| Abstract |
The paper discusses relationships between the syntactic behaviour and meaning of selected verbs, with the focus on exploiting observable syntactic similarities for uncovering of semantic kinship. The investigation is inspired by the demand in language technology for large-scale lexicons that combine morphological, syntactic and semantic descriptions of the lemmas. The development of such a lexical resource is rather demanding, therefore, an enhancement of existing resources with additional information types is a worthwhile task. The computational lexicon for Danish SprogTeknologisk Ordbase (STO) comprises a comprehensive syntactic layer which is assumed to be suitable for enhancement with semantic information. The theoretical background for the current approach is the consensus on obvious relationships between a syntactic behaviour and a particular sense of lemmas, as a surface complementation structure reflects the underlying semantic argument structure. The idea is to test the feasibility of deriving semantic information systematically from the syntactic structures encoded in syntactic patterns. In the pilot project, a sub-set of trivalent verbs that share syntactic constructions are extracted from STO; the material consists of 216 verbs subcategorising for a direct object and a prepositional object covered by eight syntactic patterns. The examination takes a syntactically based grouping of these verbs as its starting point and focuses on defining lexical classes in terms of shared prevalent meaning components. These components form the basis of the semantic label assignment to the particular groups. The material provides 20 basic semantic groups, such as force, urge, judge, consider, remove, cheat, etc. that can be refined into sub-groups along further semantic features or generalized into classes-e.g. communicate-persuade, cause-change-of , according to different degrees of granularity required. The present classifications of the verbs are also examined in relation to Levin’s English verb classes (1993). Our findings suggest that it is feasible-though within recognized limits-to exploit systematically the formalised syntactic descriptions in meaning group prediction. |
| BibTex |
@InProceedings{ELX08-013,
author = {Anna Braasch},
title = {Syntactic Behaviour and Semantic Kinship of Selected Danish Verbs},
pages = {309-321},
booktitle = {Proceedings of the 13th EURALEX International Congress},
year = {2008},
month = {jul},
date = {15-19},
address = {Barcelona, Spain},
editor = {Elisenda Bernal, Janet DeCesaris},
publisher = {Institut Universitari de Linguistica Aplicada, Universitat Pompeu Fabra},
isbn = {978-84-96742-67-3},
} |
