Recently published in the Western Journal of Communication, this OA publication co-authored with Dan Faltesek of Oregon State University offers a hybrid computational-rhetorical take on Trump’s first term. It employs a MALLET (machine learning) topic model to periodize his (almost) full Twitter corpus (c. 2010-2020).
Our key finding is that in different periods of his twitter life distinct patterns can be identified in the post-to-post flow of Trump Twitter, and that these arguments are substantially more coherent and important than any particular tweet.
You can find the full abstract below the link
https://www.tandfonline.com/doi/full/10.1080/10570314.2024.2396506
Abstract: The collective force of ex-U.S. President Donald Trump’s tweets/Xs is palpable in American public culture, political discourse, and academic rhetorical criticism. Adopting a critical and computational approach, this essay offers a novel method for the rhetorical analysis of social media-based public address by shifting emphasis from memorable exemplars of Trump’s social media discourse to the flow dynamics between those tweets/Xs. Focused on loops of Tweets as recursive argument systems, we use a Latent Dirichlet Allocation (LDA) Markov Chain analysis to offer insight into the kairotic and chronic patterning of Trump’s social media utterances to map more and less stable argument strategies across distinct periods of his first Twitter/X presidency.
Atilla Hallsby 