<![CDATA[Developing authentic English-speaking skills, defined here as the ability to communicate spontaneously and appropriately in real-world contexts, and reducing dependency on the first language (L1) are significant challenges for Indonesian EFL learners. This study investigates university students' perceptions of using AI tools like ChatGPT and Google Gemini as scaffolding to mitigate L1 reliance and foster such authentic speaking practice. A mixed-methods explanatory sequential design was employed, beginning with a descriptive survey of 50 Indonesian EFL students, followed by in-depth interviews with 6 participants. Quantitative findings revealed that students positively perceived AI-driven scaffolding, its role in reducing L1 dependency, and its contribution to enhancing authentic communicative competence. Qualitative analysis, framed by Sociocultural Theory, the Technology Acceptance Model (TAM), and Skill Acquisition Theory, uncovered three central themes: AI as a dynamic scaffold within the Zone of Proximal Development (ZPD), its effectiveness in driving learner autonomy, and its role in fostering communicative autonomy. The study concludes that AI-assisted scaffolding is perceived as an integrated technological support system that effectively addresses the cognitive, affective, and practical dimensions of L1 dependency, thereby promoting authentic English-speaking proficiency. These findings suggest practical implications for EFL pedagogy, advocating for the strategic integration of AI tools as complementary scaffolds in resource-constrained educational contexts to bridge the gap towards independent and authentic communication.]]>
Copyrights © 2025
