collocates

Collocates are words that tend to occur together with another word within a language corpus. In corpus linguistics, a collocation is a sequence of words that co-occurs more often than would be expected by chance, reflecting conventional patterns of usage. A word’s collocates form its immediate lexical environment and can influence meaning, tone, and interpretation. Collocation encompasses a range of associations, from strong to weaker, and includes cross-category pairings such as adjectives with nouns or verbs with objects.

To identify collocates, researchers analyze large text corpora and examine co-occurrence patterns around a target word. A typical approach uses a sliding window, often five to seven words to the left and right, to collect candidate collocates. Statistical measures such as mutual information, t-score, log-likelihood, or chi-squared tests assess whether observed co-occurrence is higher than expected. Some studies also incorporate syntactic relations to refine collocates, for example focusing on adjective-noun or verb-object pairings.

Examples help illustrate the concept. With coffee, common collocates include strong coffee and black coffee; with crime, collocates include commit a crime; with decision, make a decision. These illustrate noun-adjective, verb-noun, and other productive pairings. Collocations are not identical to idioms; many collocations are flexible and productive, whereas idioms convey figurative meanings not predictable from their parts.

Applications of collocation analysis include lexicography, language teaching, and natural language processing. Understanding collocations supports more natural language production, enhances dictionary entries, and improves models for language prediction, search, and automatic phrase extraction.