Scholarly method, truth, and evidence in Shakespearian textual studies

Although their raw materials are often the same, the methods of Shakespearian textual scholarship can be markedly different from those of Shakespearian literary-dramatic criticism, which are different again from those of theatre history. Ours is a diverse scholarly community. Recently a new methodological division has emerged with the rise of computational stylistics (also known by the old name of stylometry) which claims to give new insights into authorship attribution problems and the characterization of writing styles.

This article considers these methodological differences, showing that computational stylistics is closer to theatre history than literary-dramatic criticism, most noticeably in its truth claims. Most distinctive of the new computational methods is the nature of the evidence they present, which is primarily numerical (the counts of occurrences of phenomena) and statistical in its interpretation (asking whether chance might produce the numbers found). Humans are notoriously poor at judging from coincidences, and part of this talk will consider why we need the much-maligned culture of statistics to make sense of this evidence.

The article ends with some practical tips for how those unfamiliar with this new area of Shakespearian study may distinguish reliable from unreliable investigations, briefly covering the notions of scientific replicability, the empirical demonstration of a method's discriminatory power, and some examples of best scholarly practice regarding the selection of datasets and tools.