For peer-production projects to be successful, members must develop a specific and universal language that enables them to cooperate. Complicating the development of language in some projects is the lack of formalized structures (e.g., roles) that communicate to members the norms and practices around language. We address the question of how do role differences among participants interact with the adoption and dissemination of new terminologies in open peer production communities? Answering this question is crucial because we want communities to be productive even when self-managed, which requires understanding how shared language emerges. We examine this question using a structurational lens in the setting of a citizen science project. Exploring the use of words in the Gravity Spy citizen science project, we find that many words are reused and that most new words that are introduced are not picked up, showing a reproduction of structure. However, some novel words are used by others, showing an evolution of the structure. Participants with roles closer to the science are more likely to have their words reused, showing the mutually reinforcing nature of structures of signification, legitimation, and domination.