---
title: "Document-Level Text Simplification in Estonian — Muru 2026"
description: "Most text simplification research rewrites one sentence at a time and focuses on English."
authors: ["Meeri-Ly Muru", "Eduard Barbu"]
year: 2026
venue: "Proceedings of the Fifteenth Language Resources and Evaluation Conference"
source: "https://aclanthology.org/2026.lrec-1.573/"
doi: "10.63317/3ndqrcjckj32"
licence: "CC BY-NC 4.0 (ELRA, LREC proceedings)"
licence_url: "https://creativecommons.org/licenses/by-nc/4.0/"
summary_licence: "Elamur Cosmic License"
canonical: "https://research.elamur.ai/free/papers/muru-2026-document-level-text-simplification-estonian/"
---
[Elamur Research](https://md.research.elamur.ai/) · [this page in HTML](https://research.elamur.ai/free/papers/muru-2026-document-level-text-simplification-estonian/)

[home](https://md.research.elamur.ai/) / [papers](https://md.research.elamur.ai/free/papers/) / [2026](https://md.research.elamur.ai/free/papers/2026/page/2/) / Document-Level Text Simplification in Estonian Using Large Language Models

# Document-Level Text Simplification in Estonian Using Large Language Models

**Authors:** Meeri-Ly Muru, Eduard Barbu · **Year:** 2026 · **Venue:** Proceedings of the Fifteenth Language Resources and Evaluation Conference · **Licence:** [CC BY-NC 4.0 (ELRA, LREC proceedings)](https://creativecommons.org/licenses/by-nc/4.0/)

**Links:** [Publisher's page](https://aclanthology.org/2026.lrec-1.573/) · [PDF](https://aclanthology.org/2026.lrec-1.573.pdf) · [DOI 10.63317/3ndqrcjckj32](https://doi.org/10.63317/3ndqrcjckj32) · [OpenAlex](https://openalex.org/W7160142486)

**Citations:** 0 citations (openalex-snapshot · 2026-10-14)

## Summary
Most text simplification research rewrites one sentence at a time and focuses on English. Little is known about simplifying whole documents in morphologically rich, low-resource languages, where edits must also keep discourse coherent, references clear and paragraphs consistent. This paper studies document-level simplification for Estonian. The authors test five multilingual instruction-tuned LLMs on Estonian Wikipedia articles, using three prompting setups:
- one combined instruction,
- a sequential pipeline of role-based agents,
- the same pipeline enriched with simplification guidelines.

Outputs are judged with automatic measures of readability, meaning preservation and coherence, including newly proposed coherence metrics. A structured manual annotation complements these scores. Gemini-2.0 and LLaMA-3.3 performed best, producing almost native-sounding Estonian that kept the source meaning. The other models made noticeable grammatical and semantic errors. The study is an early systematic treatment of this task for Estonian. It also releases prompts, code and resources so the experiments can be reproduced.

## Key points
- Models compared: GPT-4.1, LLaMA-3.3-70B-Instruct, Claude-3.5-Sonnet, Gemini-2.0-flash-001 and Qwen-2.5-72B-Instruct.
- The Pipeline-Only strategy chains three agents: Terminology Interpreter (TI), Structural Simplifier (SS) and Anaphora Agent (AA). Each agent passes its output to the next, following the multi-agent idea of Fang et al. (2025a).
- Prompts target specific document-level operations: deleting, joining, splitting, reordering and adding sentences, plus anaphora resolution. Single-pass prompting tended to produce only light changes to the source.
- The new coherence metrics rely on sentence similarity and on whether discourse markers are preserved. Human evaluation covers fluency, grammaticality, meaning preservation and discourse coherence.
- The paper also asks whether automatic metrics alone can identify the best model (RQ2). The detailed results section is missing from the provided text. The excerpt only reports Gemini-2.0 and LLaMA-3.3 as the strongest systems.

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

Fields: [Linguistics & languages](https://md.research.elamur.ai/free/shelves/linguistics/) · [Machine translation](https://md.research.elamur.ai/free/shelves/machine-translation/)

Topics: [Discourse analysis](https://md.research.elamur.ai/free/topics/discourse-analysis/)

## Cite

The paper itself, not this page. BibTeX from the library's record of the work.

```bibtex
@inproceedings{muru2026document,
  title = {{Document-Level Text Simplification in Estonian Using Large Language Models}},
  author = {Muru, Meeri-Ly and Barbu, Eduard},
  booktitle = {Proceedings of the Fifteenth Language Resources and Evaluation Conference},
  year = {2026},
  doi = {10.63317/3ndqrcjckj32},
  url = {https://doi.org/10.63317/3ndqrcjckj32}
}
```

- [← previous Do LLMs Judge Distantly Supervised Named Entity Labels Well? Constructing the JudgeWEL Dataset](https://md.research.elamur.ai/free/papers/plum-2026-do-llms-judge-distantly-supervised/)
- [next → Domain-Aware Error Correction for Citation NER in Medieval Hebrew Responsa](https://md.research.elamur.ai/free/papers/aclbulk-2026-liebeskind-domain-aware-error-correction-citation/)
- [↑ back to papers of 2026](https://md.research.elamur.ai/free/papers/2026/page/2/)

## References

25 references, 8 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

1. Fernando Alva-Manchego, Carolina Scarton, and Lucia Specia. 2020. Data-driven sentence simplification: Survey and benchmark. Computational Linguistics, 46(1):135–187. [↗](https://doi.org/10.1162/coli_a_00370) [our card](https://md.research.elamur.ai/free/papers/manchego-2020-data-driven-sentence-simplification-survey/)
2. Anthropic. 2024. Claude 3.5 sonnet. [https://www.anthropic.com/news/claude-3-5-sonnet](https://www.anthropic.com/news/claude-3-5-sonnet). Large language model developed by Anthropic. [↗](https://www.anthropic.com/news/claude-3-5-sonnet)
3. Eduard Barbu, M. Teresa Martín-Valdivia, Eugenio Martínez-Cámara, and L. Alfonso Ureña-López. 2015. Language technologies applied to document simplification for helping autistic people. Expert Systems with Applications, 42(12):5076– 5086. [↗](https://doi.org/10.1016/j.eswa.2015.02.044)
4. Eduard Barbu, Meeri-Ly Muru, and Sten Marcus Malva. 2025. Improving estonian text simplification through pretrained language models and custom datasets. In Proceedings of the 15th International Conference on Recent Advances in Natural Language Processing - Natural Language Processing in the Generative AI era, pages 133– 142, Varna, Bulgaria. INCOMA Ltd., Shoumen, Bulgaria. [find](https://scholar.google.com/scholar?q=Eduard+Barbu%2C+Meeri%2DLy+Muru%2C+and+Sten+Marcus+Malva%2E+2025%2E+Improving+estonian+text+simplification+through+pretrained+language+models+and+custom+datasets%2E+In+Proceedings+of+the+15th+International+Confer)
5. Raman Chandrasekar, Christine Doran, and B. Srinivas. 1996. Motivations and methods for text simplification. In Proceedings of the 16th Conference on Computational Linguistics (COL-ING), pages 1041–1044. [find](https://scholar.google.com/scholar?q=Raman+Chandrasekar%2C+Christine+Doran%2C+and+B%2E+Srinivas%2E+1996%2E+Motivations+and+methods+for+text+simplification%2E+In+Proceedings+of+the+16th+Conference+on+Computational+Linguistics+%28COL%2DING%29%2C+pages+1041%E2%80%9310) [our card](https://md.research.elamur.ai/free/papers/chandrasekar-1996-motivations-methods-text-simplification/)
6. Jacob Cohen. 1968. Weighted kappa: Nominal scale agreement with provision for scaled disagreement or partial credit. Psychological Bulletin, 70(4):213–220. [find](https://scholar.google.com/scholar?q=Jacob+Cohen%2E+1968%2E+Weighted+kappa%3A+Nominal+scale+agreement+with+provision+for+scaled+disagreement+or+partial+credit%2E+Psychological+Bulletin%2C+70%284%29%3A213%E2%80%93220%2E)
7. Isabel Espinosa-Zaragoza, José Abreu-Salas, Elena Lloret, Paloma Moreda, and Manuel Palomar. 2023. A review of research-based automatic text simplification tools. In Proceedings of the 14th International Conference on Recent Advances in Natural Language Processing, pages 321–330, Varna, Bulgaria. INCOMA Ltd., Shoumen, Bulgaria. [find](https://scholar.google.com/scholar?q=Isabel+Espinosa%2DZaragoza%2C+Jos%C3%A9+Abreu%2DSalas%2C+Elena+Lloret%2C+Paloma+Moreda%2C+and+Manuel+Palomar%2E+2023%2E+A+review+of+research%2Dbased+automatic+text+simplification+tools%2E+In+Proceedings+of+the+14th+Internatio)
8. Dengzhao Fang, Jipeng Qiang, Xiaoye Ouyang, Yi Zhu, Yunhao Yuan, and Yun Li. 2025a. Collaborative document simplification using multi-agent systems. In Proceedings of the 31st International Conference on Computational Linguistics, pages 897–912, Abu Dhabi, UAE. Association for Computational Linguistics. [find](https://scholar.google.com/scholar?q=Dengzhao+Fang%2C+Jipeng+Qiang%2C+Xiaoye+Ouyang%2C+Yi+Zhu%2C+Yunhao+Yuan%2C+and+Yun+Li%2E+2025a%2E+Collaborative+document+simplification+using+multi%2Dagent+systems%2E+In+Proceedings+of+the+31st+International+Conference)
9. Dengzhao Fang et al. 2025b. Progressive document-level text simplification via large language models. arXiv preprint arXiv:2501.03857. [↗](https://arxiv.org/abs/2501.03857)
10. Núria Gala and Rodrigo Wilkens, editors. 2020. Proceedings of the 1st Workshop on Tools and Resources to Empower People with REAding DIfficulties (READI). European Language Resources Association, Marseille, France. [find](https://scholar.google.com/scholar?q=N%C3%BAria+Gala+and+Rodrigo+Wilkens%2C+editors%2E+2020%2E+Proceedings+of+the+1st+Workshop+on+Tools+and+Resources+to+Empower+People+with+REAding+DIfficulties+%28READI%29%2E+European+Language+Resources+Association%2C+Mars)
11. Colin Raffel, Noam Shazeer, Adam Roberts, Google DeepMind. 2024. Gem-Katherine Lee, Sharan Narang, Michael Matena, ini 2.0. [https://blog.google/Yanqi](https://blog.google/Yanqi) [↗](https://blog.google/Yanqi)
12. Zhou, Wei Li, and Peter J. Liu. 2020. Ex-technology/google-deepmind/ ploring the limits of transfer learning with a unified google-gemini-ai-update-december-2024/. text-to-text transformer. J. Mach. Learn. Res., Large multimodal AI model for advanced reason-21(1). ing and agentic capabilities. [find](https://scholar.google.com/scholar?q=Zhou%2C+Wei+Li%2C+and+Peter+J%2E+Liu%2E+2020%2E+Ex%2Dtechnology%2Fgoogle%2Ddeepmind%2F+ploring+the+limits+of+transfer+learning+with+a+unified+google%2Dgemini%2Dai%2Dupdate%2Ddecember%2D2024%2F%2E+text%2Dto%2Dtext+transformer%2E+J%2E+Mach%2E+L)
13. Nils Reimers and Iryna Gurevych. 2019. Sentence-J. Peter Kincaid, Robert P. Fishburne, Richard L. bert: Sentence embeddings using siamese bert-Rogers, and Brad S. Chissom. 1975. Deriva-networks. In Conference on Empirical Methods tion of new readability formulas (automated read-in Natural Language Processing. ability index, fog count and flesch reading ease formula) for navy enlisted personnel. Techni-Michael J. Ryan, Tarek Qiang, Moussa Naous, and cal Report Research Branch Report 8-75, Naval Wei Xu. 2023. Revisiting non-English text sim-Technical Training Command, Millington TN, Re-plification: A unified multilingual benchmark. In search Branch. Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Vol-Mounica Maddela, Fernando Alva-Manchego, and ume 1: Long Papers), pages 8452–8480. [find](https://scholar.google.com/scholar?q=Nils+Reimers+and+Iryna+Gurevych%2E+2019%2E+Sentence%2DJ%2E+Peter+Kincaid%2C+Robert+P%2E+Fishburne%2C+Richard+L%2E+bert%3A+Sentence+embeddings+using+siamese+bert%2DRogers%2C+and+Brad+S%2E+Chissom%2E+1975%2E+Deriva%2Dnetworks%2E+In+Co)
14. Asso-Wei Xu. 2021. Controllable text simplification with ciation for Computational Linguistics. explicit paraphrasing. In Proceedings of the 2021 Conference of the North American Chapter of the Horacio Saggion. 2017. Automatic Text Simplifica-Association for Computational Linguistics: Hu-tion. Synthesis Lectures on Human Language man Language Technologies, pages 3536–3553, Technologies. Morgan & Claypool Publishers. Online. Association for Computational Linguis-ISBN: 1627058680, 9781627058681. tics. [find](https://scholar.google.com/scholar?q=Asso%2DWei+Xu%2E+2021%2E+Controllable+text+simplification+with+ciation+for+Computational+Linguistics%2E+explicit+paraphrasing%2E+In+Proceedings+of+the+2021+Conference+of+the+North+American+Chapter+of+the+Horaci)
15. Matthew Shardlow. 2014. A survey of automated text simplification. International Journal Justina Mandravickaitė, Egle Rimkiene, Danguolė of Advanced Computer Science and Applica-Kalinauskaitė, and Danguolė Kotryna Kapkan. tions(IJACSA), Special Issue on Natural Lan-2024. Exploring automatic text simplification for guage Processing 2014, 4(1). Lithuanian. In Proceedings of the 20th Conference on Natural Language Processing (KON-Advaith Siddharthan. 2006. Syntactic simplification VENS 2024), pages 40–49, Vienna, Austria. As-and text cohesion. Research on Language and sociation for Computational Linguistics. Computation, 4(1):77–109. [↗](https://doi.org/10.14569/specialissue.2014.040109)
16. Ruslan Mitkov. 2002. Anaphora Resolution, 1st Renliang Sun, Hanqi Jin, and Xiaojun Wan. 2021. edition. Routledge, London, UK. Document-level text simplification: Dataset, criteria and baseline. In Proceedings of the 2021 [find](https://scholar.google.com/scholar?q=Ruslan+Mitkov%2E+2002%2E+Anaphora+Resolution%2C+1st+Renliang+Sun%2C+Hanqi+Jin%2C+and+Xiaojun+Wan%2E+2021%2E+edition%2E+Routledge%2C+London%2C+UK%2E+Document%2Dlevel+text+simplification%3A+Dataset%2C+criteria+and+baseline%2E+In+Pro)
17. OpenAI. 2023. Gpt-4 technical report. arXiv Conference on Empirical Methods in Natural Lan-preprint arXiv:2303.08774. guage Processing, pages 7997–8013, Online Rashmi Prasad, Nikhil Dinesh, Alan Lee, Eleni Milt-and Punta Cana, Dominican Republic. Associa-sakaki, Livio Robaldo, Aravind Joshi, and Bonnie tion for Computational Linguistics. [↗](https://arxiv.org/abs/2303.08774) [our card](https://md.research.elamur.ai/free/papers/openai-2023-gpt-4-technical-report/)
18. Webber. 2008. The Penn Discourse TreeBank Hugo Touvron, Thibaut Lavril, Gautier Izacard, 2.0. In Proceedings of the Sixth International Xavier Martinet, Marie-Anne Lachaux, Timothée Conference on Language Resources and Evalu-Lacroix, Baptiste Rozière, Naman Goyal, Eric ation (LREC’08), Marrakech, Morocco. European Hambro, Faisal Azhar, Aurelien Rodriguez, Ar-Language Resources Association (ELRA). mand Joulin, Edouard Grave, and Guillaume Qwen, An Yang, Baosong Yang, Beichen Zhang, Lample. 2023. Llama: Open and efficient foun-Binyuan Hui, Bo Zheng, Bowen Yu, Chengyuan dation language models. [find](https://scholar.google.com/scholar?q=Webber%2E+2008%2E+The+Penn+Discourse+TreeBank+Hugo+Touvron%2C+Thibaut+Lavril%2C+Gautier+Izacard%2C+2%2E0%2E+In+Proceedings+of+the+Sixth+International+Xavier+Martinet%2C+Marie%2DAnne+Lachaux%2C+Timoth%C3%A9e+Conference+on+Lang) [our card](https://md.research.elamur.ai/free/papers/miltsakaki-2004-penn-discourse-treebank/)
19. Li, Dayiheng Liu, Fei Huang, Haoran Wei, Huan Lin, Jian Yang, Jianhong Tu, Jianwei Zhang, Jianxin Yang, Jiaxi Yang, Jingren Zhou, Junyang Lin, Kai Dang, Keming Lu, Keqin Bao, Kexin Yang, Le Yu, Mei Li, Mingfeng Xue, Pei Zhang, Qin Zhu, Rui Men, Runji Lin, Tianhao Li, Tianyi Tang, Tingyu Xia, Xingzhang Ren, Xuancheng Ren, Yang Fan, Yang Su, Yichang Zhang, Yu Wan, Yuqiong Liu, Zeyu Cui, Zhenru Zhang, and Zihan Qiu. 2025. Qwen2.5 technical report. [find](https://scholar.google.com/scholar?q=Li%2C+Dayiheng+Liu%2C+Fei+Huang%2C+Haoran+Wei%2C+Huan+Lin%2C+Jian+Yang%2C+Jianhong+Tu%2C+Jianwei+Zhang%2C+Jianxin+Yang%2C+Jiaxi+Yang%2C+Jingren+Zhou%2C+Junyang+Lin%2C+Kai+Dang%2C+Keming+Lu%2C+Keqin+Bao%2C+Kexin+Yang%2C+Le+Yu%2C+Mei+Li)
20. Laura Vasquez Rodriguez et al. 2024. Simple is not enough: Document-level text simplification using readability and coherence. arXiv preprint arXiv:2412.18655. [↗](https://arxiv.org/abs/2412.18655)
21. Wei Xu, Chris Callison-Burch, and Courtney Napoles. 2015. Problems in current text simplification research: New data can help. Transactions of the Association for Computational Linguistics, 3:283–297. [↗](https://doi.org/10.1162/tacl_a_00139) [our card](https://md.research.elamur.ai/free/papers/xu-2015-problems-current-text-simplification-research/)
22. Wei Xu, Courtney Napoles, Ellie Pavlick, Quanze Chen, and Chris Callison-Burch. 2016. Optimizing statistical machine translation for text simplification. Transactions of the Association for Computational Linguistics, 4:401–415. [↗](https://doi.org/10.1162/tacl_a_00107) [our card](https://md.research.elamur.ai/free/papers/xu-2016-optimizing-statistical-machine-translation-text/)
23. Tianyi Zhang, Varsha Kishore, Felix Wu, Kilian Q. Weinberger, and Yoav Artzi. 2019. Bertscore: Evaluating text generation with bert. ArXiv, abs/1904.09675. [↗](https://arxiv.org/abs/1904.09675)
24. Xingxing Zhang and Mirella Lapata. 2017. Sentence simplification with deep reinforcement learning. In Proceedings of the 2017 Conference on Empirical Methods in Natural Language Processing, pages 584–594, Copenhagen, Denmark. Association for Computational Linguistics. [find](https://scholar.google.com/scholar?q=Xingxing+Zhang+and+Mirella+Lapata%2E+2017%2E+Sentence+simplification+with+deep+reinforcement+learning%2E+In+Proceedings+of+the+2017+Conference+on+Empirical+Methods+in+Natural+Language+Processing%2C+pages+584%E2%80%93) [our card](https://md.research.elamur.ai/free/papers/zhang-2017-sentence-simplification-deep-reinforcement-learning/)
25. Zhemin Zhu, Delphine Bernhard, and Iryna Gurevych. 2010. A monolingual tree-based translation model for sentence simplification. In Proceedings of the 23rd International Conference on Computational Linguistics (COLING), pages 1353–1361. [find](https://scholar.google.com/scholar?q=Zhemin+Zhu%2C+Delphine+Bernhard%2C+and+Iryna+Gurevych%2E+2010%2E+A+monolingual+tree%2Dbased+translation+model+for+sentence+simplification%2E+In+Proceedings+of+the+23rd+International+Conference+on+Computational+Li) [our card](https://md.research.elamur.ai/free/papers/zhu-2010-monolingual-tree-based-translation-model/)