Effects of word-frequency and social evaluation on phonetic accommodation

Abstract

Phonetic convergence between interlocutors is posited as a mechanism for long-term phonetic change (Pardo, 2006). This hypothesis requires that more frequent exposure to a phonetic variant increases the likelihood a speaker adopts said variant in their own speech. Certain instances of historical language change have, however, been theorized to represent divergence from, rather than convergence to, the speech patterns of an incoming group (Van Herk, 2008). We present the results of an artificial language learning experiment designed to investigate how social knowledge may mitigate frequency effects on phonetic convergence. Participants were exposed to nonce words produced by two speakers who differed systematically in their productions of front vowels. Participants in the experimental condition were informed that one speaker was a “native,” speaker of the language and an experienced teacher, while the other was a non-native speaker. The proportional frequency with which each speaker was heard varied across conditions. Subjects' own productions of the words were recorded after exposure. Results show an interaction between frequency and social bias. Test productions moved towards the socially upweighted, less frequent variant, but not to the same extent as one that was both upweighted and more frequent.