Governments worldwide implemented social distancing measures to curb the COVID-19 pandemic, leading to a "war against COVID-19" narrative in media and political discourse. This "new normal," characterized by remote work, telemedicine, and online learning, caused widespread anxiety and psychological distress. Individuals faced stress from confinement, economic losses, and disruptions to daily life. While emotional distress is a normal reaction to such crises, insufficient information and coping resources can exacerbate these issues. Social sharing, the process of discussing emotional events with others, plays a crucial role in coping. Social Network Services (SNS) like Instagram, Facebook, and TikTok experienced a surge in usage during the pandemic, becoming important channels for social sharing and community connection. The study aims to analyze public discourse on social media during the pandemic, utilizing text mining techniques to handle the large volume of data typical of SNS platforms. This approach combines discourse analysis, which examines language use in context and its social implications, with text mining methods to identify patterns and relationships in large text datasets. The focus is on understanding how South Korean individuals coped with and constructed meaning around the pandemic and social distancing through their Instagram posts.

Existing literature highlights the significant psychological impacts of pandemics, including anxiety, fear, and loneliness. Social sharing theory suggests that discussing emotional experiences with others aids in coping and meaning-making, strengthening community bonds. The rise of social media provides a new avenue for this social sharing. Previous research emphasizes the need for analyzing public discourse on social media platforms using computational methods like co-word analysis and topic modeling. Co-word analysis identifies frequently co-occurring words, revealing thematic relationships. Topic modeling extracts latent topics from text data, providing a broader thematic overview. This study draws upon these established methodologies to examine the specific discourse of South Korean Instagram users.

This study analyzed 8241 Instagram posts in Korean, collected using hashtag queries ("#social_distancing" and/or "#everyday_distancing") from May 6 to June 16, 2020, a period of sustained social distancing in South Korea following an initial period of intense restrictions. The data underwent rigorous preprocessing: removing reposted content and posts not containing "Corona" or "COVID." The remaining posts were categorized into three subsets: ANN (official announcements), BIZ (commercial posts), and Everyday-Life (personal experiences). Preprocessing involved tokenization, using the Korean Morphological Analyzer (KOMORAN) to extract meaningful tokens and filter stop words. Co-word analysis was conducted on each subset, creating a co-occurrence matrix and visualizing the resulting network using Gephi. For the Everyday-Life data (n=5155), a co-occurrence frequency threshold of 69 was used. DMR topic modeling, using MALLET, was applied to the Everyday-Life data, incorporating posted dates as metadata. Multiple models (10-15 topics) were compared, and a 13-topic model was selected, which were then categorized into 5 higher-order themes based on thematic similarities. The analysis focuses primarily on the Everyday-Life data, using ANN and BIZ data for contextual interpretation.

The daily distribution of Instagram posts showed peaks coinciding with periods of increased COVID-19 cases, reflecting heightened anxiety and fear. Image descriptions frequently featured "outdoor" and "indoor" settings, revealing people's desire to access nature while also spending time at home. The co-word analysis revealed two clusters: one focused on daily life management during the pandemic, with words like "workout," "play," and "communication," and another centered on the fear of COVID-19 spread linked to the Itaewon nightclub outbreak. The topic model identified five higher-order themes: self-care strategies, fears surrounding COVID-19's re-proliferation and concerns for children's education, recognition of external support, recuperation from intense social distancing, and daily life management during sustained distancing. The Itaewon outbreak led to heightened fear and the targeting of the LGBTQ community with blame and discriminatory language, illustrating how online discourse can reinforce existing societal biases. However, counter-narratives advocating for human rights and the importance of protecting public health were also present. Analysis of temporal changes in topic proportions showed that increased fear initially led to a focus on emotional regulation, followed by a shift towards self-care strategies as fear subsided. In the later period, recognition of external support declined, while the focus shifted towards rebuilding daily lives and utilizing personal resources.

The findings highlight the resilience of South Korean individuals in maintaining daily routines despite the pandemic's psychological impacts. The use of positive emotional expressions suggests a coping mechanism, potentially boosting optimism and resilience. However, the study also reveals how heightened fear and anxiety can lead to the targeting of minority groups, exemplified by the discriminatory discourse directed towards the LGBTQ community following the Itaewon outbreak. This mirrors other studies showing how pandemics can exacerbate existing societal biases and inequalities. The temporal analysis demonstrates the dynamic interplay between fear, social support, and the pursuit of self-care strategies. As the pandemic prolonged, the initial reliance on governmental support diminished, with individuals focusing more on self-reliance and personal coping mechanisms.

This study demonstrates the complexities of individual and collective responses to the COVID-19 pandemic. While social media platforms like Instagram can facilitate resilience through social sharing, they can also amplify existing societal prejudices. To enhance community support and mitigate the negative impacts of prolonged social distancing, long-term psychological interventions and public health campaigns aimed at fostering inclusivity and combating discrimination are crucial. Future research should investigate these issues with a wider range of data sources and specific focus on various demographic groups.

The study's reliance on Instagram data limits its generalizability, as certain demographics may not use the platform. The use of specific hashtags might have excluded relevant posts. Furthermore, the analysis focuses on a specific time frame and event, potentially overlooking broader trends in public perception. Future research incorporating diverse qualitative data, such as interviews or surveys, would provide a more comprehensive understanding.