This study employs Conversation Analysis (CA) as the theoretical framework and adopts a quasi- experimental design to explore the effects of CA-based instruction on enhancing classroom interactional competence among college English learners. The experimental group (n = 45) and the control group (n = 41) received systematic CA training and conventional instruction, respectively. After a 16-week teaching experiment, the experimental group demonstrated significant improvements in the contribution of short response tokens and adjacency pairs, along with a reduction in average turn length, indicating a preference for economical interaction strategies. Although the increase in the contribution of assessment was not statistically significant, its upward trend suggested enhanced complexity in evaluative expressions. The findings confirm that CA instruction optimizes interaction quality. Future research should integrate multimodal technologies to further investigate the dynamic mechanisms of interactional competence development.