Elamur Research⌕
☰
home / papers / 2026 / The Lock, Stock, and Barrel of Marathi Multiwords

The Lock, Stock, and Barrel of Marathi Multiwords

Summary

Earlier annotation work on Marathi multiword expressions covered only compound nouns and light verb constructions. The authors build a broader resource of roughly 3000 annotated MWEs for Marathi within the PARSEME 2.0 initiative, covering both verbal and non-verbal types. The data comes from three sources: the Marathi UD Treebank, which has gold annotations, and the Anuvaad corpus and web-crawled children's stories, which were tagged automatically with UDPipe. The paper does not restate the PARSEME tagset. It focuses on how Marathi phenomena were mapped onto PARSEME categories, drawing on descriptive linguistic literature. Covered phenomena include multi-verb constructions, light verbs with nominal, adjectival and adverbial preverbs, causative light verbs, verbal idioms, compounds, reduplication and multiword terms. The authors propose syntactic tests for separating light from full verb uses. They also report recurring difficulties: gradient compositionality, discontinuous and overlapping expressions, and nested compounds. The main empirical finding is that LVCs are the most productive MWE class in the data. The resource is the first broad MWE inventory for Marathi and supports both parsing and lexicon research.

Key points

  • Corpus sizes (tokens): Marathi UD Treebank 3849, Anuvaad 27956, children's stories 4287.
  • Annotation quality: the two raw corpora carry only silver-standard UDPipe tags and sentence splits, which contain known errors. A single annotator did the work, so no inter-annotator agreement was computed.
  • Light verbs identified: 13 verbs act as light verbs in MVCs and 23 in LVC.full. LVC.cause is rare and appears with only four causative verbs.
  • Diagnostics for LVCs: beyond the rephrasing test, the authors check three properties:
    • the preverb cannot be dropped;
    • event-noun preverbs cannot be coordinated;
    • preverbs combine only with a restricted set of light verbs.
  • Design decisions under PARSEME 2.0:
    • named entities are excluded;
    • nested closed compounds are flattened into a single annotation;
    • noun-adjective compounds that yield adjectives are tagged AdjID, while adjective-noun compounds that yield nouns are tagged NID.

Our summary of the paper, in our own words — not its text. The work belongs to its authors; read it at the links above.

Fields: Linguistics & languages

Cite

The paper itself, not this page. BibTeX from the library's record of the work. DOI 10.18653/v1/2026.mwe-1.11

@inproceedings{padhye2026lock,
  title = {{The Lock, Stock, and Barrel of Marathi Multiwords}},
  author = {Padhye, Aakanksha and Vaidya, Ashwini},
  booktitle = {Proceedings of the 22nd Workshop on Multiword Expressions (MWE 2026)},
  year = {2026},
  doi = {10.18653/v1/2026.mwe-1.11},
  url = {https://doi.org/10.18653/v1/2026.mwe-1.11}
}

References

12 references, 0 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. Claire Bonial. 2021. Précis of take a look at this! form, function, and productivity of english light verb constructions. Colorado Research in Linguistics, 25. find
  2. Miriam Butt. 1995. The structure of complex predicates in Urdu. Center for the Study of Language (CSLI). find
  3. Mathieu Constant, Gülen Eryiit, Johanna Monti, Lonneke van der Plas, Carlos Ramisch, Michael Rosner, and Amalia Todirascu. 2017. Multiword expression processing: A survey. Computational Linguistics, 43(4):837–892. find
  4. Neiloufar Family. 2014. Semantic spaces of Persian light verbs: A constructionist account, volume 6. Brill. find
  5. Peter Hook and Prashant Pardeshi. 2009. A taxonomy of eat expressions in marathi. Annual Review of South Asian Languages and Linguistics, pages 41– 63. find
  6. Kanishka Jain and Ashwini Vaidya. 2024. Revisiting VMWEs in Hindi: Annotating layers of predication. In Proceedings of the Joint Workshop on Multiword Expressions and Universal Dependencies (MWE-UD) @ LREC-COLING 2024, pages 98–105, Torino, Italia. ELRA and ICCL. find
  7. Ioannis Korkontzelos. 2010. Unsupervised learning of multiword expressions. Unpublished. find
  8. Aaditya Kulkarni. 2019. Semantics of hit expressions in marathi. find
  9. Yusuke Kume. 2011. On the complement structures and grammaticalization of see as a light verb. The Electronic Library, 28:206–221. find
  10. Marja Nenonen, Jussi Niemi, and Matti Laine. 2002. Representation and processing of idioms: Evidence from aphasia. Journal of Neurolinguistics, 15(1):43– 58. find
  11. Renuka Ozarkar. 2014. Structures of Marathi verbs. Ph.D. thesis, Doctoral dissertation, University of Mumbai. find
  12. Rajeshwari V Pandharipande. 1998. Marathi. Routledge. find

This page in Markdown → https://md.research.elamur.ai/free/en/papers/padhye-2026-lock-stock-barrel-marathi-multiwords/