My thesis is designed to see how adult second language learners (L2) process compounds containing plurals in non-head constituent (e.g., rats eater). These compounds have been well-studied in first language (L1) acquisition but not in L2 acquisition. They are interesting for several reasons: there is an interesting constraint such that the plural-marked noun can be irregular but not regular (mice eater vs. *rats eater); they are infrequent; Korean lacks plurals inside compounds like the English ones. They thus lend themselves to the study of three claims: 1) the ability of learners to process novel information is limited by how L1 grammatical knowledge is represented and processed; 2) L2 learners are nonetheless able to represent new linguistic distinctions; 3) the frequency of constructions in the input also constrains L2 learning. A goal is therefore to determine what Koreans learn of this construction and which acquisition mechanisms explain what they know.