A Review： The Analysis of Retained Object Construction In Mandarin Chinese Based on Case Theory

Xie+Lixian

【分類号】H146

1. Introduction

Retained object construction in Mandarin Chinese is a remarkable existence， where a verb that cannot take an object according to GB theory is actually followed by a Np. Appeared to be violating the case filter， the construction is grammatical. In previous researches two types of retained object constructions have been elaborately discussed： the Possessor-Subject Possessee-Object Sentence， abbreviated as PSPOS， and the passive sentence with "bei". This article is a review of the relevant studies and a discussion of the defects of the theories.

2. The Two Types of Retained Object Construction

PSPOS and passive sentence with “bei” are the two types of Retained Object Construction mostly discussed. Sentence （1） is an example of PSPOS：

（1） Zhangsan diao-le bijiben.

And sentence （2） is an example for passive sentence with "bei".

（2） Zhangsan bei tou-le qian.

Xu Jie （1999） summarized the four common features shares by the two types of constructions： a. In both types the subject is the possessor and the object is the possessee. b. In both types the object can be moved in front of the verb and behind the subject， with the word “de” between the two NP. Sentence （1） can be rewritten as Zhangsan de bijiben diao le. and sentence （2） can be rewritten as Zhangsan de qian bei tou le. c. In both structures， the object cannot be moved in front of the subject. e.g. Bijiben zhangsan diao le. or Qian Zhangsan bei tou-le. are not grammatical. d. Neither of the verbs in the two constructions are case-assigners.

Because of the four common features， the two kinds of constructions may be similar in their D-Structure and have been derived under similar processes.

3. Approaches to the explanation of the detained object construction

3.1 The possessor raising approach

The possessor raising Approach employs the Unaccusative hypothesis & Burzio's Generalization to account for this phenomenon. According to Unaccusative hypothesis， the verbs in passive and raising structures are unaccusative. To be licensed， the complement of the unaccusative verb has to move to [Spec， IP] to take the Nominative Case from INFL. The possessor raising approach viewed PSPOS as a kind of raising structure. The D-Structure of sentences （1） & （2） can be illustrated as：

（3） D-Structure： [IP[VPdiao-le [NPZhangsan de bijiben]]]

（4） D-Structure： [IP[VPbei tou-le [NP zhangsan de qian]]]

Because the verbs in both sentences are unaccusative， their complement has to move to [Spec， IP]. The move results sentences （5） & （6）.

（5）[IP [NP Zhangsan de bijiben i][VPdiao-le ti]]

（6）[IP [NP Zhangsan de qian i][VPbei tou-le ti]]

To account for retained object construction， Xu jie （1999） developed the hypothesis and proposed that for retained object construction， only the possessor NP is moved to [Spec， IP]， the possessee NP is retained in its base position and receives Partitive Case assigned by the unaccusative verb.

There are some obvious problems with this hypothesis. Firstly， Partitive Case is taken by NP with indefinite denotation only， as discussed by Belletti （reviewed by Pan & Han 2008）. The hypothesis suggested that retained object should be indefinite NP. This goes against language facts. Pan Hanhua & Han Jinquan （2008） pointed out that in Mandarin Chinese， retained object can be definite NP with appropriate context， e.g. Zhangsan bei tou-le zuigui de nazhi bi. The object NP is a definite NP， but the object is retained and the sentence is grammatical. Secondly， it is impossible for a NP to take only Partitive Case but not Accusative Case. e.g. [NP The destruction the town] shocks all the citizens. This sentence is not grammatical because the NP the town has only Partitive Case. This proves that Partitive Case cannot license a NP. Thirdly， if a NP can get licensed by Partitive Case assigned by raising verb and passivated verb， it does not need to move at all. But this is against the basic principles of the Unaccusative hypothesis. Fourthly， by this approach， it appears that it is optional whether to raise the whole NP of the object to [Spec， IP] or to raise only the possessor NP and the object NP retains. But it is not universal. In English retained object structure is not allowed. E.g. * Zhangsan was killed father. is a ungrammatical sentence. The approach fails to answer why its hypothesis doesnt work in other language. Finally the hypothesis has violated some basic rules for NP-movement. Pan haihua & Huang Jinquan （2008） points out that： a. it violates the Left Branch Condition. b. it violates the Chain Uniformity Principle. c. it has not provided an explanation for the disappearance of "de". In conclusion， this approach is not very satisfactory for the explanation of the existence of retained objects.

3.2 Base generation Approach

Pan haihua & Han jinquan （2008） proposed another solution. They argued that the possessor NP in retained object structure is base-generated at [Spec， CP]. The D-Structure of the sentence （1） & sentence （2） is re-analysed as following：

（1） D-Structure： [CP Zhangsan [IP e [VP diao-le [NP bijiben]]]]

（2） D-Structure： [CP Zhangsan [IP [VP bei tou-le [NP qian]]]]

The possessor NP Zhansan is situated in an A'-position and don't need to be moved. The possessee NP is case-driven to move. The movement is completed in two steps.

Step1： The object move to [Spec， IP] position， where it is assigned Nominative Case. This step produces sentence （7） & （8）.

Sentence （7）： Zhangsan bijiben diao le.

Sentence （8）： Zhansan qian bei tou le.

Step 2： The possessee NP bijiben/qian move to the end of the sentence. The motivation of this step is to make end focus. By this step， the retained object construction that concerns us is produce.

The licensing Condition of the topic： According to Pan Haihua & Hu Jianhua， is that the NP at [Spec， CP] must have a possessor-possessee semantic relation with the theme of the sentence. However this condition seems infufficient. In sentence （9）， Zhangshan and fuqin has possessor-possessee relation， but the sentence is not grammatical. On contrast， in sentence （10）， the topicalization of the possessor NP is allowed. The topic base-generation approach needs to explain what makes such difference.

（9） Wo kanjian Zhangshan de fuqin le. / * Zhangshan wo kanjian fuqin le.

（10） Wo kanjian fanzi de neibujiegou le. /Fanzi wo kanjian neibujiegou le.

4. Conclusion

The attempt to explain the allowance of retained object by Partitive Case is not satisfactory. The base generated theory better accounted for the linguistic facts， but it needs to be improved on the licensing condition for tropicalized NP.

REFERENCES

Pan， haihua & Han， jinquan. （2008） Hanyu baoliu binyu jiegou de jufa shengcheng jizhi （the generative mechanism of retained-object constructions in Chinese）. Zhongguo yuwen（Chinese Language and Literature） 6.

Xu， jie. （1999） Some theoretical issues of the two types of Chinese 'retained-Object' constructions. Xiandai waiyu （modern foreign language） 1， 61-62.