The Effect of Deliberative Quality on the Perception of Speakers

Deliberative democracy has become a subject of increasing interest within the social sciences. We propose an automated text-analysis approach for measuring the deliberative quality of political discussions. We first derive evaluative standards from theories of deliberative democracy and past attempts to measure deliberative communication. We structure them along four key dimensions of deliberative discourse: participation, respect, justification and accommodation. For each of these dimensions we develop a number of measures that can be coded automatically. Our approach combines linguistic and statistical cues to draw inferences on the deliberative quality of discourse. We apply these instruments to the transcripts of simulation-gaming experiments in which participants had to take unanimous decisions on a contested issue. For each speaker, we test the effects these standards have on their perception by other participants. We find that those speakers are perceived as having constructively contributed to the agreement who participate more, who are more eloquent and who express less opposition.