Scope of prototype categorization
A third type of question concerns the scope and applicability of prototype theory as a general explanation of natural language semantics. Most of the original work on prototypes concerned visible categories like BIRD or FURNITURE. In spite of prototype theory accounts of categories like LIE, mentioned above, several scholars have questioned whether the theory is equally justified when applied to abstract, non-visual categories. Since there is not the same perceptual basis for the analysis of attributes, it may be that the diagnosis of prototypes is more hazardous. Lehrer (1990) notes that this question is especially acute for the highly abstract categories expressed in language by prepositions and sentence connectives. Another challenge to the scope of prototype categorization is Wierzbicka’s (1990) critique of Rosch’s assimilation of taxonomic concepts like BIRD to what Wierzbicka calls ‘collective’ concepts, such as KITCHENWARE or CLOTHING. Taxonomic concepts, according to Wierzbicka, have clear boundaries (recall the bat/bird contrast just discussed) and are not open to a prototype account. (On the other hand, collective concepts, which refer to things of many different kinds, are fuzzy, and prototype approaches may well be able to contribute to their analysis.)
In this context, it is worth noting a change in the way Rosch presented the results of her research. Sometimes Rosch presents prototypes as a theory of ‘the nature of the cognitive representation’ associated with category terms (Rosch 1975: 192). Often, however, she stressed the opposite, claiming that prototype theory is not a theory of how the mind actually represents semantic content (Rosch 1978: 40–41; see also MacLaury 1991: 57). For example, she said that ‘facts about prototypes can only constrain, but do not determine, models of representation’ (1978: 40). On this view, prototype theory is a description of the structure of categories which highlights a number of prototype effects – goodness of exemplar ratings, response times, and so on. These effects are, in principle, compatible with a number of different hypotheses on the mental representation of categories, and there is no reason to believe that all words in natural language will correspond to concepts with a prototypical structure.
