Search

The purpose of doing this study is to identify the velar nasal distribution and to explain the phonological phenomena relating to the sequence ng either in words or suffixation from different angles which are based on a harmonic approach. For this, first we consider rule-based analyses such as boundaries (Wolfram and Johnson 1981), a sonority restriction (Roca and Johnson 1999), and lexical levels (Borowsky 1986). However, those showed some limitations of explanation as to the interrelation of g deletion and syllabification closely related to the sequence ng. As an alternative, we examined the constraints-based analyses depending on some constraints such as Max (More-Salient C) >> No Complex Coda >> Max (Less-Salient C) (Lee 1997) or parochial constraint (long([+tense])) and analogical constraint (OO(long, song; ɔ) & OO(long, song; ŋ)), etc. (Green 2001). But those explanations were complicated due to a lack of conciseness as well as either biased towards the deletion of the final consonants or the contexts of vowels occurring before the velar nasal. Therefore, in a harmonic approach, we could explain the issues (g deletion and syllabification of g) relating to the sequence ng more concisely through foot structures in level constructions than through the other analytical methods mentioned.

This paper explores the characteristics of ambisyllabicity in order to investigate a motive for it. After reviewing previous explanatory methods, we suggested a new approach based on prosodic structure, and through the new representation, we can account for the sonorant devoicing and the difference of segment duration in gemination by ambisyllabicity. First, we critically reviewed some approaches depending on an ambisyllabic rule and the syllable weight principle, but those analyses did not provide a satisfactory explanation regarding the reason for the intervocalic consonant to be treated as an ambisyllabic consonant. However, when we adopt the moraic structure, we can show ambisyllabicity well in the correlation between syllable weight and stress. As regards flapping and ambisyllabicity, which are regarded as occurring in the same environment, to the contrary, in this paper, we can recognize that the environments of each of ambisyllabicity and flapping are not always the same. Also, with respect to sonorant devoicing, when the intervocalic consonants are treated as ambisyllabic, we can well account for the reason why the sonorant devoicing occurs. Finally, with respect to gemination, although gemination has the same environments as ambisyllabicity, we can demonstrate that, first, gemination phonologically derived across a word boundary results in ambisyllabicity, and second, geminates are not the same in segment duration depending on the ambisyllabic structures. Thus we can say that geminates do not exhibit the same characteristics as ambisyllabicity, and geminates are longer in segment duration than a singleton.