SBIR Phase I: Intelligent Language Learning Environment

Award ID: 2112088

The broader impact/commercial potential of this Small Business Innovation Research (SBIR) Phase I project will improve the current US deficiency in foreign language skills and bring critical innovations to the $13 billion global online language learning market. Current language learning methods focus on rote memorization and not nearly enough on using the language. Experiential learning opportunities in which students have free-flowing conversations driven by interests are often not introduced until the third or fourth semester of language study. The result is that American students graduate with little ability or motivation to actually use a foreign language and maintain any degree of language fluency. Using novel machine learning technologies and game design, this project will let students learn a language while texting with their friends and help teachers facilitate conversation-based learning in their classrooms. The project will offer opportunities for students in classes located in the US to interact with students in other countries around the globe, turning textbook abstractions into real people, culture, and language. Ultimately, these machine learning solutions enable a platform that the 1.2 billion individuals learning language worldwide can use to supplement textbook instruction with experiential learning for greater long-term fluency. This Small Business Innovation Research (SBIR) Phase I project will develop a state-of-the-art machine learning solution to address critical challenges in the implementation of conversation-based language learning. Learning languages through conversation has been shown by previous research to be engaging and effective. However, beginners are often uncomfortable in the early days because they arent confident in expressing themselves. Educators can provide the needed support, but it is difficult for them to give every student personalized tutelage. This project will jumpstart these initial conversations with the help of in-chat translation, dictionaries, and grammar checking. To cement learning, the project will generate practice activities directly from chats and deliver the activities in the context of those authentic text conversations. This project will analyze the text chats and form hypotheses about student strengths and weaknesses. The system will then test these hypotheses with assessments automatically generated from the same text chats. The Phase I technical objectives are to 1) automatically identify examples of learning objectives in student text 2) align system-generated and human-authored assessment activities and 3) predict student scores on practice activities with increasing accuracy. Over time, the system will understand student needs to maximize learning and engagement. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.

NSF Program Director: Rajesh Mehta