A Mixed-method Corpus Approach to the COVID-19 Vaccination Debate

Social media have contributed to the recent proliferation of online discussions on the COVID-19 vaccines. The paper explores the evolution of this debate by analysing an ad hoc corpus of tweets (over 5.5 million words) collected from March 15th to April 14th, 2021. We deploy sentiment, emotion, and emoji analysis to uncover the users' affective states, perceptions, and reactions regarding the COVID-19 vaccination. Our results show that vaccine sentiment is influenced by real-time news and by other information that circulates on the Internet, displaying polarizations on both the negative and the positive extremities of the sentiment scale. The emotion analysis indicates that trust issues (either trust or mistrust) regarding the COVID-19 vaccination prevail in our data, amounting to 21.29% of the overall emotional valence of tweets. Furthermore, the qualitative analysis suggests that the infodemic relies primarily on strong negative emotions (e.g., fear, anger, and disgust). Finally, the emoji analysis reveals that, besides iconicity functions, emoji act as boosters of emotions, contributing to the semantic dimension of the Twitter debate on the COVID-19 vaccination.