Year: 2023
Type: article
Study design: Observational Study
Abstract: Social media promotes social connectedness, but social media\nusers can still be lonely which is an important preceding condition\nto various mental health disorders such as anxiety and\ndepression. Here we aim to describe online loneliness in individuals\nfrom the linguistic and social features of their platform\nuse. We define a sample of Twitter users who explicitly report\nbeing lonely and compare their language to a matching random\ncontrol sample. For each user, we create a text embedding - a\nnumerical representation of the content of their online posts,\nexcluding terms and expressions related to loneliness. We utilize\nprincipal component analysis on the resulting embeddings\nto condense the data into a smaller number of variables, while\nstill retaining the majority of the variance. By doing so, we\nare able to position each user within a two-dimensional space,\ndefined by the first two principal components, which capture\nthe most significant amount of variation in the data. Lonely\nindividuals are spatially separated from the control sample, indicating\nthat lonely individuals exhibit distinct language patterns\nthat is often self-referential, e.g. “I should” and “but\nI”. Indicators of online social relations, such as the number\nof online friends, favorites, mentions, show that lonely individuals\nhave fewer social relations, while a sentiment analysis\ndemonstrates that their posts have lower valence. Our results\nprovide insights into the lexical, social, and affective markers\nthat characterize loneliness online, providing a starting point\nfor the development of diagnostics and prevention.
Institutions: Indiana University Bloomington, Indiana University
Language: English
Topic: Mental Health via Writing
Subfield: Social Psychology
Field: Psychology
Domain: Social Sciences
Open Access status: green
OpenAlex ID: W7061450642
