中文雙賓動詞的母語習得

全文

(1)國立台灣師範大學英語學系碩. 士. 論文. Master Thesis Graduate Institute of English National Taiwan Normal University. 中文雙賓動詞的母語習得. L1 Acquisition of Double Object Verbs in Mandarin Chinese. 指導教授：陳純. 音. Advisor: Dr. Chun-yin Doris Chen 研究生：林馥. 蘋. 中華民國九十八年一月 January, 2009.

(2) 摘要本研究旨在探討以中文為母語的學齡前兒童雙賓動詞的第一語言習得。中文雙賓動詞可細分為三類：必加「給」的動詞、不能加「給」動詞、以及可隨意加「給」動詞。主要的研究議題包含以下五種：各類動詞的困難度、句型的標記性、非雙賓句型之句子分析、年齡因素、以及題型效應。本研究共設計兩個測驗：句子選擇以及引導造句。研究對象為四十五位學齡前兒童及十五位大學生。受試兒童再分為三組：第一組（三歲）、第二組（四歲）、和第三組（約五歲）。研究結果顯示：第一，不能加「給」的動詞對小朋友困難度最高，而另外兩類動詞困難度差異不大。第二，必加「給」的動詞中，[V-gei-IO-DO]比[V-DO-gei-IO] 更具有標記性。對於可隨意加「給」的動詞而言，[V-gei-IO-DO] 最具有標記性， [V-IO-DO]居於第二，而[V-DO-gei-IO]最無標記性。第三，從小朋友的非雙賓動詞句型使用分析得知，他們傾向於先把這些動詞當作一般及物動詞使用，，接著加入第二個賓語，最後他們說出越來越多的複雜句子。第四，在此研究中，兒童的年齡證實在語言習得中扮演著重要的角色。年齡較大的兒童表現普遍比較小的兒童來得好。此外，在他們說出的句子中也可得知，年齡越大，句子的複雜度也越高。例如，較常加入第二個賓語以及一個句子會常有兩個動詞出現。最後，在題型效應方面，兒童在理解性測驗中的表現普遍比口說表現好，證實了先前文獻中之看法：理解語言優先於使用語言。. i.

(3) ABSTRACT The present study aims to investigate Chinese children’s acquisition of double object verbs with gei.. These verbs were classified into three types, gei-required,. gei-forbidden, and gei-optional.. The following issues were examined, i.e. the. difficulty levels, unmarked/marked patterns, development of untargeted patterns, age effects, and task effects.. Two tasks were employed, a grammaticality task, i.e.. Sticker Giving Task (SG Task), and a sentence elicitation task, i.e. Picture Description Task (PD Task).. Forty-five children aged from three to five and fifteen adults. participated in the experiment, who were further divided into four groups, i.e. Group 1 (three-year-olds), Group 2 (four-year-olds), Group 3 (five-year-olds), and Group 4 (native controls). The results are as follows: Firstly, among the three types, gei-forbidden verbs posed the most difficulty for the children, while the other two were relatively easier. Secondly, with regard to gei-required verbs, [V-DO-gei-IO] was the unmarked pattern while [V-gei-IO-DO] was marked.. Similarly, for gei-optional verbs, the most. unmarked pattern was [V-DO-gei-IO], followed by [V-IO-DO], and [V-gei-IO-DO] the most marked. untargeted sentences.. Thirdly, a steady development was found in the children’s The children began with mono-transitive use of these verbs,. then added a second object, and finally generated more complex sentences.. Fourthly,. age, as expected, turned out to be a crucial factor affecting our children’s performances.. It was confirmed that the older they were, the better they performed.. Finally, a task effect was found: the SG Task was significantly easier than the PD Task, a result consistent with the previous studies that comprehension was easier and earlier than production.. ii.

(4) ACKNOWLEDGEMENTS What I have achieved is the contribution of so many people around me. Without them, I would not be able to accomplish the task. My foremost gratitude is due to my advisor and mentor Dr. Chun-yin Doris Chen, who has profound influence on me.. Her optimism has provided me with the courage. to go on when confronted with difficulties not only in my thesis writing but also in my life.. She is the person who tells me there is always a way out, and who has given me. constructive feedback on my thesis. the present study.. Her comments have contributed considerably to. I am also inspired by her enthusiasm for linguistics and teaching.. She taught me to enjoy what I am doing and live my life to the utmost. I would also like to express my appreciation for the generous and insightful comments from my committee members, Dr. Jen-I Li, and Dr. Rueih-Ling Sharon Fahn, who have been very patient with me.. My thesis would not have been. accomplished without their valuable suggestions. Many thanks go to the other teachers in the department who have given me solid training in linguistics.. Their instruction in Syntax, Pragmatics, Semantics, discourse. analysis, academic writing, and statistics has been very informative especially in times of confusion. My final appreciation is due to my family for their consideration. I am grateful for my husband and my mother-in-law. been very supportive.. I owe a great deal to them.. iii. In particular,. Through the years they have.

(5) TABLE OF CONTENTS CHINESE ABSTRACT .............................................................................................. i ENGLISH ABSTRACT ............................................................................................ ii ACKNOWLEDGEMENTS ...................................................................................... iii TABLE OF CONTENTS.......................................................................................... iv LIST OF TABLES................................................................................................... vii LIST OF FIGURES .................................................................................................. ix Chapter One Introduction .......................................................................................... 1 1.1 Motivation ................................................................................................... 1 1.2 Theoretical Background ............................................................................... 2 1.2.1 Thematic Hierarchy ........................................................................... 2 1.2.2 Markedness ....................................................................................... 3 1.2.3 Semantic Features ............................................................................. 6 1.2.4 Age Effects ........................................................................................ 7 1.3 Research Questions ...................................................................................... 8 1.4 Significance of the Study ............................................................................. 8 1.5 Organization of the Thesis ........................................................................... 9 Chapter Two Linguistic Properties and Literature Review of Double Object Verbs ...11 2.1 Theoretical Studies of Double Object Verbs ................................................11 2.1.1 Tang (1979) ......................................................................................11 2.1.2 Li & Thompson (1981).................................................................... 14 2.1.3 Chang (2005) .................................................................................. 16 2.1.4 Her (2006)....................................................................................... 19 2.2 Linguistic Properties of Double Object Verbs ............................................. 22 2.2.1 Gei-required Verbs (GRV) ............................................................... 26. iv.

(6) 2.2.2 Gei-forbidden Verbs (GFV) ............................................................. 28 2.2.3 Gei-optional Verbs (GOV) ............................................................... 29 2.3 Empirical Studies of Dative Constructions ................................................. 31 2.3.1 Osgood and Zehler (1981) ............................................................... 31 2.3.2 Mazurkewich (1984) ....................................................................... 34 2.3.3 Sadakane and Koizumi (1995) ......................................................... 36 2.3.4 Chung and Gordon (1998) ............................................................... 39 2.3.5 Campbell and Tomasello (2001) ...................................................... 43 2.3.6 Cho, Lee, O’Grady, Song, Suzuki, and Yoshinaga (2002) ................ 47 2.4 Summary of Chapter Two .......................................................................... 49 Chapter Three Research Design ............................................................................... 51 3.1 Subjects ..................................................................................................... 51 3.2 Materials and Methods ............................................................................... 52 3.3 Procedures ................................................................................................. 58 3.3.1 Pilot Study....................................................................................... 58 3.3.2 Formal Study................................................................................... 59 3.3.3 Scoring and Statistical Analysis ....................................................... 62 3.4 Summary of Chapter Three ........................................................................ 62 Chapter Four Results and Discussion....................................................................... 63 4.1 Difficulty Level of the Three Types of Double Object Verbs ...................... 63 4.2 Children’s Preferences for Certain Patterns ................................................ 71 4.3 Other Patterns Elicited ............................................................................... 84 4.4 Task Effects ............................................................................................... 96 4.5 Summary of Chapter Four .......................................................................... 98 Chapter Five Conclusion ........................................................................................101. v.

(7) 5.1 Summary of the Major Findings ................................................................101 5.2 Limitations of the Present Study and Implications for Further Research ....102 BIBLIOGRAPHY ..................................................................................................105 Appendix A: Test Structures in the Two Tasks ........................................................ 111 Appendix B: Results of the Pilot Study................................................................... 113 Appendix C: Test Sentences Used in the Sticker Giving Task ................................. 116 Appendix D: Test Pictures Used in the Picture Description Task ............................. 117 Appendix E: One-Way ANOVA Table ....................................................................123 Appendix F: Chi-square Tables ...............................................................................126 Appendix G: Consent Form ....................................................................................129. vi.

(8) LIST OF TABLES Table 2-1 Features of the Three Types of Verbs……………………………………..26 Table 2-2 The Application of the Two Rules with Verb Types………………………30 Table 2-3 Chinese Dativizable Verbs Tested in the Study…………………………..41 Table 2-4 Chinese Non-dativizable Verbs Tested in the Study……………………..41 Table 2-5 Semantics of the Children’s Dative Verbs that could be a Double-object Dative in Adult Language…………………………………………………44 Table 3-1 Basic Information of the Subjects…………………………………………51 Table 3-2 Verbs Tested in the Two Tasks………………………………………….....54 Table 3-3 Test Items Designed in the SG Task………………………………………55 Table 3-4 A Test Sample for the SG Task ……………………………………………56 Table 3-5 A Test Sample for the PD Task……………………………………………57 Table 3-6 A Test Sample of the Training Phase for the PD Task ……………………60 Table 3-7 A Test Sample of the Training Phase for the SG Task ……………………61 Table 4-1 Subjects’ Performance of Three Types of Double Object Verbs on Both Tasks (in Means)………………………………………………………….63 Table 4-2 Subjects’ Use of the Gei-required Patterns……………………………….74 Table 4-3 The Children’s Mean Scores of the Gei-required Patterns ……………….74 Table 4-4 Use of the Gei-required Patterns by Each Group …………………………75 Table 4-5 Subjects’ Use of the Gei-optional Patterns………………………………..76 Table 4-6 Subjects’ Mean Scores of the Gei-optional Patterns………………………77 Table 4-7 Subjects’ Use of the Gei-optional Patterns by Each Group……………….78 Table 4-8 Subjects’ Preferences for Patterns in Both Tasks …………………………79 Table 4-9 Patterns Elicited of Gei-required Verbs……………………………………85 Table 4-10 Unexpected Patterns of Gei-optional Verbs……………………………...88. vii.

(9) Table 4-11 Patterns Elicited of Gei-forbidden Verbs…………………………………89 Table 4-12 Subjects’ Performance in the SG and the PD Tasks……………………...96 Table 4-13 The Performance of Each Age Group in the Two Tasks…………………97. viii.

(10) LIST OF FIGURES Figure 4-1 Subjects’ Mean Scores of the Three Types of Double Object Verbs in the Two Tasks………………………………………………………………..66 Figure 4-2 Children’s Development of the Double Object Verbs…………………...95. ix.

(11) x.

(12) Chapter One Introduction 1.1 Motivation In English, double object constructions are perhaps the most widely investigated subjects.. Chinese is no exception.. In Chinese double object constructions, one of. the most crucial elements in determining the order of the direct and the indirect objects, however, is gei.. Gei is a word with multiple functions and thus can be. treated as belonging to different syntactic categories based on its meanings1.. As a. verb, for instance, gei can be used in a double object structure like (1), meaning ‘give.’ Interestingly, gei can also combine with other verbs such as song ‘give’ in the same structure, as exemplified in (2). (1) Wo I. gei le. Zhangsan. yi. GEI ASP Zhangsan. ben shu.. (Yang 1991:11). one CL book. ‘I gave Zhangsan a book.’ (2) Wo. song. gei le. Zhangsan. I. give. GEI ASP Zhangsan. yi. ben shu.. one CL book. ‘I gave Zhangsan a book.’ In both (1) and (2), Zhangsan is the indirect object and yi ben shu ‘a book’ is the direct object.. The same meaning can be expressed with gei occurring in postobject. position in (3). (3) Wo. song. le. yi. ben shu. gei. I. give. ASP one CL book GEI. Zhangsan . Zhangsan. ‘I gave a book to Zhangsan.’. 1. Gei is used as a verb to mean ‘give,’ and it means ‘for’ when used to introduce beneficiary. complements, ‘to’ to introduce goals, or ‘by’ in passive constructions, etc.. 1.

(13) As can be seen in (2) and (3), double object verbs co-occurring with gei exhibit two patterns, [V-DO-gei-IO] with gei in the postobject position, and [V-gei-IO-DO] with gei in the postverbal position. with gei.. Among them, however, not every verb can occur. These verbs, whether they can take gei or not, together with different. patterns they exhibit, will be investigated in the present study, especially from the acquisitional perspective.. 1.2 Theoretical Background In the present study, attention will be drawn to Givón’s thematic hierarchy and to the notion of markedness with regard to first language acquisition.. In addition,. Clark’s (1973) theory of children’s semantic acquisition of words, together with the age factor, will be explored. 1.2.1 Thematic Hierarchy In Chinese double object constructions, the order of the direct (DO) and the indirect object (IO) will depend on the occurrence of gei.. In order to resolve the. problem with word order, the present study adopts Givón’s (2001) thematic hierarchy as a defining criterion.. Givón differentiated the grammatical positions such as. subject and DO (direct object) from semantic roles such as agent and dative.. He. investigated the grammatical position of the DO occupied by different semantic roles, with a view to providing a hierarchy of the semantic roles that are more likely to appear in the DO position.. In his discussion of the “preferred argument structure”. (2001:199), which is “the mapping from semantic roles to the GRs subject and DO” (2001:199), Givón proposed the following hierarchy of topicality of semantic roles.. 2.

(14) (4) Hierarchy of topicality of semantic roles. (Givón 2001: 200). Agt > Dat / Ben > Pat2 > Loc > others The above hierarchy can be interpreted as (5), where GR stands for a grammatical relation. (5) Access to object GR. (Givón 2001: 200). If a simple clause has a dative / benefactive argument, it has the highest claim to the direct object GR.. Otherwise, the claim passes down hierarchy (4) in order.. In other words, in simple clauses, the Agent always plays the winning role to be the subject3.. With regard to the competition for the position as DO (direct object),. dative-benefactive wins, since it is the second most “topic-worthy” in discourse (2001:200). To support this view, Givón offered a typological perspective in which many languages “obligatorily” promote the dative-benefactive to the position as DO. To be more specific, if a clause has a dative-benefactive, it has to occur in the DO position.. Although English allows a dative-benefactive to be either the DO or IO,. frequency counts from written text still indicate a preference for a dative-benefactive to be the DO (Givón 1984).. Therefore, following Givón’s hierarchy, the preferred. argument structure for double object constructions will be [V-dative-patient], i.e. “She sent him the book” (2001:143).. 1.2.2 Markedness The notion “markedness” can also be employed to explicate a preference for. 2. In Givon (1984), ACC (accusative object, which is also the theme) was used instead of Pat (patient).. 3. According to Givón, unless the agent is in “passive, inverse or other de-transitive voice clauses,” it. will not be “demoted from subjecthood” (2001: 200).. 3.

(15) word order in double object constructions in the present study.. In the present study,. discussion of markedness will center around whether a given pattern is more frequently or widely used than another. marked 4.. That is, the more widely used, the less. Markedness Differential Hypothesis (MDH), also called the Markedness. Theory, was originally proposed by Eckman (1977) to compensate for the inadequacy of the Contrastive Analysis Hypothesis in second language acquisition.. He defined. markedness as the following, which, he claimed, should be universal. (6) Markedness. (Eckman 1977:320). A phenomenon A in some language is more marked than B if the presence of A in a language implies the presence of B; but the presence of B does not imply the presence of A. Furthermore, Eckman incorporated “degree of difficulty” (1977: 320) into language acquisition theories from a typological perspective. difficulty corresponds to the degree of markedness. more difficult.. According to him, the degree of The more marked will be the. In other words, difficult items for language learners can be predicted. from markedness relations for the reason that, generally speaking, we learn to do things which are less difficult before we learn to do more complicated things (Eckman 1977).. Therefore, marked forms are considered more difficult and will be. acquired later than unmarked forms (Eckman 1981; Pinker, 1989), and will be accompanied by more errors (Pinker 1989).. White (1986) also points out that in the. process of children’s acquisition, the initial state of their hypothesis is usually 4. The ideas of marked and unmarked categories were first proposed in Prague school phonology. (Greenberg 1966) to account for problems such as neutralization. This notion was employed by Trubetzkoy to refer to text frequency, which can be applied to various fields of language such as grammar and semantics. An example given by Greenberg (1966) was that of the category of number, e.g. in English, singular forms normally do not have overt affixes, while the plural ones are more likely to be marked by an affix.. 4.

(16) unmarked and that marked forms will not be acquired without specific evidence. Furthermore, language forms which are unmarked also have a wider range of distribution and higher frequency of occurrence than marked ones (Greenberg 1966, Liu 2001), and they usually convey concrete meanings (Comrie 1986). The notion of markedness was also one of the main issues in Liu (2001), who examined ditransitive patterns of gei from a typological point of view.. Comparing. different dialects of Mandarin Chinese, he proposed a hierarchy of principles with regard to the degree of markedness.. His hierarchy is as follows:. (7) Conceptual distance (structural and linear distance) > heavy NP > topicality > economy principle > conceptual complexity5. (Liu 2001: 396). The violation of the above principle will result in different degrees of markedness, the leftmost being the most influential.. In his observations from several languages. which differentiate the IO from the DO, the DO is usually the unmarked argument, while the IO is the marked one. the DO has no marker at all.. In Tibetan, for instance, the IO takes a marker la but. Liu discussed structural and linear distance with regard. to the conceptual distance of elements.. He claimed that if two elements are. conceptually closer, they will have a closer distance in terms of syntactic structure, i.e. structural distance.. On the other hand, linear distance refers to the fact that when. two elements are conceptually closer, they usually tend to have a closer linear distance, i.e. the distance on the surface form. In Mandarin Chinese, according to Liu, the DO and V are conceptually more closely related than the IO and V, and thus the linear distance between DO and V will be shorter than that of IO and V, which happens to be contrary to Givón’s (2001) thematic hierarchy.. Liu’s argument, as he. claimed, can account for the fact that IOs are usually introduced by prepositions or 5. Please refer to Liu (2001) for discussion of the above principle.. 5.

(17) receive Oblique case.. Following his criterion of conceptual distance, dative and. ditransitive structures can be ranked according to their degree of markedness.. Based. on Liu’s observation, the construction [V-DO-gei-IO] is the most unmarked, followed by [V-IO-DO], and the most marked form will be [V-gei-IO-DO], which is also more restricted in terms of its occurrence in Chinese as well as in other Chinese dialects.. 1.2.3 Semantic Features With respect to children’s acquisition of word meanings, a componential analysis can provide a lucid illustration (Goodluck 1991).. Among the various theories of. such approach, the present study will testify Clark’s (1973) Semantic Feature Hypothesis (SFH) by examining the semantic acquisition of double object verbs in Mandarin Chinese.. Similar to other accounts of meaning acquisition, Clark’s theory. also postulates that the meaning of a word is composed of small elements called semantic features.. As a child, the meaning of a word may consist of one or two. semantic features or components.. When he/she grows older, there will be a steady. accretion of features to that word until finally its components correspond to those listed in the adults’ entries.. In other words, the SFH assumes that at an initial stage. of word use, children have not acquired the full meaning, and that their acquisition process is gradual.. Another characteristic of the SFH is that the semantic features of. a word comprise plus (+) and minus (−) values, and positive values are more unmarked.. Children have to learn the contrasting values before they can correctly. interpret a pair of antonyms6.. In addition to the opposite feature values, Clark. proposed that there should be a distinction between general and specific features. 6. An example used by Clark was the pair more and less.. positive.. 6. More is [+Amount], which is unmarked and.

(18) The order of acquisition will be that general features are acquired first, followed by specific features gradually, i.e., a “top-down” sequence (1973:75). In the present study, Chinese double object verbs will be specified into three types according to their semantic features, as will be discussed in the following section.. It is hoped that with the help of the Semantic Feature Hypothesis, we are. able to gain insight into children’s performance on different types of verbs.. 1.2.4 Age Effects The last issue that the present study aims to address is the age factor. doubt that age plays a crucial role in children’s acquisition of language.. It is no. Researchers. are interested in age effects on acquisition of certain aspects of a language, such as syntax, semantics, or pragmatics.. In studying the acquisition of nouns and verbs in. Mandarin and English, Gelman et al. (1998) found that their English-speaking subjects’ first word for objects appeared at the mean age of 12.4 months, while their first action word appeared at the age of 16 months.. Their Mandarin-speaking. children produced first object words at around 14.6 months of age, while they said the first action words appeared at 17.3 months of age.. Chien et al. (1983 and 1985), in. their study of children’s ability to differentiate topic from subject in Mandarin Chinese, found that children at the youngest age (2;6) were able to distinguish between subject and topic, indicating that the grammatical concept of subject might not be fully acquired by children until 2;6 or at lower language levels.. Huang (2007). explored children’s acquisition of Chinese count and mass classifiers, and found that the three-year-olds were able to use two to three classifiers in their production task. At age four, the number of classifiers used increased to four or five classifiers.. At. age five, her subjects could produce five to six classifiers properly, showing that there 7.

(19) existed a developmental progress in children’s acquisition of Chinese classifiers.. As. Gentner claims, “…the meanings of many common verbs are not fully acquired until the age of 8 years or older.” (1978: 996) Although children’s use of certain words may not be equated to their full acquisition of those items, it is believed that a more general understanding of acquisition can be achieved through the investigation of their performance.. 1.3 Research Questions Based on the theoretical background discussed above, the present study aims to address the following research questions. 1. At what age do children acquire double object constructions? 2. Of different types of double object verbs, which type poses the most difficulties and which type is the easiest for children to acquire? 3. For the double object verbs that exhibit different patterns, which pattern is preferred by children? 4. Do children overgneralize any patterns in acquiring double object and dative constructions? 5. Would there be task effects between the comprehension and the production task?. 1.4 Significance of the Study Studies on Mandarin double object verbs abound in the literature (cf. Tang 1979, Li and Thompson 1981, Li 1990, Yang 1991), although most of them failed to arrive at a consensus over the syntactic categories of gei in the postobject and the postverbal positions (cf. Li 1990, Yang 1991, Her 2006). Unfortunately, none of the studies. 8.

(20) examined these verbs from an acquisitional perspective.. On the other hand, there. have been a great number of empirical studies on double object and dative constructions in languages other than Mandarin Chinese (cf. Osgood and Zehler 1981, Sadakane and Koizumi 1995, Campbell and Tomasello 2001), some of which were from the acquisitional point of view (Osgood and Zehler 1981, Mazurkewich 1984), while some took pragmatic factors into consideration (Sadakane and Koizumi 1995, Cho, Lee, O’Grady, Song, Suzuki, and Yoshinaga 2002).. The one that examined the. acquisition of dative constructions in Mandarin Chinese (Chung and Gordon 1998) revolved around the semantic constraints of such verbs and was unable to provide a whole picture of children’s acquisition of double object verbs as well as the uses of gei in its relevant structures. In the present study, Chinese double object verbs will be divided into two types based on whether they take gei or not and further into three subtypes according to their semantic features. theme as well.. Children’s acquisition of these verbs will be a recurrent. It is hoped that the present study will help us gain insights into the. acquisition of Chinese double object and dative constructions.. 1.5 Organization of the Thesis This thesis consists of five chapters.. The first chapter discusses several. common uses of gei in Mandarin Chinese and then presents the purpose of the study. In Chapter Two, a number of theoretical studies will be reviewed first on the syntactic behaviors of gei as well as the syntactic analysis of double object verbs.. Then, the. linguistic properties of gei will be discussed, together with the classification of Chinese double object verbs.. In addition, several empirical studies concerning the. research purposes of the present study will be discussed.. 9. Chapter Three presents the.

(21) research design and methods employed in the study. presented with discussion in Chapter Four. major findings of the present study.. The results of the study will be. Finally, Chapter Five summarizes the. Limitations and suggestions for further research. are provided as well.. 10.

(22) Chapter Two Linguistic Properties and Literature Review of Double Object Verbs This chapter presents linguistic properties of gei and double object verbs in Mandarin Chinese.. Theoretical backgrounds as well as some empirical studies on. double object constructions will be discussed. Section 2.1 reviews some previous studies of Chinese double object verbs and examines the properties of gei. Two constraints on the occurrence of gei and the order of direct and indirect objects will be proposed.. In Section 2.2, the linguistic properties of gei and its co-occurring verbs,. both syntactic and semantic, will be discussed.. Section 2.3 discusses empirical. studies of dative constructions in different languages.. Finally, a summary of this. chapter is given in Section 2.4.. 2.1 Theoretical Studies of Double Object Verbs 2.1.1 Tang (1979) In his book Studies in Chinese Syntax, Tang (1979) classified Chinese double object verbs into four types based on their meanings, the order of DO and IO, the requirement of gei, and the syntactic transformations they can undergo, such as passivization. The first type that denotes transfer of objects includes verbs such as ji ‘mail, send,’ chuan ‘pass,’ and jiao ‘hand over.’. The subject is the person who transfers. things; the direct object (DO) is the thing being transferred; the indirect object (IO) is the person who receives the transferred objects.. These verbs are called the. gei-required verbs, and the IO must co-occur with gei.. In addition, the IO can either. precede or follow the DO.. According to Tang, the omission of the IO is acceptable,. while the omission of the DO is not.. In addition to the order of the two objects, Tang 11.

(23) employed six types of transformations as the classification criteria of the four types of verbs: (1) ba-construction, (2) lian-construction, (3) passivization, (4) topicalization, (5) relativization, and (6) clefting.. As for the first type of verbs, only the DO can be. preposed in the ba- and lian-constructions and only the DO can undergo passivization and topicalization.. The only one which can be relativized and can also be. emphasized in cleft sentences is the DO. Like Type One verbs, the second type also denotes transfer of objects and it includes verbs like song, ‘to give as a present,’ shang, ‘bestow,’ huan, ‘return,’ fu, ‘pay,’ and shu, ‘lose.’ The IO has to co-occur with gei and it can either precede the DO or follow the DO.. The omission of the IO is acceptable, while the omission of. the DO will yield unacceptability.. Unlike Type One verbs, when the IO precedes the. DO, gei is optional, and can be omitted.. As for the six transformations, in the. ba-construction as well as passivization, only the DO can be preposed.. In the rest. four patterns, i.e., lian-construction, topicalization, relativization, and clefting, both the DO and the IO can be moved or emphasized. Different from Types One and Two, Type Three verbs denote the action of consumption, obtaining, or deprivation, such as chi ‘eat,’ ying ‘win,’ tou ‘steal,’ and hua ‘spend.’ There is only one order of the DO and IO; that is, the IO has to precede the DO.. In addition, the IO cannot co-occur with gei.. The IO can be omitted. without changing the meaning, but the omission of the DO is not acceptable.. In. addition to the fixed order of the IO and the DO, which makes Type Three different from Types One and Two, the transformations they can undergo are also different. First of all, neither the DO nor the IO can be preposed in ba-construction.. Secondly,. in the lian-construction and passivization, only the IO, not the DO, can be preposed. As for topicalization, relativization, and clefting, the movement of the IO seems more. 12.

(24) acceptable than that of the DO. Tang claimed that for Type Three Verbs, gei cannot be inserted. think that these verbs, such as tou ‘steal,’ can co-occur with gei.. However, we. The only difference. is that in the sentence with gei, the IO is the goal; in the sentence without gei, the IO is the source. The fourth type of verbs behaves like Type Three verbs in that the IO cannot co-occur with gei, and it must precede the DO.. These verbs do not denote the. transfer of objects, but they metaphorically signify the same meaning.. Type Four. verbs include wen ‘ask,’ jiao ‘teach,’ gaosu ‘tell,’ and qingjiao ‘inquire.’ Unlike the first three types, the omission of the DO is as acceptable as that of the IO.. For verbs. like jiao ‘teach,’ the DO can be preposed in ba-construction; however, for those like wen ‘ask,’ neither the DO nor the IO can be preposed in the same construction.. In. passivization, on the other hand, the IO of a verb like wen ‘ask’ can be passivized, but neither of the two objects can undergo such movement for a verb like jiao ‘teach.’ As for lian-construction, topicalization, relativization, and clefting, both the DO and the IO can undergo movement. In addition to the four types of verbs, there remains a group of verbs which Tang called “bi-directional verbs” (1979: 367).. These verbs include na ‘take,’ mai ‘buy,’. zu ‘rent,’ and jie ‘borrow / lend.’ If the direction of the action is from the subject to the IO, it is the “outward” meaning, and must take the preposition gei. the “outward” meaning will be associated with Type One verbs.. Therefore,. On the other hand,. if the direction is from the IO to the subject, it is the “inward” meaning, and can never co-occur with gei (1979: 367). Thus, the “inward” use will behave like Type Three verbs. To sum up, Tang classified Chinese double object verbs into four types based on. 13.

(25) whether they take gei or not, the order of the two objects, and certain transformations they can undergo.. Type One verbs denote the transfer of objects, and gei must be. inserted before the IO.. Type Two verbs also denote the same meaning as Type One,. but when the IO precedes the DO, the occurrence of gei is optional.. Type Three. verbs, different from Types One and Two, signify meanings of deprivation or consumption, and they cannot co-occur with gei. The fourth type, metaphorically signifying the action of transfer, behaves like Type Three in that this type can never co-occur with gei, and the IO must precede the DO. Tang’s classification was a result of detailed scrutiny of various criteria. focus of the present study is the occurrence of gei.. The. Moreover, we hold a different. view on the category of some verbs such as those of Type Three Verbs.. Therefore,. what Tang had classified is not appropriate for the purpose of the present study.. 2.1.2 Li & Thompson (1981) Different from Tang (1979), Li and Thompson (1981) classified Chinese double object verbs into three types according to whether they can take gei or not. The first type of verbs must co-occur with gei, which include verbs such as fen ‘share,’ na ‘take,’ zu ‘rent,’ shu ‘lose,’ jiao ‘call,’ di ‘pass,’ dai ‘take,’ mai ‘sell,’ and xie ‘write.’ Type Two verbs are gei-optional.. The IO can co-occur with or without gei,. such as jiao ‘teach,’ shu ‘promise,’ jie ‘lend,’ fu ‘pay,’ song ‘give (as a present),’ and shang ‘bestow.’ The first three verbs belong to different categories in Tang’s (1979) classification.. Jiao ‘teach’ is grouped into Type Four by Tang, and cannot co-occur. with gei. Shu ‘promise’ is a Type One verb in Tang’s study and has to co-occur with gei. Tang.. Jie ‘lend/borrow,’ on the other hand, is a “bi-directional verb” according to In its outward use, which means ‘lend,’ jie has to co-occur with gei; however, 14.

(26) in its inward use which means ‘borrow,’ jie cannot take gei as a Goal marker. Tang’s Type Three and Type Four verbs, all of which cannot co-occur with gei, are classified into the same type by Li and Thompson. grouped into three subtypes.. These verbs are further. The first includes gei used as a verb to mean “give.”. The second consists of verbs like wen ‘ask,’ gaosu ‘tell,’ huida ‘answer,’ and daying ‘promise.’ The third subtype, according to Li and Thompson, includes verbs such as ying ‘win,’ tou ‘steal,’ qiang ‘rob,’ and duo ‘take away,’ meaning consumption or deprivation. Source.. The subject of these verbs is usually the Goal, while the IO is the. That is to say, the transfer is from the IO to the subject.. These verbs never. co-occur with the goal marker gei for the reason that there is no transfer of an object to the IO.. Moreover, the order of the IO and the DO is not changeable.. The IO. must precede the DO. In addition to the reasons for which Type Three verbs cannot co-occur with gei, Li and Thompson (1981) have not accounted for why Type One verbs have to take gei, while Type Two are gei-optional, given that both types of verbs denote transfer from the subject to the IO. In conclusion, based on whether they have to take gei or not, Chinese double object verbs can be grouped into three types.. The first type has to take gei, the. second is gei-optional, and the third never co-occurs with gei.. Among the third type. of verbs, Li and Thompson further grouped them into three subtypes according to their meanings, together with the reasons for which these verbs can never co-occur with gei.. Although there exist subtle differences between Li and Thompson’s. classification and Tang’s, both taxonomies look alike. Although the classification was mainly based on the occurrence of gei, verbs of the third type were quite different from one another. Take gaosu ‘tell’ and ying ‘win’. 15.

(27) for examples.. The two verbs denote different kinds of transfer, especially the. direction of transfer.. Therefore, it is better to classify them into two.. 2.1.3 Chang (2005) Chang’s discussion about gei revolved around the following four structures: the V-gei sequence, [gei-NP] in sentence-final position, preverbal gei, and the structure [gei-NP-VP].. Attention has been paid to the first two structures, the V-gei sequence. and [gei-NP] in sentence-final position.. Chang not only discussed the categories of. gei in these two structures from a syntactic perspective, but he also adopted functionalists’ view to account for various constraints on the use of gei.. In addition. to the various uses of gei, Chang also discussed another type of double object verbs in Mandarin Chinese, source object verbs, a term taken from Yang (1991).. The. difference between source object verbs and other double object verbs lies in the thematic role of the IO.. The IO of source object verbs is a Source, instead of a Goal.. These source object verbs in Chang’s study coincide with Type Three verbs in Tang’s (1979) classification, which cannot co-occur with gei, as discussed in Section 2.2.1. First of all, in his discussion of the V-gei sequence, Chang claimed that the structure [V-gei-NP-NP] was parallel to that of double object constructions.. Chang. pointed out that some verbs in such sequence simply do not have a Goal in their theta-grids but they still allow the V-gei formation, e.g. zuo ‘make,’ and zhe ‘knit.’ Therefore, Chang proposed that verbs in the V-gei sequence subcategorize for an Agent and a Patient.. The addition of gei assigns a Dative role to the verb, so that the. theta-grid of the V-gei sequence will be <Agent, Dative, Patient>.. Furthermore,. verbs allowed in such a sequence are restricted to those with certain semantic contents only, i.e. verbs of transfer of possession, verbs that express causation of motion, and 16.

(28) verbs of creation and preparation.. The verbs in Tang’s (1979) second type, as. discussed in Section 2.2.1, which can either co-occur with or without gei, were also discussed by Chang.. Chang argued that gei in the V-gei sequence is never optional.. In other words, these gei-optional verbs, i.e. song ‘give,’ shang ‘bestow,’ huan ‘return,’ etc. exhibit two-argument as well as three-argument uses.. When. co-occurring with gei, these verbs are two-argument predicates; without gei, they are three-argument predicates.. Another problem with gei is that some verbs cannot. co-occur with gei, which were classified into Tang’s Type Four verbs with the metaphorical meaning of transfer.. Chang suggested that these verbs like gaosu ‘tell,’. should have already subcategorized for a Dative.. The addition of gei is hence. redundant. Moreover, Chang adopted Her’s (2006) claim that gei in [gei-NP] in sentence-final position is a preposition.. In the sentence, Wo song yi ge liwu gei ta ‘I. gave a present to him,’ yi ge liwu ‘a present’ is a Patient and it receives ACCUSATIVE Case from the verb song ‘give,’ while ta ‘him’ is a Dative and it receives DATIVE Case from the preposition gei. This is the to-dative construction in Mandarin Chinese.. In order to support his claim that gei is a preposition, Chang. introduced Bresnan’s (1982) “theory of universal controller hierarchy,” which stated that only an OBJθ can be a controller, but an OBLθ cannot1.. Since gei is not a verb. but a preposition, the NP following gei is not an OBJθbut an OBLθ, which explains why the NP in [gei-NP] cannot be a controller.. In addition to Bresnan’s theory,. Chang further referred to Zhang’s (1990) analysis of the SVC constructions to argue for his preposition account.. Zhang proposed that a true SVC construction can allow. both verbs to take the aspect marker –le. 1. In the gei structure, only the first verb can. Please refer to Chang (2005) for more review of the theory. 17.

(29) take the aspect marker, not gei. Second, in an SVC construction, the second verb can be stranded, but this is not applicable to gei.. Third, the order of verb phrases in. an SVC construction cannot be changed due to the sequencing of events.. The. [gei-NP] pattern, however, can occur in sentence-final as well as postverbal positions. Based on Ting and Chang’s (2004) study, the second verb in an SVC construction can be modified by adverbs, but the [gei-NP] cannot, which was aligned with Chang’s view.. Although the above analysis supports the prepositional claim, Chang admitted. the existence of [gei-NP] as a VP, from a historical perspective.. Based on the. diachronic development of gei, Chang suggested that the [gei-NP] was once a VP in an SVC construction, and it was later recategorized as a preposition, a process of grammaticalization from a verb to a preposition. As for the “source object verbs,” citing Li (1990), Chang discussed two characteristics.. The first is that these verbs can have the -de insertion after the IO,. which is not allowed in other double object verbs.. Second, the DO cannot undergo. topicalization, i.e., A’-movement, which is consistent with Tang’s analysis that among Type Three verbs, only topicalization of the IO is acceptable, not the DO, as discussed in Section 2.2.1.. Despite the above differences between source object verbs and. other double object verbs, Chang proposed that all these verbs have identical structures. The thematic role of the IO to be a Source or a Goal, depends upon the semantic content of the verbs. To conclude, Chang examined gei in double object and dative constructions, and proposed that in the V-gei sequence, gei is a verb, while in the sentence-final [gei-NP] formation, gei is a preposition.. In other words, according to Chang’s analysis, both. structures are base-generated and no derivation is involved.. Chang also investigated. the structures of source object verbs, and argued that they have identical structures. 18.

(30) with those of double object verbs. The only difference lies in the thematic role of the indirect object.. Concerning the question of why some verbs are gei-optional, a. problem pointed out by Li and Thompson but was not accounted for, or why some verbs cannot co-occur with gei, Chang provided reasonable accounts.. For those. so-called gei-optional verbs, Chang argued that these verbs can be two-argument and three-argument verbs. gei.. When they are used as two-argument verbs, they have to take. Therefore, the addition of gei is not optional at all.. Some verbs cannot. co-occur with gei because they have already had a Dative role in their theta-grids, making the addition of gei redundant. Chang provided detailed analysis of the syntactic and semantic structures of various patterns of double object verbs, but the issue of acquisition was not discussed. Furthermore, he conducted a frequency count of the dative and double object patterns in the book Ting Yuan Shen Shen by Chiungyao (1969).. The language in the book,. we believe, is different from the one used in modern Mandarin.. Therefore, the result. may not be completely applicable to modern Mandarin.. 2.1.4 Her (2006) Her (2006) conducted a comprehensive study of the functions and categories of gei in five different positions: verbal, postobject, postverbal, preverbal, and purposive positions.. Only the postobject and postverbal gei were reviewed here since the other. three positions were beyond the scope of the present study. The postobject use of gei, [V-NP1-gei-NP2], is prevalent in Mandarin Chinese. Her claimed that gei in the postobject position is a verb and also a preposition, as can be seen in (1):. 19.

(31) (1) Yuehan John. zhuan. qian. gei nvpengyou. liwu. earn. money. GEI girlfriend. present. ‘John earns money to give his girlfriend a present.’ He considered (1) a serial verb construction (SVC), with the second VP headed by the verb gei.. The object position of the second VP, liwu ‘present,’ can be a gap, as in. (2): (2) Yuehan John. zhuan. qiani. gei nvpengyou. earn. money. GEI girlfriend. ei.. ‘John earns money to give it to his girlfriend.’ The resulting structure is parallel to that of the postobject construction of gei. Therefore, gei can be analyzed as a verb in the postobject position.. Nevertheless,. Her (2006) argued that gei is also a preposition in the same position, which accounts for the existence of dative alternations in Mandarin Chinese.. Sentence (3) can best. illustrate the verbal and the prepositional interpretations of gei: (3) a. Yuehan jie John. yi. qian. kuai. gei Mali.. loan one thousand dollars. GEI Mary. i. ‘John loaned one thousand dollars to Mary.’ ii. ‘John borrowed one thousand dollarsi to give ei to Mary.’. If gei were a verb, the only interpretation would be (ii).. In fact, (i) is also the. common reading of the sentence, in which gei is treated as a preposition.. Based on. the prepositional account, Her (2006) claimed that (3) ai and (4) are dative alternations.. 20.

(32) (4) Yuehan John. jie. Mali. loan Mary. yi. qian. kuai. one thousand dollars. ‘John loaned Mary one thousand dollars.’ Therefore, in Her’s analysis of the postobject gei, there are two structures: the [gei-NP] as a PP and the [gei-NP-e] as a VP. it is a verb.. In other words, when gei is used to mean ‘give,’. When gei is used as a Goal marker, it is a preposition.. The postverbal gei in the V-gei structure is treated as a verb in Her’s study. Therefore, the V-gei formation is a V-V compound, which is against Huang and Ahrens’ (1999) V-suffix analysis.. Given the fact that verbs which have prepositional. dative constructions also have V-gei formation, and vice versa, and that verbs which cannot have prepositional dative constructions will not have the V-gei formation, Her thus employed two features as criteria to classify verbs into four natural classes: (i)[+Prep. Dative, -Double Obj.], such as ji ‘send’ and na ‘take;’ (ii)[+Prep. Dative, +Double Obj.], such as song ‘give’ and tigong ‘provide;’ (iii) [-Prep. Dative, +Double Obj.], such as gaosu ‘tell’ and jiao ‘teach;’ and (iv)[-Prep. Dative, -Double Obj.], such as zhan ‘stand’ and kan ‘watch.’ According to Her (2006), the verbs allowed in the V-gei formation are all [+Prep. Dative], which subcategorize for a Goal.. The V-gei. formation combines the verb gei, which subcategorizes for <Agent, Goal, Theme>, with a verb of the same argument structure. The resulting compound behaves like gei, the head of the V-V formation. To sum up, Her (2006) provided a comprehensive analysis of the functions and categories of gei in five positions: verbal, postobject, postverbal, preverbal, and purposive.. In postobject structures, gei can be treated as a verb and also a. preposition.. In postverbal positions, the V-gei formation is considered a V-V. compound headed by the verb gei, with both the verb and gei having the same. 21.

(33) argument structures. Most patterns where gei can be used were analyzed in Her’s study, but he only focused on the syntactic category of gei in different positions.. Semantic features and. preferences were not discussed at all.. 2.2 Linguistic Properties of Double Object Verbs In Mandarin Chinese, double object sentences come in various patterns, as illustrated in (5), where gei is before the object Yuehan ‘John’ as in (5) a, Yuehan ‘John’ occurs immediately after the verb without gei as in (5) b, and gei is omitted, as in (5) c.. Confusing as it may seem to differentiate these uses, an overall analysis of. double object verbs suggests that there lie differences of semantic contents among these verbs. (5) a. Mali Mary. dai. le. yi. ge. dangao. bring. ASP one CL cake. gei Yuehan. GEI John. ‘Mary brought a cake to John.’ b. Mali Mary. daying. Yuehan. promise John. yi. jian shiqing.. one CL thing. ‘Mary promised John to do something.’ c. Mali Mary. song. (gei). Yuehan. yi. ben shu.. give. GEI. John. one CL book. ‘Mary gave John a book.’ Verbs like dai as in (5) a usually take an Agent and a Patient as in (6), where Mali ‘Mary’ is the Agent and dangao ‘cake’ is the Patient.. According to Yang. (1991), these verbs imply the physical “transaction” of objects, but they do not denote the “full meaning of giving” (1991: 18-19) [－Giving].. 22. This explains why gei must.

(34) be present if a Recipient, i.e. a Goal, is added.. Therefore, they are called the. gei-required verbs in the present study. (6) Mali Mary. dai. le. yi. ge. dangao.. bring. ASP one CL cake. ‘Mary brought a cake.’ Verbs such as daying in (5) b, however, do not imply the “physical transaction” of objects.. Instead, they only denote “abstract transaction” (Yang 1991: 25).. Chang. (2005) claimed that in addition to Agent and Patient, verbs like (5) b subcategorize for Dative; thus, gei is not required in such construction. this type the gei-forbidden verb2.. In the present study, we call. Verbs like song as in (5) c are similar to verbs like. dai as in (5) a in that they can co-occur with gei. However, they differ in that gei can be omitted in (5) c, but not in (5) a.. Yang claimed that these verbs not only signify. transaction of an object, but also denote the meaning of ‘giving,’ i.e. [＋Giving]. Therefore, they can occur without gei.3 In order to account for the obligatory occurrence of gei, the following rule is proposed based on different semantic features discussed above. (7) Gei-insertion Rule For double object verbs with the feature [－Giving], gei must be inserted to form the double object construction. Verbs that are of the same type as dai in (5) a usually imply physical transfer of objects and they do not involve the full meaning of ‘giving,’ i.e. [－Giving]. Therefore, when a Goal is added, the Gei-insertion Rule must apply to enhance the giving feature of the verb. Thus, this type is called the gei-required verbs, i.e. GRV.. 2. Verbs like daying as in (5) b were called “goal object verbs” in Yang’s (1991) classification.. 3. Yang (1991) called this type the gei-optional verb. 23.

(35) As for verbs like daying as in (5) b, although there is no physical transfer of objects, they still signify transfer of “information” from the subject to the IO (Tang 1979).. That is to say, these verbs have already subsumed the meaning of transaction,. i.e. [+Giving].. In this case, the Gei-insertion Rule need not apply.. They are. defined as the gei-forbidden verbs, i.e. GFV. Verbs like song ‘give’ as in (5) c differ from the gei-required verbs in that gei can be omitted.. To account for this difference, we adopt Chang’s (2005) claim that. verbs like song ‘give’ have two lexical entries: a two-argument verb and a three-argument verb.. For some speakers, if these verbs do not strongly denote the. meaning of ‘giving,’ i.e. [－Giving], they will add gei before the IO.. Consider the. following examples. (8) a. Wo song I. le. send. yi. xie shu.. (Chang 2005: 47). ASP one CL book. ‘I sent some books.’ b. Wo song I. le. send. yi. xie shu. gei ta.. ASP one CL book. GEI he. ‘I sent some books to him.’ In other words, when these verbs are two-argument predicates, they behave like the gei-required verbs discussed above and need to insert gei to introduce the IO4. 4. Generally speaking, these verbs are involved in physical transfer of objects just like sentence (5) c.. Sometimes they are used to denote abstract transaction as (i). However, when they are involved in metaphorical uses, i.e., [+Giving], they usually occur in the same patterns as the gei-forbidden verbs. The insertion of gei in such uses, as in (ii), is not common, although it is considered acceptable for some native speakers.. Therefore, when involved in abstract transaction, the Gei-insertion rule is. blocked. (i) Wo song ta I. give he. yi. ge. one CL. wen. kiss. ‘I gave him a kiss.’ 24. If.

(36) speakers consider these verbs have the meaning of ‘giving,’ i.e. [+Giving], they will not insert gei before the IO.. When they are used as three-argument predicates, they. subcategorize for Agent, Patient and Goal. not apply.. In this case, the Gei-insertion Rule need. Since these verbs can be used either with or without the insertion of gei,. they are classified as the gei-optional verbs, i.e. GOV. In addition to the three types of verbs, there exists a fourth type as in (9): (9) Mali shou. le. Mary receive. Yuehan. ASP John. yi. ge. liwu.. one CL present. ‘Mary received a present from John.’ This type of verbs is different from the above-mentioned three types in that Yuehan ‘John’ is not a Goal but a Source.. Li (1990) and Yang (1991) called this type source. object verbs in contrast to the double object verbs whose IOs are Goals.. Although. they are also considered double object verbs (Tang 1979, Li 1990), the present study only focuses on those verbs whose IOs are Goals5. So far we have discussed three types of verbs: (1) Gei-required verbs, [－Giving], (2) Gei-forbidden verbs, [+Giving], and (3) Gei-optional verbs, [+/－Giving], which were summarized in Table 2-1 with semantic features they possess.. (ii) ? Wo song I give. yi. ge. one CL. wen gei. ta.. kiss gei. he. ‘I gave him a kiss.’ 5. Readers who are interested in source object verbs may refer to Tang (1979), Li (1990) and Yang. (1991) for more details. 25.

(37) Table 2-1 Features of the Three Types of Verbs Type. Giving. Examples. GRV. －. dai ‘bring,’ ji ‘send,’ xie ‘write,’ mai ‘sell,’ diu ‘throw,’ liu ‘leave something for someone,’ ban ‘move,’ song ‘give,’ fu ‘pay,’ huan ‘return,’ shang ‘bestow’. ＋. GFV. gausu ‘tell,’ huida ‘answer,’ daying ‘promise,’ wen ‘ask,’ qingjiao ‘ask,’ jiao ‘teach’. GOV. ＋/－. song ‘give,’ fu ‘pay,’ huan ‘return,’ shang ‘bestow,’ jie ‘lend,’ pei ‘compensate,’ shu ‘lose’. In what follows, double object verbs in Mandarin Chinese will be discussed with their patterns. 2.2.1 Gei-required Verbs (GRV) Generally speaking, after the Gei-insertion rule, the IO can either precede or follow the DO (Tang 1979), resulting in the two patterns [V-DO-gei-IO] and [V-gei-IO-DO] below. Pattern 1: V-DO-gei-IO The first pattern for GRV is [V-DO-gei-IO], where the DO precedes the IO, as in (10) a. sentence.. The opposite order, [V-IO-gei-DO] as in (10) b, results in an ungrammatical In other words, gei is a Goal marker, so it has to take the Goal, IO, as its. complement NP (Chang 2005, Her 2006).. The pattern [V-DO-gei-IO] is similar to. the English to-dative counterpart where the gei-NP is an Ajunct PP (cf. Chang 2005).. 26.

(38) (10) a. Lisi Lisi6. ti. yi. ge. qiu gei ta.. (Her 2006: 1283). kick one CL ball GEI she. ‘Lisi kicked a ball to her.’ b. *Lisi Lisi. ti ta. gei yi. ge. kick she GEI one. qiu. (*V-IO-gei-DO). CL ball. ‘Lisi kicked a ball to her.’ In (10) a, yi ge qiu ‘one ball’ is the DO, and ta ‘she’ is the IO and also the Goal.. The. verb ti ‘kick’ originally is a two-argument predicate, so it does not subcategorize for a Goal (Chang 2005), as in (11). (11) A: Ni you. If a Goal is added, gei must be obligatorily present.. gang-gang ti le just. sheme?. kick ASP what. ‘What did you just kick? ’ B: Wo ti. le. yi. ke. qiu.. I kick ASP one CL ball ‘I kicked a ball.’ In (11), wo ‘I’ is the subject; yi ge qiu ‘one ball’ is the DO.. Since there is no Goal,. the Gei-insertion Rule need not apply.. Pattern 2: V-gei-IO-DO The second pattern for GRV1 is [V-gei-IO-DO], where gei appears immediately after the verb to form a V-V compound (Li 1990, Chang 2005, Her 2006).. The. pattern [[V-gei]-IO-DO] is regarded as a double object construction in Mandarin Chinese (Chang 2005) where the IO precedes the DO as in (12) a.. 6. The opposite. In Her (2006), Lisi was translated as ‘Lee’ in English. However, for the sake of consistency in the. study, we translate it as ‘Lisi.’ 27.

(39) order of IO and DO, [[V-gei]-DO-IO] results in ungrammaticality, as in (12) b. (12) a. Zhangsan pao gei Zhangsan. Lisi. yi. toss GEI Lisi. shu hua.. (Huang and Ahrens 1999: 6). one CL flower. ‘Zhangsan tossed a bouquet to Lisi.’ b. *Zhangsan. pao gei. yi. shu hua. Zhangsan toss GEI one CL flower. Lisi. Lisi. ‘Zhangsan tossed a bouquet to Lisi.’ In (12) a, pao gei is a V-V compound, Lisi is the IO and yi shu hua ‘a bouquet’ is the DO.. This construction is similar to [V-IO-DO] in English.. 2.2.2 Gei-forbidden Verbs (GFV) Verbs with a [+Giving] feature are Gei-forbidden verbs, and typical examples of this type are gausu ‘tell,’ and huida ‘answer,’ as in (13). (13) a. Xiaoming Xiaoming. gaosu. Xiaohua yi. ge. mimi.. tell. Xiaohua one CL secret. ‘Xiaoming told Xiaohua a secret.’ b. *Xiaoming Xiaoming. gaosu. gei Xiaohua yi. ge. mimi.. tell. GEI Xiaohua one CL secret. ‘Xiaoming told Xiaohua a secret.’. Pattern 3: V-IO-DO For GFV, the DO always follows the IO, as can be seen in (14) a, which shows a grammatical pattern [V-IO-DO], not [V-DO-IO] as in (14) b: (14) a. Mali. gaosu. Mary tell. Yuehan. yi. ge. hao. John. one CL good. ‘Mary told John a piece of good news.’ 28. xiaoxi. news.

(40) b. *Mali gaosu. yi. Mary tell. ge. hao. one CL good. xiaoxi. Yuehan.. news. John. ‘Mary told John a piece of good news.’ According to Givón’s topicality hierarchy discussed in Chapter One, the “primary topic” occupies the subject position while the “secondary topic” usually occurs in the DO position.7 As for the IO (“prepositional object” in Givón 1984) and the DO (“accusative object”) in double object constructions, Givón argues that the IO is usually the more “topical” one and thus is promoted via Dative Shifting to DO position in English.. Therefore, according to Givón’s theory, we may assume that in. double object constructions, the IO has to occur before the DO. Precedence Rule, as in (15).. This is called the IO. To be more specific, for the gei-forbidden verbs to. which the Gei-insertion Rule cannot be applied, the IO Precedence Rule must apply, resulting in the pattern [V-IO-DO].. In other words, since gei does not occur with. GFV in Mandarin Chinese, they must obey the IO Precedence Rule. (15) IO Precedence Rule When gei does not co-occur with Chinese double object verbs, the IO always precedes the DO. 2.2.3 Gei-optional Verbs (GOV) The only difference between GRV and GOV is that for the latter, gei can be omitted.. Verbs such as song ‘give,’ fu ‘pay,’ and huan ‘return,’ are of this type.. As. for the order of the two objects, the IO can either precede the DO or follow it, just like GRV discussed in Section 2.2.1. Therefore, these verbs can occur in the patterns, [V-gei-IO-DO], and [V-DO-gei-IO]. 7. On the other hand, if gei is omitted, there is. According to Givón, Dative Shifting in English, i.e., the promotion of prepositional object to DO, is a. “discourse-pragmatic device” (1984:153). 29.

(41) only one order allowed; the IO must precede the DO, i.e. [V-IO-DO].. Therefore, gei. is obligatory in the dative constructions, but is optional in the double object counterparts. Examples (16) a and (16) b show such optionality. (16) a. Lisi. song. gei Xiaoling yi. zhi shoubiao.. Lisi. give. GEI Xiaoling one CL watch. ‘Lisi gave Xiaoling a watch.’ b. Lisi. song. Xiaoling yi. zhi shoubiao.. Lisi. give. Xiaoling one CL watch. ‘Lisi gave Xiaoling a watch.’. Table 2-2 summarizes the rules we previously proposed with the corresponding patterns.. For GRV, i.e. [ －Giving], the Gei-insertion Rule must apply.. application of the IO Precedence Rule to these verbs is optional.. The. On the contrary, if. the Gei-insertion Rule does not apply, then the IO Precedence Rule must apply, like GFV, [+Giving]. Table 2-2 The Application of the Two Rules with Verb Types －Gei-insertion Rule. +Gei-insertion Rule Rules +IO Precedence Patterns [V-gei-IO-DO] Types. [－Giving]. －IO Precedence. +IO Precedence. －IO Precedence. [V-DO-gei-IO]. [V-IO-DO]. NA. [－Giving]. [+Giving]. NA. Note: For the fourth type where none of the two rules is applied, there is no such verb or pattern in Mandarin Chinese, hence not applicable.. If children have acquired the two rules, they will have no difficulty in acquiring the three patterns of double object verbs.. 30. When gei is missing, based on the IO.

(42) Precedence Rule, they will know that the only pattern allowed is [V-IO-DO] instead of [V-DO-IO].. If they produce sentences with the IOs following the DOs, then the. Gei-insertion Rule must apply and the pattern will be [V-DO-gei-IO].. As for the. [V-gei-IO-DO] pattern, it should cause no difficulty since for Chinese double object verbs, as long as gei is present, the phrase [gei-IO] can occur either before or after the DO.. 2.3 Empirical Studies of Dative Constructions Empirical studies concerning dative and ditransitive constructions have been conducted not only in English but also in other languages with various research purposes.. The present study aims to investigate whether age plays a crucial role.. Therefore, studies on age of acquisition of double object verbs were discussed (Osgood and Zehler 1981, Campbell and Tomasello 2001).. In addition, children’s. preference as well as their performance on either dative or double object constructions were also reviewed (Osgood and Zehler 1981, Mazurkewich 1984, Sadakane and Koizumi 1995, and Cho et al. 2002).. The issue of markedness concerning the dative. and double object constructions in previous studies was examined (Mazurkewich 1984).. Finally, one study on children’s acquisition of double object verbs in. Mandarin Chinese was reviewed (Chung and Gordon 1998).. 2.3.1 Osgood and Zehler (1981) Osgood and Zehler examined the effects of prototypicality and sentence complexity on children’s acquisition of English double object constructions.. In. addition to the two factors, children’s performance on dative and ditransitive constructions was also the main focus of their study. 31. They proposed that children.

(43) first should use knowledge-of-the-world as a basis of language development.. That is. to say, they learn how people and things interact, which constitutes their knowledge, and the knowledge, or contextual clues, in turn will play an important role in their acquisition of syntactic structures.. English double object constructions can serve as. the basis for testing this hypothesis, since these structures exhibit various semantic changes in the source, recipient and transferred objects.. According to Osgood and. Zehler, the most typical events of transfer involve [+Human] source, [+Human] recipient, and [－Animate] objects.. The two features, humanness of actors and. animacy of transferred objects, serve as two variables that result in various levels of prototypicality in sentences.. They proposed that the to-dative, S-V-DO-to-IO, is the. Basic, and the ditransitive, S-V-IO-DO, is the Transform structure. Their subjects consisted of sixteen children in each age group, three-, four-, and five-year-olds. Thirty-two sentences were examined, sixteen of which were dative and the other sixteen ditransitive. levels of complexity were employed.. Four levels of prototypicality along with four The children were asked to perform two tasks;. the first was an acting-out task, and the second, a production task.. In the first task,. each child was asked to perform with dolls actions described by the experimenter.. In. the second task, the child had to describe what the experimenter did with dolls.. In. both tasks, one point was given for a correct response.. In the production task, all. correct responses were divided into either dative or ditransitive constructions. The results showed that, at the first complexity level, which involved one transfer event and one transferred object, no age effect was found.. In other words, at. the age of three, the children had already acquired the two alternating constructions. Second, animacy played a crucial role in their children’s performance on both tasks. Third, when the children had choices to describe actions of transfer in the production. 32.

(44) task, they displayed overwhelming preference for the Basic constructions.. If they. used the Transform structure, they tended to use it in sentences with typical situations. Therefore, prototypicality, like complexity, was a crucial factor in the children’s performance. more often.. As the children got older, they tried to use the Transform constructions Fourth, the results of the comprehension task showed that as a sentence. became more complex and less typical, the children’s performance decreased for the Transform structures much more than their performance on Basic structures.. That is. to say, the Basic structures were easier for the children than the Transform structures. To account for this, Osgood and Zehler claimed that the children tended to treat the noun closest to the verb as the DO instead of the IO.. Furthermore, the sequence of. V-DO in the Basic form expressed contiguity, while in the Transform sentences, the IO separated the verb and the DO. To further support their results, they adopted Osgood’s (1980) cognitive perspective and claimed that the transferring action and the transferred object should display “natural” linkage; the sequence [V-IO-DO], however, should be less natural. In fact, the results of the children’s performance for the comprehension task can be explained in another way.. The children did better on the Basic structures than the. Transform because in the Basic structures, i.e. [V-DO-to-IO], not only the DO receives case from the verb, but also the IO is assigned abstract case by the preposition to.. In the Transform structures, i.e. [V-IO-DO], however, we have to. account for the fact that there are two NPs but only one case assigner, the verb, is present, and this may cause difficulty in Osgood and Zehler’s children’s acquisition of the structure.. Therefore, the Basic structure, [S-V-DO-to-IO], is earlier in. acquisition than the Transform, [S-V-IO-DO].. Furthermore, although they claimed. that sentence complexity may pose difficulty in their children’s performance, it only. 33.

(45) means that sentence complexity is closely related to processing. necessarily mean that the children had not acquired the structures.. It may not. In other words,. the same complexity levels may also pose difficulties in adults’ processing of these sentences, even though adults have already acquired the two structures.. 2.3.2 Mazurkewich (1984) Mazurkewich’s research on the dative alternations was based on the Markedness Theory.. Instead of the acquisition of first language, she examined learners’. acquisition of English as a second language, with a view to exploring second language acquisition process in comparison with that of first language acquisition. First of all, Mazurkewich began with two approaches commonly employed in previous studies dealing with dative alternations in English: a transformational and a lexical approach.. Transformational approach employed the Dative Movement Rule. to account for dative verbs allowed in double object constructions.. The lexical. approach stated that alternating verbs have two subcategorial features [NP PP] and [NP NP], while non-alternating verbs only have one feature [NP PP].. The author. favored the lexical approach for the reason that if children acquired dative alternations through the Dative Movement Rule, overgeneralizations would be expected in abundance, which, however, was not true. Mazurkewich’s view was later supported by the results of her study. Following the two approaches was the definition of the marked and the unmarked structures of the two constructions.. Following Chomsky’s principles of. UG, she suggested that the dative structure [NP PP] should be the unmarked form for the following reasons.. The first is from the perspective of productivity.. That is,. most dative verbs exhibit the [NP PP] construction while only a subgroup of these 34.