A Finite-state morphological analyzer for Ge'ez verbs

Abstract

A morphological analyzer is a valuable and necessary component in many natural language processing applications such as machine translation, automatic dictionaries, spell checking, speech recognition etc. Morphological analyzer is specifically important for Semitic languages with high inflection and productive verbs. Ge’ez is one of the ancient Semitic languages in the world. It is currently the liturgical language of the Ethiopian Orthodox Tewahido Church. In this study, a finite-state morphological analyzer for the Ge’ez verbs was developed. The morphological analyzer was developed using the freely available finite-state tool Foma. The Ge’ez morphological analyzer was developed to analyze Ge’ez verbs into their root and feature tags. The analyzer also performs the generation of Ge’ez verbs from a given root and feature tags. The Ge’ez morphological analyzer was tested using manually annotated verbs by Ge’ez experts from the Ethiopic New Testaments Ge’ez Bible (books of Matthew, Luke, Mark and John) and Ge’ez prayer book - ውዳሴ ማርያም -‘wudase maryam’. The output of the Ge’ez morphological analyzer was compared with the manually annotated test data for accuracy. The result showed that the Ge’ez morphological analyzer analyzed the Ge’ez verbs with an accuracy of 97.29% and a precision of 80.24%. All in all, this research work achieved its objective by designing and implementing a Ge’ez verb morphological analyzer that performs both analysis and generation of Ge’ez verbs. The developed Ge’ez morphological analyzer will benefit the Ethiopian Orhodox Tewahido Church, interested Ge’ez language learners and the academic community that conduct researches in Ge’ez language.