Charles Explorer logo
🇨🇿

Co-reference in Czech: Information structure in the Prague Dependency Treebank

Publikace

Tento text není v aktuálním jazyce dostupný. Zobrazuje se verze "en".Abstrakt

My conference contribution intends to present my research of the influence of information and constituent structure to the co-reference chaining in Czech. Czech, as a language with free word order, offers many possibilities how to express the meaning with respect to the information structure.

The aim of my research was to show how the information structure affects the interpretation of the co-reference chaining. It was necessary to determine how are the co-reference chains shaped in authentic texts.

Prague Dependency Treebank (PDT) has been investigated instead of larger Czech national corpora (CNC) because it is syntactically and co-referentially annotated, while the CNC is not. In PDT the number of co-referential edges were measured under the several conditions (two immediately adjacent sentences, two noun phrases, no preference for one of them to be an antecedent of the pronoun etc.).

Variables considered were the information (topic and focus) and the constituent structure (subject and object).