Phonetic adaptation in non-native spoken dialogue: Effects of priming and audience design

• Korean speakers matched labeled cards with 2 confederates in successive dialogs.
• They produced more L2-like sounds (missing in L1) to the English speaker than the Korean speaker.
• This phonetic adaptation occurred after immediate priming by the English speaker.
• They also produced more English-like speech when a referent needed disambiguation.
• Both priming and the need for pragmatic disambiguation shape speakers’ adaptations.

To be understood, non-native speakers must adapt their speech in order to produce contrasts in their second language (L2) that are not present in their first language (L1). Here we examine mechanisms hypothesized to facilitate such adaptation within spoken dialogue: priming, affiliation, and audience design. In two experiments, Korean non-native speakers of English interacted in a referential communication task with a Korean English-speaking confederate (Experiment 1) and a monolingual American English-speaking confederate (Experiments 1 and 2). The task required them to spontaneously produce labels containing segments from English that do not exist in Korean (/æ/ and coda /b/), which, when spoken with a Korean accent, can result in ambiguous homophones (e.g., pat pronounced like pet, or mob pronounced like mop). The Koreans produced more English-like phonetic segments not only immediately after hearing similar segments primed by the American partner, but also when the task required the partner to distinguish two potentially ambiguous items. The first time the Koreans referred to potentially ambiguous objects, utterances took longer to initiate; once they were aware of the potential for ambiguity, initiating contrasting labels took no more time than initiating labels primed by the partner. Findings suggest that priming effects in dialogue are not obligatory but may be motivated, and that phonetic adaptation is shaped by awareness of a partner’s pragmatic needs.