What scores from monologic speaking tests can(not) tell us about interactional competence. (January 2022)

• Record Type: Journal Article
• Authors: Roever, Carsten; Ikeda, Naoki
• Is Part Of: Language testing. Volume 39:Number 1(2022)
• Page Start: 7
• Page End: 29
• Publication Date: 2022-01
• Subjects: Conversation analysis -- English as a second language -- interactional competence -- speaking test -- TOEFL; Language and languages -- Ability testing -- Periodicals; Language and languages -- Examinations -- Periodicals; 407.6
• DOI: 10.1177/02655322211003332
• Languages: English
• ISSNs: 0265-5322

Abstract

The overarching aim of the study is to explore the extent to which test takers' performances on monologic speaking tasks provide information about their interactional competence. This is an important concern from a test use perspective, as stakeholders tend to consider test scores as providing comprehensive information about all aspects of L2 competence. One hundred and fifty test takers completed a TOEFL iBT speaking section consisting of six monologic tasks, measuring speaking proficiency, followed by a test of interactional competence with three monologues and three dialogues, measuring pragmalinguistic skills, the ability to recipient design extended discourse, and interactional management skills. Quantitative analyses showed a medium to high correlation between TOEFL iBT speaking scores and interactional scores of r = .76, though with a much lower correlation of r = .57 for the subsample most similar to a typical TOEFL population. There was a large amount of variation in interactional scores for test takers at the same TOEFL iBT speaking score level, and qualitative analyses demonstrated that test takers' ability to recipient design their talk and format social actions appropriate to social roles and relationships was not well captured by speaking scores. We suggest potential improvements.