中文驢子句與光桿條件句之第一語言習得

全文

(1)國立臺灣師範大學英語學系碩. 士. 論. 文. Master’s Thesis Department of English National Taiwan Normal University. 中文驢子句與光桿條件句之第一語言習得. L1 Acquisition of Donkey Sentences and Bare Conditionals in Mandarin Chinese. 指導教授：陳純音博士 Advisor: Dr. Chun-yin Doris Chen 研究生：林韋伶 Student: Wei-ling Eileen Lin. 中華民國一百零七年六月 June 2018.

(2) 摘要本研究旨在探討，中文為母語之兒童對於量化相關句構之詮釋與發展。探討內容為兒童第一語言習得之過程中，對中文驢子句與光桿條件句詮釋的差異，以及影響其詮釋之因素。本研究包含兩階段的詮釋測驗：第一階段是單句測驗，觀察兒童對量化相關句構之理解差異；第二階段則是加入語境，用以了解兒童是否會因而影響其解讀。研究對象依年齡與年級共分為四組：幼兒園大班、小學二年級、四年級、成人，每組皆為十八人。研究結果顯示，該兩種句構的詮釋隨著年齡的增長，解讀能力逐漸與成人之詮釋一致。首先，對於這兩種句構的比較，每組兒童皆能在解讀上，顯示兩種句構並無相關性，而在這兩種句構的解讀方面，越高年級越能掌握到其意涵。在驢子句裡，量化詞對解讀有深遠的影響，但各個量化詞影響語句詮釋有不同的趨勢：「每」的意涵最為明確，因此每組孩童皆能容易解讀其語句；「不是每個」對於幼兒園大班尚有難度，但小學二年級、四年級孩童已能有成人解讀能力；「有些」所表達的意涵最為模糊，實驗所收錄之孩童，在解讀上尚未完全成熟。另外，光桿條件句則是測試句構之對等性對解讀之影響，實驗結果顯示，每組皆在對等條件句上表現較佳，且皆在對等結構上和成人有相近的解讀能力；但在不對等條件句上，只有小學二年級、四年級才有和成人一樣的解讀能力。最後，探討語境的加入是否影響兒童對於此兩種句構上的解讀，實驗結果顯示，除了幼兒園大班對於語境加入並無太大差異外，其他組皆在語句之詮釋上，有著顯著的影響。國小二年級已可以詮釋在偏置語境下的驢子句，而國小四年級則可以完全解讀支持語境與偏置語境下的驢子句，及支持語境下的光桿條件句。. 關鍵詞：驢子句、光桿條件句、第一語言習得、量化詞、中文. i.

(3) ABSTRACT The present study investigated children’s first language acquisition of donkey sentences and bare conditionals in Mandarin Chinese, which are both concerned with quantification. Aiming to discover a developmental pattern of acquisition, this study explored children’s knowledge of quantification and how each construction affects their readings by testing their interpretations of the two constructions from an empirical perspective. Four issues regarding the two constructions were taken into account, which were the construction-related factor, construction-specific factors, contextual effects, and age effects. Kindergarten, Grade 2 and Grade 4 were recruited as experimental groups, and adults as a control group to compare their interpretations, each of which consisted of eighteen subjects. Every subject finished two phases of tasks, which were sentences in isolation and sentences in context. In both phases of the experiments, the subjects were asked to determine which picture best described the target sentence to test their interpretation, a universal or existential reading. The results of this research identified a developmental pattern of the acquisition of donkey sentences and bare conditionals in Mandarin Chinese. It was found that overally, children under seven years old had difficulty interpreting quantificational sentences. First, concerning the relatedness of the two constructions, all the four groups showed a unanimous tendency in that bare conditionals were not in the same vein to donkey sentences in terms of interpretations where the latter was easier to interpret. In addition, quantifier types of donkey sentences are. ii.

(4) vital to interpretations, where the quantifier mei ‘every’ was already acquired by children as young as KS, bushi meige ‘not every’ was interpreted in an adult-like manner by Grade 2, and youxie ‘some’ could not be obtained with an adult-like interpretation by any child group. This showed that mei ‘every’ was the easiest to acquire, followed by bushi meige ‘not every’ and lastly youxie ‘some.’ Another construction-specific factor, parallelism, was identified crucial to readings where all the child groups exhibited adult-like interpretations of parallel bare conditionals, but only Grade 2 and Grade 4 could have interpretations in an adult-like manner of nonparallel ones. This indicated that nonparallel sentences were more challenging to interpret than nonparallel ones. Moreover, with respect to contextual effects, children by the stage of Grade 2 could obtain adult-like interpretations of donkey sentences in biasing context, but it was not until they were at Grade 4 could they interpret both donkey sentences and bare conditionals in supporting context with adult-like readings. As a result, contextual effects were found, and the subjects’ interpretations were greatly affected by context but in different ways to the two constructions.. Keywords: donkey sentences, bare conditionals, first language acquisition, quantification, Mandarin Chinese. iii.

(5) ACKNOWLEDGEMENTS Reaching the end of the graduate program is certainly an important milestone of my life. Throughout my process of working on this thesis, it is without any doubt that I couldn’t have finished this journey without an amount of people’s assistance and encouragement; hence, I would like to thank them all. First and foremost, I would like to express my deepest appreciation to my advisor, Dr. Chun-yin Doris Chen, who radiates strong passion for academic research and teaching that amazed me. I enjoyed the delightful atmosphere when we discussed the contents of the thesis, where her comments and feedback often gave me enlightenment and many a-ha moments. I cannot overemphasize that I could hardly reach this milestone without her constant patience, support and tenderness. Under her genuine and professional guidance, I was often fascinated and enlightened by not only her academic advice but also her positive attitude towards life. I would like to thank my committee members, Dr. Jyun-gwang Fred Chen and Dr. Rueih-lirng Sharon Fahn, by alphabetical order. They offered me insightful suggestions and kindly provided me with directions to help shape my thesis in the way it is now. Their kind encouragement also led me motivated to push myself even further. It is with immense gratitude that I acknowledge the support and help of Ms. Yu-mei May Zou, who kindly helped me arrange the subject recruitment in New Taipei Municipal BeiXin Elementary School. She kindly contacted the teachers for me in advance so that the whole process of subject recruitment was smooth and efficient. As a result, the experiments of this thesis would not have been conducted successfully and smoothly but for her help. In addition, I owe my fervent gratitude to professors who had taught me in the linguistic program: Dr. Hsaio-hung Iris Wu, Dr. Hui-shan Nissa Lin, Dr. Jen Ting, Dr. Jen-i Li, Dr. Jing-lan Joy Wu, Dr. Li-hsin Ning, Dr. Miao-ling Hsieh and Dr. Shiao-hui Chan, by alphabetical order. I was always fascinated by the interesting and insightful instructions and. iv.

(6) lessons they had taught that brought me a whole new world of linguistics. I am indebted to my many classmates and friends in NTNU who supported me in these days of graduate program: Aries Cheng, Carol Lee, Eliza Cheng, Johnny Hu, Kelvinn Kuo, Louisa Zheng, Shawn Lin and Vivien Cheng, by alphabetical order, to name a few. In particular, special thanks go to Howard Su, Mark Tu and Stephanie Yeh. The time we spent together discussing schoolwork, participating in the work of research assistants and having fun during our leisure time (which was quite limited) were certainly one of the most invaluable things that happened in my life. I was especially moved by their caring when I was discouraged and by their generous help when my computer broke down one day before my final defense. Last but not least, this thesis would not have been possible without the unconditional love and support from my beloved family and close friends: my parents, my brother Colin, Elvis Tsai, and Sally Su. They supported me thoroughly when I pursued my goals, offered me beneficial suggestions whenever I encountered difficulties, and tenderly boosted my confidence when I felt down and lost. They are certainly the best shelter I could ever ask for; therefore, I would like to dedicate this thesis to them, who have helped me and accompanied me throughout this whole journey.. v.

(7) TABLE OF CONTENTS CHINESE ABSTRACT .................................................................................................................. i ENGLISH ABSTRACT ................................................................................................................. ii ACKNOWLEDGEMENTS ........................................................................................................... iv TABLE OF CONTENTS ............................................................................................................. vi LIST OF TABLES....................................................................................................................... ix LIST OF FIGURES ...................................................................................................................... x. Chapter One Introduction ................................................................................................... 1 1.1 Motivation .................................................................................................................... 1 1.2 Research Questions ...................................................................................................... 5 1.3 Significance of the Study ............................................................................................. 5 1.4 Organization of the Thesis ........................................................................................... 6. Chapter Two Literature Review.......................................................................................... 7 2.1 Relatedness of Donkey Sentences and Bare Conditionals ........................................... 7 2.2 Constraints on Interpretations of Donkey Sentences and Bare Conditionals ............ 11 2.2.1. Quantifier Types ........................................................................................... 12. 2.2.2. Parallelism .................................................................................................. 13. 2.3 Empirical Studies of Donkey Sentences .................................................................... 17 2.3.1. Geurts (2002) ............................................................................................... 17. 2.3.2. Foppolo (2009)............................................................................................. 20. 2.3.3. Grosz et al. (2014)........................................................................................ 24. 2.3.4. Summary of the Empirical Studies .............................................................. 28. 2.4 Summary of Chapter Two .......................................................................................... 31. vi.

(8) Chapter Three Research Design ........................................................................................ 32 3.1 Subjects ...................................................................................................................... 32 3.2 Materials and Methods ............................................................................................... 34 3.2.1. Donkey Sentences ........................................................................................ 36. 3.2.2. Bare Conditionals......................................................................................... 40. 3.3 Procedures .................................................................................................................. 43 3.4 Summary of Chapter Three ........................................................................................ 46. Chapter Four Results and Discussion ............................................................................... 48 4.1 Construction-related Factor: Relatedness of the Two Constructions ........................ 48 4.1.1. Overall Findings........................................................................................... 48. 4.1.2. Discussion .................................................................................................... 52. 4.2 Construction-specific Factors: Quantifier Types and Parallelism ............................. 55 4.2.1. Donkey Sentences: Quantifier Types........................................................... 55. 4.2.2. Bare Conditionals: Parallelism .................................................................... 61. 4.2.3. Discussion .................................................................................................... 64. 4.3 Contextual Effects ...................................................................................................... 67 4.3.1. Overall Findings........................................................................................... 68. 4.3.2. Discussion .................................................................................................... 73. 4.4 Age Effects ................................................................................................................. 76 4.5 Summary of Chapter Four ......................................................................................... 78. Chapter Five Conclusion .................................................................................................... 79 5.1 Major Findings ........................................................................................................... 79 5.2 Limitations of the Present Study and Suggestions for Future Research .................... 81. vii.

(9) BIBLIOGRAPHY...................................................................................................................... 83 APPENDIX A Test Items Used in Donkey Sentences in Isolation (DSII) ........................ 88 APPENDIX B. Test Items Used in Donkey Sentences in Context (DSIC) ........................ 91. APPENDIX C Test Items Used in Bare Conditionals in Isolation (BCII) ....................... 99 APPENDIX D Test Items Used in Bare Conditionals in Context (BCIC) ..................... 103 APPENDIX E. Test Items Used in Sentences in Isolation ................................................ 113. APPENDIX F Test Items Used in Sentences in Context ................................................. 121 APPENDIX G Consent Form ............................................................................................ 140. viii.

(10) LIST OF TABLES Table 2-1 Major Findings and Limitations of the Empirical Studies....................................... 29 Table 3-1 Information of Subjects ........................................................................................... 33 Table 3-2 The Structure of Donkey Sentences......................................................................... 37 Table 3-3 An Example of Donkey Sentences in Isolation ....................................................... 38 Table 3-4 An Example of Donkey Sentences in Context ......................................................... 39 Table 3-5 The Structure of Bare Conditionals ......................................................................... 41 Table 3-6 An Example of Bare Conditionals in Isolation ........................................................ 42 Table 3-7 An Example of Bare Conditionals in Context ......................................................... 42 Table 4-1 Each Group’s Interpretations of the Two Constructions ......................................... 50 Table 4-2 The between-group Differences in DS and BC ....................................................... 51 Table 4-3 Each Group’s Interpretations of the Three Quantifier Types .................................. 57 Table 4-4 A Revision of Each Group’s Interpretations of the Three Quantifier Types .......... 59 Table 4-5 The between-type Differences in the Four Groups ................................................. 59 Table 4-6 The between-group Differences in the Three Quantifier Types .............................. 60 Table 4-7 Each Group’s Interpretations of the Two Parallelism Types .................................. 63 Table 4-8 The between-group Differences in the Two Parallelism Types .............................. 64 Table 4-9 The Overall between-type Differences .................................................................... 69 Table 4-10 Each Group’s Interpretations of DS in the Three Contextual Types ..................... 69 Table 4-11 The between-group Differences for DS in the Three Contextual Types ............... 70 Table 4-12 Each Group’s Interpretations of BC in the Three Contextual Types ..................... 71 Table 4-13 The between-type Differences for BC in Each Group .......................................... 72 Table 4-14 The between-group Differences for BC in the Three Contextual Types ............... 72 Table 4-15 Children’s Acquisition of Donkey Sentences and Bare Conditionals ................... 76. ix.

(11) LIST OF FIGURES Figure 2-1 Every boy that stands next to a girl holds her hand. .............................................. 18 Figure 2-2 Every railway line that crosses a road goes over it. .............................................. 18 Figure 3-1 Procedures of the Experiments............................................................................... 44 Figure 4-1 Subjects’ Overall Interpretations of the Two Constructions .................................. 49 Figure 4-2 Subjects’ Overall Interpretations of Quantifier Types in Donkey Sentences ........ 56 Figure 4-3 Question 5 in the DSII for Youxie ‘Some’.............................................................. 58 Figure 4-4 Subjects’ Overall Interpretations of Parallelism Types in Bare Conditionals ....... 62 Figure 4-5 Subjects’ Overall Interpretations of the Contextual Types .................................... 68. x.

(12) Chapter One Introduction 1.1 Motivation When it comes to first language acquisition, the innateness hypothesis (Chomsky 1986, Cook 1988, Gopnik 1988, Keil 1989, Gelman & Wellman 1991, Gopnik & Meltzoff 1997) is one of the most crucial theories to date. This hypothesis argues that children are born with some knowledge or principles of language where children can process and acquire a language in a short period of time. Nevertheless, not all linguistic knowledge is perceived at the same time. It has been found that children can comprehend some knowledge easier and faster, and they learn other language devices on the basis of those easier ones. In other words, there is a sequence in mastering language, and acquisition of quantification is one of the cases that can be probed into. In the process of acquisition, it has been found that children had difficulties in comprehending quantificational sentences (Herburger 1997, Cohen 2001, Geurts 2003). For example, for a sentence, like “Every kid holds an umbrella,” children tend to interpret the sentence as “Every umbrella-holder is a kid.” A donkey construction, which gets its name by its pattern of construction, is one of the important constructions concerning quantification.. (1) Every farmer who owns a donkey beats it.. (Geurts 2002:129). Sentence (1), where the pronoun it refers back to the antecedent a donkey, can be interpreted 1.

(13) as “Every farmer who owns a donkey beats at least one of the donkeys that he has,” indicating there is at least one donkey in the universe involved; hence, it is an existential reading of the sentence. On the other hand, (1) can also be interpreted as “Every farmer who owns a donkey beats all of the donkeys that he has,” implying all of the donkeys are involved, yielding a universal reading. Both readings for (1) are reasonable, but it has been found that speakers tend to have a preference for a certain reading. For example, if a sentence is headed by a universal quantifier, like every, a universal reading is more likely to be favored, but if it is headed by an existential quantifier, like some, an existential reading tends to be chosen (Kanazawa 1994, Geurts 2002, Foppolo 2009). Like donkey sentences, Chinese conditional sentences as shown in (2)-(4) are also concerned with an interplay of logic between quantifiers and variable pronouns: (2) Shei xian lai, shei xian who come first who first ‘If X comes first, X eats first.’. chi. eat (Cheng & Huang 1996:127). (3) Ruguo you shei quiao men, ni jiu jiao ta jin-lai. if have who knock door you then ask him(her) come-in ‘If someone knocks on the door, you’ll ask him/her to come in.’ (Cheng & Huang 1996:142) (4) Ni jiao shei jin-lai, wo dou jian ta. you ask who come-in I all see him(her) ‘Whoever you ask to come in, I will see him/her.’. (Cheng & Huang 1996:142). Sentence (2), an example of a bare conditional, is different from ordinary conditional sentences such as (3) and (4) in that neither a conjunction like ruguo ‘if’ in the antecedent clause as in (3). 2.

(14) nor an adverb like dou ‘all’ in the consequent clause as in (4) is present. As seen in (2), the variable, the wh-word shei ‘who,’ in the antecedent clause, Shei xian lai ‘who comes first,’ is connected to shei ‘who’ in the consequent clause shei xian chi ‘who eats first,’ meaning that whoever comes first, he (she) eats first. There have been debates over whether Chinese bare conditionals can be regarded as one type of donkey sentences (Cheng & Huang 1996, Pan & Jiang 1997, Cheung 2007, Wang 2007) because the two constructions share referentiality in readings in common. In a donkey sentence like (1), the pronoun it refers back to the number of donkeys in the antecedent, and the sentence can obtain either an existential or a universal reading. In bare conditionals, as in (2), shei ‘who’ in the consequent clause refers back to shei ‘who’ in the antecedent clause. In other words, if shei ‘who’ is referred to Zhangsan in the antecedent clause, then shei ‘who’ in the consequent clause will be identical, and shei ‘who’ can be interpreted as “someone” or “everyone” with the sentence taking either an existential or a universal reading. Hence, through a comparison of readings (universal and existential readings) in the two constructions, the researcher would like to see whether and to what extent the two constructions are related. In addition, construction-specific factors will be examined in the present study. For example, for donkey sentences, quantifier types, which elicit different readings, are taken as one crucial factor in determining the readings of donkey sentences, as in (5)1.. Examples in (5) are taken from Krifka (1996:2), but some modification has been made in order to exhibit contrasting quantifiers. 1. 3.

(15) (5) a. Every farmer who owns a donkey beats it. (universal reading) b. Some farmer who owns a donkey beats it. (existential reading). (Krifka 1996:2). For (5a), it is believed that the universal quantifier every entails a universal reading of the sentence, while for (5b), the existential quantifier some elicits an existential reading; hence, this study probed into whether different quantifiers will influence different readings. Construction-specific factors that are influential to interpretations of donkey sentences and bare conditionals respectively are put in design in each construction to see whether the results can support analysis from previous studies. Moreover, according to Prévost & Paradis (2004), Foppolo (2009), and Ireri et al. (2012), context can affect interpretations of sentences involving the syntax-semantic interface. Since it has been found that children depend on context to infer meanings of sentences, context is believed to be influential to interpretations of quantification as well. Accordingly, the contextual effects on the two constructions will be discussed in this study. Furthermore, age has been a crucial issue in first language acquisition research (Assink, van Well & Knuijt 2003, Mayberry 2007). Hence, this study will also discuss this issue in order to see if there is a derivational pattern of acquisition in quantification of donkey sentences and bare conditionals. Since little literature (Crain et al. 2009) is concerned with L1 acquisition of donkey sentences and bare conditionals, through a comparison of children’s and adults’ interpretations of these sentences, the present study aims to provide clues to the relatedness of the two 4.

(16) constructions, and to the factors that may influence readings.. 1.2 Research Questions In order to investigate whether theoretical approaches are really the mechanism operating in children’s acquisition, the researcher will explore the following research questions: 1) Are bare conditionals related to donkey sentences in readings? 2) Do construction-specific factors show significant impacts on readings of donkey sentences and bare conditionals? 3) Is there a contextual effect on the donkey construction and the bare conditional construction? 4) At what age do children actually obtain an adult-like interpretation of donkey sentences and bare conditionals?. 1.3 Significance of the Study Previous studies (Kanazawa 1994, Cheng & Huang 1996, Krifka 1996, Pan & Jiang 1997, Cheung 2007, Wang 2007) have provided some generalizations and analyses of the interpretations of English donkey sentences and of Chinese bare conditionals respectively. Nevertheless, little literature (Crain et al. 2009) is concerned with empirical research on Chinese donkey sentences and bare conditionals, and exploring the relatedness in readings of the two constructions. In addition to the innovation of conducting an empirical study of the two constructions, 5.

(17) factors that are not regarded as influential in the literature (Yoon 1994, Krifka 1996, Foppolo 2009, Geurts 2002) are also taken into consideration. Construction-specific factors (such as quantifier types) are included respectively to test whether generalizations and analyses from previous studies can be supported or not. Therefore, this study hopes to offer some insights not only into the developmental pattern of quantification in the two constructions, but also into the relation between the two constructions and properties specific to each construction that literature has not yet been agreed upon.. 1.4 Organization of the Thesis The organization of the following chapters is as follows: Chapter Two discusses the theoretical literature concerning conditionals in Mandarin Chinese, which are analyses arguing whether Chinese conditionals could be regarded as one type of donkey sentences, and also empirical studies with respect to donkey sentences, which are concerned with interpretations of donkey pronouns. In addition, Chapter Three presents the research design, providing the recruitment of subjects, materials, procedures of the experiments. Chapter Four reports the results of the experiments, and discusses the analysis of children’s acquisition of the two constructions. Lastly, Chapter Five concludes this study with a presentation of the major findings and suggestions for future research.. 6.

(18) Chapter Two Literature Review This chapter provides a review of both theoretical and empirical literature. In order to investigate the relatedness of the donkey construction and the bare conditional construction and to discuss some construction-specific constraints of each construction, theoretical studies are discussed by issues (i.e. the relatedness of the two constructions and the construction-specific factors), while empirical studies are presented in chronological order. Accordingly, Section 2.1 discusses a construction-related constraint on quantification and readings of the two constructions, and Section 2.2 presents two construction-specific constraints affecting readings of donkey sentences and bare conditionals. Section 2.3 reviews empirical studies, and finally, Section 2.4 summarizes this chapter.. 2.1 Relatedness of Donkey Sentences and Bare Conditionals With respect to the relatedness of donkey sentences and bare conditionals, some argue that the two constructions are related (Cheng & Huang 1996, Pan & Jiang 1997) while some deny the relatedness (Wang 2007). In an English donkey sentence like (1), two approaches are employed to explain the donkey pronoun it, which is anaphoric to a donkey. (1) Every farmer who owns a donkey beats it.. (Geurts 2002:129). The first approach considers it to be an E-type pronoun where “E” stands for. 7.

(19) “existential,” meaning the pronoun it expresses an existential property of the antecedent a donkey, yielding a meaning that at least one of the donkeys (but not necessarily all the donkeys) owned by a farmer is beaten by him. The other approach is the unselective binding approach, which indicates the existence of an implicit operator (called the necessity operator, NEC) that unselectively binds two anaphoric nouns (it and a donkey). This operator serves as a binder and licenser that not only binds the two variables in the sentence and also licenses a quantificational force for the variables, as can be seen in (2). (2) NEC [x, y] [farmer(x) and donkey(y) and own(x, y)  beat(x, y)] (Pan & Jiang 1997:9)1 Hence, under this approach, for (1), both the pronoun it and the antecedent a donkey are bound by the operator (NEC), which is the quantifier every that gives quantificational force of the variables. Therefore, the operator binding results in ambiguity in readings: an existential reading (i.e., for a farmer who has a donkey, there exists at least one donkey that is beaten by him) and a universal reading (i.e., for a farmer who has a donkey, all of the donkeys are beaten by him). The above two approaches to donkey sentences have been argued by Cheng & Huang (1996) to account for conditionals in Mandarin Chinese. To them, bare conditionals are one group of conditionals interpreted through the unselective binding approach, and other. 1. The formula is taken from Pan & Jiang (1997:9) but with slight changes to fit in sentence (1). 8.

(20) conditionals are the other group2 interpreted through the E-type pronoun approach. The approaches are selected through whether or not there is a wh-word in the consequent clause of a conditional sentence. For bare conditionals, which are by principle that no overt elements like ruguo ‘if’ or dou ‘all’ appear in the clauses, they can only take a wh-word in the consequent clause anaphoric to the wh-word in the antecedent clause, as illustrated below: (3) Ni xihuan shei, wo jiu piping shei/ *ta/ you like who I then criticize who/ (s)he/ ‘If you like X, I then criticize X.’. *[e]/ *na-ge-ren. [e]/ that-CL-person (Cheng & Huang 1996:128). Hence, in a bare conditional like (3), only shei ‘who,’ but not a pronoun ta ‘he(she),’ an empty category, or a definite noun phrase na-ge-ren ‘that-CL person’ can appear in the consequent clause. Based on this fact of bare conditionals, Cheng & Huang argue that they can be analyzed by the unselective binding approach since the two wh-words in the antecedent and the consequent clauses are both bound by the necessity operator. With the adoption of the unselective binding approach, the readings of bare conditionals are proposed to be universal. The two paradigms are bare conditionals as one group and ruguo- and dou-conditionals as the other group. Bare conditionals can only take a wh-word in the consequent clause, while ruguo- and dou-conditionals can allow either an overt or covert pronoun in the consequent clause but not a wh-word, as shown in (i) and (ii): (i) Ruguo ni kandao shei, qing jiao *shei/ ta/ [e]/ na-ge-ren if you see who please tell who/ him(her)/ [e]/ that-CL-person lai jian wo. come see me ‘If you see someone, please ask that person to come see me.’ (Cheng & Huang 1996:131) (ii) Ni jiao shei jin-lai, wo dou jian *shei/ ta/ [e]/ na-ge-ren. you ask who enter I all see who/ him(her)/ [e]/ that-CL-person ‘Whoever you ask to come in, I’ll see him/her.’ (Cheng & Huang 1996:130) It is proposed by Cheng & Huang that the two types of conditionals are in complementary distribution where bare conditionals adopt the unselective binding to interpret their references while the other two conditionals take the E-type pronoun approach. Nevertheless, the researcher does not find it ungrammatical with the use of wh-words in the consequent clause, and hence, is not convinced of this dichotomy. 2. 9.

(21) readings. Accordingly, they believe that since the approaches adopted to analyze donkey sentences are also available to analyze bare conditionals, the two constructions are related. Following Cheng & Huang (1996), Pan & Jiang (1997) also deem bare conditionals to be one type of donkey sentences3. To them, like donkey sentences, bare conditionals exhibit anaphoric properties. In addition, the approaches to analyzing donkey sentences can be taken to analyze bare conditionals as well. Hence, Cheng & Huang and Pan & Jiang believe that the two constructions are related. However, Wang (2007) disagrees with the idea of the relatedness of donkey sentences and bare conditionals. It is proposed by Wang that a bare conditional is actually one kind of relative clause sentences with identical wh-words anaphoric to each other where the first clause takes the role of a relative clause and the second clause is a matrix clause, as in (4), extracted from Wang (2007:71): (4) a. Sheme pianyi, ta mai sheme. what cheap he(she) buy what b. Pianyi de dongxi ta dou mai. cheap DE thing he(she) all buy ‘He(She) buys anything that is cheap.’. Although Pan & Jiang (1997) also state that conditionals are related to donkey sentences, they argue that both bare conditionals and ruguo- and dou-conditionals can take either overt pronouns or wh-words in the consequent clause, and the two approaches can both be adopted to analyze the conditionals. Hence, they claim that there is no such a complementary distribution of conditionals that Cheng & Huang (1996) have proposed. According to Pan & Jiang, the adoption of different approaches renders different readings of conditionals. The E-type pronoun approach imposes the sentence an indication that there is only one individual or entity to make the proposition tenable, while the unselective binding approach offers a universal reading. The following are readings for “Shei xian lai, shei/ta xian chi” adopted with the unselective binding approach, as stated in (i), and with the E-type pronoun approach, as in (ii), both of which are taken from Pan & Jiang (1997:13). (i) NEC [x] [come(x) first] [eat (x) first] (universal reading) (ii) a. NEC [x, s] [come(x) first in s] [eat(x) first in s] (existential reading) b. NEC [s] [Ǝx come(x) first in s] [eat(he/ she) first in s] (existential reading) 3. 10.

(22) In (4a), the first wh-word sheme ‘what’ is considered to be a relative pronoun while the second behaves like the head of the relative clause. Examples (4a) and (4b) denote the same meaning, “He will buy anything that is cheap.” In addition, Wang argues that only the first wh-word helps to trigger a universal reading (i.e., sheme pianyi ‘anything cheap’) while the second is only a component anaphoric to the first one (i.e., the head sheme ‘what’ anaphoric to sheme pianyi). The differences between the two wh-words can also be found phonetically. It is claimed that only the first wh-word can be stressed in the sentence whereas the second clause often occurs with dou ‘all’ as shown in (4b). All in all, of these three previous studies, two of them have argued in favor of the relatedness of donkey sentences and bare conditionals (Cheng & Huang 1996, Pan & Jiang 1997), and the other has argued against such an analysis (Wang 2007). Hence, the present study attempts to find out whether or not the two constructions are related by investigating the readings of these sentences.. 2.2 Constraints on Interpretations of Donkey Sentences and Bare Conditionals In this section, issues concerning two constraints on interpretations of donkey sentences and bare conditionals are discussed. Based on the issues regarding these constraints, what previous theoretical studies have proposed are presented. Section 2.2.1 introduces the first constraint, quantifier types, and Section 2.2.2 presents the second, parallelism. 11.

(23) 2.2.1 Quantifier Types According to both Rooth (1987) and Kanazawa (1994), the entailing properties of quantifiers elicit different readings of donkey sentences. Kanazawa’s Generalization (1994) states that a universal quantifier entails a universal reading of the sentence while an existential quantifier leads to an existential reading. Rooth (1987), however, presents a contrasting generalization against Kanazawa’s. For example, in (5a), the quantifier every is a universal or upward-entailing quantifier; on the other hand, in (5b), some is an existential or downward-entailing quantifier. (5) a. Every farmer who owns a donkey beats it. b. Some farmer who owns a donkey beats it.. (Krifka 1996:2). Under Kanazawa’s Generalization, (5a) yields a universal reading (i.e., every farmer beats every donkey he has) because of the universal quantifier every, and (5b) an existential reading (i.e., a farmer beats at least one of the donkeys he/she owns, but not all the donkeys) because of the existential quantifier some. However, Rooth’s Generalization argues in a reverse pattern where the upward-entailing quantifier every, as in (5a), elicits an existential reading and the downward-entailing quantifier some, as in (5b), leads to a universal reading. Even though these two generalizations diverge in interpretations, they both have their supporters in literature. For example, Krifka (1996) supports Rooth’s Generalization, while Geurts (2002) and Foppolo (2009) root for Kanazawa’s Generalization. Nevertheless, quantifier types remain a considerable factor which is believed to be the very first possible 12.

(24) one to affect readings. Hence, this study aims to discover the effect of quantifier types on Chinese donkey sentences.. 2.2.2 Parallelism Parallelism refers to a phenomenon in bare conditionals where Cheng & Huang (1996) propose that a wh-word in the antecedent clause should have a parallel wh-word in the consequent clause, as shown below. (6) Shei xian. lai,. shei/ *ta/. who come first who/ (s)he/ ‘If X comes first, X eats first.’. *[e]/ [e]/. *na-ge-ren. xian. chi.. that-CL-person first eat (Cheng & Huang 1996:127). (7) Ni xihuan shei, wo jiu piping shei/ *ta/ you like who I then criticize who/ (s)he/ ‘If you like X, I then criticize X.’. *[e]/ *na-ge-ren. [e]/ that-CL-person (Cheng & Huang 1996:128). In (6) and (7), a pronoun, an empty category or a definite noun phrase cannot occur in the consequent clause of bare conditionals. According to Cheng & Huang, only shei ‘who’ can appear in the antecedent clause and another shei in the consequent clause. Two opposite voices arise concerning this parallelism constraint on bare conditionals. Cheng & Huang (1996) are those who propose such an analysis of bare conditionals, but other studies (Pan & Jiang 1997, Cheung 2007, Wang 2007) disagree with this constraint. Cheng & Huang (1996) state that taking a wh-word or not in the consequent clause is the way to distinguish bare conditionals from other conditionals. Hence, as in (6) and (7), they argue that in the consequent clause, shei ‘who’ is more grammatical than other forms of NPs. 13.

(25) Also, they adopt the Parallelism Constraint on Operator Binding (PCOB) to support their analysis of bare conditionals. The PCOB is a constraint which illustrates that for x and y which are both variables bound by the same operator, x and y must be phonetically realized, and the variables must be identical. Hence, sentence (6) and (7) can be interpreted as in (8) and (9) respectively, where the two x-s in both clauses should be phonetically pronounced and are wh-words: (8) NECx [x comes first] [x eats first]. (Cheng & Huang 1996:135). (9) NECx [you likes x] [I criticize x] Such anaphoric wh-words can be analyzed by the unselective binding approach, where the two wh-words are bound together by the same necessity operator, NEC, as shown in (8) and (9). Nevertheless, disagreements concerning the parallel wh-words occur (Pan & Jiang 1997, Cheung 2007, Wang 2007). Pan & Jiang (1997) and Wang (2007) argue that bare conditionals with pronouns, empty categories, or definite noun phrases substituting wh-words in the consequent clause are not as ill-formed as Cheng & Huang have stated. They argue that in bare conditionals, either a wh-word or a pronoun can occur in their consequent clause, as in sentence (10). (10) Shei bu dui, wo jiu shuo shei/ta bu dui. who not correct I then say who/he(she) not correct ‘Whoever is incorrect, I will then say he/she is incorrect.’. (Wang 2007:71). As seen in (10), both shei ‘who’ and ta ‘he(she)’ can be present in the consequent clause. 14.

(26) Consequently, against Cheng & Huang’s (1996) analysis, Pan & Jiang and Wang argue that the necessity of taking a wh-word only for bare conditionals is problematic and hardly relevant to natural language. Moreover, Cheung (2007) offers another counterexample against Cheng & Huang’s parallelism. Cheung (2007) explores a syntactic and semantic analysis of bare conditionals in Mandarin Chinese and argues that the wh-word in the consequent clause as in (11a) can be omitted, and the number of the wh-words can be inconsistent as in (11b). Both examples are taken from Cheung (2007:151). (11) a. Shei xiang qu Beijing, [e] qing4 dao wo zheli who want go Beijing [e] please to me here ‘If X wants to go to Beijing, please register with me.’ b. Shei xihuan shei, shei yinggai shuo chulai. who like who who should ‘If X likes Y, X should voice out.’. baodao. register. say out. Hence, Cheung argues that the unselective binding approach to analyzing bare conditionals is with flaws, and such flaws come from Cheng & Huang’s parallelism constraint, the Parallelism Constraint on Operator Binding (PCOB). For example, since the PCOB only constrains the wh-words to be parallel in the antecedent and consequent clauses, a phenomenon where wh-words are in islands 5 or. The original gloss for qing in Cheung (2007:151) was ‘must,’ but the researcher considered it more appropriate to translate qing into ‘please.’ 5 An island refers to a constituent where phrases and their features that stay inside this island cannot move out of it. As in (12a), in principle, the first shei ‘who’ in the antecedent clause should be anaphoric to the second one in the consequent clause, yet the first shei ‘who’ is in an wh-island, that is, an island headed by a wh-word; hence, referentiality is blocked by this island. 4. 15.

(27) coordinate structures is excluded since they block the binding of wh-words, as shown in (12a). (12) a. ??Zhangsan xiangxin [shei tou-le qian de shuofa], Lisi jiu Zhangsan believe who steal-LE money DE rumor Lisi then qu daibu shei. go arrest who Intended: ‘If Zhangsan believes in the rumor that X has stolen the money, then Lisi will arrest X.’ (Cheung 2007:153) b. Zhangsan Zhangsan xiangxin [ believe. xiangxin [shei tou-le believe who steal-LE shei tou-le qian who steal-LE money. qian de shuofa], Lisi jiu money DE rumor Lisi then de shuofa/xiaoxi]. DE rumor/news. ‘If Zhangsan believes in the rumor that X has stolen the money, then Lisi will believes in the rumor/news that X has stolen the money’ (Cheung 2007:154) Cheung argues that the unselective binding approach fails to explain the above sentences and proposes that it is essential for wh-words to occur in the position (either subject or object) in the antecedent clause identical and parallel to that position in the consequent clause of an island or a coordinate structure6. Overall, concerning parallelism of wh-words in bare conditionals, although Cheng & Huang (1996) offer a solid analysis (i.e., parallelism), three out of the four theoretical studies. Cheung states that the sideward movement is an approach that can solve the situations where the unselective binding approach fails to explain (i.e. unbalanced number of wh-words and wh-words in islands or coordinate structures). It is proposed that sideward movement is licensed by theta-role assignment and the parallelism constraint. In order to fulfill the requirement of the theta-role, the sideward movement would apply to the vacuous position that needs theta-role assignment. Also constrained by the Parallelism Constraint, the movement of the wh-expressions must be identical in form; therefore, it is the whole form that undergoes Copy and Merge with the mechanism of sideward movement. The following example is taken from Cheung (2007:157). (i) Shei jinlai, wo da shei. who enter I hit who ‘If X comes in, I hit X.’ In order to satisfy the theta-role requirement, shei ‘who’ merges with the transitive verb da ‘hit,’ and becaue jinlai ‘enter’ is an intransitive verb that lacks a subject, undergoing Copy of shei and Merge with jinlaii, it forms a grammatical sentence like (i). 6. 16.

(28) on bare conditionals provide counterexamples (Pan & Jiang 1997, Cheung 2007, Wang 2007) showing problems of Cheng & Huang’s analysis. Since previous studies focus on a theoretical aspect of parallelism, this study aims to probe into this factor to see which side of analysis can be supported from an empirical aspect.. 2.3 Empirical Studies In this section, due to the lack of empirical studies on Chinese bare conditionals, only empirical English studies on the interpretations of donkey pronouns in donkey sentences are reviewed. Moreover, since there is no empirical research working on Chinese donkey sentences, English donkey sentences are examined instead, and the three studies that are reviewed are shown in chronological order (Geurts 2002, Foppolo 2009, Grosz et al. 2014).. 2.3.1 Geurts (2002) Geurts (2002) aimed to illustrate the interplay between quantifiers and world knowledge through an experiment, and objected to the strong claim of judging the interpretations of donkey pronouns as a direct and arbitrary relation with quantifiers. Geurts argues that a quantifier and world knowledge would influence the choice of a preferred interpretation. It is proposed by Geurts that weak determiners trigger an existential reading and universal determiners, a universal reading, but that universal determiners might also generate an. 17.

(29) existential reading. Subjects of the experiment were twenty native Dutch speakers who were recruited to interpret 24 test items of Dutch donkey sentences along with 75 fillers. Since there is no difference in English and Dutch donkey sentences, the Dutch test items were simply the translation of the English version. Learning from the fact that descriptions would offer biased information to the interpretations, Geurts presented with picture situations of a truth-value judgement task in judging whether the target donkey sentence could conform to the picture illustrations. Four quantifiers were utilized – every, not every, some and no – with six different situation types (railway line-road, line-square, O-K, child-balloon, girl-dog and boy-girl) setting for the test items of donkey sentences. For example, the following two examples and pictures are extracted from Geurts (2002:136).. (13) Quantifier: every Situation type: boy-girl. Figure 2-1. Every boy that stands next to a girl holds her hand. (14) Quantifier: every Situation type: railway line-road. Figure 2-2. Every railway line that crosses a road goes over it. 18.

(30) The results turned out that for sentences with weak determiners, some and no, nearly all subjects considered those sentences with an existential reading to be true in all six situations, but that for those with universal determiners, every and not every, the interpretations were not in neat consensus. In general, 60% of preference for universal readings showed that for universal determiners, there was an overall tendency for universal readings; however, context of the child-balloon, girl-dog and boy-girl situation types tended to be interpreted as existential readings that lowered the percentage of preference for universal readings. Geurts argued that the differences between situation types could offer a solid explanation for the discrepancy of readings in both a semantic and ontological aspect. From a semantic viewpoint, weak determiners obtain an intersective property that leads us to focus on positive evidence; in contrast, universal determiners lack this property. Therefore, it is expected that this property of weak determiners would render their donkey pronouns to have existential readings. On the other hand, for universal determiners, differences are shown in the noun phrase the determiner describes, which shows the ontological differences of the noun phrases. Since one of the four pictures denoted an existential reading, if it were a non-individual, like railway line, it could be counted more than just once; on the other hand, if it were an individual, like boy, our intuition would not permit us to count one individual more than once; hence, this counting difference renders sentences with nouns of individuals that universal determiners modify to obtain existential readings. This contrast in universal determiners is. 19.

(31) also named the contrast of cases, which includes events and facts, and characters, which involves individuals. As a result, it was proposed by Geurts that the interpretation of donkey sentences should rely on the initial determiners and world knowledge based on factors, such as determiners and the case or individual of the noun phrases, and that donkey sentences should not obtain definite interpretations. Geurts’ analysis is quite promising, offering us factors that are influential to an interpretation, including world knowledge and the nature of determiners and of indefinite nouns. However, she mainly focused on the generalizations of interpretations; therefore, only one group of subjects was recruited, and only one task was designed. Picture illustrations were novel for empirical materials in donkey sentences, but it might seem to be task-demanding with four pictures as a whole for the subjects to judge the correspondence with the target sentence.. 2.3.2 Foppolo (2009) Foppolo (2009) aimed to test experimentally whether there is a preferred interpretation of the dependent variable in donkey sentences, and whether there is a default one that retains the monotonicity of the quantifier based on the generalization of Kanazawa (1994), which states that a universal quantifier entails a universal reading of the sentence while an existential quantifier leads to an existential reading.. 20.

(32) According to Foppolo, the two readings of the donkey pronoun are not independently present; instead, one always entails the other, and entailment applies to the monotonicity of the quantifier. Hence, if the quantifier is the universal one every, then the strongest reading is the universal one, entailing the existential; if the quantifier is no, the strongest reading is the existential reading, entailing the universal; if the quantifier is the existential quantifier some, nevertheless, it is the universal reading that is the strongest and entails the existential. Nevertheless, the prediction is that the default reading relies on the left monotonicity of the quantifier; therefore, it is predicted that the universal reading is preferred for donkey sentences with every, and the existential reading for donkey sentences headed by no and some. Two experiments were employed by Foppolo (2009) for the examination of Kanazawa’s Generalization, exploring whether this default reading existed, whether the preference was in the same vein of Kanazawa’s analysis, and whether biased contexts influenced the reading. Three quantifiers were taken into investigation – every with the predicted preference of a universal interpretation, while no and some with the predicted preference for an existential interpretation. The first experiment was concerned with the existence of the default interpretation. Thirty subjects were asked to take a truth-value judgement task and to evaluate whether the four pictures could express the interpretation of the target sentence with a quantifier in three. 21.

(33) different situations, and their reaction time was recorded. The subjects were told to be “charitable” when doing the tasks. To eliminate the possibility of extra-linguistic or world knowledge interfering the results, fantasy names were used in order to obtain their real interpretations of the pronouns. Target sentences with the three quantifiers, every, no and some, were put into three different situations – one that was compatible with both universal and existential readings (non-differentiating true, NDT), one that was compatible with neither of the two readings (non-differentiating false, NDF), and one that was compatible with only one of the readings (differentiating-critical, DC). The results of the first experiment showed that the default of interpretation existed. First, concerning the results of the control conditions in the three quantifiers, both the control-false condition (NDF) and the control-true condition (NDT) obtained correct responses as high as 96%. Furthermore, the DC condition illustrated that for sentences introduced by no and some, the subjects tended to interpret them with an existential reading, but that no clear tendency occurred in sentences introduced by every. For the first two quantifiers, the subjects with a high percentage rejected the sentence headed by no with a universal reading and accepted the sentence headed by some with an existential reading; however, for every, only roughly about half of the subjects rejected sentences headed by every with an existential reading. Fopplo’s first experiment showed that Kanazawa’s Generalization was partially correct; in addition, Foppolo emphasized the reaction time in the three situations offered some insight.. 22.

(34) Longer reaction time was perceived in the DC condition with sentences introduced by every than with ones introduced by some, both of which were presented an existential reading of pictures. Also, there was no great difference in reaction time on sentences headed by every between in the NDT situation, which favored the universal reading, and in the DC situation, where the existential reading was presented only. Hence, as shown with the reaction time, although no high percentage was obtained to deny the existential reading of sentences introduced by every, the reaction time on the acceptance showed that the choices of truth-values were not that arbitrary for the subjects and quantifiers were indeed influential. Although the generalization has been affirmed, the second experiment was employed to further confirm whether the alternative readings were accessible. Subjects were other thirty-six students recruited to examine the test items in a truth-value judgement task. Only taking the NDF and the DC conditions into investigation, and providing with biased contexts, Foppolo was to explore whether the subjects would be induced to choose an alternative reading, which is a disfavored one, and the two situations were designed with the disfavored reading based on the generalization. The results turned out that no significant effect on context was seen in the control situation; however, a significant effect was presented in the critical DC condition for every in that with biased contexts, the subjects were greatly influenced and chose to take an existential reading, which is a disfavored one. On the other hand, biasing contexts did not bring about. 23.

(35) the subjects to change their interpretation of the donkey pronouns in sentences introduced by no and some. It was shown that the subjects were easily induced by biasing contexts with sentences introduced by every, and Foppolo reasoned that the instruction of “being charitable” could also render them to such results. Overall, it was proved by Foppolo’s two experiments that Kanazawa’s Generalization is correct, which states that the left monotonicity on the head quantifier presents default interpretations of the donkey pronouns. The experiments and the results showed the confirmation of Kanazawa’s predictions. However, due to the fact that Foppolo only explored the default interpretations of donkey pronouns, the subject recruitment was only one group in each experiment. In addition, for the task design and materials, four pictures with fantasy names in target sentences could be task-demanding, and in a total of seventeen test items, only nine of them were critical test sentences. The number of test items was small, and this design was challenging with four pictures integrated as a whole.. 2.3.3 Grosz et al. (2014) Grosz et al. (2014) employed an experiment concerning the anaphoric properties of donkey sentences involving constraints on the antecedent. Three main constraints were investigated in the experiment – the overtness, the syntactic position and the uniqueness of an NP antecedent – to see what constrains the referentiality of the antecedent to the donkey. 24.

(36) pronoun. First of all, the overtness of the NP antecedent has been proposed to be crucial for a pronoun referring back to it, and it was the first constraint named the Overt NP Antecedent Constraint. Overtness is concerned with the noun phrase as an antecedent which could not be a part of a word, for example in the following examples extracted from Grosz et al. (2014:4), the antecedent fatherless in (15b) would be a non-overt antecedent while without a father in (15a) is an overt one. (15) a. Every child who was without a father had lost him in the war. b. Every child who was fatherless had lost him in the war. Second, it was argued that the syntactic position of the antecedent would greatly influence the acceptability of referring a donkey pronoun back to its antecedent. This is named the Salient Position Condition, which states a predicate position is more salient than a modifier position. Below are examples taken from Grosz et al. (2014:4) with this Salient Position Condition, where in (16a) the predicate position is more salient than the modifier position in (16b). (16) a. b.. [Every child who was fatherless] had lost him in the war. [Every fatherless child] had lost him in the war.. Furthermore, the third constraint was associated with the uniqueness of a singular donkey pronoun. It is greatly related to the world knowledge of the antecedent, for instance, while uniqueness is fulfilled in father since everyone can only have one biological father, it is. 25.

(37) violated in friend since the presumption cannot be made regarding one would have only one friend. Through the experiment, Grosz et al. expected to see whether all three constraints would activate the referentiality of the antecedent to donkey pronouns, and whether the three interact with one another. In addition to the above three main factors in the experiment, Grosz et al. discussed one more factor (i.e., the word type) – the N-less group and the N-owner group – since the Uniqueness Condition appears to apply as a different role in the two groups. The former group will not obtain the presupposition of the existence of the N while the latter will. For example, it is self-evident that horseless does not entail the subject possessing a horse; in contrast, horse-owner then certainly entails the subject having a horse. Grosz et al. expected the results could support the theoretical analysis of donkey sentences; hence, the two competing approaches – the E-type pronoun approach and the dynamic semantic approach – were compared. The former approach discusses the existential reading of the sentence. On the other hand, the latter approach is concerned with how interpretations of sentences influence contexts; therefore, the process of interpretations is dynamic since new information in the discourse will keep updating old information7. With regard to the item types of the antecedent, the approaches make different predictions on uniqueness. Concerning the N-less type, both the approaches predict that One of the approaches to dynamic semantics is Discourse Representative Theory (DRT), which is what the unselective binding approach based on to be used to analyze Chinese conditionals. Under this approach, interpretations are discourse representatives that integrate together to offer complete information to discourse. 7. 26.

(38) uniqueness plays a role in the interpretation of donkey pronouns. In terms of the N-owner type, the predictions of the two approaches diverge – the E-type approach still predicts the uniqueness of the antecedent, however, the dynamic approach does not. A hundred and twenty-five adult subjects were recruited. An acceptability-rating task was conducted with a total of 102 sentences with 30 critical sentences and seventy-two fillers involved, and a plausibility norming task was employed consisting of thirty-nine pairs of test items. In the first task, the subjects were asked to choose the scale of five levels from “extremely unnatural” to “extremely natural.” The four factors were taken into investigation – overtness, salience of syntactic position, uniqueness and word-types. Quantifiers in the experiment were every, no and many. In the norming task, it was to test the uniqueness of the antecedent. The results showed four different effects. A prenominal-postnominal effect where the acceptability in the postnominal type was higher, indicated the salience of syntactic position did work. An effect of overtness showed the acceptability of the overt types greatly exceeded the other. An effect of word-type presented the N-owner type was much more acceptable than the N-less type, and an effect of uniqueness was obtained where the ratings in unique antecedents were higher than those in non-unique ones. Also, an interaction was obtained between uniqueness and the word-type where only the N-less type showed a uniqueness effect.. 27.

(39) All in all, it was argued by Grosz et al. that the presences of the Overt NP Constraint and the Salient Position Condition were verified. Moreover, the interaction between uniqueness and word-type confirmed the dynamic approach, which predicted uniqueness would not operate in the N-owner type. This was the crucial evidence for Grosz et al. to propose that the dynamic approach to donkey sentences suited well. The analysis and the design of Grosz et al.’s overall experiment were very novel and elegantly displayed, especially where they provided solid evidence to support the dynamic approach. Nonetheless, questions of subject recruitment and of the design in the two word-types arise. Since it was to see how the constraints interplay with referentiality, no contrasting group was compared but only one group recruited. As to the material design, it seems reasonable even without conducting an experiment to consider the N-less type has a uniqueness presupposition while the N-owner does not. Although the factor of the word-type seems appropriate, the contrast of the two designed types does not appear that valid. Hence, it might be more persuasive if the uniqueness effect could be found in all donkey sentence types instead of just the word-types that Grosz et al. designed. Also, the aim of the acceptability-rating task was designed to see the referentiality of the antecedent to the donkey pronoun; however, such a task may be challenging for the subjects to consider this issue.. 2.3.4 Summary of the Empirical Studies As presented in Table 2-1, Geurts (2002) and Foppolo (2009) were concerned with the 28.

(40) interplay between semantic entailing of quantifiers and the donkey pronoun, while Grosz et al. (2014) focused on the factors of the NP antecedent rendering the donkey pronoun interpretable.. Table 2-1. Major Findings and Limitations of the Empirical Studies Major findings Geurts (2002). Limitations.  Quantifier types and readings: Existential reading: weak determiners. 1. Subjects: only one group recruited 2. Materials: pictures with too many illustrations 3. Task: only one task (a truth-value judgement task). Foppolo (2009).  Quantifier types and readings: 1. Existential reading: no and some 2. Universal reading: every  Biased contexts: the readings of the pronouns could be biased in sentences with universal quantifiers. Grosz et al. (2014).  Constraints on the antecedent: 1. Overtnesss 2. Salience 3. Uniqueness 4. Word-types. 1. Subjects: only one group recruited 2. Materials: pictures with too many illustrations and only seventeen test items in total, where only nine of them were critical sentences. 1. Subjects: only one group recruited 2. Tasks: (1) A problematic acceptability ratings task (2) Only specific test sentences selected for the N-less and the N-owner type in the norming task. Generally, the interpretations of the antecedent from world knowledge is crucial for the donkey pronoun (Geurts 2002, Foppolo 2009, Grosz et al. 2014). Nevertheless, to what extent world knowledge influences interpretations of donkey sentences becomes a vital factor is 29.

(41) different across three previous studies. Quantifiers are one crucial factor where existential quantifiers tend to lead existential readings only, while universal quantifiers often entail universal readings but sometimes existential readings (Geurts 2002, Foppolo 2009). Moreover, properties of the NP antecedent (Geurts 2002, Grosz et al. 2014), and the uniqueness of the donkey pronoun and the antecedent (Geurts 2002, Grosz et al. 2014) are also influential. Furthermore, only one study provided evidence for context which may be successful biasing the interpretations of universal quantifiers (Foppolo 2009). Limitations of these studies show that they mainly focused on the generalizations of donkey pronouns (Geurts 2002, Foppolo 2009) and the referential properties of the antecedents (Grosz et al. 2014), so they neglected some details for experiments. First of all, concerning the subject pool, all of them recruited only one group without any other contrasting group. Second, with regard to the materials, especially those for the task design, one was a display of target sentences only (Grosz et al. 2014), and two were presented with target sentences with picture illustrations, which contained four pictures as a whole that might be too challenging (Geurts 2002, Foppolo 2009). Third, the number of tasks was different – one employed only one task (Geurts 2002), and the others conducted two tasks (Foppolo 2009, Grosz et al. 2014). Lastly, the number of test items could be vital as well, where too little of test sentences might not be evidential (Foppolo 2009). All in all, with interpretations of donkey sentences as a main concern, there is a. 30.

(42) mismatch in two different generalizations supported in previous studies. Krifka (1996) in the theoretical section shows another version of quantifier entailing generalization (Rooth’s Generalization) that is opposite to what Geurts (2002) and Foppolo (2009) have supported (Kanazawa’s Generalization). Hence, there has been no consensus to the generalizations of quantifier entailing interpretations that awaits to be further explored. In addition, biasing contexts, which is an influential factor in interpretations, deserves to be further examined as well.. 2.4 Summary of Chapter Two This chapter discusses what previous works have proposed on the topics of donkey sentences and bare conditionals. Chinese conditionals are considered as one type of donkey sentences due to their similar referential behaviors; however, opposition to this analysis arises arguing that Chinese conditionals are just one pattern of relative clause constructions rather than donkey sentences. Moreover, two constraints that are construction-specific to interpretations in each construction, quantifier types for donkey sentences and parallelism for bare conditionals, are discussed. Analyses concerning the constraints are in dispute, but that shows the two constraints are worth investigating. In addition, empirical studies on interpretations of the donkey variables are reviewed. There seems to be some preferences for a certain reading due to the influence of quantifiers, but a factor such as world knowledge may lead to a different reading. 31.

(43) Chapter Three Research Design This chapter introduces the information of the subjects and the research design that the experiments of the present study conducted. In Section 3.1, the recruitment of the subjects is described. In Section 3.2, the design of the experiments and the materials are included, while in Section 3.3, the procedures of the experiments are presented. Lastly, Section 3.4 summarizes this chapter.. 3.1 Subjects The present study compared donkey constructions with bare conditionals from an empirical perspective. Little literature has provided an empirical aspect concerning these two constructions in Mandarin Chinese; nevertheless, in these two constructions, one of the key elements that plays a crucial role in interpretation is quantification. Hence, the present study aimed to look into how quantifiers were interpreted in children’s acquisition. Since the work of Inhelder & Piaget (1958) on children’s interpretations of quantifiers, it has been found that children had difficulty obtaining an adult-like comprehension of a quantificational sentence like “Is every boy ridding an elephant?” in response to a picture illustrated with four elephants and three of them each ridden by a boy. Adults tended to answer ‘yes’ while children between four and seven answered ‘no’ since they had difficulty. 32.

(44) comprehending such a mismatch of quantification in pictures (Herburger 1997, Cohen 2001, Geurts 2003). Following previous studies, the researcher would like to see whether the age difference would show this difficulty when it was concerned with donkey sentences and bare conditionals in Mandarin Chinese where they are involved in quantification. Accordingly, the present study was to explore between the age span of four to seven, whether kindergarten children (KS) at five years old would misinterpret the sentences and whether after seven years old, an adult-like interpretation would be obtained by Grade 2 or should wait until Grade 4, that is, nine years old. Therefore, as seen in Table 3-1, the present study recruited three experimental groups at the ages of five, seven and nine, to compare their interpretations on quantification in donkey sentences and bare conditionals, each of which contained eighteen subjects. Subjects of KS were Taiwanese children studying in New Taipei Municipal BeiXin Kindergarten School, while Grade 2 and Grade 4 were those from New Taipei Municipal BeiXin Elementary School.. Table 3-1. Information of Subjects Group Experimental Control. Age range. Number of subjects. KS. 5-6. 18. Grade 2. 7-8. 18. Grade 4. 9-10. 18. Adults. 19-22. 18. As shown in Table 3-1, eighteen native Mandarin-speaking adults aged from 19 to 22, who. 33.

(45) were undergraduate students in National Taiwan Normal University, were also recruited as a control group to contrast the overall performances of children and to verify the legitimacy of the test sentences.. 3.2 Materials and Methods When it comes to testing comprehension or interpretation in acquisition, it is more suitable to choose a cross-sectional study comparing to a longitudinal one since it allows an experiment to manipulate with factors within a short period of time (Miller 1987, Larsen-Freeman & Long 1991, Gass & Selinker 1994). This present study followed the essence of the designs from previous studies (Geurts 2002, Foppolo 2009) where cross-sectional experiments were conducted and where picture illustrations were provided as a truth value judgement task, and for another phase of the experiment, biasing contexts were provided in prior to target sentences. However, unlike previous studies (Geurts 2002, Foppolo 2009), in order to avoid picture ambiguity, the present design put two picture situations for the subjects to choose from instead of four pictures situations as a whole for truth value judgement, and take a construction-related factor, like readings of sentences into account, in order to test the relatedness of donkey sentences and bare conditionals. Previous studies have not examined the interrelationship in semantic interpretations between the two constructions (as reviewed in Section 2.1), and no consensus. 34.

(46) on whether Chinese conditionals can really be regarded as one type of donkey sentences has been reached (Cheng & Huang 1996, Pan & Jiang 1997, Cheung 2007, Wang 2007). Accordingly, with the quantificational force in interpretation that they have in common, this study selected Chinese bare conditionals as a representative structure to compare with donkey sentences to see their relatedness by examining the interpretations (universal or existential readings) of the target sentences of the two constructions. Hence, the results could indicate which reading (universal or existential) was easier for children to acquire, and whether the two constructions exhibited similar tendency in readings. In addition to interpretations that held to investigate the relation between the two constructions, factors that are specific to each construction (as reviewed in Section 2.2) were taken into consideration (Cheng & Huang 1996, Krifka 1996, Pan & Jiang 1997, Geurts 2002, Cheung 2007, Wang 2007). Details of the factors would be introduced in the following subsections. Therefore, for the material design of the two constructions with factors added in, two tasks were conducted respectively – one was of sentences in isolation to see whether the subjects had unanimous interpretations for readings, and the other was of sentences in context. Unlike in literature (Foppolo 2009) where only biasing contexts were put in the task, the present study also designed supporting ones to see whether supporting contexts can reinforce subjects’ interpretations and whether biasing context could affect interpretations. In this section, the. 35.