#frAIday: Using generative pre-trained transformers to track national sentiments

The talk will be held on Zoom. You can participate via the link or watch the lecture in Galaxen at Umeå University. Welcome!  

Many recent political movements have gained traction through their leaders' savvy use of social media to promote narratives about the nation and its identity (e.g., ‘Make America Great Again’). Yet, the effectiveness of such strategies remains unclear. We sought to examine the online and offline success of politicians’ rhetoric expressing different national sentiments. To this end, we validated the use of GPT models to track the presence of different types of national identity rhetoric in social media posts and political speeches. We compared expressions of a positive national identity, which highlight  national pride and attachment to one's country, to expressions of a defensive national identity, which portray the nation as exceptional and entitled. In one application, we coded for the presence of these two types of national identity rhetoric in US and UK politicians’ posts to X (nSt1 = 234,906; nSt2 = 226,900; nSt3 = 296,416). Defensive rhetoric received more likes and more reposts when used by right-wing politicians, but it did not benefit left-wing politicians online and predicted a smaller share of their votes in a congressional election. Expressing a positive identity was linked to more likes and greater vote share for politicians on both sides. The findings can contribute to debates about the role of identity appeals amplified via social media in contemporary politics. More broadly, this work showcases how LLMs can be used to track  psychological constructs in text.