Article Per Year (5 Year)
p-Index From 2021 - 2026
0.23
P-Index
This Author published in this journals
All Journal Stilistika
I Made Agung Rai Antara
Department of Hotel Management, Triatma Mulya University
Arts Education Languange, Linguistic, Communication & Media

Published : 1 Documents Claim Missing Document
Claim Missing Document
Check
Articles

Found 1 Documents
Search

 1 
A COMPARATIVE EVALUATION OF AI-GENERATED FEEDBACK: GEMINI VS. CHATGPT IN ASSESSING EFL STUDENTS’ GRAMMATICAL RANGE AND ACCURACY I Made Agung Rai Antara; Ni Putu Yunik Anggreni
Stilistika : Jurnal Pendidikan Bahasa dan Seni Vol. 14 No. 2 (2026): Stilistika: Jurnal Pendidikan Bahasa dan Seni
Publisher : Program Studi Pendidikan Bahasa Indonesia dan Daerah Universitas PGRI Mahadewa Indonesia

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.59672/stilistika.v14i2.6280

Abstract

This study aims to compare the performance of ChatGPT 5.3 and Gemini Pro 3.1 in assessing Grammatical Range and Accuracy (GRA) in recount essays written by intermediate-level students at Triatma Mulya University. The study employed a comparative qualitative design using a content analysis approach to explore, evaluate, and compare the quality of grammatical feedback independently generated by both artificial intelligence systems. The data consisted of 15 recount essays written by intermediate-level students at Triatma Mulya University, which were analyzed based on error types, accuracy levels, and grammatical range. The findings revealed that both models demonstrated a high level of consistency in identifying major grammatical errors, particularly in tense usage, sentence structure, and capitalization. However, significant differences were found in the depth of analysis and sensitivity to minor errors. Gemini Pro 3.1 tended to provide more detailed and rule-based feedback, whereas ChatGPT 5.3 offered explanations that were simpler and easier for students to understand. Furthermore, Gemini exhibited a stricter evaluative tendency, while ChatGPT adopted a more moderate approach in classifying grammatical accuracy and range. These findings suggest that both systems possess strong potential for grammatical assessment, albeit with different orientations, making them complementary tools in English writing instruction.
Co-Authors Ni Putu Yunik Anggreni