MaTIAS, which stands for Machine Translation for Asylum Seekers, is a project designed to support multilingual communication in asylum reception centres in Belgium through the development of a multilingual messaging system. The MaTIAS-tool allows staff members at reception centres to send practical information info and instructions to centre residents, who receive these messages in their own languages through WhatsApp. The prototype was developed on the basis of concrete communication needs identified during fieldwork in Belgian reception centres.
Integrated with ModernMT, an machine translation system, MaTIAS can automatically translate short text messages from Dutch, English or French into 17 target languages. To ensure high-quality output, the system is enhanced with a context-specific translation memory built from a database of messages —along with their professional translations into all 17 languages—tailored specifically to the Belgian reception context.
To further refine the prototype, a testing phase was carried out in collaboration with six federal reception facilities in Flanders, Brussels and Wallonia.
This interdisciplinary project combines methods from ethnography and translation technology. It is a close collaboration between two research groups: MULTIPLES (July De Wilde, Katrijn Maryns, Ella van Hest and Anaïs Wouters) and LT3 (Lieve Macken, Arda Tezcan, Margot Fonteyne, Michael Lumingu and Simon Van den Meersschaut). The project is funded by the European Union.