USC researchers have developed an AI tool to help preserve and revitalize Owens Valley Paiute, an endangered language spoken by a small tribe in California. This pioneering approach shows promise for saving languages on the brink of extinction.
“My dad did not grow up speaking the language—like many families, it was forced out of use by boarding schools where speaking the language was forbidden.” —Jared Coleman, Ph.D.
Why it matters
Indigenous languages are at risk, with some estimates indicating up to 90% could vanish by 2100. Preserving them is vital for maintaining cultural diversity and identity. Ancestral languages connect people to their history, traditions, and worldviews.
- Around 40% of the world's 6,700 spoken languages are in danger of disappearing, and many of them belong to indigenous peoples.
- Some estimates predict that up to 95% of the world's languages may become extinct or seriously endangered by the end of the century.
Linguists call this a "mass language extinction", estimating that a language vanishes every two weeks. But the loss of indigenous languages is not only the loss of spoken practice but also the loss of cultural heritage passed down through generations
AI offers solutions for replicating immersive learning when few or no language resources exist, like Paiute, which was never codified or translated into English.
Harry Williams, an elder of the Bishop Paiute Tribe, observes the Owens Valley in California
The big picture
Coleman and his doctoral advisor Bhaskar Krishnamachari developed a specialized AI translation tool that simplifies complex English sentences—and uses placeholders for Paiute words not yet mapped. This structure mirrors how language learners naturally speak, supporting better retention.
What's next
Coleman plans to refine the AI tool and incorporate more Paiute history and oral narratives. This work also shows how AI can help reconstruct endangered languages globally, ensuring they're not lost forever.
“I’m lucky my great-grandparents sat down with linguists to document the language and to create recordings so I can hear their voices and words. And now, to listen to my great-grandfather and know what he is saying, there’s something very personally satisfying about that.” —Jared Coleman, Ph.D.
