This research is made possible by the support of the Higher Education Quality Council of Ontario as part of the Consortium on Generative Artificial Intelligence (AI).
As an Indigenous institute that offers an array of Indigenous language programming and where Indigenous language revitalization is a core area of focus, Six Nations Polytechnic (SNP) is committed to the language acquisition journeys of our adult learners and exploring how adaptive learning technologies might lend themselves to their success.
Indigenous Language Revitalization - The Context of the Work
In the last thirty-five years, research on the process for reversing language shift or revitalizing Indigenous languages has highlighted the importance of understanding levels of endangerment and subsequently, opportunities for intervention (Fishman, 1991; Lewis & Simons, 2020). Additional research specific to the teaching of polysynthetic languages, and Hodinohsó:ni languages in particular, has further outlined what the goals of revitalization programs might be in order to create speakers and users of Onkwehowehnéha using comprehensive, scaffolded curricula and effective teaching methods (DeCaire, 2023; Green, 2017). This research also outlines commonly occurring areas of challenge that can emerge within the language learning classrooms. While not a conclusive list, these can include challenges with pronunciation, comprehension, keeping pace with the curriculum, and varying levels of fluency in the learners (Brant, 2025). These challenges offer important opportunities for communities, Indigenous institutes, instructors, and learners to consider the use of digital tools and technologies, including artificial intelligence, to create adaptive learning pathways that meet specific learner needs.
The Role of Adaptive Learning in Successful Language Programs
SNP’s 2017 research report, “Pathways to Creating Onkwehonwehnéha Speakers at Six Nations of The Grand River Territory,” outlines several pathways that language speakers at Six Nations have taken to learn their languages and advance their proficiency. These pathways include immersion programming, mentor-apprentice study, university or other school-based programs, and independent study. In outlining the number of language contact hours required to create highly fluent speakers (approximately 1,800-3,600), the report reaffirms the notion that language programs be guided by the goal of creating speakers who can use the language confidently and meaningfully in everyday life (Green, 2017, p. 43). Centering this goal creates space for the development of adaptive learning tools to support learners along their way, no matter which path they take. Focussing this development on the polysynthetic1 structure of Hodinohsó:ni languages and the areas of challenge learners often experience, provides opportunities to focus AI and adaptive learning tools in meaningful ways.
How are Indigenous peoples engaging with AI?
Through the environmental scan, we found that Indigenous peoples across the world are engaging with AI and digital technologies in innovative and intentional ways, and exploring how these tools might contribute in a meaningful way to language revitalization, community education and the safeguarding of cultural knowledge. The following section of our blog post offers a few examples of how project teams have established frameworks for the ethical development of AI tools, worked collaboratively with language and knowledge keepers, and remained attentive to responsibilities tied to Indigenous data. And while this environmental scan is by no means a comprehensive and total accounting of Indigenous-led AI initiatives, it does offer a starting point for future discussions with our learning community.
The 2020 Indigenous Protocol and AI Workshop Position Paper - An In-Depth Look
A range of projects illustrates how Indigenous peoples are using AI in their work. The 2020 Indigenous Protocol and AI Workshop position paper documents the early stages of development of the Hua Kiʻi project, an image-recognition tool that allows users to take a picture of an object and receive the word in the Indigenous language of their region. The project was developed during the Indigenous Protocol and AI Workshops as a proof-of-concept Indigenous language technology. The position paper describes the discussion leading to the team’s choice of project—to create a mobile tool that allows users to take a photo of an object and receive the corresponding word in a local Indigenous language, beginning with Hawaiian and Northern Cheyenne. The paper then follows along as the team works through the process of designing the system as one that is culturally rooted, an image-recognition and translation app that connects place, language, and community knowledge.
The project brought together Indigenous engineers, language experts, and designers who emphasized building AI in ways that align with Indigenous values, limit reliance on English, and ensure community guidance at every stage. The end goal was not only technical innovation but the demonstration of culturally grounded protocols in action—showing how AI can support language revitalization without displacing Indigenous authority.
Automatic Speech Recognition & Verb Generators
Other Indigenous-led innovations point in similarly interesting directions. The Kawennón:nis Word Generator, a National Research Council project whose research team includes Kanyen’keha language speakers, uses machine learning to help students generate and conjugate Mohawk words, offering structured support for a polysynthetic language, supported by speaker review and correction.
Gakti LoRA, a Sámi-led fine-tuned language model, explores how Indigenous communities can customize existing AI architecture to reflect their linguistic patterns and cultural contexts without giving up control of their data.
There has also been ongoing research and experimentation into the use of automatic speech recognition and speech-based foundation models for low-resource Indigenous languages, including Seneca, Mohawk, Māori, and more (Gupta et al, 2025; Lee, 2024). As these projects continue to experiment and fine-tune their models, there may be opportunities for language instructors, learners or other community collaborators to learn, train, and help guide and inform the development of technologies designed for their languages.
Overall, these types of projects and collaborations offer examples of how AI and digital technologies can be utilized for language preservation and revitalization efforts, particularly when communities are involved in the design, visioning and implementation process—and empowered to retain authority and exercise responsibility to the data.
Conclusion
This environmental scan set out to identify the challenges language instructors face and the learning barriers experienced by students, while also examining Indigenous-led AI initiatives that point to promising future directions. The projects reviewed offered not only technical possibilities but also important considerations on how to shape these tools in ways that uphold Indigenous data sovereignty, strengthen community capacity, and honour the responsibilities that come with language work. Issues of data security, the risk of cultural appropriation, and the safeguarding of Indigenous Knowledge will be further explored in our final culminating report and blog. This foundational work positions us to move into the research phase and begin engaging our learning community in deeper dialogue.
Notes
1Polysynthetic refers to languages that form long, morphologically rich words—especially verbs—by combining numerous meaningful units (morphemes) into single complex words.
References Cited:
Brant, R. J. (2025). Shakotirihonnyén: ni Karihonnyenníhtshera: Creating a Teaching Manual for Kanyen’kéha Adult Immersion Programming.
DeCaire, O. R. (2023). The role of adult immersion in Kanien’kéha revitalization (Doctoral dissertation, University of Hawai'i at Hilo).
Green, T. (2017). Pathways to Creating Onkwehonwehnéha Speakers at Six Nations of The Grand River Territory. Six Nations, Ontario.
Gupta, R. C. S., Kazantseva, A., Tessier, M., Lothian, D., Akwiratékha’Martin, E. J., Larkin, S., & Kuhn, R. (2025). Evaluating Speech Foundation Models for Automatic Speech Recognition in the Low-Resource Kanyen’kéha Language.
Kazantseva, A., Maracle, O. B., Maracle, R. T. J., & Pine, A. (2018). Kawennón: nis: the Wordmaker for Kanyen’kéha. In Proceedings of the workshop on computational modelling of polysynthetic languages (pp. 53-64).
Lee, A. (2024). Māori Speech AI Model Helps Preserve and Promote New Zealand Indigenous Language. NVIDIA Blog.
Lewis, Jason Edward, ed. (2020). Indigenous Protocol and Artificial Intelligence Position Paper. Honolulu, Hawaiʻi: The Initiative for Indigenous Futures and the Canadian Institute for Advanced Research (CIFAR).
Lewis, M. P., & Simons, G. F. (2010). Assessing endangerment: Expanding Fishman's GIDS. Revue Roumaine de Linguistique, 55(2), 103–120.
Senior Researcher: Sara General. Research Associate: Bobby Henry. Project Lead: Lisa Dietrich. The extended project team includes Heather Bomberry, Taylor Gibson, Tania Henry, Tanis Hill and Stevie Jonathan.