2 Edited by Joanna Ut-Seong Sio & Sze-Wing Tang Society Published in 2007 by the Linguistic Society of Hong Kong G.P.O. Box 9772, Hong Kong ISBN 962-7578-08-8 "Cantonese classifier di and genericity" ©Ben Wai-Hoo Au-Yeung "Headless relatives in Cantonese: a derivational account" © Yuan-Jian He "A preliminary study on Cantonese gwai 'ghost" © Peppina Po-Lun Lee & Andy Chi-On Chin "The indexical expressions gam2, gam] in Cantonese" ©Joanna Ut-Seong Sio & Sze-Wing Tang "The uses of DUI, SHUANG and FU in Cantonese, Mandarin and in the history of the Chinese language" © Carine Yuk-man Yiu Kang-Kwong Luke The views expressed in the articles contained in this volume are those of the individual authors, and do not represent those of the editors or the publisher. Printed in Hong Kong by Wah Cheong Printing Press Ltd. TABLE OF CONTENTS Preface v Joanna Ut-Seong Sio and Szc-Wing Tang 1. Cantonese classifier di and genericity Ben Wai-Hoo Au-Yeung 2. Headless relatives in Cantonese: a derivational account 17 Yuan-Jian He 3. A preliminary study on Cantonese gu;ai 'ghost 33 Peppina Po-Lun Lee & Andy Chi-On Chin 4. The indexical expressions ,gtml2, gam} in Cantonese 55 Joanna Ut-Seong Sio & Sze-Wing Tang 5. The uses of DUI, SJ-lU./11\JG and FU in Cantonese, 75 Mandarin and in the history of the Chinese language Carine Yuk-man Yiu 6. 95 ll?g&JB\: Kang-Kwong Luke Notes on contributors 109 Preface The first volume of StHdie.r in Ccmtone.re Lii(glli.rtic.r was published by the Linguistic Society of Hong Kong in 1998. After almost a decade, linguistic studies in Cantonese in Hong Kong have flourished and progressed in many directions. The contributors to this volume, St11die.r in Can!oiie.re Ling11i.rtic.r 2, are all local linguists. This volume acts on the one hand as a forum for local linguists to share their linguistic ideas, adding to the existing richness of Cantonese linguistics and on the other hand, it stands as a milestone, providing a glimpse of the linguistic development in Hong Kong at this very stage. The papers in this volume cover a wide range of topics, providing descriptive and theoretical analyses on various issues on Cantonese linguistics. Au-Yeung's paper 'Cantonese classifier di and genericity' discusses the fuzzy classifier di in Cantonese, which unlike other regular classifiers, can have a generic reading in certain environments. He's paper 'Headless relatives in Cantonese: a derivational account' provides an analysis that accounts for the presence and the absence of the complementizer t,e in Cantonese relative clauses, an account that makes use of IP /CP extraposition and PF deletion of ge. Lee & Chin's paper 'A preliminary study on Cantonese gwai 'ghost' provides a detailed description on the distribution of the versatile Cantonese infix both on a syntactic and a discourse level. Sio & Tang's paper 'The indexical expressions gam2, gam3 in Cantonese' discusses the indexical element gam in Cantonese, the reference of which is restricted to abstract entities like properties, degrees and propositions. Yiu's paper 'The uses of DUI, 51-lUANG and .FU in Cantonese, Mandarin and in the history of the Chinese language' examines the use of three collective classifiers DUI, SHUANG and FU in different dialects and also in different time periods. One prominent characteristic of Cantonese is its rich inventory of sentence-final particles. However, not all sentence-final particles have a unique corresponding character. Luke's paper puts forth a proposal which each sentence-final particle is paired-up with a unique character. \X' e would like to thank our contributors for their contributions and our reviewers for their critical commentaries, without which, this volume would not have materialized. \'1/e are also very grateful for the support which the Linguistic Society of Hong Kong's publication program has received from the Freemason's Fund for East Asian Studies. In the preface of the first volume of Studie.r iu Cantonese Lin,gHi.rtic.r, Dr. Stephen Matthews noted that Hong Kong students often had the preconception that Cantonese has no grammar. The Linguistic Society of Hong Kong has worked very hard to prove this to be a misconception. This volume is dedicated not only to those who are interested in Cantonese linguistics but also to those who are still in doubts. Joanna Ut-Seong Sio & Sze-Wing Tang The Hong Kong Polytechnic University BEN \XIAl-HOO 1\U-YEl!NG Cantonese classifier di and genericit:y Ben Wai-Hoo Au-Yeung [email protected] The Chinese University of Hong Kong 1. Introduction Chinese languages denote genericity by bare nouns, such as xiongmao 'panda' in (1) (Cheung 1972, Li & Thompson 1981, Chen1987, lVIatthews & Yip 1994 among others). As classifier languages, they do not use classifier-noun phrase construction as generic noun phrases (2) (Krifka 1995). The following examples are in lVIandarin. (1.) a. Wo xihuan xio11gmao. (object) I like panda 'I like pandas.' b. Xiongmao xihuan shuijiao. (subject) Panda like sleep 'Pandas like sleeping.' (2.) a. # Wo xihuan z!IOIZI', I xiOIZfl,!llC/0. (object) I like Cl-kind/ Cl-sg panda NOT: 'I like pandas.' BUT: 'I like a kind of/ a panda.' b. * Zhong / Zbi xiongmao xihuan shuijiao. (subject) Cl-kind/ Cl-sg panda like sleep 'A kind of/ A panda likes sleeping.' According to Kriflza et a!. (1995), genericity can be expressed by kind-referring noun phrases (or generic noun phrases) as shown by the examples of Cantonese in (3a-b), and by generalization-referring, characterizing sentences (or generic sentences) as in (3b-c). In (3a-b), .raigJPaa 'watermelon' is a generic noun phrase, interpreted as kind-referring in the non-characterizing sentence in (3a) and in the characterizing sentence in (3b). But in (3c), genericity is expressed by the characterizing nature of the sentence although the subject is in the form of an indefinite numeral noun phrase. (3d) is a non-characterizing sentence with a classifier-noun phrase as subject, which can only be interpreted as definite, and so the sentence has no generic interpretation. In this paper, only generic noun phrases as in (3a-b) are studied. Previous versions of this paper were presented in the reading group on Yue & other Chinese dialects initiated hy Prof. Lie-Jiong Xu and Prof. Thomas Lee, and at the Annual Research Forum of LSHK, Dec14-15, 1996. I would like to thank Cindy Chan, Lawrence Cheung, Yang Gu, Dong-Fan Hua, Peppina Lee, Patricia Man, Gladys Tang, Sze-Wing Tang, Cathy Wong, Leo Wong and Virginia Yip for discussions on the topic. Special thanks go to Prof. Thomas Lee and Prof. Yang Gu for inspiring me on the issue. Any errors in the paper are n1inc. 2 CANTONESE CLASSIFIER DI AND GENERICITY (generic noun phrase, non-characterizing sentence) (3.) a. Saigwaa gamziu jiging maai saai laa.1 Watermelon this-morning already sell all Sfp 'Watermelons were all sold out this morning.' (generic noun phrase, characterizing sentence) b. Saig1vaa dosou zung hai dei soengmin. Watermelon often grow on ground surface 'Watermelons often grow on the ground surface.' (indefinite noun phrase, characterizing sentence) c. ]at go saigu;aa gau sik saam go jan. One Cl-sg watermelon enough eat three Cl-sg person 'One watermelon is enough for three persons to eat.' (definite noun phrase, non-characterizing sentence) d. Di saigJI!aa sik saai laa. Cl-pl watermelon eat all Sfp 'The watermelons were all eaten up.' Against this background, it was first observed in Au-Yeung (1996) that Cantonese can use di-noun phrases to denote genericity where di the plural classifier with a fuzzy quantity. This paper will elaborate this generic reading more and try to explain it as a consequence of the definite use of classifiers with respect to a cluster of properties developed in Au-Yeung (1997 a-b), as well as some inherent property of the plural classifier. As issues on (in)-definiteness in Cantonese will be mentioned against genericity in later sections, let me briefly introduce how it is expressed in Mandarin for the sake of comparison. Consider: (4 ) a. Keretl lai le. Guest come Pfv 'The guest(s) has/ have come.' b. Lai le kere11. Come Pfv guest 'Some guest(s) has/ have come.' Bare nouns show definite-indefinite contrast by their relative positions in a sentence (Chao 1968). In (4a), the noun phrase keren is interpreted as definite in the subject position, which marks old information. In (4b), the same noun phrase receives indefinite interpretation in the object position, which indicates new information. As for the organization of the paper, the generic problem of di in Cantonese will be first Cantonese romanization is based on the Jyutping Scheme developed by the Linguistic Society of Hong Kong in 1992. Abbreviations in the gloss are arranged as: Aux-auxiliary, Cl-pl-plural classifier, Cl-sg-singular classifier, DE-genitive marker (Mandarin), Exp-experiential marker, Foe-focus, GE-genitive marker (Cantonese), Pfv-perfective marker, Sfp-sentence final particle. BEN W,\J-Hoo Au-YEUNG 3 approached by a discussion on the properties of di as a classifier in Section 2. Section 3 will illuminate how it functions as a generic device in a number of constructions, followed by an explanation in Section 4. Section S will conclude the paper. 2. Di as a classifier 2.1. Definite and indefinite use Before discussing how the Cantonese classifier di denotes genericity, I give an overview of the properties of di that are shared \vith and parallel to other classifiers. Like other classifier-noun phrases, di-noun phrases can have definite interpretation as subject and object (Sa-b), and indefinite interpretation as object in (Se-d) (Cheung 1989, Matthews & Yip 1994). In addition, di can be used in the possessor-classifier-noun phrase, as in (6a) and in the relative clause-classifier-noun phrase, as in (6b). (definite CL-N phrase as subject) (S) a. Go / di .ra(i!,waa sik saai laa. Cl-sg/ Cl-pl watermelon eat all Sfp 'The watermelon/s was/were all eaten.' (definite CL-N phrase as object) b. Ngo sik saai di / go .ra{i!,JJ!aa. I eat all Cl-pl/ C:l-sg watermelon 'I ate all the watermelons/ the watermelon.' (non-specific/ existential indefinite C:L-N phrase as object) c. Toi soengmin fong zo go / di .raig}/J{lri. Table surface stay Pfv Cl-sg/Cl-pl watermelon 'There is/ are one/ some watermelons on the table.' (specific/ partitive indefinite CL-N phrase as object) d. Zoeng toi jau houdo saigwaa, ngo lo zo di / J!,O sik. Cl-sg table has many watermelon I take Pfv Cl-pl /Cl- sg watermelon eat 'There are many watermelons on the table. I've taken some/ a watermelon to eat.' (6) a. Ngo go / di m gin zo. I Cl-sg/Cl-pl watermelon not appear Pfv 'My watermelon/ s disappeared.' b. Lukjap lei /!,0 / di .raigJJ!aa hai ngo ge. Roll in come Cl-sg/Cl-pl watermelon be I GE 'The watermelon/ s that rolled inside is/ are mine.' In Au-Yeung (1997 a-b), the possessive construction and the relative clause construction in (6) were argued to be correlated with the definite use of classifiers in (Sa-b). Notice that the plural classifier di also possessess this series of constructions. 2.2. Non-collectivity 4 CANTONESE CLASSIFIER DI AND GENERICITY Besides the fuzzy plurality, di possesses a non-collective property that prohibits its countability. This property unspecifies how the referents denoted by the noun phrase group together and the grouping does not provide a shape or a unit for counting. In (7a), numerals larger than one cannot precede di. The referents denoted by di seem to metaphorically expell each other so that they do not form a structure for other classifiers to describe. In other words, the classifier represents its collectivity to such a minimal degree that this set of loosely-packed referents can only be counted with the number limited to one.2 (7) a. Ngo maai zo jat / * saa111 di Jvtm. buy Pfv one / three Cl-pl bowl 'I have bought one/ three unit(s) of bowls.' b. Ngo maai zo )at/s aa!ll deoiJ wtm. (collective, non-fuzzy plural) I buy Pfv one/three Cl-pair bowl 'I have bought one/three pair(s) of bowls.' c. Ngo maai zo )at/ saam deoi1 JVtltl. (collective, fuzzy plural) I buy Pfv one/ three Cl-pile bowl 'I have bought one/three pile(s) of bowls.' d. Ngo maai zo )at / saa!ll '(!lllg Jvtm. (collective, fuzzy plural) I buy Pfv one/three Cl-kind bowl 'I have bought one/three kind(s) of bowls.' However, other plural classifiers with high collectivity behave differendy. In (7b-d), collective plural classifiers, deoiJ 'pair' (with a non-fuzzy quantity), deoi1 'pile' (with a fuzzy quantity), and 'lfi!Jg 'kind' (with a fuzzy quantity) 'collect' the referents together to form a certain shape or unit for counting. So numerals larger than one can precede these classifiers. In Chierchia's (1998, 380) terms, similar collective nouns in English, such as group and bunch, are called 'singular individuals, not pluralities' - because there are groups and bunches. In this connection, the classifiers deoiJ, deoi1, and '(!ttJg can also be regarded as 'singular individuals', inclividuating the group they form as a counting unit. Thus they can follow any numeral in the nominal structure. Table 1 summarizes plural classifiers of Cantonese in terms of collectivity and quantity. Collective, non-fuzzy classifiers are countable with a non-fuzzy quantity while the fuzzy di is uncountable with a fuzzy quantity. In between these two types are collective, fuzzy classifiers, which are fuzzy in quantity and allow for counting. Table 1. Classification of plural classifiers Plural Cl Examples Countability of Cl Collective, deoiJ (a pair of), One, two, three ... non-fuzzy quantity daa (a dozen of) 2 A reviewer suggested that the numeral one preceding di might be .more like an indefinite article. I think whatever category it is called, it still carries some numerical sense that marks a small quantity of the referents. BEN WAI-Hoo 1\U-YEUNG 5 2.3. Reduplication Classifier reduplication is found to be able to denote universal quantification in Mandarin (Li & Thompson 1981), and is especially productive in Cantonese (Cheung 1972, Matthews & Yip 1994). However, collective and non-collective classifiers in reduplication are complementarily disttibuted in affirmative and negative sentences. In (Sa), all classifier-classifier phrases are grammatical with the universal quantifier do11 in affirmative sentences, except dirli. But in (Sb), the negative sentence reflects the reverse. Only dirli is allowed when the scope of the quantifier do11 is an empty set.3 (8) a. Gogo / daadaa / deoideoi / '?flllgZjmg / 1'dirli pinggwo CI-CI-sg/CI-Cl-dozen /Cl-Cl-pile /Cl-Cl-kind I ClCI-pl apple dou hou mei. all very delicious 'All units/dozens/piles/kinds/few of apples are very delicious.' b. *Gogo / *daadaa / *deoideoi / dirli pinggwo Cl-Cl-sg/ CI-Cl-dozen/ Cl-Cl-pile I Cl-Cl-kind I CI-Cl-pl apple ngo dou mou. I all no 'I do not have all units/all dozens/ all piles/ all kinds/ a bit of any apple(s).' With the property that di-noun phrases cannot be preceded by numerals larger than one, di shares other classifier properties more than it does not. Therefore, while the classifier status of di is assumed in the rest of the paper, its generic use is hoped to be explained in terms of the above properties. 3. Generic di Apart from denoting (in)definiteness as introduced in Section 2.1, rli is unicJUe in expressing genericity in di-noun phrases. Let's examine how the generic di functions in different argument positions, /Ph-questions, negation constructions, and modified nominals. 3.1. Argument positions As reported in the literature, classifiers cannot associate with genericity and so the singular classifier go in (9a-b) is not interpreted as generic, but as definite only. However when r/i is used, d;~noun phrases in the subject position can be interpreted as generic and refer to the kind of the fruit JazgJPaa 'watermelon' although the definite reading of the di-nominals may also be available. The brackets of r/i indicate that the classifier is optional for the noun phrase .wi,gJJJaa to mark genericit:y. I am not sure whether this complementariness is due to the non-collective property of di. I leave it open here.