Iranian Journal of Educational Sociology 2645-3460 In Press 2026 09 01 Comparative Effects of Adaptive and Generative AI-Supported Mind Mapping on EFL Writing Cohesion, Fluency, and Anxiety Comparative Effects of Adaptive and Generative AI-Supported Mind Mapping on EFL Writing Cohesion, Fluency, and Anxiety 1 15 10.61838/jrfb8h20 EN Journal Article 2026 03 11 <p><strong>Purpose:</strong> This study aimed to compare the effects of traditional instruction, conventional mind mapping, generative AI-supported mind mapping (GenAI), and adaptive AI-supported mind mapping (AdaptAI) on EFL learners’ writing cohesion, writing fluency, and writing anxiety.</p> <p><strong>Methods and Materials:</strong> This quasi-experimental study employed a pretest–posttest design with 140 intermediate adult EFL learners enrolled in writing courses at a private language institute in Kerman, Iran, during spring 2025. Participants were selected through purposive sampling and assigned by intact classes to four instructional conditions: traditional instruction, conventional mind mapping, GenAI-supported mind mapping, and AdaptAI-supported mind mapping, with 35 learners in each group. The intervention lasted eight weeks and included 16 ninety-minute sessions. Writing cohesion was assessed using an analytical rubric adapted from McNamara et al. (2010), writing fluency was measured through scaled word-count scores, and writing anxiety was assessed using Cheng’s Second Language Writing Anxiety Inventory. Gain scores were analyzed using one-way analysis of variance and Tukey HSD post hoc comparisons.</p> <p><strong>Findings:</strong> One-way ANOVA showed significant between-group differences in writing cohesion, F(3, 136) = 53.11, p &lt; .001; writing fluency, F(3, 136) = 48.02, p &lt; .001; and anxiety reduction, F(3, 136) = 30.38, p &lt; .001. Tukey tests indicated that AdaptAI significantly outperformed GenAI in cohesion (p = .048) and anxiety reduction (p = .025), whereas their difference in fluency was nonsignificant (p = .164). Both AI-supported groups significantly outperformed the conventional mind-mapping and traditional groups across all outcomes. Conventional mind mapping also produced significantly greater gains than traditional instruction in cohesion and fluency (both p &lt; .001), but not in anxiety reduction (p = .160).</p> <p><strong>Conclusion:</strong> AI-supported mind mapping, particularly adaptive AI-supported mind mapping, appears to provide an effective instructional scaffold for improving EFL learners’ writing cohesion and fluency while reducing writing anxiety. Adaptive, personalized feedback may offer additional advantages over generative AI for higher-order textual organization and affective outcomes, although both AI approaches appear similarly effective for improving fluency.</p> https://qijes.com/index.php/ijes/article/download/1552/1530