Drawing data from a variety of sources, this thesis compares functional evidence regarding lexical categories from a number of Salish and Wakashan languages, as well as from the Michif language. It then applies Prototype Theory to examine the structure of the lexicons of these languages. They are described in terms of prototype categories that overlap to varying extents, with each category and each area of overlap defined by a central set of prototypical features. A high degree of gradience appears to exist between categories in Salish and Wakashan languages, with no clear boundary between categories or areas of overlap, indicating that lexical categories in these languages, rather than being clearly demarcated, are instead fuzzy categories with very little distinguishing them. Categories in Michif, on the other hand, exhibit far less overlap. This variation is compared to variation in conceptual categories across languages, and challenges the notions of the universality of clearly demarcated lexical categories and the existence of separately stored language module in the human mind. In spite of the variation in lexical category demarcation observed across the languages studied, it is possible to demarcate the categories of Noun and Verb to at least some extent in all languages, as well as a category of Adjective in some languages. This supports the proposed universality of the categories of Noun and Verb, as well as the implicational universals proposed in the Amsterdam Model of Parts of Speech (Hengeveld 1992a, b). It is also possible to identify a number of defining characteristics for each lexical category that appear to hold across languages. Since similar characteristics can be identified across languages for all categories, but the categories themselves display varying degrees of overlap in individual languages, this research supports the proposal that language universals, rather than consisting of structures, rules, and categories that are identical in all languages, are rather collections of prototypical characteristics for grammatical categories that are similar across languages (Croft 2000).