An FST-driven annotation tool to analyze and process Gitxsanimx̱ words into their common interlinear annotation format, for Python.

This is a rule-based natural language processing tool that uses finite-state transducers (FSTs) to translate the word into its component morphemes and provide options for an interlinear gloss. Its strengths are in its very high accuracy rate (>95%), although it is limited to operating on words found in a central dictionary. This serves a secondary purpose of helping identify words in need of linguistic documentation. Implemented with Python and the foma CLI tool.

This project fed into collaborative machine-learning research initiatives at the University of British Columbia to support word annotation for endangered, low-resource languages.

Links

• View git_fst on GitHub
• 2021. Clarissa Forbes, Garrett Nicolai, and Miikka Silfverberg. An FST morphological analyzer for the Gitksan language. In Proceedings of the 18th SIGMORPHON Workshop on Computational Research in Phonetics, Phonology, and Morphology. 188-197.