The Language of Russian Fake Stories: A Corpus-Based Study of the Topical Change in the Viral Disinformation

Abstract

The spread of disinformation during the COVID-19 pandemic is largely associated with social media and online messengers. Viral disinformation disseminated in 2020–2021 was related to a wide range of topics that caused panic among people. Many false narratives emerged and attracted public interest over time, which mainly reflected the general public’s utmost belief in these topics. Text mining can be used to analyze the frequencies of keywords and topic-related vocabulary in order to track the changing focus of the public concerning online disinformation. In this paper, we present the results of a corpus-based study of Russian viral fake stories circulating during the first year of the COVID-19 pandemic. We propose a method for analyzing the central topics and dynamics of topical change in the context of the Russian COVID-19-fake story. In order to accomplish this objective, we make use of a set of tools to extract keywords, count their frequencies and analyze corresponding contexts. We apply these tools to the compiled specialized diachronic corpus of Russian viral false COVID-19-related stories. The obtained data is evaluated to determine the dynamic of topical shifts by tracking the changes in keyword frequencies as well as the use of other high-frequency corpus words. The findings of the work concerning topical fluctuations in the Russian viral COVID-19 disinformation agenda as well as given explanations for the identified drifts in public interest in the topics during the first year of the pandemic can contribute to developing effective strategies for combating the spread of fakes in the future.

The words that make fake stories go viral: A corpus-based approach to analyzing Russian Covid-19 disinformation

“…For our previous study of fake narratives about Covid-19 (see Monogarova et al 2021), we compiled a corpus of false Covid-19-related stories that had been virally shared by Russian social media users from March 2020 to March 2021 (hereinafter referred to as Corpus 1). However, over the next year, as the pandemic continued, the accompanying infodemic did not slow down either.…”

NLP4NLP+5: The Deep (R)evolution in Speech and Language Processing

“…The global aim of this series of studies was to investigate the speech and natural language processing (SNLP), research area through the related scientific publications, using a set of NLP tools, in harmony with the growing interest for scientometrics in SNLP [refer to Banchs, 2012 ; Jurafsky, 2016 ; Atanassova et al, 2019 ; Goh and Lepage, 2019 ; Mohammad, 2020a , b , c ; Wang et al, 2020 ; Sharma et al, 2021 and many more] or in various domains such as economics (Muñoz-Céspedes et al, 2021 ), finance (Daudert and Ahmadi, 2019 ), or disinformation (Monogarova et al, 2021 ).…”