Meraki: Encouraging Language Learning in Real-World Simulations through AI Para-Social Relationship Building

This paper presents a prototype application designed to teach language, using a dialogue-based theatre production experience. The application engages users in the production of a theatrical performance in a graphical 3D environment and incorporates AI verbal assistance to promote language learning. The application combines declarative and non-declarative memory building exercises to teach Hindi, although it is anticipated that this idea could be expanded to any language/culture. Users who have an intermediate knowledge of the language can evaluate their language skills in a social and professional virtual environment which incorporates motivational elements based around the Self-Determination Theory regulatory styles and Loci of Causality. Verbal AI assistance is used to provide an appropriate level of flow during the experience building a para-social relationship with the video game. Sandhi Viched, a Hindi euphonic junction splitting exercise has been incorporated into each language task. Context splitting the words into its constituent meanings is intended to help the players understand and remember the meaning of words and their use in different contexts. Initial Wizard of Oz testing has demonstrated that the AI verbal assistance provides a good balance between autonomy, competence and relatedness. Feedback from users show that by providing a balanced mix of external, introjected, identified and integrated regulatory motivation styles, as adopted from the Self-Determination theory, the AI assistance can speed up the language acquisition process and build a para-social relationship with the user. This initial evaluation suggests that the idea of theatre production that can incorporate many aspects of language provides a good starting point for an engaging learning game. This LLA topples the dominant belief, that the best way of learning a language through an application is to reinforce their behavior with external rewards. Further work is proposed to expand the application and build in more gamification with a view to conducting a more comprehensive evaluation.