Croatian Corpus of Non‐Professional Written Language – Typical speakers and speakers with language disorders

Corpora, as annotated archives of human communication, are objective, reliable resources for language analysis. Here we present the corpus of non-professional written Croatian, based on 1-year sampling of writings by typical speakers and speakers with language disorders. This corpus provides a unique resource because it samples language used by non-professionals, in contrast to corpora based on texts by professional writers (such as journalists, scholars or novelists) sampled over more than a century. In addition, our corpus contains written language from typical and impaired speakers sampled under identical conditions, allowing detailed analyses of language use. This paper describes the language tasks (essay, story generation, non-formal and formal letter and dictation) used to elicit text production, and procedures for sampling and annotation used to generate the corpus. Its usefulness is illustrated through language productivity analyses of transcripts of different genres produced by writers of different age and language status. This corpus may prove useful for the analysis of writing skills in typical and language-impaired speakers of Croatian