A corpus-assisted discourse analysis of epistemic stances in tweets about U.S. police from 2013 to 2023

Abstract

Recently, public debate regarding law enforcement practices has extended into digital spaces, particularly on social media platforms such as Twitter (X). Prior research has focused on police-initiated communication both offline and online, yet few studies have explored how the public discusses policing on social media or whether this discussion has changed diachronically. The current study addresses these gaps via a corpus-assisted discourse analysis of a subset of tweets from four U.S. cities (Chicago, Houston, Los Angeles, and Washington, DC) posted between 2013 and 2023 containing the word ‘police’ and the epistemic marker ‘think/thought.’ By examining these tweets, the study analyzes how Twitter users position themselves or others on an epistemic gradient as more (K+) or less (K-) knowledgeable about specific aspects of policing. Using a mixed-methods approach that combines n-gram analysis with discourse analysis of stancetakers and tweet topics, this study identifies how key events shaped Twitter users’ attitudes towards U.S. policing practices over the last decade. Findings indicate that K+ tweets most frequently discussed police services, followed by crime/victims, with particular services like calling 911 and crimes involving vehicles debated by users. In K- tweets, users critiqued others’ knowledge of policing while police services remained the dominant topic with secondary topics like race varying more than in K+ tweets. This study thus contributes to our understanding of public perceptions of policing in online contexts and highlights epistemic stancetaking strategies used by Twitter users to involve others when discussing contentious issues related to law enforcement in the U.S.