Connecting the pro-recovery eating disorder community: An analysis of the language on science, Twitter, and Reddit.

Abstract

The current article investigated and described differences in online, pro-recovery communities' linguistic themes extracted from online messaging. More specifically, the authors examined language alignment between academic journal abstracts (n = 9,744), Twitter influencers (n = 43,384), Twitter organizations (n = 52,748), and Reddit posts (n = 73,628) related to aspects of eating disorders (EDs). Natural language processing techniques (i.e., the meaning extraction method) along with a principal component analysis (PCA) were used to define themes and create novel Linguistic Inquiry and Word Count (LIWC) dictionaries specific to each platform and mental health more broadly. Using these dictionaries, along with others in LIWC previously associated with aspects of EDs, we evaluated the degree to which each platform's language reflected each dictionary. A multitude of findings arose using both common and novel LIWC dictionaries between scientific abstracts, Twitter influencers' tweets, Twitter organizations' tweets, and Reddit posts. These differences spanned language constructs, such as body, health, positive emotion, negative emotion, mental health, Twitter language, and Reddit language. These linguistic differences suggest that each platform has its own purpose and function, such that they may be considered as smaller, specialized communities within the larger pro-recovery context. Although each platform functions under different constraints, leaders and community members should project messaging that promotes recovery at the broadest level. Community leaders should take advantage of preexisting frameworks to connect mental health professionals and individuals in the recovery process. Public Policy Relevance StatementThis article reports analyses of the language used across the online pro-recovery eating disorder communities. Language from several different sources (academic journal abstracts, Twitter influencers, Twitter organizations, Reddit) differed on themes, language indicators relevant to ED recovery (i.e., body, positive emotion, negative emotion), and words related to mental health more broadly. These findings support the notion that each source serves its own function in a smaller specialized community within the larger pro-recovery community.

A search for commonalities in defining the common good: Using folk theories to unlock shared conceptions

“…LIWC's count-based algorithm identifies the most prominent words in a set of documents, after which the researcher can cluster them into themes by statistically analysing their co-occurrences using dimension reduction techniques (Blackburn et al, 2018;Chung & Pennebaker, 2008;Freyberg et al, 2014). MEM shares similar features with distributional semantic models which assume that the proximity between linguistic items can be quantified and categorized, such that different words belonging to the same theme are frequently used together and will thus co-occur in text ( Juel et al, 2023;Tausczik & Pennebaker, 2010). Due to its ability to parse vast amounts of text data efficiently and reliably, LIWC enables social psychological researchers to examine cultural patterns of thought and communication embedded in language to identify and understand 'folk concepts' such as the common good (Chung & Pennebaker, 2008).…”