The Mediation Of Language Proficiency In Syntactic Disambiguation And Syntactic Adaptation

The Shallow Structure Hypothesis proposed by Clahsen and Felser supports the viewpoint that the representations,computed by L2 learners are fundamentally shallower in contrast to L1 learners,especially the grammatical and syntactic ones.While the Declarative/Procedural Model(DPM)put forward by Ullman takes individual differences into consideration and explains that the gap between L1 and L2 may not be qualitative but a quantitative one,which can be shorten or even addressed by efficient practice and improving one’s language proficiency.These two theories clash with each other and the experimental evidence yielded by recent researches is sort of blurred and confusing,which calls more experimental results of different designs in need to support or disprove them.Therefore,the present study aims to investigate how L2 proficiency affects L2 sentence processing by answering how Chinese EFL learners respond to syntactic category and verb subcategorization information in garden-path sentences with subject-object ambiguity during natural reading,whether they can show syntactic disambiguation and syntactic adaptation in a native-like fashion,and how language proficiency mediates the processing patterns.This eye-tracing experiment adopted a 2(language proficiency: L2-H,L2-L)* 3(key word type: Transitive,Intransitive,Noun)two-factor mixed design,with language proficiency as the between-subject factor and sentence type as the within-subject factor.The experimental sentences,comprising 60 sets of sentences(180 in total)all followed by a comprehension question each.Forty-eight participants were recruited and divided into L2-H group and L2-L group according to language proficiency.L2-H group showed stronger sensitivity to verb subcategorization information and stronger competence to use them in processing,and thus stronger tendency to disambiguate and adapt compared to L2-L group.In contrast,L2-L group hardly distinguished the conditions and can only adapt in initial processing but not in the adaptation of reanalysis/disambiguation.The results support DPM.