AI meets 5E: innovating blended learning through constructivism and self-determination theory

Jingdan Liu, Yuanyuan Wu, Xujie Bao, Hazrul Abdul Hamid

In exam-oriented university English as a foreign language (EFL) settings, artificial intelligence (AI) is increasingly promoted as a means to enhance learning, yet its impact depends on how it is embedded within coherent pedagogical designs. This study proposed a 5E-sequenced AI-blended instructional model (5EAIBL), grounded in Constructivism and Self-Determination Theory (SDT), and compared it with a traditional blended learning model (BL) and a tool-driven AI-enhanced blended model (AIBL) in a Chinese tertiary EFL course. A total of 127 non-English-major sophomores from three intact classes received one of the three instructional models over a 16-week semester. English Achievement Tests were administered before and after the intervention and analyzed using one-way ANOVA, Games-Howell post hoc tests, and paired-sample t -tests. Students’ reflective journals collected from the 5EAIBL group were analyzed thematically to contextualize implementation. All groups improved significantly from pre- to post-test; however, the 5EAIBL group achieved significantly higher post-test scores than both comparison groups, whereas the AIBL group did not show a reliable advantage over traditional BL. Score distributions further indicated a raised performance floor in the 5EAIBL group. The journals further attributed the improved performance to phased preparation, guided practice, and active participation. These findings indicate that AI was more effective when orchestrated within a structured 5E sequence rather than added as a standalone tool for use, providing design-relevant evidence for AI-enhanced EFL instruction in exam-oriented higher education.