This dissertation looks into how Cantonese utterance final particles, especially question particles, and gestures are strategically employed in three televised political debates of the 2016 New Territories East by-election for the Legislative Council of Hong Kong. The study explores some of the most frequently used question particles and discusses how they are being manipulated with two prominent questioning strategies, i.e., “hypophora” (McCormick, 2003) and “question cascade” (Clayman and Heritage, 2002), to help project negative stance in this highly institutionalized discourse genre. With the help of the theoretical framework of Stance Triangle proposed by Du Bois (2007), I first investigate how various stance-taking activities are navigated through the use of question particles and questioning techniques that usually create adversarialness and hostility among the electoral candidates in the actual debates; secondly, I also look into how the use of these question particles and question designs varies when the electoral candidates are delivering their individual speeches; lastly, I specifically investigate the elected Legislative Councilor to shed light on how he strategically deploys those linguistic and gestural practices to communicate with the “remote audiences” (Linell, 2009, p. 101) and the likely voters for him to win the election. The findings show that the most frequently used question particles are le1 and aa3 in the actual debates. These are two neutral question particles used in all types of questions (Matthews & Yip, 2011). Their pragmatic functions are exceptionally crucial in the heated electoral debate as they alleviate the degree of aggressiveness and negative stance communicated through the strategic use of hypophora and question cascade. Moreover, the le1 particle becomes the only question particle that the electoral candidates rely on when their targeted audience is clearly the likely voters alone in their individual speeches. To account for the institutional data of the televised political debates where remote audiences are taken to be part of the communicative equation, a revised version of the Stance Triangle is proposed, which allows us to identify genuine stance objects in the political debate. Discovering the genuine stance objects reveals that the question designs, as well as the use of neutral question particles, are usually oriented to the “remote audiences” (Linell, 2009, p.101), who are playing an indispensable role in the actual debates and individual speeches even they are not physically present in the political debate.