Structural Silence examines the structural barriers that limit participation of underrepresented languages in modern AI systems. Using Bengali as a case study, this poster highlights four systemic constraints: web presence disparity, multilingual training token imbalance, tokenization overhead, and connectivity exclusion. Together, these factors create measurable asymmetries in model performance and accessibility. The poster argues that dataset scarcity is not merely a technical limitation but a structural infrastructure issue shaped by historical and economic forces. It calls for offline-first design strategies, transparent multilingual evaluation practices, and formal recognition of dataset construction as a core research contribution. This poster was presented at the 69th Annual Conference of the International Linguistic Association (ILA 2026) at John Jay College of Criminal Justice, New York, NY, USA.
Roy et al. (2026) studied this question.