A conversation analysis of structure and interaction in the language interview

In the push towards defining communicative competence, applied linguistics research has focused on describing and analyzing nonnative language use in communicative contexts and on issues surrounding the measurement of oral proficiency. While there is widespread agreement that the interview is the most appropriate method for measuring oral proficiency, until recently little effort has been put into combining language analysis techniques with language testing methodology, an endeavor which might help us to evaluate the widely-held assumption that a language assessment interview is a structured conversational exchange (in which oral proficiency is measured.) This study addresses the question of how the structure of and interaction in the interview bear on these oral assessments by employing conversation analysis techniques, in which the analyst looks at the talk on a turn-by-turn, sequential basis from the perspective of the participants, to describe a corpus of language interview data.; Twenty oral interviews conducted to place students in communication skills courses were audio- and videotaped on four occasions at UCLA. The tapes were transcribed using conversation analysis conventions and were examined microanalytically for several structural and interactional features. The analyses indicate that the interviews are composed of identifiable phases that are suggested by the interview agenda and oriented to by the participants. In addition, language ability assessments are a recurrent practice in these interviews; the fact that their substance is directly related to placement decisions accounts for the modifications in the preference structure of assessments in these data. Finally, interviewers routinely modify their questions in response to some perceived trouble by recompleting question turns, by suggesting alternatives to choices presented, and by reformulating the turns altogether.; These results suggest that the interviews import their fundamental structural and interactional features from conversation, but are characteristically and identifiably instances of "interviews" for the participants. Several ways in which test developers may wish to modify their instruments to engender a more conversation-like interaction are suggested; some important considerations for the analysis of native-nonnative interaction are raised as well. This study, therefore, illustrates a promising approach to the analysis of oral interaction in both testing and non-testing contexts.