Elamur Research⌕
☰
home / papers / 2026 / A Comparative Study of Multilingual Fine-tuning and Prompting for Automatic Text Readability Classification in Galician

A Comparative Study of Multilingual Fine-tuning and Prompting for Automatic Text Readability Classification in Galician

Paper · 2026 · Sandra Rodríguez Rey and Marcos Garcia

Summary

Automatic readability assessment has advanced mainly for well-resourced languages, while Galician has very little annotated material. The authors compare ways to build four-level readability classifiers for Galician. Training data come from the iRead4Skills Dataset 1 (Spanish, Portuguese, French), used both in its original form and after machine translation into Galician with SalamandraTA. Evaluation uses the native Galician Corlega corpus. Three strategies are tested:

  • fine-tuning monolingual BERT-style encoders on the translated material;
  • fine-tuning multilingual encoders on original versus translated data, to measure the value of translation-based augmentation;
  • zero-shot and few-shot prompting of several LLMs.

Prompted generative models still trail fine-tuned encoders on this task. Machine-translated data helps monolingual Galician models but adds little for multilingual ones, which already cover the source languages. The study offers practical guidance for readability modelling in low-resource settings. It also supports educational tools that could help teachers and learners find Galician reading material at the right level. The best models are released openly on Hugging Face.

Key points

  • Corlega corpus: 480 Galician texts (145,854 tokens) from genres such as social media, literature, ads, political and legal texts. One expert labelled them into four levels aligned with Celga, CEFR and the iRead4Skills scheme. It is released on Zenodo under CC BY-NC-ND 4.0.
  • Training data: iRead4Skills Dataset 1 contains roughly 2,000–3,000 texts per language (Spanish, Portuguese, French) on four complexity levels. A Galician version was produced with SalamandraTA.
  • Data split: Corlega served only for validation (25%) and evaluation (75%), with level-balanced random selection. Class imbalance in the training sets was handled through oversampling.
  • Main findings: Fine-tuned encoders outperform zero-/few-shot LLM prompting for Galician. Translation-based augmentation benefits monolingual models but has a minimal effect on multilingual ones.
  • Released models: The best-performing models are published on Hugging Face (huggingface.co/sandrarrey).

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

🎓Universidade de Santiago de CompostelaSantiago de Compostela · Spain13 works · 364 citations

 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 2Research subjects and languages the work is about.Language models & pretraining · 3,320 worksMultilingual & cross-lingual NLP · 828 works

Cite

Cite as Sandra Rodríguez Rey and Marcos Garcia. 2026. A Comparative Study of Multilingual Fine-tuning and Prompting for Automatic Text Readability Classification in Galician. Proceedings of the Joint Workshop on Readability and Text Simplification (READIxTSAR) @ LREC 2026.

The paper itself, not this page. BibTeX from the library's record of the work. DOI 10.63317/4tnwhe3r9579

@inproceedings{rey2026comparative,
  title = {{A Comparative Study of Multilingual Fine-tuning and Prompting for Automatic Text Readability Classification in Galician}},
  author = {Sandra Rodríguez Rey and Marcos Garcia},
  booktitle = {Proceedings of the Joint Workshop on Readability and Text Simplification (READIxTSAR) @ LREC 2026},
  year = {2026},
  doi = {10.63317/4tnwhe3r9579},
  url = {https://doi.org/10.63317/4tnwhe3r9579}
}

References

45 references, 9 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. Onder Coban, Mete Yağanoğlu, and Ferhat Bozkurt. 2024. Domain effect investigation for bert models fine-tuned on different text categorization tasks. Arabian Journal for Science and Engineering, 49(3):3685–3702. find
  2. Alexis Conneau, Kartikay Khandelwal, Naman Goyal, Vishrav Chaudhary, Guillaume Wenzek, Francisco Guzmán, Edouard Grave, Myle Ott, Luke Zettlemoyer, and Veselin Stoyanov. 2019. Unsupervised cross-lingual representation learning at scale. CoRR, abs/1911.02116. find
  3. Scott Andrew Crossley, David F. Dufty, Philip M. McCarthy, and Danielle S. McNamara. 2007. Toward a new readability: A mixed model approach. In Proceedings of the annual meeting of the cognitive science society, volume 29. find
  4. Edgar Dale and Jeanne Sternlicht Chall. 1948. A formula for predicting readablility. Bureau of Educational Research, Ohio State University. find
  5. Iria de Dios-Flores, Silvia Paniagua Suárez, Cristina Carbajal Pérez, Daniel Bardanca Outeiriño, Marcos Garcia, and Pablo Gamallo. 2024. CorpusNÓS: A massive Galician corpus for training large language models. In Proceedings of the 16th International Conference on Computational Processing of Portugu… find
  6. Felice Dell’Orletta, Simonetta Montemagni, and Giulia Venturi. 2014. Assessing document and sentence readability in less resourced languages and across textual genres. ITL - International Journal of Applied Linguistics, 165:163–193. find
  7. Jacob Devlin, Ming-Wei Chang, Kenton Lee, and Kristina Toutanova. 2019. BERT: Pre-training of deep bidirectional transformers for language understanding. In Proceedings of the 2019 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies,…
  8. Rudolf Flesch. 1948. A readability formula in practice. Elementary English, 25(6):344–351. find
  9. Marcos Garcia. 2021. Exploring the representation of word meanings in context: A case study on homonymy and synonymy. In Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language Processing (Volume 1: Long… find
  10. José Ángel González, Ian Borrego Obrador, Álvaro Romo Herrero, Areg Mikael Sarvazyan, Mara Chinea-Ríos, Angelo Basile, and Marc Franco-Salvador. 2026. Iberbench: Llm evaluation on iberian languages. Computer Speech & Language, 96:101899. find
  11. Joseph Marvin Imperial. 2021. BERT embeddings for automatic readability assessment. In Proceedings of the International Conference on Recent Advances in Natural Language Processing (RANLP 2021), pages 611–618, Held Online. IN-COMA Ltd. find
  12. Itziar Gonzalez-Dios, María Jesús Aranzabe, and Arantza Díaz de Ilarraza. 2018. The corpus of basque simplified texts (cbst). Language Resources and Evaluation, 52(1):217–247. find
  13. Joseph Marvin Imperial, Abdullah Barayan, Regina Stodden, Rodrigo Wilkens, Ricardo Muñoz Sánchez, Lingyun Gao, Melissa Torgbi, Dawn Knight, Gail Forey, Reka R. Jablonkai, Ekaterina Kochmar, Robert Joshua Reynolds, Eugénio Ribeiro, Horacio Saggion, Elena Volodina, Sowmya Vajjala, Thomas François, Fer… find
  14. Naman Goyal, Cynthia Gao, Vishrav Chaudhary, Peng-Jen Chen, Guillaume Wenzek, Da Ju, Sanjana Krishnan, Marc’Aurelio Ranzato, Francisco Guzmán, and Angela Fan. 2022. The Flores-101 Evaluation Benchmark for Low-Resource and Multilingual Machine Translation. Transactions of the Association for Computat… find
  15. Aaron Grattafiori, et al., 2024. The Llama 3 Herd of Models.
  16. G Harry and Mc Laughlin. 1969. Smog grading - a new readability formula. The Journal of Reading. find
  17. Nicolas Hernandez, Nabil Oulbaz, and Tristan Faine. 2022. Open corpora and toolkit for assessing text readability in French. In Proceedings of the 2nd Workshop on Tools and Resources to Empower People with REAding DIfficulties (READI) within the 13th Language Resources and Evaluation Conference, pag… find
  18. Carina Kauf and Anna A. . 2023. A better way to do masked language model scoring. In Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers), pages 925–935, Toronto, Canada. Association for Computational Linguistics.
  19. Arina Kostina, Marios D. Dikaiakos, Dimosthenis Stefanidis, and George Pallis. 2025. Large language models for text classification: Case study and comprehensive review. find
  20. Yuxuan Liu, Tianchi Yang, Shaohan Huang, Zihan Zhang, Haizhen Huang, Furu Wei, Weiwei Deng, Feng Sun, and Qi Zhang. 2024. Calibrating LLM-Based Evaluator. In Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING 2024), 263…
  21. Ion Madrazo Azpiazu and Maria Soledad Pera. 2019. Multiattentive recurrent neural network architecture for multilingual readability assessment. Transactions of the Association for Computational Linguistics, 7:421–436.
  22. Kanishka Misra. 2022. minicons: Enabling flexible behavioral and representational analyses of transformer language models. arXiv preprint arXiv:2203.13112. find
  23. Ricardo Monteiro, Raquel Amaro, Susana Correia, Alice Pintard, Roser Gauchola, Michell Moutinho, and Xavier Blanco Escoda. 2024. iRead4Skills -Complexity Levels. find
  24. Duy Van Ngo and Yannick Parmentier. 2023. Towards sentence-level text readability assessment for French. In Proceedings of the Second Workshop on Text Simplification, Accessibility and Readability, pages 78–84, Varna, Bulgaria. IN-COMA Ltd., Shoumen, Bulgaria. find
  25. Partha Pakray, Alexander Gelbukh, and Sivaji Bandyopadhyay. 2025. Natural language processing applications for low-resource languages. Natural Language Processing, 31(2):183–197. find
  26. Taja Kuzman Pungeršek, Peter Rupnik, Ivan Porup-ski, Vuk Dinić, and Nikola Ljubešić. 2025. State of the art in text classification for south slavic languages: Fine-tuning or prompting?
  27. Andre Quispesaravia, Walter Perez, Marco Sobrevilla Cabezudo, and Fernando Alva-Manchego. 2016. Coh-Metrix-Esp: A complexity analysis tool for documents written in Spanish. In Proceedings of the Tenth International Conference on Language Resources and Evaluation (LREC’16), pages 4694–4698, Portorož,… find
  28. Muhammad Rehan, Muhammad Shahid Iqbal Malik, and Mona Mamdouh Jamjoom. 2023. Fine-tuning transformer models using transfer learning for multilingual threatening text identification. IEEE Access, 11:106503–106515. find
  29. Eugénio Ribeiro, Nuno Mamede, and Jorge Baptista. 2024. Automatic text readability assessment in European Portuguese. In Proceedings of the 16th International Conference on Computational Processing of Portuguese - Vol. 1, pages 97–107, Santiago de Compostela, Galicia/Spain. Association for Computati… find
  30. Sandra Rodríguez Rey and Marcos Garcia. 2025. Clasificación automática de textos por niveis de lecturabilidade: recursos e modelos para o galego. Linguamática, 17(2):33–56. find
  31. Sarah Schwarm and Mari Ostendorf. 2005. Reading level assessment using support vector machines and statistical language models. In Proceedings of the 43rd Annual Meeting of the Association for Computational Linguistics (ACL’05), pages 523–530, Ann Arbor, Michigan. Association for Computational Lingu…
  32. Johannes Sibeko. 2024. Harnessing google translations to develop a readability corpus for sesotho: An exploratory study. Journal of the Digital Humanities Association of Southern Africa (DHASA), 5(1). find
  33. Qwen Team. 2024. Qwen2.5: A party of foundation models. find
  34. Mykola Trokhymovych, Indira Sen, and Martin Gerlach. 2024. An open multilingual system for scoring readability of Wikipedia. In Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pages 6296–6311, Bangkok, Thailand. Association for Computa… find
  35. Sowmya Vajjala. 2022. Trends, Limitations and Open Challenges in Automatic Readability Assessment Research. In Proceedings of the Thirteenth Language Resources and Evaluation Conference, 5366–5377. find
  36. Laura Vásquez-Rodríguez, Pedro-Manuel Cuenca-Jiménez, Sergio Morales-Esquivel, and Fernando Alva-Manchego. 2022. A benchmark for neural readability assessment of texts in Spanish. In Proceedings of the Workshop on Text Simplification, Accessibility, and Readability (TSAR-2022), pages 188–198, Abu Dh… find
  37. David Vilares, Marcos Garcia, and Carlos Gómez-Rodríguez. 2021. Bertinho: Galician bert representations. Procesamiento del Lenguaje Natural, 66(0):13–26. find
  38. Rodrigo Wilkens, David Alfter, Xiaoou Wang, Alice Pintard, Anaïs Tack, Kevin P. Yancey, and Thomas François. 2022. FABRA: French aggregator-based readability assessment toolkit. In Proceedings of the Thirteenth Language Resources and Evaluation Conference, pages 1217–1233, Marseille, France. Europea… find
  39. Thomas Wolf, Lysandre Debut, Victor Sanh, Julien Chaumond, Clement Delangue, Anthony Moi, Pierric Cistac, Tim Rault, Remi Louf, Morgan Funtowicz, Joe Davison, Sam Shleifer, Patrick von Platen, Clara Ma, Yacine Jernite, Julien Plu, Canwen Xu, Teven Le Scao, Sylvain Gugger, Mariama Drame, Quentin Lhoe… — Transformers: State-of-the-Art Natural Language Processing
  40. Weihao Xuan, Rui Yang, Heli Qi, Qingcheng Zeng, Yunze Xiao, Aosong Feng, Dairui Liu, Yun Xing, Junjue Wang, Fan Gao, Jinghui Lu, Yuang Jiang, Huitao Li, Xin Li, Kunyu Yu, Ruihai Dong, Shangding Gu, Yuekang Li, Xiaofei Xie, Felix Juefei-Xu, Foutse Khomh, Osamu Yoshie, Qingyu Chen, Douglas Teodoro, Na… — MMLU-ProX: A Multilingual Benchmark for Advanced Large Language Model Evaluation
  41. An Yang, Baosong Yang, Binyuan Hui, Bo Zheng, Bowen Yu, Chang Zhou, Chengpeng Li, Chengyuan Li, Dayiheng Liu, Fei Huang, Guanting Dong, Haoran Wei, Huan Lin, Jialong Tang, Jialin Wang, Jian Yang, Jianhong Tu, Jianwei Zhang, Jianxin Ma, Jin Xu, Jingren Zhou, Jinze Bai, Jinzheng He, Junyang Lin, Kai D… find
  42. Amirhossein Yousefiramandi and Ciaran Cooney. 2025. Fine-Tuning Causal LLMs for Text Classification: Embedding-Based vs. Instruction-Based Approaches. arXiv preprint arXiv:2512.12677. find
  43. A Ziyaden, A Yelenov, F Hajiyev, S Rustamov, and A Pak. 2024. Text data augmentation and pre-trained language model for enhancing text classification of low-resource languages. PeerJ Computer Science, 10:e1974. 12. find
  44. Language Resource References Alice Pintard, Thomas François, Justine Nagant de Deuxchaisnes, Sílvia Barbosa, Maria Leonor Reis, Michell Moutinho, Ricardo Monteiro, Raquel Amaro, Susana Correia, Sandra Rodríguez Rey, Marcos Garcia González, Keran Mu, and Xavier Blanco Escoda. 2024. iRead4Skills Datas… find
  45. Sandra Rodríguez Rey and Marcos Garcia. 2025. Corlega: Corpus de Lecturabilidade en Galego. 13. find

Updated

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/rey-2026-comparative-study-multilingual-fine-tuning/