Speakers
Description
This paper aims to investigate differences in syntactic complexity in the essays produced by ChatGPT and students in the Vietnamese EFL context. A quantitative corpus-based comparative design was adopted to address this aim. The data consist of 60 process essays, including 30 essays written by Vietnamese undergraduates and 30 essays produced by ChatGPT. Syntactic complexity was measured using the L2 Syntactic Complexity Analyzer (L2SCA). The study targets examining four indices, namely mean length of sentence (MLS), mean length of T-unit (MLT), mean length of clause (MLC), and clauses per sentence (C/S). The findings revealed significant differences in MLC and C/S. In contrast, there are no significant differences found in MLS and MLT. Specifically, ChatGPT-generated essays demonstrate a higher mean length of clause, which shows a greater amount of information within individual clauses. However, Vietnamese EFL students' essays exhibit higher clauses per sentence, which indicates that Vietnamese EFL students have a tendency to embed more clauses within sentences. The present study contributes to the growing body of research on the differences in syntactic complexity between ChatGPT and students’ writing in Vietnamese EFL context. In addition, the findings also provide linguistic insights that may help teachers and learners develop a more informed understanding of the linguistic characteristics of ChatGPT-generated writing and how it differs from authentic student writing, thereby supporting AI literacy in EFL education.
KEYWORDS: Syntactic complexity; L2 Syntactic Complexity Analyzer (L2SCA); Vietnamese EFL learners; ChatGPT-generated writing; corpus-based analysis
