Noa Mor, It’s a Global Village (If You Speak the Right Language): On Language Models, Digital Sidelining, and Participation, 42 Wis. Int'l L. J. 329 (2025).
The digital linguistic ecosystem is rife with disparities. While a select group among the world’s seven thousand languages enjoys the benefits of digitalization, speakers of Digitally Marginalized Languages (DMLs) have restricted or no access to these resources. Focusing on the AI fields of Natural Language Processing (NLP) and Large Language Models (LLMs), this Article explores the nature of these linguistic gaps and how they compound and exacerbate long-standing, offline linguistic hierarchies. This Article analyzes the techno-social predicaments that underpin these inequalities, addressing two key areas: (1) training data and training processes, and (2) design and evaluation choices and constraints. Drawing on Nancy Fraser’s “Parity of Participation” framework, this Article examines how these predicaments, along with cultural and regulatory criteria, prevent equal participation for DML speakers across three dimensions: distribution, recognition, and representation. This Article then explores the international human rights law framework that applies to governments and private AI companies, and outlines their duties and responsibilities in facilitating DML speakers’ participation. By tying together technological, fairness, and legal perspectives, this Article provides a comprehensive and novel look into global linguistic discrepancies in the digital age and how they can be tackled.
