Call for papers: SOAS GSMA Workshop - Exploring African language models: Bringing together African linguistics and AI research
There has been tremendous progress in the development of large language models (LLMs) and artificial intelligence (AI) applications over the last decades. However, the majority of work has focussed on a small number of well-described, mainly European languages. Yet in order to fully harness the advantages of the AI revolution, no language should be left behind – including African languages.
Africa is home to about 2,000 languages, ranging from major world languages like Swahili to many underdescribed community languages. In this workshop we explore aspects of LLM and AI for and based on African languages, bringing together AI researchers, linguists, translators, computer scientists and practitioners to exchange experiences and views for a mutually enriching dialogue, leading to a fuller understanding of the current state of the art, conceptual and practical challenges, and theoretical and methodological underpinnings of African languages and AI.
The workshop, to be held on 8th and 9th July 2026, is organised jointly by SOAS and GSMA.
Call for papers
We invite abstracts for 20-minute presentations addressing topics related to African languages and AI, addressing questions such as:
- Can linguists highlight challenges that make it hard for AI researchers to encode meaning in models?
- What are problems related to misinformation through AI and how can we build appropriate benchmarks for African languages?
- How to address multilingualism, code-switching and translanguaging in language models for African languages?
- How can language models for African languages address issues related to register and style, taboo, and offensive language?
- How can AI be harnessed for all African languages (including e.g. languages of wider communication, community languages and endangered languages)?
- What are important issues related to research ethics and research integrity in the field?
Abstracts should be no longer than one page, with optionally one additional page for data or references. Authors may submit at most two abstracts, only one which can be single-authored. Abstracts should be submitted by May 10th to cas@soas.ac.uk
Links:
Image: Ida Hadjvayanis
