The perception of speech and nonspeech analogs by first‐ and second‐language speakers.

Abstract

There are marked differences in how native speakers of Sinhala and English perceive the English /w/‐/v/ distinction; Sinhala speakers who have learned English as a second language are typically near chance at identification and have less than half the discrimination sensitivity of native English speakers. This poor discrimination ability is remarkable because the acoustic cues for /w/ and /v/ are varied, being contrastive in formant frequencies, amount of frication, and amplitude envelopes. The present project explored these differences in auditory sensitivity by manipulating speech and nonspeech stimuli, with the aim of investigating how close an auditory stimulus needs to be to natural speech in order for cross‐language perceptual differences to emerge. Synthesized VCV speech stimuli were created to model natural recordings. Nonspeech stimuli were created by removing dynamics from the stimuli (e.g., flat pitch and amplitude, no formant movement), producing a “buzz” that was acoustically similar to the speech stimuli. Discrimination results for these stimuli will be reported for native speakers of Sinhala and English, in order to evaluate whether cross‐language specialization for speech may occur at a precategorical auditory‐phonetic level of processing, or whether specialization is contingent on the stimuli being perceived as intelligible speech.