A Statistical Approach to Questions of Authorship: The Case of The Woman of Color

Speaker Name: 
Corrinne Harol & Subhash Lele
Speaker Title: 
Assoc Prof of English and Film Studies & Prof of Statistics
Speaker Organization: 
University of Alberta
Start Time: 
Monday, May 21, 2018 - 3:30pm
End Time: 
Monday, May 21, 2018 - 5:00pm
UCSC, Engineering 2, Room 180
Depts. of Literature and AMS

Abstract: The Woman of Color (1808) is an epistolary novel written from the perspective of a mixed race Caribbean woman who travels to England to fulfill the mandate of her father’s will and who comments on the racism and hypocrisy of English society. As with many novels of this time, the authorship is not revealed but rather indicated by a chain of attributions to other novels. For several reasons, literature scholars are interested in knowing who might have written the novel. The chain of attributions suggests a candidate—presumably a middle class Englishwoman—who published many other mainstream novels. But some scholars conjecture that it may have been written by a mixed race Caribbean woman, and if so, this would be the first known novel by such an author. Stylometry, a collection of statistical and computational tools, is used widely in determining the disputed authorship of a text. Our talk will outline the struggles of a literature scholar and a statistician in understanding and applying stylometry tools to this problem. As will be clear from our presentation, we have more questions and ideas than concrete results at this point. Some of the main questions we have struggled with are: (a) How do we quantify evidence for one candidate writer vis a vis an alternative and how do we quantify the errors? (b) How do we appropriately quantify variability within an author’s writing? (c) What is an appropriate sample for the analysis? (d) How do we understand the strength of stylometric evidence, especially given that it measures things that literary scholars would not consider style? (e) How do we incorporate expert opinion that is not quantified in the same way as the data?


Corrinne Harol is Associate Professor of English and Film Studies at the University of Alberta. She is a specialist in Restoration and Eighteenth Century Literature and the Literary Director of The Orlando Project, a digital resource about women writers.

Subhash Lele is professor of Statistics in the Department of Mathematical and Statistical Sciences at the University of Alberta. His research interests span various fields of statistics (Foundations of statistical inference, Hierarchical models, Markov Chain Monte Carlo Methods, Spatial Statistics) with substantive applications in ecology, physical anthropology, medicine and public health.