Voicing distinction for fricatives: Acoustic theory and measurements

Abstract

Theoretical analysis suggests that simultaneous generation of vocal-fold vibration and turbulence noise at a supraglottal constriction requires rather precise adjustment of the structures controlling the glottis, the intraoral volume, and the supraglottal constriction. The theory shows that there tends to be a reciprocal relation between voicing amplitude and the amplitude of the turbulence noise. It is not surprising, therefore, that vocal-fold vibration does not always occur throughout the constricted interval when a “voiced” fricative is produced, and that variation in the noise amplitude is often observed. Measurements of the amplitude of voicing (expressed as the spectrum amplitude in the region of the first harmonic) and of frication noise have been made for several fricative consonants spoken by several speakers in a number of phonetic environments, including utterance final and intervocalic position, and in clusters with other voiced and voiceless fricatives. Evidence for voicing can usually be found over some portion of the constricted interval, particularly in the vicinity of fricative onset or release. However, voicing may be weak or absent when the voiced fricative is followed in a cluster by a voiceless fricative. [Work supported in part by Grants NS-04332 and NS-15123 from the National Institutes of Health.]