The Digital Dictionary Database of Slovene(DDDS) has been designed as a central database with machine-readable data on Slovene. Its morphological base – the Sloleks Morphological Lexicon of Slovene – currently contains approximately 380,000 lexemes and their inflected and accentuated forms, together with pronunciations in the International Phonetic Alphabet (IPA) and its equivalent, X-SAMPA. However, except for individual manually added lexemes (e.g. žleht “evil”, pošlihtan “set in order”, zaštekati “to come to understand”), DDDS has not been systematically expanded with typically spoken vocabulary. In the paper, we present the process of extending DDDS with lexeme candidates automatically extracted from multiple language resources containing transcriptions of spoken Slovene. We describe these resources and the process of extracting and validating the candidates. Automatically tagged and lemmatized lexemes were manually corrected in terms of lemmas and morphosyntactic features. Additional information was added manually to assist further lexicographic work within DDDS. For each of the approximately 11,000 confirmed lexemes, we automatically generated their inflected and accentuated forms as well as their pronunciations using Pregibalnik, a tool developed for the automatic expansion of the morphological lexicon. We describe the challenges encountered when including typically spoken vocabulary in the database, as well as the solutions implemented. The process can be repeated when extracting vocabulary from future versions of corpora of spoken Slovene.
|