---
title: "Augmenting English Adjective Senses with Supersenses (2014)"
description: "WordNet groups adjectives into small clusters of near-synonyms but gives them no… 55% agreement. 22 references, cited by 22 papers here."
authors: ["Yulia Tsvetkov", "Nathan Schneider", "Dirk Hovy", "Archna Bhatia", "Manaal Faruqui", "Chris Dyer"]
year: 2014
venue: "Proceedings of the Ninth International Conference on Language Resources and Evaluation (LREC'14)"
source: "https://aclanthology.org/L14-1073/"
licence: "CC BY-NC 4.0 (ELRA, LREC proceedings)"
licence_url: "https://creativecommons.org/licenses/by-nc/4.0/"
summary_licence: "Elamur Cosmic License"
canonical: "https://research.elamur.ai/free/en/papers/tsvetkov-2014-augmenting-english-adjective-senses-supersenses/"
---
[Elamur Research](https://md.research.elamur.ai/free/en/) · [this page in HTML](https://research.elamur.ai/free/en/papers/tsvetkov-2014-augmenting-english-adjective-senses-supersenses/)

[home](https://md.research.elamur.ai/free/en/) / [papers](https://md.research.elamur.ai/free/en/papers/) / [2014](https://md.research.elamur.ai/free/en/papers/2014/page/2/) / Augmenting English Adjective Senses with Supersenses

# Augmenting English Adjective Senses with Supersenses

**Authors:** [Yulia Tsvetkov](https://md.research.elamur.ai/free/en/authors/yulia-tsvetkov/), [Nathan Schneider](https://md.research.elamur.ai/free/en/authors/nathan-schneider/), [Dirk Hovy](https://md.research.elamur.ai/free/en/authors/dirk-hovy/), [Archna Bhatia](https://md.research.elamur.ai/free/en/authors/archna-bhatia/), [Manaal Faruqui](https://md.research.elamur.ai/free/en/authors/manaal-faruqui/) and [Chris Dyer](https://md.research.elamur.ai/free/en/authors/chris-dyer/) · **Year:** 2014 · **Venue:** [Proceedings of the Ninth International Conference on Language Resources and Evaluation (LREC'14)](https://md.research.elamur.ai/free/en/venues/lrec/) ([LREC 2014](https://md.research.elamur.ai/free/en/venues/lrec/2014/)) · **Publisher:** [European Language Resources Association (ELRA)](https://md.research.elamur.ai/free/en/publishers/european-language-resources-association-elra/) · **Licence:** [CC BY-NC 4.0 (ELRA, LREC proceedings)](https://creativecommons.org/licenses/by-nc/4.0/)

**Links:** [Publisher's page](https://aclanthology.org/L14-1073/) · [PDF](https://aclanthology.org/L14-1073.pdf) · [OpenAlex](https://openalex.org/W265224334)

**Citations:** 35 citations (OpenAlex, 2026-10-14) · [22](https://md.research.elamur.ai/free/en/papers/tsvetkov-2014-augmenting-english-adjective-senses-supersenses/cited-by/) in our library, of the 141,840 papers whose bibliographies we read

## Summary
WordNet groups adjectives into small clusters of near-synonyms but gives them no shared top-level organisation. Nothing marks, for instance, that "ashamed" and "exasperated" both describe emotions. The authors fill this gap by adapting GermaNet's purely semantic adjective taxonomy into 13 coarse English supersenses, such as PERCEPTION, SPATIAL, FEELING, BEHAVIOR and SOCIAL.

They start from a small seed list and grow it with affix-based and WordNet-link expansion, filtering the new candidates through a classifier. On this data they train a random forest that ranks supersenses for an adjective type, using only dense distributional vectors as features.

To move from words to senses, they compare two ways of labelling synsets:
- combining classifier outputs over a synset's lemmas;
- crowdsourcing labels for lemmas in SemCor sentences.

Type-level prediction works well: the correct class is among the top four 91% of the time. Synset labelling is much harder, and even two trained annotators agree only moderately. Soft voting with the classifier beats crowdsourcing.

The released labelling of 12,304 synsets gives lexical-semantic tools a cheap coarse layer for adjectives. It has already helped metaphor detection.

## Key points
- **Taxonomy:** 13 supersenses adapted from GermaNet (Hamp and Feldweg), each with finer subclasses, e.g. PERCEPTION covers color, taste and sound.
- **Seed data:** 1,223 seed adjectives gave 1,294 instances, of which 127 were held out. After filtered expansion, 1,473 training examples remained (about 113 per class).
- **Classifier:** a scikit-learn random forest over 64-dimensional cross-lingually enriched vectors (Faruqui and Dyer 2014), trained on WMT-2011 news commentary. The vectors cover 10,793 adjective types.
- **Type accuracy:** 54% at top-1 (random baseline 7%) and 91% at top-4.
- **Synset evaluation:** 148 clusters labelled by two annotators, with 55% agreement (κ = 0.47).
  - Soft voting scored best: 44.6% accuracy against the expert, 56.1% when either annotator's label counts.
  - Crowdsourcing on AMT cost $170 and covered 5,077 sentences aggregated with MACE, but yielded only 438 labelled clusters.

## Interesting ideas
- When growing seeds through noisy links (synonyms, antonyms, affixes), keep a new candidate only if a model trained on the original seeds assigns it the seed's label; this cleans out much of the noise that polysemy introduces.
- A word-level class distribution can be turned into a sense-level label by summing the class posteriors of all the lemmas that share a sense, so synonyms resolve each other's ambiguity.
- Judging a ranker by whether the correct label falls in its top-k shows that it is useful as a candidate filter even when its top-1 accuracy is modest.
- Annotators work faster when they see the classifier's ranked choices first (top four visible), and a lenient metric that accepts either annotator's label makes the gold standard's own fuzziness explicit.
- Coarse adjective supersenses describe what kind of property an adjective expresses, and the authors note they helped tell metaphoric adjective-noun pairs from literal ones.

## For Mova

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

- **I1** (learn): silver examples produced by lexical or structural expansion could pass through the same "does the current model agree with the inherited label" gate before entering training.
- **I2** (world): when compiling the WordNet ontology, coarse classes for adjective synsets could be derived this way, giving the world model typed properties (colour, age, emotion) rather than raw adjective strings.
- **I3** (learn): evaluation of class and role predictors could report top-k recall, telling us when a later explainable step only needs to choose among a few candidates.
- **I4** (learn): LLM teacher tiers or human checks could receive pre-ranked options, and scoring could keep both a strict and an "any acceptable label" measure for inherently vague categories.
- **I5** (frames): an adjective whose supersense clashes with its noun's type (a FEELING word on an inanimate thing) could flag figurative use for the pragmatics layer.

*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

- [Carnegie Mellon University](https://md.research.elamur.ai/free/en/universities/carnegie-mellon-university/) — University, Pittsburgh, United States · 365 works · 158,048 citations
- [University of Copenhagen](https://md.research.elamur.ai/free/en/universities/university-of-copenhagen/) — University, Copenhagen, Denmark · 78 works · 14,228 citations

📚 Shelves (6): [🖥️ Computer Science](https://md.research.elamur.ai/free/en/shelves/computer-science/) · 9,085 works · [🤖 Artificial Intelligence](https://md.research.elamur.ai/free/en/shelves/computer-science/artificial-intelligence/) · 8,888 works · [🎭 Arts and Humanities](https://md.research.elamur.ai/free/en/shelves/arts-and-humanities/) · 9,296 works · [🗣️ Language and Linguistics](https://md.research.elamur.ai/free/en/shelves/arts-and-humanities/language-and-linguistics/) · 9,206 works · [💭 Psychology](https://md.research.elamur.ai/free/en/shelves/psychology/) · 798 works · [💭 Experimental and Cognitive Psychology](https://md.research.elamur.ai/free/en/shelves/psychology/experimental-and-cognitive-psychology/) · 463 works

🧭 Topics (1): [English](https://md.research.elamur.ai/free/en/topics/english/) · 1,373 works

## Cite

Cite as: Yulia Tsvetkov, Nathan Schneider, Dirk Hovy, Archna Bhatia, Manaal Faruqui and Chris Dyer. 2014. Augmenting English Adjective Senses with Supersenses. Proceedings of the Ninth International Conference on Language Resources and Evaluation (LREC'14).

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

```bibtex
@inproceedings{tsvetkov2014augmenting,
  title = {{Augmenting English Adjective Senses with Supersenses}},
  author = {Yulia Tsvetkov, Nathan Schneider, Dirk Hovy, Archna Bhatia, Manaal Faruqui and Chris Dyer},
  booktitle = {Proceedings of the Ninth International Conference on Language Resources and Evaluation (LREC'14)},
  year = {2014},
  doi = {10.63317/5edop9rocohn},
  url = {https://doi.org/10.63317/5edop9rocohn}
}
```

- [← previous Summarizing Online Forum Discussions – Can Dialog Acts of Individual Messages Help?](https://md.research.elamur.ai/free/en/papers/bhatia-2014-summarizing-online-forum-discussions-can/)
- [next → Etymological Wordnet: Tracing The History of Words](https://md.research.elamur.ai/free/en/papers/melo-2014-demelo-etymological-wordnet/)
- [↑ back to papers of 2014](https://md.research.elamur.ai/free/en/papers/2014/page/2/)

## References

22 references, 9 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

1. Breiman, Leo (2001). Random forests. Learning, 45(1):5–32. [find](https://scholar.google.com/scholar?q=Breiman%2C+Leo+%282001%29%2E+Random+forests%2E+Learning%2C+45%281%29%3A5%E2%80%9332%2E)
2. Machine Ciaramita, Massimiliano and Altun, Yasemin (2006). [Broad-coverage sense disambiguation and information extraction with a supersense sequence tagger](https://md.research.elamur.ai/free/en/papers/ciaramita-2006-broad-coverage-sense-disambiguation-information/). In Proc. of EMNLP, pages 594–602. Association for Computational Linguistics.
3. Dixon, Robert M. W. (1982). Where have all the adjectives gone? In Dixon, Robert M. W., editor, Where Have All the Adjectives Gone?: And Other Essays in Semantics and Syntax, pages 1–62. Mouton. [↗](https://doi.org/10.1515/9783110822939)
4. Dixon, Robert M. W. and Aikhenvald, Aleksandra (2004). Adjective classes. Oxford University Press. [find](https://scholar.google.com/scholar?q=Dixon%2C+Robert+M%2E+W%2E+and+Aikhenvald%2C+Aleksandra+%282004%29%2E+Adjective+classes%2E+Oxford+University+Press%2E)
5. Faruqui, Manaal and Dyer, Chris (2014). [Improving vector space word representations using multilingual correlation](https://md.research.elamur.ai/free/en/papers/faruqui-2014-improving-vector-space-word-representations/). In Proc. of EACL. Association for Computational Linguistics.
6. Fellbaum, Christiane, editor (1998). WordNet: an electronic lexical database. MIT Press. [find](https://scholar.google.com/scholar?q=Fellbaum%2C+Christiane%2C+editor+%281998%29%2E+WordNet%3A+an+electronic+lexical+database%2E+MIT+Press%2E)
7. Gross, Derek and Miller, Katherine J. (1990). Adjectives in WordNet. International Journal of Lexicography, 3(4):265–277. [↗](https://doi.org/10.1093/ijl/3.4.265)
8. Hamp, Birgit and Feldweg, Helmut (1997). [GermaNet - a lexical-semantic net for German](https://md.research.elamur.ai/free/en/papers/hamp-1997-germanet-lexical-semantic-net-german/). In Proc. of the Workshop on Automatic Information Extraction and Building of Lexical Semantic Resources for NLP Applications, pages 9–15. Association for Computational Linguistics.
9. Hovy, Dirk, Berg-Kirkpatrick, Taylor, Vaswani, Ashish, and Hovy, Eduard (2013). [Learning whom to trust with MACE](https://md.research.elamur.ai/free/en/papers/hovy-2013-learning-whom-trust-mace/). In Proc. of NAACL-HLT, pages 1120–1130. Association for Computational Linguistics.
10. Hundsnurscher, Franz and Splett, Jochen (1982). Semantik der Adjektive des Deutschen. 3137. Westdeutscher Verlag. [↗](https://doi.org/10.1007/978-3-322-87624-9)
11. Marrafa, Palmira and Mendes, Sara (2006). Modeling adjectives in computational relational lexica. In Proc. of COLING, pages 555–562. Association for Computational Linguistics. [find](https://scholar.google.com/scholar?q=Marrafa%2C+Palmira+and+Mendes%2C+Sara+%282006%29%2E+Modeling+adjectives+in+computational+relational+lexica%2E+In+Proc%2E+of+COLING%2C+pages+555%E2%80%93562%2E+Association+for+Computational+Linguistics%2E)
12. Miller, George A., Leacock, Claudia, Tengi, Randee, and Bunker, Ross T. (1993). [A semantic concordance](https://md.research.elamur.ai/free/en/papers/miller-1993-semantic-concordance/). In Proc. of HLT, pages 303–308. Association for Computational Linguistics.
13. Munro, Robert, Bethard, Steven, Kuperman, Victor, Lai, Vicky Tzuyin, Melnick, Robin, Potts, Christopher, Schnoebelen, Tyler, and Tily, Harry (2010). [Crowdsourcing and language studies: the new generation of linguistic data](https://md.research.elamur.ai/free/en/papers/burch-2010-creating-speech-language-data-amazon/). In Proc. of the NAACL HLT 2010 Workshop on Creating Speech and Language Data with Amazon’s Mechanical Turk, pages 122–130. Association for Computational Linguistics.
14. Nastase, Vivi (2008). Unsupervised all-words word sense disambiguation with grammatical dependencies. In Proc. of IJCNLP, pages 7–12. Asian Federation of Natural Language Processing. [find](https://scholar.google.com/scholar?q=Nastase%2C+Vivi+%282008%29%2E+Unsupervised+all%2Dwords+word+sense+disambiguation+with+grammatical+dependencies%2E+In+Proc%2E+of+IJCNLP%2C+pages+7%E2%80%9312%2E+Asian+Federation+of+Natural+Language+Processing%2E)
15. Paaß, Gerhard and Reichartz, Frank (2009). Exploiting semantic constraints for estimating supersenses with CRFs. In Proc. of SDM. Society for Industrial and Applied Mathematics. [find](https://scholar.google.com/scholar?q=Paa%C3%9F%2C+Gerhard+and+Reichartz%2C+Frank+%282009%29%2E+Exploiting+semantic+constraints+for+estimating+supersenses+with+CRFs%2E+In+Proc%2E+of+SDM%2E+Society+for+Industrial+and+Applied+Mathematics%2E)
16. Pedregosa, Fabian, Varoquaux, Gael, Gramfort, Alexandre, Michel, Vincent, Thirion, Bertrand, Grisel, Olivier, Blondel, Mathieu, Prettenhofer, Peter, Weiss, Ron, Dubourg, Vincent, Vanderplas, Jake, Passos, Alexandre, Cournapeau, David, Brucher, Matthieu, Perrot, M., and Duchesnay, Edouard (2011). Scikit-learn: Machine learning in Python. Journal of Machine Learning Research, 12:2825–2830. [↗](https://doi.org/10.48550/arxiv.1201.0490)
17. Peters, Ivonne and Peters, Wim (2000). The treatment of adjectives in SIMPLE: Theoretical observations. In LREC. European Language Resources Association. [find](https://scholar.google.com/scholar?q=Peters%2C+Ivonne+and+Peters%2C+Wim+%282000%29%2E+The+treatment+of+adjectives+in+SIMPLE%3A+Theoretical+observations%2E+In+LREC%2E+European+Language+Resources+Association%2E)
18. Raskin, Victor and Nirenburg, Sergei (1995). Lexical semantics of adjectives. Technical Report MCCS-95-288, New Mexico State University Computing Research Laboratory, Las Cruces, NM. [http://web.ics.purdue.edu/~vraskin/adjective.pdf](http://web.ics.purdue.edu/~vraskin/adjective.pdf) . [↗](http://web.ics.purdue.edu/~vraskin/adjective.pdf)
19. Schneider, Nathan, Mohit, Behrang, Oflazer, Kemal, and Smith, Noah A. (2012). [Coarse lexical semantic annotation with supersenses: an Arabic case study](https://md.research.elamur.ai/free/en/papers/schneider-2012-coarse-lexical-semantic-annotation-supersenses/). In Proc. of ACL, pages 253–258. Association for Computational Linguistics.
20. Snow, Rion, O’Connor, Brendan, Jurafsky, Dan, and Ng, Andrew Y. (2008). [Cheap and fast—but is it good?](https://md.research.elamur.ai/free/en/papers/snow-2008-cheap-fast-but-is-it/) Evaluating non-expert annotations for natural language tasks. In Proc. of EMNLP, pages 254– 263. Association for Computational Linguistics.
21. Torrent, Gemma Boleda, im Walde, Sabine Schulte, and Badia, Toni (2012). Modeling regular polysemy: A study on the semantic classification of Catalan adjectives. Computational Linguistics, 38(3):575–616. [↗](https://doi.org/10.1162/coli_a_00093)
22. Tsvetkov, Yulia, Boytsov, Leonid, Gershman, Anatole, Nyberg, Eric, and Dyer, Chris (2014). [Metaphor detection with cross-lingual model transfer](https://md.research.elamur.ai/free/en/papers/tsvetkov-2014-metaphor-detection-cross-lingual-model/). In Proc. of ACL. Association for Computational Linguistics.

*Updated 24 October 2026*