International Workshop on Computational Argumentation and Cognition
COGNITAR 2020 - Keynote
Argumentation, as a human activity conducted in natural language, and subject to all the irrelevancies, redundancies, and opacities which that entails, can be stubbornly difficult to bring under any sort of systematic control. It is inherently informal, and in questions of whether it is appropriate, persuasive, and clear, steadfastly subjective. That does not mean, however, that tools for the assessment and evaluation of argument have to be equally chaotic.
The closest that argumentation theory has so far come to a clear set of rules of argument practice is found in the Pragma-dialectical approach of van Eemeren and Grootendorst. Unsurprisingly, however, their set of guidelines focuses on pragmatic features of the dialogue itself, with scant mention for other elements of arguing. These rules can be combined with argument schemes - of which the late Douglas Walton was the best known exponent - to give some form of assessment procedure, but it remains incomplete and has not been clearly elaborated.
In this talk, I introduce the Comprehensive Assessment Procedure for Natural Argumentation (CAPNA), which aims to set out a clear pattern for argument evaluation through the use of a systematic set of procedural questions. The CAPNA is based on an understanding of argumentation as the Expression of Reasoning within a Process. It has, therefore, three main levels of assessment: Process - involving my Informal Argument Pragmatics, Reasoning - employing an argument identification procedure and further analysis developed with Jean Wagemans and using his Periodic Table of Arguments, and Expression - analysed under Informal Argument Semantics.
As well as describing the procedure and its theoretical background, I illustrate the method of the tool with examples, and, crucially, discuss its potential as a pseudo-algorithm which could provide the basis for, or show a path towards, a computational application for natural language argument assessment.