Although Keenan & Comrie’s NPAH (1977) hypothesis has been shown to predict the developmental order of relatives clauses (RCs) in SLA (subject RCs are easier than direct object RCs, both in production and comprehension), most studies have been based on L2 English and European languages. This study tests the prediction for L2 Korean, which differs in its language properties (cannonical word order, RCs’ position, cue strength etc.) from those languages. Study 1 analyzed RCs in spoken & written data from learners’ and native speakers’. The results of analysis revealed that SU RCs were produced easier than DO RCs in both data, in both group. However, there were also strong associations between SU and animate heads and between DO/OBL and inanimate heads. Study 2 identified these two factors(NPAH & animacy effect), including L1 transfer (English, Chinese, Japanese). The results of a listening comprehension test (involving RCs types, animacy types) revealed that Japanese & Chinese groups found DO RCs easier than SU RCs, but English group found SU RCs easier than DO RCs. Second, the animacy effect in processing RCs is evident in clear difference between sentence types (the number of animate nouns in a sentence). From these results, research suggested that analyse not only grammatical functions of RCs, but animacy, L1 properties (such as head direction, word order, and cue strength) to understand entire processing patterns of L2 Korean RCs, and suggested two hypotheses; H1) in the [-R] type sentences, SU RCs are easier than DO RCs to process, regardless of L1, but H2) in the case of [+R] type sentences, Chinese learners process DO RCs than SU RCs easier, but English learners process DO RCs than SU RCs easier.