Digital Text and Scholarship & London Forum for Authorship Studies

Patrick Juola (Duquesne University): 'Fishing the Ocean: Lessons from Large-Scale Experiments in Styometry'.

Since its resurgence, the field of stylometry has been something of a professional adhocracy; there is little understanding of what the key elements of writing style actually are, or what they mean.  While it is fairly easy to identify a new method for determining the author of a text, and such a method is likely to perform substantially better than mere chance, there has been relatively little work on analyzing the fundamentals of why the method might work.   For example, there are many different ways to assess differences in word usage, but little analysis on determining what the underlying differences mean in terms of psychology or personality.   We will show some psycholinguistically motivated methods of authorship attribution, but also show that they do not work particularly well.   We will also show our current candidates for "best practices" and try -- and probably fail -- to interpret them in interesting ways.