This paper aims to present a corpus-driven analysis of second-person address terms in Korean. Using the Sejong Corpus data, I examine the usage patterns of the most salient second-person pronouns in Korean, as well as those of other second-person markers including nouns. This paper verifies that the choice among the diverse markers is modulated by various contextual factors, such as level of speech, discourse type, and relation between speakers. In particular, it is demonstrated that the use of the non-prototypical tansin in spontaneous spoken discourse triggers certain pragmatic effects that cannot be achieved by its prototypical counterparts. The identified pragmatic effects include mitigating the threat of criticizing an addressee and signaling the change of the conversation topic by means of calling the addressee’s attention. This phenomenon is accommodated by contemporary pragmatic theories, such as Horn’s (1984, 1989) R-principle or Levinson’s (2000) M-principle. Showing that the grammatical category of pronouns, which is modelled after European languages, does not quite fit the Korean data, I open up the discussion concerning the semantic and pragmatic characteristics of Korean second-person markers, and those of personal pronouns in general.