Stylometry: Computer-Assisted Analysis of Literary Texts

Stylometry, or analysis of countable linguistic features of (literary) texts, is usually associated with authorship attribution or, more sensationally, with the sole purpose of prosecuting plagiarists. However, recent studies show that the same methods that help attribute an ancient Greek text to Plato or put a plagiarist behind bars for forging an alleged play by Shakespeare can be used in a much broader context of literary study. Patterns of stylometric similarity and difference provide new insights into relationships between different books by the same author; between books by different authors, between authors differing in terms of chronology or gender; between translations of the same author or group of authors; helping, in turn, to find new ways of looking at works that seem to have been studied from all possible perspectives.

The workshop, split into two 16-hour blocks, will try to address some of the following research questions: What is common in the language we use and what is related to cultural contexts and/or writer’s individuality? What elements of style are affected by literary period, genre, topic? What is unconsciously incorporated by the author and reflects his/her education, gender, religious background, social or historical conditions? Last but not least, which features of a written text can betray the person who wrote it despite his/her aesthetic, social, or historical conditions? The first and less advanced block, From Nothing to Networks, will acquaint the participants with stylo, the suite of stylometric tools produced by the workshop leaders, and with Gephi, a tool for network analysis. The second block, From Networks to Nearest Shrunken Centroids, will introduce some more advanced problems of classification and validation, and more advanced methods of statistical analysis of texts.

Both blocks will focus on tools based in the R statistical programming environment with user-friendly interfaces, so no expert knowledge of R in particular, or of programming in general is required for Block 1. The texts used for the workshops will be provided by the instructors and the participants are encouraged to bring their own; if necessary, the participants’ individual corpora will be expanded as needed and as available (online or elsewhere). The texts will be literary, multilingual, and include both originals and translations. Block 2 will be building on the skills acquired in Block 1, but a catching-up session is planned for participants with some R and Gephi knowledge.

Schedule:

Block 1: From Nothing to Networks (16 contact hours)

Session 1:
Presentation of participants
Introduction to the workshop
Locating/making corpora

Session 2:
A shock introduction to stylometry in R

Session 3:
Stylo: growing Cluster Analysis trees

Session 4:
Stylo: growing Bootstrap Consensus trees

Session 5:
Introduction to Gephi

Session 6:
Growing networks with Gephi

Session 7:
Introduction to Oppose

Session 8:
Oppose, stylo, and Gephi 

Block 2: From Networks to Nearest Shrunken Centroids (16 contact hours)

Session 1:
Catching up on stylometry

Session 2:
Theoretical problems: classification

Session 3:
Theoretical problems: validation

Session 4:
Crunching literary collaborations

Session 5:
Authorship attribution; advanced distances.

Session 6:
Automatic POS-tagging

Session 7:
Clustering POS-tags

Session 8:
Networking POS-tags