L2 learners' pragmatic output in a face-to-face vs. a computer-guided role-play task: Implications for TBLT

Accomplishing oral interactive workplace tasks requires various language abilities, including pragmatics. While technology-mediated tasks are thought to offer many possibilities for teaching and assessing second language (L2) pragmatics, their effectiveness – especially those facilitated by an AI agent (artificial intelligence agent) – remains to be explored. This study investigated how 47 tertiary-level learners of English as a second language (ESL) performed on an oral interactive task that required them to make requests to their boss in two distinct modalities. Each participant completed the same task with a fully automated AI agent and with a human interlocutor in a face-to-face format. Findings showed that both modalities elicited language use relevant to the pragmatics target. However, fully automated interactions were found to be more transactional, while face-to-face interactions were more functionally oriented (e.g. more frequent/varied supportive moves). Although fully automated interactive tasks may be useful for eliciting requests, replicating human-to-human interactions remains a challenge.