MEASURING WHAT MATTERS: A NOVEL METRIC-BASED EVALUATION OF CHATGPT’S EFFECTIVENESS IN EFL WRITING
Journal of Social Sciences and Humanities · 31.12.2025 · 64 (2) · 17-35
DOI: 10.46568/jssh.v64i2.811 ↗Kurz gefasst
Fachartikel in der Zeitschrift Journal of Social Sciences and Humanities, online erschienen am 31.12.2025. Verfasst von Xingxing Jia. Der Volltext ist frei zugänglich (Open Access, CC BY-NC 4.0).
Abstract (Originalsprache)
Since the introduction of ChatGPT in EFL writing, most studies have evidenced its role primarily through qualitative lens, leaving its instructional effectiveness insufficiently verified in a rigorous, measurable way. Addressing this, the study uses a metric-driven framework known as CAOS—Content, Accuracy, Complexity, and Style. It aims to quantify both the benefits and potential trade-offs to guide balanced classroom implementation. Forty essays were sampled and revised with ChatGPT. Analysis revealed significant improvements in lexical sophistication, syntactic complexity, and grammatical accuracy, as well as enhanced academic tone in the revised texts. However, some content measures showed slight declines in information density, highlighting the need for balanced use of AI tools. These findings provide empirical evidence for the pedagogical value of ChatGPT in EFL writing, while underscoring the importance of strategic, informed implementation.
Zitieren
APA
Jia, X. (2025). MEASURING WHAT MATTERS: A NOVEL METRIC-BASED EVALUATION OF CHATGPT’S EFFECTIVENESS IN EFL WRITING. Journal of Social Sciences and Humanities, 64(2), 17-35. https://doi.org/10.46568/jssh.v64i2.811
Vancouver
Jia X. MEASURING WHAT MATTERS: A NOVEL METRIC-BASED EVALUATION OF CHATGPT’S EFFECTIVENESS IN EFL WRITING. Journal of Social Sciences and Humanities. 2025;64(2):17-35. doi:10.46568/jssh.v64i2.811
BibTeX
@article{Jia2025,
author = {Jia, Xingxing},
title = {MEASURING WHAT MATTERS: A NOVEL METRIC-BASED EVALUATION OF CHATGPT’S EFFECTIVENESS IN EFL WRITING},
journal = {Journal of Social Sciences and Humanities},
year = {2025},
volume = {64},
number = {2},
pages = {17--35},
doi = {10.46568/jssh.v64i2.811}
}
EUVIMED