現代漢語浮游量詞結構及相關句型句法分析

全文

(1)國立臺灣師範大學英語學系碩士論文 Masterss Thesis Department of English National Taiwan Normal University. 現代漢語浮游量詞結構及相關句型句法分析 A Syntactic Analysis of the Floating Quantifier Construction and a Related Construction in Mandarin Chinese. 指導教授：丁仁教授 Advisor: Dr. Jen Ting 研究生：鄭榆樺 Graduate: Yu-Hua Zheng 中華民國一百零八年八月 August 2019.

(2) Chinese Abstract. 本論文的目的是分析現代漢語一組在文獻上未受到太多關注的量詞。根據這組量詞的分佈和特性，這組量詞可以被分爲兩類：名詞組量詞(ADQ)和副詞量詞(AVQ)。就目前所知，文獻上有三篇作品曾探討過這組量詞的部分：Shin (2008), Paul and Whitman (2010)和 Kuo (2016)。由於這組量詞和浮游量詞(FQ)的表現十分相似，所以本論文將回顧擱淺分析(Stranding Analysis)和副詞分析(Adverbial Analysis)的文獻並且指出對於現代漢語中這組量詞，副詞分析會是比較恰當的分析方式。根據一些觀察，本論文我提到名詞組量詞和副詞量詞具有不同的基底結構。名詞組量詞是嫁接在名詞上的量詞，結構上就像是同位語(Aoun et al. 2001)或是名詞組反身强調詞 (Siemund 200c)，而副詞量詞本身就是一個副詞，所以出現在副詞可以出現的位置。至於副詞量詞和其相關名詞之間的回指關係，本論文認爲 Doetjes (1997)所提出副詞帶有空範疇(empty category)的分析並不能夠解釋中文浮游量詞的語言事實，本論文主張將現代漢語浮游量詞視爲不帶有空範疇的副詞量詞(Kayne 1981,Jaeggli 1982)。關鍵詞：浮游量詞、名詞組量詞、副詞量詞、漢語. i.

(3) English Abstract The aim of this thesis is to analyze a set of quantifiers in Mandanin Chinese, which has not received much attention in the literature. I claim that this set of quantifier can be divided into two types: the ADQ (the adnominal quantifier) and the AVQ (the adverbial quantifier), based on their distributions and properties. To the best of my knowledge, in the literature, three studies have explored some facts of the ADQ and the AVQ: Shin (2008), Paul and Whitman (2010) and Kuo (2016). This set of quantifiers has similar behaviors to Floating Quantifier so in this thesis I review the studies on Stranding Analysis and Adverbial Analysis and point out that Adverbial Analysis is the preferred approach to this set of quantifiers in Mandarin Chinese. I suggest that the ADQ and the AVQ have different underlying forms due to some observations. The ADQ is an adnominal quantifier, which adjoins to the host nominal phrase like an appositive (Aoun et al. 2001) or an adnominal self-intensifier (Siemund 200c), structurally. The AVQ is an adverb, which can surface in the positions where an adverb can occur. In terms of the anaphoric relationship between the AVQ and its assoicated nominal phrase, I suggest that Adverbial Analysis, in which the FQ goes with an empty category (Doetjes 1997), cannot hold for the facts of the AVQ in Chinese and that the AVQ can simply be analyzed as an adverb without an empty category (Kayne 1981, Jaeggli 1982).. Keywords: floating quantifier, adnominal quantifier, adverbial quantifier, Mandarin Chinese. ii.

(4) Acknowledgements The process of doing this research was tough but meaningful to me. During the process, frustrations made me stronger and tougher. Most importantly, I am grateful to my adviser Prof. Jen Ting for her thoughtful instructions and comments. She has encouraged and supported me whenever I had difficulties. She has selflessly spent a lot of time helping me revise and finish my thesis. Then, I want to give my sincerest appreciation to my committee members, Prof. Hsiao-hung Iris Wu and Prof. Jui-heng Huang. Due to their valuable comments, I have made a great progress on my thesis. Here, I would like to thank my college friends Fang-qi Wu, Ting-wei Zhuo and Pei-xuan Chen, who gave me the warmest hugs whenever I had hard times. Also, I am really thankful to my classmates in the linguistics program of the English Department at National Taiwan Normal University, Wei-lin Eileen Lin, Bao-sheng Mark Tu, Yan-cheng Shawn Lin and Shu-fen Amy Chen for their valuable encouragement and suggestions. Finally, I want to give my deepest thanks to my dad, mom, sisters, nephew and brother-in-law, who are always by my side with all their hearts. Their companies and support are the greatest inspiration for me to get on. In the end, I will give my deepest thanks and love to those who have accompanied me to go through all these ups and downs.. iii.

(5) List of Tables Table 1 The four types of numeral phrase in Shin (2008) ........................................... 56. iv.

(6) Table of Contents Chinese Abstract ............................................................................................................................. i English Abstract ............................................................................................................................. ii Acknowledgements ........................................................................................................................iii List of Tables.................................................................................................................................. iv Table of Contents ........................................................................................................................... v Chapter 1 Introduction .................................................................................................................. 1 1.1 The FQ and the PNQ............................................................................................................ 5 1.1.1. Surface form ................................................................................................................. 5 1.1.2 Readings ....................................................................................................................... 1c 1.2 Other superficially similar but underlyingly different numeral phrases ...................... 15 Chapter 2 Literature Review....................................................................................................... 22 2.1 The FQs in English ............................................................................................................. 22 2.1.1 The stranding approach .............................................................................................. 25. 2.1.1.1 Advantages of Stranding Analysis ............................................27 2.1.1.2 Disadvantages of Stranding Analysis .......................................29 2.1.2 The adverbial approach .............................................................................................. c5. 2.1.2.1 Advantages of Adverbial Analysis ............................................c8 2.1.2.2 Disadvantages of the adverbial approach.................................40 2.2 The numeral FQs in Japanese ........................................................................................... 49 2.3 Previous analyses of the Chinese facts .............................................................................. 55 2.3.1 Shin (2008).................................................................................................................... 55 2.3.2 Paul and Whitman (2010) ........................................................................................... 62 Chapter 3 Analysis ....................................................................................................................... 7c 3.1 Two independent elements, the PNQ and the FQ............................................................ 7c 3.2 The Adnominal Quantifier (ADQ) .................................................................................... 79 3.2.1 The structure of the ADQ ........................................................................................... 84 3.3 The Adverbial Quantifier (The AVQ) ............................................................................ 10c 3.3.1 The properties and distribution of the AVQ ........................................................... 104 3.3.2 Analysis....................................................................................................................... 111 3.3.3 Two readings of the AVQ ......................................................................................... 120 3.3.4 Some remarks on other constructions which are claimed to be FQ constructions in Chinese ................................................................................................................................ 126. 3.3.4.1 DOU ......................................................................................... 127 3.3.4.2 Other numeral quantifiers ...................................................... 1c1 Chapter 4 Conclusion ................................................................................................................. 1c5 References ................................................................................................................................... 1c8. v.

(7) Chapter 1 Introduction This thesis explores a particular quantifying phenomenon, which is named Floating Quantifier (henceforth FQ), in Mandarin Chinese (henceforth Chinese). FQs are quantifiers which can quantify their associated nominal phrases even if they are not adjacent. This construction is exemplified both in English and Chinese, as shown in (1) and (2).. (1) a. The carpets will have all been being dusted for two hours. b. The carpet will all have been being dusted for two hours. (Sportiche 1988) (2) Na. liang-dui shuangbaotai xiongdi yinggai (si-ge. that two-CL twins (si-ge. ren). four-CL person. brother. should. four-CL. ren) person. yiqi. dao. taishang. biaoyan.. together. arrive. stage.top. perform. hui will. ‘Those two pairs of twin brothers, four people, will likely perform on the stage together.s. In (1), interestingly, although the quantifier all and its associated nominal phrase the carpets are not side by side, all can still quantify the carpets. Similarly, in 1.

(8) in Chinese example (2), the nominal phrase na lianghdui shuangbactai xicngdi ‘those two pairs of twin brotherss and the phrase sihge ren ‘four peoples are intervened by a modal hui ‘wills but the latter can still quantify the former. Since the phrase sihge ren ‘four peoples and all can both quantify their related nominal phrases even though they and their related nominal phrases are not adjacent, and besides, they can float in different positions in a sentence, I will follow previous studies to call the phrases sihge ren ‘four peoples in (2) and all in (1) FQs. Furthermore, it is noteworthy that that the sentences with an FQ (see ex. (1)-(2)) and the sentences with an adnominal qunatifier (see ex. (c)-(4)) share a similar meaning leads some studies to propose that there exists a derivational relationship between the sentence with an FQ and that with an adnominal quantifier.1. (c) All the carpets will have been being dusted for two hours. (4) Na. liang-dui shuangbaotai xiongdi si-ge. that two-CL twins. brother. ren. four-CL person. yiqi. dao. taishang. biaoyan.. together. arrive. stage.top. perform. yinggai. hui. should. will. ‘Those two pairs of twin brothers, four people, will likely perform on the stage together.s 1. In this thesis, I will call the nominal phrase which is associated with an FQ the associated nominal phrase and name the nominal phrase which is adjacent to a quantifier the host nominal phrase. 2.

(9) In the literature, whether (1) and (c) have a derivational relationship is still under debate. I will address this issue in this thesis as well. Of these analyses, Stranding Analysis (Sportiche 1988, Miyagawa 1989, Chiu 1990, Cirillo 2009) and Adverbial Analysis (Bobaljik 1995, 200c, Cheng 1995, Naoya 1995, Doetjes 1997, Kobuchi 2004, 2007, Fitzpatrick 2006) are rather significant. The former approach suggests that (1) and (c) have the same underlying form, [Q DP]. On the other hand, the latter claims that the FQs in (1) are base-generated in the adverbial positions. In Chinese, the quantifiers in (2) and (4) can be either numeral or non-numeral and the relationship between (2) and (4) has not been discussed thoroughly. To the best of my knowledge, only three studies have briefly discussed the FQ and its non-floating counterpart: firstly, Shin (2008) suggests that (2) and (4) have no derivational relationship and that the numeral FQ in (2) should be analyzed as an argument; secondly, Paul and Whitman (2010) concentrate on the non-numeral FQ meihren ‘everones in double object construction (DOC), as shown in (5) and they specifically indicate that the FQ in DOC should be considered as an adverb under Adverbial Analysis.. c.

(10) (5) Wo I. song-gei. haizi-men. give-GEI child-PLU. mei-ren. yibai. kuai. qian.. every(one). 100. buck. money. ‘I gave the children each 100 dollars.s (Paul and Whitman 2010:12). Lastly, Kuo (2016) is concerned about the non-numeral quantifier in (5) as well, but unlike Paul and Whitman (2010), she proposes that the quantifier meihren ‘everyones in (5) is an adnominal quantifier like the non-floating counterpart sihge ren ‘four peoples in (4) and further she claims that Stranding Analysis can account for the relationship between (2) and (4). However, these three studies do not thoroughly explore the data of the FQ in (2) and the non-floating counterpart in (4) in Chinese. In this thesis, I will adopt Adverbial Analysis for the Chinese FQ in (2). Besides, I also explore the post-nominal quantifier (henceforth PNQ) in (4), which is the non-floating counterpart of FQ in (2). Importantly, this thesis thoroughly discusses the facts of the FQ and the PNQ in Chinese, and argues for Adverbial Analysis of FQ. In this thesis, the term FQ is used to simply refer to the phenomenon that the quantifier is not adjacent to its associated nominal phrase without any theoretical implication.. 4.

(11) 1.1 The FQ and the PNQ. There are some details about the FQ and the PNQ that need to be discussed here. This section presents the surface forms of the FQ and the PNQ, and then the possible readings of the FQ and the PNQ.. 1.1.1. Surface form. The surface form of the FQ and the PNQ are basically the same. They can be either a numeral (see ex. (6)) or non-numeral quantifier (see ex. (7)-(8)).2. (6) Numeral quantifier- [number + classifier + (common noun)] a. FQ Na-dui. xiongmao hui. liang-zhi. that-CL. panda. two-CL. will. (xiongmao) panda. yiqi. chuxian.. together appear. ‘Both of that panda couple will show up together.s. 2. Thanks to Professor Hsiao-hung Iris Wu and Professor Jui-heng Huang for pointing out the exmaples to me. 5.

(12) b. PNQ Na. yi-zhi. jiandui. liangbai. ge. that. one-CL. fleet. 200. CL. hui. ren. person. yiqi will together. chuxian. appear ‘The fleet, consisting of two hundred people, will show up.s (7) Non-numeral quantifier- [meih(yi) ‘everys + classifier + (common noun)] a. FQ Na-dui. xiongmao hui. that-CL. panda. yi-ke. pingguo.. one-CL. apple. mei. (yi). will every. one. zhi. (xiongmao). dou chi. CL. panda. all. eat. ‘Each of that panda couple will eat an apple,s b. PNQ Na. yi-zhi jiandui mei. that one-CL fleet. every. (yi). ge. (one). CL. chuxi huiyi. attend. meeting. ‘Everyone in the fleet will attend the meeting.s 6. ren person. hui. dou. will. all.

(13) (8) Non-numeral quantifier- [non-numeral + common noun] a. FQ Na-xie. xuesheng. that-CL student. hui. suoyou. ren. yiqi. canjia. will. all. people. together. attend. bisai. competition ‘All of those students will attend a competition.s b. PNQ Wo. xihuan. na-xie. xuesheng. I. like. that-CL. student. suoyou ren. all. people. ‘I like all of those students.s. First, it is noteworthy that in the surface form of numeral quantifier in (6) and that of the non-numeral quantifier in (7), the common noun of the quantifiers can be omitted when the common noun of the related nominal phrase and that of the numeral quantifier are the same, as shown in (6a) and (7a); otherwise, both the common noun of the related nominal phrase and that of the quantifier must be present, as shown in example (6b) and (7b).c. 3. The surface form in (7) can be simplified as meihren ‘everyones, only when the classifier is ge and the common noun is ren ‘persons. This may be because the use of ren is wider than its original use. 7.

(14) As shown in (6a) and (7a), if the common noun of the related nominal phrase and that of the quantifier both are xicngmac ‘pandas, the common noun of the quantifier can be omitted, but if they are different, both the common noun of the related nominal phrase jiandui ‘fleets and that of the quantifier ren ‘peoples must be present (see ex. (6b)-(7b)). However, the PNQ in (10) seems to be against the generalization discussed above that when the common noun of the quantifier and that of the related nominal phrase are identical, the omission of the noun of the quantifier is optional.. (9) Na-dui that-CL. xiongmao. liang-zhi. panda. two-CL. (?xiongmao) hui panda. will together. ‘Both of that panda couple will show up together,s. 8. yiqi. chuxian. appear.

(15) In (9), the noun of the PNQ and that of the host nominal phrase are the same but the omission of the noun of the quantifier xicngmac ‘pandas is strongly preferred. The unacceptability of the occurrence of the noun xicngmac ‘pandas in (9) may be attributed to the nature of language that two identical elements are less preferred to stay too close to each other.4 Therefore, the generalization discussed above still works. As to the other pattern of the non-numeral quantifier in (8), it includes the non-numeral quantifiers like sucycu ‘alls, mcuxie ‘somes, yixie ‘somes etc. These quantifiers must be followed by a common noun (see ex.(8)) since they are not allowed to occur alone even though the common noun of the quantifier and that of the associated nominal phrase are identical, as shown in (10).. (10) FQ Na-xie. ren. that-CL people. yinggai. hui suoyou *(ren). yiqi. canjia. should. will all. together. attend. person. huiyi. meeting ‘All of those people will likely attend the meeting.s. 4. Thanks for Prof. Jen Ting for pointing out this possibility to me. 9.

(16) In (10), even though the common noun of the FQ ren ‘peoples is the same as that of the associated nominal phrase, the common noun of the quantifier is still required. However, the common noun of the PNQ of the non-numeral quantifier in (8) must be different from the common noun of the host nominal phrase, as shown in (11). This is because first, the the common noun of the non-numeral quantifier in (8) cannot be omitted and besides, the common nouns of the PNQ and that of the host nominal phrase are too close to each other to be the same.. (11) PNQ ?Na-xie that-CL. ren people. suoyou all. (ren). yinggai. person. should. hui. yiqi. canjia. will. together. attend. huiyi. meeting ‘All of those people will likely attend the meeting.s. The unacceptability of example (11) is because that the common noun of the PNQ and that of the host nominal phrase are ren ‘persons so it is unnature to put two identical elements too close; furthermore, the non-numeral quantifier sucycu ‘alls requires a common noun so the common noun of the PNQ in (11) cannot be omitted. 10.

(17) This means that the common noun of the PNQ of the non-numeral quantifier in (8) cannot be the same as the common noun of the host nominal phase and at the same time the common noun of the PNQ is necessary. I will now further discuss the first pattern of the non-numeral quantifier, [mei (yi) ‘everys + classifier + (common noun)]. In this pattern, the number is restricted to one. That means phrases like mei liang ge ren ‘every two peoples (see ex.(12)) are not included in this thesis. Even though it can surface in different positions in a sentence like the FQ at issue in (2), I claim that if the number in this pattern is not one, such phrases behave like manner adverbs and not the FQ at issue.. (12) Zhe-xie this-CL ge CL. xuesheng. (mei. student. ren) person. liang. every two. hui (mei will every. liang two. ge. ren). CL person ge. ren). yinggai. (mei. liang. should. every. two. chi. CL person eat. yi-ge one-CL. dangao. cake. ‘Every two students will likely eat a cake.s. Here, I will argue for the above claim. The function of the phrase mei liang ge ren ‘every two peoples and that of the quantifier at issue are different. The phrase mei liang ge ren ‘every two peoples in (12) is mainly used to modify the action of the 11.

(18) sentence like the function of a manner adverb but the use of the FQ mei (yi) ge ren ‘everyones in (7b) is to emphasize and quantify its associated nominal phrase.5 It is unsurprising that mei liang ge ren ‘every two peoples and the FQ mei (yi) ge ren ‘everyones have different distributions. The phrase mei liang ge ren ‘every two peoples, like the manner adverb lianghge ren ‘two peoples discussed in Lee (1986) (see ex.(1ca)), cannot occur higher than DOU (see ex.(1cb)) but the FQ mei (yi) ge ren ‘everyones at issue can (see ex.(1cc)).. (1c) a. Zhe-xie this-CL yi-ge. xuesheng. yinggai. (dou). student. should. all. hui. liang-ge. will two-CL. ren. *(dou). person all. chi eat. dangao.. one-CL cake ‘Every two students will likely eat a cake.s b. Zhe-xie this-CL *(dou) all. xuesheng. yinggai. (dou). hui. mei. student. should. all. will. every. chi eat. yi-ge. liang two. ge. ren. CL person. dangao.. one-CL cake. ‘Every two students will likely eat a cake.s 5. If the quantifier mei (yi) ge ren ‘everyones occurs lower than DOU, then it can have the same function, modifying the predicate, as the phrase mei liang ge ren ‘every two peoples while if it occurs higher than DOU, then it must be an FQ which is used to quantify its associated nominal phrase. 12.

(19) c. Zhe-xie. xuesheng yinggai. (dou). hui. student. should. all. will. (dou) chi yi-ge. dangao.. this-CL. all. mei every. ge. ren. CL. person. eat one-CL cake. ‘Each of these students will likely eat a cake.s. The manner adverb lianghge ren ‘two peoples in (1ca) is not allowed to occur higher than DOU and neither is the phrase mei liang ge ren ‘every two peoples in (1cb). However, the FQ mei (yi) ge ren ‘everyones can occur higher or lower than DOU, as shown in (1cc). It is obvious that the distribution of the phrase mei liang ge ren ‘every two peoples in (1cb) differs from that of the FQ mei (yi) ge ren ‘everyones in (1cc). With regards to the different functions and distributions of the phrase mei liang ge ren ‘every two peoples and the FQ mei (yi) ge ren ‘everyones, the former can be distinguished from the latter and will be excluded from this thesis.. 1.1.2 Readings. The FQ and the PNQ have different readings. To be specific, the FQ can yield two readings: exhaustive and partitive reading, as displayed in (14), whereas the PNQ can only be used in the exhaustive context in (15). 1c.

(20) (14) Na-xie. xuesheng. that-CL student. yiding. san-ge. definitely. three-CL person. zhe-chang. bisai. de. jiang.. this-CL. game. win. prize. ren. nenggou. zai. can. in. a. ‘Three of those students can win a prize in this game.s (FQ-partitive) b. ‘Those three students can win a prize in this game.s (FQ-exhaustive) (15) Na-xie. xuesheng san-ge. ren. a!. that-CL. student. person. EXCL. yiqi. zai. zhe-chang bisai. de. jiang.. this-CL. win. prize. together in. three-CL. game. yiding. ！ definitely. nenggou can. a. *‘Three of those students can win a prize in this game.s (PNQ-partitive) b. ‘Those three students can win a prize in this game.s (PNQ-exhaustive). In (14), the FQ sanhge ren ‘three peoples can be interpreted as ‘all of those three studentss or ‘three of those studentss. However, the PNQ in (15) can only yield a partitive reading, ‘all of those three studentss. Summarizing section 1.1, the PNQ and the FQ can have either a numeral or a non-numeral form. If the common noun of the PNQ and the FQ and that of the related nominal phrase are identical, the common noun of the quantifier can be omitted but if 14.

(21) they are different the common nouns of both the quantifier and the related nominal phrase should be present. Furthermore, the second pattern of the non-numeral quantifier in (8) is different from the pattern of the numeral quantifier in (6) and the first pattern of the non-numeral quantifier in (7). The common noun of the non-numeral quantifier in (8) must be present. The other point about the PNQ and the FQ is that their interpretations can be different. The PNQ can only yield an exhaustive reading whereas depending on the context, the FQ can have either an exhaustive or a partitive reading.. 1.2 Other superficially similar but underlyingly different numeral phrases. In addition to the FQ mentioned in the previous section (see ex.(16)), there are at least two more kinds of numeral phrase in Chinese, as shown in (17) and (18), that seem to have FQ-like behaviors. I argue that the FQ in (16) is superficially similar to the FQ-like numeral phrases in (17) and (18) but underlyingly differs from them.. 15.

(22) (16) Na liang-dui shuangbaotai xiongdi hui si-ge that two-CL twins taishang. biaoyan.. stage.top. perform. ren. brother will four-CL person. yiqi. dao. together arrive. ‘Those two pairs of twin brothers, four people, will perform on the stage together.s (17) Shang. ge. xueqi,. zhe. ji-ge. tongxue. last. CL. term. this. several-CL classmate. ren. xi. person. wash. dou. liang-ge. always two-CL. wan. dish. ‘Last term, these several classmates washed dishes in groups of two.s (Lee 1986:82) (18) Ta he. yiding. yi-ge. ren. definitely one-CL person. lai. zher.. come. here. ‘He will definitely come by himself.s (Huang et al. 2009:c05). The numeral phrase in (17) is a manner adverb, pointed out by Lee (1986) and the numeral phrase in (18) is an emphatic adverb, pointed out by Huang et al. (2009). 16.

(23) Depending on their behaviors, I will call the numeral phrase in (17) numeral manner adverb and the numeral phrase in (18) numeral emphatic adverb. The two FQ-like numeral phrases are not taken into consideration in this thesis for the following reasons. First of all, the numeral manner adverb in (17) is different from the FQ at issue, syntactically and semantically. Lee (1986) claims that it is a manner adverb since it modifies the action of the sentence. That is, in (17), the action xi wan ‘do the dishs is done in groups of two people. Syntactically, the numeral manner adverb and the FQ do not have a similar distribution. Examples are shown in (19)-(21).. (19) a. Na-dui. fuqi. yinggai. hui. kuaisude yiqi. dao. that-CL husband.wife. should. will. quickly. arrive. wenjudian. caigou.. stationery.store. do.shopping. 17. together.

(24) b.*Na-dui. fuqi. that-CL. yinggai kuaisude hui. husband.wife should. wenjudian. quickly. yiqi. will. dao. together arrive. caigou.. stationery.store do.shopping ‘That married couple will likely go shopping quickly in the stationery together.s (20) a. Xia. ge. xueqi,. zhe. ji-ge. several-CL classmate will. next. CL term. this. ren. yiqi. xi. person. together wash. *b. Xia. ge. xueqi,. next CL term. tongxue. hui. dou. liang-ge. always. two- CL. wan. dish. zhe. ji-ge. tongxue. this. several-CL. classmate two-CL person. dou. yiqi. xi. always. together wash. liang-ge ren. hui will. wan. dish. ‘Next term, these several classmates should do the dishes in groups of two.s (21) a. Na-dui. fuqi. yinggai liang-ge ren. hui. yiqi. that-CL husband.wife. should. will. together. dao. wenjudian. caigou.. arrive. stationery.store. do.shopping. two-CL person. 18.

(25) b. Na-dui. fuqi. that-CL. yinggai hui. husband.wife should. will. dao. wenjudian. arrive. stationery.store do.shopping. liang-ge ren. yiqi. two-CL person. together. caigou.. ‘Both of that married couple will likely go shopping in the stationery together.s. Consider (19)-(21). It is obvious that the typical manner adverb kuaisude ‘quicklys in (19) and the numeral manner adverb lianghge ren ‘two peoples in (20) have a similar distribution. However, the FQ lianghge ren ‘two peoples in (21) has a different distribution from them. The manner adverb kuaisude ‘quicklys and the numeral manner adverb lianghge ren ‘two peoples both have to occur in a lower position than the modal hui but the FQ can occupy a higher position than the modal hui. On the other hand, semantically, the numeral manner adverb and the FQ have different interpretations. To illustrate, the numeral manner adverb in (20) is used to express the way the action takes place like the canonical manner adverb kuaisude ‘quicklys in (19) so it is unsurprising that Lee (1986) regards it as a manner adverb. However, the FQ in (21) is not used to modify the action of the sentence but to quantify the subject. Based on the syntactical and semantic dissimilarities between the 19.

(26) FQ and the manner adverb, we can distinguish the FQ from the numeral manner adverb effortlessly. As to the numeral emphatic adverb in (18), it means ‘by oneselfYalones in Chinese and its number is only restricted to one (Huang et al. 2009). Thus, Huang et al. (2009) deems it as an emphatic adverb. Comparing the emphatic adverb and the FQ, I found that even though the FQ and the emphatic adverb have a similar distribution (see ex.(2) and (22)), they still cannot be treated as the same type of element for some reasons.. (22) Ta he. yinggai. (yi-ge. ren). hui. (yi-ge. ren). may. one-CL. person. will. one-CL. person. taishang. biaoyan.. stage.top. perform. dao arrive. ‘He will likely perform on the stage by himself.s. First, the number the FQ indicates is not restricted to one but the numeral emphatic adverb is; second, the interpretation of the FQ is different from that of the emphatic adverb. The emphatic adverb expresses the meaning ‘aloneYby oneselfs while the function of the FQ is to emphasize the quantity of the associated nominal 20.

(27) phrase and refer to the associated nominal phrase. Due to their different interpretations, we can differentiate the FQ from the numeral emphatic adverb. To sum up, first, the numeral phrase of the manner adverb type proposed by Lee (1986) has a different distribution and properties from the FQ; second, the numeral phrase of the emphatic adverb discussed by Huang et al. (2009) has a similar distribution to the FQ but has different interpretations from the FQ. Based on the above arguments, I claim that the FQ is different from the numeral manner adverb and the numeral emphatic adverb. Therefore, I will not take these two kinds of numeral phrases into consideration. In this thesis, I propose that the PNQ and the FQ are used to quantify and emphasize the quantity of their related nominal phrases. This thesis is organized as follows: chapter 2 presents the previous studies on the FQs in English and Japanese, and on the FQs in Chinese; chapter c provides the syntactic analyses for the PNQ and the FQ in Chinese; chapter 4 gives the conclusion and implications of this thesis.. 21.

(28) Chapter 2 Literature Review Regading the similar behaviors shared by the Chinese FQ and the English FQ all, this chapter first reviews the previous studies on the FQs in English. Furthermore, noticing that the FQs in Japanese and the FQ in Chinese share some properties, I will then discuss these properties. In the second half of this chapter, I will present the studies which have explored some facts of the PNQ and the FQ in Chinese: Shin (2008) on the instances of the FQ, Paul and Whitman (2010) and Kuo (2016) on the instances of non-numeral quantifiers in DOC. This chapter is organized as follow. In section 2.1, the two main approaches to FQs in English will be presented; section 2.2 provides the characteristics of the numeral FQs in Japanese; lastly, in section 2.c, I will discuss the previous studies on the PNQ and the FQ in Chinese.. 2.1 The FQs in English. The phenomenon that the FQ can quantify its associated nominal phrase when they are not adjacent to each other has been an issue for decades. The canonical examples of the FQ are shown in (2c).. 22.

(29) (2c) a. All the carpets will have been being dusted for two hours. b. The carpets will all have been being dusted for two hours. c. The carpets will have all been being dusted for two hours. (Sportiche 1988). The relationship among the sentences in (2c) is interesting since although these quantifier all are situated in different positions, the three sentences still share a similar reading. On the surface, the quantifier all appears to float in sentences (2cb-c). Therefore, the floating-like quantifiers are called Floating Quantifiers in the literature. To begin with, two fundamental properties of FQs are introduced here. Firstly, as shown in (2c), FQs can occupy different positions in a sentence; secondly, when their FQs and their associated nominal phrases are not adjacent, the FQs still can quantify their associated nominal phrases. Furthermore, this relation between the FQs and their associated nominal phrases can only be established when FQs and the associated nominal phrases yield to the Locality Constraint and c-commanding relation (Kayne 1981, Sportiche 1988, Bobaljik 1995, 200c). To be specific, the nominal phrase and the FQ cannot be intervened by finite clause boundary in (24), and the nominal phrase necessarily c-commands the FQ in (25).. 2c.

(30) (24) *My friends think that I have all left. (25) *[The mother of my friends] has all left.. It is claimed that since the relationship between the FQs and its associated nominal phrases is the same as the relationship between an anaphor and its antecedant, the relationship between the FQs and its associatd nominal phrases is called anaphoric relationship (Kayne 1981, Bobaljik 1995, 200c, Doetjes 1997). The anaphoric relationship can be shown in the feature agreement system in some languages like French, as in (26).. (26) Les femmes. estaient. the women(fem.) were. touts Y*tous. bien vetues.. all(fem.)YallY(masc.). well dressed. (Cirillo 2009:2). In French, the quantifier tcus ‘alls needs to agree with the feature of its associated nominal phrase le femmes ‘the womens, which confirms the anaphoric relationship between the FQ and its associated nominal phrase. To account for the similar interpretations among (2ca-c) and the anaphoric relationship between an FQ and its associated nominal phrase, two main approaches to 24.

(31) FQs are proposed in the literature, namely Stranding Analysis and Adverbial Analysis, to be reviewed in 2.1.1 and 2.1.2 respectively.. 2.1.1 The stranding approach. The basic idea of Stranding Analysis is that the FQ and its associated nominal phrase are underlyingly adjacent to each other. Since the FQ can be stranded in the positions that the constituent consisting of the FQ and its associated nominal phrase can pass through, the FQ can surface in different positions, as shown in (2c) and (27). Under this scenario, the proponents of Stranding Analysis claim that there exists a derivational relationship among sentences in (2ca-c) and (27a-b), as noted by Sportiche (1988), Shlonsky (1991) and Cirillo (2009).. (27) a. The children will all leave. b. All the children will leave. c.. TP [DP the children]i Ts T. vP. VP. [DP all [ti]]. leave 25.

(32) Here, I mainly discuss the examples in (27). According to Sportiche (1988), Shlonsky (1991) and Cirillo (2009), (27a) and (27b) have a derivational relationship since the FQ and its associated nominal phrase in these two sentences share the underlying structure, [QP [Qs [Q all] [DP the children]]]. To derive (27a), the FQ all is stranded in its base position and the associated nominal phrase the children is moved to a higher position, as shown in (27c). To derive (27b), the quantifier all is not stranded but moved along with its associated nominal phrase the children. Under this analysis, quantifier stranding is optional so the FQ can occur in different positions. In Sportiche (1988) and Sportiche et al. (2016), the FQ is regarded as an adjunct, which functionally modifies its associated nominal phrase, as shown in (28). 6 Moreover, Shlonsky (1991) follows Sportichess (1988) proposal and further modifies Sportichess (1988) structure for the constituent containing an FQ and its associated nominal phrase. Shlonsky (1991) claims that the FQ is the head of a Quantifier phrase and the associated nominal phrase is taken as the complement of this Quantifier phrase. This thesis will take Shlonskyss (1991) structure as the main account for Stranding Analysis since if the associated nominal phrase is a complement, then the movement of an associated nominal phrase can be accounted for without effort, as illustrated in (29). A similar point of view is also mentioned in Cirillo (2009).. 6. According to Sportichess (1988) and Sportiche et al.ss (2016) structure, the movement of the associated nominal phrase, which is not a complement, seems to be an improper movement. 26.

(33) (28). (29) QP. DP. Qs all. DP Q. the children. DP. all the children. Stranding Analysis seems successfully to account for the floating behavior of the FQ and the similar reading shared by the sentences in (27). Nevertheless, it has also received some criticisms, and I will show them in the following two sub-sections.. 2.1.1.1 Advantages of Stranding Analysis. The first advantage of Stranding Analysis is providing an appealing piece of evidence for the VP-internal Subject hypothesis (Bobaljik 200c, Cirillo 2009). That is, the FQs are able to reveal the base position of the subject, spec of vP, and mark the positions that the subject has passed through, as shown in (c0) and (c1). Regarding the positions where the FQs can occupy, the subject has passed through all the spec positions between the spec vP and its surface landing site (see (c1)), as noted by Sportiche (1988) and Sportiche et al. (2016).. 27.

(34) (c0) [DP the children] will [VP [DP all [DP the children] leave] (c1) (All) the children (all) may (all) have (all) been (all) watching the movie.. Another advantage of Stranding Analysis is that it sheds light on the quantifying relationship between the FQ and its associated nominal phrase in (2ca-c). Sportiche (1988:427) claims ‘Qs may appear in [D]P-initial positions, which indicates that the relationship between an FQ and its associated nominal phrase has been established when they are situated in the base position as a constituent. Still another advantage is that the anaphoric properties of the FQ all in (24)-(25) can be accounted for. Sportiche (1988) has noted that after the associated nominal phrase is moved away, it leaves a trace behind, which has anaphoric properties. Since the FQ is adjacent to the anaphoric trace of the associated nominal phrase, the FQ seems to have anaphoric properties.. 28.

(35) 2.1.1.2 Disadvantages of Stranding Analysis. In the literature, four criticisms on Stranding Analysis have been pointed out. The first crucial issue, noted by Sportiche (1988), Bobaljik (1995, 200c), Doetjes (1997), Fitzpatrick (2006), Cirillo (2009) etc., is that under Stranding Analysis some ungrammatical sentences would be generated, as illustrated by the following sentences in (c2):. (c2) a. *The students have arrived all t . i. i. b. *The students were seen all t . i. i. It is interesting that FQs in (c2) cannot be stranded in their base positions where they should be under Stranding Analysis. More specifically, given that the surface subjects of unaccusative and passive sentences are moved from object positions, if the associated nominal phrases the students in (c2) are moved to the subject positions, then it is expected that the FQs all can be stranded in the object positions. However, this predication is not borne out, as shown by the unacceptability of all in the sentence-final position in (c2). To deal with this problem, Sportiche (1988) is forced to propose an unwise solution that the surface subjects of unaccusative and passive sentences are 29.

(36) base-generated, which is contrary to currently predominant hypotheses on unaccusatives and passives. In addition to the unexpected examples in (c2), Stranding Analysis also cannot rule out some ungrammatical sentences. Consider (cc)-(c5). Sportiche (1988) states that the universal quantifier all can float in lower positions, as shown in (cc), while Cirillo (2009) claims that all cannot float in lower positions, as shown in (c4). Fitzpatrick (2006) is on the side of Cirillo (2009), claiming that the universal quantifier all cannot occur lower than the passive ‘beings in example (c5).. (cc) The carpets will have been being all dusted for two hours. (c4) *Y? The patients may have been being all examined. (c5) The veggies will have been *(all) being *Y?(all) roasted. (Fitzpatrick 2006:48). c0.

(37) The ungrammatical examples in (cc)-(c5) pose a problem for Stranding Analysis since all should be able to be stranded in spec of vP and spec of progressive under Stranding Analysis. To tackle the unexpected examples, Cirillo (2009), who is a proponent of Stranding Analysis, only unconvincingly attributes this problem to morphological and phonological constraints. It seems that Stranding Analysis still cannot convincingly account for these ungrammatical sentences. Another striking problem is that the FQ and its associated nominal phrase cannot form a natural sequence (see (c6)-(c7)), as noted by Bobaljik (1995, 200c) and Fitzpatrick (2006). Under Stranding Analysis, the pair, the FQ all and the nominal phrase Larry, -arryl and -arren, in (c6) and the other pair, FQ all and the nominal phrase scme (cf the), in (c7) should form a constituent underlyingly. However, the mismatches between the FQs and their associated nominal phrases in (c6b) and (c7b) pose a serious challenge to Stranding Analysis.. (c6) a. Larry, Darryl and Darren have all come into the café. b. ?*All (of) Larry, Darryl and Darren have come into the café. (c7) a. Some (of the) students might all have left in one car. b. *All (of) some (of the) student might have left in one car. (Carden 1976) c1.

(38) This problem is attributed to the morphological and phonological constraints by Sportiche (1988) and Cirillo (2009). Sepcifically, Cirillo (2009) only addresses the mismatch in (c6b) (and not ex. (c7)) and suggests that the universal quantifier all can select conjoined nominal phrases, and if the first conjoined nominal phrase is singular, all and the whole conjoined nominal phrase will be an anomalous sequence that cannot be spelled out (see (c6b)). Then, to get around the problem of the mismatch in (c6b), Cirillo is forced to claim that the FQ all in (c6a) is not a stranded quantifier but an adverb or an appositive. Admittedly, this problem is still unsolved under Stranding Analysis. Still another notable problem is that the sentence with a pre-nominal quantifier and the sentence with an FQ may not always have the same interpretation (Bobaljik 1995, 200c, Fitzpatrick 2006). Consider (c8)-(c9). According to Bobaljik (200c), the two sentences in each of the examples in (c8) and (c9) have different interpretations.. (c8) a. All lions, tigers and bears are scary. b. Lions, tigers and bears are all scary. (c9) a. All students, professors and clowns have come to the meeting. b. Students, professors and clowns have all come to the meeting.. c2.

(39) Example (c8a) has a salient meaning that every lion, every tiger and every bear is scary. (c8b) has an identical reading as (c8a), but it has one more reading that lions, tigers and bears are generally scary. Examples (c9a) and (c9b) have a similar contrast as well. The former means that every member of each of the three groups will come to the meeting while the interpretation of the latter is that each of the three groups will come to the meeting, but not all of the members of each group will come. Recall that one of the important motivations for Stranding Analysis is that the sentence with a pre-nominal quantifier and the sentence with an FQ have the same interpretation. However, examples in (c8)-(c9) would weaken this argument. The issue whether the sentence with a pre-nominal quantifier and the sentence with an FQ have the same interpretation has been widely discussed and it is a prima facie challenge for Stranding Analysis. Last but not least, Bobaljik (1995, 200c) and Fitzpatrick (2006) point out that Stranding Analysis cannot explain the fact that of English dialects, the associated nominal phrases can only undergo A movement when the FQs are left behind, as shown in (40).. cc.

(40) (40) a. *The professors who Taylor will have all met ___ before the end of term. b. *These professors, Taylor will have all met ___ before the end of term. (Bobaljik 200c:15). The associated nominal phrases prcfesscr in (40) undergoes As movement. More precisely, the nominal phrase in (40a) is relativized and that in (40b) is topicalized. As seen, the FQs in (40) are unable to refer to the associated nominal phrases the prcfesscrs, which undergo As movement. However, under Stranding Analysis, (40) should be acceptable since the associated nominal phrase is moved away from the constituent containing the FQ and its associated nominal phrase like A movement in (27). It is unexpected that the quantifying relationship between the FQ all and its associated nominal phrase the prcfesscr, cannot be established in (40). Admittedly, Stranding Analysis cannot explain the fact that the associated nominal phrase cannot undergo As movement but only A movement.7. 7. Aside from the four criticisms, Stranding Analysis cannot account for the rightward and leftward FQs in French neatly (4ca-b). As discussed in Bobaljik (1995, 200c), Doetjes (1997), Fitzpatrick (2006), (4ca) is a typical example for Stranding Analysis. Under Stranding Analysis, the FQ tcus ‘alls is stranded with the trace of the associated nominal phrase and the associated nominal phrase is moved to a higher position. However, it seems that the same analysis cannot account for (4cb) because the associated nominal phrase in (4cb) is the object les ‘thems. That is to say, there is no trace for the tcus ‘alls to adjoin. To tackle this problem, Sportiche (1988) claims that unlike the rightward FQ in (4ca), the leftward FQ in (4cb) cannot be accounted for by Stranding Analysis but by Quantifier Raising. Since the Quantifier Raising is not what this thesis focuses on, I will not provide further details of it. The most important issue is whether it is necessary to use two analyses to explain (4ca) and (4cb). c4.

(41) 2.1.2 The adverbial approach. Under Adverbial Analysis, FQs are analyzed as adverbs, and they are base-generated in adverbial positions, as proposed by Kayne (1981), Doetjes (1997) and Fitzpatrick (2006). These adverbial analyses share some characteristics. First, there is no transformational relationship between a sentence with an FQ in (41a) and a sentence with a pre-nominal quantifier in (41b), namely that the FQ and its associated nominal phrase do not form a constituent at any level of representation. Second, the FQs like adverbs adjoin to certain maximal projections like auxiliaries.. (41) a. The carpets will all have been being dusted for two hours. b. All the carpets will have been being dusted for two hours.. In terms of the anaphoric relationship between the FQ and its associated nominal phrase, Kanye (1981) and Doetjes (1997) have different proposals. Kayne (1981) states that the FQ itself is an anaphor and that the anaphoric relationship between the FQ and its associated nominal phrase is due to the nature of an anaphoric element. Doetjes (1997:206) proposes that the anaphoric relationship is held between the trace of the associated nominal phrase and the associated nominal phrase itself.. c5.

(42) Under Doetjess analysis the prc adjoined by an FQ must bind the trace of the associated nominal phrase, as shown in the configuration of (42).. (42) [XP [[QP FQ. [DP prc]]i [XP…eci…]]. Furthermore, it is notable that Doetjes (1997) argues that her analysis can capture both the rightward and leftward FQs in French in a unified way whereas Kayness (1981) analysis cannot. In Kayne (1975), the rightward and leftward FQs in French have already been treated as two distinct processes. Kayness adverbial approach cannot account for the leftward FQ in French since the FQ, which has anaphoric nature, cannot be bound by its associated nominal phrase, as shown in (4c).. (4c) a. Les the. enfants. ont. tous. dormi.. children have all. slept. ‘The children have all slept.s b. Jsai I-have. tous. voulu. les. voir.. all. wanted. them. see. ‘I want to see them all.s c6.

(43) In (4ca), the associated nominal phrase les enfants ‘the childrens can c-command the anaphoric FQ tcus ‘alls while in (4cb), the anaphoric FQ tcus ‘alls cannot be c-commanded by its associated nominal phrase les ‘thems since the position of the FQ is higher than that of the associated nominal phrase. However, (4cb) is a grammatical sentence. To deal with the unexpected example (4cb), Kayne (1975) claims that (4ca) and (4cb) in French are derived through different processes, which means that Kayness (1981) analysis cannot capture the FQ in French in a unified way. Nevertheless, Doetjess (1997) adverbial approach can account for the rightward and the leftward FQ in French in a unified way, as shown in (44a-b).. (44) a. Les the. enfantsi ont. [[QP tous [DP prc]]i. children have. ti. all. dormi. slept. ‘The children have all slept.s b. Jsai I-have. [[QP tous [DP prc]]i voulu all. anted. ‘I want to see them all.s. c7. lesi. voir. them. see. ti..

(44) As seen in (44a-b), the FQ tcu ‘alls containing a empty category prc can bind the traces of the associated nominal phrases les enfants ‘the childrens and les ‘thems and thus the quantifying relationship between the FQ and its associated nominal phrase can be established. Given these facts, the significant difference between Kayness (1981) and Doetjess (1997) adverbial approach is that the former cannot account for the leftward FQ in French but the latter can. Adverbial Analysis proposed by Kayne (1981) and Doetjes (1997) still can explain the floating behaviors of the FQ and the anaphoric relationship between the FQ and its associated nominal phrase. Nevertheless, it also has pros and cons, as shown in following sub-sections.. 2.1.2.1 Advantages of Adverbial Analysis. One strength of Adverbial Analysis is that the mismatches between the FQ and the associated nominal phrase (see ex.(c6)-(c7)), and the different interpretations between the sentence with an FQ and the sentence with a pre-nominal quantifier (see ex.(c8)-(c9)) would not pose a challenge to Adverbial Analysis since the sentence with an FQ and the sentence with a pre-nominal quantifier are independent sentences, not derived from the same underlying structure. c8.

(45) Another strength of Adverbial Analysis is that it can explain three unexpected sentences under Stranding Analysis, as shown in (45a-b). The unacceptability of (45a-b) is because the FQ is not allowed in the sentence final positions. Different kinds of adverbs have different distributions. For example, a frequency adverb cannot occur at the end of a sentence, as shown in (45c).. (45) a. *The students have arrived all. b. *The students were seen all. c. *Those students fail in exams always.. Thus, it is possible to claim that FQs are a kind of adverb which cannot occur in the end of a sentence like a frequency adverb. If the FQ is some kind of adverb and its distribution does not include the final positions like always in (45c), then the ungrammaticality of (45a-b) naturally follows. Moreover, Adverbial Analysis can also deal with the unacceptability of the sentences with As movement (see ex.(40)), which cannot be accounted for under Stranding Analysis. Based on Kayness (1981) and Doetjess (1997) analyese, there are two different accounts for the unacceptability of the sentences with As movement in (40). Under Kayness (1981) analysis the FQ all in (40a-b) as an anaphoric element c9.

(46) will choose the closest argument Taylcr as its antecedant rather than the other argument the prcfesscrs in the As position. However, the argument Taylcr cannot satisfy the requirement of plurarity of the quantifier all. Therefore, (40a-b) are unacceptable. Moreover, regarding Doetjess (1997:207) claim that the FQ can trigger principle C violation, Bobaljik (200c) and Fitzpatrick (2007) point out that As extraction across the FQ will raise a strong crossover violation while the A movement should be unproblematic. Take (40a-b) for instance. Because the associated nominal phrase the prcfesscrs is moved across the FQ all to As position, principle C is violated. So far, it seems that Adverbial Analysis can successfully rule out the ungrammatical sentences in (40), which are predicted to be legitimate under Stranding Analysis.. 2.1.2.2 Disadvantages of the adverbial approach. The first noteworthy problem of Adverbial Analysis is that it has little to say about what type of adverb an FQ is (Sportiche 1988, Cirillo 2009). Sportiche (1988) and Cirillo (2009) both compare three different types of adverbs with FQ: manner adverbs, sentential adverbs and subject-oriented adverbs. Consider examples of the manner adverb and the FQ in (46)-(47) first. 40.

(47) (46) a. The students have carefully and thoroughly read the book. b. * The students have carefully thoroughly read the book. c. * The students have carefully and all read the book. d. The students all carefully read the book.. The FQs cannot be categorized as manner adverbs since the FQs do not have the properties of the manner adverb, as demonstrated by the contrast between (46c-d). Contrast between (46a-b) shows that if two or more manner adverbs occur together, they must be conjoined; however, the universal quantifier all cannot be conjoined with the other manner adverb carefully. Furthermore, consider (47).. (47) a. The students have been careful to read the book well in advance of the test. b. * The students have been all to read the book well in advance of the test.. The manner adverbs can be converted into adjectival forms (see ex. (47a) but the FQ all cannot (see ex. (47b)). With these facts, Cirillo (2009) claims that FQ does not belong to the manner adverbial type. Sportiche (1988) also mentions that the function of manner adverbs is to modify the action of a sentence, and hence, manner adverbs. 41.

(48) adjoin to V or VP; however, FQs seem not to be the case. Next, I shall compare the FQ and the sentential adverb in (48)-(49).. (48) a. The students have probably all read the book. b. ? The students have all probably read the book (49) a. It is probable that the students have read the book. b. * It is all that the students have read the book.. Unlike sentential adverbs, if the FQ all precedes a sentential adverb, the sentence is not preferred, as shown in (48a-b). What is more, sentential adverbs can be paraphrased as (49a), but the paraphrasing is not allowed for the FQs (see (49b)). Therefore, Cirillo (2009) suggests that the FQs do not belong to the sentential adverbial type. Sportiche (1988) states that the FQs and sentential adverbs appear to modify different objects: an FQ modifies nominal phrases while a sentential adverb modifies a whole sentence. Then, consider the examples of the subject-oriented adverbs and the FQ in (50).. 42.

(49) (50) a. The students rudely and stupidly insulted the teacher who helped them. b. *The students rudely stupidly insulted the teacher who helped them. c. *The inventors have wisely and all withdraw their money. d. The inventors have wisely all withdraw their money. (Cirillo 2009:c). The FQ is different from the subject-oriented adverbs. As noted by Sportiche (1988), the behaviors of FQs are much more similar to that of subject-oriented adverbial type, since both are used to modify a nominal phrase; however, they are semantically different. Cirillo (2009) notes more obvious differences between FQs and subject-oriented adverbs. As can be seen in (50a-b), when two subject-oriented adverbs are adjacent, they should be conjoined, but when the FQ all and the subject-oriented adverb stupidly are put side by side, they cannot be conjoined, as presented in (50c-d). Given these facts, Sportiche (1988) and Cirillo (2009) argue that FQ and the three types of adverbs do not behave the same, and thus the FQ does not belong to these class of adverb. Here, Adverbial Analysis faces its first challenge. However, the supporters of Adverbial Analysis, Brisson (1995), Bobaljik (1995), and Fitzpatrick (2006) note that the distribution of FQ all is not a challenge for Adverbial Analysis. They further claim that first English FQ all has the same 4c.

(50) distribution as modal adverb easily, arguing against the criticism that the FQ does not have the nature of adverbs (see (51)).8 Consider the example in (51).. (51) a. The veggies (all) will (all) have (all) been ?*(all) being *(all) roasted. b. The veggies (easily) will (easily) have (easily) been ?*( easily) being *(easily) roasted.. In (51), the FQ all and the modal adverb easily can float among auxiliaries but both of them cannot occur lower than the functional verb been, which is against the argument that FQ does not have adverb nature since its distribution does not belong to any type of adverb. Moreover, like other adverbs, FQs can restrict the distribution of adverbs, which co-occur with the FQ in a sentence (Fitzpatrick 2006:50, c.f. Bobaljik 1995), as shown in the following sentences:9. 8. According to Fitzpatrick (2006:49), the modal adverb easily should be distinguished from the manner adverb easily, as shown in the following examples: (i) That bird could easily be a bald eagle, judging by its size. (modal easily) (ii) You could make this shot easily if you would just concentrate. (manner easily) 9 In Fitzpatrick (2006:50), he claims that an adverb can limit the distribution of the other adverbs in a sentence. Take the two subject-oriented adverbs, allegedly and willingly, for instance. (i) The students have allegedly willingly been being yelled at. (ii) *The students have willingly allegedly been being yelled at. As seen in (i-ii), willingly is not allowed to occur higher than allegedly. 44.

(51) (52) a. The gladiator bravely fought the lions. (Subject-oriented, Manner) b. The gladiator all bravely fought the lions. (Subject-oriented, Manner) c. The gladiator bravely all fought the lions. (Subject-oriented, *Manner). In (52a), the adverb bravely can yield the subject-oriented reading, indicating the brave gladiator, or the manner reading, the bravely fighting action. When all precedes bravely in (52b), the adverb bravely can have the reading of a subject-oriented adverb or that of a manner adverb. Now consider (52c). The adverb bravely is positioned above the FQ all. In this case, the adverb bravely can only be interpreted as the subject-oriented adverb, i.e.. This means that the FQ blocks the manner reading of the adverb bravely. In light of the above facts, Fitzpatrick (2006) asserts that this distributional evidence is consistent with the claim that the FQ is an adverbial element (c.f. Bobaljik 1995) since other adverbs also have the same behavior. Given these facts, Fitzpatrick (2006:52) further claims that ‘I have not (yet) suggested that these FQs belong to the adverb word class, instead, I propose that they appear in adjoined positions that are available to many types of adverbial adjuncts, including some adverbss. Therefore, for Fitzpatrick, FQs are a type of adverbial adjuncts which may not belong to the class of adverbs just like adjuncts of other. 45.

(52) categories as in (5cb-d). In sum, I agree with Fitzpatrick that even though an FQ cannot be categorized into any type of adverbs, it has the nature of an adverbial adjunct.. (5c) a. The team won the game easily. b. The team will play the day after tomorrow. c. The team will play in the morning. d. The team will play when they have passed a drug screening.. Another well-known problem of Adverbial Analysis is that it cannot account for the anaphoric relation between an FQ and its associated nominal phrase directly, as noted by Sportiche (1988) and Cirillo (2009). Unlike under Stranding Analysis, the FQ and its associated nominal phrase under Adverbial Analysis are not a constituent underlyingly. To tackle this problem, Doetjes (1997:206) proposes that the anaphoric relationship is held between the trace of the associated nominal phrase and the associated nominal phrase itself. Also, Kayne (1981) suggests that the FQ itself is an anaphor, which can refer to its associated nominal phrase.. 46.

(53) 2.1.3 Summary. Either Stranding Analysis or Adverbial Analysis has its pros and cons. Sometimes, the weaknesses of one approach are the strengths of the other approach. In (54)-(55), I summarize the advantages and disadvantages of Stranding Analysis and Adverbial Analysis.. (54) Stranding Analysis Advantages a. Supporting VP-internal subject hypothesis b. Accounting for the anaphorical relationship between an FQ and its associated nominal phrase c. Accounting for the anaphoric properties of the FQ Disadvantages d. Wrong predications for unaccusatives and passives e. The mismatch between the FQ and its associated nominal phrase f. Different interpretations between the sentence with a pre-nominal quantifier all and the sentence with an FQ all e. Wrong predications for the As movement of the associated nominal phrase. 47.

(54) (55) Adverbial Analysis Advantages a. Accounting for the mismatch between the FQ and its associated nominal phrase b. Different interpretations between the sentence with a pre-nominal quantifier all and the sentence with an FQ all c. Ruling out the ungrammatical unaccusatives and passives d. Ruling out the ungrammatical sentences with As movement Disadvantages e. Not belonging to the class of adverbs f. Indirectly accounting for the anaphorical relationship between an FQ and its associated nominal phrase. Admittedly, the challenges that Stranding Analysis faces are more critical than those for Adverbial Analysis. Recall that under Stranding Analysis, the sentence with an FQ and the sentences with a pre-nominal quantifier have the same underlying structure so they have a similar interpretation. However, the different readings of the sentences in (c8)-(c9) make us doubt the claim that the FQ and the pre-nominal quantifier are derived from the same underlying structure. Furthermore, the mismatch between the FQ and its associated nominal phrase in (c6)-(c7) also weaken Stranding 48.

(55) Analysis since under Stranding Analysis, the FQ and its associated nominal phrase form a constituent underlyingly. In this thesis, I argue in favor for Adverbial Analysis for Chinese FQs since the behaviors of Chinese FQ at issue are similar to the FQ all in English.. 2.2 The numeral FQs in Japanese. The FQs in Japanese have received extensive discussions. The FQs in Japanese can be either numeral quantifier likes sanhnin ‘three peoples in (56b) (Fujita 1994, Kobuchi 2004, 2007) or non-numeral quantifier like subete ‘alls in (57b) (Huang and Ochi 2014).10 The former has attracted much more attention than the latter. This section mainly discusses the numeral quantifier in Japanese since it shares some intriguing properties with the PNQ and the FQ in Chinese.. (56) Numeral quantifier a. Kinoo yesterday. 10. san-nin-no. kasyu-ga. three-CL-GEN singer-NOM. utat-ta. sing-PAST. In the literature, the two sentences in (57) are translated in the same way. 49.

(56) b. Kasyu-ga. kinoo. singer-NOM. san-nin. utat-ta.. yesterday three-CL. sing-PAST. ‘Three singers sang yesterday.s (Kimilo 2006) (57) Non-numeral quantifier a. Taro-wa. subete-no gyooza-o. Taro-TOP all-GEN b. Taro-wa. tabe-ta.. dumpling-ACC. gyooza-o. Taro-TOP dumpling-Ac. eat-PAST. subete tabe-ta. all. eat-PAST. ‘Taro ate all the dumplings.s (Huang and Ochi 2014). Like English FQ all, the numeral quantifier in Japanese has mainly received two approaches: Stranding Analysis (Miyagawa 1989, Miyagawa and Arikawa 2007) and Adverbial Analysis (Fujita 1994, Downing 1996, Kobuchi 2004, 2007). According to Miyagawass (1989) stranding account, the relationship between the FQ and its associated nominal phrase is established on the c-commanding requirement.. 50.

(57) (58) Mutual c-command requirement: The NP or its trace and the numeral or its trace must c-command each other. (Miyagawa 1989). Miyagawa and Arikawa (2007) modify Miyagawass (1989) stranding account and claim that the FQ and its associated nominal phrase form a constituent, which can sustain the requirement in (58).11 As shown in (59), the associated nominal phrase kasyu ‘singers and the numeral quantifier sanhnin ‘three peoples are a constituent before the associated nominal phrase is moved away and crosses the adjunct kincc ‘yesterdays. At the same time, the numeral quantifier sanhnin ‘three peoples and the trace of the associated nominal phrase kasyu ‘singers mutually c-command and then the anaphoric relationship between them is established directly.. (59) Kasyu-gai. [VP kinoo. singer-NOM. [ti san-nin]. yesterday. three-CL. utat-ta]. sing-PAST. ‘Three singers sang yesterday.s. 11. The structure Miyagawass (1989) Stranding Analysis adopts is still the ternary branching structure, and therefore, Miyagawa and Arikawa (2007) adjusts the ternary branching struture to binary branching structure. 51.

(58) The stranding approach to Japanese numeral FQs is also argued for by Fitzpatrick (2007). According to him, in some languages the FQ is adnominal and should be accouted for by Stranding Analysis like Japanese FQ, but in some other languages the FQ is adverbial and should be accounted for by Adverbial Aanalysis like English and French. The other approach to the Japanese numeral FQ is Adverbial Analysis. As aforementioned, the essence of Adverbial Analysis is that the numeral FQ is not base-generated in adnominal positions but is an adjunct like an adverb. The main point is that under adverbial account, the sentence with an FQ in (60) and the sentence with an adnominal quantifier in (61) have no derivational relationship. Regarding the quantifying relationship between the numeral FQ and its associated nominal phrase, the proponents of Adverbial Analysis provide different suggestions; for example Fujita (1994) proposes head-feature licensing, Kobuchi (2004, 2007) follows Doetjess (1997) adverbial account, and so forth. I will not further discuss the advantages and disadvantages of each approach to Japanese. Nevertheless, I notice that like FQ all in English, Chinese FQs at issue can only have the subject-oriented reading but Japanese numeral FQs can have subject-oriented or object-oriented reading (Fujita 1994).. 52.

(59) Here, I will discuss an intriguing property of Japanese FQ which is also shared by Chinese FQ at issue: the two possible readings of an FQ. According to Fujita (1994), Downing (1996), Kobuchi (2004, 2007), the Japanese FQ may have two different readings: exhaustive reading and partitive reading, as shown in (60)-(61) while the adnominal quantifier can only have an exhaustive reading.. (60) Japanese FQ a. Yakamashii noisy. gakusei-ga. go-nin ki-ta.. student-NOM. 5-CL. come-PAST. ‘Five of the noisy students came.s b. Gakusei-ga. go-nin ki-ta.. student-NOM. 5-CL. come-PAST. ‘(The) five students came.s (Fujita 1994) (61) Adnominal quantifier in Japanese Yakamashii. go-nin-no. gakusei-ga. ki-ta.. noisy. 5-CL-GEN student-NOM. ‘(The) five noisy students came.s. 5c. come-PAST.

(60) Example (60a-b) show that the FQ gchnin ‘five CLs in Japanese can have either an exhaustive reading, all the five people, or a partitive reading, five of the students, while the adnominal quantifier in (61) only allows an exhaustive reading. One more noteworthy point about the Japanese FQ is that Fujita (1994) demonstrates that partitivity is a characteristic of the FQ construction, although it is not an inherent property of the construction. More precisely, the two interpretations rely on different contexts instead of the FQ itself. That is, if in the context, there exists a relevant set which is able to be compared with the FQ, the FQ would prefer the partitive reading. If the relevant set does not exist, then the FQ can only have an exhaustive reading, as shown in (60a-b), respectively. Consequently, in Japanese, the FQ with an exhaustive reading usually goes along with an indefinite associated nominal phrase. This characteristic of Japanese FQs is also used to argue against Stranding Analysis (Fujita 1994, Kobuchi 2004, 2007) since the fundamental motivation for the derivational approach is that the sentence with an FQ and the sentence with an adnominal quantifier are underlyingly the same and these two sentences should have an identical reading. In sum, firstly, in light of the FQs in Japanese, it is not surprising that a language has both numeral and non-numeral floating quantifiers so does Chinese. Secondly, in Japanese, an FQ can have an exhaustive reading or a partitive reading depending on 54.

(61) the context, as shown in (60). I also observe that the Chinese FQ at issue also shares a similar property.. 2.3 Previous analyses of the Chinese facts. The PNQ and the FQ in Chinese have not received much attention in the literature. In this section, I will review the previous studies of them. The first study is Shin (2008), in which the numeral FQ is considered as an argument; secondly, in Paul and Whitman (2010), the non-numeral FQ is deemed as an adverb in DOC; lastly, Kuo (2016) discusses the non-numeral FQ together with the non-numeral PNQ. Following Sportiche (1988), Kuo (2016) claims that the sentence with post-nominal quantifier and the sentence with an FQ have a derivational relationship. In this section, I will discuss Shinss (2008), Paul and Whitmanss (2010) and Kuoss (2016) studies in order.. 2.3.1 Shin (2008). Shin (2008) discusses the phenomenon that the numeral phrase and its associated nominal phrase are not adjacent in Chinese. Based on the position where the numeral phrase is placed and the reading of the numeral phrase (exhaustive or partitive), Shin categorizes the phenomenon into four types, as shown in table 1, and 55.

(62) the examples of each type are shown in (62)-(67). He notes that type D must have a modal ycu. Of the four types, only some sentences of type A are relevant to the FQs at issue.. Table 1. The four types of numeral phrase in Shin (2008) Type. Numeral phrase. Reading. Example. A. preverbal. exhaustive. (62)-(64). B. postverbal. exhaustive. (65). C. postverbal. partitive. (66). D. preverbal. partitive- including ycu ‘exists. (67). (62) Zhangsan song lai. de. Zhangsan send come. shu,. san-ben dou. DE book. hen. youqu.. three-CL all very. interesting. ‘The three books sent by Zhangsan are interesting.s (6c) Tamen they. keneng. san-ge. likely. three-CL person. ren. dou qu. Taibei. le.. all go. Taipei. ASP. ‘The three people might all go to Taipei.s. 56.

(63) (64) Cong from. tushuguan. jie. lai. de. library. borrow. come. DE book. dou. kan. all. read finish-ASP. shu,. wo I. san-ben three-CL. wan-le.. ‘I have read all the three books borrowed from the library.s (65) Pingguo, Zhangsan mai-le apple. san-ge.. Zhangsan buy-ASP. three-CL. ‘Zhangsan bought three apples.s (66) Na-xie. shu,. that-CL book. Zhangsan. kan-le. Zhangsan read-ASP. san-ben. three-CL. ‘Zhangsan read three of those books.s (67) Na-xie. shu. that-CL book. you. san-ben wo kan-le.. YOU. three-CL I. read-ASP. ‘I read three of those books.s. According to Shin (2008), the difference between type B and type C is their readings, that is, the former only yield an exhaustive reading and the latter yield only a partitive reading. Besides, the difference among type A, B and C is that type A should precede verbs but type B and C should follow verbs. 57.

(64) Even though in the above examples, the numeral phrases and their associated nominal phrases are not adjacent, there exists a quantifying relationship between them. To deal with the phenomenon, Shin (2008) proposes Argument Analysis to account for all the four types above. Under Shinss analysis, the associated nominal phrase and the numeral phrase are all base-generated and the numeral phrase form a constituent with an empty category, as shown in (68).. (68) Zhangsan Zhangsan. jie. lai de. shu. san-ben. dou. borrow come DE book three-CL DOU. hen very. youqu. interesting. ‘The three books borrowed by Zhangsan are very ineresting.s TopP Zhangsani …. vP [san-ben ei] j. vP. DOU. vP tj. VP hen. youqu. (Shin 2008:54). To illustrate, in (68), the associated nominal phrase Zhangsan in the sentence occupies the topic position, called the dangling topic. Moreover, the position the 58.

(65) numeral phrase sanhben ‘three CLs occupies is an argument position.12 Then, Shin further claims that since the numeral phrase, placed in the argument position, contains an empty category, the quantifying relationship between the numeral phrase and its associated nominal phrase can be established via the empty category. Shin (2008) claims that Stranding Analysis cannot account for why the associated nominal phrase can be moved out of a complex NP island, as shown in (69), but his Argument Analysis can.. (69) a. Wo I lai. haimei. peng. yet.not. encounter arrive. de. shu. come DE book. 12. dao. [san-ben. cong. three-CL from. tushuguan. jie. library. borrow. dou. kan wan. de. ren].. DOU. read finish. DE person. In Shin (2008), he does not specifically discuss the position of the numeral phrase in the type C. Nevertheless, according to his claim, ‘he will consider all the numeral phrase of this phenomenon as the real argument of the sentences, it seems that the numeral phrase in type C is placed in the object position. 59.

(66) b. Cong. tushuguan. jie. from. library. borrow come DE book. dao. [[san-ben ti]. dou. kan. three-CL. DOU. read finish. arrive. lai. de. shui,. wan. wo haimei. peng. I. encounter. not.yet. de ren ]. DE person. Lit:‘I have not encountered anyone who has already read the three books borrowed from the library.s (Shin 2008:52). In (69a), the bracketed phrase is a complex NP island. Any element moves out of the island would give rise to island effects. Under Stranding Analysis, the numeral phrase sanhben ‘three CLs and the associated nominal phrase ccng tushuguan jiehlai de shu ‘the books borrowed from the librarys would be underlingly base-generated in the more inclusive bracketed phrase in (69b), as a constituent, so the movement of the associated nominal phrase ccng tushuguan jiehlai de shu ‘the books borrowed from the librarys, as shown in (69b) should be impossible. However, this prediction is not borne out. Therefore, Shin (2008) claims that this is a significant challenge for Stranding Analysis. According to him, on the contrary, under Argument Analysis, (69b) can be easily explained. The numeral phrase and the associated nominal phrase 60.