Affective Depolarization and Democratic Resilience in Anonymous Cross-Partisan Conversations

Affective polarization is commonly examined through individual attitudes, leaving depolarizing dynamics within cross-partisan conversations less understood. This study examines affective depolarization as a relational and temporal phenomenon using 319 anonymous, unfacilitated conversations involving 794 participants and 14,156 conversational turns. Human-validated large language model-assisted coding, quantitative ethnography, and temporal analyses examined discourse structure and sequencing. Affiliative dynamics were significantly more prevalent than antagonistic ones. Curiosity appeared relatively early and was prospectively associated with vulnerability and political identity disclosure, which were associated with later relational processes. A relational-middle composite predicted subsequent social distance reduction, with difference normalization and surprise at similarity showing particularly strong prospective associations. Findings suggest that affective depolarization is not only an attitudinal outcome but also an interactional process in which political difference remains visible while relational engagement develops. The study advances an interactional account of depolarization and identifies testable relational processes.