Language processing heuristics are one of the possible sources of divergence between first and second language systems. The Shallow Structure Hypothesis (SSH) (Clahsen and Felser, 2006) proposes that non-native language processing relies primarily on semantic, and not syntactic, information, and that second language (L2) processing is therefore necessarily less sensitive to syntactic constraints than native language (L1) processing. The SSH further predicts that regardless of whether a participant’s L1 and L2 instantiate the same structure, L2 processing will always be less sensitive to structural constraints. The present dissertation tests these claims in non-native Japanese processing. L1 English and L1 Korean speakers completed a self-paced reading task that tested their ability to rely on case particles to project clause structure in relative clauses, their sensitivity to wh- dependencies, and their ambiguity resolution preferences. Results suggest that L1 English and L1 Korean speakers rely on case particles to project structure, but that they diverge from native Japanese speakers with respect to whether projecting a second clause facilitates the processing of the head noun of the relative clause. Results also suggest that both groups of L2 Japanese speakers are sensitive to wh- dependencies in canonical wh- biclausal sentences, but only the L1 English group is in the scrambled ones. Ambiguity resolution preferences for both L2 groups converge on native-like preferences. These results are incompatible with the predictions of the SSH. The role of L2 literacy and syntactic knowledge in language processing is also discussed.
Read
