SYNTACTIC UNIFORMITY IN ESSAYS REWRITTEN BY CHATGPT

Edward, Chijioke and Nwabueze, Ijeoma Nina (2025) SYNTACTIC UNIFORMITY IN ESSAYS REWRITTEN BY CHATGPT. Advance Journal of Arts, Humanities and Social Sciences, 8 (4). pp. 267-279. ISSN 4226-6348

[img] Text
AJAHSS_2508021024.pdf

Download (571kB)

Abstract

the rapid integration of generative AI tools, particularly ChatGPT, in academic writing has raised questions about their impact on syntactic development and stylistic creativity among university students. This study investigates the degree of syntactic uniformity in essays rewritten by ChatGPT compared to humanauthored compositions by first-year computer science students at Godfrey Okoye University, Nigeria. Anchored in Systemic Functional Linguistics (SFL), the research employs a corpus-based descriptive survey design to analyze sentence complexity, clause combination, and cohesion in 60 student essays and their AI-mediated rewrites. Findings reveal that ChatGPT-generated texts exhibit consistent syntactic patterns, higher use of subordinate clauses, and logical connectors, whereas humanauthored essays demonstrate greater variability, stylistic experimentation, and context-sensitive cohesion. While AI rewrites enhance grammatical accuracy and structural clarity, they may constrain stylistic diversity and individual writing voice. The study underscores the pedagogical implications of AI-assisted writing, advocating for its use as a supportive tool to reinforce syntactic competence without diminishing students’ creative expression. Recommendations include integrating AI in writing instruction while encouraging experimentation with complex sentence structures and rhetorical variation.

Item Type: Article
Subjects: L Education > L Education (General)
Divisions: Faculty of Law, Arts and Social Sciences > School of Education
Depositing User: Uchenna Eneogwe
Date Deposited: 25 Aug 2025 11:16
Last Modified: 29 Aug 2025 12:18
URI: http://eprints.gouni.edu.ng/id/eprint/5257

Actions (login required)

View Item View Item