Noriko Kobayashi, Kiyoshi Honda
Production and Perception of the Consonantal
Voicing Distinction in Speech without VOT:
An Acoustic Study using Electric Larynx Speech
Abstract:Acoustic features and perceptual cues for the voicing distinction
of English intervocalic stops were investigated using two
alaryngeal talkers who speak with electric larynges after surgical
removal of the larynx. These talkers produced multiple
repetitions of /əCVC/ syllables. In the electric larynx speech, the
sound source was continuously generated throughout an
utterance, and systematic control of voice onset time (VOT) for
the intervocalic stops was not possible. Statistical analyses based
on the talkers' intention revealed different acoustic
characteristics for the two talkers, DB and AS. DB showed
increases in duration and amplitude of burst-like high frequency
noise following the consonant release more for the voiceless
stops than for the voiced cognates, suggesting some articulatory
efforts to generate voiceless stops. In AS, F1 and F2 were higher
and consonant closure was longer for voiceless stops than for the
voiced cognates. Correlation coefficient scores for the production
and perception of the talkers' speech suggested that perceptual
cues for the voicing distinction were different for the two
talkers. Acoustic features related to the production of burst-like
noise and pseudo-VOT (interval between the bylining of burst-like
noise and the following vowel) were highly correlated with
the perception of voiceless-ness in DB. Correlation scores were
not high in AS but shorter duration of the preceding vowel and
longer consonant closure were related to the perception of
voiceless-ness. The results suggested possible compensation
manoeuvres of electric larynx speech and individual variations
for encoding a phonetic feature.