Izvorni znanstveni članak

Morphosyntactic annotation in the Croatian Wordnet

Krešimir Šojat ; Filozofski fakultet u Zagrebu

Sažetak

The paper deals with valence theory and its application to Croatian verbs. In the first part of
the paper we briefly present the theoretical postulates upon which our description of verb valency
is based. Thereby we mainly rely on frame semantics and its assumption that valency can be stated
and described for the whole semantic frame, i. e. the same semantic valency is shared by all
the members of the frame. We extend this assumption to semantic fields in the Croatian WordNet
and treat them as frames with internal structure in terms of semantic relations, such as synonymy
and hyponymy. We use this hypothesis as a theoretical basis in the construction of verb frames.
We want to include verb frames in the Croatian WordNet in order to broaden its applicability in
various NLP tasks, such as machine or machine aided translation, rule based parsing, data mining
etc. In the second part of the paper we discuss some of the existing works dealing with Croatian
verbs and valency description. In this discussion we focuss on the descriptive inventory for the so
called syntactic valency of Croatian verbs. Particular attention is paid to some controversis in discussed
models, namely to the so called infinitive and verbative complement. Although we do not
provide a definitive solution regarding these issues, by giving the counterexamples we try to exemplify
the causes of possible misinterpretaions. In the final part of the paper we present our own
inventory for the morphosyntactic description and annotation of verbal complements used in this
phase of the of verb frame construction.

Ključne riječi

Croatian WordNet; semantic field; computational lexicon; verb valency; morphosyntactic classes; Croatian language

Hrčak ID:

47110

URI

https://hrcak.srce.hr/47110

Datum izdavanja:

16.12.2009.

Podaci na drugim jezicima: hrvatski

Posjeta: 2.803 *