Indigenous language technology in the age of machine learning

Most modern language technology for proofing tools, machine translation and other is based on machine learning. However, very few Indigenous languages the necessary amount of texts for making tools based on this technology. When language technology is based on large language models (LLMs), it bears the risk most of Indigenous language online text being produced by neural text. The result would be that online texts cannot be trusted as a source for Indigenous languages anymore. An alternative is the work done at UiT – Arctic University of Norway during the last 20 years, based on linguistics. Sámi tools have been made available for both industry and language communities, open licenses. These have been widely used by translators, teachers and various companies. The article analyzes the following four parts of language development: language data, language tool development, making the tools to users, and ethical use of available language technology tools. We make use of the CARE principles, and discuss the shortcomings of existing software data licensing schemes. Finally, we introduce a 3D table to help classify language projects with respect to their suitability for Indigenous languages.