Issue agenda-setting in the voice to parliament referendum: using big data to explain voice discourse on traditional and social media

ABSTRACT: 

This article explores issue agenda-setting in Australia’s 2023 Voiceto Parliament referendum, analysing the dynamics of Yes and Nocampaigns in a diverse media ecosystem. We identify influentialmessages across this complex media environment using softclustering of sentence embeddings from large language models(LLMs). Drawing on 3.3 million documents and over 10 millionVoice specific sentences from mainstream and social mediasources, we identify the narrative configurations of thecampaigns, their audience size and engagement. Overall, the Yescampaign dominated with more diverse messaging and greateraudience size and engagement than No. Yet, the No caseprevailed. We find the narrower, mostly negative No campaignhelped sustain the issue agenda both online and offline, forcingYes to respond to it. The findings also challenge echo chambertheory, revealing most unique topics contained content fromboth #Yes and #No perspectives, indicating robust publicengagement rather than isolated filter bubbles.