sihtkeel

Sihtkeel is a constructed language developed for linguistic typology research and for exploring how semantic roles interact with morphosyntax in natural language processing. It was designed to be typologically diverse enough to test translation models, yet regular enough to be approachable for learners and demonstrators in academic settings. The project treats sihtkeel as a testbed rather than a living language with a community of speakers.

Etymology and classification suggest a name built from Estonian roots, with siht meaning target or aim and keel meaning language. The designation signals the language’s role as a deliberate instrument for examining how grammatical systems encode meaning and how learning algorithms handle cross-linguistic transfer.

Phonology and script: Sihtkeel features a compact phoneme inventory that includes five vowel phonemes and about eighteen consonants. The language allows simple syllable structures such as CV and CVC. It is written in a Latin-based alphabet with diacritics indicating vowel quality where necessary. Stress is phonemic and typically falls on the penultimate syllable.

Grammar and syntax: Morphology is predominantly agglutinative, with nouns marked for nominative, accusative, dative, and genitive cases via suffixes. Verbs encode tense and aspect through affixes, while evidentiality is expressed through mood particles. The default word order is SVO, but the language permits flexible arrangement to highlight information structure, aided by postpositions rather than prepositions.

Usage and status: Sihtkeel has been employed in demonstration corpora, teaching materials, and software experiments to illustrate typological variation and machine translation challenges. It remains a constructed, scholarly tool rather than a language with a speaking community.