Article Preview
TopBackground
Early Speaking test theorem argued that direct tests are the better measures of speaking ability due to their close relationship between test context and real-life, and yet acknowledged that the language elicited is unrepresentative of real-life conversational discourse, because test takers are aware that they are talking to a language assessor (Clark, 1979). Hughes (2003) and Van Lier (1989) supported this view and challenged the asymmetry of the roles of test takers and interlocutors, and the relationship between them, throughout a direct test as “there are few situations in the world in which what the learner says has absolutely no effect on what he hears next” (Underhill, 1987, p. 35).
Clark’s (1979) early work acknowledged mode effect and argued that semi-direct tests involved more artificial language use and assessors missed visual aspects of test taker communication. Van Lier (1989) proposed that face-to-face talk should be regarded as the unmarked form of interaction, and communicating by telephone or speaking into a microphone as marked forms. While the suggestion that direct tests are preferable because of their perceived ability to approximate to real-life communication more closely than their semi-direct counterparts is perhaps reasonable, a reliance on face validity alone to ascertain delivery mode comparability excludes theoretical and empirical research insight.
Subsequent related research has attempted to distinguish this marked form of spoken output in semi-direct tests through numerous spoken output features with differing results found and conclusions made. This stems from both the multitude of approaches taken and the methods of analysis chosen. However, some conciliation between previous research results is to be found within the rhetorical functions and structure of test event discourse.