Abstract
Classification into word classes formed the basic descriptive frame of traditional school grammars and was also used also in foreign language teaching. However, the classical categorization based on sets of necessary and sufficient features, sharp boundaries between categories, equal status of individual members, etc. is not entirely adequate for the description of natural languages.
The paper applies newer methods of linguistic categorization based mainly on cognitive approaches and on that basis, word class peripheries are defined as subcategories, which are characterized by properties different from prototype members of the word classes. Better use of these specifically defined subcategories in the teaching of Korean grammar is proposed, which is supported by the experience that for foreign language speakers, active and passive acquisition and use of peripheral cases is more difficult than mastering prototypes.