Optimization of Non-Verbal Information for English Conversation Agents Using Interactive Evolutionary Computation
Yuma SHIMOSAKA, Emmanuel Ayedoun, Masataka Tokumaru · International Conference on Computers in Education · 2024
As English becomes increasingly important globally, many agent-based conversation practice environments struggle to maintain learner motivation due to a lack of personalized behavior. This study proposes optimizing a conversational agent's non-verbal cues—such as nodding and voice characteristics—through interactive evolutionary computation to enhance learners' motivation. Participants engaged in role- play scenarios across eight settings, providing feedback after each interaction. The agent's behavior was iteratively optimized, and approximately 90% of participants reported increased willingness to interact, suggesting that personalizing non-verbal behavior can significantly improve motivation in language learners.