Bridging Divides: Leveraging Generative AI and Translanguaging for Peace and Social Cohesion in Canadian Multicultural Math Classrooms

• In Event: Multilingual Education in K-12 Contexts

Proposal

Introduction

This proposal contributes to the CIES 2026 theme by examining how educational practices can mitigate divisions and promote social cohesion at a foundational level within the multicultural classroom. Canada's commitment to welcoming 500,000 newcomers annually (IRCC, 2023) results in classrooms that are increasingly diverse in terms of language and culture. While this diversity is a strength, it also presents challenges, particularly for multilingual learners like Chinese newcomer students who often struggle with mathematics taught in English. Research indicates that English language learners (ELLs) typically perform lower in STEM subjects, with only 34% of Grade 9 ELLs meeting provincial math standards, highlighting the difficulties of acquiring English proficiency while simultaneously mastering complex content (Crosson et al., 2020; EQAO, 2023; Riordáin & O’Donoghue, 2009). These linguistic and cultural barriers are not merely academic hurdles; left unaddressed, they can lead to feelings of exclusion, marginalization, and a lack of belonging, potentially creating micro-divisions within classrooms that undermine a harmonious learning environment and broader social cohesion.

Translanguaging, which allows students to leverage their full linguistic repertoire for learning, offers a powerful strategy to address these challenges, facilitating deeper cognitive engagement and meaning-making (Baker, 2011; Lewis et al., 2012). By disrupting traditional language hierarchies and empowering multilingual students, translanguaging can actively contribute to a more equitable and inclusive classroom, fostering an environment where diverse linguistic abilities are valued (Piccardo, 2013). The integration of Artificial Intelligence (AI) tools, such as GPT-5, presents a promising avenue to support such multilingual pedagogies by providing real-time translations, explanations, and feedback. This research argues that by ensuring equitable access to learning and validating diverse linguistic and cultural backgrounds, AI-powered translanguaging can serve as an effective educational practice for peacebuilding within the classroom, proactively addressing potential sources of division.

This study aims to investigate the effectiveness of GPT-5 in supporting Chinese newcomer students’ mathematics learning through translanguaging pedagogy, specifically focusing on how this approach contributes to fostering peace, mitigating linguistic and cultural divisions, and promoting social cohesion in Canadian multicultural classrooms. By leveraging AI to harness students' prior knowledge as valuable resources, this research contributes to the ongoing dialogue on global peace and social cohesion and how AI can enhance inclusive, cross-cultural education in Canada's increasingly diverse classrooms.

Research Question

How does the effectiveness of GPT-5 in supporting Chinese newcomer students' mathematics learning through translanguaging pedagogy contribute to fostering peace, mitigating linguistic and cultural divisions, and promoting social cohesion in Canadian multicultural classrooms?

Theoretical Framework

Sociocultural Theory(Vygotsky, 1987) emphasizes that learning is inherently a social process, with knowledge being constructed through interactions within a cultural context. This perspective is crucial for comprehending the learning trajectories of multilingual learners, such as Chinese newcomer students in Canada, as it positions learning as deeply embedded in social and cultural activities, where individuals learn through participation and interaction with others and with tools. Building on this, crosslinguistic translanguaging theory (Cummins, 2022) offers a pedagogical approach that recognizes the credibility of distinct, yet fluid and porous, language-specific features within a bilingual's cognitive system. The common underlying proficiency model, which crosslinguistic translanguaging theory supports, explains that instruction in a student's home language (L1) can facilitate the transfer of concepts, skills, and learning strategies across languages to the dominant societal language (L2). This dynamic crosslinguistic interaction encourages teachers to actively promote productive contact and transfer between languages (Cummins, 2022). It allows students to strategically utilize their full linguistic repertoire—multiple languages—for learning, thereby facilitating deeper cognitive engagement and meaning-making. This approach moves beyond traditional language hierarchies by recognizing their linguistic abilities as valuable assets and harnessing the dynamic interaction and transfer of concepts, skills, and learning strategies across languages, supported by the empirically validated common underlying proficiency.

Methodology

Qualitative Content Analysis (QCA) is employed to systematically examine how generative AI tools like GPT-5 influence the transfer of mathematical knowledge across languages and cultures. This method allows for a detailed analysis of linguistic, cultural, and pedagogical aspects, providing insights into both the opportunities and challenges AI presents in diverse educational settings. A coding scheme identifies themes such as linguistic accuracy, cultural relevance, and the effectiveness of translanguaging in student comprehension. By meticulously coding and grouping recurring themes, this approach offers an in-depth understanding of how AI-supported translanguaging can impact mathematical learning across diverse linguistic and cultural backgrounds, and implicitly, how it supports or hinders social cohesion.

Preliminary Findings

From a linguistic perspective, GPT-5 demonstrates high accuracy in mathematical terminology and problem statements, ensuring content integrity. This linguistic precision is vital not just for academic comprehension but also for reducing linguistic barriers that can alienate students. Despite linguistic accuracy, the findings reveal that certain cultural elements embedded in math problems present challenges that AI tools cannot fully address. Canadian math problems often include contextual references (local practices, geography, or typical scenarios) unfamiliar to Chinese students. The AI's inability to recognize or adjust for these cultural differences means that math problems based on Canadian sports events, holidays, or local landmarks may not resonate with Chinese students, leading to difficulties in contextualizing and engaging with the problems. This is a critical point from a peacebuilding perspective; if unaddressed, these cultural nuances can perpetuate feelings of exclusion and misunderstanding, thereby creating or reinforcing divisions within the classroom.
Conclusion

While AI-generated translations demonstrate commendable linguistic accuracy, the findings underscore the need to address cultural elements that AI may not fully adapt to. This study's implications extend beyond academic achievement to the broader goal of fostering peace and social cohesion. Educators are crucial in providing culturally relevant content to enhance student engagement and understanding. This role is vital for mitigating cultural divisions and fostering a sense of belonging, which underpins peace within the classroom and the wider community. AI tools should be used as supplementary resources, focusing on their strengths in linguistic support, while educators actively fill the cultural gaps. This hybrid approach ensures that curriculum content meets the diverse needs of students, thus reducing sources of division.

Author

• Ye Jia, Ontario Institute for Studies in Education (OISE), University of Toronto