Although a large number of studies are conducted on Japanese demonstratives, most of them explain referential functions of the three demonstrative types (the so-called proximal ko-, medial so-, and distal a-) based on sentence-level analysis, and little previous work has been directed toward the analysis of the demonstrative use in spontaneous interaction. This study employs Japanese conversational data and examines the demonstrative usages whose main function is NOT to refer to some entity in the speech situation or the discourse. From the analysis, the paper shows that the use of Japanese demonstratives can exhibit and emphasize an interactional meaning, such as the speaker’s antipathy, insult, suspicion, surprise, and affection toward the referent, and that it can be selected from among other choices, such as a noun phrase or ellipsis, when the speaker is willing to express these emotions or attitudes. In order to understand the process of expressing these emotions or attitudes, the paper applies Hanks’ (1990, 1992) ‘indexical framework’ and the interactionally defined notion of the speaker’s and addressee’s sphere proposed by Laury (1997) and Enfield (2003). Using these frameworks, this study illustrates that the relationship among the speaker’s and addressee’s spheres and the referent, as well as the context in which the three are projected, are not static or predefined but instead are flexible and do change during ongoing interaction.