Generative AI Pragmatic Scaffolding and Interaction Effects on Oral Fluency in Saudi Arabia: Self-Regulation and Proficiency Roles

Abstract

Oral fluency becomes a problem for Saudi EFL students with high foreign language speaking anxiety. Nevertheless, the mechanisms by which fluency and the quality of interactions can be influenced by a generative AI scaffold remain unclear. This study examined generative AI pragmatic scaffolding (GAPSC) and AI Interaction Effects (INTEFF) through the lens of their effects on oral fluency, with self-regulation as the mediating factor and language proficiency as the moderator. The study used a quantitative cross-sectional survey of 345 Saudi university students majoring in English. The data were analyzed using PLS-SEM. The results indicate that GAPSC and INTEFF can be used to predict oral fluency. Similarly, self-regulation played a mediating role in both relationships. Language proficiency negatively moderated the GAPSC fluency relationship and positively moderated the INTEFF fluency relationship. Theoretically, this study contributes to Sociocultural Theory (SCT) and Self-Regulated Learning (SRL) by synthesizing AI scaffolding, interaction effects, and self-regulation into a single fluency model. In practice, this study provides evidence-based recommendations from Saudi educators for creating proficiency-sensitizing AI-mediated support interventions that align with Vision 2030.