Kolokaatit

Kolokaatit is a term used in Finnish linguistics to denote collocates, i.e., words that frequently co-occur with a given target word within a text corpus. The term reflects the Finnish rendering of the English concept collocation and is employed in discussions of lexical association, semantic neighborhoods, and language modeling. Kolokaatit help researchers describe the surrounding semantic field of a word and contribute to tasks such as sense disambiguation and lexicographic description.

Identification of kolokaatit is typically done with corpus-based methods. A co-occurrence window is defined around the target word (for example, a span of a few words to several words). Statistical measures such as mutual information, t-score, likelihood ratio, or chi-squared tests are used to highlight strong kolokaatit, distinguishing them from high-frequency but weakly informative words. The resulting set often reflects domain, genre, andregister of the underlying corpus.

Examples illustrate sense-dependent variation. For the word bank, financial texts may yield kolokaatit like account, loan, interest, and credit, while geographic or natural-text corpora may show river, shore, slope, and sediment in proximity. Such differences demonstrate how kolokaatit encode meaning and usage patterns across contexts.

Applications of kolokaatit include lexicography, where collocational profiles inform dictionary entries; natural language processing and language modeling, where collocation features improve word sense disambiguation, translation, and text generation; and corpus linguistics, where kolokaatit help map semantic networks and lexical associations.

See also: collocation, corpus linguistics, distributional semantics, word sense disambiguation.