Abstract
LLM-based Voice Conversational Agents (VCAs) are emerging as social interaction partners, yet real-time immediacy can increase psychological burden for users with Communication Apprehension (CA). This study explores interaction design cues to alleviate anxiety for users with CA, grounded in their discomfort during LLM-based VCA interactions. We recruited 10 users with CA via survey screening and gathered everyday breakdown episodes through brief pre-interviews, translating them into support-card scaffolds. In a card-based co-design workshop, participants engaged in real-time voice conversations with an LLM-based VCA and reconstructed recorded moments using the cards to identify anxiety triggers and mitigation strategies. Findings show heightened burden and speech inhibition when speech continuity breaks or turn transitions become unpredictable. We propose design implications emphasizing contextual turn-taking, active listening cues, and explicit user control to support safe voice interactions for users with CA.
Author supplied keywords
Cite
CITATION STYLE
Kim, E., Ha, J., & Oh, C. (2026). Let Me Finish: Exploring Design Cues for Communication Apprehension in LLM-based Voice Conversational Agents. In Conference on Human Factors in Computing Systems - Proceedings . Association for Computing Machinery. https://doi.org/10.1145/3772363.3798435
Register to see more suggestions
Mendeley helps you to discover research relevant for your work.