Nicholas Wlezien

My research interests lie at the intersection of natural language processing and large language models. I am particularly interested in how model predictions are affected by prompts, how language models interpret stylistic evidence, and how to evaluate these systems more reliably.

In my recent work at EMNLP 2026, we examine a conservative bias in zero-shot authorship verification and investigate lightweight prompt interventions that encourage models to distinguish writing style from topic and content.