Forgot your password?
Natural Language Processing (NLP) is an area of study, in the discipline of Computer Science, that treats the manipulation and generation of human-readable language. Recently a field of persuasive technologies has emerged and NLP promises a fertile research context for this up-and-coming area of study. Few scholars in NLP, however, have included theories or results from rhetoric in their work, the traditional study of persuasion, especially linguistic persuasion. We will argue that the study of rhetoric offers profound insights into the understanding of human language both in its structure and use, insights that are essential if persuasive NLP is to make progress. In particular, rhetoric offers an extensive treatment of complex forms of expression, known as "rhetorical figures" or "figures of speech." The Rhetorical Figure Ontology Project aims to create a comprehensive database of rhetorical figures, including their history, their individualizing features, and-perhaps most importantly-their relations to one another. While the encyclopedic information being collected and assimilated in the database is significant, it is the relationship of the figures to one another that is of central importance to the discussion in this paper. The data being aggregated for each figure provides critical information that helps to elucidate and explicate the figure as a meaning-making device. Combined with the encyclopedic information, we argue that the relationships among figures provide insight into the cognitive affinities of the figures themselves. Such work has a multitude of applications, but our specific focus here is the importance of such information in understanding the role rhetorical figures as not only meaning-making devices, but as portals to the inner-workings of the human mind. The cognitive role of figures is based in the work of Mark Turner, George Lakoff, Mark Johnson, Raymond Gibbs, and Jeanne Fahnestock. Though rhetorical figures have been defined, differentiated, and categorized for millennia-from the Rhetoric ad Herennium to Peacham's Garden of Eloquence to the contemporary work of Gideon O. Burton on Silva Rhetoricae-many figures escape clear classification. While our project does not claim, nor does it aim, to provide definitive categorization for rhetorical figures, it does work towards a systematized approach to the treatment of figures in computational systems. While such efforts will, at present, remain imperfect our work will describe methods of categorization of rhetorical figures, difficulties with the categorization, and some of the methods that we might use to work around such difficulties. To this end we will treat the development, including the theoretical underpinnings for the categorization, and population of the database. Our work is situated in the Inkpot Natural Language Processing group, part of the Artificial Intelligence Laboratory in the David R. Cheriton School of Computer Science at the University of Waterloo. The team is composed of computational linguists, linguists, and rhetoricians. Together with our particular expertise we are working towards developing such a database and, subsequently, an ontology.
Created by: Brad Bowden, Marie-Agnes Pilon, Randy Harris