Computerized Diagnostic Language Assessment of Oral Requesting-in-Interaction: Proof-of-Concept

In this paper, we provide a proof-of-concept for a computerized diagnostic language assessment (C-DLA) of spoken requesting among Japanese learners of English at a Japanese computer science university. The program focuses on the pragmatics aspect of spoken communication, in which the language choices we make are affected by socio-contextual factors such as relative social status and familiarity with the interlocutor. In so doing, the C-DLA aims to address a number of challenges. Requesting is an important, but challenging skill for many learners, and yet is also undertaught in the language classroom, due to time and resource constraints. Further, assessments typically evaluate the learner holistically, providing an overall score, without providing insight into specific aspects of the learner's performance. A C-DLA addresses these challenges by employing a three-stage process: i) assessment administration; ii) provision of immediate individualized feedback to the learner that promotes learning, and iii) further instruction provision, based on the assessment results. Computerization of the DLA widens access, allowing large numbers of learners to engage with the program simultaneously. Here, we provide a rationale for the C-DLA and an outline of the key challenges - namely, speech recognition of L2 English, identification of pragmatic inappropriateness in a learner's interactive speaking performance and automated feedback delivery. We further provide a proof-of-concept for the C-DLA, in which the program administers a number of interactive spoken requesting tasks to the learner, acts as an automated spoken dialogue interlocutor, and provides immediate, automated pragmatics-focused feedback when necessary.