A Microanalytic Perspective on Discourse, Proficiency, and Identity in Paired Oral Assessment

The increasing popularity of paired format in oral testing has engendered legitimate scrutiny of its reliability and validity as compared with the more traditional interviewer–interviewee arrangement. Although characteristics such as the gender, cultural/L1 background, and language proficiency of one's interlocutor likely affects the discourse produced with a partner, the question remains whether this interlocutor effect influences scores on the test. In this article, the construct interlocutor effect is further examined. Initially, transcript data from two large-scale speaking tests were analyzed by working backward from testtaker scores to locate discourse features to support those scores; identity formulations, such as “proficient” and “competent,” as constructed in and through the discourse test takers produce, emerged as a salient feature of the talk. Specifically, we posit that a test taker brings a language proficiency identity to a test task, and this identity is constructed, mediated, and displayed in the talk. We argue that “proficiency” is fluid, in that it will shift—on a turn-by-turn basis—based on who we are talking to in a second language and what sort of identity(ies) we bring to and are mediated in that interaction. It is this aspect of the interlocutor effect that merits more attention.