ABSTRACT
This research studies how Russian authorities disseminated strategic narratives on Twitter from January 1st, 2020, to March 11th, 2021, this last date coinciding with the first anniversary of the COVID-19 pandemic. 111,077 tweets in English from 203 Russian authorities’ accounts, including Governmental profiles, embassies, and diplomats, are considered. To recognize the narratives, a sequential design is implemented. First, Bertopic is used to perform both traditional topic modeling and dynamic topic modeling. The frequency and thematic significance of the 15 most salient topics are analyzed. Second, the strategic narratives’ approach is adopted to identify thematic and structural patterns among the most representative tweets of each topic. The findings disclose that Russia’s efforts to shape perceptions on Twitter are grounded in a gradual and coordinated approach informed by identity and system narratives. They portray Russia as a long-established beneficial force to the world unlike the West, depicted as a hostile antagonist. Issue narratives during the pandemic, such as the supply of health resources, are presented as a logical outcome of this rationale. Thus, these seemingly scattered tweets can be better understood when they are interpreted within the overarching narratives they form over time.
Acknowledgments
I would like to express my gratitude to researcher Guillermo Marco (UNED) and professors Julio Gonzalo (UNED) and Manuel R. Torres Soriano (UPO) for their helpful technical advice on earlier drafts of the manuscript.
Disclosure statement
No potential conflict of interest was reported by the author(s).
Data availability statement
The data that support the findings of this study are available from the corresponding author ([email protected]) upon reasonable request.
Supplementary material
Supplemental data for this article can be accessed online at https://doi.org/10.1080/19331681.2023.2182862
Notes
1. Available at https://securingdemocracy.gmfus.org/category/hamilton-weekly-reports.
2. According to these authors propaganda is “the deliberate, systematic attempt to shape perceptions, manipulate cognitions, and direct behavior to achieve a response that furthers the desired intent of the propagandist”.
3. The tweets were downloaded in two main batches: on December 10th, 2020, and on April 19th, 2021.
4. Available at: https://securingdemocracy.gmfus.org/hamilton-monitored-accounts-on-twitter/#russia.
5. Appendix 1 provides a list of the selected accounts
6. Through word embeddings, words are represented as real-valued vectors that encode semantic information in a dimensional space. In Bertopic, these embeddings are based on Bidirectional Encoder Representations from Transformers (BERT), an advanced machine-learning model that can encode more complex contextual information, potentially enhancing the quality of topic analysis.
7. Term frequency-inverse document frequency is a statistical method to quantify the importance of a word to a document in a collection of documents.
8. The stopwords considered were the default ones from the library Spacy (https://spacy.io/) plus the characters “rt”, “amp”, “re”, “s”, “d”, “ve”, “m”, “ll”, “nt”, and “s”.
9. “covid19”, “covid 19”, “COVID-19”, covid 2019, and coronavirus were merged into covid19, whereas “US”, “U.S.”, “USA”, “U.S.A.”, and “United States” were transformed into “usa”.
10. Lemmatization was dismissed as proved detrimental for the interpretability of the results.
11. Global tuning was set to false. The rest of the parameters were as default.
12. The tweets exposed as examples in the following section have a probability of 1 (the maximum) to belong to their respective topic.
13. Topic −1 was obviated. It refers to all outliers and should typically be ignored (Grootendorst, Citation2020).
14. An interactive version of this figure showing the most representative words each month is provided as supplementary material.
15. An interactive version of this figure is provided as supplementary material.
16. On May 11th Russia reported 11,656 COVID-19 infections, according to the data collected by Johns Hopkins University. https://ourworldindata.org/explorers/coronavirus-data-explorer?facet=none&Metric=Confirmed ± deaths&Interval=7-day ± rolling ± average&Relative ± to ± Population=true&Color ± by ± test ± positivity=false&country=~RUS.
Additional information
Funding
Notes on contributors
Pablo Moral
Pablo Moral is a PhD candidate in Political Science at the Pablo de Olavide University of Seville (UPO) and a predoctoral researcher at the Research Group in Natural Language Processing and Information Retrieval of the National Distance Education University (UNED) (Spain). He was awarded with a grant for doctoral training by the Spanish Ministry of Science, under the project ‘Disinformation & Aggressiveness on Social Media: Bias, Controversy & Veracity’, hosted by UNED. In 2022 he was a visiting fellow at King’s Centre for Strategic Communications (KCSC) at King’s College London. His main areas of research are Computational Social Sciences, digital diplomacy, propaganda and disinformation.