AI-Driven Translation, Linguistic Hegemony, and Neocolonialism: Reconsidering Language Power in the Digital Age

Abstract

Artificial intelligence (AI)-driven translation is frequently presented as a technology capable of reducing linguistic barriers and widening access to information. Yet translation technologies do not operate outside histories of unequal linguistic, economic, cultural, and technological power. This conceptual research paper examines the intersection of AI-driven translation, linguistic hegemony, and neocolonialism, with particular attention to data inequality, technological ownership, market concentration, cultural homogenisation, and the digital language divide. Drawing on the conceptual material supplied for this study and a critical review of scholarship on data colonialism, algorithmic bias, low-resource machine translation, and participatory language technology, the paper argues that the central problem is not AI translation itself but the unequal structures through which languages become visible, measurable, profitable, and technically supported. The paper develops a framework linking four dimensions of power: data, infrastructure, markets, and representation. It further considers how these dimensions affect marginalised and low-resource languages through uneven training data, limited evaluation resources, dependence on externally owned infrastructures, and the possible flattening of culturally embedded meanings. At the same time, the study recognises that AI can support language preservation, multilingual education, and cross-cultural communication when communities participate meaningfully in data creation, model development, evaluation, and governance. The paper concludes by proposing a community-centred and linguistically plural model of AI translation grounded in data sovereignty, participatory design, transparent evaluation, equitable access, and human oversight.