Dimensions of inequality in broadcast political debate

About this Session


Wed. 10.04. 17:25



Authors – Zlata Kikteva, Annette Hautli-Janisz.

Abstract :

Broadcast political debates are a way to give the public easy access to the governmental decision-making process and to reduce perceived inequality: citizens are able to make informed choices and comment on the way policies are implemented. One of the UK’s most viewed political talk shows is ‘Question Time’ on BBC1, where audience members are invited to ask questions to a panel of five political and societal figures. Just looking at the superficial textual features, one could conclude that the setting is highly unequal: the audience contributes only around 12% of all speaker turns and those are generally much shorter than the responses from the panel.

However, in this talk, we discuss that the picture is not that clear-cut: based on the QT30 corpus (Hautli-Janisz et al., 2022), the largest corpus of annotated political debate to date, covering 30 episodes of ‘Question Time’ between 2020 and 2022, we derive a number of pragmatic and argumentative insights that show a more nuanced picture. For one, the moderator significantly amplifies the impact of the audience by reiterating and rephrasing their questions until she gets a response from the panel, ending up with four times more information-seeking questions than the audience. Secondly, the argumentative structures of the responses to the audience and the moderator are very similar: in both cases, the questions elicit a small number of attacks and roughly three times as many supports (either to the content of the question or the intention behind it). This means that the audience and the moderator are treated equally, despite the latter having much more leverage in the debate. Finally, the usage of different question types suggests that the audience has an impact on the discussion. Rhetorical and assertive questions have been shown to have a high argumentative impact (Kikteva et al., 2022) – the former are used to support claims, and the latter elicit conflict. Our data shows that the audience uses significantly more rhetorical questions than the moderator, who in turn uses as many assertive questions as the audience.

Overall, we claim that inequality in debate is an intricate phenomenon which requires measurements that go far beyond the linguistic surface and in fact require a detailed pragmatic and argumentative account of the debate.


  • QT30: A Corpus of Argument and Conflict in Broadcast Debate (Hautli-Janisz et al., LREC 2022)
  • The Keystone Role Played by Questions in Debate (Kikteva et al., CODI@COLING 2022)