Quantitative research on post-conflict societies has documented associations between exposure to violence and shifts in gendered social roles, labor practices, and political engagement. Yet these studies rarely show how individuals and communities verbally construct these interpretations or how meaning is negotiated through interaction. Focus groups provide access to these processes, but conventional text-as-data approaches treat transcripts as monologic and overlook the interactional structure that shapes collective sense-making. This paper introduces a computational framework for analyzing focus groups as dialogic events and applies it to 108 focus group discussions conducted across 36 communities in conflict-affected and non-affected regions of Colombia and Sri Lanka. The study treats focus groups as complex data sources by integrating turn-level text representations with contextual information and interactional patterns. Substantive themes related to gendered social roles are identified using embedding-based similarity measures that track how ideas emerge and evolve within conversations. Community conflict exposure situates each group within its broader social environment. Measures of conversational power and connectedness capture how speakers introduce, elaborate, contest, or constrain interpretations. By combining these elements, the analysis uncovers relationships between what communities discuss, how discussions unfold, and the contexts in which meaning is produced. The findings demonstrate systematic differences in meaning-making across community contexts and show how variation in interactional structure influences the articulation of post-conflict experiences. The approach highlights the value of computational text analysis for capturing the social production of meaning and provides a scalable framework for the quantitative study of qualitative data in political science.