Our mission

Language AI should work for everyone — including Zambians

Large language models are trained almost entirely on English and a handful of European languages. Bemba, Nyanja, Tonga, Lozi — the languages that 19 million Zambians speak at home, in markets, in churches — are effectively invisible to AI.

Afrobabel exists to change that. We are building the open infrastructure that language AI for Zambian languages needs: a structured, rights-cleared, culturally grounded training corpus — built with contributors, not extracted from them.

Every submission to an Afrobabel competition becomes part of the corpus. Every contributor retains full attribution. Every dataset record is released under CC-BY-SA-4.0 so researchers, developers, and future generations can build on it freely.

How it works

Three steps from story to training data.

01

Write

Contributors submit original stories, poems, folk tales, and essays in a Zambian language. Full metadata is captured — language, dialect, fluency level, cultural category.

02

Review

A panel of trained reviewers scores each submission for fluency, naturalness, cultural authenticity, and creativity. Reviewers flag anything that shouldn't enter the corpus.

03

Release

Reviewed submissions are packaged as a Hugging Face Dataset — structured, attributed, and openly licensed. Researchers can load the corpus in one line of Python.

Board & advisors

Afrobabel is governed by a multidisciplinary board of linguists, technologists, and community advocates.

Dr. Mulenga Kapwepwe

Board Chair

Zambian Language Council

Linguist and author with 20 years of work on Bemba literature and language preservation.

Prof. Charity Mwape

Academic Advisor

University of Zambia — Linguistics Dept.

Leads UNZA's Centre for African Language Studies. Specialises in computational approaches to Bantu languages.

Tendai Phiri

Technology Lead

Afrobabel Foundation

Software engineer and open-source advocate. Previously at iHub Nairobi and Mozilla Africa.

Sister Beatrice Nkonde

Community & Ethics

Churches Health Association of Zambia

Advocates for community-centred data governance. Ensures contributor rights are protected at every step.

Partners

UNESCO

Knowledge Partner

Lacuna Fund

Data Grant Recipient

Mozilla Foundation

Open Source Ally

UNZA Linguistics Dept.

Research Partner

Join the effort

Whether you're a writer, a linguist, a developer, or a researcher — there is a place for you in the Afrobabel community.

Create an account