This paper investigates variation in knowledge construction in three disciplines---sociology, economics and physics---by drawing on Halliday's notion of 'metaphenomenon'. Specifically, the study analysed the frequency and type of agentive elements (participants and processes) in textbooks to find out the extent to which knowledge in these disciplines is attributed to individual scholars, schools of thought, conventional wisdoms and the like, as opposed to being realised in a non-attributed canonical form. The findings suggest that with respect to the feature of metaphenomenon, economics textbook discourse is arguably more akin to that of physics than its social science counterpart, sociology. The implications of the findings---both pedagogical and ideological---are discussed.
English for Specific Purposes, Vol. 21, no. 4 (2002), pp. 347-366