Contemporary algospeak (algoridiome): theoretical contributions to a neology of obfuscation and connivance

Abstract

The article offers a theoretical account of algospeak, understood as a set of linguistic practices aimed at bypassing algorithmic moderation on digital platforms. It shows that these practices rely on processes similar to those found in slang or jargon, but adapted to specific technical constraints. Based on a corpus combining observations and a questionnaire conducted with adolescents, the study proposes a typology of algogisms (graphical alterations, metaphors, euphemisms, emojis, borrowings, etc.). Algospeak thus appears as a response to a form of algorithmic control that is difficult to grasp, prompting users to anticipate the risks of censorship. These practices rely on the ability to adapt language to the assumed rules of platforms, as well as on a shared implicit understanding among users. They are also characterized by their instability: forms evolve rapidly, disappear, or sometimes become integrated into everyday language. Finally, the article highlights major ethical issues: while algospeak makes it possible to maintain expression under constraint, it can also attenuate or even trivialize sensitive realities, thereby influencing how they are perceived and discussed.