Underlying Tonal Representation of Kagoshima Japanese Accentual Contrast Author

Kagoshima Japanese (KJ) is a so-called two-pattern Japanese dialect which exhibits a two-way accentual contrast. This two-way accentual contrast is [L(n)HL] (i.e. sakura LHL ‘cherry blossom’, kagaribi LLHL ‘watch fire’) and [L(n)H] (i.e. usagi LLH ‘rabbit’, kakimono LLLH ‘document’), where n = zero or a positive integer. Although the details of the KJ accentual contrast can be easily expressed, as seen from the above examples, different phonological approaches have produced different analyses of KJ’s accentual contrast. Some treat KJ as a pitch-accent dialect like Standard Japanese (SJ), whereas others treat KJ as a tone dialect (McCawley, 1970; Haraguchi, 1977; Shibatani, 1979; Hayata, 1987; Smith, 1997). Based on the acoustic-phonetic descriptions of KJ’s tonality using the z-score normalisation technique, Ishihara (2004) discussed the surface tonal representations of KJ’s accentual contrast using the Autosegmental-Metrical (AM) model of intonation (Pierrehumbert and Beckman, 1988). In the current study, I use the AM model to discuss the underlying tonal representations of KJ’s accentual contrast with particular focus on the derivational process leading to the surface tonal representations and the fundamental frequency (F0) differences of KJ’s accentual contrast. Using the lexical tonal representations of KJ’s accentual contrast as evidence, I demonstrate that in a comparison of SJ and KJ, KJ is closer to proto-typical tone languages. Finally, I argue that the F0 realisation differences of KJ’s accentual contrast can be phonetically explained by the proposed underlying representations