Winning on the Merits: The Joint Effects of Content and Style on Debate Outcomes

Winning on the Merits: The Joint Effects of Content and Style on Debate Outcomes Lu Wang1 Nick Beauchamp2,3 Sarah Shugars3 Kechen Qin1 1College of Computer and Information Science, Northeastern University 2Department of Political Science, Northeastern University 3Network Science Institute, Northeastern University luwang@ccs.neu.edu, n.beauchamp@northeastern.edu {qin.ke,shugars.s}@husky.neu.edu Abstract Debate and deliberation play essential roles in politics and government, but most models presume that debates are won mainly via su- perior style or agenda control. Ideally, how- ever, debates would be won on the merits, as a function of which side has the stronger ar- guments. We propose a predictive model of debate that estimates the effects of linguis- tic features and the latent persuasive strengths of different topics, as well as the interac- tions between the two. Using a dataset of 118 Oxford-style debates, our model’s combi- nation of content (as latent topics) and style (as linguistic features) allows us to predict audience-adjudicated winners with 74% ac- curacy, significantly outperforming linguistic features alone (66%). Our model finds that winning sides employ stronger arguments, and allows us to identify the linguistic features as- sociated with strong or weak arguments. 1 Introduction What determines the outcome of a debate? In an ideal setting, a debate is a mechanism for determin- ing which side has the better arguments and also for an audience to reevaluate their views in light of what they have learned. This ideal vision of de- bate and deliberation has taken an increasingly cen- tral role in modern theories of democracy (Haber- mas, 1984; Cohen, 1989; Rawls, 1997; Mansbridge, 2003). However, empirical evidence has also led to an increasing awareness of the dangers of style and rhetoric in biasing participants towards the most skillful, charismatic, or numerous speakers (Noelle- Neumann, 1974; Sunstein, 1999). Motion: Abolish the Death Penalty • Argument 1 (PRO): What is the error rate of convicting peo- ple that are innocent? ...when you look at capital convictions, you can demonstrate on innocence grounds a 4.1 percent error rate, 4.1 percent error rate. I mean, would you accept that in flying airplanes? I mean, really. ... • Argument 2 (CON): ... The risk of an innocent person dying in prison and never getting out is greater if he’s sentenced to life in prison than it is if he’s sentenced to death. So the death penalty is an important part of our system. • Argument 3 (PRO): ...I think if there were no death penalty, there would be many more resources and much more oppor- tunity to look for and address the question of innocence of people who are serving other sentences. Figure 1: In this segment from a debate over abolish- ing the death penalty, argument 1 is identified as hav- ing the linguistic features ‘questions’ and ‘numerical ev- idence’, while arguments 2 and 3 use ‘logical reasoning.’ Our model infers that both pro arguments are intrinsically “strong,” while the con argument is “weak”, although ar- guments 2 and 3 both use ‘reasoning’ language. In light of these concerns, most efforts …

Content truncated to 500 words