Abstract:
The paper focuses on the construction of valence lexicons for Bulgarian. It presents two approaches: a corpus-based one, which uses a syntactically annotated corpus for Bulgarian, and a grammar-based one, in which the resource is created together with the grammar language rules. BulTreeBank was used as a syntactically annotated corpus. The HPSG-oriented Bulgarian Resource Grammar was used for the grammar-based valency lexicon. Also, the interaction between the two approaches is discussed.
