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

Lu Wang, Sarah Shugars, and Kechen Qin, Nick Beauchamp
Transactions of the Association for Computational Linguistics
Volume 5 - 2017 p.219-232
May 15, 2017


Debate and  deliberation play essential roles in politics and government, but most models  presume that debates are won mainly via superior style or agenda control.  Ideally, however, debates would be won on the merits, as a function of which  side has the stronger arguments. We propose a predictive model of debate that  estimates the effects of linguistic features and the latent persuasive  strengths of different topics, as well as the interactions between the two.  Using a dataset of 118 Oxford-style debates, our model's combination of  content (as latent topics) and style (as linguistic features) allows us to  predict audience-adjudicated winners with 74% accuracy, significantly  outperforming linguistic features alone (66%). Our model finds that winning  sides employ stronger arguments, and allows us to identify the linguistic  features associated with strong or weak arguments.

Related publications