PBAG NLP Research

African languages, for Africans, by Africans

Our stories have always lived in our languages. PBAG now brings that same conviction to technology — researching and building language tools for Gĩkũyũ, Kiswahili and Sheng, with the communities who speak them.

Join the community Our research
Why it matters

A continent of languages — missing from the machines

Roughly 2,000 of the world's languages — close to a third of them — are African. Yet our languages barely register in the technology the world now runs on.

Colonialism pushed African languages out of schools, offices and print, and that neglect followed them into the digital age. The result is technology that stumbles over our names, misreads our cultures, mislabels our places and forgets our history. Voice assistants do not hear a Gĩkũyũ grandmother; translation tools flatten Sheng; maps misspell the villages on our own ridges.

PBAG grew up telling stories in the languages people actually speak — Ngaahika Ndeenda was performed in Gĩkũyũ at Kamĩrĩĩthũ for exactly that reason. Research is how we carry that work into technology.

Our goal

We want Africans to shape — and own — the language technologies that are coming, and to steer them towards human dignity, well-being and fairness. We get there by building community together, by doing research in the open with the people it serves, and by bringing artists, linguists, technologists and elders to the same table.

  • Community-built — speakers lead, contribute and benefit.
  • Open & participatory — methods, findings and (with consent) data shared openly.
  • Multidisciplinary — theatre, linguistics, history, computing and GIS together.
  • Rights-respecting — consent, credit and community ownership under Kenya's Data Protection Act.
How we work

Data archaeology

Good language technology starts with good data — and for African languages, much of it is buried in archives, memories and performances. We dig it up carefully, and we create new data with communities rather than about them.

Excavating the record

Finding and digitising Gĩkũyũ and Kiswahili text in colonial files, church and school publications, newspapers and family papers — the same digging behind Ithaka cia Kamĩrĩĩthũ.

Voices from the stage

With performers' consent, turning scripts, rehearsals and live shows into speech and text — natural, expressive language that reading corpora miss.

Inclusive gathering

Story circles, elders' sessions and youth Sheng workshops, designed so that women, rural speakers, elders and young people are all heard — so the data truly represents the culture.

Names & places

A community gazetteer of Kiambu place names, clans and landmarks, linked to the Heritage Map — so technology spells and says our places right.

Testing the models

Evaluating how today's AI translation, speech and chat models handle Gĩkũyũ, Kiswahili and Sheng — and publishing where they fail, so they can improve.

Skills & ownership

Training young Kenyans in data collection, annotation and AI so the expertise — and the value — stays in the community.

Research agenda

What we're building

Planned

Gĩkũyũ theatre speech corpus

Consented recordings and transcripts from PBAG productions and story circles.

Planned

Sheng–Kiswahili–English parallel text

Comedy and youth-content sentences translated and verified by speakers.

Planned

Kiambu names & places gazetteer

Place names, clans and historic sites with pronunciations and stories.

In progress

Bilingual website & assistant

This site in English and Kiswahili, with Ask PBAG answering in the language you write in.

Coming soon

A community of practice & the PBAG iLab

A home for researchers, artists, students and speakers working on African languages — meet-ups, shared datasets, mentorship, and an innovation lab in Limuru to prototype tools with the community.

For research partnerships and consultancy, contact PBAG on info@pbagco.com or WhatsApp 0720 960 180.