Effects of speaking style and semantic predictability on vowel production

Listener-oriented accounts of phonetic enhancement propose that talkers produce enhanced vowels to increase clarity when their interlocutor might experience communicative difficulty, e.g., for non-native interlocutors or for an unpredictable word given the semantic context. While style-driven enhancement has been shown to be a listener-oriented modification, it is less clear if semantic predictability effects are implemented by the same mechanism. Using three listener conditions, in which listeners varied in language proficiency, and two predictability contexts, the current study investigated how talkers tailor their speech to accommodate the varying proficiencies of their listeners and to what extent semantic predictability effects can interact with listener-oriented processes. Talkers produced longer and more dispersed vowels when talking to the low-proficiency non-native listener relative to the native and high-proficiency non-native listeners. Further, vowels were longer in low predictability contexts relative to high predictability contexts. Finally, talkers increased vowel F2 in low predictability contexts relative to high predictability contexts only for the low-proficiency non-native listener. No differences were found for the native and high-proficiency non-native listeners. The findings suggest that assessment of listeners' needs affects phonetic enhancement, and listener-oriented processes may influence predictability effects for at least some acoustic measures in some conditioning contexts.