Assessing the Quality of ChatGPT-Generating Arabic Subtitles: A Functional and Linguistic Analysis
Keywords:
generative artificial intelligence, auto-generated subtitling, ChatGPT 3.5, TED talk, English-to-Arabic translationAbstract
The emergence of generative artificial intelligence has reshaped the translation industry. Large language models can do more than just translate; generate, edit, and review texts. However, little is known about auto-generated subtitling. Effective subtitling requires deep technical, linguistic and cultural knowledge. This study seeks to evaluate automatic subtitling produced by the OpenAI’s ChatGPT-3.5, focusing on linguistics and functional equivalence. Using the content of TED talks: the original transcript as the source material, Arabic subtitles as reference translation, and the platform subtitle guidelines as prompts, the experiment was conducted. The automatically generated subtitles were analysed by human translators according to the parameters of FAR: functional equivalence, readability and acceptability. The findings show that omission is the most frequent error category in the auto-generated subtitling, affecting the intended audio message and informational details. Less frequent issues were found such as semantic, and stylistic errors, compressed sentences, and mistranslation. This suggests that auto-generated subtitling is generally understandable but still lacks accuracy. Further research is needed to investigate the advanced ChatGPT’s auto-generated subtitling.
References
Abdelaal, N., & Al Sawi, I. (2025). A comparative evaluation of ma-chine translation vs. human translation for legal texts: a case study of translations between English to Arabic. In Comparative Legilinguis-tics, 63, 186-223.
Al Sammarraie, H. M. (2025). Challenges and strategies in post-editing English into Arabic Neural Machine Translations of movie subtitles. In Journal of Modern Languages, 35(1), 139-164.
Al Sawi, I., & Allam, R. (2024). Exploring challenges in audiovisual translation: A comparative analysis of human-and AI-generated Arabic subtitles in Birdman. In Plos one, 19(10), e0311020.
Al-Darabee, M., Farghal, M., & Haider, A. S. (2025). Netflix versus Google Translate: A case study of the English-Arabic translation of scato-logical terms. In Studies in Linguistics, Culture & FLT, 13(2), 77-95.
Al-Harahsheh, A. M., & Rababah, R. A. H. (2025). YouTube’s Auto-mated Subtitling from English into Arabic: A Case Study of Harry Potter and the Prisoner of Azkaban. In FORUM FOR LINGUISTIC STUDIES, 7 (2).
Al-Momani, A., Haider, A. S., Dagamseh, M., & Akasheh, W. M. (2025). Audience responses to cultural and linguistic gaps in English–Arabic auto-subtitles on YouTube. In Research Journal in Advanced Humanities, 6(3).
Bahdanau, D., Cho, K., & Bengio, Y. (2015). Neural machine transla-tion by jointly learning to align and translate. In International Conference on Learning Representations, (ICLR) 2015, San Diego, California.
Bellés-Calvera, L., & Caro Quintana, R. (2022). Is academic discourse accurate when supported by machine translation. In Quaderns de Filolo-gia, (171-201).
Bojar, O., Chatterjee, R., Federmann, C., Graham, Y., Haddow, B., Huck, M., ... & Zampieri, M. (2016, August). Findings of the 2016 con-ference on machine translation. In Proceedings of the First Conference on Machine Translation: Volume 2, Shared Task Papers (pp. 131-198) Ber-lin, Germany.
Downloads
Published
How to Cite
Issue
Section
License
Copyright (c) 2026 Faculty of Languages Journal-Tripoli-Libya
This work is licensed under a Creative Commons Attribution 4.0 International License.
