This study examined how L2 learners of English process morphologically complex words (e.g., bareness) and psuedowords (e.g.,*boilness) to investigate if the L2 processing of derivational morphology is subject to critical period effects. It has been claimed that native speakers rely on procedural memory to process morphologically complex words, while L2 learners tend to utilize declarative memory to process them (Ullman, 2001, 2004, 2005). According to Ullman, L2 learners’ reliance on declarative memory for processing morphologically complex words is due to maturational changes in procedural memory. Two related experiments with a masked priming lexical decision task were conducted with two groups of Korean learners: early and late L2 learners. With regard to the results, early L2 learners showed stem priming effects for morphologically related prime-target pairs, which is indicative of morphological decomposition. On the other hand, late L2 learners did not provide any evidence of stem priming, indicating that they might have processed the morphologically complex word/pseudoword as a whole unit. These data suggest that the L2 processing of late learners is different from that of early learners.

This study examined processing of L2 English relative clauses by testing the Filler-Gap Hypothesis (Hawkins, 1999; O"Grady 1997). For that goal, we partially replicated Diessel and Tomasello (2004), who examined L1 acquisition of relative clauses by English-and German-speaking children. The Filler-Gap Hypothesis states that the structural distance between head and gap determines the processing difficulty pertained within relative clauses. Taking this hypothesis as a theoretical starting-point, we used an elicited imitation task to tap 48 L2 learners" knowledge of English relative clauses. The results of the study demonstrated that subject relatives retained greater accuracy scores than object relatives, which in turn retained greater accuracy scores than indirect object and oblique relatives. These results were largely consistent with the predictions made by the FGH, but only partially consistent with Diessel and Tomasello"s L1 data.

The present study examines the interpretation of Korean relative clauses (RC) by English speakers of L2 Korean. The purpose of the study is to see if these learners employ the same parsing strategy as native Korean speakers in the processing of a complex NP followed by a RC. Processing strategies were investigated with two different conditions, which were distinguished from each other by animacy presence in the second NP of a complex NP (e.g., chayk-ul ilk-nun apeci-uy atul ‘the son of the father who is reading a book’: [+ani, +ani] condition vs. kyosil-ey iss-nun haksayng-uy chayk ‘the book of the student who is in the classroom’: [+ani, -ani] condition). Korean speakers showed equal preference in the [+ani, +ani] condition, while they showed low attachment (LA) preference in the [+ani, -ani] condition. On the other hand, English speakers showed LA preference in both conditions. We assume that this LA preference by the English speakers might have been due to either the universal processing principle (recency) or influence from their L1, both of which make the same attachment site. The source of the Korean speakers diverging behavior is discussed on the basis of difference in verb meaning used in each condition. The discrepancy between the two language groups leads us to propose that English speakers do not rely on the same processing strategies as Korean speakers.