Korpus tabanlı ile yapay zekâ destekli çok boyutlu geri bildirimin algılanan etkilerinin EAP mühendislik öğrencilerinin yazma performansı üzerine incelenmesi

This study investigates the perceived effects of corpus-based and AI-assisted multidimensional feedback on Engineering students' English for Academic Purposes (EAP) writing performance. Using a mixed-methods approach, the research compares the effectiveness of traditional corpus tools with AI-driven feedback provided by ChatGPT, focusing on five assessment criteria: criterion-based feedback, clarity of improvement instructions, accuracy, prioritization of key features, and supportive tone. Data were collected through surveys, interviews, and the analysis of academic papers from graduate students across various engineering disciplines. Quantitative results revealed that while ChatGPT was widely appreciated for its ease of use, its effectiveness in grammar and translation, corpus tools were valued for their accuracy in identifying domain-specific expressions and grammatical patterns. Qualitative findings highlighted ChatGPT's strength in delivering instant, personalized feedback, but also raised concerns about reliability and over-reliance. In contrast, although technically complex, corpus tools were praised for promoting independent learning and deeper language awareness. The study concludes that both feedback methods offer distinct advantages: ChatGPT excels in accessibility and real-time support, while corpus tools encourage critical engagement with language. For optimal EAP writing instruction, the integration of both approaches, supported by instructor guidance, is recommended. These findings contribute to ongoing discussions on technology-enhanced language learning and provide practical insights for educators aiming to balance innovation with pedagogical rigor.