Standard Language Ideology in AI-Generated Language

The paper presents a taxonomy that explores how large language models (LLMs) reinforce standard language ideology, favoring certain language varieties over others. It introduces the concept of standard AI-generated language ideology, detailing how these models contribute to societal biases through mechanisms like legitimation and stereotyping. The findings underscore the need for accountability and community agency in AI development to protect linguistic diversity, which is crucial for practitioners aiming to build equitable AI systems.