Testing the Bilingual–AI Pedagogical Framework (BAIP Model): A Structural Equation Modeling Approach in English for Educational Technology Programs

Zakila Mardatila Ersyad (Universitas Negeri Malang)
Luthviza Nabila Putri Ersyad (Politeknik Negeri Malang)
Like Raskova Octaberlina (Universitas Islam Negeri Maulana Malik Ibrahim Malang)
Andi Asrifan (Universitas Negeri Makassar)
K. J. Vargheese (Christ College (Autonomous) Irinjalakuda, Kerala)
Sajed S. Ingilan (University of Southeastern Philippines)

IJECA (International Journal of Education and Curriculum Application)
Vol 9, No 2 (2026): August

In multilingual higher education settings, English for Specific Purposes (ESP) instruction necessitates pedagogical frameworks that can tackle language obstacles and the rising demand for AI-enhanced learning; nonetheless, empirical data on the amalgamation of bilingual pedagogy and AI is still scarce, especially within English for Educational Technology programs. This study investigated the effectiveness of the Bilingual–AI Pedagogical Framework (BAIP Model) in English for Educational Technology programs, focusing on the impact of bilingual scaffolding and AI pedagogical support on student learning outcomes in multilingual higher education settings. The study investigated the effects of multilingual scaffolding and AI-assisted pedagogy on student engagement, academic self-efficacy, and learning performance, both directly and indirectly. A quantitative explanatory cross-sectional design was employed, encompassing 218 undergraduate students from both public and private universities in South Sulawesi, Indonesia. Data were collected using a validated 5-point Likert-scale questionnaire evaluating five latent domains and analysed by PLS-SEM in SmartPLS 4, with initial screening performed in SPSS 29. The results indicated that bilingual scaffolding was a significant predictor of student engagement (β = .42) and learning performance (β = .16), whereas AI pedagogical assistance had the most substantial impact on academic self-efficacy (β = .47) and learning performance (β = .28). Mediation analysis substantiated the notable indirect impacts of student involvement (β = .13) and academic self-efficacy (β = .17), with the model accounting for 61% of the variance in learning performance (R² = .61). This study not only validates these structural correlations but also presents the BAIP Model as a cohesive instructional framework that amalgamates bilingual pedagogy and AI-assisted learning into a singular explanatory model for ESP education. This study theoretically expands the previous literature by illustrating that bilingual scaffolding and AI pedagogical support serve as complimentary processes that collectively improve engagement, self-efficacy, and learning performance in multilingual higher education settings.