Evidence for the role of acoustic boundaries in the perception of speech sounds

Abstract

Evidence from several different experiments in the past has supported the notion that cues for the perception of some of the phonetic features for consonants reside in regions of the speech signal in which there are rapid changes in the spectrum. This paper presents additional data that support this view for two features: The/s-š/ distinction and the nasal-stop distinction. In both experiments, continua of synthetic sounds were generated in which certain acoustic properties of the steady-state portion of the consonant (spectrum of fricative noise or amplitude of nasal murmur) were manipulated and acoustic events in the brief time interval immediately adjacent to the consonantal release were manipulated independently. Listener responses to these stimuli showed that the perception of the relevant phonetic dimension /s-š/ or nasal stop was influenced strongly by the acoustic properties adjacent to the consonantal release. The acoustic attributes during most of the consonantal closure interval could be manipulated over a wide range without changing the identification of the segment, as long as the attributes in the 50-odd ms in the region of the boundary were appropriate. [Work supported in part by a grant from NINCDS.]