In collaboration with the Air Force’s 517th Training Group’s Linguist Next program, we developed the Mage Duel Serious Game, a learning tool to support language proficiency for US Air Force language analysts.
In the fantasy world of Degom, word magic is erupting across the land and players must use their evolving word magic to navigate complex challenges and win quests.
Mage Duel features timed language production activities that build language fluency, accuracy, and complexity, while rewarding players for semantic similarity rather than exact formal language equivalence.
Semantic similarity is how “close” two utterances are to each other based on meaning. The game currently includes one multilingual semantic similarity engine and is built for expansion. This state of the art deep neural network-based model powers the language production activity. This means that the players in-game language inputs are evaluated based on meaning, which mirrors the way humans naturally acquire language skills.
The corpora of words, phrases, and sentences are drawn from the DLI Arabic curriculum, although they can also be drawn from other curricula or other sources.
Our initial usability study showed promising results among target end users.
With Mage Duel, Language professionals can enjoy acquiring and sustaining language proficiency at an accelerated pace, significantly reducing training costs while improving readiness and protecting national security.
Ещё видео!