Vowel onset as a rhythmic marker in Japanese single and geminate stop distinctions

Abstract

Hirata and Whiton (2005) found that the durational ratio of stop closure to a disyllabic word was the best parameter to distinguish singleton and geminate stops in spoken Japanese. However, given that vowel onset has perceptual and psychoacoustic importance as a temporal marker (Kato et al., 2003), the best unit may be from the first vowel onset of the target word to that of the following word in a carrier sentence (= Vword). In this study, 36 target disyllabic pairs including singletons and geminates (e.g. [buka] and [buk:a]) were embedded in the sentence [sokowa ____to jomimasu], spoken by four native Japanese speakers at three rates (Hirata and Whiton, 2005). The duration of Vword (e.g. the beginning of `u' in [buka] to the beginning of `o' in [to]) and of Vc(c) (e.g. `uk' from [buka] and `ukk' from [buk:a]) were measured, and the Vc(c)/Vword ratio was calculated. We examined how accurately this ratio classified singleton and geminate tokens across all rates and speakers. The classification accuracy was found to be 97.3-98.0%. This indicates that the duration of Vc(c) relative to Vword is a stable parameter for quantity distinction, and that the Vc(c) unit may constitute a meaningful rhythmic marker in production.