Stylometry
Overview
Stylometry studies recurring features of language and other creative work to compare styles, investigate authorship, and understand variation across writers, genres, and settings. Methods range from character and word n-grams, lexical and syntactic features, and statistical distance measures to embeddings and large language models; they support attribution, verification, AI-text analysis, and forensic comparison. Simple, interpretable features often remain competitive, while newer approaches combine them with learned representations and test whether results transfer across languages, domains, and unseen authors. A central tension is that style signals can be useful yet confounded by genre, topic, editing, or shared training, and the same methods that identify authors can threaten privacy or be evaded by rewriting.
Latest papers
35 papers1 paper in the last four weeks, down 80% on the four weeks before. 0.0% of all new papers.