Language Models Reveal Grammar's Epistemic Disruption

In a conceptual essay, William Benzon (with assistance from ChatGPT 5.2 and Claude 4.5) traces evidentiality from Daniel Everett’s 1980s Pirahã research to modern institutions. He argues that literacy shifts knowledge-accountability from grammatical markers to institutions like citation and peer review, and that large language models lacking lived experience create a novel epistemic rupture in how language validates knowledge.