Perception and recognition of English /s/ and /ʃ/ with varying acoustic-auditory contrast

Abstract

Seminal work (Newman et al., JASA, 2001) found that listeners’ responses to talkers with more variable fricative productions were slower, though listeners’ ability to categorize the varied fricatives was robust. The current study selects North American English-speaking talkers with /s/ and / ʃ/ productions that varying in the magnitude of the acoustic-auditory contrast. With selected speech samples, listeners were asked to categorize (1) the isolated fricative (C- only), (2) the fricative-vowel sequence (CV), or (3) to complete a speeded-shadowing task where listeners were auditorily presented with the full words and asked to identify the words by repeating them as quickly and accurately as possible. Data were analyzed with Bayesian methods. The fricatives from talkers with greater contrast were identified more accurately and more quickly, with a greater effect size for the C-only condition and /s/ productions. This suggests listeners leverage information from the formant transitions to differentiate these fricatives. The speeded-shadowing results indicate the participants are faster at identifying the words with less acoustic-auditory contrast. This is the opposite of the expected pattern. Coupling C-only, CV, and word-level responses paints a more accurate picture of how talker differences in auditory- acoustic contrast affect categorization and intelligibility.