A unified approach to nasality and voicing

This thesis proposes a merger of voicing and nasality under a single phonological feature. One main focus of this approach is a paradoxical phenomenon involving a nasal-voice affinity in Yamato Japanese. In postnasal voicing assimilation, nasals appear to be specified for voice (e.g. sin + ta-*' sinda ‘died’). On the other hand, in Lyman's Law, which allows only a single voiced obstruent in a particular domain, nasals behave as if they have no voice feature (e.g. sindo-i ‘tired’, *zindo-i). In order to resolve this paradox, I propose that nasal and voice are phonetic manifestations of the same phonological category. The different phonetic interpretations can be accommodated within a geometry-based element theory which incorporates the notion of a complement tier (Backley 1998, Backley & Takahashi 1998). When the complement of the nasal-voice element is licensed, the realisation is one of voicing. An unlicensed complement tier results in nasality. By adopting this analysis, postnasal voicing assimilation can be treated as the extension of the nasal-voice element across both positions of an NC cluster, where only the element in the second position licenses its complement tier. In addition, the transparency of nasal stops to Lyman's Law is made to follow from the element failing to license its complement tier. The validity of the proposed representations finds further support in the analysis of spontaneous prenasalisation and spontaneous velar nasalisation found in several dialects of Japanese. In accordance with the principle of Licensing Inheritance (Harris 1994, 1997), both processes are regarded as lenition of the complement tier in a prosodically weak position. This avoids the need to rely on arbitrary notions such as lexically floating nasality or nasal insertion rules which have been proposed elsewhere in the literature. Furthermore, in conjunction with the predictions of phonological licensing (Kaye 1990; Harris 1994, 1997), the representations successfully accommodate recurrent assimilatory phenomena such as nasal harmony and voicing assimilation.