The Japanese language contains several forms of conditional clauses such as "tara," "to," "ba," and "nara," and numerous researchers have attempted to syntactically and/or semantically describe their differing functions. Prior studies usually focus on the theoretical usages of conditional clauses by examining if interchanging them, changes the sentences' meanings; few studies examine how different conditional clauses are actually employed. To better understand this aspect, we analyzed the usages of conditional clauses in the Corpus of Everyday Japanese Conversation (CEJC, Koiso et al. 2017). In particular, we focused on the two most frequently appearing lexicalized expressions of Japanese conditional clauses in the CEJC – "soushitara" and "sousuruto." In Analysis 1, based on a traditional approach, we examined if the usage pattern of the two forms differed when the states of fact about antecedents varied. In Analysis 2, we selected conditional clauses associated with specific content from the corpus, subsequently examining whether the expressions of antecedents and/or consequents associated with the two conditional clauses differed. In Analysis 3, we reanalyzed data utilized in Analyses 1 and 2 by assuming that conversation is achieved by collaborative acts among agents to understand each other. Results revealed that although they may be syntactically and semantically interchangeable, "soushitara" and "sousuruto" are clearly employed in different ways, and differential usages can be attributed to variations in the states of fact about antecedents. In addition, the results of Analysis 3 provide an innovative way to differentiate between "soushitara" and "sousuruto" where no previous model or theory was applicable.