Reducing Foreign Language Anxiety in Icebreaker Sessions Using LLM-Controlled Multilingual Virtual Agents in VR
Year: 2025
Type: conference-paper
Abstract: We explored how a Large Language model-powered multilingual virtual agent can reduce foreign language anxiety (FLA) in a VRbased icebreaker scenario.Our agent supports hybrid language interaction-accepting input in either English or the user's native language, while responding in English.Ten participants interacted with the agent using both language modes.Quantitative results showed reduced subjective FLA after the English session, compared to their real-world experience, and increased external focus when using their native language (vs English).Qualitative findings suggested that participants experienced greater anxiety in understanding compared to speech production, while language switching also imposed additional cognitive costs.
Authors: Zimu Cheng, Xueni Pan, Marco Gillies
Institution: Goldsmiths University of London
Language: English
Subfield: Language and Linguistics
Field: Arts and Humanities
Domain: Social Sciences
Open Access status: bronze
OpenAlex ID: W4414959476