On the use of Jargon and Word Embeddings to Explore Subculture within the Reddit’s Manosphere

Abstract

Understanding the identities, needs, realities and development of subcultures has been a long term target of sociology and cultural studies. Socio-cultural linguistics, in particular, examines the use of language and, in particular, the existence and use of neologisms, slang and jargon. These terms capture concepts and expressions that are not in common use and represent the new realities, norms and values of subcommunities. Identifying and understanding such terms, however, is a very complex task, particularly considering the vast amount of content that is currently available online for many such groups. In this paper, we propose a combination of computational and socio-linguistic methods to automatically extract new terminology from large amounts of data, using word-embeddings to semantically contextualise their meaning. As a use case, we explore subculture on the platform Reddit. More specifically, we investigate groups considered part of the manosphere, a loose online community where men's perspectives, gripes, frustrations and desires are explicitly expressed and where women are typically targets of hostility. Characterisations of this group as a subculture are then provided, based on an in-depth analysis of the identified jargon. CCS CONCEPTS• Computing methodologies → Information extraction; Topic modeling; • Theory of computation → Random projections and metric embeddings.

Data for: "Are Anti-Feminist Communities Gateways to the Far Right? Evidence from Reddit and YouTube"

“…4 Within the Manosphere, we observe that different communities have different relationships with the Alt-right. This finding not only resonates with previous work differentiating these communities by their linguistic traces [13], but also suggests that not all Manosphere communities are equally problematic. We find that, on both Reddit and YouTube, Men Going Their Own Way (MGTOW) and Men's Rights Activists (MRA) were consistently linked with the Alt-right, both in terms of their user-base similarity and in terms of the migratory patterns observed.…”

“…The Manosphere and communities within it have been previously studied in Reddit, YouTube, and standalone websites. Farell et al have explored the use of misogyny across Manosphere-related subreddits [14] as well as extensively characterized their jargon with computational and socio-linguistic techniques [13]. Ribeiro et al [40] have mapped the evolution of the Manosphere across Reddit and several standalone websites, showing that newer and more toxic communities such as Men Going Their Own Way and Incels are overshadowing older communities such as Pick Up Artists and Men's Rights Activists.…”

Making a Radical Misogynist: How online social engagement with the Manosphere influences traits of radicalization

“…In their observational analyses of the Manosphere, Farrell et al show increasing hostility and violence towards women online. Driven by lexical analysis, their work demonstrate the Manosphere use of 'flipping the narrative' technique and playing the victim in online communities to legitimize their hatred towards women [8,91]. These findings support our results in §4.1 where we observe significant increase in out-group construction, toxicity and anger after participation in the Manosphere communities.…”

Obtaining Better Static Word Embeddings Using Contextual Embedding Models

“…We also refer the reader to this article 2 illustrating several down-sides of using BERT-like models over static embedding models for non-specialist users. Indeed, we can see continued prevalence of static word embeddings in industry and research areas including but not limited to medicine (Zhang et al, 2019;Karadeniz andÖzgür, 2019;Magna et al, 2020) and social sciences (Rheault and Cochrane, 2020;Gordon et al, 2020;Farrell et al, 2020;Lucy et al, 2020). From a cognitive science point of view, Human language has been hypothesized to have both con-1 BERT base (Devlin et al, 2019) produces 768 dimensional word embeddings using 109M parameters, requiring 29B FLOPs per inference call (Clark et al, 2020 textual as well as context-independent properties (Barsalou, 1982;Rubio-Fernández, 2008) underlining the need for continued research in studying the expressiveness context-independent embeddings on the level of words.…”