In spoken Japanese, subordinate clauses often appear without their main clauses. Ohori (1995; 1997) called them suspended clause constructions (SCCs) and formulated that a SCC occurs when “the intended message is either contextually inferable or conventional.” However, it is not very clear when and how participants in the conversation know whether the intended message is contextually inferable (or conventionalized) or not, since an SCC and an "unsuspended version" of a subordinate clause do not constitute not a completely distinct category. Therefore, in order to examine the motivation of SCCs, we need to carefully examine the details of the SCC production process. Based on corpus analysis of natural conversational recordings, I propose to modify Ohori's formulation from an interactional linguistic point of view.1 IntroductionIt is widely known that, in spoken Japanese, subordinate clauses (e.g. kedo - /kara- /node- / noni-clauses) often appear without their main clauses (Martin, 1975; Hinds, 1986). Although they are syntactically incomplete, they constitute a complete statement. For example, in (1), speaker A uses a kedo clause (“however”, “but”) without its main clause. Ohori (1995; 1997) argued that such patterns can be considered independent grammatical constructions in Fillmore's sense. et al. (1988) and called them suspended clause constructions (SCCs). Answer the question “under what conditions can a clause “marked for subordination” not be accompanied by a following main clause? ” (pp.201-202), Ohori (1995) formulated that a SCC occurs when “the intended message is either contextually inferable or conventional” (p.213). From the perspective of construction grammarians, Ohori (1995:216) argued that...... in the middle of the article ......r when and how participants in the conversation know whether the intended message is contextually inferred (or conventional) or not, since an SCC and an “unsuspended version” Subordinate clauses do not constitute an entirely distinct category. Therefore, in order to examine the motivation of SCCs, we need to carefully examine the details of the SCC production process. Based on the analysis of the corpus of natural conversational recordings, I found that it cannot be predetermined whether a subordinate clause is a subordinate clause. CSC or not. Rather, SCCs are carried out retrospectively following interactive negotiations between the participants in the conversation. Thus, I propose to modify Ohori's formulation as follows: a SCC occurs when the fact that the intended message is contextually inferable or conventional is observable interactionally by the behavior of the participants..