| The present research investigates the use of different question prompts and the discourse they generate in the SLE:OI, an ACTFL-variant second language oral proficiency interview test. One hundred and fifty-two question prompts used to elicit the test task of 'supporting an opinion,' were transcribed from 27 SLE:OI tests administered between July and November, 2000. From this, 30 categories of question prompts were identified by 6 SLE:OI raters acting as judges. Independently, the researcher and the judges determined task difficulty/complexity to be the predominant feature differentiating the categories. Using the 30 categories as a basis, the Question Prompt Complexity Questionnaire was produced and administered to the 6 judges. Analysis of the questionnaire data indicated a clear consensus for 3 categories into 'easy' and 'difficult' groups. Subsequently, candidate responses to 11 question prompts from the easy group, and 10 from the difficult group were transcribed, and discourse analyses were carried out to ascertain response levels of L2 fluency (by type-token ratio; frequency of silent and filled pauses, repetitions, and self-repairs), accuracy (by verb morphology and lexical use), and complexity (by clause subordination). The results demonstrated that those candidates tested with 'easy' and 'difficult' question prompts showed strong, significant differences in two aspects of their response fluency, but no significant differences in the accuracy or complexity of their responses. Based on these findings, several recommendations and implications for rater training were cited. |
