抄録
Japanese voiced geminates have a tendency to devoice (e.g. baggu > bakku `bag'). Voiced obstruents have inherent susceptibility for devoicing due to the aerodynamic voicing constraints (AVC), and the susceptibility is higher for geminate obstruents than singletons. As a way to investigate how Japanese speakers realize the contrast between the [+/-voice] in obstruents, we examined oral and nasal airflow patterns during intervocalic voiced and voiceless stops in singletons and geminates. The results showed that no nasal airflow appeared during voiced and voiceless stops. Oral airflow showed asymmetry between single and geminate stops in realization of the stop voicing contrast. While the oral airflow pattern clearly differentiates the voiced vs. voiceless contrast in singletons, the patterns are similar in geminates. Acoustic signals also show the same asymmetry between the singletons and geminates. The observed convergence –- a clear voicing contrast in singletons vs. a lack of the contrast in geminates, both in oral airflow and acoustic signals, indicate the tendency of neutralization of the voiced geminates into voiceless ones. Our results support the idea of phonetic and articulatory bases in phonological patterning of voicing neutralization in Japanese geminate stops.
