What's in a Bridge?: A Descriptive, Multi-Genre Analysis of the GUMBridge Corpus for Varieties of Bridging Anaphora

What it is about

The paper is a descriptive analysis of the GUMBridge corpus: 5.3k referential bridging instances in 273k tokens across 16 genres (the train, dev and test partitions). The overall density is 19.5 instances per 1k tokens. Spoken genres have fewer bridging instances than written ones. The most common subtypes are ENTITY-ASSOCIATIVE (43.2% of labels) and COMPARISON-RELATIVE (18.8%). Random forests on categorical and numeric features separate bridging anaphors from other mentions with F1 65.6, against a random baseline of 51.2. For antecedents, F1 is 54.8, against 49.6. Bridging anaphors tend to be short, definite, and to have time or place entity types. Antecedents tend to belong to larger coreference chains and to be more salient. For antecedents more than 150 tokens away, salient entities are more common (Kolmogorov–Smirnov p = 3.7e-12), but over 90% of pairs are under 150 tokens apart.

Ideas

• I1. Write the feature list as a transparent decision list in coref: entity type "time" favours a bridging anaphor, and a high salience count favours an antecedent. Measure each rule against the random baseline, not only against the others.
• I2. Build a chi-square residual report in a Rust tool under tools/, sorted with BTreeMap, to show which labels are over- or under-represented per genre. This is the paper's Table 4 and Figure 2 method, and it is a clear way to explain a result.
• I3. Use chain size and salience from coref as antecedent features, with the distance bucket as a separate check.
• I4. Keep the random baseline in every table as a negative control, as the paper does, so that a gain on a rule is shown to be above chance.

For Mova

The chapter is 35. Reference and coreference and, for the distance and genre analysis, 37. Register and variation. The code is coref/src/mention.rs and coref/src/sieve.rs, and the report tool in tools/. The measure is F1 against the random baseline per feature rule, run twice for determinism. The random forest is not for Mova; the feature list and the residual report are.

Topics: Coreference resolution

Cite

The paper itself, not this card. BibTeX from the library's record of the work. DOI 10.18653/v1/2026.codi-1.7

@inproceedings{levine2026what,
  title = {{What's in a Bridge?: A Descriptive, Multi-Genre Analysis of the GUMBridge Corpus for Varieties of Bridging Anaphora}},
  author = {Levine, Lauren and Zeldes, Amir},
  booktitle = {Proceedings of the 2nd Joint Workshop on Computational Approaches to Discourse, Context and Document-Level Inferences and Computational Models of Reference, Anaphora and Coreference (CODI-CRAC 2026)},
  year = {2026},
  doi = {10.18653/v1/2026.codi-1.7},
  url = {https://doi.org/10.18653/v1/2026.codi-1.7}
}

References

26 references, 6 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

1. Tatsuya Aoyama, Shabnam Behzad, Luke Gessler, Lauren Levine, Jessica Lin, Yang Janet Liu, Siyao Peng, Yilun Zhu, and Amir Zeldes. 2023. GENTLE: A genre-diverse multilayer challenge set for English NLP and linguistic evaluation. In Proceedings of the 17th Linguistic Annotation Workshop (LAW-XVII), pages 166–178, Toronto, Canada. Association for Computational Linguistics. ↗ our card
2. Denis Miéville. 1999. Associative anaphora: An attempt at a formalization. Journal of Pragmatics, 31:327–337. find
3. Nicholas Asher and Alex Lascarides. 1998. Bridging. Journal of Semantics, 15(1):83–113. find
4. Anna Nedoluzhko, Jiří Mírovský, Radek Ocelák, and Jiří Pergler. 2009. Extended coreferential relations and bridging anaphora in the Prague Dependency Treebank. In Proceedings of the 7th Discourse Anaphora and Anaphor Resolution Colloquium (DAARC 2009), pages 1–16, Goa, India. find
5. Stefan Baumann and Arndt Riester. 2012. Referential and lexical givenness: Semantic, prosodic and cognitive aspects. Prosody and meaning, 25:119–162. ↗
6. Anders Björkelund, Kerstin Eckart, Arndt Riester, Nadja Schauffler, and Katrin Schweitzer. 2014. The extended DIRNDL corpus as a resource for coreference and bridging resolution. In Proceedings of the Ninth International Conference on Language Resources and Evaluation (LREC’14), pages 3222– 3228, Reykjavik, Iceland. ↗
7. Malvina Nissim, Shipra Dingare, Jean Carletta, and Mark Steedman. 2004. An annotation scheme for information status in dialogue. In Proceedings of the Fourth International Conference on Language Resources and Evaluation (LREC‘04), Lisbon, Portugal. European Language Resources Association (ELRA). ↗
8. Massimo Poesio and Ron Artstein. 2008. Anaphoric annotation in the ARRAU corpus. In Proceedings of the Sixth International Conference on Language Resources and Evaluation (LREC’08), Marrakech, Morocco. European Language Resources Association (ELRA). ↗
9. Herbert H. Clark. 1975. Bridging. In Theoretical Issues in Natural Language Processing. find
10. Marie-Catherine de Marneffe, Christopher D. Manning, Joakim Nivre, and Daniel Zeman. 2021. Universal Dependencies. Computational Linguistics, 47(2):255–308. find our card
11. Massimo Poesio, Renata Vieira, and Simone Teufel. 1997. Resolving bridging references in unrestricted text. In Operational Factors in Practical, Robust Anaphora Resolution for Unrestricted Texts. ↗
12. Biaoyan Fang, Timothy Baldwin, and Karin Verspoor. 2022. What does it take to bake a cake? the RecipeRef corpus and anaphora resolution in procedural text. In Findings of the Association for Computational Linguistics: ACL 2022, pages 3481–3495, Dublin, Ireland. Association for Computational Linguistics. ↗
13. Ellen F. Prince. 1981. Toward a taxonomy of given-new information. Radical pragmatics, pages 223–255. find
14. Ina Rösiger. 2018. BASHI: A corpus of Wall Street Journal articles annotated with bridging links. In Proceedings of the Eleventh International Conference on Language Resources and Evaluation (LREC 2018), Miyazaki, Japan. European Language Resources Association (ELRA). ↗
15. John A. Hawkins. 1978. Definiteness and indefiniteness: A study in reference and grammaticality prediction. Journal of Linguistics, 27:405–442. find
16. Lauren Levine and Amir Zeldes. 2026. GUMBridge: A corpus for varieties of bridging anaphora. In Proceedings of the Fifteenth Language Resources and Evaluation Conference (LREC 2026), pages 6823– 6837, Palma, Mallorca, Spain. European Language Resources Association (ELRA). ↗ our card
17. Ina Rösiger, Arndt Riester, and Jonas Kuhn. 2018. Bridging resolution: Task definition, corpus resources and rule-based experiments. In Proceedings of the 27th International Conference on Computational Linguistics, pages 3516–3528, Santa Fe, New Mexico, USA. Association for Computational Linguistics. find our card
18. Jessica Lin and Amir Zeldes. 2025. GUM-SAGE: A novel dataset and approach for graded entity salience prediction. In Findings of the Association for Computational Linguistics: ACL 2025, pages 438–455, Vienna, Austria. Association for Computational Linguistics. ↗ our card
19. Beatrice Santorini. 1990. Part-of-speech tagging guidelines for the Penn Treebank Project. Technical report, Department of Computer and Information Science, University of Pennsylvania, Philadelphia, PA. find
20. Olga Uryupina, Ron Artstein, Antonella Bristot, Federica Cavicchio, Francesca Delogu, Kepa Joseba Rodríguez, and Massimo Poesio. 2019. Annotating a broad range of anaphoric phenomena, in a variety of genres: the arrau corpus. Natural Language Engineering, 26:95 – 128. ↗
21. Jessica Lin and Amir Zeldes. 2026. Expect the unexpected? Testing the surprisal of salient entities. In Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (ACL 2026), San Diego, CA. Association for Computational Linguistics. find
22. Klaus Von Heusinger and Petra B. Schumacher. 2019. Discourse prominence: Definition and application. Journal of Pragmatics, 154:117–127. ↗
23. Katja Markert, Yufang Hou, and Michael Strube. 2012. Collective classification for fine-grained information status. In Proceedings of the 50th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pages 795–804, Jeju Island, Korea. Association for Computational Linguistics. find
24. Amir Zeldes. 2017. The GUM corpus: Creating multilayer resources in the classroom. Language Resources and Evaluation, 51(3):581–612. SET - SPAN - INTERVAL The anaphor is a sub-span of the spatial or temporal antecedent interval (e.g, Sunday → the morning). ↗
25. Amir Zeldes. 2022. Opinion piece: Can we fix the scope for coreference? Dialogue & Discourse, 13:41–62. find our card
26. Amir Zeldes and Jessica Lin. 2026. What makes an entity salient in discourse? Corpus Linguistics and Linguistic Theory, pages 1–42. A OTHER The OTHER category is for instances which fit the information status based definition of a bridging pair but do not fall into any of the bridging subtype categories outlined above. GUMBridge Bridging Subtypes This appendix briefly details the bridging subtype varieties annotated in the GUMBridge corpus, which are reflected in its guidelines. For the full guidelines, we refer readers to Levine and Zeldes 2026. ↗

This page in Markdown → https://md.mova.elamur.ai/free/en/docs/papers/levine-2026-what-s-bridge-descriptive-multi/