Exploring the Benefits of Strategic Hesitations in Language Learning Robots

Abstract

This study investigates the impact of introducing conversational pauses and self-adaptor gestures in robots to enhance human-likeness during English conversation practice. As globalization increases the demand for English proficiency, there is increasing interest in using conversational robots for language practice. However, conventional robots often lack natural disfluencies, making interactions feel overly artificial, especially for beginners. This study focuses on enhancing the robot's non-verbal behaviors by implementing conversational pauses and self-adaptor gestures such as fidgeting at predetermined intervals during scripted conversational scenarios. The experiment involved 22 men and women in their 20s. An experimental evaluation revealed that approximately 90% of participants perceived an increased sense of human-likeness when the robot exhibited these pausing and hesitation behaviors. These results suggest that strategically incorporating naturalistic disfluencies and human-like self-adaptor gestures into a conversational robot's non- verbal repertoire can significantly increase its perceived anthropomorphism and potentially improve user engagement in language learning contexts.