Elamur Research⌕
☰
home / papers / 2025 / Compositional Syntactico-SemBanking for English as a Second or Foreign Language

Compositional Syntactico-SemBanking for English as a Second or Foreign Language

Paper · 2025 · Wenxi Li, Xihao Wang and Weiwei Sun

Summary

Non-native English (ESFL) is full of dropped articles, extra prepositions, unusual word orders and verbs with nonstandard argument patterns, yet readers still recover what the writer meant. Classic compositional models assume a fixed lexicon plus grammar rules, and they break down on such input. The authors take a constructivist stance. Every form–meaning pairing, including a so-called "error", counts as a construction, and constructions are the building blocks of composition. Each construction is encoded as a Synchronous Hyperedge Replacement Grammar (SHRG) rule. The rule pairs a CFG rule for syntax with an HRG rule that builds English Resource Semantics (ERS) graphs, so the tree and the meaning graph are derived in lockstep. Working with learner sentences from the FCE dataset, they built a gold-standard bank of 1643 ESFL sentences. They did this by checking and repairing earlier automatic analyses and by hand-composing rules for sentences the ACE parser could not handle. Annotators agreed closely. The resource suggests that learner English composes meaning much as native English does, and it offers acquisition research deeper analyses than existing learner corpora.

Key points

  • Representation: CFG trees for form and ERS hypergraphs for meaning, linked by SHRG rules.
    • Each rule pair shares a left-hand-side label.
    • When a nonterminal hyperedge is replaced, the external nodes of the inserted subgraph are fused with the nodes it was attached to.
  • Source data: FCE learner essays.
    • In earlier processing (Zhao et al., 2020), the ACE parser handled 52.50% of sentences and left 2494 unanalyzed.
    • 100 of those unanalyzed sentences were randomly sampled for manual composition.
  • Bank breakdown: 724 analyses accepted as is, 819 modified, and 100 composed from scratch, for 1643 in total.
    • Among rejected analyses, 81.2% (665 graphs) required rebuilding the syntactic tree.
  • Agreement:
    • In the review phase, 94% of one annotator's graphs were accepted by the other.
    • Independent annotation of 50 sentences reached an average S-match of 90.93.
  • Phenomena covered include omission, insertion, transposition and divergent verb argument structure.
    • The authors contrast their approach with CCG, which would need a new lexical entry, for example for "often" placed after the verb.

Interesting ideas

  • I1. Instead of adding new lexical entries for each learner deviation, loosen an existing construction (e.g. allow a post-verbal adverb while keeping the verb as its argument), so the fix generalizes across words.
  • I2. Each rule builds a syntax fragment and a meaning fragment at the same time, with explicit docking nodes saying where sub-meanings plug in.
  • I3. Treat verbs whose argument structure differs from the standard one as modified versions of the original rule, keeping both the original and the modified rule side by side.
  • I4. Triage existing automatic output into accept, modify or build from scratch, and record which tier each item came from.
  • I5. Measure agreement twice: first how often a reviewer accepts another annotator's graph, then a structural overlap score on fully independent annotations.

For Mova

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

  • I1 (tree): Non-native or noisy input could be handled by a small set of relaxed attachment patterns layered over standard UD rules, each one named and logged so the parse stays explainable.
  • I2 (frames): Pair each UD construction pattern with a frame fragment that has typed slots, so roles are filled during tree assembly rather than in a separate pass.
  • I3 (world): Verb-class knowledge from VerbNet could store attested nonstandard frames as variants linked to the canonical class, so unusual argument patterns still map to the same event type.
  • I4 (learn): Silver data from witness parsers could carry this tier label, so training can weight items differently and evaluation can report error rates by tier.
  • I5 (learn): This gives a cheap check plus a strict check for LLM-teacher or self-repair outputs, revealing when reviewers are merely rubber-stamping.

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 CambridgeCambridge · United Kingdom243 works · 121,396 citations🎓Peking UniversityBeijing · China244 works · 39,359 citations🎓Minzu University of ChinaBeijing · China5 works · 2 citations

 Fields, topics and concepts

📚 Shelves 5The 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 worksComputational Theory and Mathematics · 252 works

🧭 Topics 1Research subjects and languages the work is about.English · 2,263 works

Cite

Cite as Wenxi Li, Xihao Wang and Weiwei Sun. 2025. Compositional Syntactico-SemBanking for English as a Second or Foreign Language. Findings of the Association for Computational Linguistics: ACL 2025.

The paper itself, not this page. BibTeX from the library's record of the work. DOI 10.18653/v1/2025.findings-acl.1252

@inproceedings{li2025compositional,
  title = {{Compositional Syntactico-SemBanking for English as a Second or Foreign Language}},
  author = {Wenxi Li, Xihao Wang and Weiwei Sun},
  booktitle = {Findings of the Association for Computational Linguistics: ACL 2025},
  year = {2025},
  doi = {10.18653/v1/2025.findings-acl.1252},
  url = {https://doi.org/10.18653/v1/2025.findings-acl.1252}
}

References

41 references, 8 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. Christian Adjemian. 1976. On the nature of interlanguage systems. Language learning, 26(2):297–320. ↗
  2. Gustavo Aguilar and Thamar Solorio. 2020. From English to code-switching: Transfer learning with strong morphological clues. In Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics, pages 8033–8044, Online. Association for Computational Linguistics. find
  3. Elizabeth Bates and Brian MacWhinney. 1987. Competition, variation, and language learning. In Elizabeth Bates, editor, Mechanisms of language acquisition, pages 157–193. Erlbaum, Hillsdale, NJ. find
  4. Yevgeni Berzak, Jessica Kenney, Carolyn Spadine, Jing Xian Wang, Lucia Lam, Keiko Sophie Mori, Sebastian Garza, and Boris Katz. 2016. Universal Dependencies for learner English. In Proceedings of the 54th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pages 737–746, Berlin, Germany. Association for Computational Linguistics.
  5. Douglas Biber and Randi Reppen. 2015. The Cambridge handbook of English corpus linguistics. Cambridge University Press. find
  6. Christopher Bryant, Zheng Yuan, Muhammad Reza Qorib, Hannan Cao, Hwee Tou Ng, and Ted Briscoe. 2023. Grammatical error correction: A survey of the state of the art. Computational Linguistics, pages 643–701. ↗
  7. Yufei Chen and Weiwei Sun. 2020. Parsing into variable-in-situ logico-semantic graphs. In Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics, pages 6772–6782, Online. Association for Computational Linguistics. ↗
  8. David M. Eberhard, Gary F. Simons, and Charles D. Fennig, editors. 2024. Ethnologue: Languages of the World, twenty-seventh edition. SIL International, Dallas, Texas. Online version: http://www.ethnologue.com. ↗
  9. Dan Flickinger. 2000. On building a more efficient grammar by exploiting types. Natural Language Engineering, 6(1):15–28.
  10. Dan Flickinger, Emily M. Bender, and Stephan Oepen. 2014a. Towards an encyclopedia of compositional semantics: Documenting the interface of the English Resource Grammar. In Proceedings of the Ninth International Conference on Language Resources and Evaluation (LREC-2014), pages 875–881. European Language Resources Association (ELRA). find
  11. Dan Flickinger, Emily M. Bender, and Stephan Oepen. 2014b. Towards an encyclopedia of compositional semantics: Documenting the interface of the English Resource Grammar. In Proceedings of the Ninth International Conference on Language Resources and Evaluation (LREC’14), pages 875–881, Reykjavik, Iceland. European Language Resources Association (ELRA). find
  12. Dan Flickinger, Yi Zhang, and Valia Kordoni. 2012. Deepbank: A dynamically annotated treebank of the wall street journal. In Proceedings of the 11th International Workshop on Treebanks and Linguistic Theories, pages 85–96. find
  13. Steven M Frankland and Joshua D Greene. 2020. Concepts and compositionality: in search of the brain’s language of thought. Annual review of psychology, 71(1):273–303. ↗
  14. Jeroen Geertzen, Theodora Alexopoulou, Anna Korhonen, et al. 2013. Automatic linguistic annotation of large scale l2 databases: The ef-cambridge open language database (efcamdat). In Proceedings of the 31st Second Language Research Forum. Somerville, MA: Cascadilla Proceedings Project, pages 240– 254. find
  15. Adele E Goldberg. 1995. Constructions: A construction grammar approach to argument structure. Chicago UP. find
  16. Adele E Goldberg. 2003. Constructions: A new theoretical approach to language. Trends in cognitive sciences, 7(5):219–224. ↗
  17. Adele E Goldberg. 2013. Argument structure constructions versus lexical rules or derivational verb templates. Mind & Language, 28(4):435–465. ↗
  18. Sylviane Granger. 2003. The international corpus of learner english: a new resource for foreign language learning and teaching and second language acquisition research. Tesol Quarterly, 37(3):538–546. ↗
  19. Sylviane Granger. 2014. Learner English on computer. Routledge. ↗
  20. James A Hampton and Yoad Winter. 2017. Compositionality and concepts in linguistics and psychology. Springer Nature. ↗
  21. Thomas Hoffmann and Graeme Trousdale. 2013. The Oxford handbook of construction grammar. Oxford University Press. find
  22. Dieuwke Hupkes, Verna Dankers, Mathijs Mul, and Elia Bruni. 2020. Compositionality decomposed: How do neural networks generalise? Journal of Artificial Intelligence Research, 67:757–795. ↗
  23. Ray Jackendoff. 2000. The representational structures of the language faculty and their interactions. In The Neurocognition of Language. Oxford University Press. ↗
  24. Ray Jackendoff. 2002. Foundations of Language: Brain, Meaning, Grammar, Evolution. Oxford University Press. find
  25. Simran Khanuja, Sandipan Dandapat, Anirudh Srinivasan, Sunayana Sitaram, and Monojit Choudhury. 2020. GLUECoS: An evaluation benchmark for code-switched NLP. In Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics, pages 3575–3585, Online. Association for Computational Linguistics.
  26. Ingrid Kurz. 2008. The impact of non-native english on students’ interpreting performance. In Efforts and models in interpreting and translation research: A tribute to Daniel Gile, pages 179–192. Benjamins Publishing. ↗
  27. Ronald W Langacker. 1987. Language, pages 53–94. Nouns and verbs. find
  28. Brian MacWhinney. 2000. The CHILDES project: The database, volume 2. Psychology Press. find
  29. Marie-Catherine de Marneffe, Timothy Dozat, Natalia Silveira, Katri Haverinen, Filip Ginter, Joakim Nivre, and Christopher D. Manning. 2014. Universal Stanford dependencies: A cross-linguistic typology. In Proceedings of the Ninth International Conference on Language Resources and Evaluation (LREC’14), pages 4585–4592, Reykjavik, Iceland. European Language Resources Association (ELRA). ↗
  30. Joakim Nivre, Marie-Catherine de Marneffe, Filip Ginter, Yoav Goldberg, Jan Hajič, Christopher D. Manning, Ryan McDonald, Slav Petrov, Sampo Pyysalo, Natalia Silveira, Reut Tsarfaty, and Daniel Zeman. 2016. Universal Dependencies v1: A multilingual treebank collection. In Proceedings of the Tenth International Conference on Language Resources and Evaluation (LREC’16), pages 1659–1666, Portorož, Slovenia. European Language Resources Association (ELRA). ↗
  31. Barbara Partee. 1984. Compositionality. In Varieties of Formal Semantics: Proceedings of the 4th Amsterdam Colloquium, volume 3, pages 281–311. Foris Dordrecht. find
  32. Carl Pollard and Ivan A Sag. 1994. Head-driven phrase structure grammar. University of Chicago Press. find
  33. Peter Robinson and Nick C Ellis. 2008. Handbook of cognitive linguistics and second language acquisition. Routledge, London. find
  34. Kenji Sagae, Eric Davis, Alon Lavie, Brian MacWhinney, and Shuly Wintner. 2007. High-accuracy annotation and parsing of CHILDES transcripts. In Proceedings of the Workshop on Cognitive Aspects of Computational Language Acquisition, pages 25– 32, Prague, Czech Republic. Association for Computational Linguistics. ↗
  35. Mark Steedman. 1987. Combinatory grammars and parasitic gaps. Natural Language & Linguistic Theory, 5(3):403–439. ↗
  36. Mark Steedman. 2000. The Syntactic Process. MIT Press, Cambridge, MA. ↗
  37. Michael Tomasello. 2003. Constructing a Language: A Usage-Based Theory of Language Acquisition. Harvard University Press. find
  38. Helen Yannakoudakis, Ted Briscoe, and Ben Medlock. 2011. A new dataset and method for automatically grading ESOL texts. In Proceedings of the 49th Annual Meeting of the Association for Computational Linguistics: Human Language Technologies, pages 180–189, Portland, Oregon, USA. Association for Computational Linguistics.
  39. Yajie Ye and Weiwei Sun. 2020. Exact yet efficient graph parsing, bi-directional locality and the constructivist hypothesis. In Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics, pages 4100–4110, Online. Association for Computational Linguistics. ↗
  40. Wenxuan Zhang, Ruidan He, Haiyun Peng, Lidong Bing, and Wai Lam. 2021. Cross-lingual aspect-based sentiment analysis with aspect term code-switching. In Proceedings of the 2021 Conference on Empirical Methods in Natural Language Processing, pages 9220–9230, Online and Punta Cana, Dominican Republic. Association for Computational Linguistics. find
  41. Yuanyuan Zhao, Weiwei Sun, Junjie Cao, and Xiaojun Wan. 2020. Semantic parsing for English as a second language. In Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics, pages 6783–6794, Online. Association for Computational Linguistics. ↗

Updated

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/li-2025-compositional-syntactico-sembanking-english-as/