Discuss the role of Tribal Research Institutes in evidence-based tribal policy, with reference to recent language-technology initiatives.

Tribal Research Institutes (TRIs) are State-level bodies, and 29 of them receive financial help from the Ministry of Tribal Affairs [1]. They are the knowledge arm of tribal administration. Their role is now growing from writing studies to producing and validating data, as the AI platform Adi Vaani shows.

Role of TRIs in evidence-based tribal policy

• Think tanks for States: they advise State tribal departments. Because the Centre funds them and the States run them, they form a Centre–State link [1].
• Grassroots evidence: they have access to communities in the field. Native speakers contribute data through TRIs, and experts check it before use [1].
• Documentation: the Working Group on Tribal Languages proposed community-led documentation of oral traditions through TRI–university partnerships [2].

TRIs in language-technology initiatives

• Co-developers: Adi Vaani was built by an IIT Delhi-led consortium with BITS Pilani, IIIT Hyderabad, IIIT Nava Raipur and the TRIs of Jharkhand, Odisha, MP, Chhattisgarh and Meghalaya [3].
• Data suppliers: TRIs supplied about 13.9 lakh sentences and 300 hours of speech [1]. Nine languages now have text translation, and Adi Vaani is proposed as the National Tribal Language Repository [2].
• Pace-setters: next-phase languages such as Katkari, Koya and Kokborok depend on coordination with State TRIs [4].
• Policy use: primers and dictionaries support NEP 2020's mother-tongue instruction. Tribal-language content also helps health outreach under the Sickle Cell Anaemia Elimination Mission [1].

Limitations

• Speech gap: 300 hours collected against a target of about 1,300 hours (100 hours × 13 languages) [1]. This hurts most the oral users the platform is meant for.
• Outputs, not outcomes: Lok Sabha replies listed only four languages live in the app and gave no usage figures [4].
• Consent: a TRI is a government body, so consent given through a TRI is not the same as community consent. Choosing one dialect as the standard can push out the others.
• Duplication: MeitY's Bhashini (Bhasha Daan) collects language data separately [5].

TRIs have grown from study centres into the main producers and validators of data for tribal-language AI. To make policy truly evidence-based:

• publish data and user figures for each language;
• set speech targets that record elders first;
• adopt shared standards with Bhashini [5];
• write community consent rules before the repository opens.

Strengthened in this way, TRIs can turn Article 46's promise of special care for STs into measurable results.