Teaching Toonies: A User-Centric Approach to Designing a Robot Companion for Language and Cultural Learning

Abstract

Newcomer children face compounding challenges upon resettlement: acquiring a new language, adapting to unfamiliar social routines, and cultivating community belonging. Consequently, language and culture are deeply interconnected processes. In the context of this thesis, I focus more on English language acquisition and Canadian cultural familiarization for Newcomer children who have come to Canada. While community-based literacy programs offer vital, low-stakes environments to support this socio-cultural orientation, tutors often struggle to balance language instruction, emotional support, and attention management for highly diverse groups within limited resource constraints. SARs present an opportunity to facilitate and encourage language practice without replacing human educators. This thesis investigates the user-centered design and evaluation of Maple, a peer-like social robot engineered as a tutor-mediated companion for young newcomer children (ages between 5 to 7). Grounded in language socialization and social learning theories, along with the Design Thinking design philosophy, Maple helps children rehearse situated interactions and everyday routines, actively mitigating the anxiety and fear of evaluation often associated with additional-language acquisition. Utilizing an iterative, co-design framework, this research engaged literacy experts and HRI researchers through shadowing, interviews, and an RTA to establish critical SAR design requirements. The resulting prototype delivers story-based learning via a multi-modal "visual novel" interface featuring embedded formative assessments. To foster inclusivity and prevent cultural bias, the narrative employs anthropomorphic Canadian animal characters, grounding vocabulary in local contexts without imposing a singular human demographic. A mixed-methods user study with expert tutors evaluated Maple's pedagogical fit, usability, and classroom orchestration. Results show tutors perceived Maple as highly engaging and effective at lowering the social risk of language practice. Operating within a triadic learning architecture (tutor-child-robot), Maple is designed to function as a "second voice" for repetitive practice. However, findings also emphasize that the robot's physical embodiment and active behaviors must be carefully bounded to prevent extraneous cognitive load and classroom disruption. This thesis argues that the utility of SAR like Maple does not lie in the automation of instruction. Rather, its value lies in providing brief, playful, and socially secure moments of practice that enhance learner participation while safeguarding the human tutor’s central role in pedagogical guidance, cultural translation, and socio-emotional care.