Government Phonology: Evidence From Wu And Cantonese

This thesis analyzes data relevant to syllable structure mainly taken from Wu and Cantonese, which offer evidence to support ON-template in Government Phonology (henceforth GP).The basic assumption in GP is that: phonology and syntax are governed by the same universal grammar. It doesn't have rule any more, but limited universal principles, parameters and constraints.The research of this thesis shows: there are different opinions upon syllable internal structure in the Chinese phonological history, especially on the question what syllable positions do the glides [j], [w] and word-final nasals occupy. However, no matter how the traditional Chinese syllable structure looks like, it has no use in explaining some phonological phenomena. It leads to a most marked structure and at the same time violates Coda Licensing Principle when explains the"long diphthongs"in Cantonese. Yet in the analysis of ON-template, these problems disappear, because a"long diphthong"is treated as a long vowel followed by a glide which is the following onset licensed by an empty nucleus. GP also treats word-final segments and checked segments as onsets licensed by the following empty nuclei. In GP, a word-internal glide is treated as part of a branching onset and a word-final glide is treated as an onset. Empty positions include empty onsets and empty nuclei. The zero-initials in Mandarin Chinese are empty onsets; the empty positions which are filled by inserted vowels of loanwords in Mandarin Chinese and Cantonese are empty nuclei. There are many examples of the adjacent consonant and vowel without forming a"syllable"and many examples of non-nasal-assimilation examples in Chinese dialects. These phenomena can not be explained reasonably in the traditional syllable structure; while they can be interpreted naturally, logically and elegantly in ON-template in GP.Meanwhile, the analysis of some relating data in this thesis also proves that ON-template in GP is better than the other syllable structure assumptions, for on the one hand it has no rules, therefore it simplifies the derivation process, on the other hand it can describe and analyze some relating data in a more convincing way.