Linguistic features for Hindi light verb construction identification
4490Authors
nan
Abstract
AbstractLight verb constructions (LVC) in Hindi are highly productive. If we can distinguish a case such as nirnay lenaa ‘decision take; decide’ from an ordinary verb-argument combination kaagaz lenaa ‘paper take; take (a) paper’,it has been shown to aid NLP applications such as parsing (Begum et al., 2011) and machine translation (Pal et al., 2011). In this paper, we propose an LVC identification system using language specific features for Hindi which shows an improvement over previous work(Begum et al., 2011). To build our system, we carry out a linguistic analysis of Hindi LVCs using Hindi Treebank annotations and propose two new features that are aimed at capturing the diversity of Hindi LVCs in the corpus. We find that our model performs robustly across a diverse range of LVCs and our results underscore the importance of semantic features, which is in keeping with the findings for English. Our error analysis also demonstrates that our classifier can be used to further refine LVC annotations in the Hindi Treebank and make them more consistent across the board.
Keywords
Full Content
Proceedings of COLING 2016, the 26th International Conference on Computational Linguistics: Technical Papers, pages 1320–1329, Osaka, Japan, December 11-17 2016. Linguistic features for Hindi light verb construction identification Ashwini Vaidya IIT Delhi ird11278@ee.iitd.ac.in Sumeet Agarwal IIT Delhi sumeet@iitd.ac.in Martha Palmer University of Colorado, Boulder martha.palmer@colorado.edu Abstract Light verb constructions (LVC) in Hindi are highly productive. If we can distinguish a case such as nirnay lenaa ‘decision take; decide’ from an ordinary verb-argument combination kaagaz lenaa ‘paper take; take (a) paper’, it has been shown to aid NLP applications such as parsing (Begum et al., 2011) and machine translation (Pal et al., 2011). In this paper, we propose an LVC identification system using language specific features for Hindi which shows an improvement over previous work (Begum et al., 2011). To build our system, we carry out a linguistic analysis of Hindi LVCs using Hindi Treebank annotations and propose two new features that are aimed at capturing the diversity of Hindi LVCs in the corpus. We find that our model performs robustly across a diverse range of LVCs and our results underscore the importance of semantic features, which is in keeping with the findings for English. Our error analysis also demonstrates that our classifier can be used to further refine LVC annotations in the Hindi Treebank and make them more consistent across the board. 1 Introduction Light verb constructions (LVC) are found across languages e.g. Japanese, Korean, Persian as well as English. An LVC consists of a predicating element (usually a noun) and a verb, which is also known as a light verb. For instance, take a walk or give a sigh are LVCs consisting of light verbs take and give and their corresponding predicating nouns walk and sigh. The nouns in an LVC contribute to the event semantics and the light verb supplies additional meaning e.g. agentivity, completeness, or permis- sion. In Hindi, LVCs are productive and are also sometimes termed as ‘support verb’ or ‘conjunct verb’ constructions. Examples 1 and 2 contrast the use of a simple predicate de ‘give’ with its light verb usage. (1) Simple predicate raam=ne Ram.M.Sg=Erg mohan=ko Mohan.M.Sg=Dat kit”ab book.F.Sg d”-ii give-Perf.F.Sg ‘Ram gave Mohan a book’ (2) Noun-Verb complex predicate raam=ne Ram.M.Sg=Erg us that baat”=par topic=loc zor pressure.M.Sg d”i-yaa give-Perf.M.Sg ‘Ram put an emphasis on that topic’ LVCs form a large part of the lexicon in Hindi. In the Hindi treebank (Palmer et al., 2009) (400,000 words), there are nearly 47,163 predicates, of which 37% have been annotated as LVCs. LVCs consist of predicate types that are far more numerous than simple verbs. Hindi has approximately 700 simple verbs, but potentially many more unique LVCs. This makes them a highly productive phenomenon in Hindi. This work is licensed under a Creative Commons Attribution 4.0 International Licence. Licence details: http:// creativecommons.org/licenses/by/4.0/ 1320 Butt (2010) notes that light verbs in Hindi LVCs act as a verbalizers in order to create new predicates and incorporate borrowed items into the language e.g. email kar ‘email do; email’. Therefore, LVCs …
Content truncated to 500 words
Labeled Text
AbstractLight verb constructions (LVC) in Hindi are highly productive. If we can distinguish a case such as nirnay lenaa ‘decision take; decide’ from an ordinary verb-argument combination kaagaz lenaa ‘paper take; take (a) paper’,it has been shown to aid NLP applications such as parsing (Begum et al., 2011) and machine translation (Pal et al., 2011). In this paper, we propose an LVC identification system using language specific features for Hindi which shows an improvement over previous work(Begum et al., 2011). To build our system, we carry out a linguistic analysis of Hindi LVCs using Hindi Treebank annotations and propose two new features that are aimed at capturing the diversity of Hindi LVCs in the corpus. We find that our model performs robustly across a diverse range of LVCs and our results underscore the importance of semantic features, which is in keeping with the findings for English. Our error analysis also demonstrates that our classifier can be used to further refine LVC annotations in the Hindi Treebank and make them more consistent across the board.
Publication Details
- Paper ID:
- 4490
- Year:
- 2016
- Publication:
- TIPSTER
- Publication Type:
- conf
- DOI:
- nan