<![CDATA[The increasing use of generative artificial intelligence in language education has transformed written feedback practices, yet limited evidence exists regarding how teacher and AI-generated feedback differ in supporting distinct dimensions of second language writing. Most previous studies have emphasized feedback effectiveness rather than examining the complementary functions of human and AI feedback across higher-order and lower-order writing concerns. This study investigated the characteristics and functions of teacher and ChatGPT-3.5 feedback on paragraph writing produced by first-year English as a Foreign Language (EFL) students in Vietnam. Guided by Sociocultural Theory, feedback literacy, and the distinction between higher-order and lower-order concerns, the study employed an exploratory mixed-methods design. A total of 178 feedback instances generated from ten student paragraphs were analyzed using quantitative frequency analysis and qualitative thematic coding. The findings revealed that teacher feedback primarily addressed higher-order concerns, including idea development, organization, coherence, and meaning construction, whereas ChatGPT-3.5 predominantly focused on lower-order concerns such as grammar, vocabulary, and sentence structure. Students also perceived teacher feedback as more supportive for conceptual development, while valuing AI-generated feedback for its immediacy and linguistic accuracy. This study extends current understanding of AI-assisted writing by demonstrating that teacher and AI feedback perform complementary rather than competing instructional roles. The findings provide practical implications for integrating human expertise and generative AI to create balanced feedback practices that support both rhetorical development and linguistic accuracy in EFL writing instruction.]]>
Copyrights © 2026
