This project is on hold. We are looking for user partners to move forward together. Please contact us if you are interested in partnering on this project.
Problem
There are many extremely effective language learning tools, such as Lingoal, Duolingo, and Babbel. At core, these tools rely on natural language understanding models for tasks such as part-of-speech tagging and named entity recognition. We will be able to build similar tools for Tibetan in the wake of Project Manas.
These tools would need to work "in the wild" to be useful. There are many different learners of Tibetan: interested Westerners with reasonable computers and internet access, student translators at 84,000 and Rangjung Yeshe Institute with low budgets, Westerner monks and nuns in Dharamsala who may have only unreliable internet access and hardware, and even very poor villagers in Nepal with almost no access to technology. Even native Tibetans may use such tools to overcome the linguistic genocide being perpetrated in their homeland.
Proposal
Partnerships with organizations serving the many groups of people who are learning Tibetan will be essential for success. Once the needs of some such organizations are understood, ideally a dedicated UX team should be stood up. Tools should be developed in partnership with the modeling experts to ensure that the requisite models run on the hardware necessitated by the problem. The deliverables may take the form of a phone or iPad app, custom hardware such as a Raspberry Pi, or even teacher-only tools to help quickly prepare offline lesson materials.
Lingoal may be a good partner and model for teaching tools. They have an API for using some of their paid tools for new languages on a nonprofit basis.