This chapter examines the structure of relative clauses in Chinese and aims to develop and defend an analysis of relativization which will be made critical use of in the examination of shi-de cleft structures in Chinese in chapter 4. Section 1 of the chapter considers certain very general properties of relative clauses cross-linguistically and how these have been modeled in Chinese by standard Government and Binding/GB style analyses. Section two then goes on to examine the problem of directionality and headedness with regard to relative clause structures in Chinese, highlighting the fact that Chinese is an exceptional language which combines prenominal relative clauses with a basic V -0 word order. Concerning the 'relativizing' element de, it is argued that GB analyses of de as a complementizer are at odds with the general directionality of Category-selection in Chinese and that there is evidence that elements in C in Chinese do indeed select for rightward complements. In section 3.3 I then attempt to show how adopting a Kaynean approach to relativization may resolve these problems, following in part a discussion of relative clauses in Simpson (2002). This Kaynean analysis is subsequently compared with an alternative approach to prenominal relative clauses suggested in Murasugi (1998) for Japanese. I argue that the Kaynean account is preferable for Chinese for three essential reasons. First it allows for a more uniform account of other noun-complement clause structures which make use of the element de. Secondly it is able to resolve the headedness/directionality issue (which is not a problem of relative clauses in Japanese). Thirdly some of the critical features of relativization in Japanese which lead Murasugi to propose a fully basegenerated analysis of Japanese relative clauses are absent from relative clauses in Chinese. In section 3.4 I discuss the relevance of tone sandhi phenomena in Taiwanese for the account of relativization developed, indicating how it can be shown to provide interesting support for an analysis in which there are leftward clausal movement as part of the construction of relative clauses. Section 3.5 is an extension of the Taiwanese tone sandhi patterns discussed in section 3.4. The grammaticalization process of Taiwanese sentence-final "particle" -kong

adds further support for the leftward clausal movement proposed for relative clauses. Section 3.6 is a brief conclusion.