Perceptual scale for voice onset time (VOT) in synthetic /bae/ /pae/ syllables

Abstract

The scale used by the perceptual system to represent variations in VOT appears to be different from the physical VOT scale. The perceptual scale, ranging from voiced to voiceless, has zero point at the phonetic boundary (the zero voiced‐voiceless point). The degree of voicedness (voicelessness) increases with the distance between the VOT value and the boundary. Further, the perceptual effect of a difference in VOT near the boundary is greater than at points remote from the boundary. A rescaling of the VOT continuum that is more consistent with the properties of the perceptual scale is given by the function Dxi = k |xi − xb |β,0 < β < 1, where xi is a VOT stimulus value, xB the boundary VOT, k a constant, and Dxi the perceptual magnitude of xi along the voiced (voiceless) scale. The perceptual distance, Δ D, between two stimuli xi,xi + 1, is given by the difference between Dxi and Dxi + 1; that is, Δ D = k ‖xi + 1 − xB|β − |xi − xB|β|. The power transformation at the perceptual level is suggested by the following: (1) the relation between cumulative disciminability and cumulative distance from the boundary, in log‐log coordinates, is well described by straight lines with positive slopes smaller than one. This is observed in /bae/ /pae/ discrimination data obtained in the present and in previous studies [Elman, J. Acoust. Soc. Am. 65, 190–207 (1979)]; (2) for a constant msec difference between xi and xi + 1 the function Δ D reaches a maximum when the pair straddles the boundary, and decreases monotonically as the distance between the stimuli and the boundary increases; (3) changes in discriminability associated with changes in the size of the VOT steps Of the /bae/ /pae/ correlate strongly with changes in Δ D.