Auditory distinctiveness and the perception of voiceless stop consonants in continuous speech

Abstract

Explaining the way in which listeners cope with the apparent lack of acoustic invariance as well as the phonetic ambiguity frequently encountered in continuous speech has remained problematic for models of speech perception. The voiceless stop consonants /p t k/, for example, are frequently unreleased word-finally. Despite the reduction in articulation and presumably in the distinctiveness of the acoustics signal, listeners appear quite accurate in identifying the correct stop consonant during conversation. Sets of words such as {shop, shot, shock} and {cheap, cheat, cheek} should, given the absence of release of the final stops, be difficult (if not impossible) to distinguish in isolation. In this study, listeners were presented with such lexical stimuli (in isolation as well as in semantically priming sentences) and tested for their perception of the word-final [p t k]. It was found that the amount of distinctive information carried in the acoustic signal for this class of sounds in the environment of interest was insufficient to consistently disambiguate listener impressions; further evidence for lack of perceptual salience is provided by an extremely weak resistance to semantic context effects. The results of this study highlight the value of an auditory dimension to perceptual studies in linguistic phonetics.