The role of sentence context in perceptual learning of speech

Abstract

A key issue in speech perception is its lack of invariance: mappings between acoustic features and phonemic categories are not one-to-one. Listeners can address this problem by adapting their perception to a talker's idiosyncratic accent. Previous studies have used lexical cues to examine adaptation, but in everyday speech talkers use full sentences. We investigated whether semantic context in full sentences occurring either before or after an acoustically-manipulated target word can trigger adaptation. We found that sentence context can be used for adaptation, but findings are mixed on whether the timing of context impacts learning.