Disentangling lexical decision criterion shifts from phonetic category changes in natural accent adaptation

Le résumé fourni par la source

Listeners readily accommodate natural accent variation even when it deviates from their expectations. For example, American listeners endorse Mandarin-accented /θ/ (often produced as /s/, e.g., anthem→ ansem) after talker-specific exposure. Prior work debates the locus of perceptual flexibility: should this adaptation be attributable to lexical decision criterion shifts (H1), or updated category representations (H2)? Because previous studies typically employ a single task that is either biased towards lexical (e.g., word/nonword judgements) or prelexical (e.g., category identification) processing, it is difficult to make mechanistic conclusions unconfounded by task demands. To disentangle task demands from accent exposure effects, we query both lexical decision [LD] and category identification [CatID] of Mandarin-accented speech from the same listeners. Between subjects, we manipulate task order and contrast-biasing condition (/θ/-biased [e.g., thief] vs. /s/-biased [e.g., sudden]). Adaptation, signalled by a difference between biasing conditions, is expected in both task orders. Critically, if adaptation is driven purely by bottom-up changes in category representations (H2), no task order effect is expected. In contrast, if adaptation is mediated by a lexical decision criterion shift (H1), then we expect larger adaptation effects in categorization when it is preceded by LD tasks.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.