korpuse

Korpuse is the genitive singular form of the Estonian noun korpus, meaning a body or collection. In linguistics and language research, a korpus (plural korpused) refers to a structured collection of language data, typically consisting of texts and, in some cases, transcribed speech, assembled for linguistic analysis. A korpuse is characterized by its size, types of texts, time span, genres, and annotation level.

Corpora can be monolingual or multilingual, and may be general-purpose or domain-specific. They are often annotated with metadata such as authorship, publication date, domain, and source. Many korpused are further annotated with linguistic information, including parts of speech, lemmas, and syntactic structures, which enables both qualitative insight and quantitative analysis, as well as training and evaluation for natural language processing tools.

Constructing a korpuse involves several stages: defining the research goal, selecting and collecting material, cleaning and normalizing text, applying annotations, and establishing licensing and access terms. Quality control and documentation are important to ensure usefulness and reproducibility. Access to korpused ranges from open public corpora to restricted institutional datasets.

In scholarly and educational contexts, korpused support a wide range of activities, such as frequency studies, lexical development, grammar descriptions, and language technology development. They are foundational resources for corpus linguistics, lexical resources, and language-teaching materials, contributing to empirical insight into language use and variation within a language community.