Botaganda: examining how bots shape political discourse on twitter through the lens of interaction alignment

This study examines the influence of linguistic Twitter content posted by bots on the language of tweets posted by human users within political discourse on Twitter (now X), specifically asking whether human users aligned their linguistic constructs about the SNC Lavalin affair in Canada with content posted by bots. As a case study, all tweets containing the hashtag #SNCLavalin were collected between March 14 and April 9, 2019, resulting in a sample of 4,890 tweets. Tweets were segmented into two corpora: one containing tweets posted by bots and a second containing tweets posted by human users. The two corpora were then further segmented into six sets of temporal 12 corpora. Using a computational corpus analysis tool, semtag frequencies were collected for each corpus within the semantic fields “Government & the Public Domain” and “Emotional Actions, States, and Processes.” A linear regression analysis measured predictive relationships between bot-generated and human-generated content, followed by a series of ANOVA analyses and examination of coefficients and residuals. Results revealed that bots significantly influenced the language used by humans in tweets and retweets across the #SNCLavalin Twitter discourse, compared to human-to-human influence. Findings, particularly in emotionally charged contexts, suggest that bots actively shaped conversations and reinforced political echo chambers around the #SNCLavalin Twitter discourse in 2019 more effectively than human interlocutors. These findings underscore the need for further research across diverse political contexts and for implementing policies that regulate the use of bots for political communication within digitized spaces, safeguarding citizens’ rights to independent thought and judgment about important electoral issues.