Annotating Second Language in Universal Dependencies: a Review of Current Practices and Directions for Harmonized Guidelines

Summary

UD is increasingly used for learner (L2) texts, but grammatical errors, calques, code-switching and other interlanguage phenomena fit poorly into guidelines written for standard language. Preparing a UD treebank from the Swedish SweLL-gold corpus, the authors review UD's universal rules for errors and foreign material and compare the choices actually made in existing L2 treebanks for several target languages. They pay special attention to parallel "L1-L2" formats in which each learner sentence comes with a correction hypothesis. The central dilemma is how far analysis should follow the actual forms versus the writer's probable intent. They recommend literal token-level annotation combined with syntactic analysis informed by the correction, propose a new dependency subtype marking deliberate violations of validation rules caused by L2 phenomena, and suggest treating syntactic calques similarly to code-switching. The proposals are meant as a basis for harmonised guidelines, also useful for spoken, user-generated and dialectal data.

How Mova uses it

• Regular endings on irregular verbs: goed, buyed — source of the rule en.errors.overregularized-typo: over-regularised forms (goed, buyed) are kept as written, with Typo=Yes, the lemma and tag of the intended word and the correct form in MISC; the rule flags such forms without Typo=Yes.
• Confused words: there/their, to/too, then/than, where/were — source for five rules on homophone confusions (en.errors.there-as-their, en.errors.their-as-there, en.errors.to-as-too, en.errors.then-as-than, en.errors.where-as-were), all relying on the UD convention for errors summarised in the paper.
• Text with errors is annotated literally — the overview of L2 treebanks and UD error rules underpins Mova's principle that erroneous text is annotated literally, so agreement and form rules double as grammatical-error detectors.

Effectiveness in Mova

The paper is a direct source for six warning rules. Their gold-data checks are recorded in the seed notes: on EWT 2.18 the over-regularisation rule finds 0 such forms (no false alarms) and the spoken L2 treebank ESLSpok has 2; the homophone rules show 0 violations on EWT 2.18 (e.g. there/PRP$ 15/0, to/ADV 12/0). These are rule-consistency checks, not a gain over a baseline; the contribution of the paper is not measured separately.

👨‍🔬💥

Topics: Universal Dependencies · Treebanks

Cite

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

@inproceedings{masciolini2025annotating,
  title = {{Annotating Second Language in Universal Dependencies: a Review of Current Practices and Directions for Harmonized Guidelines}},
  author = {Masciolini, Arianna and Berdicevskis, Aleksandrs and Szawerna, Maria Irena and Volodina, Elena},
  booktitle = {Proceedings of the Eighth Workshop on Universal Dependencies (UDW, SyntaxFest 2025)},
  year = {2025},
  url = {https://aclanthology.org/2025.udw-1.17/}
}

References

16 references, 3 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

1. John Lee, Keying Li, and Herman Leung. 2017b. L1-L2 parallel dependency treebank as learner corpus. In Proceedings of the 15th International Conference on Parsing Technologies, pages 44–49, Pisa, Italy. Association for Computational Linguistics. find
2. Arianna Masciolini. 2023. A query engine for L1-L2 parallel dependency treebanks. In Proceedings of the 24th Nordic Conference on Computational Linguistics (NoDaLiDa), pages 574–587, Tórshavn, Faroe Islands. University of Tartu Library. find
3. Arianna Masciolini and Márton A Tóth. 2024. STUnD: ett Sökverktyg för Tvåspråkiga Universal Dependencies-trädbanker. In Proceedings of the Huminfra Conference, pages 95–109, Gothenburg, Sweden. find
4. Alla Rozovskaya. 2024. Universal Dependencies for learner Russian. In Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING 2024), pages 17112–17119, Torino, Italia. ELRA and ICCL. find
5. 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… find our card
6. Marie-Catherine de Marneffe, Christopher D. Manning, Joakim Nivre, and Daniel Zeman. 2021. Universal Dependencies. Computational Linguistics, 47(2):255–308. find our card
7. Elisa Di Nuovo, Cristina Bosco, Alessandro Mazzei, and Manuela Sanguinetti. 2019. Towards an Italian learner treebank in Universal Dependencies. In 6th Italian Conference on Computational Linguistics, CLiC-it 2019, volume 2481, pages 1–6. CEUR-WS. find
8. Elisa Di Nuovo, Manuela Sanguinetti, Alessandro Mazzei, Elisa Corino, and Cristina Bosco. 2022. VALICO-UD: Treebanking an Italian learner corpus in Universal Dependencies. IJCoL. Italian Journal of Computational Linguistics, 8(8-1). find
9. Kaja Dobrovoljc. 2022. Spoken language treebanks in Universal Dependencies: an overview. In Proceedings of the Thirteenth Language Resources and Evaluation Conference, pages 1798–1806, Marseille, France. European Language Resources Association. find
10. Kristopher Kyle, Masaki Eguchi, Aaron Miller, and Theodore Sither. 2022. A dependency treebank of spoken second language English. In Proceedings of the 17th Workshop on Innovative Use of NLP for Building Educational Applications (BEA 2022), pages 39–45, Seattle, Washington. Association for Computati… find our card
11. John Lee, Herman Leung, and Keying Li. 2017a. Towards Universal Dependencies for learner Chinese. In Proceedings of the NoDaLiDa 2017 Workshop on Universal Dependencies (UDW 2017), pages 67–71, Gothenburg, Sweden. Association for Computational Linguistics. find
12. Manuela Sanguinetti, Cristina Bosco, Lauren Cassidy, Özlem Çetinoğlu, Alessandra Teresa Cignarella, Teresa Lynn, Ines Rehbein, Josef Ruppenhofer, Djamé Seddah, and Amir Zeldes. 2020. Treebanking user-generated content: A proposal for a unified representation in Universal Dependencies. In Proceeding… find
13. Manuela Sanguinetti, Cristina Bosco, Lauren Cassidy, Özlem Çetinoğlu, Alessandra Teresa Cignarella, Teresa Lynn, Ines Rehbein, Josef Ruppenhofer, Djamé Seddah, and Amir Zeldes. 2023. Treebanking user-generated content: a UD based overview of guidelines, corpora and unified recommendations. Language… find
14. Hakyung Sung and Gyu-Ho Shin. 2024. Constructing a dependency treebank for second language learners of Korean. In Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING 2024), pages 3747–3758, Torino, Italia. ELRA and ICCL. find
15. Hakyung Sung and Gyu-Ho Shin. 2025. Second language Korean Universal Dependency treebank v1.2: Focus on data augmentation and annotation scheme refinement. In Proceedings of the Third Workshop on Resources and Representations for Under-Resourced Languages and Domains (RESOURCEFUL-2025), pages 13–19,… find
16. Elena Volodina, Lena Granstedt, Arild Matsson, Beáta Megyesi, Ildikó Pilán, Julia Prentice, Dan Rosén, Lisa Rudebeck, Carl-Johan Schenström, Gunlög Sundberg, and 1 others. 2019. The SweLL language learner corpus: From design to annotation. Northern European Journal of Language Technology, 6:67– 104. find

This page in Markdown → https://md.mova.elamur.ai/free/en/docs/papers/masciolini-2025-annotating-second-language-universal-dependencies/