Elamur Research⌕
☰
home / papers / 2026 / FL-MSCL: A Unified Figurative Language Detection Model Driven by Multi-Type Signals and Contrastive Learning

FL-MSCL: A Unified Figurative Language Detection Model Driven by Multi-Type Signals and Contrastive Learning

Paper · 2026 · Lu Shijia, Fumiyo Fukumoto, Huang Xiaoxi et al.

Summary

Figurative language detection is usually handled one figure at a time, as a binary choice between figurative and literal. That leaves a gap: metaphor, metonymy and simile are easy to confuse with one another. This short ACL 2026 paper treats them together as a single sentence-level task with four labels: literal, metaphor, metonymy and simile. The authors build a unified training corpus by merging existing English metaphor, metonymy and simile benchmarks and down-sampling the result into balanced classes. Their model, FL-MSCL, uses a DeBERTa encoder. For each input sentence, the prompt is extended with one retrieved example per category, chosen by similarity using Sentence-BERT and FAISS. Training combines cross-entropy with a supervised contrastive loss on the [CLS] representation, which separates the confusable categories more sharply. On the unified task the full model reaches 86.1 macro-F1. It also performs strongly on standard single-figure benchmarks, including a reported new best on MOH-X. The work suggests that explicitly modelling the boundaries between related figures helps both joint and category-specific detection, though all experiments are English-only.

Key points

  • Setup: DeBERTa-v3-base encoder. Retrieved exemplars act as "semantic anchors" in place of the static dictionary definitions used in earlier knowledge-injection work. The total loss is cross-entropy plus λ·SCL.
  • Data: VUA-18, VUA-20, VUA-Verb, MOH-X and TroFi (metaphor), ReLocaR (metonymy) and PLMs-Simile (simile). Binary evaluations follow the official splits.
  • Unified 4-way task: 86.1 F1 with precision 88.8 and recall 84.1. Contrastive learning alone mainly raises precision, while adding prompts restores recall.
  • Metaphor: 86.2 F1 on MOH-X (claimed new SOTA) and 88.9 F1 on VUA-Verb, reported as 14.8 points above WPDM. Without both prompts and contrastive loss, VUA-Verb drops to 81.9.
  • Other figures and errors: ReLocaR reaches 89.5 F1 versus 84.8 for PreWin, and simile detection reaches 99.9 F1. Remaining weaknesses are cautious metaphor recall on VUA-18, institutional metonymy such as "Washington said", and similes without explicit markers.

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

Also on the byline: Meiji University · University of Yamanashi.

 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.Idioms & multiword expressions · 217 works

Cite

Cite as Lu Shijia, Fumiyo Fukumoto, Huang Xiaoxi and Yoshimi Suzuki. 2026. FL-MSCL: A Unified Figurative Language Detection Model Driven by Multi-Type Signals and Contrastive Learning. Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers).

The paper itself, not this page. BibTeX from the library's record of the work. DOI 10.18653/v1/2026.acl-short.57

@inproceedings{shijia2026fl,
  title = {{FL-MSCL: A Unified Figurative Language Detection Model Driven by Multi-Type Signals and Contrastive Learning}},
  author = {Lu Shijia, Fumiyo Fukumoto, Huang Xiaoxi and Yoshimi Suzuki},
  booktitle = {Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers)},
  year = {2026},
  doi = {10.18653/v1/2026.acl-short.57},
  url = {https://doi.org/10.18653/v1/2026.acl-short.57}
}

References

20 references, 5 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. Saiteja Badathala, Ashish Goswami, and Radhika Mamidi. 2023. A match made in heaven: A multi-Tuhin Chakrabarty, Arkadiy Saakyan, Debanjan Ghosh, and Smaranda Muresan. 2022. FLUTE: Figurative language understanding through textual explanations. In Proceedings of the 2022 Conference on Empirical Metho…
  2. Minjin Choi, Sunkyung Lee, Eunseong Choi, Heesoo Park, Junhyuk Lee, Dongwon Lee, and Jongwuk Lee. 2021. MelBERT: Metaphor detection via contextualized late interaction using metaphorical identification theories. In Proceedings of the 2021 Conference of the North American Chapter of the Association f… find
  3. Erik-Lân Do Dinh, Steffen Eger, and Iryna Gurevych. 2018. Killing four birds with two stones: Multi-task learning for non-literal language detection. In Proceedings of the 27th International Conference on Computational Linguistics, Santa Fe, New Mexico, USA. Association for Computational Linguistics… find
  4. Mohamad Elzohbi and Richard Zhao. 2024. ContrastWSD: Enhancing metaphor detection with word sense disambiguation following the metaphor identification procedure. In Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING 202… find
  5. Fumiyo Fukumoto and Shou Asakawa. 2023. Knowledge injection with perturbation-based constrained attention network for word sense disambiguation. In Findings of the Association for Computational Linguistics: IJCNLP-AACL 2023 (Findings), pages 171– 177, Nusa Dua, Bali. Association for Computational Li… find
  6. Ge Gao, Eunsol Choi, Yejin Choi, and Luke Zettlemoyer. 2018. Neural metaphor detection in context. In Proceedings of the 2018 Conference on Empirical Methods in Natural Language Processing, pages 607–613, Brussels, Belgium. Association for Computational Linguistics.
  7. Milan Gritta, Mohammad Taher Pilehvar, Nut Limsopatham, and Nigel Collier. 2017. Vancouver welcomes you! minimalist location metonymy resolution. In Proceedings of the 55th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pages 1248–1259. find
  8. Qianyu He, Sijie Cheng, Zhixu Li, Rui Xie, and Yanghua Xiao. 2022. Can pre-trained language models interpret similes as smart as human? In Proceedings of the 60th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pages 7875–7887, Dublin, Ireland. Association fo… find
  9. Huang Heyan, Liu Xiao, and Liu Qian. 2023. Knowledge-enhanced graph encoding method for metaphor detection in text. Journal of Computer Research and Development, 60(1):140–152. find
  10. Kaidi Jia and Rongsheng Li. 2024. Metaphor detection with context enhancement and curriculum learning. In Proceedings of the 2024 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies (Volume 1: Long Papers), pages 2726–2737, Mexico Ci… find
  11. Duong Le, My Thai, and Thien Nguyen. 2020. Multi-task learning for metaphor detection with graph convolutional neural networks and word sense disambiguation. Proceedings of the AAAI Conference on Artificial Intelligence, 34(05):8139–8146. find
  12. Chee Wee (Ben) Leong, Beata Beigman Klebanov, Chris Hamill, Egon Stemle, Rutuja Ubale, and Xianyang Chen. 2020. A report on the 2020 VUA and TOEFL metaphor detection shared task. In Proceedings of the Second Workshop on Figurative Language Processing, pages 18–29, Online. Association for Computation…
  13. Saif Mohammad, Ekaterina Shutova, and Peter Turney. 2016. Metaphor as a medium for emotion: An empirical study. In Proceedings of the Fifth Joint Conference on Lexical and Computational Semantics, pages 23–33, Berlin, Germany. Association for Computational Linguistics.
  14. Sebastian Reimann and Tatjana Scheffler. 2025. Using large language models to perform MIPVU-inspired automatic metaphor detection. In Proceedings of the 2nd Workshop on Analogical Abstraction in Cognition, Perception, and Language (Analogy-Angle II), pages 10–21, Vienna, Austria. Association for Com… find
  15. Arkadiy Saakyan, Shreyas Kulkarni, Tuhin Chakrabarty, and Smaranda Muresan. 2025. Understanding figurative meaning through explainable visual entailment. In Proceedings of the 2025 Conference of the Nations of the Americas Chapter of the Association for Computational Linguistics: Human Language Tech…
  16. Elisa Sanchez-Bayona and Rodrigo Agerri. 2025. Metaphor and large language models: When surface features matter more than deep understanding. In Findings of the Association for Computational Linguistics: ACL 2025, pages 17462–17477, Vienna, Austria. Association for Computational Linguistics. find
  17. Yuan Tian, Ruike Zhang, Nan Xu, and Wenji Mao. 2024. Bridging word-pair and token-level metaphor detection with explainable domain mining. In Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), Bangkok, Thailand. Association for Computatio… find
  18. Shuhang Xu and Fangwei Zhong. 2025. CoMet: Metaphor-driven covert communication for multi-agent language games. In Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pages 7892– 7917, Vienna, Austria. Association for Computational Linguis… find
  19. Wenhao Yu, Chenguang Zhu, Ziqing Yang, and Zengchang Liu. 2021. Dict-BERT: Enhancing language representation pre-training with dictionary. In Proceedings of the 2021 Conference on Empirical Methods in Natural Language Processing, pages 1428–1439, Online and Punta Cana, Dominican Republic. Associatio… find
  20. Yucheng Li, Shunyu Wang, Chenghua Lin, and Guerin Frank. 2023. Metaphor detection via explicit basic meanings modelling. In Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers), Toronto, Canada. Association for Computational Linguistics. find

Updated

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/shijia-2026-fl-mscl-unified-figurative-language/