Harnessing Systemic Functional Linguistics (SFL) for AI-Generated Feedback on ESL Writing:

A Corpus-Based Study

Abstract

This study explores the alignment of AI-generated feedback with Systemic Functional
Linguistics (SFL) principles in English as a Second Language (ESL) writing. The research
investigates how AI tools, particularly those providing corrective feedback, adhere to SFL’s three
metafunctions, Ideational, Interpersonal, and Textual, when offering feedback to ESL learners.
The data for this study is sourced from the International Corpus Network of Asian Learners of
English (ICNALE), which provides a diverse collection of ESL student essays on various topics.
Employing a corpus-based research design, the study categorizes and analyzes AI feedback
based on the SFL framework. It examines how AI-generated feedback enhances content clarity
and precision (Ideational), improves tone and reader engagement (Interpersonal), and strengthens
the structure and coherence of ideas (Textual). The findings indicate that most AI-generated
feedback emphasizes grammatical accuracy, vocabulary, and content representation (60%),
followed by tone and engagement (25%) and structural organization (15%). The study concludes
that AI feedback, when aligned with SFL principles, significantly aids ESL learners in refining
their writing. However, the research also identifies key areas for improvement, particularly in
personalizing feedback and enhancing cohesion tools to ensure more contextually relevant
corrections. The implications of this study suggest that AI tools designed with SFL principles can
enhance ESL writing instruction by fostering more coherent, engaging, and accurate written
communication. The corpus-based approach utilizing ICNALE data provides a comprehensive
understanding of how AI feedback aligns with SFL, highlighting its effectiveness and limitations
in supporting ESL learners.