Elamur Research⌕
☰
home / papers / 2026 / Extending Creativity: Large Language Models and the Practice of Poetry Translation

Extending Creativity: Large Language Models and the Practice of Poetry Translation

Summary

Machines have long been considered unable to translate poetry. Conventional MT favours statistically likely mappings, while poems depend on ambiguity, formal constraints such as metre and rhyme, and culturally loaded imagery. Resende and Hadley propose a practical, model-agnostic framework that shows how translators can use LLMs through established prompt engineering techniques at two stages. In pre-translation, strategies such as Chain-of-Thought prompting help the translator make cultural inference explicit, break the work into steps and define goals before writing any target text. In the translation stage, prompts aim to reproduce formal features such as rhyme schemes. The authors illustrate the framework qualitatively with ChatGPT 5.2 and Claude Sonnet 4.6 on poems with complex cultural and formal features, including a classical Japanese tanka. The models gave structured, substantive analyses and usable first drafts. One model also asked the user about translation priorities without being prompted. The authors argue that LLMs can extend human creativity in translation rather than replace it. They stress that results vary by model and language pair and that human oversight remains essential.

Key points

  • Models and evaluation: Examples were generated with ChatGPT 5.2 (thinking mode) and Claude Sonnet 4.6. The evaluation is qualitative and illustrative, not a quantitative benchmark.
  • Chain-of-Thought case study: A four-step CoT sequence was applied to poem 33 of the Ogura Hyakunin Isshu by Ki no Tomonori. The steps were cultural elicitation, interpretive expansion, constraint specification and constrained translation. They target the implied cherry-blossom image of 花 combined with 散る.
  • Depth of analysis: In the first step the model described the symbolism of 花 along five dimensions, including mujō and mono no aware. It also explained how 散る dramatises the loss of beauty rather than simply describing it.
  • Zero-shot rhyme reproduction: For a rhymed poem, a plain zero-shot prompt gave an English draft that partly kept the source rhyme scheme. The Portuguese version from Claude Sonnet 4.6 matched the scheme exactly.
  • Proactive behaviour and limits: Claude Sonnet 4.6 offered a more form-faithful rendering, asked about the user's goals and analysed where its rhyme scheme failed. The authors had not seen this behaviour in their earlier work (Resende and Hadley, 2024). They expect weaker performance for less-resourced target languages, given how much of the training data is in English.

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 · Machine translation

Cite

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

@inproceedings{resende2026extending,
  title = {{Extending Creativity: Large Language Models and the Practice of Poetry Translation}},
  author = {Resende, Natalia and Hadley, James},
  booktitle = {Proceedings of the 26th Annual Conference of the European Association for Machine Translation (Volume 1)},
  year = {2026},
  url = {https://aclanthology.org/2026.eamt-1.50/}
}

References

18 references, 3 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. Bassnett, Susan. 1998. Transplanting the seed: Poetry and translation. In Constructing Cultures, pages 57–75. Multilingual Matters, Bristol and Blue Ridge Summit. find
  2. Bellos, David. 2011. Is That a Fish in Your Ear? Translation and the Meaning of Everything. Penguin Books, London. find
  3. Brown, Tom, Benjamin Mann, Nick Ryder, Melanie Subbiah, Jared D. Kaplan, Prafulla Dhariwal, Arvind Neelakantan, Pranav Shyam, Girish Sastry, and Amanda Askell. 2020. Language models are few-shot learners. Advances in Neural Information Processing Systems, 33:1877–1901. find
  4. Clunies Ross, Margaret. 2010. The Cambridge Introduction to the Old Norse-Icelandic Saga. Cambridge University Press. find
  5. Fabb, Nigel. 2015. What is Poetry?: Language and Memory in the Poems of the World. Cambridge University Press, Cambridge. find
  6. Forcada, Mikel L. 2017. Making sense of neural machine translation. Translation Spaces, 6(2):291–309. find
  7. Hume, Nancy G., editor. 1995. Japanese Aesthetics and Culture: A Reader. State University of New York Press. Accessed: 2026-04-22. find
  8. Jackson, Samantha, Barend Beekhuizen, Zhao Zhao, Yi Cheng Zhao, and Rhonda N. McEwen. 2023. Llms and linguistic competency: An exploration of gpt-4 and a non-hegemonic english variety. Newhouse Impact Journal, 1(1):Article 8. find
  9. Karpinska, Marzena and Mohit Iyyer. 2023. Large language models effectively leverage document-level context for literary translation, but critical errors persist. In Koehn, Philipp, Barry Haddow, Tom Kocmi, and Christof Monz, editors, Proceedings of the Eighth Conference on Machine Translation, page… find our card
  10. Lefevere, André. 1975. Translating Poetry: Seven Strategies and a Blueprint, volume 3 of Approaches to Translation Studies. Van Gorcum, Assen. find
  11. Li, Haochen, Jonathan Leung, and Zhiqi Shen. 2024. Towards goal-oriented prompt engineering for large language models: A survey. find
  12. Newman, John B. 1958. The meanings of poetry. Today’s Speech, 6(3):28–30. find
  13. Peng, Keqin, Liang Ding, Qihuang Zhong, Li Shen, Xuebo Liu, Min Zhang, Yuanxin Ouyang, and Dacheng Tao. 2023. Towards making the most of ChatGPT for machine translation. In Bouamor, Houda, Juan Pino, and Kalika Bali, editors, Findings of the Association for Computational Linguistics: EMNLP 2023, pag… find our card
  14. Resende, Natália and James Hadley. 2024. The translator’s canvas: Using llms to enhance poetry translation. In Proceedings of the 16th Conference of the Association for Machine Translation in the Americas (Volume 1: Research Track). find
  15. Stageberg, Norman C. 1948. The aesthetic of the petrarchan sonnet. The Journal of Aesthetics and Art Criticism, 7(2):132–137. find
  16. Suojanen, Tytti, Kaisa Koskinen, and Tiina Tuominen. 2014. User-centered translation. Routledge. find
  17. Suzgun, Mirac and Adam Tauman Kalai. 2024. Meta-prompting: Enhancing language models with task-agnostic scaffolding. find
  18. Wei, Jason, Xuezhi Wang, Dale Schuurmans, Maarten Bosma, Brian Ichter, Fei Xia, Ed H. Chi, Quoc V. Le, and Denny Zhou. 2022. Chain of thought prompting elicits reasoning in large language models. In Oh, Alice H., Alekh Agarwal, Danielle Belgrave, and Kyunghyun Cho, editors, Advances in Neural Inform… find our card

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/resende-2026-extending-creativity-large-language-models/