AI and Endangered Languages: Helping Communities Keep Their Voices

Thousands of languages are spoken worldwide, and many are at risk of disappearing. AI offers tools for documentation and learning, but communities must lead.

• Speech recognition and transcription to document recordings of fluent speakers.
• Translation tools that bridge small languages and larger ones.
• Learning apps with pronunciation feedback and games.
• Dictionaries and corpora: organising words, phrases and stories.
• Text-to-speech so learners can hear words.
• Keyboards and fonts for scripts that lack digital support.
• Educational materials: children's stories and lessons.

• Little data: many languages have limited written text or recordings, so models perform poorly.
• Dialects and variation that tools may flatten.
• Ownership and consent: who controls recordings and data?
• Cultural sensitivity: some knowledge, songs or stories are restricted or sacred.
• Extraction: outsiders collecting data without benefit to the community.
• Quality and trust: poor translation can harm rather than help.

1. Community-led: speakers decide goals, priorities and access.
2. Informed consent and clear agreements on data use.
3. Data sovereignty: communities retain control.
4. Benefit sharing: training, jobs and tools that stay with the community.
5. Honesty about limits: tools support language revival, but don't replace speakers and teachers.
6. Long-term sustainability beyond a project's funding.

• Support community-run language programmes.
• If you speak a heritage language, record and share family stories with consent and care.
• Use trusted resources developed with speakers.

AI can support language documentation and learning. Communities, speakers and teachers keep languages alive.