JORDANIAN ARABIC TO MODERN STANDARD ARABIC TRANSLATION USING A LARGE MODEL TUNED ON A PURPOSE-BUILT DATASET AND SYNTHETIC ERROR INJECTION

(Received: 2-Mar.-2025, Revised: 7-Jun.-2025 , Accepted: 8-Jun.-2025)

Authors Gheith A. Abandah, Moath R. Khaleel, Iyad F. Jafar, Mohammad R. Abdel- Majeed, Yousef H. Hamdan, Ashraf E. Suyyagh, Asma A. Abdel-Karim, Shorouq M. AlAwawdeh,

Keywords #Jordanian Arabic #Modern Standard Arabic #Dialectal translation #Large language models #Synthetic error injection #Natural-language processing #ByT5

Abstract This paper addresses the challenge of accurately translating Jordanian Arabic into Modern Standard Arabic (MSA) and correcting common linguistic errors. Although MSA is the formal standard for Arabic communication, the widespread use of local dialects in social media and everyday interactions often results in texts laden with spelling and grammatical issues. To overcome these challenges, we present an end-to-end system based on a newly constructed Jordanian Arabic dataset (JODA) comprising 59,135 sentences, as well as the Tashkeela dataset perturbed through synthetic error injection. We employ ByT5, a large pre-trained language model that processes text at the byte level, making it resilient to spelling variations and morphological complexities common in Arabic dialects. Our experimental results show that fine-tuning ByT5 on JODA and a 10% error-injected Tashkeela subset notably improves both BLEU score and character error rate (CER). Combining JODA with the synthetically modified Tashkeela data reduces the CER to 4.64% on the Test-200 test set and 1.65% on the TSMTS test set. Moreover, manual inspections reveal that the model produces correct or near-correct translations in most cases. Finally, we developed a custom smartphone keyboard and a web portal to demonstrate how the system can be made easily accessible to interested users, offering a practical solution for millions of Arabic speakers seeking to produce accurate, diacritized MSA text. This solution is currently limited to the Jordanian dialect; future work will focus on developing similar datasets and solutions for other Arabic dialects.

References

[1] G. Khan, M. P. Streck and J. C. Watson, The Semitic Languages: An International Handbook, vol. 36,Walter de Gruyter, 2011.

[2] G. Abandah, M. Khedher, W. Anati, A. Zghoul, S. Ababneh and M. S. Hattab, "The Arabic LanguageStatus in the Jordanian Social Networking and Mobile Phone Communications," Proc. of the 7th Int’l Conf. on Information Technology (ICIT 2015), pp. 449–456, 2015.

[3] B. Min et al., "Recent Advances in Natural Language Processing via Large Pre-trained Language Models:A Survey," ACM Computing Surveys, vol. 56, no., pp. 1–40, 2023.

[4] S. Castilho et al., "Is Neural Machine Translation the New State of the Art?" The Prague Bulletin ofMathematical Linguistics, vol.108, no. 1, pp. 109–120, 2017.

[5] M. S. H. Ameur, F. Meziane and A. Guessoum, "Arabic Machine Translation: A Survey of the LatestTrends and Challenges," Computer Science Review, vol. 38, DOI: 10.1016/j.cosrev.2020.100305, 2020.

[6] H. Wang et al., "Progress in Machine Translation," Engineering, vol. 18, pp. 143–153, 2021.

[7] I. Rivera-Trigueros, "Machine Translation Systems and Quality Assessment: A Systematic Review,"Language Resources and Evaluation, vol. 56, no. 2, pp. 593–619, 2022.

[8] G. Doğru, "Statistical Machine Translation Customization between Turkish and 11 Languages,"TransLogos Translation Studies J., vol. 3, no. 1, pp. 98–121, 2020.

[9] I. T. Khemakhem, S. Jamoussi and A. B. Hamadou, "Improving English-Arabic Statistical MachineTranslation with Morpho-syntactic and Semantic Word Class," Int’l J. of Intelligent Systems Technologies and Applications, vol. 19, no. 2, 172, 2020.

[10] C. España-Bonet and M. R. Costa-Jussà, "Hybrid Machine Translation Overview," Part of the BookSeries: Theory and Applications of Natural Language Processing (NLP), pp. 1–24, Springer, 2016.

[11] J. Zakraoui, M. Saleh, S. Al-Maadeed and J. M. AlJa’am, "Evaluation of Arabic to English MachineTranslation Systems," Proc. of the 2020 11th IEEE Int’l Conf. on Information and Communication Systems (ICICS), pp. 185–190, Irbid, Jordan, 2020.

[12] R. Dabre, C. Chu and A. Kunchukuttan, "A Survey of Multilingual Neural Machine Translation," arXiv,[Online], Available: https://arxiv.org/abs/1905.05395v1, 2019.

[13] S. C. Siu, "Revolutionizing Translation with AI: Unravelling Neural Machine Translation and GenerativePre-trained Large Language Models," SSRN Electr. J., DOI: 10.2139/ssrn.4499768, 2023.

[14] J. Guo, Z. Zhang, L. Xu, B. Chen and E. Chen, "Adaptive Adapters: An Efficient Way to IncorporateBERT into Neural Machine Translation," IEEE/ACM Trans. on Audio Speech and Language Processing, vol. 29, pp. 1740–1751, 2021.

[15] X. Wu, Y. Xia, J. Zhu, L. Wu, S. Xie and T. Qin, "A Study of BERT for Context-aware Neural MachineTranslation," Machine Learning, vol. 111, no. 3, pp. 917–935, 2022.

[16] C. Raffel et al., "Exploring the Limits of Transfer Learning with a Unified Text-to-Text Transformer," J.of Machine Learning Research, vol. 21, no. 140, pp. 1–67, 2020.
[17] M. R. Costa-Jussà et al., "Scaling Neural Machine Translation to 200 Languages," Nature, vol. 630, no.8018, pp. 841–846, 2024.

[18] T. B. Brown et al., "Language Models are Few-shot Learners," Advances in Neural InformationProcessing Systems, vol. 33, pp. 1877–1901, 2020.

[19] J. Devlin, M. W. Chang, K. Lee and K. Toutanova, "BERT: Pre-training of Deep BidirectionalTransformers for Language Understanding," Proc. of the 2019 Conf. of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, pp. 4171–4186, 2019.

[20] A. Vaswani, N. Shazeer, N. Parmar, J. Uszkoreit, L. Jones, A. N. Gomez, L. Kaiser and I. Polosukhin,"Attention Is All You Need," Adv. Neural Inf. Process. Syst., vol. 30, pp. 5998–6008, 2017.

[21] L. Xue et al., "ByT5: Towards a Token-free Future with Pre-trained Byte-to-Byte Models," Trans. of theAssociation for Computational Linguistics, vol. 10, pp. 291–306, 2022.

[22] B. Al-Rfooh, G. Abandah and R. Al-Rfou, "Fine-Tashkeel: Fine-tuning Byte-level Models for AccurateArabic Text Diacritization," Proc. of the 2023 IEEE Jordan Int’l Joint Conf. on Electrical Engineering and Information Technology (JEEIT), pp. 199–204, 2023.

[23] F. Alzamzami and A. E. Saddik, "OSN-MDAD: Machine Translation Dataset for Arabic Multi-dialectalConversations on Online Social Media," arXiv: 2309.12137, 2023.

[24] E. Nagoudi, A. Elmadany and M. Abdul-Mageed, "AraT5: Text-to-Text Transformers for ArabicLanguage Generation," Proc. of the 60th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pp. 628-647, 2022.

[25] A. Slim and A. Melouah, "Low Resource Arabic Dialects Transformer Neural Machine TranslationImprovement through Incremental Transfer of Shared Linguistic Features," Arabian J. for Science and Engineering, vol. 49, no. 9, pp. 12393–12409, 2024.

[26] L. H. Baniata, S. Park and S. B. Park, "A Multitask-based Neural Machine Translation Model with Part-of-Speech Tags Integration for Arabic Dialects," Applied Sciences, vol. 8, no. 12, pp. 2502, 2018.

[27] T. Alimi, R. Boujebane, W. Derouich and L. Belguith, "Fine-Tuned Transformers for Translating Multi-dialect Texts to Modern Standard Arabic," Int’l J. of Cognitive and Language Sciences, vol. 18, no. 11, pp. 679-684, 2024.

[28] S. Kchaou, R. Boujelbane and L. Hadrich-Belguith, "Parallel Resources for Tunisian Arabic DialectTranslation," Proc. of the 5th Arabic Natural Language Processing Workshop, pp. 200–206, Barcelona, Spain, 2020.

[29] S. Kchaou, R. Boujelbane and L. Hadrich-Belguith, "Hybrid Pipeline for Building Arabic TunisianDialect-Standard Arabic Neural Machine Translation Model from Scratch," ACM Trans. on Asian and Low-Resource Language Information Processing, vol. 22, no. 3, pp. 1–21, 2022.

[30] M. A. Faheem, K. T. Wassif, H. Bayomi and S. M. Abdou, "Improving Neural Machine Translation forLow-resource Languages through Non-parallel Corpora: A Case Study of Egyptian Dialect to Modern Standard Arabic Translation," Scientific Reports, vol. 14, no. 1, 2024.

[31] T. Zerrouki and A. Balla, "Tashkeela: Novel Corpus of Arabic Vocalized Texts, Data for AutoDiacritization Systems," Data Brief, vol. 11, pp. 147–151, 2017.

[32] G. Abandah and A. Abdel-Karim, "Accurate and Fast Recurrent Neural Network Solution for theAutomatic Diacritization of Arabic Text," Jordanian Journal of Computers and Information Technology (JJCIT), vol. 6, no. 2, pp. 103–121, 2020.

[33] R. Younis and G. Abandah, "Automatic Diacritization of Arabic Text and Poetry Using Pretrained Byte-to-Byte Language Models and Multiphase Training," Proc. of the 2025 1st Int’l Conf. on Computational Intelligence Approaches and Applications (ICCIAA), pp. 1-6, Amman, Jordan, 2025.

[34] G. Abandah, "Jordanian Dialect Dataset," GitHub, [online], Available: https://github.com/Gheith-Abandah/JODA, 2025.

[35] I. Alsarsou, E. Mohamed, R. Suwaileh and T. Elsayed, "DART: A Large Dataset of Dialectal ArabicTweets," Proc. of the 11th Int’l Conf. on Language Resources and Evaluation (LREC-2018), Miyazaki, Japan, 2018.

[36] C. Qwaider, M. Saad, S. Chatzikyriakidis and S. Dobnik, "Shami: A Corpus of Levantine ArabicDialects," Proc. of the 11th Int’l Conf. on Language Resources and Evaluation (LREC-2018), Miyazaki, Japan, 2018.

[37] D. Alzuheiri, H. Adwan, G. Abandah and Y. Hamdan, "Converting Colloquial to Classical Arabic aswithin the Project Developing Applications to Correct Jordanian Spoken Arabic to Proper Language Using Machine Learning Techniques," Int’l J. on Islamic Applications in Computer Science and Technology (IJASAT), vol. 12, no. 4, 2024.

[38] A. Fadel, O. Oueslati, H. Gahbiche and R. Shafik, "Neural Arabic Text Diacritization: State of the ArtResults and a Novel Approach for Machine Translation," Proc. of the 6thWorkshop on Asian Translation, pp. 215-225, Hong Kong, China, 2019.

[39] A. Abdel Karim and G. Abandah, "On the Training of Deep Neural Networks for Automatic Arabic-textDiacritization," Int’l J. of Advanced Computer Science and Applications, vol. 12, no. 8, 2021.

[40] M. Khaleel and G. Abandah, "Efficient Stochastic Error Injection for Optimizing Large Language Modelsin Arabic Spelling Correction," Proc. of the 2025 IEEE Int’l Conf. on New Trends in Computing Sciences (ICTCS), pp. 505-510, Amman, Jordan, 2025.

[41] G. Abandah, A. Suyyagh and M. Z. Khedher, "Correcting Arabic Soft Spelling Mistakes Using BiLSTM-based Machine Learning," Int’l J. of Advanced Computer Science and Applications, vol. 13, no. 5, 2022.

[42] P. Phakmongkol and P. Vateekul, "Enhance Text-to-Text Transfer Transformer with GeneratedQuestions for Thai Question Answering," Applied Sciences, vol. 11, no. 21, 2021.

[43] OpenBoard Team, "OpenBoard," GitHub, [Online], Available: https://github.com/openboard-team/openboard, 2022.

[44] KeyboardKit, "KeyboardKit," [Online], Available: https://keyboardkit.com/, 2023.

[45] B. Sadder, R. Sadder, G. Abandah and I. Jafar, "Multi-domain Machine Learning Approach of NamedEntity Recognition for Arabic Booking Chatbot Engines Using Pre-Trained Bidirectional Transformers," Jordanian Journal of Computers and Information Technology (JJCIT), vol. 10, no. 1, pp. 1–16, 2024.