This paper examines whether the concept of C-T inheritance in the sense of Chomsky 2008 and Miyagawa 2010 effective in the Korean agreement system. It further explores whether there is other possibilities than φ-feature inheritance in the C-T relation based on Miyagawa’s (2010) classification of languages into agreement-based languages and discourseconfigurational languages. We also briefly see that Korean subject honorification is a kind of φ-feature agreement, and that even Korean, a discourseconfigurational language, it can also manifest φ-agreement when there is a reason for it to do so. Then, we turn to the problem of the locus of φ-features, which since Chomsky 2008 have been assumed to originate from C. With respect to this issue, this paper argues that the seeminglyφ-feature agreements involving C can be better analyzed under the super-structure analysis for speech act a la Speas and Tenny 2003.