Understanding civic language in context
Civic discourse is shaped by language, place, and lived experience. The Civic Corpus brings together language data, local terminology, and contextual knowledge to help AI systems interpret community expression beyond literal translation.
- CIVIC DISCOURSE
- TERMINOLOGY
- RESPONSIBLE DATA
The dimensions of civic discourse
From everyday concerns to public decision-making, civic discourse reflects the issues, institutions, and experiences that shape community life.
A resident describes a collapsed footbridge on the route to the market.
- category
- infrastructure
- asset
- footbridge
- status
- collapsed
- location
- parish-level, from report
- severity
- high
Transcribes, classifies, and structures the report.
Beyond literal translation
The terminology layer connects local expressions to shared civic concepts while preserving the linguistic and institutional context that gives them meaning.
Olukiiko→committee · council→local_government.body
Olukiiko lwawa ssentebe wa disitulikiti ennaku kkumi okuwaayo kalonda yenna akwata ku abo abaakosebwa amataba.
The committee gave the district chairperson ten days to submit the details of those affected by the floods. · Source: Sunbird SALT
A deliberative body. The same word covers a committee, a council, or an assembly; the surrounding speech decides which one is meant.
The data behind the models
Explore the language datasets used to develop and adapt CommonMind models, from parallel translation corpora to instruction-tuning resources.
- Rogendo · 58,052
- MAFAND-MT · 32,414
- Inkuba · 29,991
- SALT · 23,947
- Alpaca · 14,844
- Aya · 359
- Rogendo · 60,000
- MAFAND-MT · 32,414
- Inkuba · 30,000
- SALT · 23,947
Trust is part of the infrastructure
- Informed consentPeople understand how their contributions will be collected, used, and shared.
- Purpose limitationData is collected and used for clearly defined research and development purposes.
- Privacy protectionPersonal information is minimised and protected, with additional safeguards for sensitive contributions.
- Community validationLanguage speakers and local experts help review terminology, interpretation, and contextual meaning.