stylometrie

Stylometry, also referred to as stylometrie in some languages, is the quantitative study of linguistic style in texts. It seeks to characterize and compare authorial fingerprints by analyzing features such as word frequencies (especially function words), character n-grams, punctuation patterns, syntax, and readability metrics. Using these features, researchers build models to attribute authorship, detect stylistic similarity across texts, identify genre or provenance, and monitor how an author’s style changes over time. Methods range from simple distance measures to probabilistic, machine-learning, and neural models, with representations that include bag-of-words vectors, n-gram profiles, and more sophisticated embeddings.

Stylometry has a long intellectual lineage, gaining prominence in modern corpus linguistics from the mid-20th century onward. Classic applications include authorship attribution of disputed writings and forensic linguistics, as well as broader literary and digital humanities research. Notable case studies involve attempts to assign or contest authorship of historical documents, such as debated literary canons or controversial texts, and contemporary work uses large-scale datasets to test cross-genre generalizability. Key challenges include the influence of topic and genre on style, the effects of translation or editorial intervention, and the need for sufficiently long samples to produce reliable results. Cross-domain generalization and model transfer remain active areas of investigation.

Ethical and interpretive considerations are central to stylometry. Findings are probabilistic, sensitive to data selection and methodological choices, and can have legal, scholarly, or reputational implications. Transparency, replication, and careful validation are essential, along with attention to biases, data quality, and privacy concerns when attribution carries potential harm.