Elamur Research⌕
☰
home / papers / 2026 / Contrastive Learning with Narrative Twins for Modeling Story Salience

Contrastive Learning with Narrative Twins for Modeling Story Salience

Paper · 2026 · Igor Sterner, Alex Lascarides and Frank Keller

Summary

Narrative understanding depends on telling which events drive a plot forward, yet there has been little systematic work on how to compute this "salience." The authors pose two questions: which story representations suit the task, and which narratology-inspired test best picks out salient sentences. They train a bi-directional encoder with an InfoNCE contrastive objective. Each anchor story is pulled toward a narrative twin, meaning a version with the same plot but different wording, and pushed away from a distractor, meaning a text that looks similar on the surface but tells a different story. On top of the resulting embeddings they compare four salience tests: deletion, shifting, disruption and summarization. Evaluation covers ROCStories (with newly collected human salience labels) and longer Wikipedia plot summaries. Contrastive embeddings beat a masked-LM baseline, and the summarization test, which measures how similar a sentence is to the whole story, works best. Twins can also be generated with dropout, so the method can run fully self-supervised on unannotated narratives.

Key points

  • Narrative twins are taken from the Tell me again! dataset (Hatzel and Biemann), which contains Wikipedia plot summaries from several language editions machine-translated into English. Dropout twins, made by encoding one story twice with different random dropout, perform about as well under the summarization test.
  • Distractors are produced by prompting LLMs for short stories. For long narratives, other parts of the same story serve as "in-story" negatives. Ablations that remove distractors or use entity substitution instead both lower performance.
  • On both corpora, summarization is significantly better than the other three tests (p < 0.05). Deletion and shifting come next, and disruption is clearly the weakest.
  • On Wikipedia summaries, which average about 34 sentences, narrative-twin training outperforms masked-LM training for every operation (p < 0.01). Dropout twins with in-story negatives tie for the best result.
  • Code and data, including the ROCStories salience annotations, are released at github.com/igorsterner/NarrativeTwins.

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 EdinburghEdinburgh · United Kingdom381 works · 80,747 citations

 Fields, topics and concepts

📚 Shelves 6The field of science this work belongs to (OpenAlex fields).Computer Science · 18,614 worksArtificial Intelligence · 17,965 worksArts and Humanities · 17,450 worksLanguage and Linguistics · 17,296 worksComputer Vision and Pattern Recognition · 3,486 worksLiterature and Literary Theory · 541 works

Cite

Cite as Igor Sterner, Alex Lascarides and Frank Keller. 2026. Contrastive Learning with Narrative Twins for Modeling Story Salience. Proceedings of the 19th Conference of the European Chapter of the Association for Computational Linguistics (Volume 1: Long Papers).

The paper itself, not this page. BibTeX from the library's record of the work. DOI 10.18653/v1/2026.eacl-long.71

@inproceedings{sterner2026contrastive,
  title = {{Contrastive Learning with Narrative Twins for Modeling Story Salience}},
  author = {Igor Sterner, Alex Lascarides and Frank Keller},
  booktitle = {Proceedings of the 19th Conference of the European Chapter of the Association for Computational Linguistics (Volume 1: Long Papers)},
  year = {2026},
  doi = {10.18653/v1/2026.eacl-long.71},
  url = {https://doi.org/10.18653/v1/2026.eacl-long.71}
}

References

45 references, 12 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. Jimmy Lei Ba, Jamie Ryan Kiros, and Geoffrey E. Hinton. 2016. Layer normalization. Preprint, arXiv:1607.06450v1.
  2. Roland Barthes. 1966. Introduction à l’analyse structurale des récits. Communications, 8(1):1–27. Translated in: An Introduction to the Structural Analysis of Narrative, New Literary History, 1975. find
  3. Nathanael Chambers and Dan Jurafsky. 2008. Unsupervised learning of narrative event chains. In Proceedings of the 46th Annual Meeting of the Association for Computational Linguistics, pages 789–797, Columbus, Ohio. Association for Computational Linguistics. find
  4. Neil Cohn. 2013. Visual narrative structure. Cognitive science, 37(3):413–452. find
  5. Max Conti, Manuel Faysse, Gautier Viaud, Antoine Bosselut, Celine Hudelot, and Pierre Colombo. 2025. Context is gold to find the gold passage: Evaluating and training contextual document embeddings. In Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing, pages 2260… find
  6. 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:…
  7. Jesse Dunietz and Daniel Gillick. 2014. A new entity salience task with millions of training examples. In Proceedings of the 14th Conference of the European Chapter of the Association for Computational Linguistics: Volume 2 (Short Papers), pages 205–209, Gothenburg, Sweden. Association for Computati…
  8. Tianyu Gao, Xingcheng Yao, and Danqi Chen. 2021. SimCSE: Simple contrastive learning of sentence embeddings. In Proceedings of the 2021 Conference on Empirical Methods in Natural Language Processing, pages 6894–6910, Online and Punta Cana, Dominican Republic. Association for Computational Linguistic…
  9. Michael Gutmann and Aapo Hyvärinen. 2010. Noise-contrastive estimation: A new estimation principle for unnormalized statistical models. In Proceedings of the thirteenth international conference on artificial intelligence and statistics, pages 297–304. JMLR Workshop and Conference Proceedings. find
  10. Michael Günther, Isabelle Mohr, Daniel James Williams, Bo Wang, and Han Xiao. 2025. Late chunking: Contextual chunk embeddings using long-context embedding models. Preprint, arXiv:2409.04701v3. — [arXiv:2409.04701](https://arxiv.org/abs/2409.04701) find
  11. John Hale. 2001. A probabilistic Earley parser as a psycholinguistic model. In Second Meeting of the North American Chapter of the Association for Computational Linguistics, Pittsburgh, Pennsylvania, USA. Association for Computational Linguistics.
  12. Hans Ole Hatzel and Chris Biemann. 2024a. Story embeddings — narrative-focused representations of fictional stories. In Proceedings of the 2024 Conference on Empirical Methods in Natural Language Processing, pages 5931–5943, Miami, Florida, USA. Association for Computational Linguistics. find
  13. Hans Ole Hatzel and Chris Biemann. 2024b. Tell me again! a large-scale dataset of multiple summaries for the same story. In Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation, pages 15732– 15741, Torino, Italia. ELRA and ICCL. find
  14. William Labov. 1972. The transformation of experience in narrative syntax. In Language in the Inner City, pages 354–396. University of Pennsylvania Press, Philadelphia. find
  15. William Labov. 1997. Some further steps in narrative analysis. Journal of Narrative and Life History, 7:395– 415. find
  16. William Labov and Joshua Waletzky. 1967. Narrative analysis: Oral versions of personal experience. In Essays on the Verbal and Visual Arts, pages 12–44. University of Washington Press, Seattle, WA. find
  17. Faisal Ladhak, Bryan Li, Yaser Al-Onaizan, and Kathleen McKeown. 2020. Exploring content selection in summarization of novel chapters. In Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics, pages 5043–5054, Online. Association for Computational Linguistics. find
  18. Stieg Larsson. 2005. Män som hatar kvinnor [The Girl with the Dragon Tattoo]. Norstedts, Stockholm. find
  19. Chin-Yew Lin and Eduard Hovy. 2000. The automated acquisition of topic signatures for text summarization. In Proceedings of the 2015 International Conference on Computational Linguistics: Volume 1.
  20. Fangyu Liu, Ivan Vulić, Anna Korhonen, and Nigel Collier. 2021. Fast, effective, and self-supervised: Transforming masked language models into universal lexical and sentence encoders. In Proceedings of the 2021 Conference on Empirical Methods in Natural Language Processing, pages 1442–1459, Online…
  21. Zhengzhong Liu, Chenyan Xiong, Teruko Mitamura, and Eduard Hovy. 2018. Automatic event salience identification. In Proceedings of the 2018 Conference on Empirical Methods in Natural Language Processing, pages 1226–1236, Brussels, Belgium. Association for Computational Linguistics. find
  22. Ilya Loshchilov and Frank Hutter. 2019. Decoupled weight decay regularization. In Proceedings of the 7th International Conference on Learning Representations. find
  23. Hans Peter Luhn. 1958. The automatic creation of literature abstracts. IBM Journal of Research and Development, 2(2):159–165. find
  24. William C Mann and Sandra A Thompson. 1988. Rhetorical structure theory: Toward a functional theory of text organization. Text-interdisciplinary Journal for the Study of Discourse, 8(3):243–281.
  25. Rada Mihalcea and Hakan Ceylan. 2007. Explorations in automatic book summarization. In Proceedings of the 2007 Joint Conference on Empirical Methods in Natural Language Processing and Computational Natural Language Learning (EMNLP-CoNLL), pages 380–389, Prague, Czech Republic. Association for Comput… find
  26. Rada Mihalcea and Paul Tarau. 2004. TextRank: Bringing order into text. In Proceedings of the 2004 Conference on Empirical Methods in Natural Language Processing, pages 404–411, Barcelona, Spain. Association for Computational Linguistics. find
  27. Nasrin Mostafazadeh, Nathanael Chambers, Xiaodong He, Devi Parikh, Dhruv Batra, Lucy Vanderwende, Pushmeet Kohli, and James Allen. 2016. A corpus and cloze evaluation for deeper understanding of commonsense stories. In Proceedings of the 2016 Conference of the North American Chapter of the Associati… find
  28. Feiteng Mu and Wenjie Li. 2024. Generating contrastive narratives using the brownian bridge process for narrative coherence learning. In Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics: Volume 1 (Long Papers), pages 6538–6555, Bangkok, Thailand. Association fo… find
  29. Takaki Otake, Sho Yokoi, Naoya Inoue, Ryo Takahashi, Tatsuki Kuribayashi, and Kentaro Inui. 2020. Modeling event salience in narratives via Barthes’ cardinal functions. In Proceedings of the 28th International Conference on Computational Linguistics, pages 1784–1794, Barcelona, Spain (Online). Inter… find
  30. Jessica Ouyang, Serina Chang, and Kathy McKeown. 2017. Crowd-sourced iterative annotation for narrative summarization corpora. In Proceedings of the 15th Conference of the European Chapter of the Association for Computational Linguistics: Volume 2, Short Papers, pages 46–51, Valencia, Spain. Associa… find
  31. Jessica Ouyang and Kathleen McKeown. 2015. Modeling reportable events as turning points in narrative. In Proceedings of the 2015 Conference on Empirical Methods in Natural Language Processing, pages 2149–2158, Lisbon, Portugal. Association for Computational Linguistics. find
  32. Pinelopi Papalampidi, Frank Keller, and Mirella Lapata. 2019. Movie plot analysis via turning point identification. In Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natural Language Processing (EMNLP-IJCNLP), page… find
  33. Pinelopi Papalampidi, Frank Keller, and Mirella Lapata. 2021. Movie summarization via sparse graph construction. In The Thirty-Fifth AAAI Conference on Artificial Intelligence, pages 5784–5792. AAAI Press. find
  34. Gerald Prince. 2003. A Dictionary of Narratology. University of Nebraska Press, Lincoln and London. find
  35. Rohit Saxena and Frank Keller. 2024. Select and summarize: Scene saliency for movie script summarization. In Findings of the Association for Computational Linguistics: NAACL 2024, pages 3439–3455, Mexico City, Mexico. Association for Computational Linguistics. find
  36. Karen Spärck Jones. 1972. A statistical interpretation of term specificity and its application in retrieval. Journal of Documentation, 28(1):11–21. find
  37. Tzvetan Todorov. 1968. La grammaire du récit. Langages, 12:94–102. Translated in: The Poetics of Prose, Chapter 8, Cornell University Press, 1977. find
  38. Iulia Turc, Ming-Wei Chang, Kenton Lee, and Kristina Toutanova. 2019. Well-read students learn better: On the importance of pre-training compact models. Preprint, arXiv:1908.08962v2.
  39. Aaron van den Oord, Yazhe Li, and Oriol Vinyals. 2019. Representation learning with contrastive predictive coding. Preprint, arXiv:1807.03748v2.
  40. Benjamin Warner, Antoine Chaffin, Benjamin Clavié, Orion Weller, Oskar Hallström, Said Taghadouini, Alexis Gallagher, Raja Biswas, Faisal Ladhak, Tom Aarsen, Griffin Thomas Adams, Jeremy Howard, and Iacopo Poli. 2025. Smarter, better, faster, longer: A modern bidirectional encoder for fast, memory…
  41. Orion Weller, Kathryn Ricci, Marc Marone, Antoine Chaffin, Dawn Lawrie, and Benjamin Van Durme. 2025. Seq vs seq: An open suite of paired encoders and decoders. Preprint, arXiv:2507.11412v1. — [arXiv:2507.11412](https://arxiv.org/abs/2507.11412) find
  42. David Wilmot and Frank Keller. 2021. Memory and knowledge augmented language models for inferring salience in long-form stories. In Proceedings of the 2021 Conference on Empirical Methods in Natural Language Processing, pages 851–865, Online and Punta Cana, Dominican Republic. Association for Comput… find
  43. Lanling Xu, Jianxun Lian, Wayne Xin Zhao, Ming Gong, Linjun Shou, Daxin Jiang, Xing Xie, and Ji-Rong Wen. 2022. Negative sampling for contrastive representation learning: A review. Preprint, arXiv:2206.00212v1. — [arXiv:2206.00212](https://arxiv.org/abs/2206.00212) find
  44. Xiyang Zhang, Muhao Chen, and Jonathan May. 2021. Salience-aware event chain modeling for narrative understanding. In Proceedings of the 2021 Conference on Empirical Methods in Natural Language Processing, pages 1418–1428, Online and Punta Cana, Dominican Republic. Association for Computational Ling… find
  45. Ming Zhong, Pengfei Liu, Yiran Chen, Danqing Wang, Xipeng Qiu, and Xuanjing Huang. 2020. Extractive summarization as text matching. In Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics, pages 6197–6208, Online. Association for Computational Linguistics.

Updated

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/sterner-2026-contrastive-learning-narrative-twins-modeling/