Using Digital Humanities for Understanding COVID-19: Lessons from Digital History about earlier Coronavirus Pandemic

Abstract

Background: At the time of the COVID-19 epidemic, it is useful to look at what lessons (digital) history can give us about the past pandemics and dealing with them. We show that the Google Ngram (GNV) can discover hidden patterns in history and, therefore, can be used as a window into history. By using the approach of Digital Humanities, we analysed the epidemiological literature on the development of the Russian flu pandemic for hints on how the COVID-19 might develop in the following years. Objective: Our study is searching for evidence that the COVID-19 is not a unique phenomenon in human history. We are testing the hypothesis that the flu-like illness that caused loss of taste and smell in the late 19th century (Russian flu) was caused by a coronavirus. We are aware that it is difficult to formulate a hypothesis for a microbiological aetiology of a pandemic that occurred 133 years ago. But differentiating an influenza virus infection from a COVID-19 patient purely on the clinical ground is difficult for a physician because the symptoms overlap. The most crucial observation of similarities between the Russian flu pandemic and COVID-19 is the loss of smell and taste (anosmia and ageusia). The objective was to calculate the ratio of increasing to decreasing trends in the changes in frequencies of the selected words representing symptoms of the Russian flu and COVID-19. Methods: The primary methodological concept of our approach is to analyse the ratio of increasing to decreasing trends in the changes in frequencies of the selected words representing symptoms of the Russian flu and COVID-19 with the Google NGram analytical tool. Initially, keywords were chosen that are specific and common for the Russian flu and COVID-19. We show the graphic display on the Y-axis what percentage of words in the selected corpus of books (collective memory) over the years (X-axis) make up the word. To standardise the data, we requested the data from 1800 to 2019 in English, German and Russian (to 2012) book corpora and focused on the ten years before, during and after the outbreak of the Russian flu. We compared this frequency index with non-epidemic periods to test the model analytical potential and prove the signification of the results. Results: The COVID-19 is not a unique phenomenon because the Russian flu was probably the coronavirus infection. Results show that all the three analysed book corpora (including newspapers and magazines) show the increase in the mention of the symptoms loss of smell and loss of taste during the Russian flu (1889-1891), which are today undoubtedly proven to be key symptoms of COVID-19. In the English corpus, the frequency rose from 0.0000040433 % in 1880 to 0.0000047123 % in 1889. The frequency fell sharply after the pandemic stopped in 1900 (0.0000033861%). In the Russian corpus, the frequency rises from 0 % in 1880 to 0.0000004682 % in 1889 and decreased rapidly after the pandemic (1900 = 0.0000011834 %). In the German corpus, the frequency rose from 0.0000014463 % in 1880 to 0.0000018015 % in 1889 and decreased also rapidly after the pandemic (1900 = 0.0000016600 %). According to our analysis of historical records with the approach of GNV, 1) the natural length of a pandemic is two to five years; 2) the pandemic stops on their own; 3) the viruses weaken over time; 4) the so-called herd immunity is not necessary to stop the pandemic; 5) history has shown that a significant crisis does not need to occur after the COVID-19 pandemic. Conclusion: According to our study, the Google Books Ngram Viewer (GNV) gives a clear evidence of the influence that social changes have on word frequency. The results of this study open a discussion on the usefulness of the Google Ngram insights possibilities into past socio-cultural development, i.e. epidemics and pandemics that can serve as lessons for today. We showed hidden patterns of conceptual trends in history and their relationships with current development in the case of the pandemic COVID-19. The benefit of this method could help complement historical medical records, which are often woefully incomplete. However, this method comes with severe limitations and can be useful only under cautious handling and testing. Despite the numerous indications we have shown, we are aware that this thesis still cannot be confirmed and that it is necessary to require further historical and medical research. Keywords: Google Ngram, Big Data, epidemic, COVID-19, Russian flu, Digital Humanities