The dominance of English in scientific production raises issues in relation to certain responsibilities of Swedish universities, linked to the dissemination of knowledge and the development of the Swedish language. In light of this, the article deals with Swedish-language summaries (SLSs) in English-language doctoral theses. It treats the SLS as an instrument of language regimentation, deliberately aimed at limiting the near-total dominance of English. Drawing on language policy documents, along with scholarly accounts and interview data, the article discusses the SLS as conceived by advocates in language policy and planning, university policy-makers, and active researchers. It is shown that the SLS is aimed at counteracting negative effects pertaining to knowledge outreach as well as register formation. I argue that there is a contradiction between these two aims: on the one hand, an SLS that is simple enough to bridge the gap between science and society is not likely to contribute to the expansion of advanced registers of Swedish; on the other hand, an SLS that takes seriously the task of expanding Swedish registers will be unintelligible for the wider audience. Yet, it may still serve as a reminder that languages other than English are worthy of consideration and use.