The Linguistic Marginalization of Low-Resource Languages: Confronting AI's Limits in Digital Cultural Heritage

ABSTRACT: Cultural-heritage institutions are increasingly digitizing vast collections to preserve and make accessible the histories of diverse communities, including those forcibly displaced. However, a significant and often overlooked challenge in this endeavor lies in the inherent limitations of current artificial intelligence (AI) models, particularly for non-English and low-resource linguistic communities. This article examines how the scarcity and inadequacy of AI tools for languages such as Palestinian Arabic impede deep scholarly and public engagement with crucial historical narratives. Drawing on a detailed case study of the Palestinian Oral History Archive, this article illustrates how a community-governed, linguistically attentive digital archive exposes the limitations of English-centric, high-resource natural language processing pipelines. This article also addresses broader structural inequities shaped by market incentives, data colonialism, institutional capacity gaps, and the geopolitical constraints surrounding Palestinian linguistic data. These dynamics constitute forms of testimonial and hermeneutic injustice. This article concludes by outlining pathways toward linguistic justice in cultural-heritage AI, emphasizing community-led model development, dialect-sensitive tools, ethical governance frameworks, and collaborative institutional infrastructures that can sustain equitable access for marginalized linguistic communities.