NCTE 2024

More than one billion students globally are learning English (Beare, 2016) with an underlying motive of being able to communicate with others. Thus, speaking in English is a primary goal for many students who learn English as a second and/or foreign (L2) language. However, providing learners with opportunities to practice speaking, and more specifically, interactional aspects of oral language, can be time- and resource intensive. Fully automated Spoken Dialogue Systems (SDSs) and more recently large language models (LLMs) such as ChatGPT, could provide a solution, allowing learners to practice their interactional skills in English anytime, anywhere, even when an instructor or English-speaking peer is not available (Bibauw et al., 2019; Litman et al., 2018; Timpe-Laughlin et al., 2022).
In recent years, research on the use of SDSs and AI-powered LLMs for the practice, learning, and assessment of L2 interactional skills has grown substantially. In this presentation, we will critically review the research landscape, focusing in particular on aspects such as (a) the process of designing and operationalizing tasks in SDS and ChatGPT, (b) how learners interact with the systems, (c) the performances of human interlocutors vis-à-vis SDSs vis-à-vis ChatGPT, and (d) what learners are thinking about using ChatGPT for practicing their English language skills. We will discuss the affordances and limitations that these new types of AI systems have for practicing oral English and for deployment in English language teaching and assessment.