Digital Diplomacy in the Age of COVID-19Strategic narratives and propaganda spread by China, Russia, the US, and the EU on Twitter

Abstract

The COVID-19 pandemic led to a "battle of narratives" among the world's major powers. This doctoral thesis studies how China, Russia, the United States, and the European Union used Twitter to influence the perception of international audiences. Through a multidisciplinary approach, it explores the digital diplomacy carried out by 163 Chinese, 217 Russian, 314 American, and 291 European authorities between January 1, 2020, and March 11, 2021. These accounts include the Twitter profiles of presidents, foreign affairs ministers, ambassadors, consuls, embassies, consulates, and other diplomatic missions. On the one hand, this research examines the spread of strategic narratives on Twitter using two computational methods: topic modeling, which extracts the most discussed topics by the authorities; and social network analysis, allowing exploration of content dissemination and identification of the most influential agents. The results help to understand the formation of narratives on Twitter, confirm the coordination among authorities, and shed light on the strategic nature of these narratives. For instance, China and Russia used the pandemic to position themselves as alternatives to Western powers, the EU focused on repairing its reputation, and the US was hindered by the confrontational style of President Trump. On the other hand, this thesis analyzes the use of propaganda techniques by these actors and explores the design and viability of an automated tool to identify and categorize propagandistic language on social media. A novel taxonomy of propaganda techniques based on rhetorical features is proposed, and two datasets containing tweets from the authorities in English and Spanish are created. A content analysis is then conducted to manually classify the tweets according to this taxonomy. This classification not only enables an examination of the systematic use of propagandistic language but also facilitates the training of machine learning systems. Finally, this thesis presents DIPROMATS, a natural language processing challenge that lays the groundwork for artificial intelligence systems capable of detecting and categorizing propaganda in messages from public authorities.