The aim of this work is to investigate the popularization of science by looking at the differences in the discourse of research articles (RA's) in the field of cancer research and popularizations in news magazines. A corpus of twenty articles dealing with news in cancer research in 1994 was taken from Newsweek, Time and US News & World Report. Firstly, a genre analysis of the popularization is carried out to produce a new genre analysis model of popular bioscience texts, and new definitions of move and genre are put forward. Following this, a new model of popularization, the narrative of society, is proposed. This suggests that the genre of the popular science article is a completely different type from the RA, since it includes two different types of moves, scientific and social. The differences between the popularization and RA genres are demonstrated further by a detailed comparison of sample popularizations with the source RA's. The model is then contrasted with the standard model of the popularization of science, the diffusion model, and various deficiencies of the current model are highlighted. The current model is shown to be not only incompatible with the model put forward here but also to be dependent on a code model of communication which fails to reflect the way popularization is achieved. A non-coding, more cognitive model of communication, Sperber and Wilson's relevance theory (1986/95), is then discussed. This theory is then used to further support the narrative of society model proposed and various difficulties with the use of relevance theory to analyse genres are discussed. The thesis thus a) proposes a new genre model of popular bioscience texts, b) puts forward a substantive new model of science popularization, and c) demonstrates that different approaches to discourse analysis can be combined to yield a richer theory.