Amanda Resti Maulidiya, - and Maman Abdurahman, - and Nalahuddin Saleh, - (2025) Exploring AI Capabilities in Arabic Grammar: Comparative Analysis of ChatGPT and Gemini. S1 thesis, Universitas Pendidikan Indonesia.
Abstract
This study aims to compare the performance of ChatGPT and Gemini in analyzing i'rāb Marfû’ât al-Asmâ’', a critical aspect of Arabic grammar for determining the grammatical roles of words in sentences. The analysis uses 11 examples from the book Mulakhash Qawâ'id al-Lugah al-'Arabiyyah by Fuad Ni'mah, focusing on identifying grammatical components such as mubtada', khabar, isim ashbah, isim kāda, khabar inna, fā'il, nā'ib al-fā'il, na'at, 'atf, tawkid, and badal. The study employs the Mann-Whitney test to assess statistical significance and the Cosine Similarity Index (CSI) to measure semantic similarity. Results show that ChatGPT outperforms Gemini with a significant value of 0.019, while the CSI score of 0.800 indicates high similarity between the models' outputs. ChatGPT excels in providing detailed and accurate analyses, while Gemini is more suitable for concise answers but may lack precision. These findings highlight the unique strengths of each model and underscore the necessity of manual correction to ensure the accuracy and relevance of results, particularly in technology-based Arabic grammar learning.
|
Text
TA_ART_S_ARB_2107498_ART - AMANDA RESTI.pdf Restricted to Staf Perpustakaan Download (580kB) |
|
|
Text
TA_ART_S_ARB_2107498_SK - AMANDA RESTI.pdf Download (739kB) |
| Item Type: | Thesis (S1) |
|---|---|
| Additional Information: | ID SINTA Dosen Pembimbing Maman Abdurahman: 257947 Nalahuddin Saleh: 6746413 |
| Uncontrolled Keywords: | ChatGPT, Gemini, Generative AI, I'râb analyzing |
| Subjects: | L Education > L Education (General) |
| Divisions: | Fakultas Pendidikan Bahasa dan Sastra > Jurusan Pendidikan Bahasa Arab |
| Depositing User: | Alsa Kurnia Muhammad Rizky |
| Date Deposited: | 09 Dec 2025 04:13 |
| Last Modified: | 09 Dec 2025 04:13 |
| URI: | http://repository.upi.edu/id/eprint/134451 |
Actions (login required)
![]() |
View Item |
