Emotional Triggers and Cognitive Manipulation in Romanian Social Media: A NLP Analysis

Abstract

Emotionally charged speech has become a key component of disinformation efforts in the era of social media. To elicit strong emotional responses, speakers strategically employ cultural and historical allusions in their highly engaging media appearances on various internet platforms and so is the case for Romania as well. This work uses Natural Language Processing (NLP) tools to identify and analyze these speeches' emotional signals and language patterns, especially those intended to evoke negative emotional reactions like hatred, fury, and fear. We aim to highlight how historical allusions are used to influence public opinion by analyzing the interaction between quick, automatic emotional reactions (System 1 thinking) and more thoughtful, rational replies (System 2 thinking) when culture and history are embedded. The results of this investigation show how speakers adeptly lead listeners from an initial emotional connection to controversial beliefs that are intellectually defined and frequently aligned with polarized societal or political goals. Our findings contribute to the understanding of how emotionally charged language fosters ideological division and perpetuates disinformation in the digital space. Moreover, the research offers a critical lens on the cognitive mechanisms that make individuals susceptible to such manipulative tactics, providing insights that could inform strategies to counter disinformation and promote more informed public discourse. This paper is part of a broader PhD research program focused on analyzing narratives and narrative strategies in disinformation, with all findings contributing to this purpose.