1 Oral Language Tests as Interactional Data Eric Hauser University of Electro-Communications and University of Hawaiʻi at Mānoa In this talk, I will be looking at oral language tests. However, as it says in the title, I will not be looking at these tests from the perspective of their status as tests. Rather, I will be looking at them from the perspective of their status as interaction between test-taker and tester, and occasionally as interaction between test-takers. Specifically, I will be looking at 1) how the test performance is begun and ended, that is, how it is framed; 2) how the test performance itself is interactionally organized; 3) how a certain kind of correction that sometimes occurs within test performances is organized; and 4) how test performances are sometimes scaffolded. (I am not the first person, of course, to look at oral language tests as interactional data. See, for example, the collection of papers in Young & He, 1998). The Data: Procedures for the Construction of a Geometric Figure The data consist of 30 oral test performances in English of students at a Japanese university with a focus on engineering and applied science. The students are first-year students and have two required English classes, one focusing on written academic English and the other on spoken academic English. The curriculum for all required classes is genre based and one of the genres taught in the first-year classes is the genre of procedure. I was the students’ teacher for their spoken class. In order to draw on the students’ strength in mathematics, I chose to use procedures for the construction of geometric figures, using a compass and a straightedge, to teach students how to organize a procedure, to teach them what sort of language to use, and to test their ability to produce a spoken procedure. In class, the procedure for one geometric figure was used to explicitly model and teach the organization and language of a procedure, after which the students and I worked together to jointly produce additional procedures. The students were then instructed to make notes for a spoken procedure of a geometrical figure of their choice, as long as it was not one of the simpler figures that were used in class. They then came to my office at their appointed time and used their notes to produce the procedure, which I attempted to follow using a compass and a straightedge. These test performances were video-recorded for assessment purposes. At the end of the semester, the students were asked for permission to use the recordings for research. As I talk about the data, I will refer to the tester as the teacher and the test-takers as students. In the transcripts, the teacher is labeled “T” and the students are labeled as “S” plus a number. Framing the Test Performance Beginnings. Looking across the test data, it becomes apparent that there is a canonical, at least for this particular test, way to begin. It is canonical not because it is necessarily the most common way to begin the test performance, but because when there are no oriented to problems in beginning the test, this is how it begins. Also, test beginnings which are different from the canonical beginning can be seen as deviations from the canonical beginning and generally involve repair work. Examples of the canonical beginning can be seen in extracts 1 and 2.