Return to Article Details When Alignment becomes Reinforcement: Linguistic Mechanisms of Epistemic Drift in Human-LLM Dialogue Download Download PDF