Generic propositions are statements that make general claims about ‘kinds’ that are found in a wide variety of written genres and speech. By definition, generics do not include in their structure any reference to the conditions under which they hold true. Their mis-use in popular scientific writing, however, can erode the public’s confidence in the process of science itself when they discover that conclusions are highly contingent on certain truth conditions. The language used in scholarly scientific papers often includes qualifiers and hedges, the epistemological consequences of which have been explored by Bruno Latour, Thomas Kuhn, Ian Hacking and others. Some research shows that abstracts, however, of- ten include generic statements that are not war- ranted by the scientific evidence described in the full text. Similarly, when accounts of scientific discoveries appear in popular media, journalists of- ten remove qualifiers, hedges and context markers that existed in the original study. Studies in anthropology by Joseph Dumit, Annemarie Mol, Harris Solomon and others explore the human re- actions to such pronouncements. One possible solution to the over-use of generics in scientific abstracts, especially for studies that rely on human subjects, is the inclusion of a mandatory section entitled “Constraints on Generality,” as suggested by Gutiérrez and Rogoff (2003). Other suggestions include using less nominalized verbs and more past-tense descriptions of what actually occurred in the particular study.

Keywords: generics, journalism, linguistics, publishing

Click here for a copy of the preprint.

“This research project analyzes the language used in a corpus of academic papers from 1956- 1976 generally considered to be the foundational documents of the field of artificial intelligence. Thirty-one papers were assembled and examined for evidence of the use of structural metaphors (Lakoff & Johnson, 1980), first manually with an adapted version of the Metaphor Identification Procedure (Pragglejaz, 2007), and then through key-word-in-context searches (Deignan, 2008) with online corpus analysis tool Sketch Engine. Concordance data shows that the scientists frequently used metaphors to make sense of their work. Some structural metaphors used imagery from the same source domain, suggesting underlying root metaphors (Pepper, 1972), evidence of particular perspectives that comes to constitute the academic field. Root metaphors such as A MACHINE IS A BRAIN or RESEARCH IS A JOURNEY were extremely successful in communicating non- observable phenomena between scientists. Other structural metaphors appeared briefly in the literature but soon disappeared from discourse.”

Oliver Selfridge’s “Pandemonium” (1959) metaphor (inspired by Milton’s vision of hell in Paradise Lost) was created as a model for Artificial Intelligence research. It didn’t stick. The depiction above is from Human Information Processing, 2nd edition by Peter Lindsay and Donald Norman, 1977, Academic Press, Inc., New York (via David Bozak)