Search

Morphological Guesser as a Tool for Analyzing Field Data: Experiences with The Naukan Yupik Language // Tomsk Journal of Linguistics and Anthropology. 2025. Issue 2 (48). P. 9-19

The paper presents the development and evaluation of two automated morphological analysis tools for Naukan Yupik (Yupik Eskimo Eskimo-Aleut): a dictionary-based morphological analyzer and a dictionary-free morphological guesser. Both tools are implemented with a two-stage approach to morphological modeling based on finite state automata. The study examines in detail the morphological features of Naukan Yupik that influence the development of automated analysis tools, including rich inflection and derivation, homonymy of morphological markers, and complex morphophonological processes. The effectiveness of both tools will be evaluated using a corpus of oral texts from 2022–2023. Particular attention is paid to the problem of overgeneration in the output of the morphological guesser and to ways of solving this problem through part-of-speech-based analysis separation. The results show that when working with field data, the use of a guesser can be more effective despite its known limitations.

Keywords: Naukan Yupik, morphological analyzer, language documentation, natural language processing