SEMANTIC ALIGNMENT AND MISALIGNMENT IN AUTOMATED WRITING FEEDBACK: AN NLP-BASED STUDY OF MEANING CONSTRUCTION IN EFL WRITING

Abstract

Recent advances in Natural Language Processing (NLP) have led to the widespread use of automated writing feedback tools in English as a Foreign Language (EFL) instruction. While existing research in applied linguistics has primarily evaluated the accuracy and usability of such tools, comparatively little attention has been given to the semantic validity of automated feedback, particularly in relation to teachers’ interpretations of meaning in student writing. Addressing this gap, the present study investigates the extent to which NLP-generated feedback aligns with or diverges from EFL teachers’ semantic evaluations of student texts. Adopting a mixed-methods design, the study analyzes a corpus of undergraduate EFL student essays using NLP-based tools capable of generating meaning-related feedback on coherence, clarity, and semantic relevance. These computational outputs are systematically compared with evaluations provided by experienced EFL teachers using a semantic assessment rubric. Semi-structured interviews further explore teachers’ perceptions of semantic mismatches between automated feedback and pedagogical judgment. The findings are expected to reveal recurring patterns of semantic misalignment, particularly at the discourse and pragmatic levels, where contextual and rhetorical meaning plays a crucial role. This study contributes to applied linguistics by foregrounding the semantic limitations of current NLP-based feedback systems and by offering insights into how automated tools can be better aligned with pedagogical understandings of meaning in EFL writing.