earticle

영어

This paper uniquely bridges the technical and socio-cultural dimensions of ChatGPT integration in university-level English language teaching (ELT). Unlike prior research that often isolates technical efficacy (e.g., error correction) or ethical concerns, we innovatively combine quantitative outcomes with qualitative insights into equity, ethics, and human-AI dynamics. Employing a quasi-experimental design (n=50 undergraduates) and in-depth interviews (4 instructors, 15 students), we evaluated ChatGPT’s impact on writing proficiency, engagement, and critical thinking. Our quantitative results demonstrate significant improvements for ChatGPT users in grammatical accuracy (34%) and task achievement (29%), alongside a notable 42% increase in writing time—a metric seldom explored previously. However, our qualitative findings reveal critical, underexplored limitations: risks to learner autonomy, erosion of teacher-student rapport, and exacerbated equity gaps. Distinct from studies narrowly focused on AI's technical or ethical facets, we frame these issues within sociocultural theory and blended learning. We identified cultural biases in AI feedback and new academic integrity challenges, issues significantly underexplored, highlighting the necessity and originality of our study. Based on these findings, we propose a balanced pedagogical model integrating AI with structured human mentorship—a novel contribution to addressing current AI-ELT paradigm gaps. This model enhances critical thinking and inclusivity, offering actionable institutional strategies for ethical technology integration in ELT, prioritizing the humanistic core of language education and advancing the discourse beyond earlier transactional approaches.