Linguistic Analysis of Online Group Communication Reveals Insights into Collective Values and Behavior

New research conducted by psychologists from the Universities of Exeter, Bath, and Lancaster has revealed a novel approach to understanding social groups and their dynamics. 

By examining the language style used by online communities, researchers can gain valuable insights into the values and principles held by these groups. The study sheds light on how groups view themselves and how their values evolve over time, offering social scientists a new tool to investigate group behavior and dynamics.

Moreover, it holds the potential to assist authorities in tackling detrimental behaviors like the propagation of conspiracy theories and enhance predictions of radicalization and criminal activities within groups.

Dr. Alicia Cork from the University of Bath explained how the team approaches the study. 

"The study of group dynamics often involves asking people to tell us about their group – for example, what does being an England football supporter mean to you?" Cork asked. "The drawbacks of this, in simple terms, is that these questions can be quite ambiguous to answer, and it is tricky to articulate what particular group memberships may mean to a person."

"However, when you concentrate on the style of communication used by group members, you can access the overarching principles that guide the behavior of those members," Cork said.

Supported by funding from the Engineering and Physical Sciences Research Council, the doctoral research encompassed three distinct phases. 

In the initial stage, the researchers aimed to determine if groups with similar characteristics exhibited similar communication patterns. They selected three online Reddit communities for each of the five group types: vocational, political, religious/ethnic, relational, and stigmatized. For instance, within the relational category, they chose communities focused on fatherhood, motherhood, and general relationships. As for the political type, a political party (Conservative), a social movement (Feminist), and an ideology (Libertarian) were selected. In the stigmatized category, they included a transgender community, a community addressing alcohol abuse, and one discussing homelessness.

Over the course of a year, the researchers collected Reddit comments from these 15 online communities. The gathered text underwent analysis using specialized software, which assessed elements such as formality and the emotional or reflective nature of the language. By comparing the writing styles of the groups, the researchers sought to understand the extent of their similarities or dissimilarities.

Upon graphing the groups' relative positions, the researchers observed that groups tended to cluster closely with their counterparts, indicating that similar groups indeed displayed similar communication patterns. However, an exception emerged with communities associated with stigma, as they exhibited more significant variation in their communication styles. In the second phase of the study, the research team delved deeper into the relationship between a group's communication and its explicit values. 

By analyzing posts based on ten well-established values such as benevolence, security, conformity, power, and achievement, they gained insights into how the groups likely depicted themselves. According to the University of Exeter, these findings were then compared to their respective clustering positions, providing a comprehensive understanding of the interplay between communication style and group values.

Dr. Miriam Koschate-Reis, Associate Professor of Computational Social Psychology at the Institute for Data Science and AI (IDSAI) at Exeter, discussed the groups in more detail.

"The results indicated that our vocational identities such as entrepreneurs and white-collar salespeople expressed high levels of achievement but fairly little benevolence," Koschate-Reis said. "Groups that expressed benevolence were consistent with the relational types – mothers, fathers, and those focused on relationships. Interestingly, we also found that while the political and religious/ethnic groups tended to espouse conformity, the latter tended to express greater benevolence in their language."

In the final stage of the project, the researchers investigated whether this linguistic analysis could track the evolution and development of a group's collective self-understanding over time. Focusing on transgender identity, they analyzed Reddit data dating back to 2011 and observed a migration away from stigmatized groups discussing homelessness or alcohol abuse toward communities defined by collective action, such as political and religious/ethnic groups. This shift corresponded to the gradual politicization and de-stigmatization of the transgender group, mirroring the progress of LGBTQ+ rights during that period.

Prof. Mark Levine of the Department of Psychology at Lancaster shared more changes seen in the study.

"The changes we observe in the linguistic style of this transgender group appear to correspond with the gradual politicization and de-stigmatization of that group," Levine said. "And this is consistent with what we know of the development of LGBTQ+ rights over that period."

The researchers argue that this linguistic method could aid in identifying emerging political movements, including those that initially manifest as mutual support groups but eventually become radicalized. Additionally, it could help uncover groups attempting to conceal their true nature, such as criminal organizations operating on online forums.

Prof. Richard Everson, a professor of machine learning at the University of Exeter, added the results of their findings.

"Understanding what groups stand for is integral to a diverse array of social processes, ranging from understanding political conflicts to promoting public health behaviors," Everson adds. "The results from these three studies provide compelling evidence that suggests that linguistic style can be used to understand the collective self-understanding of social groups."

The research, titled "Collective self-understanding: A linguistic style analysis of naturally occurring text data," has been published in the journal Behavior Research Methods. The study was conducted by Cork, Koschate-Reis and Levine. The project received funding from the Engineering and Physical Sciences Research Council.