African Language Tokenization Workshop

​African languages remain significantly disadvantaged by AI systems designed primarily around high-resource languages. One of the overlooked layers driving this challenge is tokenization — the process through which language is broken down into the tokens that AI models use to understand and generate text.

​Join Qhala for a half-day, co-creation workshop on the sidelines of UNGA 2026, bringing together researchers, linguists, AI practitioners, infrastructure builders, policymakers, funders and industry stakeholders to explore what it would take to build an open, community-oriented African Language Tokenizer.

​This will be a working session, not a traditional panel. Together, we will:

• ​Build a shared understanding of the opportunities and challenges in African language tokenization.

• ​Identify and prioritise languages and language families for further exploration.

• ​Explore key technical, data, governance and sustainability considerations.

• ​Identify opportunities for collaboration across research, engineering, infrastructure and funding.

• ​Agree on priority next steps for advancing the initiative beyond the workshop.

​The goal is to move from fragmented efforts and research to greater coordination and collective action around African language infrastructure.

​Who should attend?
Researchers, NLP and AI practitioners, linguists, technology and infrastructure leaders, policymakers, funders, investors, open-source communities and organisations working on African language technology.