Situated learning for collaboration across language barriers

Many Australian Indigenous communities are experiencing language shift in favour of English. This phenomenon is not new but its acceleration is. Most of these languages are spoken in remote communities where western workers collaborate with locals. The main language is not systematically known by the members of the community, and the language barrier might cause issues for collaboration. This situation presents an opportunity to preserve the endangered language through its documentation, and then use this documentation as a new material to help locals and western workers learning each other native tongues.
This work will be divided into four parts: the first part will take place in a remote community in west Arnhem where a data collection will be made in English and Kunwok. The purpose is to record usable data related to the task in which the two communities have to collaborate. In a second part, we will use recent NLP (Natural Language Processing) technologies in order to automatically extract relevant knowledge such as lexicon or catch phrases. From this information, we will design a prototype which can be used in an uncontrolled environment for helping the mutual comprehension. Finally, we will evaluate this prototype in terms of the effectiveness of the collaboration and the mutual language learning. If successful, this work could not only enhance cross-lingual collaboration in these communities, but also help to preserve the minority language.