An acoustic and perceptual investigation of the geminate and singleton stop contrast in Japanese

Abstract

Segmental durations are known to be heavily influenced by changes in speech rate such that the durations of members of two contrasting length categories may overlap [Pickett et al. ‘‘Effects of speaking rate on singleton/geminate contrast in Italian,’’ Phonetica 56, 135–157 (1999); Hirata & Whiton, ‘‘Relational acoustic invariance in the single/geminate stop distinction in Japanese,’’ J. Acoust. Soc. Am. 117, 2569 (2005)]. Given this, questions arise as to how category distinction is achieved for contrasts signaled by differences in segmental duration. This study investigates the question by examining the production and perception of the Japanese singleton versus geminate stop contrast by adult native speakers. Six native Japanese speakers produced disyllabic words with medial singleton and geminate stops at three speaking rates. Duration of stop closure and a number of relational timings between the closure duration and other segments were examined as potential rate-independent measures distinguishing the two categories. The ratios of stop closure to the preceding mora as well as stop closure to the following vowel remained stable across speaking rate and accurately classified naturally produced tokens. Ten native Japanese listeners participated in an identification task to examine the perceptual relevance of these relational timings. The results indicated an influential role of preceding mora duration and a small role of following vowel duration. The results also suggested that nondurational factor(s) influence perception of the stop length contrast.