Voice offset characteristics for voiceless fricatives and stops

This paper is concerned with whether the nature and timing of the glottal abduction gesture might differ for voiceless fricatives and stops. The preconsonantal vowel in VCV sequences were examined, where C =/p(:) b(:) f(:) v(:)/ in French (single consonants), Italian, and Swedish (geminates). The speech pressure waveform was inverse filtered using interactive software and the LF voice source model [Fant et al., STL-QPSR 4/1985, 1–13 (1985)] was matched to the source signal. Results show a long breathy-voiced transition from a preceding vowel to the voiceless fricative, suggesting that glottal abduction occurs early. For the voiceless stops, however, results confirm a previous study [A. Ni Chasaide and C. Gobl, Lang. Speech 36, 303–330 (1993)] indicating that the timing of glottal abduction is quite variable and language/dialect dependent. A tentative conclusion is that early glottal abduction in voiceless fricatives may be a universal feature, necessitated by the dual production requirements of effecting devoicing with an open vocal tract and achieving the high oral pressure required for generating oral frication. This clearly is not the case for voiceless stops where the precise control appears to depend on language-specific rules. [Work supported by ESPRIT Project No. 6975.]