| This thesis is the first to analyze the segmental phonology of Cantonese within the theoretical framework of Government Phonology (GP). It mainly aims to formulate the distributional constraints for phonological elements in Cantonese, and based on these constraints program a computer procedure to generate all the grammatical Cantonese 0-N pairs. I first give a brief introduction to the key concepts and theories of GP, which is based on universal principles and language-specific parameters. Next I review the traditional analyses of Cantonese phonological system. I find out that the literature confines itself to the structural phonemic model. All the discussions and controversies are centered on so-called nitial?n-glide?taody?and ff-glide? Within this model, explanation for ungrammatical yllables?can never be provided. In the third part I present a formal analysis of the segmental phonology of Cantonese in terms of the universal representations provided by GP. I explore the onset and nuclear inventories and their phonological expressions, examine how phonological elements spread in Cantonese, and formulate a series of distributional constraints for Cantonese phonological elements. I assume from its constrained distribution that is not a lexical nucleus in Cantonese. It is the result of the element spreading from the adjacent onsets into an empty nucleus. I also draw a comparison between Cantonese and Putonghua in a number of aspects from the GP perspectives. The computer program is theoretically based on the Free Distributional Principle. I choose the algorithm of ufirst eliminate, then combine?from a few alternatives. The key step is to utranslaten the distributional constraints into computer language, so that the ungrammatical 0-N pairs will be filtered out by the restrictions, and only the grammatical ones are generated by the procedure. In the concluding part I discuss some problems of my study and indicate the potential of developing computational analyses in phonological study.
|
PDF Full Text Request