Corpus-based collocation research targeted at Japanese language learners

Abstract

This paper discusses corpus-based research on collocations, introduces various tools for querying and extracting Japanese collocations and presents an analysis of Japanese collocations using language corpora and related tools. First, major corpus query tools such as Sketch Engine, NINJAL-NLP, Natsume, Chunagon, which can be used by learners and teachers of Japanese language, are briefly described. Focus then shifts to adjectival and nominal collocates and the resource "Collocation data of adjectives and nouns" which consists of adjective headwords and their nominal collocates extracted from two large corpora, BCCWJ and JpTenTen: 500 adjectives and 9,218 collocate nouns, and 500 adjectives and 23,220 collocate nouns from each corpus respectively. Finally, it is shown that corpus-based resources can be used in the creation of reference materials for learners of the Japanese language. The benefits of empirical research into collocations are also shown by comparing the obtained results with collocations in textbooks for Japanese as foreign language.