Influence of lexical and acoustic context on phonetic categorization depends on listening situation

Abstract

Phonetic categorization is influenced by multiple sources of contextual information, but little is known about how different sources of information interact. We examined the relative influence of lexical versus acoustic contexts on phonetic categorization of sounds along [s]-[S] continua embedded in word-nonword pairs (e.g., a[S]amed-a[s]amed, ca[s]ino-ca[S]ino). These categorization targets were preceded by sequences of 12 nonspeech tones with mean frequencies a standard deviation above or below the spectral means of the endpoint fricatives. Listeners’ [s]-[S] categorization was influenced by lexical information, exhibiting a Ganong effect with categorization shifted toward responses consistent with words, and also by acoustic context. The effect of the tone sequence was spectrally contrastive; there were more [S] responses (low spectral mean) following higher-frequency tones and more [s] responses (high spectral mean) following lower-frequency tones. In addition, the influence of acoustic relative to lexical context was modulated by listening environment. When the informational load of the lexical context was low (four word-nonword continua) acoustic context exerted a relatively greater influence than when the informational load of the lexical context was high (forty word-nonword continua). Multiple sources of context interact to influence phonetic categorization and the relative influence of different information sources is flexibly modulated by listening environment.