A Study On Mandarin Sluices

This thesis makes an in-depth study on MS (Mandarin Sluices) with some relevant theories, based on a discussion of the following three research questions:(1) Is there any asymmetric phenomenon among the wh-remnants in MS?(2) What are the constraints on the correlates in MS?(3) Does Sluicing in Mandarin exhibit island repair effects?After making a review of some previous literature on Sluicing in English, Japanese as well as Mandarin, the thesis adopts Focus-Movement analysis to MS as overt movement does exist in Mandarin even though it is allegedly a wh-in-situ language. It maintains that a FP lies between CP and IP, wh-words moves to [Spec, FP] to check the strong [+focus] feature in the functional head F. Shi is a focus marker lying in the functional head F which licenses IP deletion.Concerning the three research questions, evidence shows that there exists an asymmetry between wh-arguments and wh-adjuncts in MS in that shi support is obligatory before wh-arguments but optional before wh-adjuncts. Further evidence shows that there also exhibits an asymmetry between zenmeyang'how'and the other wh-adjuncts in MS. The asymmetry lies in that zenmeyang can't feed Sluicing as the other wh-adjuncts do. The thesis suggests the reason lie in the different syntactic positions of them:zenmeyang is located in VP while the other adjuncts in IP. Thus the former gets deleted at PF.As to the second research question, Chung et al. (1995) argues that the inner antecedent (or correlate) of a sluice must be indefinite to provide a free variable for the operator to bind, thus wh-pronouns and weak DPs are licit correlates while proper name, strong quantificational DPs and demonstratives are not. Yet study shows that some'illicit correlates' -proper name and quantificational phrases can feed Sluicing. The thesis explains this phenomenon with Mats Rooth's (1992,1996) Alternative Semantics and Merchant's (1999) e-GIVENness:focused correlates and disjunctive correlates trigger choice functions which derive alternative sets and provides free variable for binding, obeying the Ban on Vacuous Variable Binding. In addition, the antecedent and Sluicing clauses in the problematic examples observe e-GIVENness, thus explaining the deviancy.With respect to the third question, research indicates that MS do exhibit island repair effects, but no such effect is observed in MS with implicit correlates. Further investigation reveals that D-linking plays an important role in island repair. At last, the author makes a summary of the whole thesis, in hopes of providing some useful evidence for further research on MS.