Cumulative effects of phonetic context on speech perception

Abstract

Different speakers produce speech sounds differently. The phonetic context is known to facilitate the recovering of phonological categories from productions with talker variation. However, whether the context effect originates from central auditory processing or speech-related processing remains debated. It is worth noting that the context effect may be a combined effect, contributed by both auditory and speech-related processing. To investigate this question, we compared the effect of four types of contexts with incrementally more cues (nonspeech, reversed speech, meaningless speech and meaningful speech) on perception of Cantonese level tones. Results indicate that the context effect is a product of multiple levels of processing, with the primary contribution from phonological cues (meaningless speech context). The contribution of auditory cues is negligible, and that of phonetic cues and semantic+syntactic cues is both moderate. Phonological cues likely enable listeners to calibrate the acoustic-to-phonological mapping of speech sounds for each talker, facilitating the categorization.