All the lonely people: Effects of social isolation on self-disclosure of loneliness on Twitter

Abstract

This study explores the effect of unprecedented mass isolation during COVID-19 lockdowns through the lens of self-disclosure of loneliness on Twitter. Using a dataset of 30 million public tweets, we use machine learning to identify tweets that contain self-disclosure of loneliness. We find that thousands more people turned to Twitter to express their loneliness during the lockdowns; however, this effect normalized within a month, demonstrating the “ordinization” effect on a collective level. Furthermore, lockdown brought a marked shift in the weekly timings of posting and a change in the accompanying emotions, which were more positive and other-focused. Finally, based on a qualitative analysis, we propose an updated typology of loneliness that captures the possibilities offered by the affordances of social media. Our findings illustrate the profound effect lockdowns had on the societal psychological state and emphasize the importance of mental health resources during extreme and isolating events.