Investigating self-efficacy as a mediating mechanism between AI-dialogic scaffolding and Saudi learners’ speaking anxiety and confidence

Foreign language speaking anxiety (FLSA) remains a widespread problem in the spoken development of EFL learners. This cross-sectional study examined the associations between three AI-dialogic scaffolding interventions (GenAI Chatbots, digital storytelling, and collaborative reasoning), the speaking anxiety and confidence of Saudi EFL learners, with self-efficacy as a mediating variable. A sample of 287 valid responses (68.3% response rate) from Saudi undergraduate EFL students was collected through a cross-sectional survey and analysed using partial least squares structural equation modelling (PLS-SEM). The findings confirmed significant associations between the three scaffolds and reduced speaking anxiety and increased speaking confidence, both directly and indirectly, through the mediation of enhanced self-efficacy. Digital Storytelling showed the strongest direct association with anxiety reduction, while GenAI Chatbots showed the strongest association with self-efficacy. Due to the cross-sectional design, causal claims cannot be inferred from these associations; the results reflect covariance among perceptions at a single time point. The findings provide a validated structural model indicating that self-efficacy is a significant psychological process. The study offers implications for developing affective-oriented, technology-based EFL programmes, though longitudinal research is needed to establish causal direction.