Breaking Language Barriers in Open Education: An IoT-Powered AI Device for Inclusive Access

• In Event: Language, Technology, and Cross-Cultural Competence for Inclusive, Equitable, and Peace-Oriented Education

Proposal

Open Educational Resources (OER) hold promises for education, allowing educators and learners to reuse and share content without copyright restrictions (Hilton et al., 2016). They help reduce costs and expand access, and research shows student performance with OER is often equal or better than with traditional resources (Cho & Permzadian, 2024; Colvard et al., 2018; Hilton III, 2016).

Despite such promises of OER, challenges remain. For instance, these applications and research on OERs are more prevalent in English-speaking countries and institutions. A UNESCO report revealed that nearly 92% of the materials on a major global OER platform are in English (UNESCO, 2023). This concentration excludes millions of students whose schooling is in other languages, as content outside their mother tongue creates major barriers to comprehension and engagement.

To better understand these challenges, we conducted a quantitative content analysis of OER to map the linguistic availability in the OER Commons library, and the findings confirm a striking imbalance in linguistic availability usually highlighted in UNESCO reports. Almost all content in the OER Commons global library can be found in English; languages like French and Chinese combined represent only 3.47% of the total content. This confirms that English is overwhelmingly dominant across all education levels, while the vast majority of the world’s languages are scarcely represented.

When broken down by education level, the disparities persist: in the resources for Lower Primary Schools, English holds over 79% of available content, while French, Arabic, Chinese and Russian combined do not reach 2%; for Upper Primary Schools, English accounts for over 97% of the share, with French, Arabic, Chinese and Russian combined are below 3%; Middle school, English remains the dominant language with more than 89% of the content, while French, Arabic, Chinese and Russian together account for under 6%. High school, English represents over 78% of the share, with French, Arabic, Chinese and Russian reaching 12.8%—the only educational level where these languages combined surpass 10%.

Local languages, especially indigenous African and Asian ones, remain largely absent from OER repositories. This imbalance has been highlighted for over a decade through UNESCO initiatives: the 2012 Paris Declaration called for OER in diverse languages, the 2017 Ljubljana Action Plan stressed closing the “language gap,” and the 2019 UNESCO Recommendation on OER made linguistic diversity a core principle. Yet, the 2023 Global Education Monitoring Report shows little progress, with English still dominating and millions of learners excluded.

Compounding the language gap is the digital divide. Nearly two-thirds of school-age children—about 1.3 billion—lack internet access at home (UNICEF, 2020). During COVID-19, this meant disadvantaged learners, especially in rural areas such as Mozambique, could not join online classes or access digital content due to limited electricity, hardware, and connectivity.
The question arises: how can children and youth who do not speak English access quality OERs with limited technology? This challenge is acute in sub-Saharan Africa, Latin America, and rural Asia, where local languages are essential. The dominance of English in OER reinforces systemic barriers, excluding learners twice— by language and by the digital divide limiting access to multilingual tools.

Our proposed IoT-powered AI device directly responds to these compounding barriers, offering multilingual access to OER in offline environments, making the movement toward equitable education more than just an aspiration.

This project proposes a solution response through the integration of Artificial Intelligence (AI) and Internet of Things (IoT) technologies. Specifically, we present the design and pilot development of an IoT-enabled AI device that automatically translates OER, including audiobooks, and digital textbooks into multiple languages. For this specific study, we focused on Portuguese, since it was conducted in the context of Mozambique. The device is designed to function in low-resource settings, operating offline once the content is loaded, and updated only when internet access is available. By coupling translation models with IoT-enabled delivery, the system provides a pathway to equitable access, bridging both linguistic and infrastructural divides.

The theoretical grounding of this project draws on equity frameworks in global education, emphasizing inclusion, participation, and the right to learn in one’s own language. We frame language accessibility as a fundamental right, not a privilege. The device also aligns with UNESCO’s vision of OER as a tool to democratize knowledge, ensuring open access reaches all learners rather than reinforcing hierarchies.

Methodologically, the project follows a design-based research approach. First, we mapped the linguistic landscape of OER and identified critical gaps, using both quantitative data and qualitative insights from educators from Mozambique, a country used as reference for representing developing countries which cannot speak English as mother-tongue language. Second, we prototyped the device using a Raspberry Pi microcomputer, integrating ai-translation, and ai-voice synthesis modules.

We hypothesize that the IoT-AI device will improve access and learning in pilot contexts, particularly by increasing availability of curriculum-aligned materials in local languages. We expect students to engage more with quality content compared to baseline groups without the device, since it removes two major barriers: language and internet dependence. The study will measure and compare learning gains (for example, better understanding of subject matter and improved literacy skills) when taught without the OER, when taught with OER delivered as it is, and OER delivered in their native language via the device. Teacher feedback will also be assessed, focusing on whether AI-driven personalization helps save time and support differentiated instruction. While generalization from a limited pilot is premature, we expect results to show that technology designed to bridge linguistic and digital gaps can markedly improve inclusion in education.

The significance of this research lies in its contribution to reframing the global OER movement around inclusion and accessibility. Current patterns reinforce linguistic inequities, limiting the transformative potential of open education. By demonstrating a practical solution that merges AI and IoT, the project offers an alternative model where technology actively dismantles barriers rather than exacerbating them. The device challenges the dominance of English in open education by providing learners with immediate, localized access, thereby supporting Sustainable Development Goal 4: ensuring inclusive and equitable quality education for all.

Authors

• Joao Filipe Papel, United Nations University - Institute for the Advanced Study of Sustainability

• Jonghwi PARK, United Nations University - Institute for the Advanced Study of Sustainability

• Tatsuji Munaka, Tokai University