As a collaborative effort from GIZ (atingi and FAIR forward) with implementing partners CLEAR Global and Digital Umuganda, we present an innovative plugin that integrates a machine translation tool for English to Kinyarwanda. Our plugin was developed by fine-tuning an open model for research use – NLLB 200[1] (No Language Left Behind)– on datasets of parallel text; manually collected by native speakers of Kinyarwanda in the domains of tourism[2] and e-learning content for business management in SMEs[3]. By leveraging community-sourced data and open-source technology, we have trained a model[4] and developed a plugin that can facilitate digital learning and bridge the language divide starting with learners in Rwanda as we aim, through collaboration with local Partners, to develop it further for learners in other regions (e.g. Ethiopia and Tanzania).
Our plugin tackles a critical challenge in digital learning; that course materials are often not available in local languages. By seamlessly integrating a machine translation tool within the Moodle LMS which incorporates automated course localization and interactive (inline) translation, we empower learners to access content in their preferred language enhancing their learning experience and promoting inclusivity.
The plugin’s utilization of AI technology not only enhances the accuracy of translations but also underscores the potential of AI in education. This paves the way for further development so that more languages will become accessible allowing learners to fully engage in their learning experience. Our initial step demonstrates the feasibility and impact of integrating neural machine translation models in open-source learning management systems.
We invite the Moodle community to explore our plugin and collaborate in advancing AI-driven solutions for education. Together, we can break down language barriers and foster a more inclusive learning environment for all learners.