Elamur Research⌕
☰
home / papers / 2026 / Multimodal OXYmorons: A Comprehensive Introduction and Computational Analysis Using a Dataset of Oxymoronic Memes in Italian and Spanish

Multimodal OXYmorons: A Comprehensive Introduction and Computational Analysis Using a Dataset of Oxymoronic Memes in Italian and Spanish

Paper · 2026 · Eliana Di Palma, Giulia Rizzi, Francesca Masini et al.

Summary

Oxymorons have received far less research attention than metaphors or idioms, especially when the contradiction arises across modalities rather than within a sentence. The authors propose the notion of a multimodal oxymoron, in which one pole of an antonymic contrast appears in the image and the other in the caption. An example is a cheerful caption over a visibly miserable face. They separate this figure from paradox, antithesis (which relies on parallel structure), irony and antonymy. They also release MOXY, a curated collection of Italian and Spanish memes that contain oxymorons. Antonym pairs guided both its construction and the later recognition step. The paper tests whether vision-language models and an LLM can identify these oxymorons. It also designs a pipeline that generates new ones: an image is created for one concept, and a caption is then prompted for the opposing concept. The provided excerpt omits the quantitative results, so performance figures cannot be reported here. The work matters because it gives figurative-language and multimodal AI research a theoretical framework, a benchmark resource and a generation method.

Key points

  • Research questions: the paper addresses four, covering
    • what defines multimodal oxymorons,
    • how visuals change the perceived contrasts,
    • how well current models detect them,
    • whether a dataset can be built automatically.
  • Resource: MOXY (Multimodal OXYmoron) is a multilingual meme dataset in Italian and Spanish. It is annotated with attention to how readers perceive and interpret the contrast.
  • Models evaluated:
    • CLIP-ViT-B-32-multilingual-v1 and mBLIP-mt0-xl, both fine-tuned with 10-fold cross-validation on MOXY for 5 epochs with batch size 4;
    • ChatGPT-4o-mini.
  • Generation pipeline: the prompt supplies an image concept (CI) and an opposing text concept (CT). The caption must state CT explicitly, refer to the same subject, and present both concepts as simultaneous. Alternative, less constrained prompts were also compared.
  • Prior textual resources: earlier Italian work produced 377 oxymorons from 17 antonym pairs (La Pietra and Masini 2020) and 207 oxymorons derived from De Mauro adjectives (Bolognesi et al. 2024).

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 TurinTurin · Italy33 works · 6,637 citations🎓Universitat Politècnica de ValènciaValencia · Spain34 works · 3,575 citations🎓University of BolognaBologna · Italy19 works · 2,067 citations🎓University of Milano-BicoccaMilan · Italy15 works · 4,946 citations

 Fields, topics and concepts

📚 Shelves 4The field of science this work belongs to (OpenAlex fields).Arts and Humanities · 17,450 worksLanguage and Linguistics · 17,296 worksPsychology · 1,649 worksSocial Psychology · 330 works

🧭 Topics 1Research subjects and languages the work is about.Vision & language · 811 works

Cite

Cite as Eliana Di Palma, Giulia Rizzi, Francesca Masini, Paolo Rosso and Elisabetta Fersini. 2026. Multimodal OXYmorons: A Comprehensive Introduction and Computational Analysis Using a Dataset of Oxymoronic Memes in Italian and Spanish. Computational Linguistics.

The paper itself, not this page. BibTeX from the library's record of the work. DOI 10.1162/coli.a.586

@article{palma2026multimodal,
  title = {{Multimodal OXYmorons: A Comprehensive Introduction and Computational Analysis Using a Dataset of Oxymoronic Memes in Italian and Spanish}},
  author = {Eliana {Di Palma}, Giulia Rizzi, Francesca Masini, Paolo Rosso and Elisabetta Fersini},
  journal = {Computational Linguistics},
  year = {2026},
  doi = {10.1162/coli.a.586},
  url = {https://doi.org/10.1162/coli.a.586}
}

References

14 references, 4 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. Athanasiadou, Angeliki. 2024. On the margins of figurative thought and language. Lingua, 299:103655. https://doi.org/10.1016/j.lingua.2023.103655 Barthes, R. and S. Heath. 1977. — [doi:10.1016/j.lingua.2023.103655](https://doi.org/10.1016/j.lingua.2023.103655) ↗
  2. Image, Music, Text. Fontana Press. Benammar, Silya. 2023. Multimodal figuration in internet memes. metaphorik.de, 34. find
  3. Bolognesi, Marianna M., Claudia Roberta Combei, Marta La Pietra, and Francesca Masini. 2024. What makes an awfully good oxymoron? Language and Cognition, 16(1):242–262. https://doi.org/10.1017/langcog.2023.68 Cerini, Ludovica, Eliana Di Palma, and Alessandro Lenci. 2022. From speed to car and back:… — [doi:10.1017/langcog.2023.68](https://doi.org/10.1017/langcog.2023.68) ↗
  4. Choi, Wan Chong and Chi In Chang. 2025. A survey of techniques, key components, strategies, challenges, and student perspectives on prompt engineering for large language models (LLMs) in education. https://doi.org/10.20944/preprints202503.1808.v1 Dancygier, Barbara and Eve Sweetser. 2014. ↗
  5. Figurative Language, Cambridge University Press. De Mauro, Tullio. 2016. Il nuovo vocabolario di base della lingua italiana. Internazionale. https://www.internazionale.it/opinione/tullio-de-mauro/2016/12/23/il-nuovo-vocabolario-di-base -della-lingua-italiana Dynel, Marta. 2016. Two layers of overt… ↗
  6. Forceville, Charles. 2006. Non-verbal and Multimodal Metaphor in a Cognitivist Framework: Agendas for Research. De Gruyter Mouton. https://doi.org/10.1515/9783110197761.5.379 Forceville, Charles and Eduardo Urios-Aparisi. 2009. Multimodal Metaphor, volume 11, Walter de Gruyter. https://doi.org/10.… ↗
  7. Kress, Gunther and Theo Van Leeuwen. 2020. Reading Images: The Grammar of Visual Design. Routledge. https://doi.org/10.4324/9781003099857 La Pietra, Marta and Francesca Masini. 2020. Oxymorons: A preliminary corpus investigation. In Proceedings of the Second Workshop on Figurative Language Processi… ↗
  8. Li, Junnan, Dongxu Li, Caiming Xiong, and Steven Hoi. 2022. BLIP: Bootstrapping language-image pre-training for unified vision-language understanding and generation. arXiv preprint arXiv:2201.12086. https://doi.org/10.48550/ARXIV.2201.12086 Lou, Adrian. 2017. Multimodal simile: The “when” meme in s… ↗
  9. Nadeau, Claude and Yoshua Bengio. 2003. Inference for the generalization error. Machine Learning, 52(3):239–281. https://doi.org/10.1023/a:1024068626366 Radford, A., J. W. Kim, C. Hallacy, A. Ramesh, G. Goh, S. Agarwal, G. Sastry, A. Askell, P. Mishkin, J. Clark, G. Krueger, and I. Sutskever. 2021.… ↗
  10. Reimers, Nils and Iryna Gurevych. 2019. Sentence-BERT: Sentence embeddings using Siamese BERT-networks. In Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing. https://doi.org/10.18653/v1/D19-1410 Richard, I. A. 1936. The Philosophy of Rhetoric. ↗
  11. Oxford University Press. Rizzi, Giulia, Francesca Gasparini, Aurora Saibene, Paolo Rosso, and Elisabetta Fersini. 2023. Recognizing misogynous memes: Biased models and tricky archetypes. Information Processing & Management, 60(5):103474. https://doi.org/10.1016/j.ipm.2023.103474 Ruiz, Javier Herrer… ↗
  12. Scott, Kate. 2021. Memes as multimodal metaphors: A relevance theory analysis. Pragmatics & Cognition, 28(2):277–298. https://doi.org/10.1075/pc.21010.sco — [doi:10.1075/pc.21010.sco](https://doi.org/10.1075/pc.21010.sco) ↗
  13. Shen, Yeshayahu. 1987. On the structure and understanding of poetic oxymoron. Poetics Today, 8(1):105–122. https://doi.org/10.2307/1773004 Tomás, David, Reynier Ortega-Bueno, Guobiao Zhang, Paolo Rosso, and Rossano Schifanella. 2022. Transformer-based models for multimodal irony detection. Journal… ↗
  14. Zhong, Zenan, Suijun Wen, and Shukun Chen. 2023. Research trends in multimodal metaphor: A bibliometric analysis. Frontiers in Psychology, 14. https://doi.org/10.3389/fpsyg.2023.1144725, PubMed: 37138971 — [doi:10.3389/fpsyg.2023.1144725](https://doi.org/10.3389/fpsyg.2023.1144725) ↗

Updated

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/palma-2026-multimodal-oxymorons-comprehensive-introduction-computational/