Assessing L2 interactive speaking ability has become a key pursuit for assessment developers, as it taps into the pragmatic dimensions of speech beyond traditional psycholinguistic properties (Roever, 2024). However, operationalizing such assessments faces challenges ranging from the inherent complexity of measuring communicative competence to the labor-intensive and inconsistent nature of human-delivered tests (Galaczi & Taylor, 2018). Recent studies have explored technologies such as spoken dialogue systems (SDS) to automate interactive speaking assessments (Ockey et al., 2023). Although these systems can elicit features comparable to those from human interlocutors, they still face major challenges such as reliance on scripted outputs and lack of authenticity compared to face-to-face conditions. Emerging technologies such as Extended Reality (XR) and Artificial Intelligence (AI) hold the potential to deliver authentic, immersive, and automated speaking assessments (Voss, 2024). Employing AI-powered humanoid avatars simulating human interlocutors, we developed CAIVR (Conversational AI in Virtual Reality), an interactive speaking assessment platform delivered via a high-immersion VR HMD (Head Mounted Display), aiming to assess L2 interactional competence that maximally mirrors traditional in-person oral assessments. We hope this study lays the groundwork for future innovative speaking assessments that integrate XR and AI.

Reference:
Galaczi, E., & Taylor, L. (2018). Interactional competence: Conceptualisations, operationalisations, and outstanding questions. Language Assessment Quarterly, 15(3), 219-236.
Ockey, G. J., Chukharev-Hudilainen, E., & Hirch, R. R. (2023). Assessing interactional competence: ICE versus a human partner. Language Assessment Quarterly, 20(4-5), 377-398.
Roever, C. (2024). Assessing Interactional Competence and. The Concise Companion to Language Assessment, 230.
Voss, E. (2024). Language assessment and artificial intelligence. The Concise Companion to Language Assessment, 112-125.