Korpusteadust

Korpusteadust, corresponding to corpus linguistics, is a field within linguistics that studies language by analyzing large collections of authentic texts and spoken transcripts, known as corpora. It aims to reveal patterns of usage, frequency of forms, and the distribution of linguistic features across genres, registers, and time periods.

Core methods involve designing and building corpora with explicit sampling criteria, annotating texts with part-of-speech tags, lemmata, syntactic structures, and metadata, and applying computational tools to retrieve and quantify data. Analysts use concordancers, frequency lists, collocation and multi-word analysis, and statistical modeling to test hypotheses about language.

Historically, korpusteadust emerged with early computer-assisted text analysis in the mid-20th century and grew through large, annotated corpora such as the Brown Corpus, the British National Corpus, and contemporary corpora like COCA. Researchers continue to develop multilingual corpora, spoken corpora, and domain-specific collections to capture variation and change.

Applications span lexicography, language teaching, sociolinguistics, and natural language processing, including language model evaluation and search feature development. Limitations include sampling bias, representativeness, copyright and privacy concerns, and dependence on annotation quality. When used carefully, korpusteadust provides empirical evidence about how language is actually used.