AI Large Language Models, Algorithmic Language Policies and Social Stratification

The rapid integration of AI Large Language Models (LLMs) into daily and institutional life constitutes a critical new frontier for the sociology of language. These systems are not neutral technological tools; they are active agents that mediate linguistic capital and reshape social stratification, particularly along the intersecting lines of social class, gender, and age. By automating text generation, translation, and interaction, AI models invariably enforce specific linguistic norms, effectively acting as hidden, top-down mechanisms of language policy.

This session critically examines how Generative AI architectures construct, reproduce, and enforce these "algorithmic language policies." Trained predominantly on massive corpora that reflect the linguistic hegemony of the Global North and standardized, elite sociolects, LLMs implicitly encode prescriptive language ideologies. We must interrogate how these models distribute linguistic capital and reproduce systemic biases, potentially widening the gap between dominant groups and communities marginalized by their class origins, gender identities, or generational divides. When corporate tech giants become the de facto language planners of the digital age, how do algorithms negotiate or exacerbate existing social inequalities?

We invite theoretical, methodological, and empirical contributions that address the stratification effects of AI.