Previous researches have shown that subject relative clauses are easier to process than object relative clauses in many languages. Roland et al. (2007), however, claimed that the processing ease of subject relative clauses does not depend on structural complexity or linear distance, but it depends on the difference of discourse function between subject and object relative clauses. They confirmed the appropriate context reduced the processing load of object relative clauses. In this paper, we examined the validity of the discourse-function account (Roland et al. 2007, 2008) using a combination of a corpus and an experimental research. The results showed that the discourse-function account was not successful in explaining Japanese data. This suggests that the reduction of processing load in Roland et al. (2007) might be derived from the repetition of noun phrases rather than the relative clause processing itself.