The purpose of this article is to consider the distribution of the three conditional forms in Japanese: reba, tara and nara. Comrie (1985) claims that a given construction should be identified in terms of prototype rather than in terms of necessary-and-sufficient conditions. Dancygier (1993) classifies the conditionals into two types: predictive conditional and non predictive conditional. The former type is treated as typical conditional sentence.
Here I propose that the predictive conditional as typical conditional sentence carries the following two characteristics:
1) It denotes cause-effect relation between its protasis and apodosis.
2) Its tense form does not designate real time.
These characteristics are sufficiently significant to reflect the distribution of reba, tam and nara in predictive conditional in Japanese. And this is explained by the same motivation as the backshifting patterns in English.
The Linguistic Society of Japan