유능감이 발화 특성, 동기, 사회적 상호의존성 태도에 미치는 영향: 멀티모달 데이터 분석

초록

This study examined how competence relates to differences in learners’ interactions and learning-related psychological characteristics by integrating behavior-based speech data with self-reported indicators (motivation and attitudes). A total of 130 university students were assigned to dyads, and competence was manipulated at high and low levels in a within-subject design with two discussion sessions. Thirteen speech features were extracted from the transcribed discussions, and surveys measured competence threat, interest, and social interdependence attitudes after each discussion. Among self-report variables, only competitive attitude significantly correlated with speech features. In the low-competence condition, higher competitive attitude was associated with increased morphemes per sentence and inter-sentence similarity, but decreased total sentence count. A repeated-measures MANOVA tested differences in speech characteristics, motivation, and social interdependence attitudes across competence conditions. Results showed that the high-competence condition produced greater speech quantity at the sentence level and more content words at the word level than the low-competence condition. Keyword analysis also revealed a higher frequency of commonly occurring nouns in the high-competence condition. Self-report results showed higher levels of interest and competitive attitudes under high competence. Overall, the findings suggest that competence may influence not only the quantity of learner interaction but also the quality of meaning construction, highlighting the value of multimodal data approaches in educational research.This study contributes both academically and practically by proposing instructional design principles and chatbot prompts that educators can readily apply in classroom practice.

키워드