Aspects of Kejia phonology

Abstract

This thesis is concerned with four aspects of Kejia phonology: initial consonants, diphthong representations and syllabification, the domain of tone sandhi, and the notion of geminate inalterability. The analysis is in line with the non-linear framework of generative phonology.

Two issues are investigated in terms of initial consonants: palatalization and onset friction. We conclude that the nasal palatal n has two UR's: n and ng. Moreover, we argue that the fricatives v and j are derived from high vowel spreading. As for diphthong representations, we propose that there are two different representations: one for a rising diphthong and one for a falling diphthong. In addition, we propose that there are at most three skeletal slots for a Kejia syllable. These proposals account for two generalizations: no consonant can follow a falling diphthong, and a falling diphthong differs from a branching rime. As far as the domain of tone sandhi is concerned, we postulate that it is defined by the syntactic structure on the basis of domain-c-command (Kaisse 1985). The last topic is gemination and geminate inalterability. We conclude that gemination in Kejia results from spreading and that there is no inalterability effect in Kejia geminates.