An analysis of journal articles from physics and molecular biology carried out with the help of six specialists in those disciplines reveals that scientists read journal articles by searching for the most newsworthy information, a behavior similar to that of newspaper readers. For this reason the scientific journal article is gradually taking on text features that promote surprise value, conventions analogous to those of news reports. Titles have become more informative, section headings and subheadings are more frequent, abstracts (which enable writers to foreground their most important claims) are becoming standard, visual aids are more rhetorically focused and attention-getting, introductions more frequently contain results, and methodology sections are de-emphasized, with background information often published in a separate genre--the review article. To meet the scientific community's imperatives of factuality and personal credibility, such fast-paced surprise-value oriented modifications of the traditional layout are offset by the discussion section. Here the writer restates the results, redescribes the gap in knowledge filled, compares findings with earlier results in a selective literature review, proposes implications for larger issues, and makes acknowledgements. Thus situating his novel findings within the body of knowledge previously accepted, the scientist reasserts his membership in the scientific discourse community. Helping students see such form-function correlations of discourse conventions is one way to "demystify" academic discourse. (JG)