large language models (LLMs) are trained on corpora disproportionally
weighted in favor of Standard American English. As a result, speakers of other
dialects experience significantly more failures when interacting with these
technologies. In practice, these speakers often accommodate t