| Summary: | This paper deals with the origins of the theory of categorization developed by Eleanor Rosch in the 1970s. It is divided into two parts. The first part describes the theoretical context which underpinned Rosch's initial investigation into the structure of categories. A proper understanding of this context requires that we go back to the research conducted by Lenneberg and his co-authors on Whorf’s principle of relativity. The path which led from this principle to its reformulation as a testable hypothesis is retraced in detail, from early research at Harvard University on the categorization of colours to Berlin and Kay’s book on basic colour terms. For reasons that are explained in the paper, the relativist spirit which prevailed during the initial period gave way to universalism, to which Berlin and Kay’s book contributed in no small measure. The second part narrates how Rosch elaborated her own theory of categorization in this universalist context. As she extended the scope of her theory from colours and spatial shapes to “semantic categories”, so she moved from an “analogical”, Gestalt-like vision of categories to an analysis in terms of discrete features. Her concept of prototype initially incorporated various features borrowed from Gestalttheorie and from the notion of schema, as it had been operationalized in psychology after Bartlett, and became progressively more complex. In the course of this evolution, she remained faithful to a realist conception of categories, envisaged as “natural” clusters of properties. This realist stance of Rosch's built on previous work in several domains, which supplied her with key theoretical constructs such as the cue validity index (originally put forward by Egon Brunswik), and the notion of “structural correlation” (taken from Garner’s informational theory of categorization). As for her views on the different levels of categorization, notably on the basic level, they owe much to Brent Berlin’s anthropological work on folk taxonomies. The influence of Roger Brown is perhaps discernible too, and may have played a role in Rosch’s emphasis on action patterns. Thus, it is shown that Rosch’s theory makes up a “conceptual patchwork” which progressively gained coherence and unity by incorporating constructs of various origins.
|
|---|