---
title: "Metaphor in Textual Entailment — paper by Agerri (2008)"
description: "Agerri proposes using the Recognizing Textual Entailment (RTE) framework as the evaluation setting… 11 references, cited by 15 papers here."
authors: ["Rodrigo Agerri"]
year: 2008
venue: "Coling 2008: Companion volume: Posters"
source: "https://aclanthology.org/C08-2001/"
licence: "CC BY-NC-SA 3.0 (ACL Anthology, before 2016)"
licence_url: "https://creativecommons.org/licenses/by-nc-sa/3.0/"
summary_licence: "Elamur Cosmic License"
canonical: "https://research.elamur.ai/free/en/papers/agerri-2008-metaphor-textual-entailment/"
---
[Elamur Research](https://md.research.elamur.ai/free/en/) · [this page in HTML](https://research.elamur.ai/free/en/papers/agerri-2008-metaphor-textual-entailment/)

[home](https://md.research.elamur.ai/free/en/) / [papers](https://md.research.elamur.ai/free/en/papers/) / [2008](https://md.research.elamur.ai/free/en/papers/2008/page/2/) / Metaphor in Textual Entailment

# Metaphor in Textual Entailment

**Authors:** [Rodrigo Agerri](https://md.research.elamur.ai/free/en/authors/rodrigo-agerri/) · **Year:** 2008 · **Venue:** [Coling 2008: Companion volume: Posters](https://md.research.elamur.ai/free/en/venues/coling/) ([COLING 2008](https://md.research.elamur.ai/free/en/venues/coling/2008/)) · **Publisher:** [International Committee on Computational Linguistics (ICCL)](https://md.research.elamur.ai/free/en/publishers/international-committee-on-computational-linguistics-iccl/) · **Licence:** [CC BY-NC-SA 3.0 (ACL Anthology, before 2016)](https://creativecommons.org/licenses/by-nc-sa/3.0/)

**Links:** [Publisher's page](https://aclanthology.org/C08-2001/) · [PDF](https://aclanthology.org/C08-2001.pdf) · [OpenAlex](https://openalex.org/W22280556)

**Citations:** 17 citations (OpenAlex, 2026-10-14) · [15](https://md.research.elamur.ai/free/en/papers/agerri-2008-metaphor-textual-entailment/cited-by/) in our library, of the 149,439 papers whose bibliographies we read

## Summary
Computational work on metaphor has mostly produced standalone prototypes tested on a handful of hand-picked examples. There are no shared corpora or common evaluation exercises, so metaphor has stayed outside mainstream NLP. Agerri proposes using the Recognizing Textual Entailment (RTE) framework as the evaluation setting for metaphor interpretation.

The task changes from "give the correct reading of a metaphor" to "decide whether a hypothesis H follows from a text T containing a metaphor". This is easier, because H supplies context that narrows the possible readings.

To test whether metaphor actually hurts entailment systems, the author collected RTE-1 and RTE-2 pairs where resolving a metaphor matters for the judgment. He then compared participating systems' accuracy on those pairs with their official overall accuracy. On these pairs accuracy dropped clearly, and a Fisher test found the gap significant for most runs. The metaphors found were mostly conventional ones, which shallow methods partly handle. Open-ended metaphors probably need deeper reasoning, for example along the lines of ATT-Meta. Building RTE-style datasets with metaphors would give metaphor research a large-scale, theory-neutral benchmark.

## Key points
- The author identified 10 metaphor-relevant pairs in RTE-1 and 9 in RTE-2. Examples include "gastronomic capital of France", "incubated in the mind" and "stocks rallied".
- RTE-1 analysis covered 7 available runs, including 4 of the top 5 systems. Overall accuracies were about 0.55–0.59, while metaphor-pair accuracy ranged from 0.1 to 0.6.
- RTE-2 analysis covered 10 runs: 8 top-scoring systems plus 2 weaker ones. Overall accuracy was about 0.53–0.75, while metaphor-pair accuracy ranged from about 0.11 to 0.56.
- A Fisher's test of independence found the difference significant at the 0.05 level for 5 of the 7 RTE-1 runs and 7 of the 10 RTE-2 runs.
- All but one RTE-1 run wrongly judged "Lyon is the capital of France" as entailed by the "gastronomic capital" text. The author notes it is unclear whether metaphor or noun-modifier handling caused this.

## Interesting ideas
- Turn open-ended interpretation into a binary judgment: instead of asking for "the" meaning of a figurative phrase, ask whether one specific consequence follows, so the hypothesis itself disambiguates.
- Minimal hypothesis perturbation as a probe: adding one word ("accidentally") to H flips the judgment, because it clashes with the connotation of careful nurturing carried by "incubate".
- Metaphorical views (e.g., MIND AS PHYSICAL SPACE) model metaphor as a mapping that transfers properties and relations from the domain the figure is borrowed from onto the domain it describes.
- Slice evaluation by phenomenon: compare accuracy on pairs that hinge on a phenomenon against overall accuracy, with a significance test, to expose blind spots that aggregate scores hide.
- Separate conventional from open-ended metaphor: lexicalized metaphors can be handled shallowly, but novel ones need source-domain reasoning.

## For Mova

How these ideas could serve Mova, our small open language model — its files, stages and tests:

- **I1** (answer): Readers could test figurative sentences in fables against candidate consequences rather than committing early to a single literal reading.
- **I2** (learn): Build negative controls by perturbing a hypothesis with one modifier that contradicts a metaphor's carried connotation, and check that the system's verdict flips.
- **I3** (world): Store a small set of source→target mappings in the compiled knowledge, so that "incubated in the mind" can produce an explainable state like "developed slowly, carefully" in the text's world.
- **I4** (learn): Tag evaluation items by phenomenon (metaphor, noun modifiers, negation) and report per-slice accuracy with Fisher tests so regressions become visible.
- **I5** (frames): Map conventional metaphoric verb senses directly to their literal frames, and route only unrecognized figurative uses to the deeper source-to-target reasoning path.

*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 Birmingham](https://md.research.elamur.ai/free/en/universities/university-of-birmingham/) — University, Birmingham, United Kingdom · 25 works · 62,564 citations

📚 Shelves (6): [🖥️ Computer Science](https://md.research.elamur.ai/free/en/shelves/computer-science/) · 16,667 works · [🤖 Artificial Intelligence](https://md.research.elamur.ai/free/en/shelves/computer-science/artificial-intelligence/) · 16,194 works · [🎭 Arts and Humanities](https://md.research.elamur.ai/free/en/shelves/arts-and-humanities/) · 16,039 works · [🗣️ Language and Linguistics](https://md.research.elamur.ai/free/en/shelves/arts-and-humanities/language-and-linguistics/) · 15,891 works · [💭 Psychology](https://md.research.elamur.ai/free/en/shelves/psychology/) · 1,552 works · [💭 Experimental and Cognitive Psychology](https://md.research.elamur.ai/free/en/shelves/psychology/experimental-and-cognitive-psychology/) · 870 works

🧭 Topics (2): [🧮 Reasoning](https://md.research.elamur.ai/free/en/topics/reasoning/) · 1,426 works · [🧷 Idioms & multiword expressions](https://md.research.elamur.ai/free/en/topics/multiword-expressions/) · 217 works

## Cite

Cite as: Rodrigo Agerri. 2008. Metaphor in Textual Entailment. Coling 2008: Companion volume: Posters.

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

```bibtex
@inproceedings{agerri2008metaphor,
  title = {{Metaphor in Textual Entailment}},
  author = {Rodrigo Agerri},
  booktitle = {Coling 2008: Companion volume: Posters},
  year = {2008},
  url = {https://aclanthology.org/C08-2001/}
}
```

- [← previous Anaphora Resolution Exercise: an Overview](https://md.research.elamur.ai/free/en/papers/orasan-2008-anaphora-resolution-exercise-overview/)
- [next → Arabic WordNet: Semi-automatic Extensions using Bayesian Inference](https://md.research.elamur.ai/free/en/papers/rodriguez-2008-arabic-wordnet-semi-automatic-extensions/)
- [↑ back to papers of 2008](https://md.research.elamur.ai/free/en/papers/2008/page/2/)

## References

11 references, 1 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

1. Agerri, R., J.A. Barnden, M.G. Lee, and A.M. Wallington. 2007. Metaphor, inference and domain independent mappings. In Proceedings of Research Advances in Natural Language Processing (RANLP 2007), pages 17–24, Borovets, Bulgaria. [find](https://scholar.google.com/scholar?q=Agerri%2C+R%2E%2C+J%2EA%2E+Barnden%2C+M%2EG%2E+Lee%2C+and+A%2EM%2E+Wallington%2E+2007%2E+Metaphor%2C+inference+and+domain+independent+mappings%2E+In+Proceedings+of+Research+Advances+in+Natural+Language+Processing+%28RANLP+2007%29%2C+pag)
2. Barnden, J., S. Glasbey, M. Lee, and A. Wallington. 2003. Domain-transcending mappings in a system for metaphorical reasoning. In Companion Proceedings of the 10th Conference on the European Chapter of the Association for Computational Linguistics (EACL-03), pages 57–61. [find](https://scholar.google.com/scholar?q=Barnden%2C+J%2E%2C+S%2E+Glasbey%2C+M%2E+Lee%2C+and+A%2E+Wallington%2E+2003%2E+Domain%2Dtranscending+mappings+in+a+system+for+metaphorical+reasoning%2E+In+Companion+Proceedings+of+the+10th+Conference+on+the+European+Chapter+o)
3. Bos, J. and K. Markert. 2006. Recognizing textual entailment with robust logical inference. In Quiñonero-Candela, J., I. Dagan, B. Magnini, and F. d’Alché Buc, editors, MLCW 2005, volume 3944 of LNAI, pages 404–426. Springer-Verlag. [find](https://scholar.google.com/scholar?q=Bos%2C+J%2E+and+K%2E+Markert%2E+2006%2E+Recognizing+textual+entailment+with+robust+logical+inference%2E+In+Quin%CC%83onero%2DCandela%2C+J%2E%2C+I%2E+Dagan%2C+B%2E+Magnini%2C+and+F%2E+d%E2%80%99Alche%CC%81+Buc%2C+editors%2C+MLCW+2005%2C+volume+3944+of+LNA)
4. Bos, J. 2008. Lets not argue about semantics. In Proceedings of the Sixth International Language Resources and Evaluation (LREC’08), Marrakech, Morocco. [find](https://scholar.google.com/scholar?q=Bos%2C+J%2E+2008%2E+Lets+not+argue+about+semantics%2E+In+Proceedings+of+the+Sixth+International+Language+Resources+and+Evaluation+%28LREC%E2%80%9908%29%2C+Marrakech%2C+Morocco%2E)
5. Dagan, I., O. Glickman, and B. Magnini. 2007. The PASCAL Recognising Textual Entailment challenge. In Quiñonero-Candela, J., I. Dagan, B. Magnini, and F. d’Alché Buc, editors, MLCW 2005, volume 3944 of LNAI, pages 177–190. Springer-Verlag. [find](https://scholar.google.com/scholar?q=Dagan%2C+I%2E%2C+O%2E+Glickman%2C+and+B%2E+Magnini%2E+2007%2E+The+PASCAL+Recognising+Textual+Entailment+challenge%2E+In+Quin%CC%83onero%2DCandela%2C+J%2E%2C+I%2E+Dagan%2C+B%2E+Magnini%2C+and+F%2E+d%E2%80%99Alche%CC%81+Buc%2C+editors%2C+MLCW+2005%2C+volume+3944)
6. Falkenhainer, B., K.D. Forbus, and D. Gentner. 1989. The structure-mapping engine: algorithm and examples. Artificial Intelligence, 41(1):1–63. [↗](https://doi.org/10.1016/0004-3702(89)90077-5)
7. Fass, D. 1997. Processing metaphor and metonymy. Ablex, Greenwich, Connecticut. [find](https://scholar.google.com/scholar?q=Fass%2C+D%2E+1997%2E+Processing+metaphor+and+metonymy%2E+Ablex%2C+Greenwich%2C+Connecticut%2E)
8. Hobbs, J.R. 1992. Metaphor and abduction. In Ortony, A., J. Slack, and O. Stock, editors, Communication from an Artificial Intelligence Perspective, pages 35– 58. Springer-Verlag, Berlin. [↗](https://doi.org/10.1007/978-3-642-58146-5_3)
9. Lakoff, G. 2004. Conceptual metaphor home page. [http://cogsci.berkeley.edu/lakoff/MetaphorHome.html](http://cogsci.berkeley.edu/lakoff/MetaphorHome.html). [↗](http://cogsci.berkeley.edu/lakoff/MetaphorHome.html)
10. Martin, J.H. 1990. [A computational model of metaphor interpretation](https://md.research.elamur.ai/free/en/papers/fass-1991-computational-model-metaphor-interpretation/). Academic Press, New York.
11. Zaenen, A., L. Karttunen, and R. Crouch. 2005. Local textual inference: Can it be defined or circumscribed? In Proceedings of the ACL 05 Workshop on Empirical Modelling of Semantic Equivalence and Entailment, pages 31–36. [find](https://scholar.google.com/scholar?q=Zaenen%2C+A%2E%2C+L%2E+Karttunen%2C+and+R%2E+Crouch%2E+2005%2E+Local+textual+inference%3A+Can+it+be+defined+or+circumscribed%3F+In+Proceedings+of+the+ACL+05+Workshop+on+Empirical+Modelling+of+Semantic+Equivalence+and+Ent)

*Updated 1 November 2026*