Multilingual CPA: Linking Verb Patterns across Languages

Logo poskytovatele

Varování

Publikace nespadá pod Ekonomicko-správní fakultu, ale pod Fakultu informatiky. Oficiální stránka publikace je na webu muni.cz.
Autoři

BAISA Vít MOŽE Sara RENAU Irene

Rok publikování 2016
Druh Článek ve sborníku
Konference Proceedings of the XVII EURALEX International congress
Fakulta / Pracoviště MU

Fakulta informatiky

Citace
Obor Informatika
Klíčová slova Corpus Pattern Analysis; corpus lexicography; multilingual resources; verb patterns
Popis This paper presents the results of a pilot study in linking corresponding English and Spanish verb patterns using both automatic and manual procedures. Our work is rooted in Corpus Pattern Analysis (CPA) (Hanks 2004, 2013), a corpus-driven technique that was used in the creation of existing monolingual pattern dictionaries of English and Spanish verbs, which were used in our experiment to design a gold standard of manually annotated verb pattern pairs. Research in CPA has inspired parallel projects in English, Spanish, Italian and German. Our study represents the first attempt to build a multilingual lexical resource by linking verb patterns in these languages. Verb have special difficulties related to grammar and argument structure that we do not find in other parts-of-speech, and for that reason we think that it is necessary to create a specific resource for them. After applying the automatic matching to a set of 87 Spanish verbs linked to 176 English verbs, an evaluation of a random selection of 50 of these pairs show 80% precision
Související projekty:

Používáte starou verzi internetového prohlížeče. Doporučujeme aktualizovat Váš prohlížeč na nejnovější verzi.