Workshop Explores the Fundamentals of AI in Advancing Indigenous Languages

BY: .

A two-day programme on the Introduction to Artificial Intelligence (AI) in Language was hosted by the University Language Planning and Development Office (ULPDO) in collaboration with the South African Centre for Digital Language Resources (SADiLaR).

The workshop introduced participants to the fundamentals of AI allowing them to explore the relationship between language and technology, available term glossaries, data management tools, machine translation and the ethical considerations of using AI in translation and linguistic contexts.

Acting Director of ULPDO, Mr Khumbulani Mngadi, said the event was held in partnership with SADiLaR which is a centre that supports the creation, management and distribution of digital language resources, in particular in universities. Mngadi said the workshop would provide language practitioners, academic scholars and students with the knowledge and skills required to advance indigenous languages using AI.

Mngadi highlighted the achievements of UKZN as the only Higher Education Institution with an isiZulu corpus boasting 40 million tokens/words - making it the first African language to achieve such a milestone. “With this workshop we aim to explore how we can use these tokens to train AI to translate and edit texts in indigenous languages, and this can only be done if we accumulate and feed AI enough datasets for machine learning to take place,” he said.

He urged language educators, practitioners and students to play a part in advancing indigenous languages in the field of technology and AI. Mngadi said: “Language plays an important role in preserving one’s identity and culture and advancing languages also means protecting the indigenous knowledge systems carried by our African languages. Engage your social media platforms using isiZulu or your indigenous languages, that’s how AI will pick up more datasets and start to improve their currency.”

Ms Jessica Mabaso, Leader of the ESCALATOR Programme, which is a SADiLaR flagship initiative, said the workshop was a response to a need for individuals to be educated in the fundamental ways in which AI can be used to advance African languages.

A research institute funded by the Department of Science, Technology and Innovation, SADiLaR is based at the North-West University with a national mandate of advancing the digitisation of all South Africa’s official languages.

Mabaso said the ESCALATOR Programme sought to bridge the gap in the work being done by driving digital scholarship and facilitating the integration of AI literacy in academia, and creating pathways and communities for them to engage with.

In his presentation, Dr Benito Trollip, SADiLaR’s Digital Humanities Researcher, offered an insightful introduction to data management, demonstrating how effective data practices formed the foundation for AI applications in language technology and digital humanities research.

SADiLaR’s researcher and node project manager Ms Marissa Griesel provided an interactive and practical session that explored:

•    Available AI and language tools - demonstrating how to download glossaries from the SADiLaR repository and integrate them into Autshumato, making digital linguistic resources accessible for everyday use

•    An introduction to machine translation (MT) - highlighting the vital role of linguists in refining MT systems, and guiding participants through comparisons of free MT and AI tools for translation and language work

•    Ethical considerations for using AI-generated translations, emphasising responsible data practices, quality assurance, and preserving the human touch in digital translation

In her closing remarks, senior lecturer at UKZN’s School of Commerce Dr Lindiwe Kunene, thanked the diverse group of academics, students, language practitioners, government officials and media representatives for their attendance.

Kunene also thanked the organising committee for making the event possible and the presenters for sharing their insights on how AI can be harnessed to advance innovation, inclusivity and linguistic diversity in South Africa’s digital and educational ecosystem.

Robust discussions took place on AI’s limitations in constructing linguistic meaning, the importance of feeding AI platforms indigenous languages to increase their data, and how digital resources can remain current with the ever-evolving world of AI.

•    Visit UKZN’s Human Language Technologies and SADiLaR’s Language Resource Catalogue for more information.

Words: Hlengiwe Khwela and Lihle Sosibo

Photograph: Andile Ndlovu