This paper attempts to explore whether frequency can play a role in accounting for difficulties in processing English relative clauses (RCs) by L2 learners. By analyzing a corpus of essays written by Korean adult learners of English, this paper shows that the distribution patterns of the RCs in the learner corpus are similar to those in L1 English corpora Specifically, object-extracted RCs almost exclusively occur with inanimate head nouns, while subject-extracted RCs occur about 2 times more frequently with animate head nouns than inanimate head nouns. It is also found that the object-extracted RCs with inanimate head nouns often contain animate subjects and have the animacy configuration reflecting typical thematic structures of transitive clauses. These results are consistent with the experiment findings that the processing difficulty of RCs is modulated by the syntactic function of the head noun and animacy configuration. This corpus study provides support to frequency-based approaches for L2 processing, claiming that a higher frequency of a particular RC form makes it easier for L2 learners to acquire and process.