Hungarian Corpus of Light Verb Constructions

Proceedings of the 23rd International Conference on Computational Linguistics (Coling 2010), pages 1110–1118, Beijing, August 2010 Hungarian Corpus of Light Verb Constructions Veronika Vincze University of Szeged Department of Informatics vinczev@inf.u-szeged.hu J´anos Csirik Hungarian Academy of Sciences Research Group on Artificial Intelligence csirik@inf.u-szeged.hu Abstract The precise identification of light verb constructions is crucial for the successful functioning of several NLP applications. In order to facilitate the development of an algorithm that is capable of recogniz- ing them, a manually annotated corpus of light verb constructions has been built for Hungarian. Basic annotation guidelines and statistical data on the corpus are also presented in the paper. It is also shown how applications in the fields of machine translation and information extraction can make use of such a corpus and an algo- rithm. 1 Introduction In this paper, we report a corpus containing light verb constructions in Hungarian. These expres- sions are neither productive nor idiomatic and their meaning is not totally compositional (the noun is usually taken in one of its literal senses but the verb usually loses its original sense to some extent), as it can be seen in the examples from dif- ferent languages shown below. Since their mean- ing is the same, only literal translations are pro- vided: • English: to give a lecture, to come into bloom, the problem lies (in) • German: halten eine Vorlesung to hold a pre- sentation, in Bl¨ute stehen in bloom to stand, das Problem liegt (in) the problem lies (in) • French: faire une pr´esentation to make a pre- sentation, ˆetre en fleur to be in bloom, le probl`eme r´eside (dans) the problem resides (in) • Hungarian: el˝oad´ast tart presentation- ACC holds, vir´agba borul bloom-ILL falls, probl´ema rejlik (vmiben) problem hides (in sg) Several terms like complex verb structures, sup- port verb constructions or light verb constructions have been used1 for these constructions in the lit- erature (Langer, 2004). In this paper, the term light verb constructions will be employed. The structure of the paper is as follows. First, the importance of the special NLP treatment of light verb constructions is emphasized in section 2. The precise identification of such constructions is crucial for the successful functioning of NLP applications, thus, it is argued that an algorithm is needed to automatically recognize them (sec- tion 4). In order to facilitate the development of such an algorithm, a corpus of light verb construc- tions has been built for Hungarian, which is pre- sented together with statistical data in section 5. Finally, it is shown how NLP applications in the fields of machine translation and information ex- traction can profit from the implementation of an algorithm capable of identifying light verb con- structions (section 6). 2 Light verb constructions in NLP In natural language processing, one of the most challenging tasks is the proper treatment of col- 1There might be slight theoretical differences in the usage of these terms – e.g. semantically empty support verbs are called light verbs in e.g. Meyers et …

Content truncated to 500 words