ScholarWorks@숙명여자대학교

초록

This study explores the interplay between emotional dynamics, self-reflection, and thematic changes in psychotherapy through a multidimensional text-mining analysis of six session transcripts. Methods include sentiment analysis (valence-arousal), self-reflection analysis, LIWC and Empath category analysis, and topic modeling with LDA. Results indicate a dynamic emotional trajectory-initial positive arousal, mid-phase negative arousal, and final positive low arousal. A negative correlation (r = -0.55) was found between self-reflection and emotional expression; their co-occurrence corresponded with thematic transitions. Thematic analysis revealed consistent patterns across sessions-family relationships (sessions 1, 6), life experiences (session 3), and expression of emotions/needs (sessions 2, 4, and 5). These findings suggest psychotherapy progresses not through linear change but through the deepening of emotional processing, development of self-reflection, thematic shifts, and gradual integration. The study highlights the potential for text mining to provide objective measures of psychotherapy processes, informing customized treatment evaluation and planning.

키워드