The Korean conditional markers myen and tamyen have been distinguished in terms of the speaker’s epistemic stance: while myen can be used with any speaker attitude, tamyen is only used with a hypothetical or irrealis attitude. However, tamyen-antecedents do not always express such an irrealis speaker attitude. This paper analyses the difference between myen and tamyen in terms of the modes of language use, i.e., descriptive and metarepresentational uses. It introduces the concept of interpretive use as defined and used in relevance theory and suggests that the ta in tamyen functions as an interpretive use marker. The presence of this interpretive use marker is responsible for the different felicities between myen- and tamyen-conditionals. Since a tamyen-antecedent is a metarepresentation of another representation, it is not used to describe a state of affairs in the content domain. This explains why deictic, generic, and temporal conditionals are more often used with myen, and given conditionals (i.e., conditionals whose antecedents are contextually given) with tamyen.