The present study argues against the predominant view in the literature that the default semantic denotation of common nouns in numeral classifier languages,i.e., table, is similar to the denotation of the mass nouns in English, i.e.,furniture. Based on this mass noun view, many linguists argue that there is no count-mass distinction in numeral classifier languages. I show that the mass noun view based on simple morpho-syntactic analogies between English and numeral classifier languages has a number of problems both in theory and in the interpretation of the data. I provide evidence for the existence of a count-mass distinction in numeral classifier languages from various areas of grammar based on the Korean data and supporting evidence from the researches in other disciplines. My study will be able to contribute to the renewed and correct understanding on the nature of common noun semantics of the numeral classifier languages.