Elamur Research⌕
☰
home / papers / 2026 / AI-TraLow: AI-Driven Translation for Low-Resource Languages and Cultures

AI-TraLow: AI-Driven Translation for Low-Resource Languages and Cultures

Paper · 2026 · Antoni Oliver, Maite Melero, Felipe Sanchez-Martinez et al.

Summary

AI-TraLow is a three-year Spanish coordinated research project that started on 1 September 2025. It aims to close the gap between well-resourced and low-resource languages in LLM-based machine translation. Modern decoder-only models depend on large datasets, and their capacity is spread thin across many languages, so poorly digitised varieties get weak translations. The project plans to address this in three ways. First, it will curate better corpora, including classifiers that flag machine-generated or machine-translated web text. Second, it will feed symbolic resources such as Apertium dictionaries, transfer rules and grammar books into prompts and fine-tuning. Third, it will distil large models into compact systems that run on ordinary desktop hardware. It will also explore multimodal, pixel-based and byte-level models for image-to-text translation and for languages without a standard orthography. The project focuses on minority Iberian Romance varieties, several languages of the Mayan family, and the tongues spoken by at-risk migrant groups. This paper describes the project's plan, so it reports no experimental results yet. All models and tools are promised as open-source releases, which matters for digital inclusion and for preserving cultural heritage.

Key points

  • Consortium: Universitat d'Alacant coordinates the project and leads CEHR-TraLowLa (data curation and linguistic integration). Barcelona Supercomputing Center leads MLLM4TRA (multimodal LLMs and adaptation). UOC handles Iberian Romance languages, resources and evaluation.
  • Target languages: Aranese, Aragonese, Asturian, Eonavian, a subset of Mayan languages, and migrant languages such as Wolof and Amazigh.
  • Structure: nine objectives (O1–O9) and five work packages:
    • WP1: data acquisition and curation
    • WP2: linguistic information in LLMs
    • WP3: image-to-text translation with multimodal LLMs
    • WP4: training and adaptation, including gender-inclusive MT, synthetic data and distillation into encoder-decoder models
    • WP5: release of optimised models
  • Method cycle: a three-phase loop (data compilation, model training and fine-tuning, model release) that will run twice. Models will be tracked with automatic metrics, and the best ones will receive human evaluation.
  • Funding: MICIU, the Spanish Research Agency and FEDER, through grants PID2024-158157OB-C31, -C32 and -C33.

Our summary of the paper, in our own words — not its text. The work belongs to its authors; read it at the links above.

 Institutions

🎓University of AlicanteAlicante · Spain21 works · 1,098 citations🔬Barcelona Supercomputing CenterBarcelona · Spain14 works · 53 citations🎓Universitat Oberta de CatalunyaBarcelona · Spain3 works

 Fields, topics and concepts

📚 Shelves 4The field of science this work belongs to (OpenAlex fields).Computer Science · 16,667 worksArtificial Intelligence · 16,194 worksArts and Humanities · 16,039 worksLanguage and Linguistics · 15,891 works

🧭 Topics 1Research subjects and languages the work is about.Low-resource languages · 267 works

Cite

Cite as Antoni Oliver, Maite Melero, Felipe Sanchez-Martinez and Víctor M. Sánchez-Cartagena. 2026. AI-TraLow: AI-Driven Translation for Low-Resource Languages and Cultures. Proceedings of LANLP: Bridging Ibero and Latin American NLP Communities.

The paper itself, not this page. BibTeX from the library's record of the work. DOI 10.63317/32hkgizmdnhw

@inproceedings{oliver2026ai,
  title = {{AI-TraLow: AI-Driven Translation for Low-Resource Languages and Cultures}},
  author = {Antoni Oliver, Maite Melero, Felipe Sanchez-Martinez and Víctor M. Sánchez-Cartagena},
  booktitle = {Proceedings of LANLP: Bridging Ibero and Latin American NLP Communities},
  year = {2026},
  doi = {10.63317/32hkgizmdnhw},
  url = {https://doi.org/10.63317/32hkgizmdnhw}
}

References

16 references, 6 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. Aarón Galiano-Jiménez, Juan Antonio Pérez-Ortiz, Felipe Sánchez-Martínez, and Víctor M Sánchez-Cartagena. 2025. Multi-hypothesis distillation of multilingual neural translation models for low-resource languages. arXiv preprint arXiv:2507.21568. find
  2. Mehdi Ali, Michael Fromm, Klaudia Thellmann, Richard Rutmann, Max Lübbering, Johannes Leveling, Katrin Klug, Jan Ebert, Niclas Doll, Jasper Schulze Buschhoff, Charvi Jain, Alexander Arno Weber, Lena Jurkschat, Hammam Abdelwahab, Chelsea John, Pedro Ortiz Suarez, Malte Ostendorff, Samuel Weinbach, Ra… — Tokenizer Choice For LLM Training: Negligible or Crucial?
  3. Javier Garcia Gilabert, Carlos Escolano, Xixian Liao, and Maite Melero. 2025. Terminology-constrained translation from monolingual data using GRPO. In Proceedings of the Tenth Conference on Machine Translation, pages 1335–1343, Suzhou, China. Association for Computational Linguistics. find
  4. Duarte M. Alves, José Pombal, Nuno M. Guerreiro, Pedro H. Martins, João Alves, Amin Farajian, Ben Peters, Ricardo Rei, Patrick Fernandes, Sweta Agrawal, Pierre Colombo, José G. C. de Souza, and André F. T. Martins. 2024. Tower: An open multilingual large language model for translation-related tasks.
  5. Cristian García-Romero, Miquel Esplà-Gomis, and Felipe Sánchez-Martínez. 2025. Automatic machine translation detection using a surrogate multilingual translation model. arXiv preprint arXiv:2511.02958.
  6. Cristian García-Romero, Miquel Esplà-Gomis, and Felipe Sánchez-Martínez. 2026. When translations surprise: Human awareness of predictability in translations. In Proceedings of the 15th biennial Language Resources and Evaluation Conference (in press), Palma de Mallorca, Spain.
  7. Ozan Caglayan, Pranava Madhyastha, Lucia Specia, and Loïc Barrault. 2019. Probing the need for visual context in multimodal machine translation. In Proc. of the 2019 Conference of the Naman Goyal, Cynthia Gao, Vishrav Chaudhary, Peng-Jen Chen, Guillaume Wenzek, Da Ju, Sanjana Krishnan, Marc’Aurelio…
  8. Jonathan Hus and Antonios Anastasopoulos. 2024. Back to school: Translation using grammar books. In Proc. of the 2024 Conference on Empirical Methods in Natural Language Processing, pages 20207–20219, Miami, Florida, USA. find
  9. Aleksandar Petrov, Emanuele La Malfa, Philip Torr, and Adel Bibi. 2024. Language model tokenizers introduce unfairness between languages. Advances in Neural Information Processing Systems, 36.
  10. Elizabeth Salesky, Philipp Koehn, and Matt Post. 2024. Benchmarking visually-situated translation of text in natural images. In Proc. of the Ninth Conference on Machine Translation, pages 1167–1182, Miami, Florida, USA. find
  11. Garrett Tanzer, Mirac Suzgun, Eline Visser, Dan Jurafsky, and Luke Melas-Kyriazi. 2024. A benchmark for learning to translate a new language from one grammar book. find
  12. Haoran Xu, Young Jin Kim, Amr Sharaf, and Hany Hassan Awadalla. 2024. A paradigm shift in machine translation: Boosting translation performance of large language models. find
  13. Linting Xue, Aditya Barua, Noah Constant, Rami Al-Rfou, Sharan Narang, Mihir Kale, Adam Roberts, and Colin Raffel. 2022. ByT5: Towards a token-free future with pre-trained byte-to-byte models. Transactions of the Association for Computational Linguistics, 10:291–306. find
  14. Wen Yang, Chong Li, Jiajun Zhang, and Chengqing Zong. 2023. Bigtranslate: Augmenting large language models with multilingual translation capability over 100 languages. find
  15. Chen Zhang, Xiao Liu, Jiuheng Lin, and Yansong Feng. 2024a. Teaching large language models an unseen language on the fly. find
  16. Kexun Zhang, Yee Choi, Zhenqiao Song, Taiqi He, William Yang Wang, and Lei Li. 2024b. Hire a linguist!: Learning endangered languages in LLMs with in-context linguistic descriptions. In Findings of the Association for Computational Linguistics: ACL 2024, pages 15654–15669, Bangkok, Thailand. find

Updated

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/oliver-2026-ai-tralow-ai-driven-translation/