漢語光桿分類詞組之分析 - 政大學術集成

全文

(1)國立政治大學語言學研究所碩士論文 National Chengchi University Graduate Institute of Linguistics Master Thesis. 指導教授：何萬順博士 Advisor：Dr. One-Soon Her. 立. 政治大. ‧ 國. 學. 漢語光桿分類詞組之分析. On the Bare Classifier Phrase in Mandarin Chinese. ‧. n. er. io. sit. y. Nat. al. Ch. engchi. i n U. 研究生：陳景芃撰 Student: Ching-Perng Chen 中華民國一○二年一月 January, 2013. v.

(2) ON THE BARE CLASSIFIERS PHRASES IN MANDARIN CHINESE. BY. Ching-Perng Chen. 立. 政治大. ‧ 國. 學 ‧. A Thesis Submitted to the Graduate Institute of Linguistics in Partial Fulfillment of the Requirements for the Degree of Master of Arts. n. er. io. sit. y. Nat. al. Ch. engchi. January, 2013. i n U. v.

(3) The members ofthe Committee approve the thesis of Ching-Perng Chen. defended on October 22, 2012.. 二卜. One-Soon Her. Advisor. 立. 政治大. "。" "仁二 ‧. ‧ 國. 學. Husn-Huei Chang. Committee Member. T山匕. n. er. io. sit. y. Nat. al. Ch. engchi. i n U. v. Committee Member. Kawai Chu, Director, Graduate Institute of Linguistics.

(4) 立. 政治大. ‧. ‧ 國. 學. n. er. io. sit. y. Nat. al. Ch. engchi. Copyright © 2013 Ching-Perng Chen All Rights Reserved III. i n U. v.

(5) ACKNOWLEDGEMENTS. 能夠完成這份研究以及碩士學位論文，最感謝的當然是我的指導老師何萬順老師。雖然我到最後都還是無法追上老師腦中飛快的推理邏輯，卻從何老師嚴謹的科學態度學到對研究的執著。在此之前，我從來不敢相信對語言學沒什麼天份的自己可以成為班上第一個寫完論文的人。何老師不僅在學術上不厭其煩地給予指導，同時也讓我從助理工作裡學到許多，更謝謝熱情的何老師多次邀我們一群學生到家裡聚會，與我們分享生活碰到的大小事情。. 立. 政治大. ‧. ‧ 國. 學. 在政大語言所的兩年半，真的可以說是豐收的時光。我喜歡政大語言所多元的課程設計以及自由的學風，讓我在語言學這個領域收穫不少。也喜歡語言所的所有任課師長們，傾囊相授之餘也不忘時時給學生們關懷。在此感謝所有教過我的師長們，史佳琳老師、林祐瑜老師、莫建清老師、張郇慧老師、賴惠玲老師，更不用說所上的專任老師們，私底下笑容很多的語言所所長徐老師，總是像仙女一樣舉止優雅的黃老師，大眼眨呀眨、美麗又活潑的萬老師，熱情也愛熱鬧搞笑的蕭老師，以及上課幽默、魅力四射、立領為註冊商標的何老師，不管是課堂上或課後總給我好多好多的鼓勵與支持。另外，要特別感謝助教學姐，很多時候需要麻煩她幫我在大小文件上核章，在所辦工讀的兩年半時時間，助教學姊也教了我很多行政工作上要注意的大小事。. n. er. io. sit. y. Nat. al. Ch. engchi. i n U. v. 論文的完成，除了指導教授外，感謝兩位參與的口試委員們，張郇慧老師和謝富在老師。感謝你們不吝嗇給我許多專業的指導和建議，在我露出疑惑神情時給個溫暖的微笑，不厭其煩地再講解一次。也感謝你們願意親臨政大參與我論文計畫書口試以及論文口試，學生我十分感激。在此，我還要感謝研究生生活認識的學長姐和學弟妹，不管是所上或是外系的大家，或是和我相處最多時間的同屆同學們，和你們一起上課到包種茶節、期末聯歡、大大小小的出遊以及畢業旅行，這些全部都是我非常美好的研究生回憶。感謝吹吹和維維不時的關心和加油打氣。感謝原本是電腦工讀生後來卻參與我們好多活動的小明，在某次論文快寫完電腦卻當機時拯救了我的筆電。感謝 Tim&Linda 夫婦以及藍爸爸，擴大了我小小的生活圈，每一次和你們聊天都笑到一個不行。感謝後來和我一起搬進大公寓的 DH 花，所有開心難過的時候都可以 IV.

(6) 跟你們一起分享，不管是大哭大笑或是大鬧~ 感謝姊妹淘涵吉對生活上大小事情的協助，在我碰到難過事情或苦惱時聽我講心事。感謝瘦(ㄆㄤˋ)亨積極參與我們大小活動、分享他的特別辭彙，以及在論文的音韻分析上指導我。感謝大衛，研究生生活有了你才有意義（噁~~~），感謝你參與了我研究所生活的大部份，在論文口試前夕耐心聽我多次練習。感謝在台北工作念書的中山外文 98 朋友們，阿莓、Momo、魚丸、悅綾、小 P…，每一次和你們聚會總可以好好放鬆，分享除了語言學外的生活趣事。也感謝在高雄錦忠老師、阿白和梨盎幫我加油打氣。然後，我想跟還在台大外文所奮鬥的大貓說，好好加油！學術的路不好走，但撐過就是你的~離畢業之路不遠了。另外，謝謝我在安親班碰到的陳老師和小朋友們，對我來說，家教這份工作除了賺錢之外，更是我找到自己長處與成就感的地方。謝謝你們溫暖的笑容以及每一次的出遊邀約，在你們身上我也學到了好多。. 立. 政治大. ‧ 國. 學. ‧. 最後，也是最重要的，感謝支持我選擇以及完成這個學位的家人們。感謝爸媽給了好多溫暖，在碰到瓶頸想要放棄的時候鼓勵我堅持下去，也感謝你們經濟上給我部分援助，讓我不致陷在無限的打工循環裡。感謝妹在臉書上時不時跟我分享生活大小事情。感謝在台北對我照顧無微不至，幾乎每週末都讓我去打擾的二姨和四姨。感謝四伯父伯母、大姑姑、表姊和外甥女采采，在我無法回高雄的大節日找我去你們天母的家叨擾，共享天倫之樂。感謝現在在天堂的愛犬波卡，好多時候你是讓我想回高雄的動力，相信你現在過得很好。. n. er. io. sit. y. Nat. al. Ch. engchi. i n U. v. 即將完成論文之際，心中可說是百感交集。開心快要可以拿到一個學位，卻又對無憂無慮的學生生活充滿不捨。畢業之後馬上要踏入人生的新階段了，對未知的未來既興奮又帶了點不安的感覺。但在此之前，容我對這一路上碰到的大家說聲：真的好感謝你們！！！. 景芃於論文完成前夕 V.

(7) VITA. EDUCATION. M.A. in Linguistics, December 2012 National Chengchi University B.A. in English, June. 2010 National Sun Yat-sen University. 政治大. 學. ‧ 國. 立 GRANTS and SCHOLARSGIPS. ‧. Oct. 2012. sit. er. al. Research Assistant of NSC Research Project: Numerals and Classifiers: A Typological Study of Word Orders and Structures in the World's Languages. n. Sep. 2011June 2012. io. Aug. 2012Jan. 2013. y. Nat. National Science Council Travel Grant for attending 45th International Conference on Sino-Tibetan Languages and Linguistics. Project No. NSC-101-2922-I-004-065. Ch. engchi. i n U. v. Teaching Assistant of Language and World Civilization National Chengchi University. Aug. 2011July 2012. Research Assistant of NSC Research Project: Classifiers in Taiwan Mandarin: Bibliography, Corpus, and LFG Analysis. Sep. 2010June 2011. Syntax Laboratory Assistantship National Chengchi University. VI.

(8) PUBLICATION Chen, Ching-Perng. 2012. On the Bare Classifier Phrases in Mandarin Chinese. Paper presented in 45th International Conference on Sino-Tibetan Languages and Linguistics. Nanyang Technological University. Singapore.. 立. 政治大. ‧. ‧ 國. 學. n. er. io. sit. y. Nat. al. Ch. engchi. VII. i n U. v.

(9) 國. 立. 政. 治. 大. 學. 研. 究. 所. 碩. 士. 論. 文. 題. 要. 研究所別：語言學研究所論文名稱：On the Bare Classifier Phrase in Mandarin Chinese 漢語光桿分類詞組之分析指導教授：何萬順博士研究生：陳景芃. 立. 政治大. 論文內容提要：(共一冊，21,967 字，分五章節). ‧ 國. 學 ‧. 本篇論文分析光桿分類詞組[CL+N]在漢語的分佈位置。過去文獻中，學者. Nat. io. sit. y. 們認為光桿分類詞組[CL+N]為不定數量詞組[yi+CL+N]的音韻弱化（Lu 1990,. er. Chen 2004），且因為出現在主詞或主題的光桿分類詞組數量詞組[yi+CL+N]並不. al. n. v i n Ch 是不定語意，不定數量詞組音韻弱化的光桿分類詞組[CL+N]不得出現在這兩個 engchi U 位置 (Cheng & Sybesma 1999/2005, Li 1998, Huang, Li & Li 2009)。在此前提下，可以得到兩個假設，第一，所有的不定數量詞組[yi+CL+N]都能弱化為光桿分類詞組[CL+N]。第二，所有的光桿分類詞組[CL+N]都只能在不定語意的位置出現。然而，在漢語裡，即使指涉數量的數量詞組[yi+CL+N]還是能弱化為光桿分類詞組[CL+N]，且並非所有的光桿分類詞組[CL+N]都出現在不定語意的位置。換句話說，過去文獻只能解釋而非預測光桿分類詞組[CL+N]在漢語中的分佈位置。 VIII.

(10) 進一步檢視漢語裡光桿分類詞組[CL+N]的分佈位置，我認為指涉數量的數量詞組[yi+CL+N] 除了在比較語境，大多時候也能弱化為光桿分類詞組[CL+N]。另外，由於漢語裡的分類詞有許多附著語素的特徵，我認為應該將其分析為附著語素較為適當。從構詞的角度來看，漢語裡的分類詞需附著於前方的成分，不能. 政治大. 單獨出現。句法上來說，漢語分類詞可以自由黏著在某幾詞類的宿主。音韻上，. 立. 漢語的分類詞以最大音韻重量限制所選宿主。綜合以上幾點，我主張，限定或非. ‧ 國. 學. 限定的句法位置並非決定漢語光桿分類詞組[CL+N]出現的主要原因；決定漢語. ‧. 光桿分類詞組[CL+N]的出現因素應是分類詞是否能找到適當的宿主附著。有關. Nat. n. al. er. io. sit. y. 合法宿主的制約條件，在我的論文裡也有詳細的討論。. Ch. engchi. IX. i n U. v.

(11) Abstract. This thesis investigates the distribution of the bare classifier phrase [CL+N] in Mandarin Chinese. Previous studies argue that [CL+N] is the phonological reduction of the indefinite numeral classifier phrase [yi+CL+N] (Lu 1990, Chen 2004), and it cannot occur in subject or topic position because [yi+CL+N] in these positions are never considered indefinite (Cheng & Sybesma 1999/2005, Li 1998, Huang, Li & Li 2009). Two predictions are made from these analyses. First, all indefinite [yi+CL+N] can be reduced to [CL+N], and second, [CL+N] only occurs in positions which are considered indefinite. However, it is easily found in Mandarin Chinese that numeral classifier phrase denoting quantity can also be reduced to bare classifier phrase [CL+N], and that not all [CL+N] occur in indefinite positions. In other words, previous studies simply explain the occurrence of [CL+N], but fail to predict the occurrence of [CL+N]. With a close examination, I suggest that numeral classifier phrase [yi+CL+N] denoting quantity can also be reduced to bare classifier phrase [CL+N] except occurring in contrastive contexts. Also, I adopt the clitic account on classifiers in Mandarin since they share many properties with clitics. Morphologically, classifiers in. 立. 政治大. ‧. ‧ 國. 學. n. er. io. sit. y. Nat. al. Ch. engchi. i n U. v. Mandarin never appear in isolation but attach to preceding elements. Syntactically, they are free to cliticize to hosts belonging to different categories but are not subject to syntactic roles. Phonologically, classifiers in Mandarin subcategorize their hosts within a maximal number of moras. Therefore, I argue that it is not (in)definiteness that decides the occurrence of the bare classifier [CL+N] but whether it can find a legal host to cliticize. Constraints concerning to the legal hosts will also be discuss in detail in my thesis.. X.

(12) List of Abbreviations:. The following abbreviations are used in the glosses. ASP = aspect marker. 政治大. BA = ba-construction CL = classifier. 立. DE = Structural particle in Mandarin. ‧ 國. 學. EXP = Experiential aspect marker. V = verb. y. sit. al. n. PAR= particle. io. P = preposition. er. Num = numeral. Nat. N = noun. ‧. M = measure word. Ch. engchi. XI. i n U. v.

(13) TABLE OF CONTENTS. Acknowledgement ................................................................................................................ IV VITA. ...................................................................................................................................... VI. Chinese Abstract. ................................................................................................... VIII English Abstract. ......................................................................................................... X. 政治大. List of Abbreviations..................................................................................................XI. 立. Table of Contents. .................................................................................................... XII. ‧ 國. 學. CHAPTER 1. INTRODUCTION............................................................................................ 1. ‧. 1.1 Motivation and Purpose........................................................................................................ 2. sit. y. Nat. 1.2 Conventions of the data ...................................................................................................... 5. io. n. al. er. 1.3 Organization of the thesis ................................................................................................... 7. Ch. i n U. v. CHAPTER 2. LITERATURE REVIEW................................................................................ 9 2.1 Definiteness and Specificity. engchi. ......................................................................................... 11. 2.1.1 Lyons (1999)............................................................................................................ 11 2.1.2 Frawley (1992) ........................................................................................................ 12 2.2 The interpretation and the distribution of the numeral classifier phrase [Num+CL+N] .... 14 2.2.1 Li (1998) and Huang, Li and Li (2009) 2.2.2 Cheng and Sybesma (1999). ............................................................... 16. ................................................................................. 21. 2.3 The interpretation and the distribution of the bare classifier phrase [CL+N]..................... 22 2.3.1 Lu (1990) and Chen (2004) ................................................................................... 22 2.3.2Cheng and Sybesma (1999) ...................................................................................... 24 2.3.3 Yip (2008) ............................................................................................................. 28 2.4 Other analyses on the distribution of [Num+CL+N] and [CL+N] ..................................... 31 XII.

(14) 2.4.1 Ji (2007)................................................................................................................... 31 2.4.2 Yang (2002)............................................................................................................. 34 2.5 Remarks .............................................................................................................................. 36. CHPATER 3. Unsolved Problems for [yi+CL+N] and [CL+N] in Mandarin Chinese ..... 39 3.1 Assumptions from previous studies.................................................................................... 39 3.1.1 [CL+N] only occurs in indefinite positions? ........................................................... 40. 政治大. 3.1.2 All [CL+N] are allowed in indefinite positions? ..................................................... 42. 立. 3.1.3 Other special contexts which block the occurrence of [CL+N]............................... 47. ‧ 國. 學. 3.2 Distributions of [CL+N]: an overview ............................................................................... 51 3.2.1 Positions allowing the occurrence of [CL+N] ......................................................... 51 3.2.2 Positions disallowing the occurrence of [CL+N] .................................................... 54. ‧. 3.3 Problems for adopting previous accounts ......................................................................... 55. Nat. sit. y. 3.3.1 Properties that deny previous accounts .................................................................. 55. io. er. 3.3.2 The selections on the phrase preceding [CL+N] ..................................................... 56 3.3.3 Classifiers cannot be too far away from the preceding verb .................................. 57. n. al. Ch. i n U. v. 3.3.4 Phonological restrictions on the preceding element and [CL+N] itself ............ 58. engchi. 3.3.5 Problems for adopting any syntactic approach alone ........................................... 59. CHAPTER 4. A Clitic Account ............................................................................................. 61 4.1 Classifiers in Mandarin as clitics? ...................................................................................... 61 4.1.1 Definitions and Properties of clitics .......................................................................... 62 4.1.2 Different types of clitics ............................................................................................ 64 4.2 A clitic account for classifiers in Mandarin Chinese .......................................................... 66 4.2.1 Properties shared by clitics and classifiers in Mandarin Chinese .............................. 67 4.2.2 Subcategorizations of classifiers on their preceding elements .................................. 69 4.3 Processes and principles involved in the clitic analysis of classifiers in Mandarin ........... 70 4.3.1 Classifiers in [CL+N] should be lexically governed by legal governors ................... 70 XIII.

(15) 4.3.2 The adjacency of classifiers in [CL+N] and the governors ....................................... 73 4.3.3 The unsolved problems in ditransitive clauses .......................................................... 75 4.3.4 The phonological constraints on the hosts and the classifier ..................................... 77 4.3.5 [CL+N] in distransitive structures ........................................................................ 81. 4.3.6 Conclusion of this section....................................................................................... 84 4.4 The remained unsolvable .................................................................................................. 85. 治政 4.4.2 The prohibition of disyllabic classifiers in [CL+N] structure 大 ...................................88 4.5 The interpretation of ‘one’立 in the bare classifier phrase [CL+N] ..................................... 92 4.4.1 The prohibition of [CL+N] in contrastive contexts ................................................... 86. ‧ 國. 學. 4.5.1 The redundant 1 value of Chinese classifiers .............................................................. 93 4.5.2 The redundant 1 value of numerals ........................................................................... 96. ‧. 4.6 Conclusion of the chapter ................................................................................................. 98. y. Nat. io. sit. CHAPTER 5. CONCLUSION ............................................................................................ 100. n. al. er. REFERENCES ..................................................................................................................... 104. Ch. engchi. XIV. i n U. v.

(16) Chapter 1 Introduction. Categorization is considered one of the most prominent aspects of human. 政治大. cognition. Jakendoff (1983) has stated that to judge whether a particular thing belongs. 立. to a particular category is an essential aspect of cognition. The importance of. ‧ 國. 學. categorization is also given by Lakoff (1987) that the understanding of how people. ‧. categorize is prior to any understanding of how people think and how people function.. Nat. io. sit. y. Studies have also proved that the way an entity is categorized depend on how it is. n. al. Language encodes. er. conventionally perceived (Adams, 1986; Tai and Chao, 1994).. v i n Ch representations i U e n g cofhcategorizing. sets and individuals. In. languages like English and other Indo-European languages, words are used directly with numerals or with various determiners. On the other hand, many languages, for example, Chinese, Japanese, Malay and Thai, use classifiers to signal the references. Being one of the classifier languages, the classifier system in Chinese remains a hot linguistics topic since Chinese is extremely rich in the use of classifiers. In recent decades, research related to classifiers is predominantly on cognitive approaches or 1.

(17) semantic approaches. While cognitive approaches and semantic approaches compose the majority of the data, analyses on syntactic aspect are still investigated by many syntacticians. Different syntactic analyses on classifier in Mandarin Chinese have been proposed. The major three issues are devoted to defining classifiers, finding out whether classifier form a constituent with Num or N first, and interactions between. 政治大. numeral classifier phrases [Num+CL+N] and bare classifier phrases [CL+N] (e.g., Li. 立. and Thompson 1981, Huang 1982, Tang 1990, Li 1998, Cheng and Sybesma 1999,. ‧ 國. 學. 2005, Chen 2004, Huang 1982, Li and Li 2009, Her 2010). My thesis concerns more. ‧. to the interaction between numeral classifier phrases [Num+CL+N] and bare classifier. n. al. er. io. sit. y. Nat. phrases [CL+N].. 1.1 Motivation and Purpose. Ch. engchi. i n U. v. Numeral classifier phrases [Num+CL+N] in Mandarin Chinese, according to most studies, have generally been regarded as indefinite expressions, and they are not allowed to occur in subject or topic positions because these positions are considered definite (e.g., Chao 1968, Li and Thompson 1981, Li 1996, Tsai 1994a, 1996). But some hold the different opinion that the prohibition of [Num+CL+N] in some positions is due to its quantity-denoting expression (Li 1998, Huang, Li and Li 2009). 2.

(18) While many studies discuss the distributions of [Num+CL+N], some studies also discuss a related structure, the bare classifier phrase [CL+N] (Chen 2003, 2004, Cheng and Sybesma 1999, 2005, Yang 2002, Her 2010). Since the two structures have a lot of overlapping distributions (for example, postverbal positions), some works claim that [yi+CL+N] and [CL+N] belong to distinct structures because the only interpretation of [CL+N] is indefinite and nonspecific (Cheng and Sybesma 1999); yet. 政治大. more studies suggest that bare classifier phrases [CL+N] are the phonological. 立. reduction of numeral classifier phrases [Num+CL+N] when the numeral is yi ‘one’. ‧ 國. 學. (Lu 1990, Chen 2003, 2004, Her 2010).. ‧. Discrepancies are also discussed concerning the distribution of bare classifier. Nat. io. sit. y. phrases [CL+N]. Compared with [Num+CL+N], the occurrences of [CL+N] are more. n. al. er. limited. Various explanations are provided for the prohibition of [CL+N] in certain. Ch. engchi. contexts. Cheng and Sybesma (1999), who. iv n hold U the opinion. that [CL+N] and. [yi+CL+N] being two different structures, state that the prohibition of [CL+N] is concerning boundedness and indefiniteness. Lu (1990) and Chen (2004) argue that only when [yi+CL+N] occur in indefinite position, being specific or nonspecific, can [yi+CL+N] be reduced to [CL+N]. Since it has been proved that [CL+N] do occur in indefinite and nonspecific positions, in this thesis, I adopt Lu (1990) and Chen’s (2004) account that the bare classifier phrase [CL+N] is the phonological suppressed 3.

(19) form of [yi+CL+N]. Two assumptions are made under Lu (1990) and Chen’s (2004) account. First, classifier phrases [CL+N] never occur in definite positions such as subject and topic positions. Second, all indefinite [yi+CL+N] can be reduced to [CL+N]. So far the first assumption has proven to be correct, but counterexamples for the second assumption are easily found in Mandarin Chinese. Yi ‘one’ in [yi+CL+N] cannot be omitted in. 政治大. some indefinite positions. Also, some [yi+CL+N] which are not interpreted as. 立. indefinite also undergo yi omission and become [CL+N]. All these facts indicate that. ‧ 國. 學. the previous accounts simply explain the occurrences of [CL+N] in some contexts yet. ‧. fail to “predict” the occurrences of [CL+N].. Nat. io. sit. y. The above counterexamples bring up the question: are there any other reasons. er. that might decide the occurrences of [CL+N] instead of indefiniteness? Yang (2002). al. n. v i n Cinha chapter of his dissertation suggests a rather different viewpoint that the classifiers engchi U. in Mandarin Chinese are actually clitics. He posits a morpho-syntactic structure for a Mandarin Chinese full NP with a classifier, where the combination of the numeral and the classifier is assumed to be a morphological complex rather than two independent words. In Yang’s (2002) study, he uses Zwicky and Pullum’s (1983) six criteria of distinguishing affixes and clitics to examine the classifier in Mandarin Chinese by comparing to clitics in English. In addition to Yang’s (2002) study, I have also 4.

(20) observed that classifiers in Mandarin share many properties with clitics syntactically, phonologically, and morphologically, and these properties seem to provide a better solution for the prohibition of [CL+N] in certain contexts. Therefore, in my thesis, first I am going to argue that the bare classifier phrase [CL+N], as a phonological reduction of [yi+CL+N], not only occurs in indefinite positions but also in quantity-denoting positions. Second, I will adopt a clitic account. 政治大. for classifiers in [CL+N] in Mandarin Chinese and argue that the prohibition of. 立. [CL+N] is due to the lack of a legal host but not indefiniteness.. ‧. ‧ 國. 學. 1.2 Conventions of the Data. Nat. io. sit. y. The data presented in my thesis is based on Taiwan Mandarin, a dialect of. er. Mandarin Chinese. The reason that I narrow the data down to Taiwan Mandarin is to. al. n. v i n Cissue. minimize the grammaticality h e Examples i U are from both on-line corpus, n g c h provided. the Academia Sinica Balanced Corpus of Mandarin Chinese, google searches and the utterances of Taiwan Mandarin native speakers. Note that the bare classifier structure [CL+N] discussed in this thesis does not include the verbal classifier such as he ge shui ‘to drink water’ or guang ge jie ‘to go shopping’. There are several differences between the verbal usage of classifiers and the nominal usages of classifiers. First, the nominal ge is necessarily inserted between 5.

(21) a quantifier and certain classes of count nouns, while the absence of the verbal ge does not cause ungrammaticality. Second, unlike the nominal ge, the verbal ge usually does not follow numerals and demonstratives. Third, the verbal ge can occur before noun phrases that are typically non-classifiable. For example, in he ge shui ‘to drink water’, ge is not the classifier for water. The three reasons above show that the verbal classifier construction [VP+ge+N] differs from the construction given in this thesis.. 政治大. In addition, the “classifier phrase” used in this thesis includes both classifiers and. 立. measure words defined by Her (2010). In other words, the classifier phrases in my. ‧ 國. 學. thesis have a broad meaning. The reason for counting measure words such as jin. ‧. ‘Taiwanese kilogram’, bao ‘pack’ into the classifier phrase category is that Chinese. Nat. io. sit. y. classifiers and measure words behave similarly in syntax yet differ in semantics and. er. mathematical value (Her 2010). The major two semantic differences lie in adjectival. al. n. v i n C(Her modification and mathematical value h e2010). h i U since adjectival modifiers n g cHowever, are not allowed in bare classifier phrases in Mandarin, semantic difference will not be the focus I am going to discuss in the following section. Therefore, in my thesis, both true classifiers and measure words in Mandarin Chinese will count as “classifier phrases” used in my examples.. 6.

(22) 1.3 Organization of the Thesis After the motivation and the purposes of my thesis are introduced, chapter two will recapitulate some main points of explanations about the occurrences of two classifier constructions, [Num+CL+N] and [CL+N], in previous studies. Later, in chapter three, distributions of [yi+CL+N] and [CL+N] will be presented, and some counterexamples against traditional analyses will be given to suggest modifications on. 政治大. previous studies. In chapter four, properties of clitics are first introduced, then a. 立. detailed analysis on supporting classifiers in Mandarin being clitics will be proposed.. ‧ 國. 學. Finally, chapter V concludes the study by summarizing the main points of the thesis. ‧. and pointing out the implications for future study.. n. er. io. sit. y. Nat. al. Ch. engchi. 7. i n U. v.

(23) 立. 政治大. ‧. ‧ 國. 學. n. er. io. sit. y. Nat. al. Ch. engchi. 8. i n U. v.

(24) Chapter 2 Literature Review. It is well known that Chinese languages allow both nominals appearing with. 政治大. classifiers and nominals appearing without classifiers to occur as arguments. It is. 立. believed in earlier studies that in Mandarin Chinese a classifier must cooccur with an. ‧ 國. 學. overt numeral or a demonstrative (see, e.g., Li and Thompson 1981:104, Tang 1990).. ‧. But in recent years, contrary to what is generally assumed, this statement is proven to. Nat. io. sit. y. be incorrect. The following example shows that in Mandarin Chinese, a classifier is. n. al. er. allowed to appear without the occurrence of a numeral or a demonstrative.. (1) 我買了. 本. 書. Ch. engchi. i n U. v. wo mai le ben shu I buy ASP CL book ‘I bought a book.’. A classifier appearing with a numeral is referred to as a numeral classifier phrase [Num+CL+N], and the one without a numerals or a demonstrative is referred to as a bare classifier phrase [CL+N]. Since both structures have many overlapping distributions, bare classifier phrases [CL+N] are generally considered to be the 9.

(25) omitted form of [Num+CL+N] when the numeral is 1. While most studies agree that numeral classifier phrases [Num+CL+N] have at least two interpretations (Tsai 1994a, 1996 and Li 1996, Huang, Li and Li 2009), issues on the interpretation and the distribution of bare classifier phrases [CL+N] have long been controversial because the occurrences of [CL+N] are more limited. Some claim that bare classifier phrases [CL+N] are always interpreted as nonspecific. 政治大. indefinite (Cheng and Sybesma 1999), others state that [CL+N] can occur in. 立. indefinite specific positions (Lu 1990, Chen 2004). Explanations for the occurrences. ‧ 國. 學. of [CL+N] also vary among proposed studies, yet a consensus has been reached that. ‧. [CL+N] never occurs in subject and topic position (Cheng and Sybesma 1999, Chen. io. sit. y. Nat. 2004, Yang 2002).. n. al. er. In this chapter, several issues will be concerned. First, in 2.1, related to the. Ch. engchi. interpretation of [Num+CL+N] and [CL+N], the. i n concept U. v. of definiteness and. specificity will be introduced. Then in 2.2, I will review some studies concerning the interpretation and the distributions of the numeral classifier phrases [Num+CL+N]. Later in 2.3, some studies with respect to the issue on whether [CL+N] being a suppressed form of [yi+CL+N] or a totally different structure from [yi+CL+N] will be captured. Also, I will deal with some works on discussing the interpretation of the bare classifier phrase [CL+N] and its interactions with the general constraint in 10.

(26) Mandarin Chinese that disallows the occurrence of [CL+N] in subject or topic positions. 2.4 will introduce two different viewpoints proposed by Ji (2007) and Yang (2002). Ji (2007) argues against Cheng and Sybesma (1999) that the prohibition of [CL+N] in certain positions is due to phonological factor but not definiteness. Yang (2002), rather than treating classifiers as phrases, thinks that classifiers in Mandarin Chinese behave more like clitics. Finally, few remarks about this section will be presented in 2.5.. 學. ‧ 國. 立. 政治大. 2.1 Definiteness and Specificity. ‧. Definiteness and specificity are two important concepts when discussing the. Nat. io. sit. y. syntax and semantics of nominal expressions. In this section, I will first briefly. n. al. er. introduce the definitions and some assumptions under these two concepts.. Ch. engchi. i n U. v. 2.1.1 Lyons (1999) Lyons (1999) claims that definiteness is universal. In any language, the referents of some nominal are identifiable to the hearer, no matter there is any grammatical device to indicate the definiteness of the nominal or not. Though universal, definiteness still varies syntactically among languages and it may not be encoded in some languages (Lyons 1999). In English, definiteness and indefiniteness are marked 11.

(27) by articles and other determiners. It is easy to determine whether a nominal expression is definite or indefinite in English, as definiteness is always encoded by definite determiners. However, in a language like Chinese, definiteness of the nominal usually is not encoded syntactically; the definite interpretation of the nominal expression usually depends on pragmatic factors such as shared background knowledge or prior discourse (Lyons 1999).. 政治大. Lyons (1999) also classifies definites into simple and complex ones. Simple. 立. definites are definiteness signaled by grammatical or functional morphemes (e.g., the. ‧ 國. 學. in English). Complex definites are proper names, personal pronouns, noun phrases. ‧. with possessives or demonstratives. In Chinese, there is no grammatical marker to. Nat. sit. n. al. er. io. demonstratives.. y. mark definiteness, but it is also encoded by proper names, personal pronouns and. Ch. engchi. i n U. v. 2.1.2 Frawley (1992) A close concept to definiteness is specificity. Frawley (1992) defines specificity as a concept which “refers to the uniqueness, individuation, or referential accessibility of an entity in a mentally projected world.” Unlike definiteness, a nominal can be both specific and non-specific in the same syntactic position, as the below example shows. (2) I’m looking for a man who speaks Mandarin. 12.

(28) In (2), there are two readings. It can mean that (i) the speaker is finding a particular man who speaks Mandarin, or (ii) the speaker is finding anyone who speaks Mandarin. Also, noted by Frawley (1992: 74), specificity “can be functionally induced by tense, mood, definiteness, context of utterance or assumed knowledge base.” For example, in English, the past tense usually induces a specific reading of the nominal and the habitual present tense a nonspecific reading. He states that this is due to actuality that. 政治大. past tense denotes what really happened in the past.. 立. (3) a. John bought the book.. ‧ 國. 學. b. John buys the book.. ‧. While the object in (3a) refers to a specific book, the object in (3b) does not have to. sit. y. Nat. io. n. al. er. refer to a specific book. It can either mean John habitually buys certain books, or it can also refer to any book.. Ch. engchi. i n U. v. Definiteness and specificity are closely related. Definite nominals tend to be specific, as they tend to encode old information (Frawley 1992). Still, not all definite descriptions are specific; similarly, not all indefinite descriptions are nonspecific. The judgment of a nominal being definite or specific is depending on the contextual factors.. 13.

(29) 2.2 The interpretation and the distribution of the numeral classifier phrase [Num+CL+N] Numeral classifier phrases [Num+CL+N] have generally been regarded as indefinite expressions and are not allowed to appear in subject or topic position in Mandarin Chinese (Chao 1968, Li and Thompson 1981, Tsai 1994, 1996).. (4) a. ?? 三. 個. 小孩. 吃了. 冰淇淋. san ge xiaohai chi-le bingqilin three CL kid eat ASP ice cream ‘Three kids ate the ice cream.’. 立. 小孩. 很. ‧ 國. 個. 可愛. 學. b. ?? 三. 政治大. y. 以為吃了. 冰淇淋. io. san ge xiaohai wo yiwei chi-le bingqilin three CL kid I think eat ASP ice cream ‘Three kids , I thought (they) ate the ice cream.’. n. al. b. ?? 三. 個. Ch. 小孩，我以為. engchi 很. sit. 小孩，我. er. 個. Nat. (5) a. ?? 三. ‧. san ge xiaohai hen keai three CL kid very cute ‘Three kids are cute.’. i n U. v. 可愛. san ge xiaohai wo yiwei hen keai three CL kid I think very cute ‘Three kids, I thought they are smart.’ These ungrammatical sentences are saved by adding the existential marker you ‘have, exist’ before the numeral classifier phrase.. 14.

(30) (6) a. 有. 三. 個. 小孩. 吃了. 冰淇淋. you san ge xiaohai chi-le bingqilin have three CL kid eat ASP ice cream ‘There were three kids who ate the ice cream.’ b. 有. 三. 個. 小孩. 很. 可愛. you san ge xiaohai hen keai have three CL kid very cute ‘There are three kids who are cute.’ (7) a. 有. 三. 個. 小孩，我. 以為吃了. 冰淇淋. you san ge xiaohai wo yiwei chi-le bingqilin have three CL kid I think eat ASP ice cream ‘There are three kids whoI thought (they) ate the ice cream.’ 三. 立小孩，. 個. 我以為. 很. 可愛. 學. you san ge xiaohai wo yiwei hen keai have three CL kid I think very cute ‘There are three kids who I think (they) are cute.’. ‧. ‧ 國. b. 有. 政治大. Nat. io. sit. y. However, there are many counterexamples to this generalization. [Num+CL+N] are. er. also found to occur in subject and topic positions (e.g., Tsai 1994a, 1996 and Li 1996).. al. n. v i n Ch Examples below show that [Num+CL+N] i U in subject and topic positions. e n g ccanhappear (8) 一. 張. 床. 擠了. 五. 個. 人. yi zhang chuang ji-le wu ge ren one CL bed squeeze ASP five CL person ‘One beds were crowded with five people. (9) 三. 個. 保母. 就. 照顧. 他一個小孩？. san ge baomu JIU zhaogu ta yi ge xiaohai three CL babysitter only care he one CL child ‘Three babysitters took care of him, only one child?’. 15.

(31) (10) 一個老師. 就. 把那群. 壞孩子. 控制住. 了. yi ge laoshi jiu ba na qun huaixiaohai kongzhizhu le one CL teacher JIU BA that group bad kids control-hold PAR ‘One teacher have controlled that group of bad kids.’ (11) 三. 張. 衛生紙. 夠. 你. 擦. 屁股嗎. san zhang weishengzhi guo ni ca pigu ma three CL tissue paper enough you wipe butt Q ‘Are three pieces tissue enough for you to wipe after bm?’. Two aspects explaining the occurrence of [Num+CL+N] are related to. 政治大. definiteness and the quantity-denoting expression. These two aspects will be introduced in 2.2.1 and 2.2.2.. 立. ‧. ‧ 國. 學. 2.2.1 Li (1998) and Huang, Li and Li (2009). Nat. io. sit. y. Li (1998) argues that there are two types of numeral classifier phrases. er. [Num+CL+N] which exhibit different syntactic behaviors. Those [Num+CL+N] that. al. n. v i n do not occur at sentence initial areC indeed indefinite; yetUfor those occur at sentence hen gchi initial, Li (1998) provides further explanations that [Num+CL+N] in these sentences are actually quantity-denoting expressions, not indefinite expressions. She argues that the verb like gou ‘enough/sufficient’ in (11) denotes the sufficiency of a certain amount. The verb kongzhizhu ‘to control over’ in (10) expresses the amount of children being controlled over. The different structural representations are supported from the evidence related to pronominal binding and scope interaction. 16.

(32) Li (1998) argues that a quantity-denoting expression [Num+CL+N] never corefers with a pronoun, nor can it serve as an antecedent of a bound pronoun like ta ‘he’ or tamen ‘they.’ Compare the indefinite expression in (12a) and (13a) with the quantity denoting expression (12b) and (13b).. 個. (12) a. *三. 人. 抬不動. 這. 架鋼琴。. 他們. san ge ren i taibudong zhe jia gangqi. three CL people lift-not-move this CL piano. 力量. 太. 小. 的. tamen i their. (Li 19a). de DE. 政治大. liliang tai xiao (Quantity-denoting) strength too small. ‘Three people cannot lift up this piano. Their strength is too weak.’ 明天. 會. 看到. 三個. 人. 學. b. 他. ‧ 國. 立. 會. 跟. 他們. (Li 19b). ‧. ta mingtian hui kandao san ge ren i , hui gen tamen i he tomorrow will see three CL people will with them 做朋友. y. Nat. sit. n. al. er. io. zuo pengyou (Indefinite expression) make friends ‘He will meet three people tomorrow and will make friends with them.’ 個. (13) a. *三. 人. Ch. i n U. e n g c你h i 給. 吃不完. v. 他們. 的五碗. 飯 (Li 20a). san ge ren i chi-bu-wan ni gei tamen i de wu wan fan three CL men eat-not-finish you give them DE five bowl rice ‘Three men cannot finish five bowls of rice you gave to them.’ b. 有. 三. 個. 人. 吃不完. 你. 給. 他們. 的五碗飯 (Li 20b). you san ge ren i chi-bu-wan ni gei tamen i de wu wan fan have three CL men eat-not-finish you give them DE five bowl rice ‘There are three men unable to finish five bowls of rice you gave to them.’. Quantity-denoting expressions not only behave differently from indefinite ones with respect to pronominal coreferences but also with respect to scope interaction. A 17.

(33) quantity-denoting expression does not have scope interaction. For example, there is only one reading in (14) that the amount of soup consumed by the three people in five bowls:. (14) 三個人喝得完. 五碗. 湯. san ge ren he-de-wan wo wan tang three CL person drink-de-finish five bowl soup ‘Three people can finish three bowls of soup.’. 政治大. But indefinite [Num+CL+N] can have scope interaction. See the example below.. 讓三. 個. 立. 人. 喝完. 五碗. 湯. ‧. wo rang san ge ren he-wan wo wan tang I let three CL person drink-finish five bowl soup ‘I let three people eat five bowls of soup.’. 學. ‧ 國. (15) 我. Nat. io. sit. y. Li (1998) also points out that the indefinite expression, but not the. er. quantity-denoting expression, can co-occur with dou ‘all’, which refers to an entire set. al. n. v i n of individuals to derive a universal C expression, U have’, which asserts the h e n gandcyou h i ‘exist, existence of individuals (an existential expression).. (16) 三. 個學生. 都. 來. 這兒了. san ge xuesheng dou lai zher le three CL student all come here PAR ‘Three students all came here.’ (17) 有. 三. 個. 學生. 來. 這兒. (Li 14). 了. you san ge xuesheng lai zher le (Li 15) have three CL student come here PAR ‘There are three students that came here.’ 18.

(34) Because the two types of [Num+CL+N] exhibit different syntactic behaviors, Li (1998) claims that the quantity-denoting expression has a Num as its head, and the Num projects a NumP. An indefinite [Num+CL+N], on the other hand, has a null D head and the D head projects a DP. Huang, Li and Li (2009) support Li’s (1998) work state that the two structures capture the nature of syntax. If there is no D in its structure, [Num+CL+N] can only. 政治大. be interpreted as quantity-denoting since the D of a DP is generally the locus of. 立. reference. The two structures for yi ge ren ‘one person’ are presented below.. ‧ 國. 學. (18) a. structure for a quantity-denoting yi ge ren. ‧. NumP. sit. y. CLP NP. n. al. CL. er. io. yi. Nat. Num. Cgeh. e nNg c h i ren. 19. i n U. v.

(35) b. structure for an indefinite yi ge ren DP D yi. NumP Num. CLP. ge. CL. NP. ti. ren. 政治大. When yi ge ren is interpreted as quantity-denoting, there is no need to specify the. 立. indefiniteness of the whole phrase, thus it has the structure as shown in (18a). But. ‧ 國. 學. when yi ge ren is interpreted as indefinite, the classifier ge will move to Num head,. ‧. and yi to D head to receive an indefinite feature.. Nat. io. sit. y. Since there are two possible structures for [Num+CL+N], in order to make sure. er. only one interpretation is possible, being it indefinite or quantity-denoting, there must. al. n. v i n C hthe null D head( Huang, be a governor which properly governs e n g c h i U Li and Li 2009). For. subject and topic positions, because no lexical item is available to govern them, it is always impossible for a [Num+CL+N] to occur. To be more specific, according to Huang, Li and Li (2009), an indefinite [Num+CL+N] can never be in positions like subject or topic because these positions are too high to have a governor which can lexically govern the null D head.. 20.

(36) 2.2.2 Cheng and Sybesma (1999) While Li (1998) and Huang, Li and Li (2009) suggest a DP analysis for the prohibition of indefinite [Num+CL+N] in subject and topic positions, Cheng & Sybesma (1999) argue that the prohibition of indefinite [Num+CL+N] in subject or topic positions is because the only interpretation of [Num+CL+N] in these positions is definite. For instance,. 政來了治大(Cheng and Sybesma 1999: 57) na san ge xuesheng lai le 立 that three CL student come PAR 三. 個. 學生. 學. ‧ 國. (19) a. 那. ‘Those three students came.’ 個. 學生. 都. 來了. ‧. b. 三. n. sit er. io. al. y. Nat. san ge xuesheng dou lai le three CL student DOU come PAR ‘All three students came.’. i n U. v. Cheng and Sybesma (1999) claim that the only way for [Num+CL+N] to appear in. Ch. engchi. sentence initial is by adding a demonstrative zhe ‘this’ or na ‘that’, since demonstrative marks the definiteness of the whole phrase, or by using the universal quantifier dou ‘all’, which makes the sentence interpreted as ‘for all x, x is a member of three students, x came’. “All three students” in (19b) must refer to all the members of a certain group, thus [Num+CL+N] gets the definite interpretation here.. 21.

(37) 2.3 The interpretation and the distribution of the bare classifier phrase [CL+N] As mentioned earlier, in Mandarin, a classifier does not always occur with a numeral preceding it. Phrases with such form are called bare classifier phrases. While most agree that [CL+N] is the phonological reduction of [yi+CL+N] (Lu 1990, Chen 2004), some hold the opinion that they belong to different structures (Cheng and Sybesma1999). Different interpretations they proposed, most works agree that bare. 政治大. classifier phrases [CL+N], similar to numeral classifier phrases [Num+CL+N],. 立. usually do not occur in definite positions.. ‧. ‧ 國. 學. 2.3.1 Lu (1990) and Chen (2004). Nat. io. sit. y. There is a special use of the indefinite determiner in Chinese that has not been. er. reported for indefinite articles or the numeral ‘one’ in other languages. It was first. al. n. v i n observed in Lu (1990 [1944], 1990C [1948]) can be used, normally h e nthatg ‘yi+classifier’ chi U with yi omitted, denoting a definite reference. Consider the following examples from Lu (1990[1944] : 164). In both examples, ‘yi+classifier’ introduce a definite referent.. (20) 只. 這一句，. 把. (一) 個江. 平. 唬了一跳. zhi zhe yi ju, ba yi ge Jiang Ping hu le yi tiao only this one utterance BA one CL Jiang Ping scare PFV one jump ‘Just this one utterance gave Jiang Ping a fright.’. 22.

(38) (21) 他. 被朋友. 把. (一) 個太太給. 騙走. 了. ta bei pengyou ba yi ge taitai gei pianzou le he BEI friend BA one CL wife GEI cheat-away PRA ‘he was cheated by his friend out of his wife’ or ‘He suffered from his friend cheating his wife away from him.’. However, Lu (1990) points out that this usage is different from the following indefinite expression.. (22) 有. (一) 個江. 平. 想要. 見你. you yi ge Jiang Ping xiangyao jian ni. have one CL Jiang Ping think-want see you ‘There is a certain Jiang Ping who wants to see you.’. 立. 政治大. ‧ 國. 學. It is also observed by Lu (1990) that, besides prosodic constraints, yi is more likely to. ‧. be omitted when used as a marker of nonidentifiable nonspecific reference. But it can. Nat. sit. al. n. 事，昨天. 我請了. Ch. er. io. (23) 這件. y. also be omitted in a nonidentifiable specific position.. (一) 個人. v nreni. 來. zhe jian shi zuotian wo qing le yi ge lai this CL matter yesterday I invite ASP one CL person come ‘For this issue, I invited a person yesterday. ‘. engchi U. Chen (2004) adopts Lu’s (1990) analysis that [CL+N] is the omitted form of [yi+CL+N], and further points out under what condition can [yi+CL+N] omit yi. He argues that only when [yi+CL+N] is interpreted as indefinite instead of quantity-denoting can yi be omitted. Chen (2004) also provides a piece of phonological evidence to support Lu’s (1990) argument. When [yi+CL+N] has a 23.

(39) numeral reading, yi is always stressed and thus cannot be omitted; while when it has an indefinite reading, it is unstressed and commonly omitted. Chen (2004) proposes a more inclusive and revealing account by agreeing with Heine’s (1997) grammaticalization processes. Heine (1997) argues that the more stages an item has gone through from numeral to indefinite article, the more it is affected by grammaticalization processes such as bleaching, cliticization, and. 政治大. phonetic erosion. Chen (2004) states that the more grammaticalized yi is, the more. 立. weakened is the morphological and phonological weight of yi in [yi+CL+N]. To be. ‧ 國. 學. more specific, the extent of the phonological reduction of the Chinese numeral yi. ‧. correlates perfectly with the order of its development through the five stages along the. Nat. n. al. er. io. sit. y. continuum of grammaticalization.. 2.3.2Cheng and Sybesma (1999). Ch. engchi. i n U. v. Cheng and Sybesma (1999) also discuss the relationship between [yi+CL+N] and [CL+N]. Unlike many believe that [CL+N] is the phonological reduction of [yi+CL+N], they claim that these two structures are totally different structures for some reasons. The first reason is related to the bounded predicate. Sybesma (1992) argues that with predicates that are bounded for reasons independent of the object, a strong reading is forced upon the object: a bare noun is interpreted as definite, an 24.

(40) indefinite NP as specific. For instance,. (24) a. 我吃完. 了. 一. 塊. 餅乾. (Cheng and Sybesma 1999: 29, 30). wo chiwan le yi kuai binggan I eat-finish ASP one CL cookie ‘I finished a cookie.’ b. *我吃完. 了. 塊. 餅乾. wo chiwan le kuai binggan I eat-finish ASP CL cookie ‘I finished a cookie.’. 治政 They claim that in above example, [yi+CL+N] must 大be interpreted as specific, thus 立 ‧ 國. 學. the indefinite and nonspecific [CL+N] cannot appear in this position. The other bounded object position they point out is the predicate following ba. They argue that. ‧. BA construction provides a similar context of boundedness, so [CL+N] cannot occur. n. al. er. io. sit. y. Nat. here.. (25) a. 我把一. 碗湯. Ch. 喝完. 了. engchi. wo ba yi wan tang hewan le I BA one CLs oup drink-finish PAR ‘I finished a (particular) bowl of soup.’ b. * 我把. 碗. 湯. 喝完. i v and Sybesma 1999: 31) (Cheng n U. 了. wo ba wan tang hewan le I BA CL soup drink-finish PAR ‘I finished a (particular) bowl of soup.’. The second reason is related to the secondary predication. Huang (1987) argues that object NP in positions like the subject of the secondary object must be indefinite 25.

(41) but specific. Therefore, [CL+N] cannot occur here, either.. (26) a. 我教. 過. 一個學生. 很. 聰明. wo jiao guo yi ge xuesheng hen congming I teach EXP one CL student very intelligent ‘I once taught a student who was very intelligent.’ b. *我教. 過. 個. 學生. 很. 聰明. wo jiao guo ge xuesheng hen congming I teach EXP CL student very intelligent ‘I once taught a student who was very intelligent.’. 治政 According to the evidence provided by Cheng and大 Sybesma (1999), [CL+N] 立 ‧ 國. 學. phrases cannot simply be considered as the reduced form of [yi+CL+N] phrases. In addition, they have found that [CL+N] in Mandarin only occurs postverbally but not. ‧. preverbally or at sentence initial, as shown in (27).. y. sit. al. n. wo xiang kan ben shu I want read CL book ‘I want to read a book.’. er. 本書. io. 看. Nat. (27) a. 我想. b.*隻狗. 嚇到. Ch. engchi. i n U. v. 他. zhe guo xiadao ta CL dog scare he ‘A dog scared him.’. Compared with Mandarin, Cheng & Sybesma (1999) have found that the bare classifier phrase [CL+N] in Cantonese can appear at sentence initial, denoting a definite expression, yet it can also appear postverbally, indicating both a definite and 26.

(42) an indefinite expression.. (28) a. zek gau zungji sek juk CL dog like eat meat ‘The dog likes to eat meat.’. (Cantonese). b. ngo zungji tong zek gau waan I like with CL dog play ‘I like to play with the dog.’ c. keoi seung maai gaa ce he want buy CL car ‘He wants to buy a car.’. 立. 政治大. ‧ 國. 學. A brief summary about the differences between the bare classifier phrase in Mandarin and in Cantonese is made: in Mandarin, [CL+N] is always interpreted as indefinite. ‧. while in Cantonese [CL+N] can be definite or indefinite. Cheng and Sybesma (1999). sit. y. Nat. io. n. al. er. also propose different structures, a CLP for the definite [CL+N], and a NumP for the indefinite [CL+N], as presented in (29). Ch. engchi. (29) a. CLP for the definite [CL+N] CLP CL. i n U. v. b. NumP for the indefinite [CL+N] NumP. NP. Num. N. CLP CL. NP N. They further provide a constraint explaining why [CL+N] cannot be interpreted as definite in Mandarin. Cheng and Sybesma (1999) claim that in Mandarin Chinese, 27.

(43) overt classifiers are always accompanied by a numeral, and the numeral can be overt or nonovert. In other words, in Mandarin, whenever there is an overt classifier, it must occur in the structure in (29b). Cantonese, however, lacks such a constraint on the occurrence of overt classifiers. In Cantonese, the occurrence of a classifier does not automatically imply the presence of a numeral. Thus, Cantonese nouns involving a classifier are not necessarily indefinite. This explains why [CL+N] phrases in. 政治大. Mandarin cannot be interpreted as definite. Since subject position and the bounded. 立. predicate in Mandarin Chinese are forced to be interpreted as definite, it is impossible. ‧ 國. 學. for [CL+N] to occur in these positions.. ‧. To sum up, Cheng and Sybesma (1999) argue that the different syntactic. Nat. io. sit. y. behaviors of [CL+N] in Mandarin Chinese and in Cantonese are due to the feature. er. [+definite]. Besides, they argue that classifiers in Chinese specify definiteness,. al. n. v i n C hthe functions of determiners individuation, and number, which are in languages with engchi U. overt Ds like English. Therefore, argument nominals in Chinese should be represented as CLPs.. 2.3.3 Yip (2008) Yip (2008) provides arguments against Cheng and Sybesma (1999)’s CLP hypothesis. Yip (2008) first gives the definitions and distinguishes between sortal 28.

(44) classifiers (what Her 2010 defines as classifiers), mensural classifiers (what Her 2010 refers to as measure words), and true measures, which represent a unit of measure like dimensions (weight, height, and length), and length of time, for example, gongjin ‘kimogram’ and mi ‘meter’.. (i) A sortal classifier ‘individuates whatever it refers to in terms of the kind of entity that it is.’ (Lyons 1977: 463), (ii) A mensural classifier ‘individuates in terms of quantity’. (Lyons 1977: 463) (iii) A true measure does not individuate and only assigns to the noun the measurement that the numeral specifies.. 政治大. 立. ‧ 國. 學. Yip (2008) starts by pointing out the different syntactic behaviors true measures have. First, unlike sortal classifiers and mensural classifiers, true measures do not. ‧. have postverbal [TM+N] construction as shown in (20).. io. al. (sortal classifiers). n. wo xiang mai ba dao I want buy CL knife ‘I want to buy a knife’. b. 我想買打蛋. Ch. engchi. er. sit. y. Nat. (30) a. 我想買把刀. i n U. v. wo xiang mai da dan (mensural classifiers) I want buy M egg ‘I want to buy a dozen of eggs.’ c. *我. 想. 走. 哩路. wo xiang zou li lu I want walk TM road ‘I want to walk a mile.’. (true measures). Second, although it has been known to many people that in the literature 29.

(45) Cantonese classifiers mark definiteness (Cheng and Sybesma 1999, Au Yeung 2005), Yip (2008) points out that Cantonese true measures actually do not appear at sentence initial.. (31) a. bui caa hou jit CL tea very hot ‘The tea is very hot’. (Yip 2008: 5). b. *cek dei hou gwai TM land very expensive ‘The square feet of land is very expensive’. 政治大. 立. ‧ 國. 學. Third, unlike sortal classifiers mensural classifiers, true measures are disallowed to appear directly following demonstratives.. ‧. 很鋒利. y. Nat. (32) a. 那 (一) 把刀. sit. n. al. er. io. na yi ba dao hen fengli that one CL knife very sharp ‘That knife is very sharp.’ b. 那 *(一) 哩路. 很. Ch. 難. engchi. 走. i n U. v. na yi li lu hen nan zou Dem one TM road very difficult walk ‘That mile is hard to walk.’. Yip (2008) thus argues against Cheng and Sybesma’s (1999) overgeneralization that definiteness is the only reason which decides the occurrences of all classifiers in Mandarin Chinese. The prohibition of [TM+N] occurring in the sentence initial position is not just due to its definiteness, but its incompetence to individuate. 30.

(46) 2.4 Other analyses on the distribution of [Num+CL+N] and [CL+N] The dominant analyses for the distribution of [Num+CL+N] and [CL+N] are concerning (in)definiteness and quantity-denoting expressions as mentioned above. However, some researchers propose analyses beyond the dominant analyses. Two recent views will be introduced in this section.. 2.4.1 Ji (2007). 立. 政治大. Ji (2007) argues against Cheng and Sybesma (1999) that he thinks [CL+N]. ‧ 國. 學. pattern is the result of the phonological reduction of the numeral yi and the two. ‧. patterns have the same meanings and distributions. He opposes to Cheng and. Nat. io. sit. y. Sybesma’s (1999) argument that indefinite [CL+N] can only have nonspecific. er. reference, while indefinite [yi+CL+N] can have both specific and nonspecific. al. n. v i n Cevidence. references by providing some Ji (2007) points out that the specific and h e n gFirst, chi U. nonspecific reading of [CL+N] is depending on the verb tense. See (33a) and (33b) for the contrast.. (33) a. 我想. 看. 本書. wo xiang kan ben shu I want read CL book ‘I would like to read a book.’ b. 昨晚 zuowan. 我看了. 本書. wo kan-le. ben shu 31.

(47) yesterday evening I read ASP CL book ‘I read a book yesterday evening.’. In (33a), the indefinite [CL+N] phrase ben shu tends to have a nonspecific reference. Yet ben shu can only have a specific indefinite interpretation in example (33b). Therefore, Ji (2007) argues that the verb tense plays an important role in deciding the [CL+N] phrase has a specific or nonspecific indefinite interpretation, and same for [yi+CL+N].. 治政 Ji (2007) also argues against the two contexts Cheng 大 and Sybesma (1999) 立 ‧ 國. 學. provide. One context Cheng and Sybesma (1999) propose is that the object of bounded predicates which are bounded for reasons independent of the object cannot. ‧. be nonspecific indefinite. Thus, according to this analysis, [CL+N] cannot occur as an. sit. y. Nat. io. n. al. er. object in such a context. Take a pair of examples from Cheng and Sybesma (1999).. (34) a. 我喝完. 了. 一. Ch. 碗. 湯. engchi. wo hewan le yi wan tang I drink-finish ASP one bowl soup ‘I finished a bowl of soup.’ b* 我喝完. 了. 碗. i n U. v. 湯. wo hewan le wan tang I drink-finish ASP bowl soup ‘I finished a bowl of soup.’. Ji (2007) argues that the ungrammaticality of (34a) is due to the syllable of both N and V compound. It sounds natural by substituting the N to more syllables or the verb 32.

(48) compound to monosyllable.. (35) a. 我喝完. 了. 碗. 酸辣湯. wo he-wan le wan suanlatang I drink-finish ASP CL sour-hot-soup ‘I finished a bowl of sour-hot soup.’ b. 我喝了. 碗. 湯. wo he le wan tang I drinkASP CL soup ‘I drank a bowl of soup.’. 治政 Ji (2007) provides an explanation for (34) and (35)大 by adopting Duanmu & Lu (2002) 立 ‧ 國. 學. and Duanmu’s (2000, 2001) studies on stress and word length. Duanmu & Lu (2002) argue that word choices in Chinese are concerning stress. Nonhead Stress Rule states. ‧. that syntactic nonheads should have stress. To obey this rule, objects in Chinese. sit. y. Nat. io. n. al. er. should have more stress than the verb. A word with more stress should not be shorter. i n U. v. than a word with less stress. Therefore, an object should not be shorter than its. Ch. engchi. preceding verb. Ji (2007) extends the analysis to the use of [CL+N] and claims that classifiers in Mandarin are usually unstressed and numerals are stressed. The occurrence of [CL+N] or the deletion of yi in [yi +CL+N] can be explained by the Nonhead Stress Rule. Yi in [yi+CL+N] phrases can be deleted when the object is a compound noun, but it cannot be deleted when the verb is a compound with the object noun being monosyllabic, as the above examples show. Another context that Cheng and Sybesma (1999) provide is related to the 33.

(49) secondary predication. The object nominal in such sentences functions as the subject of the secondary predicate. [yi+CL+N] in such positions cannot have their yi deleted. Ji (2007) argues that the grammaticality of (36a) ad (36b) is very subjective. Some of his informants think both sentences are ungrammatical. For instance,. (36) a. 我教過. 一. 個. 學生. 很. 聰明. wo jiao guo yi ge xuesheng hen congming I teach ASP one CL student very intelligent ‘I once taught a student who was very intelligent’. 政治大聰明. 很立 wo jiao guo ge xuesheng hen congming 個. 學生. 學. ‧ 國. b.*我教過. I teach ASP CL student very intelligent ‘I once taught a student who was very intelligent’. ‧. In conclusion, Ji (2007) states that both [yi +CL+N] and [CL+ N] have two. sit. y. Nat. io. n. al. er. indefinite interpretations: specific and nonspecific. The appropriate interpretation in. i n U. v. particular contexts is related to the verb form. Contrary to Cheng & Sybesma’s (1999). Ch. engchi. view, Ji (2007) supports Lu (1990) and Chen’s (2004) analyses that [CL+ N] results from the deletion of yi in [yi+Cl+N], and yi cannot be deleted in certain contexts for phonological reasons.. 2.4.2 Yang (2002) Yang (2002) discusses [CL+N] in one of the chapter in his dissertation. He agrees with Lu (1990) and Chen (2004) that the bare classifier phrase [CL+N] is the 34.

(50) suppressed form of [yi+CL+N]. In addition, he suggests a new hypothesis that classifiers in Mandarin Chinese are in fact clitics and are freely alternate with [CL+N] in a variety of contexts. Yang (2002) proposes two arguments related to this assumption. He claims that in Chinese a numeral may be optionally omitted from [yi+CL+N] context if and only if 1) there is another hosting word (a demonstrative, a quantifier, or a verb) preceding [CL+N], and 2) a strict locality condition is satisfied. 政治大. between the hosting word and [CL+N]. Evidence provided to support his argument is. 立. mainly based on the study Zwicky & Pullum (1983) propose. Zwicky and Pullum. ‧ 國. 學. (1983) propose some criteria to capture the characteristic distinction between clitics. ‧. and affiexes in general. Six lines of the criteria are A. the degree of selection between. Nat. io. sit. y. the dependent morpheme the word to which it is attached, B. arbitrary lexical gap, C.. er. phonological idiosyncrasies, D. semantic idiosyncrasies, E. syntactic operations. al. n. v i n affecting the combinations,Cand on the combinability of clitics with h eF nrestrictions gchi U inflectional affixes. Yang (2002) fits the classifiers in Mandarin into these criteria and gets the conclusion that Chinese classifiers do perform like clitics. A crucial evidence Yang (2002) adopts is the fact that yi-omission has restrictions not only on syntactic positions but also on the preceding elements to attach to. For example,. 35.

(51) (37) a. 那. 一. 本書. 很. 貴. na (yi) ben shu hen gui that one CL book very expensive ‘That book is very expensive.’ b. 那. 厚厚的. 本書. *(一). 很貴. na houhoude *(yi) ben shu hen gui that thick-DE one CL book very expensive ‘That thick book is very expensive.’. Example (37) shows that within a full NP, if the demonstrative (or a quantifier). 政治大. is separated from [yi+CL+N] by another modifier, the omission of the numeral. 立. becomes illegal. However, if we compare (36a) and (36b), we would find that. ‧ 國. 學. [yi+CL+N] in both sentences occupy the same syntactic position. The only difference. ‧. lies in the preceding element [yi+CL+N] attaches to. Therefore, Yang (2002). Nat. sit. n. al. er. io. 2.5 Remarks. y. concludes that classifiers in Mandarin are clitics.. Ch. engchi. i n U. v. Summarized from the above studies, the proposed issues concerning classifiers in [Num+CL+N] and [CL+N] are 1) whether [CL+N] is the omitted form of [yi+CL+N], and 2) the distribution of [yi+CL+N] and [CL+N]. While most analyses (Lu 1990, Chen 2004) agree that [CL+N] is the phonological reduction of [yi+CL+N], Cheng and Sybesma (1999) hold the opinion that they belong to different structures. Since Lu (1990) and Chen (2004) provide the contexts that [CL+N] can be interpreted 36.

(52) both as definite and indefinite, there is no reason to argue that [CL+N] and [yi+CL+N] belong to different structures. Also, the only value we can get from [CL+N] is one, but not other numbers. If we claim [CL+N] and [yi+CL+N] being different structures, why is it impossible to get other values rather than one from [CL+N]? As for the second issue, most suggest that the occurrences of [yi+CL+N] and [CL+N] are due to indefiniteness and quantity-denoting expressions (Cheng and. 政治大. Sybesma 1999, Li 1998, Chen 2004, Huang, Li and Li 2009). Cheng and Sybesma. 立. (1999) argue that because subject positions and topic positions in Mandarin Chinese. ‧ 國. 學. are always definite, it is impossible for [CL+N] to occur since [CL+N] can only be. ‧. understood as indefinite. For those who claim that [CL+N] being the phonological. Nat. io. sit. y. reduction of [yi+CL+N], the prohibition of [CL+N] from topic or subject positions is. er. due to both indefiniteness and quantity-denoting expressions. Those [yi+CL+N]. al. n. v i n C hpositions can never appearing in topic or subject e n g c h i U be interpreted as indefinite but. quantity-denoting expressions. And when [yi+CL+N] is interpreted as numeral expression, it cannot delete yi to become [CL+N]. It seems plausible that the above analyses provide reasonable solutions for the second issue. Two assumptions are made according to the above analyses. First, in positions where they are realized as definite, [CL+N] would never appear. Second, in positions. where. they. are. realized. as 37. indefinite,. [CL+N]. should. occur..

(53) Counterexamples have been found and proved that these two assumptions overgenerate the language use in Mandarin Chinese. Thus in the following chapter, I will point out the problems under the above analyses to show that previous studies fail to account for many counterexamples found in Mandarin Chinese. I will argue that neither indefiniteness nor quantity-denoting expression can account for the distribution of the bare classifier phrase [CL+N].. 立. 政治大. ‧. ‧ 國. 學. n. er. io. sit. y. Nat. al. Ch. engchi. 38. i n U. v.

(54) Chapter 3 Unsolved Problems for [yi+CL+N] and [CL+N] in Mandarin Chinese. 3.1 Assumptions from previous studies. 政治大. In last chapter, studies related to the interpretation, the distribution and the. 立. internal structure of [Num+CL+N] and [CL+N] have been reviewed. As mentioned. ‧ 國. 學. before, two factors that decide the distribution of [yi+CL+N] and [CL+N] are. ‧. definiteness and quantity-denoting expressions. When [yi+CL+N] is interpreted as. Nat. io. sit. y. indefinite (whether specific or non-specific), it can be reduced to [CL+N]. The. er. exclusion of [CL+N] occurring in subject or topic positions is that [yi+CL+N] is. al. n. v i n C h when it occursU in subject quantity-denoting engchi. realized as. position, and that. quantity-denoting [yi+CL+N] can never be reduced to [CL+N]. Two assumptions are made under the above analyses. First, the bare classifier phrase [CL+N], as a suppressed form of indefinite [yi+CL+N], can only occur in indefinite positions. Second, in a syntactic position which is realized as indefinite, [CL+N] should always be allowed to occur. However, the idiosyncratic examples are found that [CL+N] appear in positions which are considered definite or quantity-denoting. Besides, some 39.

(55) indefinite positions do not allow [CL+N] to appear. In this chapter, I am going to list these counterexamples and point out that indefiniteness and quantity-denoting expressions suggested by previous studies can only explain the occurrence of [CL+N] in some cases but cannot correctly predict the positions for the bare classifier phrase [CL+N] to occur.. 政治大. 3.1.1 [CL+N] only occurs in indefinite positions?. 立. The first assumption made from the previous studies, as mentioned above, is that. ‧ 國. 學. the bare classifier phrase [CL+N] can only occur in indefinite positions. Supporting. ‧. evidence from the previous studies is that [CL+N] never occurs in definite positions. Nat. io. sit. y. such as subject or topic positions. Cheng and Sybesma (1999) argue that [CL+N] in. n. al. er. Mandarin carries a [-definite] feature that prohibits it from definite positions, while. Ch. engchi. Lu (1990) and Chen (2004), claiming that [CL+N]. iv n being U the. reduced form of. indefinite [yi+CL+N], suggest that [CL+N] only appears in indefinite position where indefinite [yi+CL+N] occurs. However, with a closer examination, I have found that it is not the case. First, in the definite position where D head is occupied by a demonstrative zhe ‘this’ or na ‘that’, [CL+N] is always allowed to appear. (38a), (38b) and (38c) show the three distributions for [Dem+CL+N]. 40.

(56) (38) a. 這/那. (一) 本. 書我想. 看. (Topic position). zhe/na yi ben shu wo xiang kan this/that one CL book I want see ‘I want to read this/that book.’ b. 這/那. (一) 本. 書很好. 看. (Subject position). zhe/na yi ben shu hen hao kan this/that one CL book very good see ‘This/that book is very good.’ c. 我買. 了這/那. (一) 本書. (Postverbal position). wo mai le zhe/na yi 政ben shu治大 I buy ASP this/that one CL book 立 ‘I bought this/ that book.’. ‧ 國. 學. When [yi+CL+N] follows a demonstrative, it no longer serves as an indefinite. ‧. expression but a definite expression since the D head is occupied by an element. sit. y. Nat. io. n. al. er. carrying a definite feature. But (38a), (38b) and (38c) clearly show that even. i n U. v. quantity-denoting [yi+CL+N] can undergo yi omission. This fact fails out of the. Ch. engchi. prediction under the first assumption.. Another counterexample found in Mandarin is that, contrary to what it is assumed in the first assumption, yi in many quantity-denoting [yi+CL+N] can also be reduced.. (39) 我買了. (一)磅. 蘋果. wo mai-le yi bang pinguo I buy-ASP one pound apple ‘I bought one pound of apples.’ 41.

(57) (40) 三. 個人. 各. 吃了. (一) 碗飯. san ge ren ge chi-le yi wan fan three CL person each eat-ASP one M rice ‘For these three people, each of them has a bowl of rice.’ (41) 他送了. (一) 箱. 芒果. 給我們. ，但顯然. 不夠. ta song-le yi xiang mangguo gei women dan xianran bugou he give-ASP one M mango for we but obviously not-enough ‘He gave one box of mangoes for us, but obviously they are not enough.’. In (39), yi in yi bang can only refer to the number “one” but not an indefinite. 政治大. expression which refers to “any pound of apples.” Similarly, in (40), yi in yi wan. 立. denotes the number “one”, meaning these three people each has one bowl of rice. ‧ 國. 學. instead of any bowl of rice. In (41), yi in yi xiang specifies the number ’one’ rather. ‧. than the indefinite reading as well. To briefly sum up, all [yi+CL+N] in above. Nat. io. sit. y. examples mark numbers instead of indefiniteness, yet they are all allowed to be. n. al. er. reduced to [CL+N], which is not what the first assumption predicts.. Ch. engchi. Though I present some counterexamples above. iv n that U some. quantity-denoting. [yi+CL+N] can be reduced to [CL+N], there are some examples which do not allow yi omission.. (42) 我買了. *(一)磅. 蘋果，不是兩磅. wo mai-le yi bang pinguo bu shi liang bang I buy-ASP one pound apple not be two pound ‘I bought one pound of apples, not two pounds.’. 42.

(58) (43) 三. 個人. 各. 吃了. *(一) 碗飯，不是兩碗. san ge ren ge chi-le yi wan fan bu shi liang wan three CL person each eat-ASP one M rice not be two M ‘For these three people, each of them has a bowl of rice.’ (44) 叫. 你帶. *(一) 個朋友. 來，你帶了. 三. 個. jiao ni dai yi ge pongyou lai ni dai-le san ge call you bring one CL friend come you bring-ASP three CL ‘I told you to bring one friend here, but you brought three.’. Actually, (42) and (43) are very similar to (39) and (40), but both (42) and (43) have a. 政治大. second half of the sentence with the other quantity-denoting [Num+CL+N] in it.. 立. Contexts in (42), (43) and (44) are known as contrastive contexts since two. ‧ 國. 學. [Num+CL+N] are being compared. Again, the question pops out: what might be the. ‧. reason for similar sentences to be grammatical in (39) and (40) but not in (42) and. er. io. sit. y. Nat. (43)?. n. al. i n C 3.1.2 All [CL+N] are allowed h in indefinite positions? engchi U. v. The second assumption from previous studies concerns the occurrence of the bare classifier phrase [CL+N]. Cheng and Sybesma (1999) claim that all [CL+N] in Mandarin are interpreted as indefinite and thus occur in indefinite positions. Lu (1990) and Chen (2004), suggesting [CL+N] being the suppressed form of indefinite [yi+CL+N], also claim that [CL+N] occurs in indefinite positions. The implication from both analyses is that wherever a position is indefinite, [CL+N] can occur. 43.