#59 Stylometry with R

Joanna Byszuk, Maciej Eder and Artjoms Šeļa

This is a workshop in stylometry, or the analysis of countable linguistic features of texts. While stylometry has been usually associated with authorship attribution, the same methods are successfully applied to more general text analysis, and, recently, even analysis of other modes such as music and image. The statistics of such features as word, word n-gram or character n-gram frequencies, are not only a highly precise tool for identifying authorship, but can in fact reveal patterns of similarity and difference between various works by one author, works by various authors, finally between authors differing in terms of chronology, gender, genre or narrative styles, between translations of the same author or group of authors, or specific voices such as idiolects of characters in novels. This provides a new opening in literary studies, and the results of a stylometric analysis can be compared and confronted with the findings of traditional stylistics and interpretation. It also opens a new set of questions about style and its transfer, as well as the nature of particular features and language.

Auditors will learn major stylometric tools and methods, from simple keywords extraction to machine-learning classification based on text features, followed by visualization techniques ranging from dendrograms to networks. Auditors will learn how to identify the problem, define relevant research questions, and design an experiment.

We will use our own package written for the R statistical programming environment — ‘stylo’, which allows us to avoid R’s usually steep learning curve – we don’t expect advanced programming skills. We will provide text corpora to use for training purposes, but also hope and expect participants to bring their own data and problems to work on. In response to the pandemic, we offer an online and shortened version of the course. The focus will be on understanding basic stylometric methods and learning how to conduct simple experiments on one’s own. We still hope to provide a broad overview of the main stylometric problems and techniques, with some of the more detailed technical discussions moved to auxiliary materials.

Duration: 7-12 hours

Auditor Option: Available

Related Materials: instructor biographies