɛ213/seiʔ23/luoʔ23] → tsui31 [[tshiŋ51 tsheiŋ213] # niɛ213/neiʔ23/nuoʔ23] 31 h 51 h 51 *tsui [[ts iŋ ts iŋ ] niɛ213/neiʔ23/nuoʔ23] water (very) cold AdjR ‘The water is very cold.’ (40) Blocking of FC in “host + aspect marker” I. Host + durative aspect marker 𠲥 [lɛ0] a. [[坐]ω 𠲥C]CG 讲 [[soy242] lɛ0] kouŋ31 242 0 → [[soy ] # lɛ ] kouŋ31 242 0 *[[søy ] lɛ ] kouŋ31 sit DUR speak ‘to speak sitting down’ II.

Host + experiential aspect marker 过 [kuo213] /着 [tuoʔ5] a. [[试]ω 过C]CG 只 块 衣裳 [[tshei213] kuo213] tsi31 toy213 ʔi44 suoŋ51 → [[tshei213] # ʔuo213] tsi44 loy213 ʔi44 luoŋ51 h 51 213 44 213 *[[ts i ] ʔuo ] tsi loy ʔi44 luoŋ51 try EXP this Cl clothes ‘to have tried this piece of clothes’ b. [[办]ω 着C]CG 护照 [[paiŋ242] tuoʔ5] xou242 tsiu213 242 → [[paiŋ ] # nuoʔ5] xu51 ʒiu213 *[[peiŋ44] nuoʔ5] xu51 ʒiu213 do EXP passport ‘to have applied for a passport’

III.

Host + perfective aspect marker 𠲥 [lɛ0] a. [[睏]ω 𠲥C]CG 大 半 [[khouŋ213] lɛ0] tuai242 puaŋ213 → [[khouŋ213] # nɛ0] tuai21 βuaŋ44 h 213 0 *[[k uŋ ] nɛ ] tuai21 βuaŋ44 sleep PERF big half ‘slept for most of the day’

日 niʔ5 niʔ5 niʔ5 day

Fuzhou enclitics and the Type A clitic group  115 IV.

V.

VI.

Host + perfective aspect marker 去 [kho0] a. 水缸 [[必]ω 去C]CG tsui31 kouŋ44 [[peiʔ23] kho0] → tsui21 ʔouŋ44 [[peiʔ23] # ʔo0] 21 44 *tsui ʔouŋ [[piʔ23] ʔo0] water jar crack PERF ‘The water jar is cracked.’ Host + sentence final particle 了 [lau31] a. 伊 [[出]ω 了C]CG ʔi44 [[tshouʔ23] lau31] 44 h 23 → ʔi [[ts ouʔ ] # lau31] 44 *ʔi [[tshuʔ51] lau31] he go out CRS ‘He has gone out.’ Host + delimitative aspect marker 囇 [la242] a. [[坐]ω 囇C]CG [[soy242] la242] 242 → [[soy ] # la242] 51 *[[søy ] la242] sit DLM ‘to sit awhile’

(41) Blocking of FC in “host + post-verbal particle” a. 通知 敆C]CG 外斗 [[贴]ω thuŋ44 ti44 [[thaiʔ23] kaʔ0] ŋiɛ242 tau31 h 44 44 h 23 0 → t uŋ ni [[t aiʔ ] # kaʔ ] ŋiɛ51 lau31 h 44 44 h 23 0 *t uŋ ni [[t eiʔ ] kaʔ ] ŋiɛ51 lau31 notice paste PVP outside ‘The notice is posted outside.’ b. [[睏]ω 遘C]CG 三 点 [[khouŋ213] kau213] saŋ44 teiŋ31 → [[khouŋ213] # ŋau213] saŋ51 neiŋ31 *[[khuŋ213] ŋau213] saŋ51 neiŋ31 sleep PVP three o’clock ‘to sleep until three o’clock’ c. 衣裳 [[颂]ω 𠲥C]CG 破破 ʔi44 suoŋ51 [[søyŋ242] lɛ0] phuai213 phuai213 → ʔi44 luoŋ51 [[søyŋ242] # nɛ0] phuai51 phuai213 *ʔi44 luoŋ51 [[syŋ242] nɛ0] phuai51 phuai213 clothes wear PVP (very) ragged ‘The clothes are worn out.’

式 seiʔ23 leiʔ23 leiʔ23 AdjR

116  Fuzhou enclitics and the Type A clitic group (42) Blocking of FC in “host + locative marker 𠲥 [lɛ0]” a. [[面]ω 𠲥C]CG [[meiŋ213] lɛ0] → [[meiŋ213] # nɛ0] *[[miŋ213] nɛ0] face LOC ‘on the face’ (43) Blocking of FC in recursive clitic group with enclitics a. 伊 羊毛裤 [[[颂]ω 𠲥C]CG ʔi44 yoŋ51 mo51 khou213 [[[søyŋ242] lɛ0] 44 21 21 213 242 → ʔi yoŋ mo ʔou [[[søyŋ ] # nɛ0] # 44 21 21 213 *ʔi yoŋ mo ʔou [[[syŋ242] nɛ0] he woolen pant wear DUR ‘He has been wearing the woolen pants already.’ b. 我 [[[办]ω 过C]CG 了C]CG ŋuai31 [[[paiŋ242] kuo213] lau31] 31 242 213 → ŋuai [[[paiŋ ] # ŋuo ] # lau31] 31 51 51 *ŋuai [[[peiŋ ] ŋuo ] lau31] I do EXP CRS ‘I have done (that).’

了C]CG lau31] lau31] lau31] CRS

It can be seen from the examples in (38)–(43) that FC is blocked in the Type A  clitic group domain in the Fuzhou dialect. Thus, we can find that the second distinctive phonological property of the Type A  clitic group in this dialect is the obligatory blocking of the FC rule between the host and the enclitic inside the Type A clitic group domain. Since the FC rule can apply within the prosodic word domain and the domain formed by certain phrasal-level strings of sounds (i.e., the phonological phrase), the obligatory blocking of FC is characteristic only of the domain formed by the Type A clitic group. 4.3.3  CL within the Type A clitic group As we have seen in Chapter 3, the CL rule in the Fuzhou dialect does not apply within the domain of the prosodic word formed by all types of Fuzhou morphosyntactic words, although it refers to the prosodic word as its domain of application at the lexical level. For example, it does not apply to the prosodic word domain formed by “diminutive” nouns and reduplicated forms of monosyllabic adjectives, as shown in (44), where sandhi initials in question are marked in bold. (44)

a. b. c. d.

杯杯 桶桶 白白 凊凊

pui44 pui44 → pui31 pui44 *pui31 βui44 *thøyŋ31 nøyŋ31 thøyŋ31 thøyŋ31 → thøyŋ31 thøyŋ31 paʔ5 paʔ5 → paʔ31 paʔ5 *paʔ31 βaʔ5 h 213 h 213 h 51 h 213 ts eiŋ ts eiŋ → ts iŋ ts eiŋ *tshiŋ51 ʒeiŋ213

‘cup’ ‘bucket’ ‘(very) white’ ‘(very) cold’

Fuzhou enclitics and the Type A clitic group  117 The CL rule also fails to apply within the domain formed by some phrasal-level strings, as in (45).8 (45) a. →

b.

食 饭 siɛʔ5 puoŋ242 21 siɛʔ # puoŋ242 21 *siɛʔ βuoŋ242 eat rice ‘to eat food’

→

买 锅 mɛ31 kuo44 21 mɛ # kuo44 21 *mɛ ʔuo44 buy pan ‘to buy a pan’

By contrast, it has been found that some elements in the Fuzhou dialect always undergo CL (cf. Chen 1998; Li & Liang 2001; Li 2002, among others). Compare (45b) with the example in (46). (46) →

买 过 mɛ31 kuo213 31 mɛ ʔuo213 buy EXP ‘to have bought (something)’

We can find that (45b) and (46) share the same phonological environment, namely an open syllable followed by the initial [k]. However, CL does not apply in (45b), with the initial [k] of 锅 ‘pan’ remaining unchanged, while the experiential aspect marker 过 in (46) undergoes the CL rule, with the initial [k] changed to the glottal stop (or zero initial). Elements like 过 [kuo213] are enclitics in the Fuzhou dialect, according to the discussion in Section 4.2. A thorough investigation of Fuzhou data reveals that the Type A clitic group formed by the host plus the enclitic serves as a domain of application for the CL rule in this dialect, which can be illustrated by the examples in (47)–(52). (47) Application of CL in marker 其 [ki0]” a. [[我]ω 其C]CG [[ŋuai31] ki0] → [[ŋuai31] ʔi0] I POSS ‘my book’ c. [[食]ω 其C]CG [[sieʔ5] ki0] → [[sieʔ5] ʔi0] eat NOM ‘food’

“host + possessive/modificational/nominalization 书 tsy44 tsy44 book

b. →

其C]CG [[红]ω [[ʔøyŋ51] ki0] [[ʔøyŋ51] ŋi0] red MOD ‘red flower’

花 xua44 xua44 flower

118  Fuzhou enclitics and the Type A clitic group (48) Application of CL in “host + adjective reduplication marker” a. 伊 [[悬 悬]ω 势/式/喏C]CG ʔi44 [[keiŋ51 keiŋ51] siɛ213/seiʔ23/luoʔ23] → ʔi44 [[keiŋ31 keiŋ51] niɛ213/neiʔ23/nuoʔ23] he (very) tall AdjR ‘He is rather tall.’ b. [[舒舒畅畅]ω 势/式/喏C]CG [[tshy44 tshy44 thuoŋ213 thuoŋ213] siɛ213/seiʔ23/luoʔ23] → [[tshy21 ʒy21 luoŋ51 nuoŋ213] niɛ213/neiʔ23/nuoʔ23] (very) comfortable AdjR ‘very comfortable’ (49) Application of CL in “host + aspect marker” I. Host + durative aspect marker 𠲥 [lɛ0] a. 门 [[关]ω 𠲥C]CG mouŋ51 [[kuoŋ44] lɛ0] → mouŋ51 [[kuoŋ44] nɛ0] door close DUR ‘The door is closed.’ II.

Host + experiential aspect marker 过 [kuo213] /着 [tuoʔ5] a. [[去]ω 过C]CG 天津 [[kho213] kuo213] thiɛŋ44 kiŋ44 → [[kho213] ʔuo213] thiɛŋ44 kiŋ44 go EXP Tianjin ‘to have been to Tianjin before’ b. [[办]ω 着C]CG 护照 [[paiŋ242] tuoʔ5] xou242 tsiu213 → [[paiŋ242] nuoʔ5] xu51 ʒiu213 do EXP passport ‘to have applied for a passport’

III.

Host + perfective aspect marker 𠲥 [lɛ0] a. [[睏]ω 𠲥C]CG 大 半 [[khouŋ213] lɛ0] tuai242 puaŋ213 → [[khouŋ213] nɛ0] tuai21 βuaŋ44 sleep PERF big half ‘slept for most of the day’

IV.

Host + perfective aspect marker 去 [kho0] a. 伊 [[病]ω 去C]CG ʔi44 [[paŋ242] kho0] → ʔi44 [[paŋ242] ŋo0] he sick PERF ‘He is sick.

日 niʔ5 niʔ5 day

Fuzhou enclitics and the Type A clitic group  119 b. →

V.

VI.

水缸 [[必]ω 去C]CG tsui31 kouŋ44 [[peiʔ23] kho0] tsui21 ʔouŋ44 [[peiʔ23] ʔo0] water jar crack PERF ‘The water jar is cracked.’

Host + sentence final particle 了 [lau31] a. 伊 [[[生]ω [囝]ω]φ 了C]CG ʔi44 [[[saŋ44] [kiaŋ31]] lau31] → ʔi44 [[[saŋ51] [kiaŋ31]] nau31] she give birth child CRS ‘She has given birth to a child.’ Host + delimitative aspect marker 囇 [la242] a. [[听]ω 囇C]CG [[thiaŋ44] la242] → [[thiaŋ44] na242] listen DLM ‘to listen awhile’

(50) Application of CL in “host + post-verbal particle” I. Host + post-verbal particle 敆 [kaʔ0] a. [[定]ω 敆C]CG 今旦 [[tiaŋ242] kaʔ0] kiŋ44 taŋ213 242 0 → [[tiaŋ ] ŋaʔ ] kiŋ51 naŋ213 set PVP today ‘to be scheduled today’ b. [[排]ω 敆C]CG 厅中 [[pɛ51] kaʔ0] thiaŋ44 touŋ44 51 0 → [[pɛ ] ʔaʔ ] thiaŋ44 nouŋ44 put PVP drawing room ‘to be put in the drawing room’ II.

Host + post-verbal particle 遘 [kau213] a. [[收]ω 遘C]CG 批 [[siu44] kau213] phiɛ44 → [[siu44] ʔau213] phiɛ44 receive PVP letter ‘to receive a letter’ b. [[等]ω 遘C]CG 三 点 [[tiŋ31] kau213] saŋ44 teiŋ31 → [[tiŋ31] ŋau213] saŋ51 neiŋ31 wait PVP three o’clock ‘to wait until three o’clock’

120  Fuzhou enclitics and the Type A clitic group III.

Host + post-verbal particle 𠲥 [lɛ0] a. [[看]ω 𠲥C]CG 野 清楚 [[khaŋ213] lɛ0] ʔia31 tshiŋ44 tshu31 → [[khaŋ213] nɛ0] ʔia21 tshiŋ51 ʒu31 look PVP very clear ‘saw (something) very clearly’

(51) Application of CL in “host + locative marker 𠲥 [lɛ0]” a. [[面]ω b. [[桶桶]ω 𠲥C]CG [[meiŋ213] lɛ0] [[thøyŋ31 thøyŋ31] → [[meiŋ213] nɛ0] → [[thøyŋ31 thøyŋ31] face LOC bucket ‘on the face’ ‘in the bucket’

𠲥C]CG lɛ0] nɛ0] LOC

(52) Application of CL in recursive clitic group with enclitics a. [[[看]ω 过C]CG 其C]CG 书 [[[khaŋ213] kuo213] ki0] tsy44 h 213 213 0 → [[[k aŋ ] ŋuo ] ʔi ] tsy44 read EXP MOD book ‘the book that has been read’ b. [[[收]ω 遘C]CG 其C]CG 批 [[[siu44] kau213] ki0] phiɛ44 44 213 0 → [[[siu ] ʔau ] ʔi ] phiɛ44 receive PVP MOD letter ‘the letter that was received’ Empirical evidence presented in (47)–(52) suggests that the CL rule consistently applies between the host and the enclitic within the Type A clitic group domain in the Fuzhou dialect. If we compare the application of CL within the Type A  clitic group domain with its application within the prosodic word domain and the domain formed by certain strings of sounds at the phrasal level, we can find that the CL rule exhibits different degrees of application within these domains. In the prosodic word domain and the domain formed by certain phrasal-level strings, CL is not an obligatory rule, as shown in (44) and (45). By contrast, within the Type A clitic group domain, the application of CL is mandatory, which is another important phonological property of the Type A  clitic group domain in Fuzhou. Due to the fact that Fuzhou enclitics can never stand alone and that CL is mandatory between the host and the enclitic, the syllable initial of an enclitic is always decided by the final of the preceding syllable, showing that Fuzhou enclitics are phonologically dependent. 4.3.4 Summary On the basis of the discussion in Section 4.3, the application/blocking of TS, FC, and CL within the Type A clitic group domain in the Fuzhou dialect can be summarized as in Table 4.1.

Fuzhou enclitics and the Type A clitic group  121 Table 4.1  Phonological rules and the Type A clitic group in the Fuzhou dialect Type A clitic group

Application

Rules

Host + possessive/modificational/nominalization marker Host + adjective reduplication marker Host + aspect marker Host + post-verbal particle Host + locative marker Recursive clitic group with enclitics

TS

FC

CL

× × × × × ×

× × × × × ×

√ √ √ √ √ √

We can find that the Type A  clitic group formed by the group of “host + enclitic” in the Fuzhou dialect is different from the prosodic word and the domain formed by certain phrasal-level constructions (i.e., the phonological phrase; as to be discussed in Chapter 5) with respect to the application of phonological rules. We have seen that both TS and FC apply within the prosodic word domain formed by some subtypes of morphosyntactic words and the domain formed by certain phrasal-level strings but are obligatorily blocked within the Type A clitic group domain. This is an issue of whether a particular phonological rule applies within a given domain. In terms of the CL rule, although it can apply within the prosodic word domain formed by most morphosyntactic words, it is consistently blocked in the prosodic word domain formed by “diminutive” nouns and reduplicated forms of monosyllabic adjectives. Similarly, as we will see in Chapter  5, although CL applies within the domain of the phonological phrase, it is blocked in phrasal-level strings that form the phonological phrase through φ-restructuring. In contrast, CL can apply within the Type A clitic group domain with no exceptions. This is an issue of the degree of application of a particular phonological rule in different prosodic domains. Therefore, we can draw the conclusion that there are phonological phenomena characteristic only of the Type A clitic group domain composed of “host + enclitic” in the Fuzhou dialect, which serves as important evidence for the existence of enclitics and the Type A clitic group in this dialect.

4.4  Violation of SLH within the Type A clitic group As mentioned in Chapter  1, the SLH in prosodic phonology has been challenged by evidence from various languages. Many examples of the violation of Exhaustivity, Nonrecursivity, and even Layeredness have been found across languages. In the Fuzhou dialect, as we have seen in Section  4.2 and Section 4.3, there are also examples in which the constraints of Nonrecursivity and Layeredness are violated. Examples of the violation of Nonrecursivity in (52) are reproduced in (53).

122  Fuzhou enclitics and the Type A clitic group (53) Violation of Nonrecursivity within the Type A clitic group a. [[[看]ω 过C]CG 其C]CG 书 [[[khaŋ213] kuo213] ki0] tsy44 h 213 213 0 → [[[k aŋ ] ŋuo ] ʔi ] tsy44 read EXP MOD book ‘the book that has been read’ b. [[[收]ω 遘C]CG 其C]CG 批 [[[siu44] kau213] ki0] phiɛ44 44 213 0 → [[[siu ] ʔau ] ʔi ] phiɛ44 receive PVP MOD letter ‘the letter that was received’ (53) shows that the Type A clitic group in the Fuzhou dialect, which is composed of “host + enclitic”, may dominate another clitic group of the same type. Take (53a) as an example. In the inner Type A clitic group, the enclitic 过 [kuo213] attaches to the prosodic word [看]ω ‘read’ as the host, and in the outer Type A clitic group, the enclitic 其 [ki0] attaches to the inner Type A clitic group [看过]CG as the host. Similarly, in (53b), the enclitic 遘 [kau213] attaches to the prosodic word [收]ω ‘receive’ as the host in the inner Type A clitic group, and the inner Type A  clitic group is dominated by the outer Type A clitic group, in which the enclitic 其 [ki0] takes the inner Type A clitic group [收遘]CG as the host. Examples in (20b), (26d), and (49Va) violate the constraint of Layeredness, which are re-presented in (54a–c) respectively, with both the citation form and the surface form provided. (54) Violation of Layeredness within the Type A clitic group a. [[[食]ω 了C]CG [饭]ω]φ [[[siɛʔ5] [puoŋ242]] lau31] → [[[siɛʔ21] [puoŋ242]] nau31] eat rice CRS ‘It’s time to eat.’ b. [[[旧]ω [书]ω]φ 𠲥C]CG [[[kou242] [tsy44]] lɛ0] → [[[ku44] [ʒy44]] lɛ0] old book LOC ‘in/on the old book’ c. 伊 [[[生]ω [囝]ω]φ 了C]CG ʔi44 [[[saŋ44] [kiaŋ31]] lau31] → ʔi44 [[[saŋ51] [kiaŋ31]] nau31] she give birth child CRS ‘She has given birth to a child.’ From the examples in (54), we can find that a Fuzhou enclitic can attach to a phrasal-level unit consisting of more than one prosodic word. As to be discussed in Chapter 5, phrasal-level units like [食饭] ‘eat food’ in (54a), [旧书] ‘old book’ in (54b), and [生囝] ‘to give birth to a child’ in (54c) form the domain of the phonological phrase in the Fuzhou dialect. The examples in (54) thus show that

Fuzhou enclitics and the Type A clitic group  123 an enclitic in Fuzhou can attach to a prosodic constituent higher than the prosodic word and the clitic group. In other words, a Type A clitic group in Fuzhou can dominate a phonological phrase, which clearly violates the Layeredness constraint. Examples like those in (53) and (54) constitute a great challenge to the SLH. Nonetheless, this can be accounted for if we assume a weakened SLH that allows the violation of Nonrecursivity and Layeredness in a given language. By so doing, the problem caused by the attachment of enclitics to constituents higher than the prosodic word in the Fuzhou dialect can be nicely captured. This problem is not due to the (Type A) clitic group itself, but only due to the restrictions imposed by the original SLH, showing that a weakened SLH is required in the prosodic phonology of the Fuzhou dialect.

4.5 Summary Based on the discussion on clitics and the clitic group across languages, this chapter presents a thorough investigation of enclitics and the clitic group composed of the host plus the enclitic (Type A clitic group) in the Fuzhou dialect, from the perspectives of morphosyntactic functions and phonological behavior. The following properties of enclitics and the Type A clitic group in this dialect have been identified. (55) Properties of enclitics in the Fuzhou dialect a. Fuzhou enclitics all belong to functional categories. b. Fuzhou enclitics never occur as the only element of an utterance and must attach to the adjacent prosodic unit (ω, CG, or φ) on the left as the host. c. The meaning of the string of the host plus the enclitic is predictable from the meaning of the host and that of the enclitic. d. Fuzhou enclitics can attach to material already containing the affix or the enclitic. e. Fuzhou enclitics are phonologically dependent—the initial of an enclitic is always decided by the final of the preceding syllable. (56) Properties of the Type A clitic group in the Fuzhou dialect a. TS and FC are obligatorily blocked between the host and the enclitic within the Type A clitic group domain. b. CL obligatorily applies between the host and the enclitic within the Type A clitic group domain. Thus, we can find that, on the one hand, enclitic-like elements reported in the literature of the Fuzhou dialect are indeed enclitics, since they share some common properties with clitics in other languages. On the other hand, the group of “host + enclitic” in this dialect does have peculiar phonological behavior as compared to morphosyntactic words (i.e., the prosodic word in this dialect) and certain phrasal-level constructions (i.e., the phonological phrase in this dialect). By establishing a prosodic constituent that contains the host plus the enclitic in the Fuzhou dialect, I distinguish the “host + enclitic” group from the prosodic word

124  Fuzhou enclitics and the Type A clitic group and the phonological phrase. I have thus accounted for the phonological behavior exhibited by the group of “host + enclitic”, part of which has been noticed in previous studies with no further explanation. The distinctive phonological behavior of the Type A clitic group in the Fuzhou dialect, in turn, provides evidence and motivation for the existence of the clitic group within the prosodic hierarchy. Moreover, a Type A clitic group in Fuzhou can not only dominate a prosodic word, but also dominate another clitic group or a phonological phrase, which indicates that the violation of Nonrecursivity and Layeredness is allowed in this dialect. This can be accounted for by assuming a weakened SLH, instead of excluding the clitic group from the prosodic hierarchy. Therefore, the cases of Fuzhou enclitics and the Type A clitic group provide evidence for not only the existence of the clitic group, but also the necessity of a weakened SLH. In the following chapters, we will first move on to the discussion of the phonological phrase in the Fuzhou dialect and phonological rule application at the phrasal level in Chapter  5. Then, in Chapter  6, based on the discussion in Chapter 4 and Chapter 5, we will proceed to the investigation of proclitics in the Fuzhou dialect and the phonological behavior of the Type B clitic group formed by the group of “proclitic + host”. In Chapter 5 and Chapter 6, we will see more examples that challenge the original SLH, showing that a weakened SLH is well supported by the evidence from the Fuzhou dialect.

Notes 1 In addition to these two main types of clitics, mesoclitics and endoclitics are also reported in a few languages. A mesoclitic occurs between the stem of the host and the affix(es), as can be found in Portuguese (Zwicky 1987; Halpern 1995, among others). An endoclitic is inserted within a single word, as reported in languages like Udi, Degema, and Pashto (Harris 2002; Kari 2003; Korpris & Davis 2005, among others). 2 It is noteworthy that in Hayes (1984/1989) and Nespor & Vogel (1986), prosodic words like “please” and “alone” are marked as clitic groups in the cases (5) and (6), in accordance to the SLH. 3 Notice that, in this example and many of the following examples, a monosyllabic monomorphemic word is assigned the prosodic word (ω) status according to the definition of the prosodic word domain in the Fuzhou dialect proposed in Chapter 3, by means of which a clitic group can be constructed by grouping together the prosodic word host and the adjacent clitic in accordance with the definition of the clitic group domain in (8). 4 Elements like 依- in (9b) and -囝 in (26c) are affixes in the Fuzhou dialect, according to the discussion in Chapter 3. Thus, we can find that a Fuzhou enclitic can attach to a string of sounds that already contains an affix. 5 喏 [luoʔ23] is also recorded as 偌 or 若 in the literature (cf. Chen 1998; Li & Liang 2001, among others). 6 It will be shown in Chapter 5 that strings of sounds like those in (28II) form the phonological phrase in the Fuzhou dialect. 7 The strings in (37II), like those in (28II), also form the phonological phrase in the Fuzhou dialect, as to be seen in Chapter 5. 8 The strings in (45) constitute phonological phrases through φ-restructuring, which will be discussed in Chapter 5.

5 Phonological phrase Domain of phrasal-level rule application in the Fuzhou dialect

This chapter is an investigation of the phonological phrase in the Fuzhou dialect through the examination of how TS, FC, and CL operate at the phrasal level. It begins with the long-standing issue of the phrasal-level application of TS (Section 5.1). On the basis of the detailed inspection of the phrasal-level constructions in which the TS rule is triggered, we will see that the domain of application for TS at the phrasal level is the phonological phrase. I will argue that neither the EBA (including the original parameterized algorithms, the Align/Wrap Theory, and the Match Theory) nor the RBA alone can successfully capture the domain formation of the phonological phrase in Fuzhou, which is why an alternative analysis combining the EBA and the RBA is proposed. Besides, a “no look-back” device called the Domain Impenetrability Condition (DIC) is introduced in Section 5.1, which, together with the notion of the phonological phrase, accounts for the complicated application of TS at the phrasal level. Based on the definition of the phonological phrase in Fuzhou, Section 5.2 and Section 5.3 discuss the phrasal-level application of FC and CL respectively and show that it is also the phonological phrase that forms the domain of phrasal-level application for these two rules. It is demonstrated that the degree of application of CL is different from that of TS and FC within the phonological phrase domain, in that CL does not apply in all phonological phrases—it is blocked in phonological phrases that are constructed through φ-restructuring. The DIC is further elaborated upon in Section 5.2 and Section 5.3 with the examples of FC and CL. A summary of the major findings in this chapter is given in Section 5.4.

5.1  Application of TS at the phrasal level 5.1.1  The nature of the problem It has long been noticed that the application of TS at the phrasal level in the Fuzhou dialect is a quite elusive problem (cf. Chen  & Norman 1965a; Chan 1980; Wright 1983; Shih 1986; Hung 1987, 1990; Zhang 1992, 2017; Chan 1998, among others). As mentioned in Chapter 1, TS can apply within some phrasallevel constructions, as in (1), while it is somehow blocked in some other constructions at the phrasal level, as in (2). Sandhi tones in question are marked in bold. The positions where TS applies are marked with “=”, and the positions where TS is blocked are denoted by “#”.

126  Phonological phrase in the Fuzhou dialect (1)

a. → c. →

(2)

a. →

食 饭 siɛʔ5 puoŋ242 siɛʔ21 = puoŋ242 eat rice ‘to eat food’ 煎 鱼尾 tsiɛŋ44 ŋy51 mui31 tsiɛŋ51 = ŋy31 mui31 fry fish tail ‘to fry fish tails’ 跳 野 悬 thiu213 ʔia31 keiŋ51 thiu213 # ʔia21 ʔeiŋ51 *thiu21 ʔia21 ʔeiŋ213 jump very high ‘(somebody) jumps very high’

b. → d. →

b. →

野 俊 ʔia31 tsouŋ213 44 ʔia = ʒouŋ213 very beautiful ‘very beautiful’ 好 队员 xo31 tui242 ʔuoŋ51 21 xo = tui44 ʔuoŋ51 good team member ‘good team member’ 研究 ŋiɛŋ31 kiu213 ŋiɛŋ44 kiu213 # *ŋiɛŋ21 kiu44 study ‘to study insects’

虫 thøyŋ51 thøyŋ51 thøyŋ51 insect

The “asymmetry” in the phrasal-level application of TS illustrated by the examples in (1) and (2) was first observed in Chen & Norman (1965a), in which they distinguish four types of “junctures” in the Fuzhou dialect, namely terminal junctures, plus junctures, intermediate junctures, and close junctures. According to Chen & Norman, TS is blocked before a terminal juncture and a plus juncture but applies before an intermediate juncture and a close juncture. They assume that these four types of junctures are correlated with grammatical features, which can be summarized in (3). (3)

TS and Chen & Norman’s junctures I. Terminal juncture (blocking of TS) # End of sentence/clause II. Plus juncture (blocking of TS) a. Subject # Predicate b. Determiner-Classifier # Noun c. Verb # Resultative complement d. Reduplicated Adjective # adjective reduplication marker e. Some marked words (e.g., 共 [koyŋ242] ‘and, with’; 乞 [khøyʔ23] ‘to give’; 着 [tuoʔ5] ‘to be in or at’) # III. Intermediate juncture (application of TS) a. Verb = Object (both monosyllabic) b. 会 [ʔa242] (‘can’) = Verb c. 𠲥 [lɛ31] (progressive action) = Verb d. Adjective = Adjective IV. Close juncture (application of TS) a. Determiner = Classifier b. Adjective = Noun (both monosyllabic) c. Adverb = Verb (both monosyllabic) d. 𣍐 [ma242] (‘cannot’) = Verb

Phonological phrase in the Fuzhou dialect  127 Chen & Norman’s observation admirably summarizes a wide variety of syntactic contexts in which TS is triggered or blocked, which has inspired many subsequent analyses. However, the summary in (3) is simply a list that contains several lexically marked contexts. The junctures listed in (3) may share some common characteristics, and what we need is an adequate linguistic analysis to capture their common nature and make correct predictions about where the TS rule should be triggered. A closer examination of Chen & Norman’s list reveals that they mix up constructions at different levels, and the application/blocking of TS in some of these constructions should actually be handled with respect to prosodic domains other than the phonological phrase. For example, a “terminal juncture”, which is associated with intonation according to Chen & Norman (1965a), should be treated as a notion related to the intonational phrase domain (to be discussed in detail in Chapter 7). By contrast, a string containing a reduplicated adjective and the adjective reduplication marker (with a “plus juncture” in between) should be considered to be a Type A clitic group domain, as we have seen in Chapter 4. The application of TS at an “intermediate juncture” between the members of a reduplicated adjective should not be dealt with in this chapter either because a reduplicated adjective forms a prosodic word domain, as shown in Chapter 3. Hence, these constructions should be excluded from the discussion in this chapter. There are some syntactic contexts that are not covered in Chen  & Norman’s summary. For example, according to (3IIIa), TS occurs between a monosyllabic verb and a monosyllabic noun. Nonetheless, as we can find from (1c) and (2b), the TS rule can occur between a monosyllabic verb and a disyllabic noun, while it is blocked between a disyllabic verb and a monosyllabic noun, both of which are not listed in Chen & Norman’s summary. Chen & Norman also do not mention whether TS applies between a verb and a noun phrase containing an embedded adjective or “determiner/numeral-classifier” construction. Other phrasal-level constructions, such as prepositional phrase, serial verb construction, double-object construction, and so forth, are not covered in (3) either. In order to account for Fuzhou TS at the phrasal level, we need to take more phrasal-level constructions into consideration. By excluding the lexically marked or irrelevant contexts and adding some constructions not included in Chen & Norman’s list, we can have a more complete list of phrasal-level constructions we are going to explore, as presented in (4). An adequate linguistic analysis must be able to account for the application/blocking of TS within all these contexts. (4)

Phrasal-level constructions in the Fuzhou dialect a. Subject-predicate b. Modifier-head (attributive-noun and adverbial-adjective/verb) c. Verb-object (including various subtypes) d. Verb-resultative complement e. Determiner/numeral-classifier-noun f. Preposition-object g. Serial verb construction h. Double-object construction i. Pivotal construction j. Passive construction

128  Phonological phrase in the Fuzhou dialect The crux of the problem we need to address thus can be encapsulated in the following questions: (a) How is the TS rule triggered or blocked in these phrasallevel constructions? (b) What is the common feature of the positions where TS is triggered or blocked? (c) How to define the domain of application for TS at the phrasal level? A good deal of literature has been published on these questions. In the subsections that follow, I  will first briefly review the most important previous analyses and discuss their advantages and shortcomings (Section 5.1.2). We will see that a satisfactory analysis has yet to be achieved, and hence an alternative analysis is required. In Section  5.1.3, I  will explore whether the two major approaches to phonological phrasing, namely the RBA and the EBA (including the original parameterized algorithms, the Align/Wrap Theory, and the Match Theory), can account for the phrasal-level application of TS. Then, I will provide an alternative analysis on the basis of a thorough investigation of relevant Fuzhou data in Section  5.1.4. A  brief summary is presented in Section 5.1.5. 5.1.2  Previous analyses 5.1.2.1  Chan (1980) Chan (1980) is the first attempt that directly relates the issue of Fuzhou TS at the phrasal level to syntactic structure. She points out that in the Fuzhou dialect the head of a major syntactic category coincides with the last and dominant syllable of the TS domain at the phrasal level. Based on such an observation, Chan proposes the Head Dominance Condition, as presented in (5), according to which basic TS domains at the phrasal level in Fuzhou can be represented as in (6). (5)

Head Dominance Condition (Chan 1980) Let y be a monosyllabic word immediately dominated by a preterminal category symbol Y, and Y be the head of X. Daughters of Z are within the tone sandhi domain of y if and only if Z is the first node to the left of Y, and the daughters of Z are monosyllabic words.

(6)

a.

X Z z

=

b.

X

Y

Z

y

z1 =

z2 =

Y

Y: head of X

y

y, z: monosyllabic

The following example in (7) (with minor changes to Chan’s example) illustrates how the Head Dominance Condition works.

Phonological phrase in the Fuzhou dialect  129 S1

(7) NP

AP

A N 旧 书 kou242 tsy44 → ku44 = y44 # old book ‘Old books are very expensive.’

Adv 真 tsiŋ44 tsiŋ51 very

=

A 贵 kui213 ŋui213 expensive

As we can see in (7), 书 ‘book’ is the head of NP, and 旧 ‘old’ is a monosyllabic word dominated by the first node to the left of the head N. Similarly, 贵 ‘expensive’ is the head of AP, and 真 ‘very’ is a monosyllabic word dominated by the first node to the left of the head A. Hence, TS applies between 旧 ‘old’ and 书 ‘book’ as well as between 真 ‘very’ and 贵 ‘expensive’. In contrast, TS is blocked between 书 ‘book’ and 真 ‘very’ because the syntactic relationship between them does not accord with either (6a) or (6b). Chan’s Head Dominance Condition can account for the application of TS between a monosyllabic pre-head modifier and its head, as well as the blocking of TS between a monosyllabic head and its post-head complements. However, her analysis runs into difficulty since the Head Dominance Condition is too powerful—it wrongly excludes the possibility of the application of TS within all the verb-object constructions. This can be illustrated in (8), where the object NP 饭 ‘rice’ is a complement following the head V 食 ‘eat’, but the head V still undergoes TS. (8)

VP V’ NP2

V 食

饭 5

si → si 21 = *si 5 # eat ‘to eat food’

puoŋ242 puoŋ242 puoŋ242 rice

(predicted by Chan 1980)

Since Chan’s analysis excludes the possibility of the application of TS within all the verb-object constructions, it is impossible for her analysis to account for the contrast presented in (9) and (10)—the Head Dominance Condition works perfectly for the blocking of TS between the verb and the object

130  Phonological phrase in the Fuzhou dialect in examples like (10), but it obviously cannot account for the application of TS in examples like (9). (9)

a. →

(10) a. →

b.

趁 钱 theiŋ213 tsiɛŋ51 thiŋ44 = tsiɛŋ51 earn money ‘to earn money’

→

浪费 louŋ242 xiɛ213 louŋ51 xiɛ213 # waste ‘to waste money’

b.

钱 tsiɛŋ51 tsiɛŋ51 money

→

食 葡萄 siɛʔ5 pu51 to51 31 siɛʔ = pu31 lo51 eat grape ‘to eat grapes’ 讨厌 tho31 ʔiɛŋ213 tho44 ʔiɛŋ213 # hate ‘to hate grapes’

葡萄 pu51 to51 pu31 lo51 grape

Another problem of Chan’s analysis, as noticed by some subsequent studies (e.g., Hung 1987; Chan 1998), is its failure to explain the behavior of TS in the construction of determiner-classifier-noun. The example in (11) is adapted from Hung (1987). (11)

NP DP D 只 tsi31 → tsi44 = *tsi44 = this ‘this person’

N Cl 隻 tsi i i Cl

23 23 44

# =

侬 nøyŋ51 nøyŋ51 nøyŋ51 (predicted by Chan 1980) people

In (11), according to the NP structure assumed by Hung (1987), 侬 ‘person’ is the head of NP, and daughters of the first node to the left of the head N are both monosyllabic. Hence, Chan’s Head Dominance Condition fails to predict the blocking of TS between the classifier and the noun. Instead, Chan’s analysis will wrongly predict that TS should apply to both the first two syllables in (11). According to the DP theory that determiners are not inside the NP but rather the NP is the complement to the determiner head D (Abney 1987), the current tree of a determiner-classifier-noun construction differs from that in (11), as shown in (12).3

Phonological phrase in the Fuzhou dialect  131 (12)

DP D’ D

ClP Cl’ Cl

只 隻 tsi 23 tsi31 → tsi44 = i 23 *tsi31 # tsi 23 this Cl ‘this person’

NP

# #

侬 nøyŋ51 nøyŋ51 nøyŋ51 people

(predicted by Chan 1980)

We can find that even when the Fuzhou determiner-classifier-noun construction is updated to accommodate the DP hypothesis, incorrect predictions still persist. In (12), Chan’s analysis fails not because it cannot account for the blocking of TS between the classifier and the noun, but because it has difficulty explaining the application between the head of DP and the head of its complement ClP, namely between the determiner and the classifier. 5.1.2.2  Wright (1983) The contrast between (1c) and (2b) and the contrast between (9) and (10) show that TS seems to be blocked between a disyllabic verb and the following object but triggered between a monosyllabic verb and the object. Wright (1983) proposes a prosodic analysis based on this observation and claims that the asymmetry exhibited by the monosyllabic verb and the disyllabic verb can be explained with regard to different prosodic structures involved in the formation of a tone sandhi domain. Based on the theory developed in Liberman  & Prince (1977), Prince (1983), Hayes (1980), and Selkirk (1980b), she divides a Fuzhou sentence into four levels of prosodic domains, as in (13). (13) Phonological phrase Super-foot Foot Syllable In Wright’s analysis, a phonological phrase is defined in such a way that the end of a phonological phrase coincides with the end of a noun phrase (NP) or a clause

132  Phonological phrase in the Fuzhou dialect (S). She argues that a weak-strong (w-s) binary foot must be constructed within a phonological phrase and must be built above syllable strings from right to left at the lexical level first and then the phrasal level. A remaining unfooted syllable to the left of a binary foot will be marked with ‘w’ and form a super-foot with the binary foot, while an unfooted syllable to the right of a binary foot will form an independent foot by itself. Thus, given a string of syllables, examples in (14a) are well-formed, while those in (14b) are not (Σ = foot, Σ’ = super-foot, σ = syllable, w = weak, s = strong). (14) a. (14)  

Σ w σ

b.

Σ’ s σ

*Σ s σ

w σ

Σ s

w σ

w σ * Σ’

s σ

s σ

s σ

w σ

Σ s σ

σ

w w σ

In Wright’s analysis, it is the foot or the super-foot that forms the domain of application for TS at the phrasal level. The examples in (15) illustrate how her analysis works, in which “[. . .]” indicates boundaries of a phonological phrase, and “ (. . .) ” indicates a foot or a super-foot (i.e., a tone sandhi domain). (15) (15) a.

Σ

Σ’ s

b.

w 伊 食 [ i44]NP [si 5 → ( 44 i ) # (si 21 = he eat ‘He ate chicken eggs.’ Σ Σ

w 鸡 ki 44 ki 51 = chicken

w 伊 看 [ i44]NP [khaŋ213 → ( 44 i ) # (khaŋ51 = he see ‘He saw the person.’

s 见 ki ŋ213 ŋi ŋ213) #

s 卵 louŋ242]NP, S louŋ242) egg Σ

侬 nøyŋ51]NP, S (nøyŋ51) people

Phonological phrase in the Fuzhou dialect  133 From the above examples, we can find two advantages of Wright’s analysis. First, by defining the phonological phrase by marking the end of an NP or S, her analysis avoids joining the subject and its predicate into one foot and thus accounts for the blocking of TS between the subject and the predicate. Second, her analysis nicely captures the contrast between disyllabic verbs and monosyllabic verbs, as can be seen from the behavior of TS in 看见侬 ‘see the person’ and 食鸡卵 ‘eat chicken eggs’. Nevertheless, there are some outstanding problems with Wright’s analysis. Notice that syntactic information is heavily relied on in Chan (1980) while the syntactic information used for the formation of the phonological phrase in Wright (1983) is very simple. This simplicity makes her analysis run into difficulty when dealing with constructions other than subject-predicate and verb-object, for example in (16). (16) (16)

Σ’ s w w s 跳 野 悬 ai 31 keiŋ51]S [thiu213 → thiu213 # ia21 = eiŋ51 h 21 21 *(t iu = ia = eiŋ51) jump very high ‘(somebody) jumps very high’

(predicted by Wright 1983)

We can find that Wright’s analysis is not able to account for the blocking of TS between 跳 ‘jump’ and 野悬 ‘very high’ in (16). In order to handle such constructions, Wright (1983) is forced to argue that a verb will undergo TS only if it is followed by an argument with theta-role, but not with other theta-roles such as , , or . She claims that this explanation helps solve the problem posed by (16). However, it actually does not solve the problem since a syntactic constituent like 野悬 ‘very high’ in (16) is an adjunct instead of an argument to the head verb. Besides, this explanation is an ad hoc solution and makes Wright’s whole analysis suspicious since the rest of her analysis is syntax-blind while only a sub-part of it is forced to rely on syntactic and semantic information. Wright’s analysis also has a theoretical drawback from the perspective of the prosodic phonology theory. In the four-level prosodic hierarchy assumed by Wright, there are two levels of the foot domain (i.e., regular foot and super-foot), while the prosodic word domain is not found between the foot and the phonological phrase. From (15a) and (15b), we can find that both 食鸡卵 ‘to eat chicken eggs’ and 看见 ‘to see’ are treated as only one TS domain according to Wright’s analysis because the super-foot and the foot are both considered as the domain of TS. According to the discussion in Chapter 3, 看见 ‘to see’ is a morphosyntactic

134  Phonological phrase in the Fuzhou dialect word in Fuzhou and forms a prosodic word domain on its own, while 食鸡卵 ‘to eat chicken eggs’ must form a prosodic domain larger than the prosodic word as it contains two prosodic words 食 ‘to eat’ and 鸡卵 ‘chicken egg’. Wright’s analysis thus actually mixes up the difference between the prosodic word domain and the phonological phrase domain with the difference between the foot and the super-foot. Moreover, as argued in Chapter 2, Wright’s theory that argues for the existence of the foot domain in the Fuzhou dialect is ultimately problematic, which makes her analysis of the phrasal-level application of Fuzhou TS even less convincing. 5.1.2.3  Shih (1986) Being convinced that neither syntactic nor prosodic properties alone can solve the issue of Fuzhou phrasal tone sandhi domain, Shih (1986) takes an approach that involves both syntactic and prosodic information. Based on Chan’s (1980) Head Dominance Condition, Shih proposes the Revised Head Dominance Condition, which defines where TS is blocked, rather than triggered, in the Fuzhou dialect, as presented in (17). (17) Revised Head Dominance Condition (HDC) (Shih 1986) Mark the right edge of every X0, except where XP is an adjunct. As mentioned in Chapter 2, Shih argues that, after the Revised Head Dominance Condition inserts a tone group boundary to the right of the head of every nonadjunct XP, the Foot Formation Rule, as presented in (18), operates within each tone group to construct prosodic feet and super-feet. (18) Foot Formation Rule (FFR) (Shih 1986) Foot Construction a. Immediate Constituency (IC): link immediate constituents into disyllabic feet. b. Duple Meter (DM): scanning from left to right, string together unpaired syllables into binary feet, unless they branch to the opposite direction. Super-foot Construction Join any leftover monosyllable to a neighboring binary foot according to the direction of syntactic branching. Similar to Wright (1983), Shih (1986) also assumes that the foot and the super-foot form the domain of TS. The examples in (19) adapted from Shih (1986) illustrate how Shih’s analysis works, in which “. . .] ” denotes the right boundary of a tone group, and “ (. . .) ” denotes a foot or a super-foot (i.e., a tone sandhi domain).

Phonological phrase in the Fuzhou dialect  135 (19) a.

丁 先生 住 福州 tiŋ44 siŋ44 saŋ44 tiu242 xuʔ23 tsiu44 HDC: tiŋ44 siŋ44 saŋ44] tiu242] xuʔ23 tsiu44 44 44 44 242 FFR: (tiŋ siŋ saŋ )] tiu ] (xuʔ23 tsiu44 21 44 44 242 → (tiŋ = siŋ naŋ )] # tiu ] # (xuʔ21 tsiu51 Ding Mr. live Fuzhou ‘Mr. Ding lives on Fuzhou Road.’ b. 丁 先生 住 福州 tiŋ44 siŋ44 saŋ44 tiu242 xuʔ23 tsiu44 44 44 44 242 HDC: tiŋ siŋ saŋ ] tiu ] xuʔ23 tsiu44 44 44 44 242 FFR: (tiŋ siŋ saŋ )] tiu ] (xuʔ23 tsiu44) 21 44 44 242 → (tiŋ = siŋ naŋ )] # tiu ] # (xuʔ21 tsiu44) # Ding Mr. live Fuzhou ‘Mr. Ding lives on Fuzhou East Road.’

=

路 tuo242 tuo242] tuo242)] luo242)] road

东 tøyŋ44 tøyŋ44 (tøyŋ44 (tøyŋ51 east

路 tuo242 tuo242] tuo242)] nuo242)] road

According to Shih’s Revised Head Dominance Condition, a pre-head adjective and a pre-head adverb, which are heads of AP and AdvP respectively, are not marked off by a tone group boundary since AP and AdvP are both adjuncts. Thus, the adjunct AP or AdvP is joined into the same tone group with the following head, which nicely accounts for the application of TS in the following examples in (20). (20) a. (20)

NP

b.

N’

AP A’

AP

N’

AdvP

A’

A’

N

Adv’

A

A 旧 书 kou242 tsy44 → ku 44 = y44 old book ‘old book’

Adv 真 贵 tsiŋ44 kui213 → tsiŋ51 = ŋui213 very expensive ‘very expensive’

Shih’s analysis can also correctly predict the blocking of TS between the head and the post-head adjunct, such as 跳野悬 ‘(somebody) jumps very high’ in (16)— the right edge of the head verb is marked by the tone group boundary, and hence the TS rule is not triggered between the head verb and the post-head adjunct. However, Shih’s analysis wrongly predicts the blocking of TS between the ­determiner/numeral and the classifier, since the right edge of the determiner/numeral must be marked as a tone group boundary according to (17), as exemplified in (21).

136  Phonological phrase in the Fuzhou dialect (21) a. (21)

DP D’ D

ClP Cl’ Cl

NP

只 隻 tsi 23 tsi31 → tsi44 = i 23 # *tsi31 # i 23 # this Cl ‘this person’ b. NumP

侬 nøyŋ51 nøyŋ51 nøyŋ51 people

(predicted by Shih 1986)

Num’ Num

ClP Cl’ Cl

蜀 隻 tsi suo 5 → suo 21 = i *suo 5 # i one Cl ‘one person’

NP

23 23 23

# #

侬 nøyŋ51 nøyŋ51 nøyŋ51 people

(predicted by Shih 1986)

Moreover, like Chan’s (1980) analysis, Shih’s analysis also fails to account for the application of TS between a monosyllabic verb and its object, e.g., 食饭 [siɛʔ5 puoŋ242 → siɛʔ21 puoŋ242] ‘to eat food’, since the right edge of the head V is marked as a tone group boundary according to (17). In order to deal with such examples, Shih (1986) has to propose an expansion rule to expand the domain of tone group to include verb and its object, as shown in (22). (22) Tone Group Expansion (Shih 1986) V # O → V = O This expansion rule makes Shih’s analysis self-contradictory in terms of the explanation of TS within VP constructions—without the expansion rule, Shih is not able to account for the application of TS between a monosyllabic verb and its object, while the existence of this additional rule indicates that rule (17) always

Phonological phrase in the Fuzhou dialect  137 has exceptions. Shih (1986) leaves unexplained the issue of when the expansion rule applies, which makes this additional stipulation more arbitrary. Furthermore, in order to account for the application of TS at the phrasal level, feet and super-feet in Shih’s analysis are constructed by making use of syntactic information such as immediate constituents and the direction of syntactic branching, which is actually on the wrong track, as I have argued in Chapter 2. Despite its weaknesses, Shih (1986) relates the distinction between adjunct constituents and non-adjunct constituents to the application of Fuzhou TS at the phrasal level, which has been assumed in several subsequent works (e.g., Hung 1987; Zhang 1992, among others). As we will see in our alternative analysis in Section 5.1.4, the adjunct/non-adjunct distinction would be a key factor in defining the domain of phrasal-level application of TS in the Fuzhou dialect. 5.1.2.4  Hung (1987) Similar to Shih (1986), Hung (1987) also observes that there is an asymmetrical status of adjuncts (modifiers) and arguments. Hung points out that a pre-head modifier in the Fuzhou dialect is always incorporated into a tone group with the head while a pre-head argument is not. By contrast, in post-head environments, there is a reverse situation—a post-head modifier never undergoes TS with its head while a post-head argument does. Hung’s observation is schematized by Chen (1990: 42), as provided in (23). (23) Hung’s (1987) observation (also cf. Chen 1990) a. adjunct = head b. argument # head c. head = argument d. head # adjunct Hung argues that only when two constituents stand in the “adjunct-head” or the “head-argument” relationship can they form sense units (cf. Selkirk 1984), which is why TS only occurs in “adjunct-head” and “head-argument” constructions. Based on the above observation, Hung defines the tone sandhi domain in Fuzhou as in (24). (24) Fuzhou tone sandhi domain (Hung 1987) XP [ . . . tg[(M*) X (A)]tg (M)]XP (where M = Modifier, A = Argument, XP = maximal projection of X (a lexical head), tg = tone group, () = optional, * = any arbitrary number) In addition to this syntactic-semantic condition, Hung also points out that there should be a prosodic constraint in Fuzhou tone sandhi, as in (25). Furthermore, in agreement with Wright (1983) and Shih (1986), Hung proposes a similar Foot Formation Rule, as in (26). (25) Prosodic constraint on Fuzhou external tone sandhi (Hung 1987) The constituent(s) of a tone group to the left of a determinant is/are monosyllabic.

138  Phonological phrase in the Fuzhou dialect (26) Fuzhou Foot Formation Rules (Hung 1987)4 a. Link the syllables in polysyllabic lexical items into freely structured feet. b. Scanning from left to right, link heads to their arguments to form disyllabic or right-branching feet. c. Scanning from left to right, link modifiers to their heads to form disyllabic or right-branching feet. Hung’s analysis, especially his distinction between adjuncts (modifiers) and arguments can explain the application of Fuzhou TS in a number of phrasal-level constructions, some of which are illustrated as in (27). (27) a.

NP

b.

N’

TP DP

AP

N’

D’

A’

N

D

A

T’ T

VP V’ V

旧 [adjunct] = kou242 → ku44 old ‘old book’ c. VP

书 [head] tsy44 y44 book

伊 食 [argument] # [head] = 44 i si 5 → i44 # si 21 = he eat ‘He ate food.’

V’ V’

AP

V

A’ AdvP

A’

A 跳 野 悬 [adjunct] = [head] ] [head] # [ adjunct thiu213 ai 31 keiŋ51 → thiu213 # ia21 = eiŋ51 jump very high ‘(somebody) jumps very high’

NP 饭 [argument] puoŋ242 puoŋ242 rice

Phonological phrase in the Fuzhou dialect  139 However, notice that it is unclear whether the “modifier” and “argument” in Hung’s definition in (24) refer to maximal projections or zero projections, which may lead Hung’s analysis into some empirical obstacles. Let us take a look at the example in (28) where the NP is modified by an AP. (28) (28)

VP V’ V

NP N’ AP

N’

N 食 大 葡萄 tuai242 pu51 to51 si 5 → si 5 # tuai51 = pu31 lo51 eat big grape ‘to eat big grapes’ Prediction 1 [adjunct] = [head] [head] = [ argument ] = pu31 lo51 *si 21 = tuai51 Prediction 2 [head] # [adjunct] = [head] si 5 # tuai21 = pu31 lo51

As we can see from (28), Hung’s analysis could have two predictions for this example. If the modifier and argument are interpreted at the maximal projection level, then TS is predicted to apply between the V 食 ‘eat’ and the NP 大葡萄 ‘big grapes’ since they stand in the relationship of “head-argument”, as shown in Prediction 1, which is an incorrect prediction. We could give Hung the benefit of the doubt and assume that the modifier and argument are interpreted at the zero projection level, as shown in Prediction 2, which is a correct prediction. However, one can find that Prediction 2 can hardly serve as a convincing example to support Hung’s analysis since the AP 大 ‘big’ in this case is the adjunct to the head of NP, namely 葡萄 ‘grape’, rather than the adjunct to the head of VP, namely 食 ‘eat’. Thus, Prediction 2 runs against the spirit of syntax since one is forced to allow the function of part of a constituent to supersede the function of the entire constituent. By the same token, Hung’s analysis fails to account for the blocking of TS between the verb and the DP/NumP object in (29). Moreover, since the term “argument” in syntax usually refers to the constituent that functions as the subject or the object and completes the meaning of a predicate, the adjunct/argument distinction in Hung’s analysis is not able to deal with the relationship among determiner/numeral, classifier, and head noun. Hence, his analysis also fails to account for the blocking of TS between the classifier and the NP as well as the application of TS between the determiner/numeral and the classifier, which can be exemplified in (29) as well.

140  Phonological phrase in the Fuzhou dialect (29) (29) a.

VP V’ V

DP D’ D

ClP Cl’ Cl

b.

隻 食 只 tsi si 5 tsi31 → si 5 # tsi44 = i eat this Cl ‘to eat this grape’ VP

NP

23 23

#

葡萄 pu51 to51 pu31 lo51 grape

V’ V

NumP Num’ Num

ClP Cl’ Cl

食 蜀 隻 suo 5 tsi si 5 → si 5 # suo 21 = i eat one Cl ‘to eat one grape’

NP

23 23

#

葡萄 pu51 to51 pu31 lo51 grape

5.1.2.5  Zhang (1992, 2017) Zhang (1992, 2017) argues that, apart from the functional relations in Hung’s (1987) observation mentioned in (23), Fuzhou TS at the phrasal level is also sensitive to syntactic branching structure. To be specific, only a right-branching structure that bears functional relation “adjunct-head” or “head-argument” will form a tone sandhi domain, while a left-branching structure will turn out to have two domains. Different from Hung (1987), who takes the difference between right/

Phonological phrase in the Fuzhou dialect  141 left-branching structures as a prosodic constraint as in (25), Zhang argues that it is a syntactic condition, namely the c-command condition (defined by Reinhart 1976, 1983), as presented in (30). (30) C-command (Reinhart 1976, 1983) α c-commands β iff every branching node dominating α dominates β. Based on the assumption that functional categories are affected by c-command relation in terms of the application of TS at the phrasal level in the Fuzhou dialect, Zhang proposes the domain formation rule of phrasal TS in Fuzhou as in (31). (31) Phrasal TS domain formation (Zhang 1992, 2017) The phrasal tone sandhi rule is applied iteratively right-to-left to the syllable that is either the adjunct or the head of an argument when the syllable c-commands a following syllable. Zhang’s analysis nicely captures many cases of Fuzhou TS at the phrasal level with the domain formation rule in (31), as illustrated with the examples adapted from Zhang (1992, 2017) in (32). (32) a. (32) 书

真 贵 [adjunct] = [head] [argument] # [ head ] tsiŋ44 kui213 tsy44 → tsy44 # tsiŋ51 = ŋui213 book very expensive ‘Books are very expensive.’ b. 悬

六 尺 [adjunct] = [head] ] [head] # [ adjunct keiŋ51 løy 5 tshuo 23 → keiŋ51 # løy 21 = tshuo 23 high six foot ‘six-foot-high’ c. 食 鸡 卵 [head] = [ argument ] ki 44 louŋ242 si 5 → si 21 = ki 51 louŋ242 eat chicken egg ‘to eat chicken eggs’ d. 相 [ head suoŋ44

信 ]

seiŋ213

#

我 [argument] ŋuai31

[head] = [ argument ] ki 44 louŋ242 si 5 → si 21 = ki 51 louŋ242 eat chicken egg 142  Phonological phrase in the Fuzhou dialect ‘to eat chicken eggs’ d. 相 信 [ head ] # suoŋ44 seiŋ213 → suoŋ51 seiŋ213 # believe ‘to believe me’

我 [argument] ŋuai31 ŋuai31 me

In (32a), 真 ‘very’ is the adjunct and c-commands the head 贵 ‘expensive’, so it obtains a sandhi tone. 书 ‘book’ is the argument of the head and therefore cannot undergo TS. In (32b), 六 ‘six’ obtains a sandhi tone because it is the adjunct and c-commands the head 尺 ‘foot’. In contrast, although 悬 ‘high’ c-commands the following constituent, it stands in a head-adjunct relation with the following constituent, which is why it maintains its citation tone. (32c) and (32d) illustrate how Zhang’s analysis deals with the contrast between the monosyllabic verb and the disyllabic verb. In (32c), the monosyllabic verb 食 ‘eat’ c-commands the following constituent. In addition, 食 ‘eat’ and its object 鸡卵 ‘chicken eggs’ stand in a head-argument relation. Therefore, 食 ‘eat’ obtains a sandhi tone. In contrast, in (32d), according to Zhang, the object 我 ‘me’ is not c-commanded by the preceding syllable 信 ‘to believe’, and hence TS is blocked between the last two syllables, although the structure between the verb 相信 ‘to believe’ and the object 我 ‘me’ is a head-argument structure. Despite its strengths, Zhang’s analysis is not perfect. First, the definition in (31) needs refinements. Stating that “when the syllable c-commands a following syllable”, the definition in (31) seems to take the concept of c-command as a relationship between syllables but not between nodes of a syntax tree. In actual fact, if a syntactic constituent is composed of more than one syllable, it is the entire polysyllabic constituent instead of one of the syllables that serves as either the adjunct or the head of an argument. By stating that “the syllable which is either the adjunct or the head of an argument”, the definition in (31) has already implied that only monosyllabic constituent can undergo phrasal-level TS. Second, like Hung’s (1987) analysis, Zhang’s analysis also has difficulty in dealing with verb-object constructions like (28) and (29). On the one hand, since it is the “syllable” that plays an important role in Zhang’s analysis, we assume that the “adjunct” and the “argument” in his analysis refer to zero projection. Thus, Zhang’s analysis can account for cases like (28) in which TS is blocked between the head verb and the following adjunct of the NP. Nonetheless, as discussed in Section 5.1.2.4, this indicates that the function of part of a constituent supersedes the function of the entire constituent, which is open to question. On the other hand, like Hung’s analysis, the adjunct-argument distinction that is relied heavily on in Zhang’s analysis is also unable to deal with the relationship among determiner/numeral, classifier, and head noun in cases like (29). 5.1.2.6  Chan (1998) As mentioned in Chapter  2, Chan (1998) claims that the Fuzhou tone sandhi domain corresponds to the foot. She argues that, at the phrasal level, a phonological

Phonological phrase in the Fuzhou dialect  143 phrase must be constructed first by making use of syntactic information, and then feet are built within the phonological phrase by referring to the length and/or the position of the morphosyntactic word. We have seen in Chapter  2 that her definition of the foot in Fuzhou is actually problematic, so let us focus on her construction of the phonological phrase in this subsection. Chan’s construction of the phonological phrase is based on the idea of lexical government advanced by Hale & Selkirk (1987) and Lin (1994). The term “government” is defined as in (33) (cf. Chomsky 1981, 1982, 1986; Hale & Selkirk 1987, among others), based on which Lin (1994) proposes the algorithm of tone group formation in the Xiamen dialect of Chinese, as given in (34). (33) Government A governs B iff A m-commands B and every barrier for B dominates A.5 (34) Xiamen Chinese phrasing parameter (Lin 1994) ]Xmax, Xmax not lexically governed. Inspired by the approach advanced by Lin (1994), Chan argues for a lexicalgovernment-based analysis for Fuzhou phrasal TS domain. She proposes a Fuzhou Phonological Phrase rule, which inserts a boundary at the right edge of any X0 if its maximal projection XP is not lexically governed, as presented in (35). (35) Fuzhou Phonological Phrase (Chan 1998) {right, X0}, where XP is not lexically governed. Chan assumes that the adverbial modifier and the adjectival modifier are adjoined to V’ and N’ respectively and thus are located inside the domain of V0 and N0. Therefore, the adverbial modifier and the adjectival modifier are lexically governed by the verb and the noun respectively, and thus no phonological phrase boundary is inserted at the right edge of Adv0 and A0 according to (35). Hence, the TS rule is correctly predicted to apply between the adverbial modifier and the verb, as well as between the adjectival modifier and the noun, namely between the pre-head adjunct and the head, as shown in (36) (examples are adapted from Chan 1998). (36) a. (36)

VP

b.

NP

V’

N’

AdvP

V’

AP

N’

Adv’

V0

A’

N0

Adv0 齐 来 ts 51 li51 → ts 31 = li51 together come ‘to come together’

A0 蓝 laŋ51 → laŋ44 = blue ‘blue books’

书 tsy44 y44 book

144  Phonological phrase in the Fuzhou dialect In order to account for the behavior of TS in determiner/numeral-­classifiernoun constructions, Chan adopts Lin’s (1994) assumption of DP structure, as presented in (37) (QP = quantifier phrase, Q = quantifier). (37) (37) DP (1994) DPstructure structure in in Lin (1994) DP

D’ D

QP Q’ Q Num

NP Cl

With the DP structure in (37), Chan claims that the rule in (35) can deal with determiner/numeral-classifier-noun constructions that have posed difficulties for Hung (1987) and Zhang (1992). Two examples given by Chan are presented in (38). (38) (38) a. DP

b. V0

D’ D0

VP DP

QP Q0

Num

D’ D0

NP Cl

蜀 本 suo 5 puoŋ31 → suo 31= βuoŋ31 # one Cl ‘one book’

N’ N0 书 tsy44 tsy44 book

QP Q0

只 tsi31 → tsi24 = this ‘this book’

NP Cl

N’

本 puoŋ31 βuoŋ31 # Cl

N0 书 tsy44 tsy44 book

Chan argues that in both examples, the QPs are non-lexically governed by the D0, and therefore the right margin of their head Q0 is marked with a tonal domain boundary according to (35). In (38b), the DP is lexically governed by the V0, and thus, the right margin of D0 is not marked with a TS domain boundary. This analysis, at first sight, perfectly solves the problems caused by determiner/ numeral-classifier-noun constructions. A  closer examination, however, reveals

Phonological phrase in the Fuzhou dialect  145 that there are at least three problems in Chan’s analysis. First, it is not clear why, in both examples provided by Chan, Q0 appears in the specifier position of QP and Q’ is deleted from the tree. Second, the DP structure in (37) adopted by Lin (1994) and Chan (1998) is quite debatable because two terminal nodes Num and Cl are located directly under Q/Q0, which is another terminal node. This appears that Num and Cl form a compound that is located under Q0. The last and the most important problem is that Chan does not discuss whether Num and Cl have their own maximal projections and whether their maximal projections, namely NumP and ClP, are lexically governed or not. If they are both non-lexically governed by D0, then the right margin of Num and Cl should be marked with a tonal domain boundary. In that case, Chan’s analysis fails to account for the application of TS between Num and Cl. If they are both lexically governed by some nodes in the tree, then the right margin of neither Num nor Cl would be marked with a tonal domain boundary. In that case, Chan’s analysis fails to account for the blocking of TS between the classifier and the noun. In terms of VP constructions, Chan points out that the rule in (35) requires that the right edge of the head V of the VP must be marked with a tonal domain boundary since the VP is non-lexically governed by INFL, which is a functional category, as illustrated in (39). (39) (39)

IP I

VP V

#

XP

According to this analysis, the head V of the VP must be marked with a tonal domain boundary no matter what the post-verbal XP is. This nicely predicts the blocking of TS between the verb and the post-verbal adjunct. Nevertheless, when the verb is followed by an argument, two types of TS behavior of the verb are observed in Chan (1998). Compare the examples in (40). (40) a. (40)

VP

b.

VP

V’ V 食 si 5 → si 31 = eat ‘to eat grapes’

V’ NP 葡萄 pu51 to51 pu31 lo51 grape

V

NP

回 福州 xui51 xu 23 tsiu44 → xui51 # xu 21 tsiu44 return Fuzhou ‘to return to Fuzhou’

The blocking of TS in (40b) can be predicted by (39) while the application of TS in (40a) cannot. Thus, Chan is forced to introduce a distinction between

146  Phonological phrase in the Fuzhou dialect two types of argument NPs, namely [+theme] NP and [−theme] NP. She claims that when the argument NP bears a [+theme] theta-role, the verb and the argument NP form one incorporated noun through the process of Noun Incorporation (cf. Mithun 1984; Baker 1988). Thus, both the verb and the argument NP are dominated by the V0 node, and hence the tonal domain boundary is inserted at the right edge of V0 that coincides with the right edge of the NP, but not inserted between the verb and the argument NP. In contrast, if the argument NP is marked as [−theme], it forms a syntactic phrase with the verb, and thus, a tonal domain boundary is inserted between the verb and the NP according to (39), which causes TS to be blocked. Chan’s schematization of these two types of V-NP constructions is presented in (41). (41) a. (41)

V0 V

=

b. NP[+theme]

V’ V0

#

NP[-theme]

This analysis, which treats the V-NP[+theme] as an incorporated noun constructed through Noun Incorporation while V-NP[−theme] as a syntactic phrase, seems to account for the contrast between the two types of TS behavior of the verb in (40). However, such an analysis is somewhat ad hoc and counter-intuitive. The Fuzhou dialect, like many other Chinese dialects, has almost no overt morphology and hence is very different from polysynthetic languages such as Mohawk and Southern Tiwa (cf. Baker 1988) in which unique affixation patterns can be used to distinguish an incorporated noun from its corresponding syntactic phrase. Therefore, it is very difficult to figure out whether a direct object in the Fuzhou dialect participates in the process of Noun Incorporation or not. Moreover, according to Chan’s analysis, the structural position of nouns in the syntactic tree is actually determined by the theta-roles assigned to the nouns. Chan’s analysis recalls Baker’s (1988) Uniformity of Theta Assignment Hypothesis (UTAH), which states that identical thematic relationships between items are represented by identical structural relationships between these items at the level of D-structure and thus associates each theta-role with a constant structural position. The UTAH, however, has been argued to be at odds with evidence across languages (see Li 2005 and Borer 2005 for problems with and arguments against the UTAH; also cf. Hale & Keyser 1993, 2002, among others, for a different approach to the correspondence between theta-role assignment of arguments and the syntactic positions of arguments). Therefore, the analysis in (41) is a theoretical pitfall in Chan’s theory, although it seems to nicely capture the difference in TS behavior exhibited by different types of argument NPs. Also worth noting is that the notion of “lexical government”, as a key concept in Chan’s analysis, is rooted in the assumptions of the Government and Binding Theory (Chomsky 1981, 1982). The notion of “government”, nonetheless, has been eliminated from the syntactic theory in the Minimalist Program (Chomsky 1993, 1995, 2000), which further undermines the validity of Chan’s analysis.

Phonological phrase in the Fuzhou dialect  147 5.1.2.7 Summary In Section 5.1.2, I have reviewed the most representative previous studies concerning the issue of Fuzhou TS at the phrasal level. Starting from Chen & Norman (1965a), several different analyses have been proposed to deal with this issue. Some of them heavily rely on syntactic notions, e.g., Chan (1980) (Head Dominance Condition) and Zhang (1992) (adjunct/argument distinction and c-command); some of them mainly employ prosodic notions, e.g., Wright (1983) (weak-strong foot and super-foot); and some of them combine both syntactic and prosodic notions, e.g., Shih (1986) (Revised Head Dominance Condition and Foot Formation Rule), Hung (1987) (adjunct/argument distinction and Foot Formation Rule), and Chan (1998) (Lexical Government and weak-strong foot). Semantic notions such as theta-roles are also involved in some analyses, e.g., Wright (1983) and Chan (1998). All these analyses successfully handle some cases of Fuzhou TS at the phrasal level, but all of them run into some empirical and/or theoretical obstacles. Therefore, to better account for the issue of Fuzhou TS at the phrasal level, an alternative analysis is required. 5.1.3  Which is a better solution: RBA or EBA? From the discussion on previous studies in Section 5.1.2, we can find that the key to dealing with the issue of Fuzhou TS at the phrasal level is to define the phrasal-level domain of application for this rule. As we can see from all the relevant examples presented in Section 5.1.2, the so-called “tone group” within which TS applies is always composed of more than one prosodic word in the Fuzhou dialect. Thus, we can assume that the phrasal-level domain of Fuzhou TS must be a prosodic domain larger and located higher than the prosodic word in the prosodic hierarchy, as I have also argued in Chapter 2. Since the relevant examples of phrasal-level Fuzhou TS we have seen so far do not contain any clitic elements, the domain of application for Fuzhou TS at the phrasal level is not the clitic group. Moreover, these examples at the phrasal level, such as 食葡萄 ‘to eat grapes’ and 齐来 ‘to come together’, are not related to intonation contours. Hence, the domain formed by such constructions is not the intonational phrase either. Therefore, the prosodic constituent that can serve as the domain of application for phrasal-level Fuzhou TS must be the phonological phrase. Thus, the key to solving the notorious problem raised by Fuzhou TS at the phrasal level is to define the phonological phrase domain in this dialect. As reviewed in Chapter 1, two major approaches, namely the Relation-Based Approach (RBA) and the Edge/End-Based Approach (EBA; including the original parameterized algorithms, the Align/Wrap Theory, and the Match Theory), have been developed within the framework of prosodic phonology. These two approaches differ from each other with respect to the definition of the phonological phrase since they employ different types and amounts of syntactic information in the formation of the phonological phrase domain. Both approaches have been adopted by many linguists and have successfully accounted for a number of phonological phenomena related to the phonological phrase across languages. In this

148  Phonological phrase in the Fuzhou dialect section, I am going to explore whether these two major approaches to phonological phrasing can provide a satisfactory solution to the long-standing problem of the phrasal-level application of TS in Fuzhou. I will show that neither of these two approaches alone can define the phonological phrase in this dialect. 5.1.3.1  RBA to phonological phrasing in the Fuzhou dialect The definition of the phonological phrase advanced by the RBA is given in (42), in which the lexical head X only refers to V, N, and A according to Nespor & Vogel (1986). As I have pointed out in Chapter 1, the clitic group status is assigned to every prosodic word in this definition in accordance to the original SLH, and the “clitic group” here is actually better to be understood as “clitic group or prosodic word” if we assume a weakened SLH. (42) Phonological phrase (φ) formation (Nespor & Vogel 1986) The domain of φ consists of a clitic group that contains a lexical head (X) and all clitic groups on its non-recursive side up to the clitic group that contains another head outside of the maximal projection of X. In addition, an optional rule for restructuring φ is proposed, as presented in (43), which has the effect of eliminating non-branching φs. (43) φ-restructuring (optional) (Nespor & Vogel 1986) A non-branching φ, which is the first complement of X on its recursive side, is joined into the φ that contains X. Although some notions in (42) and (43), e.g., head, complement, branching, and maximal projection, have been employed in some previous studies we have reviewed in Section 5.1.2, we have not seen any previous studies dealing with the issue of Fuzhou phrasal TS strictly following the mapping algorithm of RBA. If we adopt the RBA definition of the phonological phrase in (42) and assume that the restructuring rule in (43) is obligatory in the Fuzhou dialect, we can find that a number of Fuzhou data concerning phrasal-level TS can be accounted for, as illustrated in (44)–(47). (44) Modifier-Head (Attributive-Noun and Adverbial-Adjective/Verb) (44) a.

NP

b.

N’ AP

AP

c.

A’ N’

N [旧 书]φ [kou242 tsy44] → [ku44 = y44] old book ‘old book’

AdvP

VP V’

A’

A [真 贵]φ [tsiŋ44 kui213] → [tsiŋ51 = ŋui213] very expensive ‘very expensive’

AdvP

V’

V [齐 来]φ [ts 51 li51] → [ts 31 = li51] together come ‘to come together’

Phonological phrase in the Fuzhou dialect  149 (45) Subject-Predicate (45)     (46) (46) Verb-Resultative complement TP

NP

T’

N’ AP

VP

T N’ N

V’ AP

V’

AP

A’

V

A’

AdvP

A’

AdvP

A [真 贵]φ [旧 书]φ [kou242 tsy44] [tsiŋ44 kui213] → [ku44 = y44] # [tsiŋ51 = ŋui213] old book very expensive ‘Old books are very expensive.’ (47) Verb-Object (47)

VP

a.

A [野 悬]φ [跳]φ [thiu213] [ ia31 keiŋ51] → [thiu213] # [ ia21 = eiŋ 51] jump very high ‘(somebody) jumps very high’

b.

VP

V’ V

V’ NP

V

N’ AP

A’

NP N’

N’

N [食]φ [大 葡萄]φ [si 5] [tuai242 po51 to51] → [si 5] # [tuai51 = po31 lo51] eat big grape ‘to eat big grapes’

N [食]φ [葡萄]φ [食 葡萄]φ (restructuring) [si 5 pu51 to51] → [si 31 = pu31 lo51] eat grape ‘to eat grapes’

From (44)–(47), we can find that the phonological phrase defined with the RBA forms the domain of application for the TS rule. By defining the phonological phrase domain with the formation rule in (42), we actually group the pre-head adjunct and the head into one phonological phrase since they are both inside the maximal projection of the head, as in (44). This rule also separates the pre-head argument from the head since the pre-head argument usually belongs to a maximal projection that is outside of the maximal projection of the head, as in (45). Furthermore, this rule separates the post-head adjunct from the head since the post-head adjunct is on the head’s recursive side in Fuzhou, as in (46). As for the post-head argument, we separate the head and a branching complement while link the head and a non-branching complement by adopting the φ-restructuring rule in (43), as in (47).

150  Phonological phrase in the Fuzhou dialect However, the analysis based on the RBA is not perfect, and it fails to account for the behavior of TS in three types of phrasal-level constructions. First, since only V, N, and A are viewed as lexical heads in (42), nothing in the RBA-based analysis can deal with the behavior of TS within determiner/numeral-classifiernoun constructions, such as 只隻葡萄 [tsi31 tsiɛʔ23 pu51 to51 → tsi44 = ʒiɛʔ23 # pu31 lo51] ‘this grape’ and 蜀隻葡萄 [suoʔ5 tsiɛʔ23 pu51 to51 → suoʔ21 = ʒiɛʔ23 # pu31 lo51] ‘one grape’. In addition, the RBA also fails to explain the blocking of TS between a disyllabic verb and its non-branching object NP, such as 浪费钱 [louŋ242 xiɛ213 tsiɛŋ51 → louŋ51 xiɛ213 # tsiɛŋ51] ‘to waste money’ and 讨厌葡萄 [tho31 ʔiɛŋ213 pu51 to51 → tho44 ʔiɛŋ213 # pu51 lo51] ‘to hate grapes’, since the φ-restructuring rule will always group the disyllabic verb and the non-branching object NP into one phonological phrase. Furthermore, the RBA-based analysis cannot handle the blocking of TS between a verb and an NP indicating the location (“[−theme] argument” in Chan’s 1998 terminology). For example, in 回福州 [xui51 xuʔ23 tsiu44 → xui51 # xuʔ21 tsiu44] ‘to return to Fuzhou’, the verb should be incorporated into one phonological phrase with the following NP 福州 ‘Fuzhou’ by the φ-restructuring rule, which means that TS should be triggered in this case. This is clearly an incorrect prediction. 5.1.3.2  EBA to phonological phrasing in the Fuzhou dialect Let us now move on to the EBA and begin with its original parameterized algorithms. As we have seen in Chapter 1, the phonological phrase is defined in the way that the right or left edge of any phonological phrase coincides with the corresponding edge of a head or a maximal projection according to the EBA. End parameter settings of phonological phrasing proposed within the EBA are given in (48). (48) End parameter settings for the phonological phrase (Selkirk 1986) (I)

a.

]Xmax

b.

[

Xmax

(II)

a.

]Xhead

b.

[

Xhead

There are several EBA-like analyses in previous studies on Fuzhou phrasallevel TS, e.g., Chan (1980), Shih (1986), and Chan (1998). From the discussion in Section 5.1.2, we can find that Chan (1980), Shih (1986), and Chan (1998) all mark the right edge of X as the phrasal-level TS domain boundary, although they place different restrictions on the XP. Their analyses can thus all be interpreted as employing (different versions of) the end-setting ]Xhead. As shown in Section 5.1.2, these analyses have some similar problems, among which the most outstanding one is their failure to explain the application of TS between a monosyllabic verb and the following argument. To incorporate the monosyllabic verb into the same TS domain with the following argument, one has to propose an ad hoc condition or stipulation, such as Shih’s (1986) expansion rule and Chan’s (1998) Noun Incorporation analysis. Since the end-setting ]Xhead always marks the right edge of the head V, the analyses based on ]Xhead is also bound to fail when dealing with the

Phonological phrase in the Fuzhou dialect  151 contrast between monosyllabic verbs and disyllabic verbs, if no additional condition or stipulation exists. The end-setting Xhead[ is not a good choice for the issue of Fuzhou TS at the phrasal level either. Since it defines the phonological phrase by marking the left edge of a head, the end-setting Xhead[ not only fails to explain the possible application of TS between a monosyllabic verb and its object noun, but also wrongly predicts that TS would be blocked between a pre-head adjunct constituent and the head. As we have seen in Section 5.1.2, a pre-head adjunct constituent is usually included in the same tone sandhi domain with the head in the Fuzhou dialect, and that is why the left edge of the head should not be marked with a phonological phrase boundary in this dialect. If the phonological phrase in Fuzhou is defined with the end-setting ]Xmax, similar to the domain of tone sandhi in Xiamen and the domain of stress assignment in ChiMwiini (cf. Selkirk 1986), more problems would arise. First, since ]Xmax invariably groups everything within the same VP, it is impossible to account for the blocking of TS between (a) a monosyllabic verb and a branching object, (b) a disyllabic verb and its object, and (c) a monosyllabic verb and a resultative complement. The second problem is how to deal with the application of TS in modifier-head constructions (attributive-noun and adverbial-verb/adjective). The end-setting ]Xmax inserts a phonological phrase boundary on the right edge of an AP or an AdvP, and thus we are not able to explain the application of TS between the adjunct AP/AdvP and the head. The last end-setting Xmax[ has difficulty dealing with the phonological phrasing in Fuzhou as well. By assuming Xmax[, we mark the left edge of every XP as the left edge of a phonological phrase. Thus, the left edge of an object NP will coincide with the left edge of a phonological phrase, and the NP is never incorporated into a domain with the preceding verb. By doing so, the possibility of application of TS in some verb-object constructions is again wrongly excluded. To sum up, we have seen that no matter which of the four end parameter settings for phonological phrasing proposed in EBA is assumed in the definition of Fuzhou phonological phrase, there are always some cases of Fuzhou phrasal TS that cannot be covered. This shows that the original parameterized algorithms of EBA are not more effective than the RBA-based analysis. Let us proceed and see whether the recent development of the EBA within the framework of OT, namely the Align/Wrap Theory and the Match Theory, can help address the issue of Fuzhou TS at the phrasal level. We have seen in Chapter 1 that the parametric formulation of the EBA is recast into a group of violable constraints in the Align/Wrap Theory, such as ALIGN-XP (ALIGN-XP,L and ALIGN-XP,R), WRAP-XP, and NONREC (Selkirk 1996; Truckenbrodt 1995, 1999). It has been argued that these constraints interact in terms of the phonological phrasing and that differences in the ranking of these constraints give rise to the cross-linguistic variation in the formation of the phonological phrase domain (Truckenbrodt 1995, 1999). The Align/Wrap Theory, however, cannot solve the problem posed by Fuzhou TS at the phrasal level. Take the verb-resultative complement construction as

152  Phonological phrase in the Fuzhou dialect an example. We have seen that TS is always blocked between the verb and the resultative complement, which indicates that the verb and the resultative complement must form two separate phonological phrases in the Fuzhou dialect. In other words, given a verb-resultative complement construction [V [AP]]VP, the correct result of phonological phrasing must be [V]φ [AP]φ. Nonetheless, in an Align/Wrap Theory-based analysis, if the constraint ALIGN-XP,R is adopted, we will inevitably select the incorrect optimal candidate [V AP]φ, whether ALIGN-XP,R is ranked higher than WRAP-XP or lower.6 As shown in (49), the candidate (a) violates WRAP-XP, while the candidate (b) fulfills both constraints and hence will be selected as the optimal candidate, which is in actual fact an incorrect prediction. (49)

[V [AP]]VP

ALIGN-XP,R

a. [V]φ [AP]φ

WRAP-XP *!

*  b. [V AP]φ If we adopt the constraint ALIGN-XP,L, we then need to rank ALIGN-XP,L higher than WRAP-XP to select the correct optimal candidate, as can be seen in (50). (50)

[V [AP]]VP

ALIGN-XP,L

 a. [V]φ [AP]φ b. [V AP]φ

WRAP-XP *

*!

Nevertheless, the constraint ranking ALIGN-XP,L >> WRAP-XP will run into difficulty when dealing with the case of verb-object constructions in Fuzhou, since an Align/Wrap Theory-based analysis is not able to distinguish monosyllabic verbs and disyllabic verbs. With such a constraint ranking, we will always select the phonological structure [V]φ [NP]φ as the optimal candidate for the input syntactic structure [V [NP]]VP , as shown in (51), but this is clearly not the case for examples like 食饭 [siɛʔ5 puoŋ242 → siɛʔ21 puoŋ242] ‘to eat food’. (51)

[V [NP]]VP

ALIGN-XP,L

 a. [V]φ [NP]φ b. [V NP]φ

WRAP-XP *

*!

To handle the different TS behavior between monosyllabic verbs and disyllabic verbs, one has to propose additional restrictions or conditions in the Align/Wrap Theory-based analysis, which shows that the Align/Wrap Theory is not superior to the original parameterized algorithms of EBA in terms of Fuzhou phrasal-level TS. Now, let us apply the Match Theory to the case of Fuzhou phrasal-level TS and see if it provides a better solution to the phonological phrasing in Fuzhou. As we have seen in Chapter 1, according to the Match Theory, the phonological phrasing in languages is obtained through the interaction of the faithfulness constraint Match (Phrase, φ), which calls for any syntactic phrase to have a corresponding

Phonological phrase in the Fuzhou dialect  153 phonological phrase, with the markedness constraint BinMin (φ, ω), which requires that a phonological phrase should be minimally composed of two prosodic words (Selkirk 2000, 2005, 2009, 2011). In the case of Fuzhou TS at the phrasal level, we have seen that TS applies between the pre-head adjunct and the head, which means that the pre-head adjunct and the head must belong to the same phonological phrase in the Fuzhou dialect. Hence, given a modifier-head construction like [[AP] N]NP , the correct phonological phrasing must be [AP N]φ. Thus, we have to rank Match (Phrase, φ) lower than BinMin (φ, ω) in order to select the correct optimal candidate, as illustrated in (52). (52)

[[AP] N]NP

BinMin (φ, ω)

 a. [AP N]φ

Match (Phrase, φ) *

b. [AP]φ [N]φ

*!*

c. [[AP]φ N]φ

*!

*

The constraint ranking BinMin (φ, ω) >> Match (Phrase, φ), however, renders only one prediction about the optimal candidate of the input verb-object construction whose object is not branching (i.e., composed of one prosodic word), as given in (53). This prediction is correct only when the verb is monosyllabic but does not work for verb-object constructions in which the verb is disyllabic. (53)

[V [NP]]VP (NP is not branching)

BinMin (φ, ω)

 a. [V NP]φ

Match (Phrase, φ) *

b. [V]φ [NP]φ

*!*

c. [V [NP]φ] φ

*!

*

This constraint ranking does not work for the subject-predicate construction either. As can be seen from the discussion in Section 5.1.2, TS is always blocked between the subject and the predicate. Hence, the subject NP must be contained in a phonological phrase that is different from the phonological phrase containing the predicate VP/AP. Assuming that both the subject NP and the predicate VP/AP consist of only one prosodic word, the ranking BinMin (φ, ω) >> Match (Phrase, φ) will select [NP VP/AP]φ as the optimal candidate, as in (54), which goes against the facts. (54)

[NP] [VP/AP]

BinMin (φ, ω)

 a. [NP VP/AP]φ b. [NP]φ [VP/AP]φ

Match (Phrase, φ) **

*!*

From the above discussion, we can find that the Match Theory, like other approaches to phonological phrasing developed in the prosodic phonology theory, also has problems when defining the phonological phrase domain in Fuzhou. Therefore, as the most recent development in the study of the syntax-phonology

154  Phonological phrase in the Fuzhou dialect interface, the Match Theory does not exhibit stronger explanatory power as compared to the Align/Wrap Theory and the original parameterized algorithms of EBA. 5.1.3.3 Summary So far, I have presented a discussion on RBA-based and EBA-based analyses of the definition of the phonological phrase, namely the domain of TS at the phrasal level, in the Fuzhou dialect. We can find that neither of these two major approaches to phonological phrasing can provide a completely satisfactory account for the issue of Fuzhou phrasal TS. An analysis based on RBA or EBA (including its original parameterized algorithms and the recent developments) alone may be able to account for the formation of phonological phrases in some phrasal-level constructions, but it may also make incorrect predictions about the phonological phrasing of some other constructions. Therefore, an alternative analysis, which is based on both RBA and EBA, may turn out to be a better solution, as we will see in Section 5.1.4. 5.1.4 An alternative approach to phonological phrasing in the Fuzhou dialect In this section, I am going to propose and argue for an alternative approach to phonological phrasing in the Fuzhou dialect, which combines the RBA and the EBA. This approach, as will be demonstrated in the following subsections, can cover more linguistic facts of Fuzhou TS at the phrasal level and offer greater predictive power as compared to the previous analyses. 5.1.4.1  Phonological phrase in Fuzhou: a three-step definition From the discussion in Sections 5.1.2 and 5.1.3, we can find that there are several key factors we need to take into consideration when dealing with the phrasal-level application of TS in Fuzhou. The first one is the right edge of the word. As mentioned in Chapter 1, tones in Fuzhou maintain their citation tonal values only when occurring on monosyllabic syllables or on the terminal syllable in a given domain. Therefore, at the phrasal level, if the terminal syllable of a word (whether monosyllabic or polysyllabic) keeps its citation tonal value, the right edge of this word must coincide with the right edge of the tone sandhi domain, namely the right edge of a phonological phrase. In previous sections, we have seen that, by assuming the endsetting ]Xhead and making the right edge of a head X correspond to the right edge of a phonological phrase, a number of data of Fuzhou phrasal TS can be accounted for, e.g., the blocking of TS between the subject and the predicate, between the verb and the resultative complement, as well as between the verb and a branching object. Nevertheless, if we assume the end-setting ]Xhead but do not add any restrictions or stipulations, we can only account for the positions where the blocking of TS happens but cannot explain and predict the positions where TS applies. There are two major empirical problems here. First, the end-setting ]Xhead inserts a phonological phrase domain boundary at the right edge of the head X of the pre-head adjunct XP. This incorrectly separates the adjunct from the following head and makes the

Phonological phrase in the Fuzhou dialect  155 application of TS between the pre-head adjunct and the head unexplained. Second, since the right edge of the head V is always marked with a phonological phrase domain boundary according to the end-setting ]Xhead, such an analysis wrongly excludes the possibility of the application of TS within all verb-object constructions. To solve the first problem, we need to take the second key factor into account, namely the adjunct/non-adjunct distinction. In order not to insert a boundary at the right edge of the head of pre-head adjuncts, we only need to assume that only the right edge of the head X of a non-adjunct XP will be marked, as Shih (1986) suggests. Fuzhou data show that the branchiness of the complement (i.e., object) of the head verb is another key factor relevant to phonological phrasing in this dialect, and the key to solving the second problem is to remove the boundary inserted by ]Xhead at the right edge of the head V when the head V is followed by a nonbranching object. The φ-restructuring rule advanced in the RBA thus can be of help, since it requires a non-branching complement of the head V be grouped into the phonological phrase containing the V. Thus, taking into consideration three key factors relevant to phonological phrasing in the Fuzhou dialect, namely the right edge of the word, the adjunct/ non-adjunct distinction, and the branchiness of the complement, we can tentatively define the phonological phrase in the Fuzhou dialect as in (55). (55) Phonological phrase (φ) domain in the Fuzhou dialect (1st approximation) a. Mark the right edge of every head X, except where XP is an adjunct; b. φ-restructuring: a non-branching φ, which is the first complement of X on its recursive side, is joined into the φ that contains X. The definition in (55), which combines the RBA and the EBA, covers more data as compared to any analysis that is solely based on the RBA or the EBA. Nevertheless, this analysis fails to account for the application of TS between the determiner/ numeral and the classifier since the right edge of the head D or Num would be marked with a boundary according to (55a). This problem can be resolved if the definition in (55) can be refined by stating that only the right edge of the lexical head is marked in the first step, in which the lexical head refers to verbs (V), nouns (N), adjectives (A), prepositions (P), and adverbs (Adv) (cf. Cowper 1992; Culicover 1997; Falk 2001; Carnie 2007; Huang et al. 2009, among others). Thus, we can have the following definition of the phonological phrase in the Fuzhou dialect. (56) Phonological phrase (φ) domain in the Fuzhou dialect (2nd approximation) a. Mark the right edge of every lexical head X, except where XP is an adjunct; b. φ-restructuring: a non-branching φ, which is the first complement of X on its recursive side, is joined into the φ that contains X. By not marking the right edge of the functional head, however, the definition in (56) fails to prevent the head D of a subject DP from being grouped into a phonological phrase with the following predicate and thus fails to predict the blocking of TS between the head D and the predicate, as illustrated in (57).

156  Phonological phrase in the Fuzhou dialect (57) (57)

TP DP

T’

D’

T

VP

D

V’ V’

AP

V

A’

[伊 跳]φ 44 i thiu213 → [ 44 i] # [thiu213] # 51 = thiu213] # *[ i he jump ‘He jumps very high.’

AdvP

A’

[野 ai 31 [ ia21 [ ia21 very

A 悬]φ (predicted by (56a)) keiŋ51 eiŋ51] (correct TS result) eiŋ51] (predicted by (56a)) high

= =

By the same token, the definition in (56) fails to predict the blocking of TS between the classifier and the noun, as exemplified in (58), since classifiers (Cl) are also considered as functional heads. (58) (58)

DP D’ D

ClP Cl’ Cl

[只 隻 tsi31 tsi → [tsi44 = i i *[tsi21 = this Cl ‘this grape’

NP

23 23 24

] # =

葡萄]φ pu51 to51 [pu31 lo51] pu31 lo51] grape

(predicted by (56a)) (correct TS result) (predicted by (56a))

To deal with the problems posed by (57) and (58), we need to insert a boundary at the right edge of the head D in (57) and the right edge of the head Cl in (58). From (57) and (58), we can find that the head D and the head Cl are the

Phonological phrase in the Fuzhou dialect  157 first non-empty head outside of the maximal projection of the following lexical head V and N respectively. Recall that, in the RBA definition of the phonological phrase in (42), a phonological phrase domain starts from a lexical head X and extends until it reaches another head outside of the XP, which can nicely capture the blocking of TS in (57) and (58). Hence, we adapt the RBA definition and incorporate it into (56) and thus arrive at a three-step definition of the Fuzhou phonological phrase, as given in (59). (59) Phonological phrase (φ) domain in the Fuzhou dialect a. Mark the right edge of every lexical head X, except where XP is an adjunct. b. On the non-recursive side of the lexical head X, mark the right edge of the first phonetically overt head Y (if any; either lexical or functional) outside of XP; ωs/CGs that are separated by the right edge of X or Y belong to different φs. c. φ-restructuring: a non-branching φ, which is the first complement of X on its recursive side, is joined into the φ that contains X. Note that in (59) the clitic group status is no longer assigned to the prosodic word. Instead, according to this definition, a phonological phrase in Fuzhou can immediately dominate a prosodic word, which violates the Exhaustivity constraint entailed in the SLH. This is well supported by the Fuzhou data we have seen so far, showing that this constraint is not observed in this dialect, as in many other languages (cf. Inkelas 1989/1990; Kanerva 1989; Itô & Mester 1992/2003; Prince & Smolensky 1993; Mester 1994; Hayes 1995; Vogel 2009; Zhang 2014, 2017, among others). Examples of verb-object constructions in (60) illustrate step by step how the definition of the phonological phrase in (59) works in the Fuzhou dialect. (60) a. (60)

VP

b.

VP

V’ V

V’ NP

V

N’ AP

NP N’

N’

N 食] 大 葡萄] (59a) [食]φ [大 葡萄]φ (59b) [si 5] [tuai242 pu51 to51] → [si 5] # [tuai51 = pu31 lo51] eat big grape ‘to eat big grapes’

N 食] 葡萄] (59a) [葡萄]φ (59b) [食]φ [食 葡萄]φ (59c) [si 5 pu51 to51] → [si 31 = pu31 lo51] eat grape ‘to eat grapes’

158  Phonological phrase in the Fuzhou dialect In (60a), (59a) first marks the right edge of the lexical head V 食 ‘to eat’ and the lexical head N 葡萄 ‘grape’. The right edge of the head A 大 ‘big’ is not marked since its maximal projection AP functions as the adjunct of the head N. Then, (59b) again marks the right edge of the head V since it is the first head that is on the non-recursive side of the lexical head N and outside of NP. Hence, prosodic words 大 and 葡萄 are grouped into one phonological phrase, while the leftover prosodic word 食 forms another phonological phrase on its own. Thus, the application of TS between 大 and 葡萄 and the blocking of TS between the head V and its complement are both accounted for. Since the complement of the head V is branching, (59c) does not play a role here. As can be seen from (60a), a verb followed by a branching object is not to be grouped into the same phonological phrase with the object due to the non-application of the φ-restructuring rule, which is why it does not undergo TS. By contrast, (59c) applies in the phonological phrasing in (60b). After (59a) and (59b) establish two phonological phrases in (60b), the φ-restructuring rule (59c) groups these two phonological phrases into one single φ, since the NP 葡萄 ‘grape’, which forms a non-branching φ, is the only (and hence the first) complement of the head V 食 ‘to eat’ in (60b). By restructuring (60b) into a single phonological phrase, the application of TS between 食 and 葡萄 is accounted for. Besides the examples in (60), the three-step definition in (59) can also be well supported by most of the other Fuzhou data we have discussed in previous sections, as illustrated in (61)–(64). (61) (61) Subject-Predicate a.

TP

NP

T’

N’ AP

T N’ N

AP A’

AdvP

A’

A 旧 书] 真 贵] [旧 书]φ [真 贵]φ [kou242 tsy44] [tsiŋ44 kui213] → [ku44 = y44] # [tsiŋ51 = ŋui213] old book very expensive ‘Old books are very expensive.’

(59a) (59b)

Phonological phrase in the Fuzhou dialect  159 b.

TP DP D’

T’ T

AP

D

A’ AdvP

伊 真 [伊]φ [真 [ i44] [tsiŋ44 → [ 44 i ] # [tsiŋ21 he very ‘He is very smart.’

A’

=

A 聪明] 聪明]φ tshuŋ44 miŋ51] tshuŋ44 miŋ51] smart

(59a) (59b)

(62) (62) Modifier-Head (Attributive-Noun and Adverbial-Adjective/Verb) NP

a.

b.

N’ AP

AP

c.

A’ N’

N 旧 书] [旧 书]φ [kou242 tsy44] → [ku44 = y44] old book ‘old book’

AdvP

VP V’

A’

A 真 贵] [真 贵]φ [tsiŋ44 kui213] → [tsiŋ51 = ŋui213] very expensive ‘very expensive’

AdvP

V’

V 齐 来] (59a) [齐 来]φ (59b) [ts 51 li51] → [ts 31 = li51] together come ‘to come together’

160  Phonological phrase in the Fuzhou dialect (63) (63) Verb-Resultative complement7 a.

VP V’

V’

AP

V

A’ AdvP

A’

A 跳] 野 悬 [跳]φ [野 悬]φ [thiu213] [ ia31 keiŋ51] h 213 21 → [t iu ] # [ ia = eiŋ51] jump very high ‘(somebody) jumps very high’ b. VP c.

(59a) (59b)

VP

V’ V’

V’ AP

V’

V 食] 饱 (59a) [食]φ [饱]φ (59b) [si 5] [pa31] → [si 5] # [pa31] eat full ‘to eat one’s fill’

(64) (64) Determiner/Numeral-Classifier-Noun a.

DP

V 食] 完 (59a) [食]φ [完]φ (59b) [si 5] [ uoŋ51] → [si 5] # [ uoŋ51] eat finish ‘to finish eating’

b.

NumP

D’ D

VP

Num’ ClP

Num

ClP

Cl’ Cl

Cl’ NP

只 隻 葡萄] [葡萄]φ [只 隻]φ [tsi31 tsi 23] [pu51 to51] → [tsi44 = i 23] # [pu31 lo51] this Cl grape ‘this grape’

Cl

NP

蜀 隻 葡萄] (59a) [蜀 隻]φ [葡萄]φ (59b) [suo 5 tsi 23] [pu51 to51] → [suo 21 = i 23] #[pu31 lo51] one Cl grape ‘one grape’

Phonological phrase in the Fuzhou dialect  161 In addition to (60)–(64), the definition in (59) can also account for the behavior of TS in other phrasal-level constructions that have been mentioned but have not been discussed in detail in previous studies, as exemplified in (65)–(66). (65) Preposition-object (65) Preposition-Object a.

PP

b.

VP

P’ P

V’ NP

向] 北] (59a) [向]φ [北]φ (59b) (59c) [向 北]φ [xyoŋ213 pøy 23] → [xyoŋ51 = pøy 23] towards north ‘towards the north’

c.

P

PP

V’

P’

V NP

朝 西] 行] (59a) [朝]φ [西]φ [行]φ (59b) [朝 西]φ [行]φ (59c) [tiu51 s 44] [kiaŋ51] 44 44 → [tiu = s ] # [kiaŋ51] towards west walk ‘to walk towards the west’

PP P’ P

DP D’ D

ClP Cl’ Cl

NP

按] 只 芘 药方] (59a)8 [按]φ [只 芘]φ [药方]φ (59b) [ aŋ51] [tsi31 pi 5] [yo 5 xuoŋ44] → [ aŋ51] # [tsi21 = βi 5] # [yo 44 uoŋ44] according to this Cl prescription ‘according to this prescription’

162  Phonological phrase in the Fuzhou dialect (66) Serial verb construction (66) a.

VP V’

VP

V’

V’

V

NP

V 去 食] [去 食]φ [去 食 [kho213 si 5 h 51 → [k o = si 21 go eat ‘to go to eat food’ b.

饭] (59a) [饭]φ (59b) (59c) 饭]φ puoŋ242] puoŋ242] rice

=

VP V’ VP

V’

V’ V

V

NP

NP

(59a) 去 店] 食] 饭] [去]φ [店]φ [食]φ [饭]φ (59b) [去 店]φ [食 饭]φ (59c) [kho213 taiŋ213] [si 5 puoŋ242] [si 21 = puoŋ242] → [kho21 = taiŋ213] # go restaurant eat rice ‘to go to the restaurant to eat food’

Besides the phrasal-level constructions discussed above, the three-step definition of the phonological phrase in Fuzhou can also cover examples with more complicated syntactic structures that are rarely discussed in previous studies. It has been claimed that some words in Fuzhou never form a tone sandhi domain with any following constituents, such as 乞 [khøyʔ23] ‘to give’, 叫 [kiu213] ‘to tell’, and 乞 [khøyʔ23] ‘passive marker’9 (Chen & Norman 1965a; Wright 1983). Examples of these three words are given in (67).

Phonological phrase in the Fuzhou dialect  163 (67)

a. → c. →

乞 我 khøyʔ23 ŋuai31 khøyʔ23 # ŋuai31 # give me ‘to give me money’ 乞 我 khøyʔ23 ŋuai31 khøyʔ23 # ŋuai31 # PASS me ‘to be hit by me’

钱 tsiɛŋ51 tsiɛŋ51 money

b. →

叫 我 kiu213 ŋuai31 kiu213 # ŋuai31 # tell me ‘to tell me to go’

去 kho213 kho213 go

拍 phaʔ23 phaʔ23 hit

In Chen  & Norman (1965a), 乞 [khøyʔ23] ‘to give’ is treated as a lexically marked exception to the TS rule. In Wright (1983), the blocking of TS between these words and their following constituents is ascribed to the nontheme theta-role they are assigned with. In the following, I am going to show that the blocking should be attributed to the syntactic structure of relevant constructions and thus can be predicted by the definition of Fuzhou phonological phrase in (59). As can be seen from (67a), 乞 ‘to give’ in the Fuzhou dialect is a ditransitive verb that takes two objects (direct object and indirect object). The double-object construction has long been recognized as a problem for the theory of syntax. Since Larson’s (1988) analysis of ditransitive verbs involving a layered V (VPshell), there have been a number of proposals regarding the idea of an extra head in the V domain (Grimshaw & Mester 1988; Bowers 1993; Hale & Keyser 1993; Kratzer 1993; Chomsky 1995, among others). Following the idea on (v-) Voice as the head whose specifier hosts the external argument of a verb (Kratzer 1993), Chomsky (1995) first proposes the concept of the light verb, or little v. According to Chomsky (1995), the job of Agent-introducing is attributed to v, and VP is taken as the complement of v. Thus, the basic D-structure of a vP can be presented as follows. (68)

vP Spec v

v’ VP Spec V

V’ …

By adopting the structure in (68), the D-structure of the double-object construction of (67a) can be presented as follows (cf. Tang 2003, 2010).

164  Phonological phrase in the Fuzhou dialect (69)

vP1 Spec

v1’

v1

vP2 v2’

DP v2

VP NP

V’

N’ V

DP

N 钱 乞 我 money give me ‘to give me money’

In order to yield the correct S-structure of sentence presented in (67a), the indirect object 我 ‘me’ must move to the specifier position of vP2, and the head V 乞 ‘give’ must move to the head v2 position and then to the head v1 position. Thus, the S-structure of (67a) can be presented as in (70). Traces of moved elements are indicated by t. Following the definition in (59), the phonological phrasing of (67a) is also given in (70). (70) (70)

vP1 v1’

Spec v1i

vP2 v2’

DPj D’

v2i

D

VP NP N’ ti

乞] 我 [乞]φ [我]φ [khøy 23] [ŋuai31] h 23 → [k øy ] # [ŋuai31] # give me ‘to give me money’

N 钱] [钱]φ [tsi ŋ51] [tsi ŋ51] money

V’ tj

(59a)10 (59b)

Phonological phrase in the Fuzhou dialect  165 As we can see from (70), by adopting the concept of vP and the structure of vP in (68), the blocking of TS between the verb 乞 ‘to give’ and its following constituent can be correctly predicted by the definition in (59). 乞 ‘to give’ is not the only ditransitive verb in the Fuzhou dialect. Other ditransitive verbs, such as 送 [soyŋ213] ‘to give’ and 借 [tsuoʔ23] ‘to lend’, actually exhibit the same behavior as 乞 ‘to give’ in terms of the application of TS, as exemplified in (71). (71) a.

b.

借 我 tsuoʔ23 ŋuai31 23 → tsuoʔ # ŋuai31 # lend me ‘to lend me money’

送 我 钱 soyŋ213 ŋuai31 tsiɛŋ51 soyŋ213 # ŋuai31 # tsiɛŋ51 give me money ‘to give me money’

→

钱 tsiɛŋ51 tsiɛŋ51 money

Double-object constructions in (71) share the same surface syntactic structure of (70), so the ditransitive verb and the following constituent belong to separate phonological phrases according to the definition in (59). This is why we can find systematic blocking of TS between a ditransitive verb and its following constituent. Now, let us move on to the behavior of 叫 ‘to tell’. Examples like (67b) are so-called “pivotal constructions” in Chinese, in which the object of the first verb functions as the subject of the second verb/verb phrase at the same time. Like ­double-object constructions, pivotal constructions can also be parsed with the concept of vP. Thus, the surface syntactic structure of (67b) can be presented as in (72) (cf. Tang 2010; He 2011). The phonological phrasing of (67b) is also given in (72). (72) (72)

vP v’

Spec vi

VP V’

DPj ti 叫] 我 [叫]φ [我]φ [kiu213] [ŋuai31] → [kiu213] # [ŋuai31] tell me ‘to tell me to go’

TP

#

PROj 去] [PROj 去]φ [kho213] [kho213] go

(59a)11 (59b)

From (72), we can find that by adopting the concept of vP and the structure of vP in (68), the tone sandhi behavior of 叫 ‘to tell’ can be handled—the definition of the phonological phrase in (59) correctly predicts the blocking of TS between the verb 叫 ‘to tell’ and the following object.

166  Phonological phrase in the Fuzhou dialect Other verbs that can form pivotal constructions in Fuzhou include 告 [ko213] ‘to tell’; 请 [tshiaŋ31] ‘to invite’; 劝 [khuoŋ213] ‘to advise, to persuade’; 干 [kaŋ44] ‘to force’; and 逼 [peiʔ23] ‘to force’. Some examples are presented in (73). We can find that the TS rule, again, is blocked between these verbs and their objects, which can be nicely captured by the structure in (72) and the definition in (59). (73) a. → c. → e. →

告 我 ko213 ŋuai31 213 ko # ŋuai31 # tell me ‘to tell me to go’ 劝 我 khuoŋ213 ŋuai31 h 213 k uoŋ # ŋuai31 # advise me ‘to advise me to go’ 逼 我 peiʔ23 ŋuai31 23 peiʔ # ŋuai31 # force me ‘to force me to go’

去 kho213 kho213 go

b.

去 kho213 kho213 go

d.

→

→

请 我 tshiaŋ31 ŋuai31 h 31 ts iaŋ # ŋuai31 # invite me ‘to invite me to go’ 干 我 kaŋ44 ŋuai31 44 kaŋ # ŋuai31 # force me ‘to force me to go’

去 kho213 kho213 go 去 kho213 kho213 go

去 kho213 kho213 go

The last phrasal-level construction explored in this subsection is the passive construction. Similar to bei 被 in Mandarin Chinese, 乞 as the passive marker in Fuzhou is also used to introduce the agent preceding the VP. The object of the verb (i.e., the receiver of the action) in a passive sentence is placed in the sentence subject position. Feng (1995) proposes an analysis for Mandarin passive sentences, with the notions of null operator (NOP) movement and predication. Many other works have appeared in support of Feng’s proposal, including Chiu (1995); Cheng et  al. (1993, 1996); Ting (1995, 1996); Tang (2008, 2010); and Huang et al. (2009), among others. Since the passive construction in Fuzhou is similar to that in Mandarin Chinese, I follow the previous studies and assume the surface syntactic structure of (67c) as in (74). We can find that the blocking of TS between 乞 and the agent can be explained since they belong to different phonological phrases according to the definition in (59).

Phonological phrase in the Fuzhou dialect  167 (74) (74)

TP DPi

... v’

D’ D

vj

VP V’ tj

TP NOPi

TP

DP D’

乞] [乞]φ [khøy 23] # → [khøy 23] PASS ‘to be hit by me’

D 我 [我]φ [ŋuai31] [ŋuai31] # me

... V’ V ti 拍] [拍]φ [pha 23] [pha 23] hit

(59a) (59b)

From the above discussion and the examples in (67)–(74), we can find that, by adopting the concept of vP, the blocking of TS between 乞 [khøyʔ23] ‘to give’/叫 [kiu213] ‘to tell’/乞 [khøyʔ23] ‘passive marker’ and their following constituents can be nicely explained with the definition of Fuzhou phonological phrase in (59). The blocking of TS in these cases has nothing to do with the theta-role but should be attributed to the phonological phrasing based on the syntactic structure. The blocking of TS exhibited by these words is not lexically marked either, since other words with the same syntactic function also have the same TS behavior, which provides further evidence for the analysis assuming the notion of vP and the definition in (59). So far, I have shown how the three-step definition of the phonological phrase in Fuzhou is proposed and have demonstrated that this alternative approach to phonological phrasing, which combines the EBA and the RBA, is a better solution to the long-standing problem of Fuzhou TS at the phrasal level. This approach not only provides a satisfactory account for many examples we have seen in previous studies, but also explains some of the relevant data that has rarely been discussed before. It is noteworthy that we are now still left with two issues that need to be addressed. One is the blocking of TS between a verb and an NP indicating the

168  Phonological phrase in the Fuzhou dialect location (“[−theme] argument” in Chan’s 1998 terminology), and the other is the contrast exhibited by monosyllabic verbs and disyllabic verbs, both of which have been noticed in some previous studies but have not received satisfactory explanation. It seems that these two issues may pose a challenge to the alternative analysis we have just advanced, but as we will see in Section 5.1.4.2, this would not be a problem at all. 5.1.4.2  Residual issues of Fuzhou TS at the phrasal level We have seen that a monosyllabic verb in Fuzhou can be incorporated into a phonological phrase with its object when the φ-restructuring rule is triggered according to the three-step definition in (59), which explains why a monosyllabic verb may undergo TS when followed by a non-branching object. However, it has long been noticed that not every verb in the Fuzhou dialect undergoes TS before a non-branching object (Chen & Norman 1965a; Wright 1983; Hung 1987; Zhang 1992; Chan 1998, among others). From the discussion in Section 5.1.2 and Section 5.1.3, we can find that the TS rule is never triggered between a verb and an NP indicating the location, nor does it apply between a disyllabic verb and the object no matter the object is branching or not. These two cases seem to pose difficulties for the alternative analysis proposed in Section 5.1.4.1. Nonetheless, as I will demonstrate in the following, the blocking of TS in these two cases does not invalidate the alternative analysis and can be nicely handled with the Domain Impenetrability Condition (DIC, You 2018a). 5.1.4.2.1  VERB + LOCATION

Chen & Norman (1965a) and Wright (1983) have reported that there are a few verbs that never undergo TS with any following constituent, among which there are verbs such as 住 [tiu242] ‘to live’, 坐 [soy242] ‘to sit’, 徛 [khiɛ242] ‘to stand’, 放 [pouŋ213] ‘to put’, and 着 [tuoʔ5] ‘to be in or at’. Chan (1998) has also noticed the distinctive TS behavior exhibited by such verbs, as we have seen in (40). The examples in (40) are reproduced in (75), in which we can find that TS applies between the verb and the following NP in (75a), while is blocked in (75b). (75) a. (75)

VP

b.

VP

V’ V 食 si 5 → si 31 = eat ‘to eat grapes’

V’ NP 葡萄 pu51 to51 pu31 lo51 grape

V

NP

回 福州 xui51 xu 23 tsiu44 → xui51 # xu 21 tsiu44 return Fuzhou ‘to return to Fuzhou’

Phonological phrase in the Fuzhou dialect  169 To deal with the contrast presented in (75), Wright (1983) argues that a verb will undergo TS only if its following argument is assigned thetarole, but not other theta-roles such as , , or . I have pointed out in Section 5.1.2.2 that this is an ad hoc solution in Wright’s ­prosody-based analysis. Chan (1998) also introduces a distinction between two types of argument NPs, namely [+theme] NP and [−theme] NP, and argues that the V-NP[+theme] construction forms an incorporated noun through Noun Incorporation while V-NP[−theme] is a syntactic phrase. This analysis, as argued in Section 5.1.2.6, is counter-intuitive and is also on an ad hoc basis. A closer examination reveals that verbs like 食 ‘to eat’ and verbs like 回 ‘to return’ and 住 ‘to live’ actually belong to different subcategories of verbs. Verbs like 食 ‘to eat’ are typical transitive verbs that are able to take their own direct object. By contrast, verbs like 回 ‘to return’ and 住 ‘to live’ are intransitive verbs and hence do not take the direct object. The NP that follows an intransitive verb, such as 福州 ‘Fuzhou’ in (75b), is not the direct object of the verb. Furthermore, an intransitive verb such as 回 ‘to return’ and 住 ‘to live’ can only assign the thetarole to its own syntactic argument, which occurs in the subject position. Hence, the argument following the intransitive verb actually does not receive the thetarole from the verb. Thus, we have to answer the following question: where does the argument like 福州 ‘Fuzhou’ in (75b) obtain the theta-role? Notice that I  have mentioned two Fuzhou enclitics that serve as post-verbal particles in Chapter  4, namely 敆 [kaʔ0] and 遘 [kau213]. Both of them can be attached to the verb to introduce the location. Basically, the former indicates the location where something is located or some actions take place, e.g., 住敆福州 ‘to live in Fuzhou’. In contrast, the latter indicates the destination/result of a movement/action, e.g., 回遘福州 ‘to return to Fuzhou’. 敆 [kaʔ0] and 遘 [kau213] are not indispensable in the sentences, as exemplified in (76) and (77). (76) a. →

(77) a. →

敆 住 tiu242 kaʔ0 tiu242 # ʔaʔ0 # live PVP ‘to live in Fuzhou’

b.

福州 xuʔ23 tsiu44 xuʔ21 tsiu44 Fuzhou

回 遘 福州 xui51 kau213 xuʔ23 tsiu44 51 213 xui # ʔau # xuʔ21 tsiu44 return PVP Fuzhou ‘to return to Fuzhou’

→

b. →

住 福州 tiu242 xuʔ23 tsiu44 tiu242 # xuʔ21 tsiu44 live Fuzhou ‘to live in Fuzhou’

回 福州 xui51 xuʔ23 tsiu44 51 xui # xuʔ21 tsiu44 return Fuzhou ‘to return to Fuzhou’

We can find that (76a) and (77a) are similar to (76b) and (77b) respectively in terms of their semantic meanings as well as their tone sandhi behaviors. This can also be illustrated by examples of other verbs of this type mentioned in the literature, namely 坐 [soy242] ‘to sit’, 徛 [khiɛ242] ‘to stand’, 放 [pouŋ213] ‘to put’, and 着 [tuoʔ5] ‘to be in or at’, as presented in (78)–(81).

170  Phonological phrase in the Fuzhou dialect (78) a.

b.

敆 坐 厅中 242 soy kaʔ0 thiaŋ44 touŋ44 242 0 soy # ʔaʔ # thiaŋ44 nouŋ44 sit PVP drawing room ‘to sit in the drawing room’

→

(79) a.

徛 敆 厅中 kaʔ0 thiaŋ44 touŋ44 khiɛ242 h 242 0 → k iɛ # ʔaʔ # thiaŋ44 nouŋ44 stand PVP drawing room ‘to stand in the drawing room’

(80) a.

→

b.

徛 厅中 khiɛ242 thiaŋ44 touŋ44 h 242 → k iɛ # thiaŋ44 nouŋ44 stand drawing room ‘to stand in the drawing room’ b.

敆 放 厅中 pouŋ213 kaʔ0 thiaŋ44 touŋ44 pouŋ213# ŋaʔ0 # thiaŋ44 nouŋ44 put PVP drawing room ‘to put in the drawing room’

→

(81) a.

→

敆 b. 着 厅中 tuoʔ5 kaʔ0 thiaŋ44 touŋ44 tuoʔ5 # ʔaʔ0 # thiaŋ44 nouŋ44 → be in PVP drawing room ‘to be in the drawing room’

→

坐 厅中 soy242 thiaŋ44 touŋ44 soy242 # thiaŋ44 nouŋ44 sit drawing room ‘to sit in the drawing room’

放 厅中 pouŋ213 thiaŋ44 touŋ44 pouŋ213 # thiaŋ44 nouŋ44 put drawing room ‘to put in the drawing room’

着 厅中 tuoʔ5 thiaŋ44 nouŋ44 5 tuoʔ # thiaŋ44 touŋ44 be in drawing room ‘to be in the drawing room’

Since an “intransitive V + 敆/遘 + location” construction shares the same semantic meaning and the tone sandhi behavior with its corresponding “intransitive V + location” construction, it is reasonable to assume that they share the same syntactic structure as well and the latter construction is derived from the former by not articulating the enclitic PVP 敆 or 遘. It is the omitted or soundless PVP that assigns the theta-role to the following locative argument. The syntactic structure of “intransitive V + 敆/遘 + location” and “intransitive V + location” constructions can be presented as in (82). (82) (82) a. Intransitive V+敆/遘+location VP

b.

Intransitive V+location VP

V’ V Verb

V’ NP (location)

敆/遘 (PVP)

NP (location)

V Verb

ø (PVP)

Notice that the non-branching NP in (82a) must be incorporated into a phonological phrase with the preceding head V according to the φ-restructuring rule, and thus phrasal-level TS is expected to apply within the phonological phrase.

Phonological phrase in the Fuzhou dialect  171 From the examples in (76a)–(81a), nevertheless, we can find that neither the verb nor the PVP (敆/遘) undergoes TS. Since the PVP is an enclitic in the Fuzhou dialect and forms a Type A clitic group with the preceding prosodic word, as discussed in Chapter  4, there are two possible explanations for the blocking of TS in a structure like (82a): (a) an enclitic never triggers TS on the preceding prosodic word, nor does it undergo TS itself (notice that 遘 [kau213] does not carry a neutral tone and hence phonologically it should be able to undergo TS); (b) a Type A  clitic group embedded within a phonological phrase is not affected by a rule that applies within the domain of the phonological phrase. I will leave the discussion of these two explanations to Section 5.1.4.2.2, and it will be shown that the second explanation is more general than the first one. At this moment, nonetheless, no matter which explanation is adopted, they can both account for the blocking of TS in (76a)–(81a). Since (82b) shares the same structure with (82a), it is reasonable to expect that TS is blocked in (76b)–(81b) by the same token. To recapitulate, the example in either (75a) or (75b) forms only one phonological phrase domain according to the definition in (59). The difference in the tone sandhi behavior of the verbs in (75a) and (75b) is ascribed to the different subcategories of the verbs. The verb in (75a) is transitive and takes two arguments, while the verb in (75b) is intransitive and has only one argument. The argument following the verb in (75b) is not the argument of the verb and is not able to receive the theta-role from the verb, so there must be a (soundless) constituent that assigns the theta-role to this post-verb argument. Since the “intransitive V + 敆/遘 + location” construction and the “intransitive V + location” construction share the same semantic meaning and tone sandhi behavior, it is natural to assume that they have the same syntactic structure. Hence, the verb in cases like (75b) must be followed by an omitted, soundless enclitic 敆/遘, which plays a crucial role in blocking the TS within the phonological phrase domain formed in cases like (75b), as demonstrated by the examples in (76b)–(81b). Thus, by distinguishing the verbs in (75a) and (75b) as well as their syntactic structures, the problem raised by verbs like 住 ‘to live’, 坐 ‘to sit’, 徛 ‘to stand’, 放 ‘to put’, 回 ‘to return’, and 着 ‘to be in or at’ can be well handled. 5.1.4.2.2  MONOSYLLABIC VERB VS. DISYLLABIC VERB

The second residual issue concerning Fuzhou phrasal-level TS results from the contrast in tone sandhi behavior between monosyllabic verbs and disyllabic verbs. It has long been recognized that monosyllabic verbs and disyllabic verbs in Fuzhou have different TS behavior at the phrasal level (Chen & Norman 1965a; Wright 1983; Shih 1986; Zhang 1992, among others). A monosyllabic verb can undergo TS when followed by an object as long as the object is the first nonbranching complement of the verb, while a disyllabic verb (to be more specific, the tone of the second syllable in a disyllabic verb) never undergoes TS, as exemplified in (83) and (84) respectively.

172  Phonological phrase in the Fuzhou dialect (83) a. →

(84) a. →

趁 钱 theiŋ213 tsiɛŋ51 thiŋ44 = tsiɛŋ51 earn money ‘to earn money’

b.

浪费 louŋ242 xiɛ213 louŋ51 xiɛ213 # waste ‘to waste money’

钱 tsiɛŋ51 tsiɛŋ51 money

→

食 葡萄 siɛʔ5 pu51 to51 31 siɛʔ = pu31 lo51 eat grape ‘to eat grapes’ b. →

讨厌 tho31 ʔiɛŋ213 tho44 ʔiɛŋ213 # hate ‘to hate grapes’

葡萄 pu51 to51 pu31 lo51 grape

We can find that examples in (83) and (84) share the same syntactic structure, namely a verb head plus a non-branching complement. Thus, each phrasal-level construction in (83) and (84) forms a single phonological phrase according to the definition in (59). In previous sections, I have demonstrated that the phonological phrase is the domain of application for Fuzhou TS at the phrasal level. Hence, there must be a condition that can manipulate the application and/or blocking of TS within the phonological phrase domain and thus distinguishes monosyllabic verbs from disyllabic verbs with respect to their tone sandhi behavior. The concept of the foot has been adopted by several linguists (e.g., Wright 1983; Shih 1986, among others) to deal with the contrast presented in (83) and (84). However, it is noteworthy that the domain of the phonological phrase should be constructed on the basis of syntactic notions, rather than phonological notions, as can be seen in the EBA, the RBA, and our alternative analysis. Moreover, as discussed in Chapter 2, no independent evidence has been identified in the Fuzhou dialect for the existence of the stress or the foot. Therefore, it is invalid to ascribe the contrast between (83) and (84) to the formation of the foot. Different from the foot analysis, Zhang (1992, 2017) introduces the concept of the c-command to account for the problem caused by the contrast between (83) and (84). As discussed in Section 5.1.2.5, taking the c-command condition as a relationship between syllables somehow resolves the problem, but such an analysis is not unquestionable because c-command actually refers to a relationship between nodes of a syntactic tree, instead of between syllables. As we can find from (83) and (84), the only difference between these two pairs of examples lies in the number of syllables in the verbs. If such a difference is not related to the formation of the foot and also has nothing to do with c-command, then what is the key factor that prevents the second syllable within a disyllabic verb from undergoing TS? Let us take a look at the internal prosodic structure of the examples in (83a) and (84a), as shown in (85a) and (85b) respectively. We can see that the phonological phrases formed by these two examples are both composed of two prosodic words.

Phonological phrase in the Fuzhou dialect  173 (85) a. →

[[趁]ω [钱]ω]φ [[theiŋ213] [tsiɛŋ51]] [[thiŋ44] = [tsiɛŋ51]] earn money ‘to earn money’

b. →

[[浪费]ω [[louŋ242 xiɛ213] [[louŋ51 xiɛ213] # waste ‘to waste money’

[钱]ω]φ [tsiɛŋ51]] [tsiɛŋ51]] money

As discussed in Chapter 3, the prosodic word in the Fuzhou dialect is defined as the terminal node of the syntactic tree. Therefore, whether monosyllabic or disyllabic, every morphosyntactic word in Fuzhou forms a prosodic word, which serves as the lexical-level domain of application for TS. Nevertheless, since tone sandhi occurs when two or more syllables are contained within a given domain, the TS rule only applies within the prosodic word domain formed by polysyllabic morphosyntactic words but is never triggered within the prosodic word domain formed by monosyllabic words. Hence, in (85a), the syllable in the monosyllabic verb does not undergo TS within the prosodic word domain due to the lack of appropriate phonological context. By contrast, in (85b), the syllables in the disyllabic verb provide the applicable phonological environment for TS, which is why TS has been triggered once within the prosodic word domain before the derivation moves to the outer domain. After the derivation moves to the phonological phrase domain, which is the phrasal-level domain of application for TS, the TS rule applies to the syllable in the monosyllabic verb in (85a) (notice that the application happens in the phonological phrase domain where the monosyllabic verb “meets” its object, but not in the prosodic word domain), but not to the syllables in the disyllabic verb in (85b). Therefore, it seems that once the derivation has moved from the prosodic word domain to the phonological phrase domain, the TS rule that is expected to apply within the phonological phrase domain is no longer able to go into effect within the prosodic word domain. In other words, the TS rule fails to “look back” to the prosodic word domain when the derivation has moved to the phonological phrase domain. Some “no look-back” devices have been proposed in some previous studies on tone sandhi in Chinese dialects, such as One Step Principle (Hsu 1994), NoBacktracking (Chen 2000), and Moving Window (Chen et al. 2004), as presented in (86). (86) a. One Step Principle (Hsu 1994) A tone that has undergone change cannot change again. b. No-Backtracking (Chen 2000) Do not backtrack. c. Moving Window (Chen et al. 2004) Two-tone sandhi may not apply to the same local window more than once. These devices, however, cannot account for the “no look-back” phenomena in the Fuzhou dialect. Take (85b) as an example. The tone of 费 [xiɛ213] does not undergo TS within the first prosodic word domain [浪费]ω, so it should be able

174  Phonological phrase in the Fuzhou dialect to change within the phonological phrase [浪费钱]φ ‘to waste money’ according to Hsu’s (1994) One Step Principle, but this is patently an incorrect prediction. Chen’s (2000) No-Backtracking is not able to deal with this example either. Since No-Backtracking imposes a unidirectional scanning, the movement of the “local window” (in Chen’s terminology) of TS in (85b) will begin from left to right, namely from 浪费 [louŋ242 xiɛ213] to 费钱 [xiɛ213 tsiɛŋ51], which does not prevent the tone of 费 [xiɛ213] from undergoing TS. The Moving Window constraint (Chen et al. 2004) makes a wrong prediction for (85b) as well: 费 [xiɛ213] and 钱 [tsiɛŋ51] form a new two-tone window of TS when the derivation has moved to the phonological phrase domain, so TS should not be blocked here according to the Moving Window constraint. Inspired by other “no look-back” devices that have been proposed in the literature, such as different versions of the Strict Cycle Condition (Chomsky 1973; Mascaró 1976; Halle 1978; Kiparsky 1982a, 1982b; Kaye 1995) and Chomsky’s (2000, 2001, 2004 et passim) Phase Impenetrability Condition, You (2018a) proposes a Domain Impenetrability Condition (DIC) as a “no look-back” device in prosodic phonology to account for the “no look-back” phenomena in Fuzhou, by assuming that phonological rules like TS apply cyclically within successively larger prosodic domains (also cf. Shih 1986, 1997; Wee 2004). The DIC is formulated as in (87), which conditions the cyclic application of phonological operations across domain boundaries. (87) Domain Impenetrability Condition (DIC) (You 2018a) For a given string in prosodic domain i, in which the structural description of any phonological rules is met, after the phonological computation has been done in i and the derivation has moved to the next prosodic domain j, properties of the string that has been spelled out in i are not further modifiable within j. The DIC in (87) nicely explains the contrast between monosyllabic verbs and disyllabic verbs in Fuzhou in terms of their tone sandhi behavior. Let us illustrate how the DIC works with the two examples in (85). In (85b), both 浪 [louŋ242] and 费 [xiɛ213] are contained in the first prosodic word domain, where the structural description of TS is met. Thus, in this domain, TS applies to 浪 [louŋ242] and changes it into [louŋ51], while the tone of 费 [xiɛ213] does not undergo TS. Then, the string composed of 浪 [louŋ242] and 费 [xiɛ213] is spelled out as [louŋ51 xiɛ213] and the derivation moves to the phonological phrase domain. Within the phonological phrase, all the properties of the string of 浪费 [louŋ51 xiɛ213] are immune against further modification according to the DIC, which is why the TS rule, though expected to apply within the phonological phrase domain, is unable to change the tone of either 浪 [louŋ51] or 费 [xiɛ213]. By contrast, in (85a), the first prosodic word contains only one syllable, namely 趁 [theiŋ213], so in this prosodic word domain, the structural description

Phonological phrase in the Fuzhou dialect  175 of TS is not satisfied. After the syllable 趁 [theiŋ213] in the prosodic word domain is spelled out and the derivation moves to the phonological phrase domain, further phonological modification on this syllable is not prohibited, which is why it undergoes TS in the phonological phrase domain. The DIC is not an ad hoc solution for the problem regarding the contrast in the TS behavior between monosyllabic verbs and disyllabic verbs—it also handles the blocking of TS in the construction of “intransitive V + 敆/遘 + location” discussed in Section 5.1.4.2.1. In such a construction, “intransitive V + 敆/遘 ” forms a Type A clitic group domain before it is incorporated into the phonological phrase domain, and the TS rule that is expected to apply within the phonological phrase does not affect any constituent within the Type A clitic group. I proposed two possible explanations in Section 5.1.4.2.1 for this case, namely (a) an enclitic never triggers TS on the preceding prosodic word, nor does it undergo TS itself; (b) a Type A clitic group embedded within a phonological phrase is not affected by a rule that applies within the domain of the phonological phrase. Clearly, the first explanation is so specific that it only works for the Type A clitic group but cannot be applied to other cases. By contrast, the second explanation can be subsumed under the DIC in (87). Take (88) as an example. (88) →

[[[回]ω 遘C]CG [[[xui51] kau213] 51 [[[xui ] # ʔau213] return PVP ‘to return to Fuzhou’

#

[福州]ω]φ [xuʔ23 tsiu44]] [xuʔ21 tsiu44]] Fuzhou

In (88), 回 [xui51] and 遘 [kau213] are not able to “see” each other until the derivation moves to the Type A clitic group. In the Type A clitic group, the structural description of TS is met, but TS does not apply between 回 [xui51] and 遘 [kau213] because the Type A clitic group cannot serve as the domain of application for TS in Fuzhou, as discussed in Chapter 4. Then, the derivation moves to the phonological phrase, in which the Type A clitic group has been spelled out as 回遘 [xui51 ʔau213]. Although TS is expected to be triggered within the phonological phrase domain, no further phonological modification can be done on the properties of the string of 回遘 [xui51 ʔau213] according to the DIC, which is why 遘 [ʔau213] does not undergo TS. Thus, we can find that the DIC can nicely explain the TS behavior in the construction of “intransitive V + 敆/遘 + location”. Since the “intransitive V + location” construction shares the same structure with the “intransitive V + 敆/遘 + location” construction, as discussed in Section 5.1.4.2.1, TS is blocked between the intransitive verb and the NP indicating the location by the same token. Let us take a look at some additional examples of “no look-back” phenomena in Fuzhou, as presented in (89).

176  Phonological phrase in the Fuzhou dialect (89) I.

II.

φ with embedded ω a. [[马上]ω [来]ω]φ [[ma31 suoŋ242] [li51]] → [[ma44 luoŋ242] # [li51]] immediately come ‘to come immediately’

b. →

[[直接]ω [[tiʔ5 tsieʔ23] [[tiʔ21 tsieʔ23] # directly ‘to go directly’

[去]ω]φ [kho213]] [kho213]] go

φ with embedded Type A CG a. [[[去]ω 过C]CG [天津]ω]φ [[[kho213] kuo213] [thiɛŋ44 kiŋ44]] → [[[kho213] # ʔuo213] # [thiɛŋ44 kiŋ44]] go EXP Tianjin ‘to have been to Tianjin before’ b. [[[食]ω 着C]CG [鳗鱼]ω]φ [[[siɛʔ5] tuoʔ5] [muaŋ51 ŋy51]] 5 5 → [[[siɛʔ ] # luoʔ ] # [muaŋ31 ŋy51]] eat EXP eel ‘to have eaten eels before’

Examine the examples in (89I) first. We have known that TS is usually triggered in adverbial-verb constructions, because the adverbial as an adjunct is incorporated into the same phonological phrase with its head verb according to the definition in (59), such as 齐来 [tsɛ51 li51 → tsɛ31 li51] ‘to come together’. However, in both examples in (89I), where the first prosodic word formed by the adverbial is composed of two syllables, TS does not apply in the phonological phrase, leaving the second syllable in the adverbial unchanged. The contrast in TS behavior between monosyllabic and disyllabic adverbials thus recalls the contrast between monosyllabic and disyllabic verbs and can be easily accounted for by the DIC. The examples in (89II) do not pose any challenge to the DIC either. The TS behavior of both examples is similar to that of (88)—although they do not serve as the domain of application for TS, the Type A clitic groups in these examples contain more than one syllable and thus satisfy the structural description of TS. Once the derivation moves to the phonological phrase domain, the TS rule that is expected to apply cannot affect the tones of the enclitics. The examples of phonological phrases with embedded prosodic words and Type A clitic groups in (89) thus further support the idea of DIC. From Section  5.1.4.2.1 and Section  5.1.4.2.2, we can find that the two residual issues regarding the application of Fuzhou TS at the phrasal level, namely the blocking of TS between a verb and an NP indicating the location, and the contrast between monosyllabic verbs and disyllabic verbs, are both caused by the internal prosodic structure of relevant constructions. With the DIC proposed in You (2018a), we can easily handle these two issues, showing that they are not a challenge to the three-step definition of Fuzhou phonological phrase advanced in Section 5.1.4.1. 5.1.4.3 Summary In Section 5.1.4, I propose a three-step definition for the phonological phrasing in the Fuzhou dialect, which combines the RBA and the EBA. This alternative

Phonological phrase in the Fuzhou dialect  177 approach, together with the DIC proposed in You (2018a), can nicely cover all the linguistic facts of Fuzhou TS at the phrasal level we have seen so far, as I have demonstrated. Therefore, this alternative analysis has explanatory power that is superior to that of any previous analyses. 5.1.5 Summary In Section 5.1, I have revisited the issue of Fuzhou TS at the phrasal level, which has long been recognized as a problem for linguists. I have reviewed and compared the most important previous analyses and discussed their advantages as well as their shortcomings, based on which I suggest that an alternative analysis is required. I have also examined the two major approaches to phonological phrasing developed within the framework of prosodic phonology, namely the RBA and the EBA (including the original parameterized algorithms, the Align/Wrap Theory, and the Match Theory), and showed that neither of these two approaches alone can correctly define the phonological phrase in Fuzhou. On the basis of the discussion of previous studies and the examination of the RBA and the EBA, I have proposed a three-step definition of Fuzhou phonological phrase, which combines the RBA and the EBA, and have demonstrated that most relevant Fuzhou data discussed in previous studies can be re-analyzed and well accounted for with this alternative approach to phonological phrasing. In addition to the phonological phrase, another important concept is also introduced into the analysis of Fuzhou TS at the phrasal level, i.e., the Domain Impenetrability Condition (DIC) proposed in You (2018a). Two important issues regarding the phrasal-level application of Fuzhou TS, namely the blocking of TS between a verb and an NP indicating the location, and the contrast between monosyllabic verbs and disyllabic verbs, are discussed with the DIC. As we have seen, these two issues are actually typical examples of “no look-back” phenomena in the Fuzhou dialect, which are caused by the internal prosodic structure of relevant constructions and can be successfully solved by the DIC. In the following sections, I will discuss the phrasal-level application of another two major phonological rules in Fuzhou, FC and CL, and we will see that the three-step definition of Fuzhou phonological phrase and the “no look-back” device DIC are further substantiated.

5.2  Application of FC at the phrasal level The discussion in Section 5.1 reveals that the TS rule refers to the phonological phrase as its phrasal-level domain of application, but the application of TS is not the only phonological phenomenon applying within the phonological phrase domain in this dialect. In Section 5.2 and Section 5.3, I am going to examine the application of FC and CL at the phrasal level, and it will be shown that these two rules also make crucial reference to the domain of the phonological phrase defined in (59). Let us begin with the FC rule. FC is a tonally conditioned rule in Fuzhou, and it applies along with the application of TS obligatorily, as we have seen in previous chapters. Therefore, it is reasonable to assume that the application or blocking of

178  Phonological phrase in the Fuzhou dialect the FC rule has the same distribution with the TS rule at the phrasal level. In other words, FC applies within phrasal-level constructions in which TS applies as long as the syllables in the sandhi position contain Group B variants of alternating finals, while it is blocked whenever TS is blocked. Thus, we can expect that the FC rule also refers to the phonological phrase domain defined in (59) as its domain of application at the phrasal level and does not apply across the phonological phrase boundaries. Data from the Fuzhou dialect well support this assumption, as shown in (90)– (98), in which the finals in question are marked in bold.12 In the examples, “=” and “#” indicate the key positions where TS and FC are triggered and blocked respectively. For the sake of brevity, syntactic trees and formation steps of the phonological phrases are not presented. (90) Subject-predicate a. [店]φ [真 大]φ [taiŋ213] [tsiŋ44 tuai242] → [taiŋ213] # [tsiŋ51 nuai242] store very big ‘The store is very big.’

b. →

[野 俊]φ [裤]φ [khou213] [ʔia31 tsouŋ213] [khou213] # [ʔia44 ʒouŋ213] pants very beautiful ‘The pants are very beautiful.’

(91) Modifier-head (attributive-noun and adverbial-adjective/verb) a. [旧 b. [固 书]φ 食力]φ [kou242 tsy44] [kou213 siɛʔ5 liʔ5] 44 44 51 → [ku = ʒy ] → [ku = siɛʔ31 liʔ5] old book more tired ‘old book’ ‘more tired’ c. [硬 爬]φ [ŋaiŋ242 pa51] 44 → [ŋeiŋ = ma51] with difficulty climb ‘to climb with some difficulty’ (92) Verb-object (monosyllabic verb-object) a. [办]φ b. [新 护照]φ [paiŋ242] [siŋ44 xou242 tsiu213] → [paiŋ242] # [siŋ21 xu51 ʒiu213] → do new passport ‘to apply for a new passport’

[办 护照]φ [paiŋ242 xou242 tsiu213] 21 [peiŋ = xu51 ʒiu213] do passport ‘to apply for a new passport’

(93) Verb-resultative complement a. [办]φ [野 好]φ [paiŋ242] [ʔia31 xo31] → [paiŋ242] # [ʔia24 ʔo31] do very good ‘(somebody) did very well’

[碰]φ [呆]φ [phouŋ242] [ŋai51] h 242 [p ouŋ ] # [ŋai51] hit bad ‘to hit and break’

b. →

Phonological phrase in the Fuzhou dialect  179 c. →

[试]φ [完]φ [tshei213] [ʔuoŋ51] [tshei213] # [ʔuoŋ51] try finish ‘to finish trying’

(94) Determiner/numeral-classifier-noun13 a. [只 b. 块]φ [衣裳]φ [tsi31 toy213] [ʔi44 suoŋ51] → [tsi44 loy213]# [ʔi44 luoŋ51] → this Cl clothes ‘this piece of clothes’

[七 块]φ [衣裳]φ [tsheiʔ23 toy213] [ʔi44 suoŋ51] h 44 213 [ts iʔ = toy ]# [ʔi44 luoŋ51] seven Cl clothes ‘seven pieces of clothes’

(95) Serial verb construction a. [去 办 护照]φ [kho213 paiŋ242 xou242 tsiu213] → [kho21 peiŋ21 = xu51 ʒiu213] go do passport ‘to go to apply for a passport’ b. [去 店]φ [食 饭]φ [kho213 taiŋ213] [siɛʔ5 puoŋ242] → [kho51 taiŋ213] # [siɛʔ21 puoŋ242] go restaurant eat rice ‘to go to the restaurant to eat food’ (96) Double-object construction a. [乞]φ [我]φ [钱]φ [khøyʔ23] [ŋuai31] [tsiɛŋ51] → [khøyʔ23] # [ŋuai31] [tsiɛŋ51] give me money ‘to give me the money’ (97) Pivotal construction a. [逼]φ [我]φ [peiʔ23] [ŋuai31] → [peiʔ23] # [ŋuai31] force me ‘to force me to go’ (98) Passive construction a. [乞]φ [我]φ [khøyʔ23] [ŋuai31] → [khøyʔ23] # [ŋuai31] PASS me ‘to be hit by me’

[去]φ [kho213] [kho213] go

[拍]φ [phaʔ23] [phaʔ23] hit

b. →

[送]φ [我]φ [钱]φ [soyŋ213] [ŋuai31] [tsiɛŋ51] [soyŋ213] # [ŋuai31] [tsiɛŋ51] give me money ‘to give me the money’

180  Phonological phrase in the Fuzhou dialect Similar to the TS rule, the application of FC within the phonological phrase domain is also conditioned by the DIC, as exemplified in (99) and (100), in which the internal prosodic structure is presented. (99) V + location14 a. [[[坐]ω øC]CG [厅中]ω]φ [[[soy242] ] [thiaŋ44 touŋ44]] 242 → [[[soy ] ]# [thiaŋ44 nouŋ44]] sit drawing room ‘to sit in the drawing room’ b. [[[放]ω øC]CG [厅中]ω]φ [[[pouŋ213] ] [thiaŋ44 touŋ44]] 213 → [[[pouŋ ] ] # [thiaŋ44 nouŋ44]] put drawing room ‘to put in the drawing room’ (100) Disyllabic verb + object a. [[中意]ω [伊]ω]φ [[tøyŋ213 ʔei213] [ʔi44]] → [[tyŋ51 ʔei213] # [ʔi44]] like him ‘to like him’

b. →

[泅水]ω]φ [[兴趣]ω [[xeiŋ213 tshøy213] [siu51 tsui31]] [[xiŋ51 ʒøy213] # [siu31 ʒui31]] interest swimming ‘to be interested in swimming’

As we can see in (99) and (100), although all the strings in these examples form only one phonological phrase according to the definition in (59), FC does not apply to the marked syllables, following the requirement of the DIC. From the examples in (90)–(100), we can find that the FC rule in the Fuzhou dialect also refers to the phonological phrase domain defined in (59) as its domain of application at the phrasal level, similar to the TS rule. Like the TS rule, FC also applies within, but not across, the phonological phrase domain. Whenever tones 213, 242, and 23 are changed due to the application of TS within a phonological phrase, their corresponding Group B variants of alternating finals will undergo FC. Within the phonological phrase domain, the application of FC is also restricted by the DIC presented in (87), as can be seen in (99) and (100), where the application of FC in the phonological phrase domain cannot not “look back” and modify any properties within the embedded Type A clitic group domain or prosodic word domain.

5.3  Application of CL at the phrasal level Let us now move on to the application of CL at the phrasal level. It has been noticed that TS and FC apply within more phrasal-level constructions than CL does (Chen & Norman 1965a; Shih 1986). As mentioned in Section 5.1.1, Chen &

Phonological phrase in the Fuzhou dialect  181 Norman (1965a) distinguish four types of “junctures” in Fuzhou, i.e., terminal junctures, plus junctures, intermediate junctures, and close junctures. According to Chen & Norman, CL applies only at a close juncture, while TS (and FC) can apply at both intermediate junctures and close junctures. Hence, the difference between the application of CL at the phrasal level and that of TS and FC lies in the blocking of CL at the intermediate juncture. Among the four cases of intermediate juncture listed in (3), the most important case is actually the position between a monosyllabic verb and a monosyllabic object. According to the discussion in Section 5.1.4, it would be more accurate to describe this position as between a monosyllabic verb and a non-branching object. Since a monosyllabic verb and a non-branching object form a phonological phrase through φ-restructuring according to the definition in (59), it is reasonable to assume that CL is blocked within the phonological phrase that is constructed through φ-restructuring. This is well supported by the examples in (101)–(102), in which initials in question are marked in bold, and the positions where CL is blocked are marked by “#”. (101) Verb-Non-branching object (101) a. VP

b.

VP

V’ V

V’ NP

[食 [si 5 → [si 21 # *[si 21 = eat ‘to eat food’ c. VP

饭]φ (59c) puoŋ242] puoŋ242] βuoŋ242] rice

V [趁 [theiŋ213 → [thiŋ44 # *[thiŋ44 = earn ‘to earn money’ d. VP

V’ V [喜欢 [xi31 xuaŋ44 → [xi21 uaŋ44 # *[xi21 uaŋ44 = like ‘to like books’

NP 钱]φ (59c) tsi ŋ51] tsi ŋ51] ʒi ŋ51] money

V’ NP 书]φ (59c) tsy44] tsy44] ʒy44 ] book

V [看见 [khaŋ213 ki ŋ213 → [khaŋ51 ŋi ŋ213 # *[khaŋ51 ŋi ŋ213= see ‘to see him’

NP 伊]φ (59c) 44 i] ʔi44] ŋi44] he

182  Phonological phrase in the Fuzhou dialect (102) (102) Preposition-Non-branching object a. PP

b.

PP

P’ P

P’ NP

[向 北]φ (59c) [xyoŋ213 pøy 23] # pøy 23] → [xyoŋ51 = møy 23] *[xyoŋ51 towards north ‘towards the north’

P

NP

[朝 西]φ (59c) [tiu51 s 44] → [tiu44 # s 44] *[tiu44 = l 44] towards west ‘towards the west’

The examples in (101)–(102) show that CL is blocked not only between a monosyllabic verb and a non-branching object, as recorded in Chen & Norman (1965a), but also in the position between a disyllabic verb or a preposition and a non-branching object. Since each string in (101)–(102) forms one phonological phrase through φ-restructuring, we can find that CL is consistently blocked in such phonological phrases. Besides the cases in (101)–(102), the CL rule, like TS and FC, is also blocked across phonological phrase boundaries, even though the combination of sounds can trigger its application, as in (103)–(111). By contrast, in the phonological phrase domain constructed without φ-restructuring, CL applies, as can be seen in (103)–(111) as well. Key positions in the following examples are marked by “=” or “#”, and initials in question are marked in bold. For the sake of brevity, syntactic trees are not presented. (103) Subject-predicate a. [侬]φ [侈]φ [nøyŋ51] [sɛ242] 51 → [nøyŋ ] # [sɛ242] people many ‘There are many people.’

b. →

[真 贵]φ [书]φ [tsy44] [tsiŋ44 kui213] [tsy44] # [tsiŋ51 ŋui213] book very expensive ‘Books are very expensive.’

(104) Modifier-head (the head is monosyllabic) a. [旧 书]φ [kou242 tsy44] 44 → [ku = ʒy44] old book ‘old book’ c. [硬 爬]φ [ŋaiŋ242 pa51] 44 → [ŋeiŋ = ma51] with difficulty climb ‘to climb with some difficulty’

b. →

[真 贵]φ [tsiŋ44 kui213] 51 [tsiŋ = ŋui213] very expensive ‘very expensive’

Phonological phrase in the Fuzhou dialect  183 (105) Verb-branching object a. [看]φ [旧 书]φ [khaŋ213] [kou242 tsy44] h 213 44 → [k aŋ ] # [ku ʒy44] read old book ‘to read old books’ b. [办]φ [新 护照]φ [paiŋ242] [siŋ44 xou242 tsiu213] → [paiŋ242] # [siŋ21 xu51 ʒiu213] do new passport ‘to apply for a new passport’ (106) Verb-resultative complement a. [办]φ [野 好]φ [paiŋ242] [ʔia31 xo31] → [paiŋ242] # [ʔia24 ʔo31] do very good ‘(somebody) did very well’ c. [看]φ [完]φ [khaŋ213] [ʔuoŋ51] → [khaŋ213] # [ʔuoŋ51] read finish ‘to finish reading’ (107) Determiner/numeral-classifier-noun a. [只 b. 块]φ [糕]φ [tsi31 toy213] [ko44] → [tsi44 = loy213] # [ko44] → this Cl cake ‘this piece of cake’

b. →

[食]φ [饱]φ [siɛʔ5] [pa31] [siɛʔ5] # [pa31] eat full ‘to eat one’s fill’

[蜀 块]φ [suoʔ5 toy213] 21 [suoʔ = loy213] # one Cl ‘one piece of cake’

[糕]φ [ko44] [ko44] cake

(108) Serial verb construction a. [去 店]φ [食 饭]φ [kho213 taiŋ213] [siɛʔ5 puoŋ242] → [kho51 taiŋ213] # [siɛʔ21 puoŋ242] go restaurant eat rice ‘to go to the restaurant to eat food’ (109) Double-object construction a. [乞]φ [贼]φ [钱]φ [khøyʔ23] [tsheiʔ5] [tsiɛŋ51] → [khøyʔ23] # [tsheiʔ5] # [tsiɛŋ51] give thief money ‘to give the thief the money’

b. →

[送]φ [伊]φ [钱]φ [soyŋ213] [ʔi44] [tsiɛŋ51] [soyŋ213] # [ʔi44] # [tsiɛŋ51] give him money ‘to give him the money’

184  Phonological phrase in the Fuzhou dialect (110) Pivotal construction a. [劝]φ [伊]φ [khuoŋ213] [ʔi44] h 213 → [k uoŋ ] # [ʔi44] # advise him ‘to advise him to go’

b.

[去]φ [kho213] [kho213] go

(111) Passive construction [贼]φ [乞]φ [khøyʔ23] [tsheiʔ5] → [khøyʔ23] # [tsheiʔ5] # PASS thief ‘to be hit by the thief’

→

[干]φ [伊]φ [kaŋ44] [ʔi44] 44 [kaŋ ] # [ʔi44] # force him ‘to force him to go’

[去]φ [kho213] [kho213] go

[拍]φ [phaʔ23] [phaʔ23] hit

The above examples demonstrate that the CL rule also refers to the phonological phrase as its domain of application at the phrasal level, similar to TS and FC. The difference between the phrasal-level application of CL and that of TS and FC lies in their degree of application within the phonological phrase domain—TS and FC can be triggered within all phonological phrases, while CL only applies within phonological phrases that are not constructed through φ-restructuring. Thus, the examples mentioned in the end of Section 1.3.5.5 can be well explained, as presented in (112). (112) a. b.

做细 做细

tso213 sa213 → tso51 la213 tso213 sa213 → tso51 sa213

‘(in) childhood’ ‘to be a concubine’

We can find that 做细 ‘(in) childhood’ in (112a) is a noun while 做细 ‘to be a concubine’ in (112b) is a verb-object construction in which the object is nonbranching. According to the discussion in Chapter  3, (112a) forms a prosodic word, which serves as the lexical-level domain of application for CL. In contrast, (112b) forms a phonological phrase through φ-restructuring, similar to the examples in (101). This is why the CL rule applies in (112a) but does not in (112b). It is noteworthy that “no look-back” phenomena can also be found in the examples of the phrasal-level application of CL, as exemplified in (113). (113) a. →

c. →

[[新]ω [护照]ω]φ [[siŋ44] [xou242 tsiu213]] [[siŋ21] # [xu51 ʒiu213]] *[[siŋ21] [ŋu51 ʒiu213]] new passport ‘new passport’ [中意]ω]φ [[野]ω [[ʔia31] [tøyŋ213 ʔei213]] [[ʔia21] # [tyŋ51 ŋei213]] *[[ʔia21] [lyŋ51 ŋei213]] very like ‘to like (sth.) very much’

b. →

[[野]ω [[ʔia31] [[ʔia21] # *[[ʔia21] very ‘very happy’

[欢喜]ω]φ [xuaŋ44 xi31]] [xuaŋ51 ŋi31]] [ʔuaŋ51 ŋi31]] happy

Phonological phrase in the Fuzhou dialect  185 In (113a), CL is expected to apply within both the second prosodic word domain and the phonological phrase domain. However, after CL is triggered within the second prosodic word domain and the derivation moves to the phonological phrase domain, CL is blocked and fails to change the initial of the first syllable in the second prosodic word domain. The derivations in (113b) and (113c) are similar to that in (113a). The “no look-back” phenomena relevant to the phrasal-level application of CL in (113) can be nicely captured by the DIC. Take (113a) as an example. In (113a), within the second prosodic word domain, the structural description of the CL rule is met, and the initial of 照 [tsiu213] is changed. Then, the derivation moves to the phonological phrase domain, which constitutes the phrasal-level domain of application for CL. Within this domain, since the string of 护照 [xou242 tsiu213] has been spelled out as [xu51 ʒiu213], nothing inside this string can be further modified, and hence the surface form of (113a) is [siŋ21 xu51 ʒiu213] instead of [siŋ21 ŋu51 ʒiu213]. The “no look-back” phenomena in (113b) and (113c) can be accounted for in the same way.15 The above discussion in Section 5.3 reveals two outstanding characteristics of the phrasal-level application of CL. First, like TS and FC, CL does not apply across the phonological phrase domain boundaries. Second, unlike TS and FC, which are almost obligatory within the phonological phrase domain and are only restricted by the DIC, the application of the CL rule in the phonological phrase domain is subject to more restrictions. As we have seen from the examples in (101)–(113), although CL refers to the phonological phrase as its domain of application at the phrasal level, it is not only restricted by the DIC, as in (113), but also blocked within the phonological phrase domain constructed through φ-restructuring, as in (101) and (102). Therefore, the CL rule has a different degree of application within the phonological phrase domain as compared to TS and FC.

5.4 Summary In this chapter, I have discussed the application of TS, FC, and CL at the phrasal level in the Fuzhou dialect. According to the discussion in this chapter, all these three Fuzhou phonological rules make crucial reference to the phonological phrase domain as their phrasal-level domain of application. Based on the review of previous studies on the issue of Fuzhou TS at the phrasal level and the examination of both the EBA and the RBA approaches to phonological phrasing, I have proposed an alternative analysis and defined the phonological phrase in the Fuzhou dialect with a three-step definition, which allows a phonological phrase in Fuzhou to immediately dominate a prosodic word and thus violates the Exhaustivity constraint entailed in the SLH. This three-step definition is reproduced in (114). (114) Phonological phrase (φ) domain in the Fuzhou dialect = (59) a. Mark the right edge of every lexical head X, except where XP is an adjunct.

186  Phonological phrase in the Fuzhou dialect b. On the non-recursive side of the lexical head X, mark the right edge of the first phonetically overt head Y (if any; either lexical or functional) outside of XP; ωs/CGs that are separated by the right edge of X or Y belong to different φs. c. φ-restructuring: a non-branching φ, which is the first complement of X on its recursive side, is joined into the φ that contains X. This alternative approach to phonological phrasing is a combination of the EBA and the RBA, and it correctly defines the phonological phrase domain for the application of TS, FC, and CL at the phrasal level. Data from the Fuzhou dialect show that this alternative analysis is superior to previous studies and any analyses that are solely based on the EBA or the RBA, in that it covers more linguistic facts and offers greater predictive power. Although TS, FC, and CL can apply within the phonological phrase domain defined by the three-step definition in (114), these rules are not triggered within the phonological phrase domain without any restriction. We have found some “no look-back” phenomena in the phonological phrase in Fuzhou, which can be nicely handled with the DIC introduced in You (2018a), showing that the DIC should be treated as an integral part of the prosodic phonology of the Fuzhou dialect. Moreover, I have shown that TS, FC, and CL exhibit different degrees of application within the phonological phrase domain in the Fuzhou dialect. On the basis of the discussion in this chapter, the application of TS, FC, and CL within the phonological phrase domain can be summarized as in Table 5.1. As can be seen in Table 5.1, TS and FC are both obligatory within the phonological phrase domain (only restricted by the DIC). By contrast, CL is not triggered within all phonological phrases. If a phonological phrase is constructed through φ-restructuring, CL would be blocked within such a phonological phrase. Thus, we can find that not only is the phonological phrase, as the phrasal-level domain of application for these rules, established on the basis of syntactic notions, but also the different degrees of application of these rules within this domain are conditioned by the syntactic structure, which makes the phrasal-level application of TS, FC, and CL in Fuzhou a very typical example of the syntax-phonology interface. Table 5.1  Phonological rules and phonological phrase in the Fuzhou dialect Application

Rules

TS

FC

CL

Defined by (59/114a, b) (without φ-restructuring)

√

√

√

Defined by (59/114a, b, c) (with φ-restructuring)

√

√

×

Phonological phrase

Phonological phrase in the Fuzhou dialect  187

Notes 1 Today’s syntactic tree is quite different from that presented in (7). Here, I just follow Chan’s example. 2 For the sake of brevity, intermediate constituents between VP and NP, e.g., DP, are not presented in the tree in this chapter if they are not phonetically overt. 3 Thanks to Prof. Victor Junnan Pan (pers.comm.) for discussion. 4 As I have pointed out in Chapter 2, Hung’s (1987) definition of the foot in the Fuzhou dialect also wrongly refers to syntactic information (e.g., head, argument, modifier), similar to what we have seen in Shih (1986). 5 Chomsky (1986) defines m-command as follows: A  m-commands B iff A  does not dominate B and every maximal projection C that dominates A dominates B. A barrier is defined in Chomsky (ibid.) as follows: G is a barrier for B iff (a) or (b): (a) G immediately dominates D, D a blocking category for B; (b) G is a blocking category for B, and G is not IP (inflectional phrase). 6 For the sake of brevity, other constraints are not included in the discussion. 7 It has been noticed that whether TS applies in some verb-resultative complement examples in Fuzhou can indicate the difference in meaning. Liang (1983b) reported two such pairs: 抱倒 [po242 to31 → po242 to31] ‘to pull down’ vs. [po242 to31 → po51 to31] ‘to wrestle’ and 做死 [tso213 si31 → tso213 si31] ‘to work to death’ vs. [tso213 si31 → tso51 si31] ‘to act without shame’. We can see that the first example in each pair does not undergo TS while the second does. This can be nicely accounted for with the definition in (59). According to (59), similar to the examples in (63), the first example in each pair forms two phonological phrases, which is why TS is blocked between the verb and the resultative complement. In contrast, the second example in each pair is a prosodic word formed by a verb-complement compound and hence constitutes a domain of application for TS, according to the discussion in Chapter 3. 8 It is noteworthy that in (65a) and (65c), the right edge of the head P is marked according to (59a), while the right edge of the head P in (65b) is not since the entire PP in (65b) serves as the adjunct of the head V. Also note that φ-restructuring applies in (65a) and (65b) but not in (65c), since the objects of the head P in (65a) and (65b) are not branching, while the object of the head P in (65c) is branching. 9 乞 is a versatile word in the Fuzhou dialect, which means ‘to give’ as a verb and can also be used as the passive marker. 10 Though the little v is somewhat “less lexical” than V, a boundary is still inserted at the right edge of the little v in (70) since the v position is filled by the verb 乞 ‘to give’ that originates in the lexical V position.  11 In syntax, PRO is a phonetically empty pronoun that serves as the subject of a nonfinite clause. In (72), the PRO and 我 ‘me’ are coindexed since 我 ‘me’ is the real subject of the verb 去 ‘go’. 12 No examples of the preposition-object construction are presented in this section since none of the prepositions in Fuzhou contain Group B variants of alternating finals. 13 Since no determiner in Fuzhou contains Group B variants of alternating finals, there are no examples for the application of FC between the determiner and the classifier. 14 The symbol “ø” here stands for the omitted or soundless PVP. 15 One informant suggests that [ʔia21 lyŋ51 ŋei213] may be acceptable for (113c) as well, but she believes that [ʔia21 tyŋ51 ŋei213] is more accurate and authentic. I  will return to this example in Chapter 7 and discuss the variability in the acceptability of such examples.

6 Proclitics and the Type B clitic group in the Fuzhou dialect

As mentioned in Chapter  1 and Chapter  4, clitics in the Fuzhou dialect can be further divided into proclitics and enclitics, and there are consequently two types of clitic groups in this dialect, namely the Type A clitic group and the Type B clitic group. This chapter will focus on proclitics and the Type B clitic group composed of “proclitic + host” in this dialect. I will start with the investigation into the morphosyntactic functions of Fuzhou proclitics in Section 6.1 and show that these elements, like Fuzhou enclitics, share some common properties of clitics across languages. I will then examine the phonological behavior of the Type B clitic group in Section 6.2. I will show that the Type B clitic group in the Fuzhou dialect exhibits different phonological behavior as compared to the prosodic word and the phonological phrase. We will see that although the Type B clitic group differs from the Type A clitic group in terms of their phonological behavior, the clitic group domain in Fuzhou as a whole is distinct from both the prosodic word domain and the phonological phrase domain, which provides further evidence for the existence of the clitic group in this dialect. Then, I present a discussion in Section 6.3 on the violation of the SLH observed in the cases of the Type B clitic group and again argue for a slight weakening of the SLH. Finally, Section 6.4 closes this chapter with some concluding remarks.

6.1 Proclitics in the Fuzhou dialect and their morphosyntactic functions It has been noticed that the Fuzhou dialect has some proclitic-like elements, such as the progressive aspect marker 𠲥 [lɛ31] and auxiliary verbs 有 [ʔou242] ‘to have, to own’ and 是 [sei242] ‘to be’, as well as prepositions 将 [tsuoŋ44] (similar to ba 把 in Mandarin Chinese) and 跟 [kyŋ44] ‘with’ (cf. Chen & Norman 1965a; Wright 1983; Shih 1986; Chen 1998; Li & Liang 2001, among others). In the following subsections in Section 6.1, I will discuss the morphosyntactic functions of these proclitics. In the examples, the proclitic is labeled with “C”; the group of “proclitic + host” with “CG”; the prosodic word host, with “ω”; and the phonological phrase host, with “φ”. For the sake of brevity, only the citation/underlying segmental structure is presented in the examples in this section.

Fuzhou proclitics and the Type B clitic group  189 6.1.1  Progressive aspect marker 𠲥 [lɛ31] As discussed in Chapter 4, 𠲥 in the Fuzhou dialect can be used as an enclitic that has multiple morphosyntactic roles including the durative aspect marker, perfective aspect marker, post-verbal particle, and the locative marker. Besides, 𠲥 can also be used preceding a verb or verb phrase to indicate an ongoing action (cf. Zheng 1988b; Chen 1998; Li  & Liang 2001; Li 2002, among others). Like its enclitic counterpart 𠲥 [lɛ0], the proclitic 𠲥 [lɛ31] never occurs on its own. The proclitic 𠲥 [lɛ31] differs from the enclitic 𠲥 [lɛ0] in that it has to attach to the verb or verb phrase on its right, rather than on its left. In addition to their direction of attachment, the proclitic 𠲥 [lɛ31] is also different from the enclitic 𠲥 [lɛ0] in terms of the tone—it carries a shang tone instead of a neutral tone. Examples of the progressive aspect marker 𠲥 [lɛ31] are presented in (1). It is noteworthy that the Layeredness constraint is violated in examples like (1b) here and (2c) in Section 6.1.2. Take (1b) as an example: the two prosodic words 洗 and 碗 form a phonological phrase according to the discussion in Chapter 5, and the proclitic 𠲥 takes the phonological phrase as its host. (1)

a.

伊 [𠲥C [睏]ω]CG ʔi [lɛ [khouŋ]] he PROG sleep ‘He is sleeping.’

b.

[[洗]ω [碗]ω]φ]CG 伊 [𠲥C ʔi [lɛ [[sɛ] [ʔuaŋ]]] he PROG wash bowl ‘He is washing dishes.’

6.1.2  Auxiliary verbs 是 [sei242] and 有 [ʔou242] 是 [sei242] and 有 [ʔou242] are two commonly used auxiliary verbs in the Fuzhou dialect. These two are slightly different from other Fuzhou clitics in that they can be used alone when answering questions. Nevertheless, they are treated as proclitics in this study for the following reasons: (a) they belong to a functional category, i.e., the auxiliary verb; (b) their semantic meanings are “less concrete” than verbs that indicate actions; and (c) the group of “是/有 + host” exhibits similar phonological behavior as other groups composed of the proclitic plus the host in the Fuzhou dialect, which can distinguish such groups from other prosodic constituents in this dialect, as will be demonstrated in Section 6.2. Let us take a look at 是 [sei242] first. Like be in English, the proclitic 是 [sei242] is a copula that links nominal and/or pronominal elements. It attaches to the left of the adjacent element, as exemplified in (2). (2)

a.

c.

嚽 [是C [豆]ω]CG b. 嚽 [是C [葡萄]ω]CG tsui [sei [tau]] tsui [sei [pu to]] this be bean this be grape ‘This is a bean.’ ‘This is a grape.’ 伊 [是C [[好]ω [侬]ω]φ]CG ʔi [sei [[xo] [nøyŋ]]] he be good people ‘He is a nice person.’

190  Fuzhou proclitics and the Type B clitic group Compared with 是 [sei242], the auxiliary verb 有 [ʔou242] is more versatile. The basic meaning of the proclitic 有 [ʔou242] is ‘to have, to own’, and thus it can be used to indicate possession, as exemplified in (3). In addition to the basic meaning, 有 [ʔou242] is usually used before a verb or verb phrase to emphasize the existence of an action or an event in the Fuzhou dialect (cf. Zheng 1985; Li et al 1994, among others). In this case, the proclitic 有 [ʔou242] is similar to English do/did that is used before the verb and expresses the idea of ‘indeed, really’, as exemplified in (4). (3)

a.

我 [有C [事计]ω]CG ŋuai [ʔou [tai kiɛ]] I have affair ‘I have something to do.’

(4)

a.

伊 [有C [去]ω]CG ʔi [ʔou [kho]] he have go ‘He did go.’ 只 间 店 暝晡 [有C tsi kaŋ taiŋ maŋ puo [ʔou this Cl store evening have ‘This store is open at night.’ 伊 下昼 [有C [[看]ω [书]ω]φ]CG ʔi ʔa tau [ʔou [[khaŋ] [tsy]]] he afternoon have read book ‘He did do some reading in the afternoon.’

b.

c.

b.

伊 [有C [[好]ω [书]ω]φ]CG ʔi [ʔou [[xo] [tsy]]] he have good book ‘He has good books.’

[开]ω]CG [khui]] open

6.1.3 Prepositions 将 [tsuoŋ44] and 跟 [kyŋ44] 将 [tsuoŋ44] and 跟 [kyŋ44] have been recorded as clitics in some previous studies (Wright 1983; Shih 1986). Neither of them can be used alone, and both have to attach to the constituent on the right. 将 [tsuoŋ44] is similar to ba 把 in Mandarin Chinese. It is used to introduce a noun or noun phrase that is the object of the following verb or verb phrase. The noun or noun phrase following 将 [tsuoŋ44] is the receiver of the action denoted by the verb or verb phrase. This object noun or noun phrase is “disposed” or “affected” in the event described, like its counterpart in the ba construction (cf. Wang 1954; Chao 1968; Huang et al. 2009, among others), as exemplified in (5).1 (5)

a.

b.

汝 [将C 完 [饭]ω]CG 食 ny [tsuoŋ [puoŋ]] siɛʔ ʔuoŋ you tsuoŋ rice eat finish ‘You, finish the rice.’ 伊 [将C 排 敆 [书桌]ω]CG 厅中 ʔi [tsuoŋ [tsy toʔ]] pɛ kaʔ thiaŋ touŋ he tsuoŋ desk put PVP drawing room ‘He put the desk in the drawing room.’

Fuzhou proclitics and the Type B clitic group  191 c.

我 [将C [[蜀]ω [头]ω]φ]CG ŋuai [tsuoŋ [[suoʔ] [thau]]] I tsuoŋ one Cl ‘I hit and killed a worm.’

虫 thøyŋ worm

拍 phaʔ hit

死 si die

去 kho PERF

The preposition 跟 [kyŋ44] in the Fuzhou dialect is similar to gen 跟 in Mandarin Chinese, which means ‘with’. It is used before a noun or noun phrase to express the idea of ‘doing something together with somebody’, as in (6).2 (6)

a.

我 [跟C [伊]ω]CG 齐 来 ŋuai [kyŋ [ʔi]] tsɛ li I with he together come ‘I will come together with him.’ b. 我 [跟C [依妹]ω]CG 齐 行 ŋuai [kyŋ [ʔi mui]] tsɛ kiaŋ I with younger sister together walk ‘I will walk together with the younger sister.’ c. 伊 [跟C [[两]ω [隻]ω]φ]CG 先生 齐 ʔi [kyŋ [[laŋ] [tsiɛʔ]]] siŋ saŋ tsɛ he with two Cl teacher together ‘He will go together with two teachers.’

去 kho go

6.1.4  Recursive clitic group with proclitics There are some cases of Fuzhou proclitics in which the violation of Nonrecursivity can be observed, as exemplified in (7). (7)

a.

伊 [有C [跟C [我]ω]CG]CG 齐 去 ʔi [ʔou [kyŋ [ŋuai]]] tsɛ kho he have with I together go ‘He did go with me.’ b. 我 [有C [将C [[只]ω [件]ω]φ]CG]CG 事计 交代 伊 ŋuai [ʔou [tsuoŋ [[tsi] [ʔyoŋ]]]] tai kiɛ kau tai ʔi I have tsuoŋ this Cl affair hand over he ‘I did hand over this work to him.’

6.1.5 Summary On the basis of the above discussion and examples presented in Section 6.1, we can find that those proclitic-like elements in the Fuzhou dialect, similar to enclitics in this dialect, share some of the most common morphosyntactic properties of clitics across languages: (a) they all belong to functional categories; (b) most of them never occur as the only element of an utterance and must attach to the adjacent prosodic unit on the right; (c) the meaning of the string of the proclitic plus the host is predictable from the meaning of the host and that of the proclitic; and (d) they can attach to material already containing the affix, as in (6b), or the proclitic, as in (7). Therefore, according to the discussion of clitics across languages in Chapter 4, we have enough reason to assume that these elements in the Fuzhou

192  Fuzhou proclitics and the Type B clitic group dialect are proclitics. According to the definition of the clitic group domain in the Fuzhou dialect proposed in Chapter 4, the group of the proclitic plus the host also forms a type of clitic group (Type B clitic group) in this dialect.

6.2 Phonological phenomena within the Type B clitic group in the Fuzhou dialect Now, let us proceed to the investigation of the phonological behavior of the Type B clitic group in the Fuzhou dialect. Through the examination of the application of three Fuzhou phonological rules, TS, FC, and CL, we will see in the following subsections that some phonological phenomena refer crucially to the domain of the Type B clitic group. It will also be shown that although the Type B clitic group in Fuzhou exhibits different phonological behavior as compared to the Type A clitic group, the clitic group in this dialect as a whole can be distinguished from both the prosodic word and the phonological phrase in terms of their roles as the domain of rule application. This further demonstrates that there are phonological phenomena characteristic only of the domain of the clitic group, but not in any other context. 6.2.1  TS within the Type B clitic group Different from enclitics in the Fuzhou dialect, proclitics in this dialect are invariably incorporated into a TS domain with the host they attach to, as has been observed in some previous studies (Wright 1983; Shih 1986). Specifically, the proclitic within a Type B clitic group obligatorily undergoes TS as long as the appropriate phonological context is provided. Empirical evidence from the Fuzhou dialect supports this observation, as exemplified in (8)–(11), in which sandhi tones in question are marked in bold. (8)

Application of TS in “progressive aspect marker 𠲥 [lɛ31] + host” a. 伊 [𠲥C b. 伊 [𠲥C [睏]ω]CG [[洗]ω [碗]ω]φ]CG ʔi44 [lɛ31 [khouŋ213]] ʔi44 [lɛ31 [[sɛ31] [ʔuaŋ31]]] 44 44 h 213 44 21 → ʔi [lɛ [k ouŋ ]] → ʔi [lɛ [[sɛ24] [ʔuaŋ31]]] he PROG sleep he PROG wash bowl ‘He is sleeping.’ ‘He is washing dishes.’

(9)

Application of TS in “auxiliary verb + host”3 I. Auxiliary verb 是 [sei242] + host a. 嚽 [是C [豆]ω]CG b. tsui51 [sei242 [tau242]] → tsui51 [si51 [tau242]] → this be bean ‘This is a bean.’ c. 伊 [是C [[好]ω [侬]ω]φ]CG ʔi44 [sei242 [[xo31] [nøyŋ51]]] → ʔi44 [si21 [[xo21] [nøyŋ51]]] he be good people ‘He is a nice person.’

嚽 [是C [葡萄]ω]CG tsui51 [sei242 [pu51 to51]] tsui51 [si51 [pu31 lo51]] this be grape ‘This is a grape.’

Fuzhou proclitics and the Type B clitic group  193 II.

Auxiliary verb 有 [ʔou242] + host a. 我 [有C [事计]ω]CG ŋuai31 [ʔou242 [tai242 kiɛ213]] 31 21 → ŋuai [ʔu [tai51 ʔiɛ213]] I have affair ‘I have something to do.’ b. 伊 [有C [[好]ω [书]ω]φ]CG ʔi44 [ʔou242 [[xo31] [tsy44]]] → ʔi44 [ʔu21 [[xo21] [ʒy44]]] he have good book ‘He has good books.’ c. 伊 [有C [去]ω]CG ʔi44 [ʔou242 [kho213]] → ʔi44 [ʔu51 [kho213]] he have go ‘He did go.’ d. 只 间 店 暝晡 [有C [开]ω]CG tsi31 kaŋ44 taiŋ213 maŋ51 puo44 [ʔou242 [khui44]] → tsi21 ʔaŋ44 taiŋ213 maŋ44 muo44 [ʔu44 [khui44]] this Cl store evening have open ‘This store is open at night.’ e. 伊 下昼 [有C [[看]ω [书]ω]φ]CG ʔi44 ʔa242 tau213 [ʔou242 [[khaŋ213] [tsy44]]] → ʔi44 ʔa51 lau213 [ʔu21 [[khaŋ44] [tsy44]]] he afternoon have read book ‘He did do some reading in the afternoon.’

(10) Application of TS in “preposition + host”4 I. Preposition 将 [tsuoŋ44] + host a. 汝 [将C [饭]ω]CG 食 完 siɛʔ5 ʔuoŋ51 ny31 [tsuoŋ44 [puoŋ242]] → ny31 [tsuoŋ51 [puoŋ242]] siɛʔ5 ʔuoŋ51 you tsuoŋ rice eat finish ‘You, finish the rice.’ b. 伊 [将C [书桌]ω]CG 排 敆 ʔi44 [tsuoŋ44 [tsy44 toʔ23]] pɛ51 kaʔ0 → ʔi44 [tsuoŋ21 [tsy51 loʔ23]] pɛ51 ʔaʔ0 he tsuoŋ desk put PVP ‘He put the desk in the drawing room.’ c. 我 [将C [[蜀]ω [头]ω]φ]CG 虫 ŋuai31 [tsuoŋ44 [[suoʔ5] [thau51]]] thøyŋ51 → ŋuai31 [tsuoŋ51 [[suoʔ31] [lau51]]] thøyŋ51 I tsuoŋ one Cl worm ‘I hit and killed a worm.’

厅中 thiaŋ44 touŋ44 thiaŋ44 nouŋ44 drawing room 拍 phaʔ23 phaʔ23 hit

死 si31 si31 die

去 kho0 ʔo0 PERF

194  Fuzhou proclitics and the Type B clitic group II.

Preposition 跟 [kyŋ44] + host a. 伊 [跟C [我]ω]CG 齐 来 ʔi44 [kyŋ44 [ŋuai31]] tsɛ51 li51 → ʔi44 [kyŋ51 [ŋuai31]] tsɛ31 li51 he with I together come ‘He will come together with me.’ b. 我 [跟C [依妹]ω]CG 齐 行 ŋuai31 [kyŋ44 [ʔi44 mui213]] tsɛ51 kiaŋ51 → ŋuai31 [kyŋ21 [ʔi51 mui213]] tsɛ31 ʔiaŋ51 I with younger sister together walk ‘I will walk together with the younger sister.’ c. 伊 [跟C [[两]ω [隻]ω]φ]CG 先生 齐 ʔi44 [kyŋ44 [[laŋ242] [tsiɛʔ23]]] siŋ44 saŋ44 tsɛ51 → ʔi44 [kyŋ21 [[laŋ51] [ʒiɛʔ23]]] siŋ44 naŋ44 tsɛ21 he with two Cl teacher together ‘He will go together with two teachers.’

(11) Application of TS in recursive clitic group with proclitics a. 伊 [有C [跟C [我]ω]CG]CG 齐 ʔi44 [ʔou242 [kyŋ44 [ŋuai31]]] tsɛ51 44 21 51 31 → ʔi [ʔu [kyŋ [ŋuai ]]] tsɛ21 he have with I together ‘He did go with me.’ b. 我 [有C [将C [[只]ω [件]ω]φ]CG]CG ŋuai31 [ʔou242 [tsuoŋ44 [[tsi31] [ʔyoŋ242]]]] → ŋuai31 [ʔu21 [tsuoŋ21 [[tsi44] [ʔyoŋ242]]]] I have tsuoŋ this Cl 交代 伊 kau44 tai242 ʔi44 → kau51 lai242 ʔi44 hand over he ‘I did hand over this work to him.’

去 kho213 ʔo213 go

去 kho213 ʔo213 go 事计 tai242 kiɛ213 tai51 ʔiɛ213 affair

The examples in (8)–(11) show that TS obligatorily applies in the domain formed by the Type B clitic group in the Fuzhou dialect with no exceptions. To be specific, TS is triggered between the proclitic and the host, thus changing the tone of the proclitic. Hence, one of the most outstanding phonological properties of the Type B clitic group in the Fuzhou dialect is the obligatory application of TS within the domain. Since TS is obligatory between the proclitic and the host CG-internally, the tone of a Fuzhou proclitic is always dependent on the tones of syllables in the host, which shows that Fuzhou proclitics, like Fuzhou enclitics, are also phonologically dependent. 6.2.2  FC within the Type B clitic group As a tonally conditioned rule, the application/blocking of the FC rule in the Fuzhou dialect within different prosodic domains exhibits a similar distribution

Fuzhou proclitics and the Type B clitic group  195 as the TS rule. Hence, it is no wonder that FC obligatorily applies within the domain formed by the Type B clitic group in which the application of TS is obligatory. This can be exemplified in (12)–(13).5 Sandhi finals in question are marked in bold. (12) Application of FC in “auxiliary verb + host”6 I. Auxiliary verb 是 [sei242] + host a. 嚽 [是C [豆]ω]CG b. tsui51 [sei242 [tau242]] → tsui51 [si51 [tau242]] → this be bean ‘This is a bean.’ c. 伊 [是C [[好]ω [侬]ω]φ]CG ʔi44 [sei242 [[xo31] [nøyŋ51]]] → ʔi44 [si21 [[xo21] [nøyŋ51]]] he be good people ‘He is a nice person.’ II.

嚽 [是C [葡萄]ω]CG tsui51 [sei242 [pu51 to51]] tsui51 [si51 [pu31 lo51]] this be grape ‘This is a grape.’

Auxiliary verb 有 [ʔou242] + host a. 我 [有C [事计]ω]CG ŋuai31 [ʔou242 [tai242 kiɛ213]] → ŋuai31 [ʔu21 [tai51 ʔiɛ213]] I have affair ‘I have something to do.’ b. 伊 [有C [[好]ω [书]ω]φ]CG ʔi44 [ʔou242 [[xo31] [tsy44]]] 44 21 → ʔi [ʔu [[xo21] [ʒy44]]] he have good book ‘He has good books.’ c. 伊 [有C [去]ω]CG ʔi44 [ʔou242 [kho213]] → ʔi44 [ʔu51 [kho213]] he have go ‘He did go.’ d. 只 间 店 暝晡 [有C [开]ω]CG tsi31 kaŋ44 taiŋ213 maŋ51 puo44 [ʔou242 [khui44]] → tsi21 ʔaŋ44 taiŋ213 maŋ44 muo44 [ʔu44 [khui44]] this Cl store evening have open ‘This store is open at night.’ e. 伊 下昼 [有C [[看]ω [书]ω]φ]CG ʔi44 ʔa242 tau213 [ʔou242 [[khaŋ213] [tsy44]]] → ʔi44 ʔa51 lau213 [ʔu21 [[khaŋ44] [tsy44]]] he afternoon have read book ‘He did do some reading in the afternoon.’

196  Fuzhou proclitics and the Type B clitic group (13) Application of FC in recursive clitic group with proclitics a. 伊 [有C [跟C [我]ω]CG]CG 齐 去 [ʔou242 [kyŋ44 [ŋuai31]]] tsɛ51 kho213 ʔi44 → ʔi44 [ʔu21 [kyŋ51 [ŋuai31]]] tsɛ21 ʔo213 he have with I together go ‘He did go with me.’ b. 我 [有C [将C [[只]ω [件]ω]φ]CG]CG 事计 ŋuai31 [ʔou242 [tsuoŋ44 [[tsi31] [ʔyoŋ242]]]] tai242 kiɛ213 → ŋuai31 [ʔu21 [tsuoŋ21 [[tsi44] [ʔyoŋ242]]]] tai51 ʔiɛ213 I have tsuoŋ this Cl affair 交代 伊 kau44 tai242 ʔi44 → kau51 lai242 ʔi44 hand over he ‘I did hand over this work to him.’ The examples in (12)–(13) demonstrate another distinctive phonological property of the Type B clitic group in the Fuzhou dialect, namely the obligatory application of the FC rule between the proclitic and the host inside the Type B clitic group domain. 6.2.3  CL within the Type B clitic group Unlike the Type A clitic group, the Type B clitic group in the Fuzhou dialect does not form the domain of application for the CL rule. To be more specific, the initial of the first syllable of the host never undergoes CL, as can be seen from the examples in (14)–(17). The position where CL is blocked is indicated by the symbol “#”, and initials in question are marked in bold. (14) Blocking of CL in “progressive aspect marker 𠲥 [lɛ31] + host” a. 伊 [睏]ω]CG [𠲥C ʔi44 [lɛ31 [khouŋ213]] 44 44 → ʔi [lɛ # [khouŋ213]] 44 44 *ʔi [lɛ [ʔouŋ213]] he PROG sleep ‘He is sleeping.’ b. 伊 [𠲥C [[洗]ω [碗]ω]φ]CG ʔi44 [lɛ31 [[sɛ31] [ʔuaŋ31]]] 44 21 24 → ʔi [lɛ # [[sɛ ] [ʔuaŋ31]]] 44 21 24 *ʔi [lɛ [[lɛ ] [ʔuaŋ31]]] he PROG wash bowl ‘He is washing dishes.’

Fuzhou proclitics and the Type B clitic group  197 (15) Blocking of CL in “auxiliary verb + host” I. Auxiliary verb 是 [sei242] + host a. 嚽 [是C [豆]ω]CG tsui51 [sei242 [tau242]] 51 51 → tsui [si # [tau242]] 51 51 *tsui [si [lau242]] this be bean ‘This is a bean.’ b. 嚽 [是C [葡萄]ω]CG tsui51 [sei242 [pu51 to51]] 51 51 → tsui [si # [pu31 lo51]] 51 51 *tsui [si [βu31 lo51]] this be grape ‘This is a grape.’ c. 伊 [是C [[好]ω [侬]ω]φ]CG ʔi44 [sei242 [[xo31] [nøyŋ51]]] 44 21 → ʔi [si # [[xo21] [nøyŋ51]]] 44 21 *ʔi [si [[ʔo21] [nøyŋ51]]] he be good people ‘He is a nice person.’ II.

Auxiliary verb 有 [ʔou242] + host a. 我 [有C [事计]ω]CG ŋuai31 [ʔou242 [tai242 kiɛ213]] 31 21 → ŋuai [ʔu # [tai51 ʔiɛ213]] 31 21 *ŋuai [ʔu [lai51 ʔiɛ213]] I have affair ‘I have something to do.’ b. 伊 [有C [[好]ω [书]ω]φ]CG ʔi44 [ʔou242 [[xo31] [tsy44]]] 44 21 → ʔi [ʔu # [[xo21] [ʒy44]]] 44 21 *ʔi [ʔu [[ʔo21] [ʒy44]]] he have good book ‘He has good books.’ c. 伊 [有C [去]ω]CG ʔi44 [ʔou242 [kho213]] → ʔi44 [ʔu51 # [kho213]] *ʔi44 [ʔu51 [ʔo213]] he have go ‘He did go.’ d. 只 间 店 暝晡 [有C tsi31 kaŋ44 taiŋ213 maŋ51 puo44 [ʔou242 → tsi21 ʔaŋ44 taiŋ213 maŋ44 muo44 [ʔu44 *tsi21 ʔaŋ44 taiŋ213 maŋ44 muo44 [ʔu44 this Cl store evening have ‘This store is open at night.’

#

[开]ω]CG [khui44]] [khui44]] [ʔui44]] open

198  Fuzhou proclitics and the Type B clitic group e. →

伊 下昼 [有C [[看]ω ʔi44 ʔa242 tau213 [ʔou242 [[khaŋ213] ʔi44 ʔa51 lau213 [ʔu21 # [[khaŋ44] *ʔi44 ʔa51 lau213 [ʔu21 [[ʔaŋ44] he afternoon have read ‘He did do some reading in the afternoon.’

[书]ω]φ]CG [tsy44]]] [tsy44]]] [tsy44]]] book

(16) Blocking of CL in “preposition + host” I. Preposition 将 [tsuoŋ44] + host a. 汝 [将C [饭]ω]CG 食 完 ny31 [tsuoŋ44 [puoŋ242]] siɛʔ5 ʔuoŋ51 → ny31 [tsuoŋ51 # [puoŋ242]] siɛʔ5 ʔuoŋ51 31 *ny [tsuoŋ51 [muoŋ242]] siɛʔ5 ʔuoŋ51 you tsuoŋ rice eat finish ‘You, finish the rice.’ b. 伊 [将C [书桌]ω]CG 排 敆 厅中 ʔi44 [tsuoŋ44 [tsy44 toʔ23]] pɛ51 kaʔ0 thiaŋ44 touŋ44 → ʔi44 [tsuoŋ21 # [tsy51 loʔ23]] pɛ51 ʔaʔ0 thiaŋ44 nouŋ44 44 *ʔi [tsuoŋ21 [ʒy51 loʔ23]] pɛ51 ʔaʔ0 thiaŋ44 nouŋ44 he tsuoŋ desk put PVP drawing room ‘He put the desk in the drawing room.’ c. [将C [[蜀]ω [头]ω]φ]CG 虫 拍 死 去 [tsuoŋ44 [[suoʔ5] [thau51]]] thøyŋ51 phaʔ23 si31 kho0 → [tsuoŋ51 # [[suoʔ31] [lau51]]] thøyŋ51 phaʔ23 si31 ʔo0 *[tsuoŋ51 [[nuoʔ31] [lau51]]] thøyŋ51 phaʔ23 si31 ʔo0 tsuoŋ one Cl worm hit die PERF ‘(Someone) hit and killed a worm.’ II.

Preposition 跟 [kyŋ44] + host a. 我 [跟C [伊]ω]CG 齐 来 ŋuai31 [kyŋ44 [ʔi44]] tsɛ51 li51 31 44 44 31 → ŋuai [kyŋ # [ʔi ]] tsɛ li51 31 44 44 31 *ŋuai [kyŋ [ŋi ]] tsɛ li51 I with he together come ‘I will come together with him.’ b. 我 [跟C [依妹]ω]CG 齐 行 ŋuai31 [kyŋ44 [ʔi44 mui213]] tsɛ51 kiaŋ51 31 21 51 213 31 → ŋuai [kyŋ # [ʔi mui ]] tsɛ ʔiaŋ51 31 21 51 213 31 *ŋuai [kyŋ [ŋi mui ]] tsɛ ʔiaŋ51 I with younger sister together walk ‘I will walk together with the younger sister.’ c. [跟C [[两]ω [隻]ω]φ]CG 先生 齐 去 [kyŋ44 [[laŋ242] [tsiɛʔ23]]] siŋ44 saŋ44 tsɛ51 kho213 → [kyŋ21 # [[laŋ51] [ʒiɛʔ23]]] siŋ44 naŋ44 tsɛ21 ʔo213 *[kyŋ21 [[naŋ51] [ʒiɛʔ23]]] siŋ44 naŋ44 tsɛ21 ʔo213 with two Cl teacher together go ‘(Someone) goes together with two teachers.’

Fuzhou proclitics and the Type B clitic group  199 (17) Blocking of CL in recursive clitic group with proclitics a. 我 [有C [跟C [伊]ω]CG]CG 齐 去 ŋuai31 [ʔou242 [kyŋ44 [ʔi44]]] tsɛ51 kho213 31 21 51 44 21 → ŋuai [ʔu # [kyŋ # [ʔi ]]] tsɛ ʔo213 31 21 51 44 21 *ŋuai [ʔu [ʔyŋ [ŋi ]]] tsɛ ʔo213 I have with he together go ‘I did go with him.’ b. [有C [将C [[只]ω [件]ω]φ]CG]CG 事计 交代 [[tsi31] [ʔyoŋ242]]]] tai242 kiɛ213 kau44 tai242 [ʔou242 [tsuoŋ44 → [ʔu21 # [tsuoŋ21 # [[tsi44] [ʔyoŋ242]]]] tai51 ʔiɛ213 kau51 lai242 *[ʔu21 [ʒuoŋ21 [[ʒi44] [ʔyoŋ242]]]] tai51 ʔiɛ213 kau51 lai242 have tsuoŋ this Cl affair hand over ‘(Someone) did hand over this work to him.’

伊 ʔi44 ʔi44 ʔi44 he

From examples in (14)–(17), we can find that CL is consistently blocked in the domain of the Type B clitic group—it never applies to the initial of the first syllable of the host to which a proclitic attaches to. Hence, the mandatory blocking of CL can be considered as the third distinctive phonological property of the Type B clitic group. 6.2.4  Summary and discussion According to the discussion in Section 6.2, the application/blocking of TS, FC, and CL in the Type B clitic group domain in the Fuzhou dialect can be summarized as in Table 6.1. Comparing Table 6.1 with Table 4.1 in Chapter 4, we can find that the Type A and Type B clitic groups in the Fuzhou dialect exhibit quite different phonological properties. As discussed in Chapter 4, within the Type A clitic group domain, both TS and FC are obligatorily blocked between the host and the enclitic, while CL is consistently triggered. In contrast, in the Type B clitic group domain, both TS and FC obligatorily apply between the proclitic and the host as long as there is an applicable phonological context, while CL is consistently blocked. Although the Type A  and Type B clitic groups have different phonological behavior in terms of the application and blocking of TS, FC, and CL, the clitic group domain in Fuzhou as a whole can be distinguished from both the prosodic

Table 6.1  Phonological rules and the Type B clitic group in the Fuzhou dialect Application Type B clitic group Progressive aspect marker + host Auxiliary verb (是 [sei242]/有 [ʔou242]) + host Preposition (将 [tsuoŋ44]/跟 [kyŋ44]) + host Recursive clitic group with proclitics

Rules

TS

FC

CL

√ √ √ √

N/A √ N/A √

× × × ×

200  Fuzhou proclitics and the Type B clitic group word domain and the phonological phrase domain. From Chapter 3 to Chapter 6, we have examined the application and blocking of TS, FC, and CL with respect to the three prosodic domains located in the morphosyntax-based part of the prosodic hierarchy, namely the prosodic word, the clitic group, and the phonological phrase. The findings can be briefly summarized as in Table 6.2, in which “√/×” indicates that the rule does not apply within all the subtypes of a given domain. We can see that these three rules behave differently within different ­morphosyntax-based prosodic domains, and their application/blocking is affected by different factors. In the prosodic word domain, although TS, FC, and CL can all apply within the domain, they are not obligatory in all the subtypes of the prosodic word domain. The application and blocking of these rules in the prosodic word domain is conditioned by morphological processes of morphosyntactic word formation in the Fuzhou dialect, as discussed in Chapter 3. In the phonological phrase domain, although the application of both TS and FC is compulsory, the application of CL is not. As we have seen in Chapter 5, the application of the CL rule within the phonological phrase domain is determined by the syntactic structure—it applies only within a subtype of the phonological phrase, namely the phonological phrase that is not constructed through φ-restructuring. By contrast, the application of these three rules in the clitic group domain is decided by the direction in which the clitic attaches to its host, which means that the difference in the phonological properties between Type A and Type B clitic groups is simply due to the difference in attachment direction between enclitics and proclitics. Within either the Type A or Type B clitic group, nevertheless, the application/blocking of the phonological rules is very clear-cut, with no exceptions or options. One might argue that the Type B clitic group is similar to the phonological phrase formed through φ-restructuring, in which the CL rule is also blocked, and hence the Type B clitic group should be considered as a subtype of the phonological phrase. However, there are phonological disparities between the Type B clitic group and the phonological phrase regarding the application of TS and FC in constructions containing branching object nouns/noun phrases. As mentioned in Note 3 and Note 4, the auxiliary verbs 是 [sei242] and 有 [ʔou242] as well as the prepositions 将 [tsuoŋ44] and 跟 [kyŋ44] undergo TS whenever they

Table 6.2  Phonological rules and morphosyntax-based domains in the Fuzhou dialect Application

Rules TS

FC

CL

√/×

√/×

√/×

Type A clitic group (host + enclitic)

×

×

√

Type B clitic group (proclitic + host)

√

√

×

√

√

√/×

Morphosyntax-based prosodic domains Prosodic word Clitic group

Phonological phrase

Fuzhou proclitics and the Type B clitic group  201 are followed by an object noun/noun phrase, whether the object is branching or not. In contrast, when regular monosyllabic verbs like 食 [siɛʔ5] ‘to eat’ and regular monosyllabic prepositions like 向 [xyoŋ213] ‘towards’ precede a branching object, the TS rule fails to apply to these verbs and prepositions since they are not grouped into the same phonological phrase with the following branching object. Furthermore, as discussed in Note 6, 是 [sei242] and 有 [ʔou242] also undergo FC when followed by an object noun/noun phrase, regardless of the internal branchingness of the object, whereas regular monosyllabic verbs are not incorporated into the same phonological phrase with the following branching object and thus do not undergo FC. Therefore, in terms of the phonological behavior, the Type B clitic group is still quite dissimilar from the phonological phrase and thus should not be treated as a subtype of the phonological phrase. Based on the above discussion, we can find that the three morphosyntax-based prosodic domains in the Fuzhou dialect, namely the prosodic word, the phonological phrase, and the clitic group (including Type A and Type B), can be distinguished from each other in terms of their roles as the domain of rule application. These three prosodic domains, though all belong to the morphosyntax-based hierarchy, are constructed by appealing to different types of information (morphological processes for ω, syntactic structure for φ, attachment direction of clitics for CG) and hence exhibit different phonological properties.

6.3  Violation of SLH within the Type B clitic group We have seen in Chapter 4 that the SLH in the theory of prosodic phonology is violated in some cases of the Type A clitic group in Fuzhou. Both the constraints of Nonrecursivity and Layeredness can be violated since an enclitic can attach to not only a prosodic word, but also a Type A clitic group or a phonological phrase. Similarly, there are also examples of the Type B clitic group in Fuzhou in which the SLH is violated, as we have seen in Section 6.1 and Section 6.2. Some examples of the violation of Nonrecursivity are presented in (18). (18) a. → b. →

→

伊 [有C [跟C [我]ω]CG]CG 齐 ʔi44 [ʔou242 [kyŋ44 [ŋuai31]]] tsɛ51 44 21 51 31 ʔi [ʔu [kyŋ [ŋuai ]]] tsɛ21 he have with I together ‘He did go with me.’ 我 [有C [将C [[只]ω [件]ω]φ]CG]CG ŋuai31 [ʔou242 [tsuoŋ44 [[tsi31] [ʔyoŋ242]]]] ŋuai31 [ʔu21 [tsuoŋ21 [[tsi44] [ʔyoŋ242]]]] I have tsuoŋ this Cl 交代 伊 kau44 tai242 ʔi44 kau51 lai242 ʔi44 hand over he ‘I did hand over this work to him.’

去 kho213 ʔo213 go 事计 tai242 kiɛ213 tai51 ʔiɛ213 affair

202  Fuzhou proclitics and the Type B clitic group The examples in (18) demonstrate that a Type B clitic group in the Fuzhou dialect, which is composed of “proclitic + host”, may dominate another Type B clitic group. Take (18a) as an example. In the inner Type B clitic group, the proclitic 跟 [kyŋ44] attaches to the prosodic word [我]ω ‘I’ as the host, and the proclitic 有 [ʔou242] attaches to the entire inner Type B clitic group [跟我]CG as the host in the outer Type B clitic group. In a similar way, the proclitic 将 [tsuoŋ44] in (18b) attaches to the phonological phrase [只件]φ ‘this piece’ as the host in the inner Type B clitic group, and the inner Type B clitic group is dominated by the outer Type B clitic group, in which the proclitic 有 [ʔou242] takes the inner Type B clitic group [将只件]CG as the host. In addition to the above examples, we can also find some examples in which a Type B clitic group in Fuzhou dominates a Type A clitic group, as presented in (19). (19) a.

嚽 [是C [[我]ω 其C]CG]CG tsui51 [sei242 [[ŋuai31] ki0]] → tsui51 [si51 [[ŋuai31] ʔi0]] this be I POSS ‘This is mine.’ b. 伊 [有C [[买]ω 过C]CG]CG ʔi44 [ʔou242 [[mɛ31] kuo213]] → ʔi44 [ʔu51 [[mɛ31] ʔuo213]] he have buy EXP ‘He did have bought (something).’ c. 伊 [将C [[依妹]ω 其C]CG]CG 拈 ʔi44 [tsuoŋ44 [[ʔi44 mui213] ki0]] niɛŋ44 44 21 51 213 0 → ʔi [tsuoŋ [[ʔi mui ] ʔi ]] niɛŋ44 he tsuoŋ younger sister POSS take ‘He took away the younger sister’s (things).’

走 tsau31 tsau31 go

去 kho0 ʔo0 PERF

Take (19a) as an example. The enclitic 其 [ki0] attaches to the prosodic word [我]ω ‘I’ as the host, and then the group of “我 + 其” forms a Type A  clitic group, in which CL applies between the host and the enclitic. Then, the proclitic 是 [sei242] attaches to the Type A  clitic group [我其]CG as the host and the group of “是 + [我其]CG” forms a Type B clitic group, in which TS and FC apply to the proclitic. The formation process of the prosodic structure in (19b) and (19c) is the same as that in (19a). In these examples, the constraint of Nonrecursivity is clearly violated. In Chapter 4, we have seen that the constraint of Layeredness can be violated in Fuzhou Type A clitic group. The violation of this constraint is also allowed in the cases of the Type B clitic group in Fuzhou, as can be found in some examples presented in Section 6.1 and Section 6.2. Take a look at the following examples in (20).

Fuzhou proclitics and the Type B clitic group  203 (20) a. → b → c. → d. → e. →

→

伊 [𠲥C [[洗]ω [碗]ω]φ]CG ʔi44 [lɛ31 [[sɛ31] [ʔuaŋ31]]] 44 21 ʔi [lɛ [[sɛ24] [ʔuaŋ31]]] he PROG wash bowl ‘He is washing dishes.’ 伊 [是C [[好]ω [侬]ω]φ]CG ʔi44 [sei242 [[xo31] [nøyŋ51]]] ʔi44 [si21 [[xo21] [nøyŋ51]]] he be good people ‘He is a nice person.’ 伊 [有C [[好]ω [书]ω]φ]CG ʔi44 [ʔou242 [[xo31] [tsy44]]] ʔi44 [ʔu21 [[xo21] [ʒy44]]] he have good book ‘He has good books.’ 伊 下昼 [有C [[看]ω [书]ω]φ]CG ʔi44 ʔa242 tau213 [ʔou242 [[khaŋ213] [tsy44]]] ʔi44 ʔa51 lau213 [ʔu21 [[khaŋ44] [tsy44]]] he afternoon have read book ‘He did do some reading in the afternoon.’ 我 [有C [将C [[只]ω [件]ω]φ]CG]CG ŋuai31 [ʔou242 [tsuoŋ44 [[tsi31] [ʔyoŋ242]]]] ŋuai31 [ʔu21 [tsuoŋ21 [[tsi44] [ʔyoŋ242]]]] I have tsuoŋ this Cl 交代 伊 kau44 tai242 ʔi44 kau51 lai242 ʔi44 hand over he ‘I did hand over this work to him.’

事计 tai242 kiɛ213 tai51 ʔiɛ213 affair

According to the discussion in Chapter  5, phrasal-level units such as [洗碗] ‘wash dishes’, [好侬] ‘nice person’, [好书] ‘good books’, [看书] ‘read books’, and [只件] ‘this piece’ are all phonological phrases. Hence, the proclitic(s) in each example in (20) attach to the phonological phrase as the host, and the phonological phrase host is dominated by the Type B clitic group, which violates the constraint of Layeredness. The examples of the Fuzhou Type B clitic group presented in (18)–(20) thus pose a challenge to the original SLH, similar to what we have seen in the cases of the Fuzhou Type A clitic group. If we adopt a weakened SLH and allow the constraints of Nonrecursivity and Layeredness to be violated in a given language like the Fuzhou dialect, the violation of these two constraints in (18)–(20) would not be a problem at all. Once again, there is no need to exclude the clitic group from the prosodic hierarchy. The problem caused by the attachment of proclitics

204  Fuzhou proclitics and the Type B clitic group to constituents higher than the prosodic word in the Fuzhou dialect can be easily accounted for by slightly weakening the original SLH, without any undesirable theoretical consequences.

6.4 Summary This chapter presents an investigation of Fuzhou proclitics and the Type B clitic group from the perspectives of morphosyntactic functions and phonological behavior. From the discussion in this chapter, we have seen the following properties of proclitics and the Type B clitic group in the Fuzhou dialect. (21) Properties of proclitics in the Fuzhou dialect a. Fuzhou proclitics all belong to functional categories. b. Most Fuzhou proclitics never occur as the only element of an utterance and must attach to the adjacent constituent (ω, CG, or φ) on the right as the host. c. The meaning of the string of the proclitic plus the host is predictable from the meaning of the proclitic and that of the host. d. Fuzhou proclitics can attach to material already containing the affix or the proclitic. e. Fuzhou proclitics are phonologically dependent—the tone of a proclitic is always decided by the tones of following syllables. (22) Properties of the Type B clitic group in the Fuzhou dialect a. TS and FC obligatorily apply between the proclitic and the host within the Type B clitic group domain. b. CL is obligatorily blocked between the proclitic and the host within the Type B clitic group domain. On the one hand, we can find from (21) that Fuzhou proclitics share common morphosyntactic and phonological properties with clitics across languages, including Fuzhou enclitics, and thus should not be treated as prefixes or independent words. On the other hand, we can see from (22) that the Type B clitic group in Fuzhou exhibits different phonological behavior as compared to the Type A clitic group with respect to the application of TS, FC, and CL. Nonetheless, as I have argued in Section 6.2.4, the clitic group in this dialect as a whole can be distinguished from the prosodic word and the phonological phrase. Therefore, the clitic group (including both Type A and Type B) should be maintained and established as an indispensable prosodic constituent in the prosodic phonology of the Fuzhou dialect. Moreover, as we have seen from the examples in Chapter 4 and this chapter, both the Type A clitic group and the Type B clitic group in the Fuzhou dialect can dominate a prosodic constituent higher than the prosodic word. This shows that Nonrecursivity and Layeredness are both violable in this dialect, which constitutes a great challenge to the original SLH. This challenge can be solved by

Fuzhou proclitics and the Type B clitic group  205 assuming that the violation of these two constraints is allowed in the Fuzhou dialect, which further substantiates the idea that a weakened SLH is required in the theory of prosodic phonology. Based on the discussion in Chapter 4 and Chapter 6, the domain formation of the clitic group in the Fuzhou dialect can be formally presented as in (23). (23) Clitic group (CG) formation in the Fuzhou dialect The domain of the CG in the Fuzhou dialect consists of one independent (i.e., non-clitic) prosodic constituent (ω, CG, or φ), plus any adjacent: a. directional clitic(s), or b. plain clitic/nondirectional clitic such that there is no possible host with which they share more category memberships.

Notes 1 It is noteworthy that a Fuzhou Type B clitic group is not necessarily isomorphic to any morphosyntactic structures. For example, [将蜀头]CG in (5c) does not correspond to any syntactic constituent because the head noun [虫] ‘worm’ in this example is excluded from the Type B clitic group. 2 Notice that 跟 [kyŋ44] in (6b) attaches to the prosodic word 依妹 ‘younger sister’ that already contains an affix 依-. 3 Notice that 是 [sei242] and 有 [ʔou242] undergo TS no matter what kind of object noun they are followed by. They undergo TS when followed either by a non-branching object, such as (9Ia, b) and (9IIa), or by a branching object, such as (9Ic) and (9IIb). Recall that a regular monosyllabic verb like 食 [siɛʔ5] ‘to eat’ never undergoes TS when followed by a branching object, because in that case the branching object is not able to be incorporated into the same phonological phrase with the verb, as discussed in Chapter 5. This implies that (a) auxiliary verbs 是 [sei242] and 有 [ʔou242] are different from regular verbs like 食 [siɛʔ5] in terms of the phonological behavior and that (b) the group composed of the auxiliary verb 是 [sei242] or 有 [ʔou242] and the following object is different from the group composed of a regular verb like 食 [siɛʔ5] plus its object. 4 As shown in Chapter 5, in a prepositional phrase, the tone of the regular monosyllabic preposition like 向 [xyoŋ213] ‘towards’ is usually incorporated into the same tone sandhi domain with that of the following non-branching object and thus undergoes TS, while it does not undergo TS when it is followed by a branching object. This is due to the fact that the branching object is not incorporated into the same phonological phrase with the preposition. By contrast, prepositions 将 [tsuoŋ44] and 跟 [kyŋ44] undergo TS no matter what kind of object noun phrases they are followed by, as exemplified in (10). This indicates that prepositions 将 [tsuoŋ44] and 跟 [kyŋ44] behave more like auxiliary verbs 是 [sei242] and 有 [ʔou242] in the Fuzhou dialect with respect to their phonological behavior, while different from regular prepositions. Hence, it would be more reasonable to treat 将 [tsuoŋ44] and 跟 [kyŋ44] as proclitics rather than regular prepositions. 5 Since the progressive aspect marker 𠲥 [lɛ31] and the prepositions 将 [tsuoŋ44] and 跟 [kyŋ44] do not contain Group B variants of alternating finals, no examples of these proclitics are presented in Section 6.2.2. 6 Notice that in the examples in (12), 是 [sei242] and 有 [ʔou242] both undergo the application of FC when followed by an object noun/noun phrase, whether the object is branching or not, which again distinguishes these two proclitics from regular monosyllabic verbs we have seen in Chapter 5.

7 Intonational phrase in the Fuzhou dialect Variability in domain formation and rule application

This chapter investigates the intonational phrase in the Fuzhou dialect, which is the last prosodic constituent discussed in this book. As the domain of an intonation contour, the formation and the restructuring of the intonational phrase are influenced by semantic and pragmatic factors such as length, rate of speech, style, and contrastive prominence, exhibiting a large degree of variability. Based on the introduction of the intonational phrase in general, Section 7.1 discusses the definition and the restructuring of the intonational phrase in the Fuzhou dialect. Section 7.2 probes phonological phenomena that are relevant to the domain of the intonational phrase in Fuzhou. It will be demonstrated that it is the semantic and pragmatic factors that create the great degree of variability in the rule application within the intonational phrase domain, which distinguishes the intonational phrase from the other prosodic constituents in this dialect. Section 7.3 summarizes the findings in this chapter.

7.1 Definition and restructuring of the intonational phrase in the Fuzhou dialect 7.1.1  Definition of the intonational phrase in the Fuzhou dialect The intonational phrase, which is located above the phonological phrase in the prosodic hierarchy, is defined by Selkirk (1978) as the domain over which an intonation contour is spread. Nespor & Vogel (1986) also suggest that the intonational phrase is the domain of an intonation contour and that the ends of intonational phrases always coincide with the positions where grammar-related pauses may be introduced in a sentence. Thus, a root sentence in syntactic representation, which is defined as an S (“sentence”) that is not dominated by a node other than S (Emonds 1976), corresponds to an intonational phrase regardless of its length, as exemplified in (1). (1)

a. [Lions are dangerous]ι b. [Jennifer discovered that her attic had been invaded last winter by a family of squirrels]ι

Intonational phrase in the Fuzhou dialect  207 In addition to root sentences, a number of specific types of constructions obligatorily form intonational phrases, such as those marked in bold in (2) (Nespor & Vogel 1986; Jensen 1993). Nespor & Vogel point out that these constructions are all elements that are linearly represented but not structurally attached to the sentence tree (also cf. Safir 1985). They also suggest that the same types of constructions will form intonational phrases in all languages that make use of these constructions. As can be seen in the following examples, strings adjacent to the obligatorily formed intonational phrase must also automatically form intonational phrases on their own.1 (2)

a. [Lions]ι [as you know]ι [are dangerous]ι   (parenthetical expression) b. [My brother]ι [who absolutely loves animals]ι  [just bought himself an exotic tropical bird]ι (nonrestrictive relative clause) c. [That’s Theodore’s cat]ι [isn’t it?]ι        (tag question) d. [Clarence]ι [I’d like you to meet Mr. Smith]ι    (vocative) e. [Good heavens]ι [there’s a bear in the back yard]ι  (expletive) f. [Duncan and Mary]ι [our next-door neighbors]ι   [have a lovely Labrador retriever]ι             (appositive) g. [They are so cute]ι [those Australian koalas]ι   (right dislocation)

On the basis of the observation presented above, the intonational phrase is defined in Nespor & Vogel (1986) as in (3). (3)

Intonational phrase (IPh) domain An Intonational Phrase domain may consist of a. all the φs in a string that is not structurally attached to the sentence tree at the level of s-structure, or b. any remaining sequence of adjacent φs in a root sentence.

Since the intonational phrase in Fuzhou can directly dominate not only the phonological phrase, but also the clitic group, as will be evidenced by some examples in this chapter, the general definition of the intonational phrase in (3) can be adapted and formulated as in (4), by assuming that the Exhaustivity constraint entailed in the SLH is violable in this dialect. (4)

Intonational phrase (IPh) domain in the Fuzhou dialect An Intonational Phrase domain may consist of a. all the CGs and φs in a string that is not structurally attached to the sentence tree at the level of s-structure, or b. any remaining sequence of adjacent CGs/φs in a root sentence.

According to (4), root sentences in Fuzhou form independent intonational phrases because they are associated with intonation contours, as exemplified in (5).

208  Intonational phrase in the Fuzhou dialect (5)

a. → b. → c. → d. →

[侬 真 侈]ι [nøyŋ51 tsiŋ44 sɛ242] [nøyŋ51 tsiŋ51 nɛ242] people very many ‘There are a lot of people.’ [汝 其 衣裳 野 俊]ι [ny31 ki0 ʔi44 suoŋ51 ʔia31 tsouŋ213] [ny31 ʔi0 ʔi44 luoŋ51 ʔia44 ʒouŋ213] you POSS clothes very beautiful ‘Your clothes are very beautiful.’ [伊 𠲥 洗 碗]ι [ʔi44 lɛ31 sɛ31 ʔuaŋ31] [ʔi44 lɛ21 sɛ24 ʔuaŋ31] he PROG wash bowl ‘He is washing dishes.’ [许 隻 侬 生 𠲥 两 [xi31 tsiɛʔ23 nøyŋ51 saŋ44 lɛ0 laŋ242 [xi44 ʒiɛʔ23 nøyŋ51 saŋ44 nɛ01 laŋ51 that Cl people give birth PERF two ‘That person gave birth to two children.’

隻 tsiɛʔ23 ʒiɛʔ23 Cl

囝]ι kiaŋ31] kiaŋ31] child

There are also some syntactic constructions in Fuzhou that can obligatorily form intonational phrases, including parenthetical expressions, vocatives, expletives, appositives, and certain moved elements, as illustrated in (6). These sentences are grammatical only if pauses are inserted on one or both sides of the marked constructions. Similar to what we have seen in (2), strings adjacent to the marked constructions in (6) also automatically form intonational phrases on their own. (6)

a.

[汝 会 仈 其]ι [许 隻 侬]ι [xi31 tsiɛʔ23 nøyŋ51] [ny31 ʔa242 paiʔ23 ki0] → [xi44 ʒiɛʔ23 nøyŋ51] [ny31 ʔa51 paiʔ23 ʔi0] that Cl people you can know NOM [生 𠲥 两 隻 囝]ι [saŋ44 lɛ0 laŋ242 tsiɛʔ23 kiaŋ31] → [saŋ44 nɛ01 laŋ51 ʒiɛʔ23 kiaŋ31] give birth PERF two Cl child ‘That person, as you know, gave birth to two children.’ (parenthetical expression)

[陈 先生]ι [汝 其 衣裳 野 俊]ι [tiŋ51 siŋ44 saŋ44] [ny31 ki0 ʔi44 suoŋ51 ʔia31 tsouŋ213] → [tiŋ21 siŋ44 naŋ44] [ny31 ʔi0 ʔi44 luoŋ51 ʔia44 ʒouŋ213] Chen teacher you POSS clothes very beautiful ‘Mr. Chen, your clothes are very beautiful.’ (vocative) b.

Intonational phrase in the Fuzhou dialect  209 [当歹 去]ι [水缸 碰 必 去]ι [touŋ44 tai213 kho0] [tsui31 kouŋ44 phouŋ242 peiʔ23 kho0] 51 213 0 21 44 h 242 23 → [touŋ nai ʔo ] [tsui ʔouŋ p ouŋ peiʔ ʔo0] terrible PERF water jar hit crack PERF ‘Damn! The water jar was hit and developed a crack.’ (expletive) d. [我 其 骹溜帮]ι [陈 先生]ι [真 本事]ι [ŋuai31 ki0 kha44 liu51 pouŋ44] [tiŋ51 siŋ44 saŋ44] [tsiŋ44 puoŋ31 søy242] → [ŋuai31 ʔi0 kha44 liu44 βouŋ44] [tiŋ21 siŋ44 naŋ44] [tsiŋ21 puoŋ44 nøy242] I POSS friend Chen teacher very capable ‘My friend, Mr. Chen, is very capable.’ (appositive) e. [生 野 俊]ι [许 隻 诸娘囝]ι [saŋ44 ʔia31 tsouŋ213] [xi31 tsiɛʔ23 tsy44 noyŋ51 kiaŋ31] → [saŋ44 ʔia44 ʒouŋ213] [xi44 ʒiɛʔ23 tsy51 noyŋ31 ŋiaŋ31] grow very beautiful that Cl girl ‘She is very beautiful, that girl.’ (right dislocation) c.

7.1.2  Restructuring of the intonational phrase in the Fuzhou dialect As pointed out by Nespor & Vogel (1986), some semantic and pragmatic factors are involved in the formation of the intonational phrase domain, including length, rate of speech, style, and contrastive prominence, which affect the ultimate number of intonational phrases within an utterance. This results in the restructuring of the intonational phrase and the relatively large degree of variability in the formation of intonational phrases, which distinguishes the intonational phrase from the other prosodic constituents since no other constituents allow for this type of variability in terms of their domain formation. In the restructuring of the intonational phrase, the length plays such a role that the longer the original intonational phrase, the more likely it is to be divided into smaller intonational phrases. There seems to be a tendency to avoid series of very short intonational phrases or sequences of intonational phrases of very different lengths. Hence, a long intonational phrase like (1b) could be divided into smaller intonational phrases as in (7a) and (7b), while the divisions in (7c) and (7d) are less acceptable. (7)

a. [Jennifer discovered]ι [that her attic had been invaded last winter]ι [by a family of squirrels]ι b. [Jennifer discovered that her attic]ι [had been invaded last winter by a family of squirrels]ι c. ? [Jennifer]ι [discovered]ι [that her attic]ι [had been invaded]ι [last winter]ι [by a family]ι [of squirrels]ι (series of very short ιs) d. ? [Jennifer]ι [discovered that her attic had been invaded last winter by a family of squirrels]ι (sequences of ιs of very different lengths)

The restructuring of the intonational phrase caused by the length is quite common in the Fuzhou dialect. Let us take a look at the sentences in (8a) and (9a).

210  Intonational phrase in the Fuzhou dialect Although both sentences can form long intonational phrases as in (8b) and (9b) respectively, the divisions in (8c) and (9c) are more natural and more acceptable to native speakers. (8)

a.

伊 讲 做 贼 其 许 隻 侬 ʔi44 kouŋ31 tso213 tsheiʔ5 ki0 xi31 tsiɛʔ23 nøyŋ51 → ʔi44 kouŋ31 tso44 tsheiʔ5 ʔi0 xi44 ʒiɛʔ23 nøyŋ51 he speak do thief MOD that Cl people 昨暝 敆 街中 乞 侬 拍 去 soʔ5 maŋ51 kaʔ5 kɛ44 touŋ44 khøyʔ23 nøyŋ51 phaʔ23 kho0 31 51 21 44 44 → soʔ maŋ kaʔ kɛ louŋ khøyʔ23 nøyŋ51 phaʔ23 ʔo0 yesterday in on the street PASS people hit PERF ‘He said that that person who was a thief was hit by others on the street yesterday.’ b. [伊讲做贼其许隻侬昨暝敆街中乞侬拍去]ι c. [伊讲]ι [做贼其许隻侬]ι [昨暝敆街中]ι [乞侬拍去]ι

(9)

a.

礼拜 早头 我 跟 爸奶 齐 去 lɛ31 pai213 tsa31 thau51 ŋuai31 kyŋ44 pa242 nɛ31 tsɛ51 kho213 44 213 21 51 31 21 51 31 21 → lɛ βai tsa lau ŋuai kyŋ pa nɛ tsɛ ʔo213 Sunday morning I with parents together go 许 间 店 买 新 衣裳 xi31 kaŋ44 taiŋ213 mɛ31 siŋ44 ʔi44 suoŋ51 21 44 213 31 21 → xi ʔaŋ taiŋ mɛ siŋ ʔi44 luoŋ51 that Cl store buy new clothes ‘I went together with my parents to that store on Sunday morning to buy new clothes.’ b. [礼拜早头我跟爸奶齐去许间店买新衣裳]ι c. [礼拜早头]ι [我跟爸奶齐去许间店]ι [买新衣裳]ι

The rate of speech is another key factor in ι-restructuring—the faster the rate of speech, the longer the intonational phrases of a given utterance tend to be, and vice versa, which can be evidenced by the comparison of the speech of my informants. The speech of the informant aged 85 is slower than that of the other two informants, who are around 55 years old, and more pauses are introduced in the older informant’s speech, especially when producing a long utterance. Hence, a sentence like (5d) is divided into more than one intonational phrase in the older informant’s speech, as in (10a), while the same sentence forms only one intonational phrase in the speech of the other two informants, as in (10b). (10) a. b.

[许隻侬]ι [生𠲥两隻囝]ι [许隻侬生𠲥两隻囝]ι ‘That person gave birth to two children.’

Since a formal speech is often given at a slow tempo, the more formal the style of speech, the slower the rate of speech tends to be. This is why the style of

Intonational phrase in the Fuzhou dialect  211 speech is also an important factor in determining the restructuring of the intonational phrase—a sentence uttered as part of a formal presentation tends to contain more than one intonation contours, while a sentence with the similar length and syntactic structure uttered in an informal colloquial way will contain a single intonational phrase. For example, the two Fuzhou sentences in (11) have the same length and syntactic structure, but the formal one in (11a) can be divided into three small intonational phrases, while the less formal one in (11b) forms one single intonational phrase when uttered in a colloquial way. (11)

a.

[我 希望]ι [各 隻 选手]ι [保持 [ŋuai31 xi31 ʔuoŋ242] [koʔ23 tsiɛʔ23 souŋ31 tshiu31] [po31 thi51 → [ŋuai31 xi44 ʔuoŋ242] [koʔ44 tsiɛʔ23 souŋ24 ʒiu31] [po21 thi51 I hope every Cl contestant keep ‘I hope that every contestant can behave sportingly.’ b. [我 见觉 许 隻 先生 兴趣 [ŋuai31 kiɛŋ213 koyʔ23 xi31 tsiɛʔ23 siŋ44 saŋ44 xeiŋ213 tshøy213 → [ŋuai31 kiɛŋ51 ŋoyʔ23 xi44 ʒiɛʔ23 siŋ44 naŋ44 xiŋ51 ʒøy213 I think that Cl teacher interest ‘I think that that teacher is interested in swimming.’

风度]ι xuŋ44 tou242] xuŋ51 nou242] deportment 泅水]ι siu51 tsui31] siu31 ʒui31] swimming

Another key factor in the process of ι-restructuring is the contrastive prominence. Bing (1979) distinguishes contrastive prominence from contrastive stress and argues that contrastive prominence is not predictable by rule and the speaker may add it where he or she wishes. Unlike contrastive stress, which does not require a modification of ι structure, the assignment of contrastive prominence on a particular constituent in the string will lead to the addition of an intonation contour within the string. Let us examine the examples in (12) from Nespor & Vogel (1986). We can see that (12a) is assigned one intonational phrase, while the single intonational phrase in (12b) can be restructured as three smaller intonational phrases as in (12c). The sentences in (12a) and (12b) have the same syntactic structure, but the presence of the pronouns in (12b) requires the listener to co-refer the pronouns to the nouns and assign contrastive prominence to the pronouns, which creates a different interpretation from that of (12a). (12) a. b. c.

[Paul called Paula before Carla called Carl]ι [Paul called Paula before she called him]ι [Paul called Paula]ι [before she]ι [called him]ι

Examples from Fuzhou further support the idea that contrastive prominence may trigger the ι-restructuring. For instance, the sentence in (13) can form one intonational phrase, as in (13a), but would be restructured as two smaller intonational phrases in (13b) if prominence is placed on the pronoun 伊 ‘him’. Similarly, the sentence in (14) forms one intonational phrase in (14a), while it is broken down into two intonational phrases when the noun 诸娘囝 ‘girl’ is assigned prominence, as in (14b).

212  Intonational phrase in the Fuzhou dialect (13) a. → b. (14) a.

[我 看见 伊 下昼 有 看 书]ι [ŋuai31 khaŋ213 kiɛŋ213 ʔi44 ʔa242 tau213 ʔou242 khaŋ213 tsy44] [ŋuai31 khaŋ51 ŋiɛŋ213 ʔi44 ʔa51 lau213 ʔu21 khaŋ44 tsy44] I see he afternoon have read book ‘I saw him do some reading in the afternoon.’ [我看见伊]ι [下昼有看书]ι ‘I saw HIM (but not others) do some reading in the afternoon.’

[侬家 班 其 诸娘囝 野 [naŋ51 ka44 paŋ44 ki0 tsy44 noyŋ51 kiaŋ31 ʔia31 → [naŋ44 ŋa44 paŋ44 ŋi0 tsy51 noyŋ31 ŋiaŋ31 ʔia21 we class POSS girl very ‘Girls in our class are very smart.’ b. [侬家班其诸娘囝]ι [野聪明]ι ‘GIRLS (but not boys) in our class are very smart.’

聪明]ι tshuŋ44 miŋ51] tshuŋ44 miŋ51] smart

Besides the above-mentioned semantic and pragmatic factors, it has also been noticed that the presence of lists usually causes ι-restructuring in languages. Consider the possible restructuring of intonational phrases in sentences containing lists presented in (15). We can find that each of the items in a list can form an independent intonational phrase since pauses are usually introduced on both sides of each item. (15)

a. →

→ b. →

→

c. →

→

[伊 是 蜀 隻]ι [好疼]ι [聪明]ι [ʔi44 sei242 suoʔ5 tsiɛʔ23] [tshuŋ44 miŋ51] [xo31 thiaŋ213] [ʔi44 si51 suoʔ21 ʒiɛʔ23] [tshuŋ44 miŋ51] [xo44 liaŋ213] she be one Cl smart cute [听嘴 其 诸娘囝]ι [thiaŋ44 tshui213 ki0 tsy44 noyŋ51 kiaŋ31] [thiaŋ51 ʒui213 ʔi0 tsy51 noyŋ31 ŋiaŋ31] obedient MOD girl ‘She is a smart, cute, obedient girl.’ [包 粽]ι [挂 艾]ι [做 香包]ι [pau44 tsoyŋ213] [kua213 ŋiɛ213] [tso242 xyoŋ44 pau44] [pau51 tsoyŋ213] [kua51 ŋiɛ213] [tso21 xyoŋ44 mau44] wrap rice dumpling hang argyi (Chinese mugwort) do scent bag [扒 龙船]ι [是 五月节 其 风俗]ι [pa51 luŋ51 suŋ51] [sei242 ŋu242 ŋuoʔ5 tsaiʔ23 ki0 xuŋ44 syʔ5] [pa31 luŋ31 nuŋ51] [si21 ŋu21 ŋuoʔ5 tsaiʔ23 ʔi0 xuŋ44 nyʔ5] row dragon boat be Dragon Boat Festival POSS custom ‘Making rice dumplings, hanging argyi grass, making scent bags, and racing dragon boats are customs of the Dragon Boat Festival.’ [福州 有 三山]ι [于山]ι [乌石山]ι [xuʔ23 tsiu44 ʔou242 saŋ44 saŋ44] [ʔy44 saŋ44] [ʔu44 suoʔ5 saŋ44] [xuʔ21 tsiu44 ʔu21 san44 naŋ44] [ʔy44 laŋ44] [ʔu21 luoʔ44 laŋ44] Fuzhou have Three Mountains Mt. Yu Mt. Wushi [共 屏山]ι [koyŋ242 piŋ51 saŋ44] [koyŋ242 piŋ44 naŋ44] and Mt. Ping ‘Fuzhou has Three Mountains: Mt. Yu, Mt. Wushi, and Mt. Ping.’

Intonational phrase in the Fuzhou dialect  213 The cases of lists can be well handled by the rule of List Restructuring proposed by Nespor & Vogel (1986), as presented in (16), according to which an intonational phrase boundary is inserted before each item in the list. (16) List Restructuring (optional) In a sequence of more than two constituents of the same type, i.e., x1, x2, . . . xn, an intonation break may be inserted before each repetition of the node X. 7.1.3 Summary So far, I have discussed how the intonational phrase is defined and restructured in general and in the Fuzhou dialect, and I have shown that the intonational phrase in Fuzhou is similar to the intonational phrase in other languages in terms of the definition and the restructuring. On the one hand, the Fuzhou intonational phrase is also the domain of intonation contours, and the domain boundaries should be inserted in the positions where pauses may be introduced. Hence, root sentences and certain syntactic constructions, such as parenthetical expressions, vocatives, expletives, appositives, and certain moved elements, can form intonational phrases on their own in this dialect. On the other hand, similar to ι-restructuring in other languages, the restructuring of the intonational phrase in Fuzhou depends on several key factors including the length of the original intonational phrase, the rate and the style of speech, and the assignment of contrastive prominence. The restructuring of the intonational phrase also happens when there is a list in a given utterance. The discussion in Section  7.1 reveals that the intonational phrase exhibits a large degree of variability in terms of its formation and restructuring, which distinguishes the intonational phrase from the other prosodic constituents. In Section 7.2, I will show that such variability also exists in the application of phonological rules within the intonational phrase domain, as can be seen in many languages including the Fuzhou dialect. The special phonological properties of the intonational phrase thus provide further support for the establishment of this prosodic constituent.

7.2 Rule application within the intonational phrase in the Fuzhou dialect 7.2.1  Introduction: variability in ι-domain rule application across languages As mentioned in Section 7.1, Selkirk (1978) defines the intonational phrase as the domain over which an intonation contour is spread. In addition to serving as the domain for intonation contours, the intonational phrase also represents the domain of application for a number of segmental phonological rules across languages, as Nespor & Vogel (1986) suggest. Take Gorgia Toscana in Tuscan Italian as an example. Gorgia Toscana is a phonological rule in Tuscan Italian, which changes the voiceless stops /p, t, k/ into the corresponding fricatives [ɸ, θ, h] between two

214  Intonational phrase in the Fuzhou dialect [−consonantal] segments within and across words. This rule applies throughout the intonational phrase domain but does not apply across the domain boundaries, as exemplified in (17) (examples in Section  7.2.1 are adapted from Nespor  & Vogel 1986), where the stops undergoing the rule are marked by “_” and those resisting the rule are marked by “=”. (17) [Almerico]ι [quando dorme solo]ι [cade spesso dall’amaca]ι Almerico when he-is-sleeping alone falls often the hammock ‘Almerico, when he sleeps alone, often falls out of the hammock.’ We have seen in Section 7.1 that several factors may cause the restructuring of the intonational phrase and break a long intonational phrase down into shorter ones. When this happens in Tuscan Italian, Gorgia Toscana may still apply in the smaller intonational phrases as long as the entire segmental context of this rule is within an intonational phrase domain, while it may be blocked if the restructuring destroys its segmental context. Consider the examples in (18). We can see that (18a) forms a single intonational phrase, and thus Gorgia Toscana applies throughout the domain. In contrast, since the List Restructuring rule applies and breaks the original intonational phrase down into smaller ones in (18b), Gorgia Toscana only applies in the first small intonational phrase in (18b), while it is blocked in the other two positions. [Quel giardino ha una gabia piena di corvi canarini that garden has a cage full of crows canaries colibrì e pellicani]ι hummingbirds and pelicans ‘That garden has a cage full of crows, canaries, hummingbirds, and pelicans.’ b. [Quel giardino ha una gabia piena di corvi]ι [canarini]ι [colibrì]ι [e pellicani]ι

(18) a.

Another phonological rule in Tuscan Italian, Intervocalic Spirantization, which changes the affricates /tʃ/ and /dʒ/ into the corresponding fricatives [ʃ] and [ʒ] between [-consonatal] segments within and across words, is similar to Gorgia Toscana in that it can only be triggered within the intonational phrase domain. The application and blocking of this rule can be exemplified in (19), in which the affricate at the domain boundaries, as marked by “=”, is not affected by this rule. (19) [Santo cielo]ι [c’è un verme in questa holy heaven there is a worm in this ‘Good heavens, there’s a worm in this cherry.’

ciliegia]ι cherry

Like Gorgia Toscana, Intervocalic Spirantization also exhibits the variability in its application. As can be seen in (20a), the entire sentence forms one intonational phrase, and Intervocalic Spirantization applies throughout the domain.

Intonational phrase in the Fuzhou dialect  215 Nonetheless, in (20b), the same sentence is restructured into shorter intonational phrases due to List Restructuring, which destroys the segmental context of this rule in the positions marked by “=” and results in the non-application of this rule in these positions. (20) a. [Hanno citato cinque recensioni cinquanta libri they cited five reviews fifty books cento articoli e duecento lavori inediti]ι one hundred articles and two hundred works unpublished ‘They cited five reviews, fifty books, one hundred articles, and two hundred unpublished works.’ b. [Hanno citato cinque recensioni]ι [cinquanta libri]ι [cento articoli]ι [e duecento lavori inediti]ι The Nasal Assimilation rule in Spanish is another typical example that exhibits a certain degree of flexibility in its application with respect to the restructuring of the intonational phrase. This rule assimilates a nasal in place of articulation to a following obstruent within the intonational phrase, such as those marked by “_” in (21), but it does not apply to nasals at the boundaries, such as those marked by “=”. (21) [Carmen]ι [cántanos una nueva canción]ι Carmen sing-us a new song ‘Carmen, sing us a new song, please.’

[por favor]ι for favor

When restructuring occurs and breaks a long intonational phrase down into shorter ones, this rule can still apply within shorter intonational phrases, but it is blocked if the restructuring interrupts the context of its application, which is similar to the two rules in Tuscan Italian we have just seen. This can be illustrated by the examples in (22), in which the sentence in (22a) is uttered more quickly than the one in (22b). We can find that the Nasal Assimilation rule fails to apply to the nasal at the ι domain boundaries in (22b), which is marked by “=”. (22)

a.

[Usa su sombrero carísimo con seis plumas de tucán wear her hat very expensive with six feathers of toucan cuando desea crear la impresión que es una persona muy when want create the impression that is a person very importante]ι important ‘(She) wears her very expensive hat within six toucan feathers when she wants to create the impression that she is a very important person.’ b. [Usa su sombrero carísimo con seis plumas de tucán]ι [cuando desea crear la impresión que es una persona muy importante]ι

As we can see from the above discussion, the ι-bounded rules, such as Gorgia Toscana and Intervocalic Spirantization in Tuscan Italian and Nasal Assimilation

216  Intonational phrase in the Fuzhou dialect in Spanish, may apply in a given position when a particular string is uttered in one way but may be blocked in the same position when the string is uttered in another way. This shows that the application of these rules is significantly affected by the formation and the restructuring of the intonational phrase and has a great degree of variability. We will see in Sections 7.2.2 and 7.2.3 that similar variability in the rule application can also be found within the intonational phrase domain in the Fuzhou dialect. 7.2.2  Blocking of phonological rules across ι boundaries in the Fuzhou dialect Like other languages in the world, the major function of the intonational phrase in the Fuzhou dialect is to act as the domain for intonation contours. Besides that, the intonational phrase in Fuzhou is also relevant to certain phonological phenomena in this dialect. In this section and Section 7.2.3, I will investigate two phonological properties distinctive of the intonational phrase domain in Fuzhou, which can distinguish the intonational phrase from other prosodic constituents in this dialect. Let us examine the first phonological property of the intonational phrase domain in Fuzhou. It has long been noticed that Fuzhou phonological rules cannot apply across intonational phrases. As mentioned in Chapter 5, Chen & Norman (1965a) distinguish four types of junctures, one of which is called a terminal juncture, and they point out that a terminal juncture is a pause or actual stop in the flow of speech and is always associated with intonation contours. According to the discussion in Section 7.1, we can find that this so-called terminal juncture is actually the boundaries of the intonational phrase in Fuzhou. Chen & Norman argue that TS is blocked before this type of juncture and the syllable following it will not undergo CL. Since FC always applies together with TS, it will be blocked before this juncture as well. Data from the Fuzhou dialect provide evidence for these observations and demonstrate that all these three phonological rules are blocked across the boundaries of intonational phrases, as exemplified in (23), where the positions of the blocking relevant to the discussion are marked by “#”. In (23), three types of Fuzhou intonational phrases are presented, namely the intonational phrase formed by a root sentence, as in (23a), the intonational phrase formed by a syntactic construction that forms an obligatory intonational phrase on its own, as in (23b), and the intonational phrase formed by ι-restructuring, as in (23c) and (23d). In these examples, although the sequences of the marked sounds provide the applicable phonological context for TS, FC, and/or CL, none of these rules applies, since such sequences all occur at the intonational phrase boundaries.

Intonational phrase in the Fuzhou dialect  217 (23)

a.

[只 间 店 是 好 店]ι [陈 先生 [tsi31 kaŋ44 taiŋ213 sei242 xo31 taiŋ213] [tiŋ51 siŋ44 saŋ44 → [tsi21 ʔaŋ44 taiŋ213 si21 xo44 laiŋ213] # [tiŋ21 siŋ44 naŋ44 this Cl store be good store Chen teacher 每日 都 来]ι mui31 niʔ5 tu44 li51] 21 5 44 → mui niʔ tu li51] every day all come ‘This is a very nice store. Mr. Chen comes here every day.’ b.

[生 野 俊]ι [许 隻 [saŋ44 ʔia31 tsouŋ213] [xi31 tsiɛʔ23 → [saŋ44 ʔia44 ʒouŋ213] # [xi44 ʒiɛʔ23 grow very beautiful that Cl ‘She is very beautiful, that girl.’

诸娘囝]ι tsy44 noyŋ51 kiaŋ31] tsy51 noyŋ31 ŋiaŋ31] girl

c.

[伊 讲]ι [做 贼 其 许隻 侬]ι [昨暝 [ʔi44 kouŋ31] [tso213 tsheiʔ5 ki0 xi31 tsiɛʔ23 nøyŋ51] [soʔ5 maŋ51 → [ʔi44 kouŋ31] # [tso44 tsheiʔ5 ʔi0 xi44 ʒiɛʔ23 nøyŋ51] # [soʔ31 maŋ51 he speak do thief MOD that Cl people yesterday 敆 街中]ι [乞 侬 拍 去]ι kaʔ5 kɛ44 touŋ44] [khøyʔ23 nøyŋ51 phaʔ23 kho0] → kaʔ21 kɛ44 louŋ44] # [khøyʔ23 nøyŋ51 phaʔ23 ʔo0] in on the street PASS people hit PERF ‘He said that that person who was a thief was hit by others on the street yesterday.’ d. →

[包 [pau44 [pau51 wrap

[挂 [kua213 [kua51 rice dumpling hang 粽]ι tsoyŋ213] tsoyŋ213] #

艾]ι ŋiɛ213] ŋiɛ213] # argyi (Chinese mugwort)

[做 [tso242 [tso21 do

香包]ι [扒 xyoŋ44 pau44] [pa51 xyoŋ44 mau44] # [pa31 scent bag row

[是 五月节 其 风俗]ι 龙船]ι luŋ51 suŋ51] [sei242 ŋu242 ŋuoʔ5 tsaiʔ23 ki0 xuŋ44 syʔ5] → luŋ31 nuŋ51] # [si21 ŋu21 ŋuoʔ5 tsaiʔ23 ʔi0 xuŋ44 nyʔ5] dragon boat be Dragon Boat Festival POSS custom ‘Making rice dumplings, hanging argyi grass, making scent bags, and racing dragon boats are customs of the Dragon Boat Festival.’

Besides the above examples, the three major Fuzhou phonological rules are also blocked across the boundaries of intonational phrases in some special cases of ι-restructuring. It has been reported in the literature that when a string is uttered at a slow tempo or a particular part in a string is emphasized in the Fuzhou dialect, the application of phonological rules at certain positions tends to be blocked (Chen 1998; Li & Liang 2001; Li 2002). Consider the examples in (24) and (25).

218  Intonational phrase in the Fuzhou dialect (24) a.

Normal rate of speech and no contrastive prominence 我 买 菜 伊 卖 菜 ŋuai31 mɛ31 tshai213 ʔi44 mɛ242 tshai213 → ŋuai31 mɛ44 tshai213 ʔi44 mɛ51 tshai213 I buy vegetable he sell vegetable ‘I buy vegetables. He sells vegetables.’ b. Slow rate of speech and prominence on 买 ‘to buy’ and 卖 ‘to sell’ 我 买 菜 伊 卖 菜 ŋuai31 mɛ31 tshai213 ʔi44 mɛ242 tshai213 → ŋuai31 mɛ31 tshai213 ʔi44 mɛ242 tshai213 I buy vegetable he sell vegetable ‘I BUY vegetables. (But) he SELLS vegetables.’

(25)

Speaker A: 掏 甚乇? to51 sieʔ23 noʔ23 → to21 sieʔ51 noʔ23 take what ‘What (do you want) to take?’ Speaker B (Normal rate of speech and no contrastive prominence): 旧 碗 两 块 新 碗 三 块 kou242 ʔuaŋ31 laŋ242 toy213 siŋ44 ʔuaŋ31 saŋ44 toy213 → ku51 ʔuaŋ31 laŋ51 noy213 siŋ51 ŋuaŋ31 saŋ51 noy213 old bowl two Cl new bowl three Cl ‘Two old bowls. Three new bowls.’ Speaker A: 甚乇 碗 两 块? 甚乇 碗 三 块? sieʔ23 noʔ23 ʔuaŋ31 laŋ242 toy213 sieʔ23 noʔ23 ʔuaŋ31 saŋ44 toy213 → sieʔ21 noʔ51 ʔuaŋ31 laŋ51 noy213 sieʔ21 noʔ51 ʔuaŋ31 saŋ51 noy213 what bowl two Cl what bowl three Cl ‘What bowl for two and what bowl for three?’ Speaker B (Slow rate of speech and prominence on 旧 ‘old’ and 新 ‘new’): 旧 碗 两 块 新 碗 三 块 kou242 ʔuaŋ31 laŋ242 toy213 siŋ44 ʔuaŋ31 saŋ44 toy213 → kou242 ʔuaŋ31 laŋ51 noy213 siŋ44 ʔuaŋ31 saŋ51 noy213 old bowl two Cl new bowl three Cl ‘Two OLD bowls. Three NEW bowls.’

We can find that one sentence in (24) and (25) may have two surface readings. In (24), when there is no contrastive prominence in the sentence and the sentence is produced at a normal rate, TS occurs between 买 ‘to buy’ and 菜 ‘vegetable’, and it is also triggered between 卖 “to sell” and 菜 ‘vegetable’, since both 买菜 ‘to buy vegetables’ and 卖菜 ‘to sell vegetables’ form phonological phrases according to the discussion in Chapter 5. In contrast, when prominence is assigned to the two verbs to emphasize the contrast between the two subjects, the sentence may be uttered slower and neither 买 ‘to buy’ nor 卖 ‘to sell’ undergoes TS. In (25),

Intonational phrase in the Fuzhou dialect  219 which is a dialogue between two speakers, Speaker B utters the same sentence twice. The first time, this sentence is uttered at a normal rate with no contrastive prominence, and both TS and FC apply to 旧 ‘old’, which is followed by 碗 ‘bowl’, and both TS and CL occur between 新 ‘new’ and 碗 ‘bowl’, because both 旧碗 ‘old bowls’ and 新碗 ‘new bowls’ form phonological phrases. The second time, however, Speaker B assigns prominence to both 旧 ‘old’ and 新 ‘new’, and the sentence is uttered at a slow rate. This time, none of the three rules is triggered in the above-mentioned positions. Clearly, semantic and pragmatic factors such as the rate of speech and the assignment of contrastive prominence affect the application of TS, FC, and CL in (24) and (25). As we have seen in Section 7.1, these factors play an important role in the restructuring of the intonational phrase in the Fuzhou dialect. Therefore, it is very natural to assume that it is the ι-restructuring caused by these factors that blocks the application of these rules. The formation and the restructuring of intonational phrases in relevant sentences in (24) and (25) are presented in (26) and (27) respectively. (26)   a. Normal rate of speech and no contrastive prominence (root sentence) [我 买 菜]ι [伊 卖 菜]ι tshai213] [ʔi44 mɛ242 tshai213] [ŋuai31 mɛ31 → [ŋuai31 mɛ44 = tshai213] [ʔi44 mɛ51 = tshai213] I buy vegetable he sell vegetable ‘I buy vegetables. He sells vegetables.’ b. Slow rate of speech and prominence on 买 ‘to buy’ and 卖 ‘to sell’ [菜]ι [伊 卖]ι [菜]ι [我 买]ι (ι-restructuring) [ŋuai31 mɛ31] [tshai213] [ʔi44 mɛ242] [tshai213] 31 31 h 213 44 242 → [ŋuai mɛ ] # [ts ai ] [ʔi mɛ ] # [tshai213] I buy vegetable he sell vegetable ‘I BUY vegetables. (But) he SELLS vegetables.’ (27) a.

Normal rate of speech and no contrastive prominence [新 碗 三 块]ι [旧 碗 两 块]ι (root sentence) [kou242 ʔuaŋ31 laŋ242 toy213] [siŋ44 ʔuaŋ31 saŋ44 toy213] 51 31 51 213 51 → [ku = ʔuaŋ laŋ noy ] [siŋ = ŋuaŋ31 saŋ51 noy213] old bowl two Cl new bowl three Cl ‘Two old bowls. Three new bowls.’ b. Slow rate of speech and prominence on 旧 ‘old’ and 新 ‘new’ [碗]ι [两 块]ι [新]ι [碗]ι [三 块]ι [旧]ι (ι-restructuring) [kou242] [ʔuaŋ31] [laŋ242 toy213] [siŋ44] [ʔuaŋ31] [saŋ44 toy213] → [kou242] # [ʔuaŋ31] [laŋ51 noy213] [siŋ44] # [ʔuaŋ31] [saŋ51 noy213] old bowl two Cl new bowl three Cl ‘Two OLD bowls. Three NEW bowls.’

As we can see in (26) and (27), TS, FC, and/or CL can apply in the phonological phrases 买菜, 卖菜, 旧碗, and 新碗 as long as the phonological context is satisfied.

220  Intonational phrase in the Fuzhou dialect When ι-restructuring occurs, however, the boundaries of intonational phrases are inserted inside these phonological phrases, dividing these phonological phrases into two parts. These two parts are thus contained in two adjacent intonational phrases, which results in the non-application of TS, FC, and CL at the boundaries. The discussion in Section 7.2.2 reveals one peculiar phonological property of the intonational phrase domain in Fuzhou, namely the blocking of the phonological rules across the domain boundaries. The blocking not only can be seen at the boundaries of regularly formed intonational phrases, but also happens when a string is uttered at a slow tempo or when contrastive prominence is assigned to a particular part in a string. This shows a large degree of variability in the application of Fuzhou phonological rules within the intonational phrase domain, which results from the variability in the formation and the restructuring of intonational phrases. 7.2.3  Optional rule application within intonational phrase in the Fuzhou dialect Let us now move on to another distinctive phonological property of the intonational phrase in Fuzhou. Data from the Fuzhou dialect show that when an intonational phrase is uttered rapidly, rules that are expected to be blocked within or across the boundaries of the embedded domains may be optionally, though not always, triggered. Consider the examples in (28). Positions where rules are expected to be blocked are marked by “#” while positions where rules optionally apply are marked by “=”. (28) a. → → b. → →

[[食]φ [大 葡萄]φ]ι [[siɛʔ5] [tuai242 pu51 to51]] [[siɛʔ5] # [tuai51 pu31 lo51]] (normal rate of speech) [[siɛʔ21] = [tuai51 pu31 lo51]] (rapid rate of speech) eat big grape ‘to eat big grapes’ [[天]φ [长]φ [地]φ [久]φ]ι (adapted from Li et al. 1994) [[thiɛŋ44] [tuoŋ51] [tei242] [kiu31]] [[thiɛŋ44] # [tuoŋ51] # [tei242] # [kiu31]] (normal rate of speech) [[thiɛŋ21] = [nuoŋ21] = [ni51] = [ʔiu31]] (rapid rate of speech) sky long earth long ‘enduring as the universe (everlasting and unchanging)’

The example in (28a) shows that, when the intonational phrase is uttered at a normal rate of speech, no phonological rules can apply between the verb 食 ‘to eat’ and the branching object 大葡萄 ‘big grape’, since the verb and the object are contained in two separated phonological phrases according to the discussion in Chapter 5. However, when these two phonological phrases are included in an intonational phrase that is produced quickly, TS can be optionally triggered. In (28b), the noun 天 ‘sky’ and the adjective 长 ‘long’ form a subject-predicate construction, and so do 地 ‘earth’ and 久 ‘long’. Hence, the string in (28b) forms four

Intonational phrase in the Fuzhou dialect  221 phonological phrases following the definition of Fuzhou phonological phrase proposed in Chapter 5, and thus no phonological rules can apply across their boundaries. However, when all these four phonological phrases are included within one intonational phrase domain, TS, FC, and CL may all be triggered throughout the intonational phrase when it is uttered rapidly. The optional triggering of phonological rules within the intonational phrase domain can also be found in the examples in (29). (29) a.

(adapted from Shih 1986) [[看见 老板]φ]ι [[khaŋ213 kiɛŋ213 lo31 peiŋ31]] → [[khaŋ51 ŋiɛŋ213 lo24 peiŋ31]] (normal rate of speech) → [[khaŋ21 ŋiɛŋ21 lo24 peiŋ31]] (rapid rate of speech) see boss ‘to see the boss’ b. [[野 中意]φ]ι [[ʔia31 tøyŋ213 ʔei213]] → [[ʔia21 tyŋ51 ŋei213]] (normal rate of speech) → [[ʔia21 lyŋ51 ŋei213]] (rapid rate of speech) very like ‘to like (sth.) very much’

We can see that these two examples are typical examples of “no look-back” phenomena in the Fuzhou dialect. When these two examples are uttered at a normal rate of speech, the application of phonological rules in the examples are restricted by the DIC introduced in Chapter 5. In (29a), since the verb 看见 ‘to see’ forms a prosodic word in which the structural description of TS is satisfied, the tone of its second syllable cannot be further modified when the derivation has moved to the phonological phrase domain composed of 看见 ‘to see’ and its object 老板 ‘boss’. In (29b), an example mentioned in Chapter 5, 中意 ‘to like’ constitutes a prosodic word where the structural description of the CL rule is met. So, when the derivation moves to the phonological phrase domain containing 野 ‘very’ and 中意 ‘to like’, the initial of the first syllable in 中意 ‘to like’ is immune against further modification. However, similar to the blocking of phonological rules across the phonological phrase boundaries, the blocking caused by the effect of the DIC can also be “erased” in the intonational phrase domain when the intonational phrase is uttered at a rapid rate. As shown in (29), TS can be optionally “activated” and applies to the tone of the second syllable in 看见 ‘to see’ in (29a), and CL can also optionally change the initial of the first syllable in 中意 ‘to like’ in (29b), both of which take place when the examples are produced rapidly. From the examples in (28) and (29), we can notice that the rate of speech, as a pragmatic factor, can affect the rule application within the intonational phrase domain in the Fuzhou dialect and that a rapidly uttered intonational phrase can lead to the optional triggering of phonological rules in cases where the rules are not expected to apply when the strings are produced at a normal rate. These

222  Intonational phrase in the Fuzhou dialect examples again indicate that there is a large degree of variability in the application of phonological rules within the intonational phrase domain in this dialect. The above-discussed optional application of Fuzhou phonological rules within a rapidly uttered intonational phrase, together with the blocking of phonological rules across the ι boundaries we have seen in Section 7.2.2, can well account for the two alternative readings in the example in (30) (adapted from Shih 1986). (30) → →

信 固𠲥 ku213 lɛ31 seiŋ213 [ku51 = lɛ31] # [siŋ51 = [ku21 = lɛ44 = leiŋ213] # still believe ‘to still believe in Buddhism’

佛教 xuʔ5 kau213 xuʔ21 kau213] [xuʔ21 kau213] Buddhism

(Reading 1) (Reading 2)

In (30), 固𠲥 ‘still’ in Reading 1 forms a domain of rule application and 信佛 教 ‘believe in Buddhism’ forms the other. By contrast, in Reading 2, 固𠲥 ‘still’ and 信 ‘to believe’ are incorporated into one domain, and 佛教 ‘Buddhism’ forms a domain on its own. Shih (1986) argues that the two readings in (30) should be attributed to the application/non-application of the optional Tone Group Expansion rule (see Chapter 5)—when this rule applies, 信 ‘to believe’ and 佛教 ‘Buddhism’ are grouped together, while they will be separated when the expansion rule does not apply. However, this does not explain why the second syllable in the disyllabic adverb 固𠲥 ‘still’ in Reading 2 can undergo TS when followed by the verb (TS should not be able to apply here due to the restriction of the DIC). Following the discussion of the phonological properties of the intonational phrase domain in the Fuzhou dialect, I would argue that the variability in the reading of the example in (30) is simply due to the relatively large degree of variability in the formation of intonational phrases. The two possible ways of the intonational phrase formation of (30) are given in (31). (31) a.

Normal rate of speech and no contrastive prominence (root sentence) 信 佛教]ι [固𠲥 [ku213 lɛ31 seiŋ213 xuʔ5 kau213] → [ku51 lɛ31 siŋ51 xuʔ21 kau213] still believe Buddhism ‘to still believe in Buddhism’ b.

Rapid rate of speech and prominence on 信 ‘to believe’ (ι-restructuring) [佛教]ι 信]ι [固𠲥 [ku213 lɛ31 seiŋ213] [xuʔ5 kau213] → [ku21 lɛ44 leiŋ213] [xuʔ21 kau213] still believe Buddhism ‘to still believe in Buddhism’ As we can see in (31a), the string forms one intonational phrase on its own since it is a root sentence. The string within this intonational phrase domain also forms

Intonational phrase in the Fuzhou dialect  223 only one phonological phrase domain according to the discussion in ­Chapter 5. Within this phonological phrase domain, on the one hand, since the adverb 固𠲥 ‘still’ forms a prosodic word domain on its own where the structural description of TS is met, its second syllable can no longer be affected by TS when the derivation has moved to the phonological phrase domain, as required by the DIC. On the other hand, the verb 信 ‘to believe’ undergoes both TS and FC because it is incorporated into the same phonological phrase with the following object. In (31b), by contrast, prominence is assigned to the verb 信 ‘to believe’, and the string is broken down into two smaller intonational phrases. On the one hand, with the ι boundaries inserted between 信 ‘to believe’ and 佛教 ‘Buddhism’, no phonological rules can be triggered across the boundaries. On the other hand, when the first intonational phrase 固𠲥信 ‘to still believe’ is uttered rapidly, the TS rule, which is originally blocked between 固𠲥 ‘still’ and 信 ‘to believe’ due to the DIC, can be optionally triggered. The variability in the formation and the restructuring of the intonational phrase and the phonological properties of the intonational phrase in Fuzhou thus successfully explain the two different readings of this example. This, once again, shows that semantic and pragmatic factors such as the rate of speech and the assignment of contrastive prominence are essential in the Fuzhou dialect, which give rise to the variability in the rule application within the intonational phrase domain. 7.2.4 Summary From the discussion in Section 7.2, we have identified two outstanding phonological properties of the intonational phrase domain in the Fuzhou dialect. First, phonological rules in Fuzhou are blocked across the boundaries of intonational phrases. When ι-restructuring is caused by semantic and pragmatic factors such as the rate of speech and the assignment of contrastive prominence, the insertion of ι boundaries may block the phonological rules that are supposed to apply. As we have seen in Section 7.2.2, this may result in the variable reading of a given string. Second, when an intonational phrase in Fuzhou is uttered at a rapid tempo, phonological rules that are originally blocked within the embedded domains or across the boundaries of these domains may be optionally triggered, as we have seen in Section 7.2.3. Therefore, it can be concluded that there is a great degree of variability in the application of phonological rules within the intonational phrase domain in the Fuzhou dialect, which is similar to what we have seen in some other languages, such as Tuscan Italian and Spanish. Data from this dialect demonstrate that semantic and pragmatic factors play a crucial role in creating such type of variability and giving rise to the two outstanding phonological properties of the intonational phrase in Fuzhou. Since the variability in the rule application and these two phonological properties are characteristic only of the intonational phrase but cannot be found in any other prosodic domains in the Fuzhou dialect, it is reasonable to establish the intonational phrase as an indispensable prosodic constituent in this dialect.

224  Intonational phrase in the Fuzhou dialect

7.3 Summary In this chapter, I  have explored the formation and the phonological properties of the intonational phrase domain in the Fuzhou dialect. Overall speaking, the intonational phrase in Fuzhou shares some common features with the intonational phrase in other languages with respect to its domain formation: (a) the intonational phrase in Fuzhou is also the domain of intonation contours, and the domain boundaries should be inserted in the positions where pauses may be introduced; (b) root sentences and certain syntactic constructions in this dialect, such as parenthetical expressions, vocatives, expletives, appositives, and certain moved elements, can also form intonational phrases on their own; (c) an intonational phrase in Fuzhou can be broken down into smaller intonational phrases, and the restructuring is usually caused by semantic and pragmatic factors such as the length of the original intonational phrase, the rate and the style of speech, and contrastive prominence; and (d) the restructuring of the intonational phrase in Fuzhou may also take place when there is a list in a given utterance. In terms of phonological properties, I  have shown that the variability in the application of phonological rules within the intonational phrase domain can also be observed in the Fuzhou dialect, which is another common feature between the Fuzhou dialect and other languages. Based on the evidence from the Fuzhou dialect, I have demonstrated that such type of variability in the rule application is created by semantic and pragmatic factors such as the rate of speech and the assignment of contrastive prominence. On the one hand, ι-restructuring caused by these factors may insert ι domain boundaries into positions where phonological rules such as TS, FC, and CL are expected to apply and thus block the application of these rules at the newly inserted domain boundaries. On the other hand, when an intonational phrase is uttered at a rapid rate in the Fuzhou dialect, phonological rules that are originally blocked within or across the embedded domains may be optionally triggered. The discussion in this chapter reveals not only the variability in the formation and the restructuring of the intonational phrase domain in the Fuzhou dialect, but also the variability in the rule application within this domain. From the investigation in previous chapters, we can notice that no other prosodic constituents in Fuzhou allow for such a great degree of variability in terms of their domain formation, nor does this type of variability occur in the rule application within other prosodic domains. This, therefore, provides a significant motivation for the establishment of the intonational phrase in the prosodic phonology of the Fuzhou dialect.

Note 1 It is noteworthy that intonational phrases formed by the strings on one or both sides of an intervening obligatory intonational phrase are not always isomorphic to any syntactic constituent. For instance, the intonational phrases to the left of the obligatory intonational phrases in sentences “[They have]ι [as you know]ι [been living together for years]ι” and “[He will never]ι [as I said]ι [accept your proposal]ι” do not correspond to any syntactic constituent (Nespor & Vogel 1986).

8 Concluding remarks

This book presents a study on the phonological system of the Fuzhou dialect within the framework of prosodic phonology. On the one hand, it aims at providing a thorough description and formal analysis of the phonological system of the Fuzhou dialect by employing the concepts and approaches developed in the theory of prosodic phonology. On the other hand, it is an attempt to further the general understanding of prosodic phonology and the relationship between phonology and other components of the grammar through the investigation of the Fuzhou dialect. The main findings and conclusions drawn in this study are summarized and discussed below, followed by suggestions for future research.

8.1 Prosodic domains and rule application in the Fuzhou dialect In this book, I have studied most prosodic constituents in the universal prosodic hierarchy, including the syllable, the foot, the prosodic word, the clitic group, the phonological phrase, and the intonational phrase, in terms of their definitions and phonological properties in the Fuzhou dialect. From Chapter 2 to Chapter 7, I have demonstrated that most of these universal prosodic constituents should be established as indispensable prosodic domains in the prosodic phonology of the Fuzhou dialect, since there are phonological phenomena that make crucial reference to these domains. The discussion in Chapter 2 shows that the only exception is the foot because no independent evidence and valid arguments can be provided to demonstrate its existence and its role as a domain of rule application in this dialect. In Chapter 2, I also reviewed some previous studies that treat the mora as the tone bearing unit in the Fuzhou dialect. Although I have demonstrated that the employment of the notion of mora in these studies is on the wrong track, I have not found evidence to exclude this unit from the prosodic hierarchy in this dialect. Similarly, in spite of the fact that no phonological phenomena relevant to the utterance have been reported in the literature or observed by my informants or myself, no decisive evidence has been identified so far to prove the non-existence of this domain in Fuzhou either. Therefore, both the mora and the utterance are still maintained in the prosodic hierarchy in the Fuzhou dialect.

226  Concluding remarks Semantic & pragmatic information

Utt/υ IPh/ι

(Discourse/focus-based) Morphosyntactic information (Morphosyntax-based)

PPh/φ CG

PW/ω Phonological information σ (Rhythm-based)

(IPh/ι) (Intonational Phrase)

(PPh/φ) (CG)

(PW/ω) (σ)

( )

(Utterance)

(Phonological Phrase) (Clitic Group) (Prosodic Word) (Syllable) (Mora)

Figure 8.1  Prosodic hierarchy in the Fuzhou dialect

The complete model of prosodic hierarchy in the Fuzhou dialect can be presented as in Figure 8.1, which is divided into three parts according to the types of information involved in the definition of different prosodic constituents. I have explored the phonological properties of these prosodic domains through a detailed inspection of the application and the blocking of major phonological rules in the Fuzhou dialect, including Phonological Tone Sandhi (TS), Morphological Tone Sandhi (MTS), Final Alternation (FA), Final Change (FC), and Initial Consonant Lenition (CL). Based on the evidence from the Fuzhou dialect, I have shown that a given rule in this dialect may be triggered within a particular domain but be blocked within another and that a given rule may apply within two or more domains with different degrees of application. Moreover, the application of the rules in a particular domain is conditioned by the relevant factors involved in the domain formation. The formation of each major prosodic domain and the summary of the rule application within different domains in the Fuzhou dialect are given in Table 8.1. Figure 8.1 and Table 8.1 have summarized our major findings about the phonological system of the Fuzhou dialect. In the course of the investigation into Fuzhou prosodic domains and relevant rule application, this book has made a number of contributions to the studies on this dialect. It has offered a detailed description of a lot of new primary data from the Fuzhou dialect, presented a formal analysis of the application of Fuzhou phonological rules, and re-analyzed and proposed explanations for long-standing controversial issues such as the phrasallevel tone sandhi and the existence of the foot and clitics. It has also discussed many aspects of Fuzhou morphology, syntax, semantics, and pragmatics, including but not limited to the morphological processes of morphosyntactic word formation, the morphosyntactic functions of clitics, and the syntactic structure of phrasal constructions. Overall, this book presents a comprehensive analysis of the

Table 8.1  Prosodic domains and rule application in the Fuzhou dialect Prosodic domains

Domain formation and rule application

Syllable (σ)

Domain formation: the domain of σ is a syllable. Rule application (conditioned by syllable structure): Application of FA (in syllables with alternating finals).

Prosodic word (ω)

Domain formation: the domain of ω is the terminal node of the syntactic tree.

Clitic group (CG)

Domain formation: the domain of CG consists of one independent (i.e., nonclitic) prosodic constituent (ω, CG, or φ), plus any adjacent a. directional clitic, or b. plain clitic/nondirectional clitic such that there is no possible host with which they share more category memberships.

Rule application (conditioned by morphological factors, i.e., subtypes of morphological processes/morphosyntactic words): a. Application of TS (in polysyllabic monomorphemic words, derived words, compounds, and reduplicated adjectives); b. Application of MTS (in “diminutive” nouns and sound-splitting words); c. Application of FC (in all subtypes of morphosyntactic words; optionally in sound-splitting words); d. Application of CL (in polysyllabic monomorphemic words, derived words, compounds, and reduplicated disyllabic adjectives).

Rule application (conditioned by the attachment direction of clitics): a. Blocking of TS and FC in Type A CG (between the host and the enclitic); b. Application of CL in Type A CG (between the host and the enclitic); c. Application of TS and FC in Type B CG (between the proclitic and the host); d. Blocking of CL in Type B CG (between the proclitic and the host). Phonological Domain formation: the domain of φ is constructed with three steps: phrase (φ) a. Mark the right edge of every lexical head X, except where XP is an adjunct; b. On the non-recursive side of the lexical head X, mark the right edge of the first phonetically overt head Y (if any; either lexical or functional) outside of XP; ωs/CGs that are separated by the right edge of X or Y belong to different φs; c. φ-restructuring: a non-branching φ, which is the first complement of X on its recursive side, is joined into the φ that contains X. Rule application (conditioned by syntactic factors, e.g., head, branchiness, adjunct/non-adjunct distinction, recursive/non-recursive sides, etc.): a. Application of TS and FC (in all φs); b. Application of CL (in φs NOT constructed through φ-restructuring); c. Blocking of CL (in φs constructed through φ-restructuring). Intonational phrase (ι)

Domain formation: the domain of ι may consist of a. all the CGs and φs in a string that is not structurally attached to the sentence tree at the level of s-structure, or b. any remaining sequence of adjacent CGs/φs in a root sentence. Rule application (conditioned by semantic and pragmatic factors, e.g., rate of speech, contrastive prominence, etc., and exhibiting a large degree of variability): a. Blocking of TS, FC, and CL across the boundaries of ιs (including regular ιs formed by root sentences, obligatory ιs formed by certain syntactic constructions, and ιs formed through ι-restructuring); b. Optional application of originally blocked TS, FC, and CL (in rapidly uttered ιs).

228  Concluding remarks Fuzhou phonological system within the framework of prosodic phonology, and it is my hope that the discoveries in this book have provided a clearer and more complete picture of Fuzhou phonology.

8.2  Theoretical issues regarding prosodic phonology On the basis of an extensive discussion of the prosodic domains and the rule application in the Fuzhou dialect, this book has provided supporting evidence for some basic tenets of the prosodic phonology theory. First of all, data from the Fuzhou dialect have shown that there are phonological rules that refer crucially to prosodic constituents in the Fuzhou dialect as their domains of application, which provides significant motivation and evidence for the existence of prosodic domains and the prosodic hierarchy. In addition, some prosodic constituents in the Fuzhou dialect, e.g., the clitic group, though constructed by making use of morphosyntactic information, are not necessarily isomorphic to any morphosyntactic structures. This has not only confirmed the main claim of prosodic phonology that syntactic and phonological representations are not isomorphic, but also offered another important motivation for the establishment of prosodic domains and the prosodic hierarchy. Moreover, the definitions and domain formations of prosodic constituents in Fuzhou have substantiated that a given constituent should be built with reference to a specific type of phonological or non-phonological information, which is not a free choice. In addition to providing empirical evidence for the basic tenets of the prosodic phonology theory, I have also explored several outstanding issues regarding this theory, including the weakened Strict Layer Hypothesis (SLH), the status of the clitic group, and the approaches to phonological phrasing. Besides, I  have employed the Domain Impenetrability Condition (DIC) to account for the “no look-back” phenomena in the Fuzhou dialect. The conclusions reached on these theoretical issues are discussed below. I  The weakened Strict Layer Hypothesis The original SLH, since it was proposed in the 1980s, has been challenged by evidence across languages, which is why it has been made less strict and has been factored out into four constraints on prosodic domination with the advent of OT. The discussion in this book has demonstrated the necessity of a weakened SLH on the basis of the linguistic facts in the Fuzhou dialect. As we have seen in Chapter 4 and Chapter 6, a clitic group in the Fuzhou dialect may dominate another clitic group, which indicates that the Nonrecursivity constraint is violable in this dialect. In addition, a clitic group in Fuzhou may dominate a phonological phrase, which is located higher than the clitic group in the prosodic hierarchy, showing that the Layeredness constraint is not inviolable either. In this book, we have also identified a number of examples in which a prosodic constituent of level n in the Fuzhou dialect directly dominates constituents of the level n-2, e.g., a phonological phrase dominating a prosodic word (Chapter 5) and an intonational phrase

Concluding remarks  229 dominating a clitic group (Chapter 7). These examples suggest that the violation of the Exhaustivity constraint should be allowed in the prosodic phonology of this dialect as well. These Fuzhou linguistic facts pose a great, unsolvable challenge to the original SLH, which, therefore, could serve as strong evidence that a weakened SLH allowing the violation of Exhaustivity, Nonrecursivity, and even Layeredness is necessary in the theory of prosodic phonology. II  The status of the clitic group The status of the clitic group is a long-standing problem for linguists, and a vast amount of research has been devoted to arguing for or against the existence of the clitic group as a prosodic constituent. The examination of Fuzhou clitics and clitic groups in this book has given us some hints about how to solve the problems with the clitic group. We have seen in Chapter 4 and Chapter 6 that there are two types of clitics and consequently two types of clitic groups in the Fuzhou dialect. Data from this dialect reveal two important properties of Fuzhou clitics: (a) Fuzhou clitics may attach to constituents higher than the prosodic word, and (b) asymmetries exist between enclitics and proclitics, as well as between the Type A clitic group and the Type B clitic group, in terms of their phonological behavior. These two properties would be viewed as counterevidence against the existence of the clitic group according to some previous studies. However, the first property of Fuzhou clitics, on the one hand, is nicely accounted for by assuming a weakened SLH and allowing the violation of Nonrecursivity and Layeredness. The second property, on the other hand, would not be a problem either. By comparing the clitic group in Fuzhou with the other prosodic constituents in this dialect, such as the prosodic word and the phonological phrase, I have shown that the clitic group as a whole (including both Type A  and Type B) can be distinguished from the other constituents with respect to their roles as the domain of rule application—it is only within the clitic group domain that the application of Fuzhou phonological rules is conditioned by the attachment direction of clitics. Therefore, different from previous studies that exclude the clitic group from the prosodic hierarchy, this study maintains and establishes the clitic group as an indispensable prosodic constituent, on account of the fact that some phonological phenomena are characteristic only of the clitic group domain. III  Approaches to phonological phrasing Two main approaches to phonological phrasing have been advanced in the prosodic phonology theory, namely the RBA and the EBA. The former was developed by Nespor & Vogel (1986) and Hayes (1989), and the latter was proposed by Selkirk (1986) and Chen (1985, 1987) and further developed into the Align/ Wrap Theory and the Match Theory with the advent of OT. These two approaches can be distinguished according to the types and amount of syntactic information they implement in the formation of the phonological phrase, but it is not easy to demonstrate the superiority of the EBA over the RBA, or vice versa. In this book,

230  Concluding remarks I have evaluated these two approaches by examining their explanatory power in accounting for the phrasal-level application of TS in the Fuzhou dialect. Data from the Fuzhou dialect have shown that the Align/Wrap Theory and the Match Theory do not exhibit stronger explanatory power as compared to the RBA and the original parameterized algorithms of the EBA. It has also been demonstrated that neither the RBA nor the EBA alone can correctly define the phonological phrase in this dialect. Based on such observations, I have proposed an alternative analysis that defines the Fuzhou phonological phrase through a three-step algorithm. This alternative approach is a combination of the EBA and the RBA—it makes reference to not only the edges of syntactic heads, but also relation-based considerations such as adjunct/non-adjunct distinction and syntactic branching. The Fuzhou linguistic facts have proved that this approach is a better solution to the problem of Fuzhou TS at the phrasal level. This, therefore, suggests that the prosodic phonology theory may need to allow a third possibility of phonological phrasing, namely that the phonological phrase in a particular language may be constructed by combining the EBA and the RBA, instead of relying on only one of them. IV  The Domain Impenetrability Condition In Chapter 5, I discussed several examples of “no look-back” phenomena in the Fuzhou dialect and introduced the DIC, which was first proposed in You (2018a), to account for these phenomena. It has been shown in Chapter  5 that the DIC imposes restrictions on the cyclic application of Fuzhou phonological rules across domain boundaries and nicely explains the “no look-back” phenomena we have seen within the phonological phrase domain. Different from You (2018a), whose arguments for the DIC are only about the levels of prosodic word, clitic group, and phonological phrase, the discussion on DIC in this book also touches upon the intonational phrase level. I have demonstrated in Chapter 7 that, when a phonological phrase is contained in an intonational phrase, the blocking of rules within the phonological phrase domain, which is caused by the effect of the DIC, can be “erased” in the intonational phrase domain if the intonational phrase is uttered at a rapid tempo. Actually, in the Fuzhou dialect, the effect of the DIC does not hold across syllable boundaries within a prosodic word domain either. Since the syllable serves as the domain of application for FA in the Fuzhou dialect, when the derivation moves from the syllable to the prosodic word, properties of the syllable should not be further modified according to the DIC, which clearly goes against the facts. Notice that, in the two cases where the DIC fails to apply (i.e., ι with embedded φ and ω with embedded σ), the embedded domain and the larger domain are constructed by referring to different types of information. By contrast, in the cases where the DIC plays a role (i.e., φ with embedded ω and φ with embedded CG), both the embedded domain and the larger domain are defined by making use of morphosyntactic information. Therefore, we can conclude that the DIC, as an integral part of the prosodic phonology of the Fuzhou dialect, holds only within the morphosyntax-based hierarchy in this dialect.

Concluding remarks  231 The investigation of the Fuzhou phonological system in this book have thus not only provided persuasive evidence for the basic tenets of prosodic phonology, but also shed new light on several outstanding theoretical issues. With the data from the Fuzhou dialect, this book has offered suggestions to revise the SLH, confirmed the status of the clitic group, pointed out the possibility of combining the RBA and the EBA, and proposed a condition to restrict the application of phonological operations across domain boundaries. The discussion in this book, in sum, has contributed to furthering our understanding of prosodic phonology and the relationship between phonology and other components of the grammar.

8.3  Future research This book is the first attempt to establish a systematic analysis of the Fuzhou phonological system from the perspective of prosodic phonology. Although many important topics regarding the Fuzhou phonological system and the prosodic phonology theory have been included in this book, there are still topics of interest and relevance to Fuzhou phonology and prosodic phonology to be explored and addressed in future research. On the one hand, the discussion of Fuzhou phonological system in this book focuses on the construction of prosodic domains and the behavior of phonological rules within different domains but does not touch on the linguistic mechanisms underlying these rules. After reading this book, the reader may be curious about questions such as (a) why Group B variants of Fuzhou alternating finals co-occur only with tones 213, 242, and 23 but not with the other four tones in this dialect; (b) why the difference in the attachment direction of clitics leads to the difference in rule application between the two types of clitic groups; and (c) what makes the CL rule different from TS and FC in terms of the application within the phonological phrase domain and why only CL is blocked within phonological phrases constructed through φ-restructuring. To obtain a deeper understanding of the phonological system of the Fuzhou dialect, special attention needs to be paid to these unsolved questions in future research. On the other hand, this book has offered suggestions about how to deal with some theoretical issues in the prosodic phonology theory, but it is necessary to point out that these suggestions are made based on evidence generated from the Fuzhou dialect only. Is the Layeredness constraint of the SLH violable crosslinguistically? Can we employ the approach combining the RBA and the EBA to construct the phonological phrase domain in other languages as well? Is the DIC a universal “no look-back” device that conditions the cyclic application of phonological rules across domain boundaries? None of these questions can be easily answered if we do not have any support from languages other than the Fuzhou dialect. Therefore, in-depth analysis of relevant phonological phenomena in other languages needs to be done in the future, which may yet produce further evidence for the explanations and hypotheses proposed in this book.

References

Abney, Steven. 1987. The English noun phrase in its sentential aspect. Ph.D. Dissertation, MIT. Adam, T. B. 1891. An English-Chinese dictionary of the Foochow dialect. Fuzhou: Methodist Episcopal Mission Press. ———. 1905. An English-Chinese dictionary of the Foochow dialect. Second edition. Fuzhou: Methodist Episcopal Mission Press. Baker, Mark. 1988. Incorporation: A theory of grammatical function changing. Chicago, IL: The University of Chicago Press. Baldwin, Caleb C. 1871. A manual of the Foochow dialect. Fuzhou: Methodist Episcopal Press. ———. 1909. A manual of the Foochow dialect. Revised and enlarged edition. Fuzhou: Fuzhou College Press. Beckman, Mary E. and Janet B. Pierrehumbert. 1986. Intonational Structure in Japanese and English. Phonology Yearbook, 3: 15–70. Bickmore, Lee. 1990. Branching nodes and prosodic categories. In Sharon Inkelas and Draga Zec (eds.), The phonology-syntax connection, 1–17. Chicago, IL: The University of Chicago Press. Bing, Janet M. 1979. Aspects of English prosody. Ph.D. Dissertation, University of Massachusetts, Amherst. Booij, Geert. 1981. Generatieve fonologie van het Nederlands. Utrecht: Het Spectrum. ———. 1983. Principles and parameters in prosodic phonology. Linguistics, 21: 249–280. ———. 1984. Neutral vowels and the autosegmental analysis of Hungarian vowel harmony. Linguistics, 22: 629–641. ———. 1985a. The interaction of phonology and morphology in prosodic phonology. In Edmund Gussmann (ed.), Phono-morphology: Studies in the interaction of phonology and morphology, 23–34. Lublin: Katolicki Universytet Lubelski. ———. 1985b. Coordination reduction in complex words: A  case for prosodic phonology. In Harry van der Hulst and Norval Smith (eds.), Advances in nonlinear phonology, 143–160. Dordrecht: Foris. ———. 1986. Two cases of external sandhi in French: Enchainement and liaison. In Henning Andersen (ed.), Sandhi phenomena in the languages of Europe, 93–103. Berlin: Mouton de Gruyter. ———. 1996. Cliticization as prosodic integration: The case of Dutch. The Linguistic Review, 13: 219–242. ———. 1999. The role of the prosodic word in phonotactic generalizations. In T. Alan Hall and Ursula Kleinhenz (eds.), Studies on the phonological word, 47–72. Amsterdam & Philadelphia: John Benjamins Publishing Company.

References  233 Booij, Geert and Rochelle Lieber. 1993. On the simultaneity of morphological and prosodic structure. In Sharon Hargus and Ellen M. Kaisse (eds.), Phonetics and phonology 4: Studies in lexical phonology, 23–44. San Diego, CA: Academic Press. Booij, Geert and Jerzy Rubach. 1984. Morphological and prosodic domains in lexical phonology. Phonology Yearbook, 1: 1–27. ———. 1987. Postcyclic versus postlexical rules in lexical phonology. Linguistic Inquiry, 18: 1–44. Borer, Hagit. 2005. Structuring sense. (Vol. 1: In name only; Vol. 2: The normal course of events.) New York: Oxford University Press. Bowers, John. 1993. The syntax of predication. Linguistic Inquiry, 24: 591–656. Carnie, Andrew. 2007. Syntax: A generative introduction. Malden, MA  & Oxford: Blackwell. Chan, L-L. Lily. 1998. Fuzhou tone sandhi. Ph.D. Dissertation, UCSD. Chan, Marjorie K-M. 1980. Syntax and phonology interface: The case of tone sandhi in the Fuzhou dialect of Chinese. MS., University of Washington. ———. 1985. Fuzhou phonology: A non-linear analysis of tone and stress. Ph.D. Dissertation, University of Washington. Chao, Yuen-Ren. 1930. A system of tone letters. Le Maître Phonétique, 45: 24–27. ———. 1968. A grammar of spoken Chinese. Berkeley: University of California Press. Chen, Leo and Jerry Norman. 1965a. An introduction to the Foochow dialect. San Francisco: San Francisco State College. ———. 1965b. Foochow-English glossary. San Francisco: San Francisco State College. Chen, Matthew Y. 1984. “Unfolding latent principles of literary taste”—poetry as a window onto language. The Tsing Hua Journal of Chinese Studies, 16: 203–240. ———. 1985. The syntax of Xiamen tone sandhi. MS., UCSD. ———. 1987. The syntax of Xiamen tone sandhi. Phonology Yearbook, 4: 109–150. ———. 1990. What must phonology know about syntax? In Sharon Inkelas and Draga Zec (eds.), The phonology-syntax connection, 19–46. Chicago, IL: The University of Chicago Press. ———. 2000. Tone sandhi: Patterns across Chinese dialects. Cambridge & New York: Cambridge University Press. Chen, Matthew Y., Xiuhong Yan, and Lian-Hee Wee. 2004. Hakka tone sandhi: Corpus and analytical challenges. Journal of Chinese Linguistics Monograph Series, 21: 1–354. Chen, Zeping 陈泽平. 1998. Fuzhou fangyan yanjiu 福州方言研究 [A study of the Fuzhou dialect]. Fuzhou: Fujian People’s Publishing House. Chen, Zhangtai 陈章太 and Rulong Li 李如龙. 1991. Minyu yanjiu 闽语研究 [A study of Min dialects]. Beijing: Yuwen chubanshe. Cheng, Lisa L-S., C-T. James Huang, Y-H. Audrey Li, and C-C. Jane Tang. 1993. Three ways to get passive. MS., University of California, Irvine, USC, Academia Sinica. ———. 1996. Hoo, hoo, hoo: The causative, passive, and dative in Taiwanese. In PangHsin Ting (ed.), Contemporary studies on the min dialects, Journal of Chinese Linguistics Monograph Series, vol. 14, 146–203. Chiu, Bonnie. 1995. An object clitic projection in Mandarin Chinese. Journal of East Asian Linguistics, 4: 77–117. Cho, Young-Mee Yu. 1990. Syntax and phrasing in Korean. In Sharon Inkelas and Draga Zec (eds.), The phonology-syntax connection, 47–62. Chicago, IL: The University of Chicago Press. Chomsky, Noam. 1973. Conditions on transformations. In Stephen Anderson and Paul Kiparsky (eds.), A festschrift for Morris Halle, 232–286. New York: Holt, Rinehart & Winston.

234  References ———. 1981. Lectures on government and binding. Dordrecht: Foris. ———. 1982. Some concepts and consequences of the theory of government and binding. Cambridge, MA: MIT Press. ———. 1986. Barriers. Cambridge, MA: MIT Press. ———. 1993. A minimalist program for linguistic theory. In Kenneth Hale and Samuel Jay Keyser (eds.), The view from building 20, 1–52. Cambridge, MA: MIT Press. ———. 1995. The minimalist program. Cambridge, MA: MIT Press. ———. 2000. Minimalist inquiries: The framework. In Roger Martin, David Michaels, and Juan Uriagereka (eds.), Step by step: Essays on minimalist syntax in honor of Howard Lasnik, 89–155. Cambridge, MA: MIT Press. ———. 2001. Derivation by phase. In Michael Kenstowicz (ed.), Ken Hale: A life in language, 1–52. Cambridge, MA: MIT Press. ———. 2004. Beyond explanatory adequacy. In Adriana Belletti (ed.), Structures and beyond: The cartography of syntactic structures, vol. 3, 104–131. Oxford: Oxford ­University Press. Clements, George N. 1978. Tone and syntax in Ewe. In Donna Jo Napoli (ed.), Elements of tone, stress, and intonation, 21–99. Washington, DC: Georgetown University Press. Cocchi, Gloria. 2000. Free clitics and bound affixes: Towards a unitary analysis. In Birgit Gerlach and Janet Grijzenhout (eds.), Clitics in phonology, morphology and syntax, 85–119. Amsterdam & Philadelphia: John Benjamins Publishing Company. Cohn, Abigail C. 1989. Stress in Indonesian and bracketing paradoxes. Natural Language and Linguistic Theory, 7: 167–216. Condoravdi, Cleo. 1990. Sandhi rules of Greek and prosodic theory. In Sharon Inkelas and Draga Zec (eds.), The phonology-syntax connection, 63–84. Chicago, IL: The University of Chicago Press. Cowper, Elizabeth A. 1992. A concise introduction to syntactic theory: The governmentbinding approach. Chicago, IL: The University of Chicago Press. Cowper, Elizabeth A. and Keren D. Rice. 1987. Are phonosyntactic rules necessary? ­Phonology Yearbook, 4: 185–194. Crysmann, Berthold. 1997. Cliticization in European Portuguese using parallel morphosyntactic constraints. In Miriam Butt and Tracy Holloway King (eds.), Proceedings of the LFG97 conference. Stanford, CA: CSLI Publications. ———. 2000. Clitics and coordination in linear structure. In Birgit Gerlach and Janet ­Grijzenhout (eds.), Clitics in phonology, morphology and syntax, 121–159. Amsterdam & Philadelphia: John Benjamins Publishing Company. Crystal, David. 2008. A dictionary of linguistics and phonetics. Sixth edition. Malden, MA & Oxford: Blackwell. Culicover, Peter W. 1997. Principles and parameters: An introduction to syntactic theory. New York, NY: Oxford University Press. Davis, Stuart. 2011. Quantity. In John Goldsmith, Jason Riggle, and Alan C. L. Yu (eds.), The handbook of phonological theory. Second edition, 103–140. Malden, MA & Oxford: Blackwell. Dixon, Robert. 1977a. A grammar of Yidiɲ. Cambridge & New York: Cambridge University Press. ———. 1977b. Some phonological rules of Yidiɲ. Linguistic Inquiry, 8: 1–34. Donohue, Cathryn. 2013. Fuzhou tonal acoustics and tonology. Muenchen: LINCOM Europa. Downing, Laura J. 1999. Prosodic stem ≠ prosodic word in Bantu. In T. Alan Hall and Ursula Kleinhenz (eds.), Studies on the phonological word, 73–98. Amsterdam & Philadelphia: John Benjamins Publishing Company.

References  235 Duanmu, San. 1993. Rime length, stress, and association domains. Journal of East Asian Linguistics, 2: 1–44. Elfner, Emily. 2012. Syntax-prosody interactions in Irish. Ph.D. thesis, University of Massachusetts, Amherst. ———. 2013. Recursivity in prosodic phrasing: Evidence from Conamara Irish. In Seda Kan, Claire Moore-Cantwell, and Robert Staubs (eds.), Proceedings of the 40th annual meeting of the North-East linguistic society, 191–204. Amherst, MA: GLSA. Emonds, Joseph E. 1976. A transformational approach to English syntax: Root, structure preserving and local transformations. New York: Academic Press. Falk, Yehuda N. 2001. Lexical-functional grammar: An introduction to parallel constraintbased syntax. Stanford, CA: CSLI Publications. Feng, Aizhen 冯爱珍. 1998. Fuzhou fangyan cidian 福州方言词典 [The dictionary of the Fuzhou dialect]. Nanjing: Jiangsu Education Publishing House. Feng, Shengli 冯胜利. 1995. Guanyue lilun yu Hanyu de beidong ju 管约理论与汉语的 被动句 [GB theory and passive sentences in Chinese]. Zhongguo yuyanxue luncong 中国语言学论丛, 1: 1–28. Frota, Sonia. 2000. Prosody and focus in European Portuguese. New York: Garland Publishing. Gao, Ming-kai. 1947. The sandhi phenomenon of the Foochow dialect. Yenching Journal of Chinese Studies, 33: 129–144. Ghini, Mirco. 1993. Phonological phrase formation in Italian: A  new proposal. Toronto Working Papers in Linguistics, 12: 41–78. Gordon, Matthew. 2011. Stress systems. In John Goldsmith, Jason Riggle, and Alan C. L. Yu (eds.), The handbook of phonological theory. Second edition, 141–163. Malden, MA & Oxford: Blackwell. Grimshaw, Jane and Armin Mester. 1988. Light verbs and θ-marking. Linguistic Inquiry, 19: 205–232. Gussenhoven, Carlos and Haike Jacobs. 1998. Understanding phonology. London: Arnold. Hale, Kenneth and Samuel Jay Keyser. 1993. On argument structure and the lexical expression of syntactic relations. In Kenneth Hale and Samuel Jay Keyser (eds.), The view from building 20, 53–109. Cambridge, MA: MIT Press. ———. 2002. Prolegomenon to a theory of argument structure. Cambridge, MA: MIT Press. Hale, Kenneth and Elisabeth Selkirk. 1987. Government and tonal phrasing in Papago. Phonology Yearbook, 4: 151–183. Hall, T. Alan. 1999. The phonological word: A review. In T. Alan Hall and Ursula Kleinhenz (eds.), Studies on the phonological word, 1–22. Amsterdam & Philadelphia: John Benjamins Publishing Company. Halle, Morris. 1978. Formal vs. functional considerations in phonology. Studies in the Linguistic Sciences, 8: 149–176. Halpern, Aaron L. 1995. On the placement and morphology of clitics. Stanford, CA: CSLI Publications. ———. 1998. Clitics. In Andrew Spencer and Arnold M. Zwicky (eds.), The handbook of morphology, 101–122. Malden, MA & Oxford: Blackwell. Hammond, Michael. 2011. The foot. In Marc van Oostendorp, Colin J. Ewen, Elizabeth Hume, and Keren Rice (eds.), The Blackwell companion to phonology, vol. 2, 949–979. Malden, MA & Oxford: Blackwell. Hannahs, Stephen J. 1995a. Prosodic structure and French morphophonology. Tübingen: Max Niemeyer Verlag.

236  References ———. 1995b. Glide formation, prefixation, and the phonological word in French. In Jon Amastae, Grant Goodall, Mario Montalbetti, and Marianne Phinney (eds.), Contemporary research in romance linguistics, 13–24. Amsterdam & Philadelphia: John Benjamins Publishing Company. Harris, Alice C. 2002. Endoclitics and the origins of Udi morphosyntax. New York: Oxford University Press. Harris, James W. 1983. Syllable structure and stress in Spanish: Nonlinear analysis. Cambridge, MA: MIT Press. Haspelmath, Martin and Andrea D. Sims. 2010. Understanding morphology. London: Hodder Education. Hayes, Bruce. 1980. A metrical theory of stress rules. Ph.D. Dissertation, MIT Press. ———. 1984/1989. The prosodic hierarchy in meter. In Paul Kiparsky and Gilbert ­Youmans (eds.), Rhythm and meter, 201–260. Orlando, FL: Academic Press. ———. 1995. Metrical stress theory: Principles and case studies. Chicago, IL: The University of Chicago Press. Hayes, Bruce and Aditi Lahiri. 1991. Bengali intonational phonology. Natural Language and Linguistic Theory, 9: 47–96. He, Yuanjian 何元建. 2011. Xiandai Hanyu shengcheng yufa 现代汉语生成语法 [A generative grammar of Mandarin Chinese]. Beijing: Beijing daxue chubanshe. Hogg, Richard and C. B. McCully. 1987. Metrical phonology: A coursebook. Cambridge & New York: Cambridge University Press. Hsu, Hui-Chuan. 1994. Constraint-based phonology and morphology: A survey of languages in China. Ph.D. Dissertation, UCSD. Huang, C-T. James, Y-H. Audrey Li, and Yafei Li. 2009. The syntax of Chinese. Cambridge & New York: Cambridge University Press. Hulst, Harry van der. 1984. Syllable structure and stress in Dutch. Dordrecht: Foris. Hulst, Harry van der and Norval Smith (eds.). 1982. The structure of phonological representations (Part I). Dordrecht: Foris. Hung, Tony T-N. 1987. Syntactic and semantic aspects of Chinese tone sandhi. Ph.D. Dissertation, UCSD. ———. 1990. Syntax-phonology interaction in Chinese tone sandhi: An overview. Journal of Chinese Language Teachers Association, 15: 43–59. Hyman, Larry M., Francis Katamba, and Livingstone Walusimbi. 1987. Luganda and the strict layer hypothesis. Phonology Yearbook, 4: 87–108. Inkelas, Sharon. 1989. Prosodic constituency in the lexicon. Ph.D. Dissertation, Stanford University. Published 1990, Outstanding Dissertations in Linguistics Series. New York: Garland Publishing. Inkelas, Sharon and Draga Zec. 1995. Syntax-phonology interface. In John Goldsmith (ed.), The handbook of phonological theory, 535–549. Malden, MA & Oxford: Blackwell. Itô, Junko. 1986. Syllable theory in prosodic phonology. Ph.D. Dissertation, University of Massachusetts, Amherst. Itô, Junko and Armin Mester. 1992/2003. Weak layering and word binarity. In Takeru Honma, et al. (eds.), A new century of phonology and phonological theory: A festschrift for Professor Shosuke Haraguchi on the occasion of his sixtieth birthday, 26–65. Tokyo: Kaitakusha. ———. 2012. Recursive prosodic phrasing in Japanese. In T. Borowsky, S. Kawahara, T. Shinya, and M. Sugahara (eds.), Prosody matters: Essays in honor of Elisabeth Selkirk, 280–303. London: Equinox.

References  237 Jensen, John T. 1993. English phonology. Amsterdam  & Philadelphia: John Benjamins Publishing Company. Jun, Sun-Ah. 1993. The phonetics and phonology of Korean prosody. Ph.D. Dissertation, the Ohio State University, Columbus. Kabak, Bariş and Irene Vogel. 2001. The phonological word and stress assignment in Turkish. Phonology, 18: 315–360. Kager, René. 1996. The metrical theory of word stress. In John Goldsmith (ed.), The handbook of phonological theory, 367–402. Malden, MA & Oxford: Blackwell. ———. 2007. Feet and metrical stress. In Paul de Lacy (ed.), The Cambridge handbook of phonology. Cambridge & New York: Cambridge University Press. Kahn, Daniel. 1976. Syllable-based generalizations in English phonology. Ph.D. Dissertation, MIT Press. Kaisse, Ellen M. 1985. Connected speech: The interaction of syntax and phonology. Orlando, FL & London: Academic Press. Kanerva, Jonni. 1989. Focus and phrasing in Chichewa phonology. Ph.D. Dissertation, Stanford University. ———. 1990. Focusing on phonological phrases in Chichewa. In Sharon Inkelas and Draga Zec (eds.), The phonology-syntax connection, 145–161. Chicago, IL: The University of Chicago Press. Kang, Ongmi. 1992. Word-internal prosodic words in Korean. Proceedings of the North East Linguistic Society, 22: 243–257. Kari, Ethelbert. 2003. Clitics in Degema: A meeting point of phonology, morphology, and syntax. Ph.D. Dissertation, Tokyo University of Foreign Studies. Kaye, Jonathan. 1995. Derivations and interfaces. In Jacques Durand and Francis Katamba (eds.), Frontiers of phonology, 289–332. London & New York: Longman. Kenstowicz, Michael J. 1994. Phonology in generative grammar. Cambridge, MA  & Oxford: Blackwell. Kenstowicz, Michael J. and Hyang-Sook Sohn. 1997. Phrasing and focus in Northern Kyungsang Korean. In Pier M. Bertinetto, et  al. (eds.), Certamen Phonologicum III, 137–149. Torino: Rosenberg & Sellier. Keyser, Samuel J. and Paul Kiparsky. 1984. Syllable structure in Finnish phonology. In M. Aronoff and R. Oehrle (eds.), Language sound structure, 7–31. Cambridge, MA: MIT Press. Kidima, Lukowa. 1990. Tone and syntax in Kiyada. In Sharon Inkelas and Draga Zec (eds.), The phonology-syntax connection, 195–216. Chicago, IL: The University of Chicago Press. Kim, No-Ju. 1997. Tone, segments, and their interaction in North Kyungsang Korean: A correspondence theoretic account. Ph.D. Dissertation, The Ohio State University. Kiparsky, Paul. 1979. Metrical structure assignment is cyclic. Linguistic Inquiry, 10: 421–442. ———. 1982a. Lexical morphology and phonology. In In-Seok Yang (ed.), Linguistics in the morning calm, 3–91. Seoul: Hanshin. ———. 1982b. From cyclic phonology to lexical phonology. In Harry van der Hulst and Norval Smith (eds.), The structure of phonological representations, vol. 1, 131–175. Dordrecht: Foris. ———. 1985. Some consequences of lexical phonology. Phonology Yearbook, 2: 83–138. Klavans, Judith L. 1982. Some problems in a theory of clitics. Bloomington: Indiana ­University Linguistics Club.

238  References ———. 1985. The independence of syntax and phonology in cliticization. Language, 61: 95–120. Kleinhenz, Ursula. 1996. The prosody of German clitics. In Artemis Alexiadou, Nanna Fuhrhop, Paul Law, and Sylvia Löhken (eds.), ZAS Papers in Linguistics, vol. 6, 81–95. Korpris, Craig A. and Anthony R. Davis. 2005. Endoclitics in Pashto: Implications for lexical integrity. Paper presented at The Fifth Mediterranean Morphology Meeting. Fréjus, France. Kratzer, Angelika. 1993. On external arguments. In Elena Benedicto and Jeff Runner (eds.), University of Massachusetts occasional papers 17: Functional projections, 103–130. Amherst, MA: Graduate Students Linguistic Association. Ladd, D. Robert. 1986. Intonational phrasing: The case for recursive prosodic structure. Phonology Yearbook, 3: 311–340. Lan Ya-shiu. 1953. Phonetic system of the Foochow dialect. Bulletin of the College of Arts, National Taiwan University, 5: 241–331. Larson, Richard K. 1988. On double object constructions. Linguistic Inquiry, 19: 335–391. Lees, Robert B. 1961. The phonology of modern standard Turkish. The Hague: Mouton. Li, Charles N. and Sandra A. Thompson. 1981. Mandarin Chinese: A functional reference grammar. Berkeley: University of California Press. Li, Rulong 李如龙 and Yuzhang Liang 梁玉璋. 2001. Fuzhou fangyan zhi 福州方言志 [A record of the Fuzhou dialect]. Fuzhou: Haifeng chubanshe. Li, Rulong 李如龙, Yuzhang Liang 梁玉璋, and Tianquan Chen 陈天泉. 1979. Fuzhouhua yuyin yanbian gaishuo 福州话语音演变概说 [An outline of phonetic changes in the Fuzhou dialect]. Zhongguo yuwen 中国语文, 4: 287–293. Li, Rulong 李如龙, Yuzhang Liang 梁玉璋, Guangchun Zou 邹光椿, and Zeping Chen 陈泽平. 1994. Fuzhou fangyan cidian 福州方言词典 [The dictionary of the Fuzhou dialect]. Fuzhou: Fujian renmin chubanshe. Li, Yafei. 2005. X0: A theory of the morphology-syntax interface. Cambridge, MA: MIT Press. Li, Zhuqing. 1998. Fuzhou-English dictionary. Kensington, MD: Dunwoody Press. ———. 2002. Fuzhou phonology and grammar. Hyattsville: Dunwoody Press. Liang, Yuzhang 梁玉璋. 1982. Fuzhou fangyan de “qiejiao ci” 福州方言的“切脚词” [Fuzhou word derivation by the principle of fanqie]. Fangyan 方言, 3.1: 37–46. ———. 1983a. Fuzhou fangyan chongdieshi mingci 福州方言重叠式名词 [Nominal reduplications in the Fuzhou dialect]. Zhongguo yuwen 中国语文, 3: 177–184. ———. 1983b. Fuzhou fangyan liandu yinbian yu yuyi fenbie 福州方言连读音变与语义 分别 [Semantic distinctions of sandhi in the Fuzhou dialect]. Fangyan 方言, 3: 166–169. ———. 1986. Fuzhou fangyan de yuliu yinbian 福州方言的语流音变 [Sandhi in the Fuzhou dialect]. Yuyan yanjiu 语言研究, 2: 85–97. Liberman, Mark. 1975. The intonational system of English. Ph.D. Dissertation, MIT Press. Liberman, Mark and Alan Prince. 1977. On stress and linguistic rhythm. Linguistic Inquiry, 8: 249–336. Lin, Hansheng 林寒生. 2002. Mindong fangyan cihui yufa yanjiu 闽东方言词汇语法研究 [A study of vocabulary and grammar of the Eastern Min dialects]. Kunming: Yunnan University Press. Lin, Jo-wang. 1994. Lexical government and tone group formation in Xiamen Chinese. Phonology, 11: 237–275. Maclay, Robert S. and Caleb C. Baldwin. 1870. An alphabetic dictionary of the Chinese language in the Foochow dialect. Fuzhou: Methodist Episcopal Mission Press. ———. 1898. An alphabetic dictionary of the Chinese language in the Foochow dialect. Revised edition. Fuzhou: Methodist Episcopal Mission Press.

References  239 ———. 1929. Dictionary of the Foochow dialect. Revised and enlarged edition of Maclay & Baldwin 1898 by Samuel H. Leger. Shanghai: The Presbyterian Mission Press. Manzini, Rita. 1983. Syntactic conditions on phonological rules. MIT Working Papers in Linguistics, 5: 1–9. Mascaró, Joan. 1976. Catalan phonology and the phonological cycle. Ph.D. Dissertation, MIT Press. Matthews, Peter H. 1991. Morphology. Cambridge & New York: Cambridge University Press. McCarthy, John and Alan Prince. 1993. Generalized alignment. In Geert Booij and Jaap van Marle (eds.), Yearbook of morphology, 79–153. Dordrecht: Kluwer. ———. 1995. Faithfullness and reduplicative identity. In J. Beckman, L. W. Dickey, and S. Urbancxyk (eds.), University of Massachusetts occasional papers in linguistics 18: Papers in optimality theory, 249–384. Amherst, MA: GLSA. Mester, Armin. 1994. The quantitative trochee in Latin. Natural Language and Linguistic Theory, 12: 1–61. Miller, Philip. 1992. Clitics and constituents in phrase structure grammar. New York: Garland Publishing. Mithun, Marianne. 1984. The evolution of noun incorporation. Language, 60.4: 847–894. Mohanan, Karuvannur P. 1982. Lexical phonology. Ph.D. Dissertation, MIT Press. Monachesi, Paola. 1999. A lexical approach to Italian cliticization. Stanford, CA: CSLI Publications. Neijt, Anneke. 1985. Clitics in arboreal phonology. In Harry van der Hulst and Norval Smith (eds.), Advances in nonlinear phonology, 179–192. Dordrecht: Foris. Nespor, Marina. 1999. The phonology of clitic groups. In Henk van Riemsdijk (ed.), Clitics in the languages of Europe, 865–887. Berlin: Mouton de Gruyter. Nespor, Marina and Angela Ralli. 1996. Morphology-phonology interface: Phonological domains in Greek compounds. The Linguistic Review, 13: 357–382. Nespor, Marina and Irene Vogel. 1986. Prosodic phonology. Dordrecht: Foris. ———. 2007. Prosodic phonology: With a new foreword. Berlin: Mouton de Gruyter. Odden, David. 1987. Kimatuumbi phrasal phonology. Phonology Yearbook, 4: 13–26. ———. 1990. Syntax, lexical rules and postlexical rules in Kimatuumbi. In Sharon Inkelas and Draga Zec (eds.), The phonology-syntax connection, 259–277. Chicago, IL: The University of Chicago Press. ———. 1996. The phonology and morphology of Kimatuumbi. Oxford: Clarendon Press. Parker, Edward H. 1879a. Tonic and vocal modification in the Foochow dialect. China Review, 7: 182–187. ———. 1879b. New Foochow colloquial words. China Review, 7: 415–418. ———. 1881a. Foochow syllabary. China Review, 9: 63–82. ———. 1881b. Characterless Chinese words. China Review, 9: 85–88. Peking University. 1962. Hanyu fangyin zihui 汉语方音字汇 [The syllabary of Chinese dialects]. Beijing: Wenzi gaige chubanshe. ———. 1964. Hanyu fanyan cihui 汉语方言词汇 [The vocabulary of Chinese dialects]. Beijing: Wenzi gaige chubanshe. ———. 1989. Hanyu fangyin zihui 汉语方音字汇 [The syllabary of Chinese dialects]. Second edition. Beijing: Wenzi gaige chubanshe. ———. 1995. Hanyu fanyan cihui 汉语方言词汇 [The vocabulary of Chinese dialects]. Second edition. Beijing: Yuwen chubanshe. Peperkamp, Sharon. 1997. Prosodic words. Ph.D. Dissertation, University of Amsterdam. Pierrehumbert, Janet B. and Mary E. Beckman. 1988. Japanese tone structure. Cambridge, MA: MIT Press.

240  References Prieto, Pilar. 2005. Syntactic and eurhythmic constraints on phrasing decisions. Studia Linguistica, 59: 194–222. ———. 2006. Phonological phrasing in Spanish. In Fernando Martínez-Gil and Sonia Colina (eds.), Optimality-theoretic studies in Spanish phonology, 39–61. Amsterdam & Philadelphia: John Benjamins Publishing Company. Prince, Alan. 1983. Relating to the grid. Linguistic Inquiry, 14: 19–100. ———. 1984. Phonology with tiers. In M. Aronoff and R. Oehrle (eds.), Language sound structure, 234–244. Cambridge, MA: MIT Press. Prince, Alan and Paul Smolensky. 1993. Optimality theory: Constraint interaction in generative grammar. Cambridge, MA: MIT Press. Pulleyblank, Douglas. 1986. Tone in lexical phonology. Dordrecht: Reidel. Raffelsiefen, Renate. 1993. Relating words: A model of base-recognition. Linguistic Analysis, 23: 3–159. ———. 1999. Phonological constraints on English word formation. In Geert Booij and Jaap van Marle (eds.), Yearbook of morphology, 225–287. Dordrecht: Kluwer. Reinhart, Tanya. 1976. The syntactic domain of anaphora. Ph.D. Dissertation, MIT Press. ———. 1983. Anaphora and semantic interpretation. London: Croom Helm. Rice, Keren D. 1993. The structure of the slave (Northern Athabaskan) verb. In Sharon Hargus and Ellen M. Kaisse (eds.), Phonetics and phonology 4: Studies in lexical phonology, 145–171. San Diego, CA: Academic Press. Riemsdijk, Henk van. 1999. Clitics: A state of the art report. In Henk van Riemsdijk (ed.), Clitics in the languages of Europe, 1–30. Berlin: Mouton de Gruyter. Rizzi, Luigi and Leonardo Savoia. 1993. Conditions on /u/ propagation in Southern Italian dialects: A  locality parameter for phonosyntactic processes. In Adriana Belletti (ed.), Syntactic theory and the dialects of Italy, 252–318. Torino: Resenberg and Sellier. Russell, Kevin. 1999. The “word” in two polysynthetic languages. In T. Alan Hall and Ursula Kleinhenz (eds.), Studies on the phonological word, 203–221. Amsterdam  & Philadelphia: John Benjamins Publishing Company. Russi, Cinzia. 2008. Italian clitics: An empirical study. Berlin: Mouton de Gruyter. Safir, Ken. 1985. Binding in relatives and LF. Glow Newsletter, 14: 77–79. Sandalo, Filomena and Hubert Truckenbrodt. 2002. Some notes on phonological phrasing in Brazilian Portuguese. MIT Working Papers in Linguistics, 42: 285–310. Sapir, Edward. 1930. Southern Paiute, a Shoshonean language. Proceedings of the American Academy of Arts and Sciences, 65.1: 1–296. Selkirk, Elisabeth. 1972. The phrase phonology of English and French. Ph.D. Dissertation, MIT Press. ———. 1978. On prosodic structure and its relation to syntactic structure. Published 1981, in Thorstein Fretheim (ed.), Nordic Prosody II, 111–140. Trondheim: Tapir. ———. 1980a. Prosodic domains in phonology: Sanskrit revisited. In Mark Aronoff and Mary-Louise Kean (eds.), Juncture, 107–129. Saratoga, CA: Anma Libri. ———. 1980b. The role of prosodic categories in English word stress. Linguistic Inquiry, 11: 563–605. ———. 1984. Phonology and syntax: The relation between sound and structure. Cambridge, MA: MIT Press. ———. 1986. On derived domain in sentence phonology. Phonology Yearbook, 3: 371–405. ———. 1996. The prosodic structure of function words. In James L. Morgan and Katherine Demuth (eds.), Signal to syntax: Bootstrapping from speech to grammar in early acquisition, 187–214. Mahwah, NJ: Lawrence Erlbaum Associates. ———. 2000. The interaction of constraints on prosodic phrasing. In Merle Horne (ed.), Prosody: Theory and experiments, 231–262. Dordrecht: Kluwer.

References  241 ———. 2005. Comments on intonational phrasing in English. In S. Frota, M. Vigário, and M. J. Freitas (eds.), Prosodies, 11–58. Berlin: Mouton de Gruyter. ———. 2009. On clause and intonational phrase in Japanese: The syntactic grounding of prosodic constituent structure. Gengo Kenkyu, 136: 35–73. ———. 2011. The syntax—phonology interface. In J. Goldsmith, J. Riggle, and A. C. L. Yu (eds.), The handbook of phonological theory. Second edition, 435–484. Oxford: Wiley-Blackwell. Selkirk, Elisabeth and Tong Shen. 1990. Prosodic domains in Shanghai Chinese. In Sharon Inkelas and Draga Zec (eds.), The phonology-syntax connection, 313–337. Chicago, IL: The University of Chicago Press. Selkirk, Elisabeth and Koichi Tateishi. 1988. Constraints on minor phrase formation in Japanese. In Gary Larson and Diane Brentari (eds.), Papers from the 24th annual regional meeting of the Chicago linguistic society, 316–336. Chicago, IL: The Chicago Linguistic Society. Shih Chi-Lin. 1986. The prosodic domain of tone sandhi in Chinese. Ph.D. Dissertation, UCSD. ———. 1997. Mandarin third tone sandhi and prosodic structure. In Jialing Wang and Norval Smith (eds.), Studies in Chinese phonology, 81–124. New York: Mouton de Gruyter. Smith, Norval. 1986. Evidence for the foot as a hierarchical unit in Žuǀ’hõasi. MS., University of Amsterdam. Sun, Chaofen. 2006. Chinese: A linguistic introduction. Cambridge & New York: Cambridge University Press. Tang, Sze-Wing 邓思颖. 2003. Hanyu fangyan yufa de canshu lilun 汉语方言语法的参 数理论 [A parametric theory of Chinese dialectal grammar]. Beijing: Peking University Press. ———. 2008. Hanyu beidongju jufa fenxi de chongxi sikao 汉语被动句句法分析的重新 思考 [Syntactic analysis of Chinese passives: revisited]. Dangdai yuyanxue 当代语言学, 10: 308–319. ———. 2010. Xingshi Hanyu jufaxue 形式汉语句法学 [Formal Chinese syntax]. Shanghai: Shanghai Educational Publishing House. Tao, Yu-min 陶燠民. 1930. Minyin yanjiu 闽音研究 [A study of Min phonology]. Bulletin of the National Research Institute of History and Philology, 1.4: 445–470. (Also published in 1956 by Beijing: Kexue chubanshe). Ternes, Elmar. 1973. The phonemic analysis of Scottish Gaelic: Based on the dialect of Applecross, Ross-shire. Hamburg: Helmut Buske Verlag. Ting, Jen. 1995. A non-uniform analysis of the passive construction in Mandarin Chinese. Ph.D. Dissertation, University of Rochester. ———. 1996. A non-uniform analysis of the passive construction in Mandarin Chinese. Paper presented at The 8th North American Conference on Chinese Linguistics, University of Illinois. Trommelen, Mieke. 1983. The syllable in Dutch. Dordrecht: Foris. Truckenbrodt, Hubert. 1995. Phonological phrase: Their relation to syntax, focus and prominence. Ph.D. Dissertation, MIT Press. ———. 1999. On the relation between syntactic phrases and phonological phrases. Linguistic Inquiry, 30: 219–255. ———. 2002. Variation in p-phrasing in Bengali. In Pierre Pica and Johan Rooryck (eds.), Linguistic variation yearbook, vol. 2, 259–303. Amsterdam & Philadelphia: John Benjamins Publishing Company. Vigário, Marina. 2003. The prosodic word in European Portuguese. Berlin: Mouton de Gruyter.

242  References Vogel, Irene. 1990. The clitic group in prosodic phonology. In Joan Mascaró and Marina Nespor (eds.), Grammar in progress, 447–454. Dordrecht: Foris. ———. 1991. Level ordering in Italian lexical phonology? In Pier Marco Bertinetto, Michael Kenstowicz, and Michele Loporcaro (eds.), Certamen phonologicum II, vol. 11, 81–101. Torino: Rosenberg & Sellier. ———. 2009. The status of the clitic group. In Janet Grijzenhout and Barış Kabak (eds.), Phonological domains: Universals and deviations, 15–46. Berlin: Mouton de Gruyter. Wang, Li 王力. 1954. Zhongguo yufa lilun 中国语法理论 [Theory of Chinese grammar]. Beijing: Zhonghua shuju. Wee, Lian-Hee. 2004. Inter-tier correspondence theory. Ph.D. Dissertation, Rutgers, State University of New Jersey. Wright, Martha S. 1983. A metrical approach to tone sandhi in Chinese dialects. Ph.D. Dissertation, University of Massachusetts, Amherst. Xu, Debao. 2001. Lexical third tone sandhi and the lexical organization of Mandarin. In Debao Xu (ed.), Chinese phonology in generative grammar, 45–94. San Diego, CA: Academic Press. Yip, Moira. 1980. The tonal phonology of Chinese. Ph.D. Dissertation, MIT Press. You, Shuxiang. 2017. Prosodic phonology of the Fuzhou dialect. Ph.D. Dissertation, University of Wisconsin–Madison. ———. 2018a. No look-back devices in prosodic phonology: A case study of the Fuzhou dialect. Paper presented at The 26th annual conference of international association of Chinese linguistics (IACL-26), Madison, WI. ———. 2018b. Lun Fuzhou fangyan de nianfu chengfen yu nianfuzu 论福州方言的粘附 成分与粘附组 [Clitics and the clitic group in the Fuzhou dialect]. International Journal of Chinese Linguistics, 5.1: 124–165. ———. 2018c. Two types of tone sandhi rules in the Fuzhou dialect. Paper presented at The 13th phonetic conference of China. Guangzhou, China. ———. 2019. Enclitics and the clitic group consisting of “host + enclitic” in the Fuzhou dialect. In Hongming Zhang and Youyong Qian (eds.), Prosodic studies: Challenges and prospects, 80–107. London & New York: Routledge. Yu, Hui 于辉 and Yuxia Yin 尹玉霞. 2014. Chongdieshi sansheng biandiao de cihui yinxixue jieshi 重叠式三声变调的词汇音系学解释 [A lexical phonological explanation on third tone sandhi of reduplicates in Mandarin]. Yuyan yanjiu 语言研究, 34: 60–64. Yuan, Jiahua 袁家骅. 1960. Hanyu fangyan gaiyao 汉语方言概要 [A survey of Chinese dialects]. Beijing: Wenzi gaige chubanshe. ———. 2001. Hanyu fangyan gaiyao 汉语方言概要 [A survey of Chinese dialects]. Revised second edition. Beijing: Yuwen chubanshe. Zec, Draga. 1988. Sonority constraints on prosodic structure. Ph.D. Dissertation, Stanford University. ———. 1993. Rule domains and phonological change. In Sharon Hargus and Ellen M. Kaisse (eds.), Phonetics and phonology 4: Studies in lexical phonology, 365–405. San Diego, CA: Academic Press. Zhang, Hongming. 1992. Topics in Chinese phrasal tonology. Ph.D. Dissertation. UCSD. ———. 2017. Syntax-phonology interface: Argumentation from tone sandhi in Chinese dialects. London/ & New York: Routledge. Zhang, Hongming 张洪明. 2014. Yunlu yinxixue yu Hanyu yunlu yanjiu zhong de ruogan wenti 韵律音系学与汉语韵律研究中的若干问题 [Some issues on prosodic phonology and Chinese prosodic studies]. Dangdai yuyanxue 当代语言学, 16: 303–327.

References  243 Zhang, Hongming 张洪明 and Hui Yu 于辉. 2009. Cihui yinxixue yu Hanyu chongdieshi de yinxi yanjiu 词汇音系学与汉语重叠式的音系研究 [Lexical phonology and the phonological study of the reduplicated forms in Chinese]. Yuyanxue luncong 语言学论丛, 39: 506–521. Zheng, Yide 郑懿德. 1983. Fuzhou fangyan danyin dongci chongdieshi 福州方言单音动 词重叠式 [Reduplicated forms of monosyllabic verbs in the Fuzhou dialect]. Zhongguo yuwen 中国语文, 1: 30–39. ———. 1985. Fuzhou fangyan de “you” zi ju 福州方言的“有”字句 [“You” sentences in the Fuzhou dialect]. Fangyan 方言, 4: 309–313. ———. 1988a. Fuzhou fangyan xingrongci chongdieshi 福州方言形容词重叠式 [Reduplicated forms of adjectives in the Fuzhou dialect]. Fangyan 方言, 4: 301–311. ———. 1988b. Fuzhou fangyan “li” de cixing jiqi yongfa 福州方言“𠲥”的词性及其用法 [Part of speech and usages of “lɛ” in the Fuzhou dialect]. Zhongguo yuwen 中国语文, 6: 450–452. ———. 1995. Fuzhou fangyan de fangweici 福州方言的方位词 [Locatives in the Fuzhou dialect]. Fangyan 方言, 2: 151–155. Zwicky, Arnold M. 1977. On clitics. Bloomington: Indiana University Linguistics Club. ———. 1987. Suppressing the Z’s. Journal of Linguistics, 23: 133–148. Zwicky, Arnold M. and Geoffrey Pullum. 1983. Cliticization vs. inflection: English n’t. Language, 59: 502–513.

Index

adjective reduplication marker 98, 100, 109, 114, 118, 121, 126 – 127 adjunct 30, 52, 133 – 135, 137 – 143, 145, 147, 149, 151, 153 – 155, 157 – 158, 176, 185, 227 affix 64, 67 – 68, 70, 89 – 90, 94 – 95, 98, 107, 123 – 124, 191, 204 – 205n2; prefix 64 – 66, 68, 71, 76, 86, 89, 91n1 – 92n10, 204; suffix 49, 51, 64 – 67, 69 – 71, 76, 83, 86, 89, 92n11, 94 – 95 Align/Wrap Theory 19 – 20, 32, 34, 125, 128, 147, 151–152, 154, 177, 229 – 230 alternating final 6 – 7, 13 – 14, 34, 36n4, 40 – 43, 48, 58, 62, 82, 84, 92n10, 178, 180, 187n12, 205n5, 227, 231 application (of rules) 2, 9, 11, 15 – 17, 19 – 20, 29, 34 – 36, 38 – 40, 42 – 43, 45 – 49, 51 – 52, 55 – 63, 65 – 66, 68, 75 – 93, 95 – 97, 106 – 108, 113, 116 – 121, 124 – 139, 141, 145, 147 – 151, 154 – 155, 158, 165, 172 – 178, 180 – 182, 184 – 187, 192 – 196, 199 – 201, 204 – 206, 213 – 217, 219 – 231 argument 53, 57, 133, 137 – 142, 145 – 147, 149 – 150, 163, 168 – 171, 187n4 aspect marker: delimitative aspect marker 104, 111, 115, 119; durative aspect marker 100 – 101, 105, 109, 114, 118, 189; experiential aspect marker 101, 110, 114, 117, 118; perfective aspect marker 100, 102 – 103, 105, 110, 114 – 115, 118, 189; progressive aspect marker 188 – 189, 192, 196, 199, 205n5 binary 46, 49, 50, 52, 56, 60, 132, 134 blocking (of rules) 12, 81, 87, 96, 108 – 109, 111 – 116, 120, 126 – 127, 129 – 131, 133, 135, 139, 145, 150 – 151,

154 – 158, 163, 165 – 168, 171 – 172, 175 – 177, 181, 187n5, 194, 196 – 200, 214, 216, 220 – 222, 226 – 227, 230 branching 18, 28 – 29, 31, 33, 52 – 53, 57, 134, 137 – 138, 140 – 141, 148 – 151, 153 – 155, 157 – 158, 168, 170 – 172, 181 – 184, 186, 187n8, 200 – 201, 205, 220, 227, 230 c-command 141 – 142, 147, 172 citation tone 3 – 5, 10 – 12, 40, 49 – 51, 53 – 54, 56, 58, 78, 80 – 81, 142 clitic 26 – 28, 36 – 37n12, 64 – 65, 93 – 99, 107, 123 – 124n3, 147, 188 – 191, 200 – 201, 204 – 205, 226 – 227, 229, 231; enclitic 26 – 28, 35 – 36, 52, 75, 93 – 94, 97 – 101, 103 – 104, 106 – 108, 112 – 113, 116 – 117, 120 – 124, 169 – 171, 175 – 176, 188 – 189, 191 – 192, 194, 199 – 202, 204, 227, 229; proclitic 27 – 28, 35, 93 – 94, 97 – 98, 124, 188 – 192, 194, 196, 199 – 200, 202 – 205, 227, 229 clitic group 18, 21 – 22, 25 – 28, 34 – 37n14, 55, 63, 91, 93 – 98, 106 – 108, 112 – 113, 116 – 117, 120 – 124, 127, 147 – 148, 157, 171, 175 – 176, 180, 188, 191 – 192, 194 – 196, 199 – 205, 207, 225 – 231 complement (under X-bar framework) 18, 28 – 30, 92n4, 129 – 131, 148 – 149, 155, 157 – 158, 163, 171 – 172, 186, 227 compound 53 – 56, 64 – 66, 71 – 72, 76 – 77, 79, 81 – 84, 86, 88 – 91, 145, 187, 227 constraint 18 – 19, 24 – 26, 31 – 33, 40 – 41, 43, 45, 65 – 66, 97, 121 – 123, 137, 141, 151 – 153, 157, 174, 185, 187n6, 189, 201 – 203, 205, 207, 228 – 229, 231 contrastive prominence 1, 35, 206, 209, 211, 213, 218 – 220, 222 – 224, 227

Index  245 derived word 66, 68 – 70, 76, 79, 81, 83 – 84, 86, 88 – 90, 92n10, 227 diminutive noun 72 – 73, 77, 79 – 82, 84, 87 – 92n8, 116, 121, 227 domain formation 22, 63 – 64, 67, 125, 141, 205 – 206, 209, 224, 226 – 228 Domain Impenetrability Condition (DIC) 35, 125, 168, 174 – 177, 180, 185 – 186, 221 – 223, 228, 230 – 231 Edge/End-Based Approach (EBA) 18 – 20, 28, 30 – 35, 125, 128, 147, 150 – 155, 167, 172, 176 – 177, 185 – 186, 229 – 231 Exhaustivity 18, 24 – 26, 28, 31, 37n14, 121, 157, 185, 207, 229 Final Alternation (FA) 13, 16, 34, 38, 40 – 43, 62, 226 – 227, 230 Final Change (FC) 13 – 14, 16, 35, 40 – 43, 46 – 49, 52, 56 – 63, 75, 82 – 85, 88 – 93, 107, 113 – 116, 120 – 121, 123, 125, 177 – 178, 180 – 182, 184 – 187n13, 192, 194 – 196, 199 – 202, 204 – 205, 216, 219 – 221, 223 – 224, 226 – 227, 231 foot 17 – 18, 21 – 22, 24, 34 – 35, 37n11, 38, 43 – 50, 51 – 63, 75, 91, 131 – 134, 142 – 143, 147, 172, 187n4, 225 – 226 Foot Formation Rule 52 – 53, 62n1, 134, 137 – 138, 147 head 18, 28 – 30, 37n15, 53, 57, 71, 77, 83, 86, 89, 92n4, 108, 127 – 131, 133 – 145, 148 – 151, 153 – 159, 163 – 164, 170, 172, 176, 178, 182, 185 – 187n4, 205n1, 227, 230 Head Dominance Condition 128 – 130, 134, 147; Revised Head Dominance Condition 52, 134 – 135, 147 Headedness 18, 24 Initial Consonant Lenition (CL) 15 – 16, 35, 40, 52, 57, 63, 75, 85 – 93, 99, 104, 107, 116 – 121, 123, 125, 177, 180 – 182, 184 – 186, 192, 196 – 200, 202, 204, 216, 219 – 221, 224, 226 – 227, 231 intonational phrase 17 – 22, 24, 27, 33 – 36, 127, 147, 206 – 217, 219 – 225, 228, 230 intonation contour 147, 206 – 207, 211, 213, 216

Layeredness 18, 24 – 26, 31, 97, 121 – 124, 189, 201 – 204, 228 – 229, 231 length (of intonational phrase) 1, 35, 206, 209, 211, 213, 224 lexical government 143, 146 – 147 lexical phonology 81 – 82 locative marker 100, 106, 112, 116, 120 – 121, 189 Match Theory 19 – 20, 24, 32 – 34, 125, 128, 147, 151–154, 177, 229 maximal projection 28, 30, 137, 139, 143, 145, 148 – 150, 157 – 158, 187n5 monomorphemic word 66 – 67, 75 – 76, 79, 81 – 86, 88 – 92n9, 124n3, 227 mora 21 – 22, 25, 34, 36n11 – 38, 46 – 48, 53, 55, 66, 225 – 226 morphological process 9, 12, 35, 67 – 68, 72 – 73, 75, 81, 91, 200 – 201, 226 – 227 Morphological Tone Sandhi (MTS) 9, 12 – 14, 16, 35, 40, 63, 75, 77 – 78, 80 – 82, 84 – 85, 88 – 91, 113, 226 – 227 morphosyntactic word 1, 35, 54 – 55, 57 – 61, 63, 66 – 68, 73, 75, 80 – 82, 85, 88 – 91, 107, 116, 121, 123, 143, 173, 200, 226 – 227 neutral tone 49, 51, 92n8, 99, 103 – 104, 108, 171, 189 Optimality Theory (OT) 18 – 20, 24, 31 – 32, 151, 227 – 229 parameter/parameterized algorithm 30 – 31, 34, 125, 128, 143, 147, 150 – 154, 177, 230 pause 94, 206, 208, 210, 212 – 213, 216, 224 phonological phrase 17 – 22, 24 – 37n14, 52, 54 – 55, 57, 63, 91, 93, 97 – 98, 113, 116, 121 – 125, 127, 131 – 134, 142–143, 147 – 155, 157 – 158, 162 – 163, 165 – 168, 170 – 178, 180 – 182, 184 – 189, 192, 200 – 207, 218 – 221, 223, 225 – 231 Phonological Tone Sandhi (TS) 9 – 14, 16, 35, 40 – 43, 46 – 54, 56 – 63, 75 – 82, 85, 88 – 91, 93, 106 – 109, 111 – 113, 120 – 121, 123, 125 – 158, 161, 163, 165 – 178, 180 – 182, 184 – 187n7, 192 – 195, 199 – 202, 204 – 205, 216, 218 – 224, 226 – 227, 230 – 231

246 Index possessive/modificational/nominalization marker 98 – 99, 109, 113, 117, 121 post-verbal particle 100, 104 – 105, 111, 115, 119, 121, 169, 189 prosodic hierarchy 17 – 28, 36 – 38, 43, 45, 55, 57, 61 – 65, 89, 93, 96 – 97, 124, 133, 147, 200, 203, 206, 225 – 226, 228 – 229 prosodic phonology 1 – 3, 17 – 26, 28, 32 – 36, 38, 43, 45, 52, 55, 57 – 58, 61, 93, 121, 123, 133, 147, 153, 174, 177, 186, 201, 204 – 205, 224 – 225, 228 – 231 prosodic structure 17 – 20, 24 – 25, 30, 33, 131, 172, 176 – 177, 180, 202 prosodic word/phonological word 17 – 22, 24 – 30, 33 – 37n14, 55 – 67, 89 – 91, 93, 96 – 98, 103, 106 – 107, 113, 116, 120 – 124, 127, 133 – 134, 147 – 148, 153, 157 – 158, 171 – 176, 180, 184 – 185, 187n7 – 189, 192, 200 – 202, 204 – 205n2, 221, 223, 225 – 230 rate (of speech) 1, 35, 206, 209 – 210, 213, 218 – 224, 227 recursion/recursivity 24 – 25, 106; Nonrecursivity 18, 24 – 26, 31 – 32, 97, 103, 121 – 124, 191, 201 – 204, 228 – 229 reduplication 9, 12, 16, 72 – 74, 78, 80 – 82, 87, 89, 92n8, 98, 100, 109, 113 – 114, 118, 121, 126 – 127 Relation-Based Approach (RBA) 18, 20, 28 – 31, 33 – 35, 125, 128, 147 – 151, 154–155, 157, 167, 172, 176 – 177, 185 – 186, 229 – 231 restructuring 29, 35 – 36, 121, 124n8 – 125, 148 – 150, 155, 157 – 158, 168, 170, 181 – 182, 184 – 186, 200, 206, 209 – 217, 219 – 220, 222 – 224, 227, 231

sandhi tone 1, 5, 10 – 12, 14, 46, 49 – 51, 58, 76 – 78, 80 – 81, 107, 125, 142, 192 sentence final particle 103, 110, 115, 119 sound-splitting word 73 – 74, 78 – 82, 84 – 85, 87 – 90, 92, 227 stress 26 – 27, 29 – 30, 43 – 44, 46 – 47, 49 – 57, 59, 61, 66, 91n2, 94, 96 – 97, 151, 172, 211 Strict Layer Hypothesis (SLH) 17 – 18, 23 – 28, 31, 35, 37n14, 64, 93, 97, 121, 123 – 124, 148, 157, 185, 188, 201, 203 – 205, 207, 228 – 229, 231 style (of speech) 1, 35, 206, 209 – 210, 213, 224 syllable 1, 3, 5 – 9, 11 – 12, 15 – 17, 21 – 22, 24 – 26, 34, 36, 38 – 47, 49 – 51, 53 – 56, 58, 60, 62 – 63, 66 – 67, 73, 80, 82, 84 – 85, 87, 92, 99, 106, 108, 117, 120, 123, 128, 131 – 132, 141 – 142, 154, 171, 173 – 176, 185, 196, 199, 221 – 223, 225 – 227, 230 syllable duration 46 – 47, 49 – 51, 56 syllable structure 3, 9, 43, 48, 73, 227 syllable weight 47, 55 theta-role 133, 146 – 147, 163, 167, 169 – 171 tone bearing unit 34, 46, 53 – 54, 56, 225 tone group 52, 134 – 137, 143, 147, 222 utterance 17, 21 – 22, 25, 34, 55, 94, 107, 123, 191, 204, 209 – 210, 213, 224 – 226 well-formedness condition 40, 43, 62 word formation rule (WFR) 81 – 82