Delayed effects of speech and non-speech stimuli on sibilant categorization

Abstract

Adaptation to the speech of a novel talker can involve at least two types of mechanism: phonetic learning (e.g., Samuel & Kraljic, 2009) and spectral contrast (e.g., Lotto & Kluender, 1998). While phonetic learning persists over long time periods, auditory effects such as spectral contrast have been demonstrated to occur for sounds that are temporally adjacent or separated by brief delays (e.g., 1.3s; Holt, 2005). The present study examined whether phonetic and auditory mechanisms can be distinguished by introducing a substantially longer delay between exposure and test. Five exposure conditions were examined across participants: exposure to syllables beginning with a relatively high or low COG [z], exposure to white noise matched in LTAS to the high or low COG [z]-initial syllables, and a baseline (no exposure) condition. Following exposure, participants performed a one-back image repetition detection task for 15 minutes, and then categorized members of the same 10-point [ʃ]-[s] continuum over 6 blocks. In comparison to baseline, exposure to speech had a significant and expected influence on sibilant categorization (e.g., the [ʃ]-[s] boundary shifted toward [s] following high COG [z] exposure). Speech and non-speech exposures were compared to determine whether spectral contrast is a viable explanation for long-lasting sibilant adaptation.