A Reading-time Based Study Of L2Sentence Gap Processing

For decades, first language (L1) sentence processing has been a popular area ofresearch among linguists and psycholinguists. On the other hand, Second LanguageAcquisition (SLA) research is primarily concerned with explaining how individualsacquire proficiency in a second language (L2) and focuses on both the learning processand individual outcomes (Dussias,2003:531). The divergent goals between SLA researchand sentence processing research, coupled with insufficient technical resources andmethodological expertise among SLA researchers, have kept L2research on sentenceprocessing in a peripheral situation.However, the processing of input materials is crucially involved in the acquisition ofgrammatical system in both L1and L2, because the construction of the target grammarhas to be driven by the input, and the way L2learners process sentences in real time isimportant to a thorough understanding of the development of the L2grammar. Besides,many researchers in L2acquisition (Clahsen&Felser,2006a, b; Felser&Roberts,2007;Juffs,2005; Papadopoulou,2005) have raised the question of whether the incompleteacquisition broadly observed in L2learners is due to non-optimal parsing mechanismsemployed by the learners.The present study arises out of the concern over the current situation in L2sentenceprocessing studies. Its objective is to investigate Chinese EFL learners’ parsingmechanism by examining their online reading of filler-gap dependencies that containIntermediate Gap and Subject Filled Gap, compare their parsing performance with that ofnative speakers of English, explore the role of proficiency levels in L2sentenceprocessing, evaluate the existing theoretical models and seek alternative explanations forthe L2sentence processing that bears insight to SLA research and foreign language teaching.Prichett’s parsing theory which is based on Principles and Parameters frameworkand Clifton&Frazier’s Active Filler Hypothesis provide the theoretical framework forthe present study, within which the experimental sentences are designed and theassumptions and implications thereof are drawn.The empirical study consists of two parts, with the first part testing IntermediateGap Effect and the second part the Subject Filled-Gap Effect. To explore how L2learnersprocess the two kinds of syntactic gaps, experimental sentences used for both parts aredesigned in such a way that the syntactic gap processing could be separated fromlexical/semantic association. The two parts of the empirical study follow roughly thesame procedure.The participants of the present study are second-year university students grouped byCET4scores into two proficiency groups, the High Proficiency Group and the LowProficiency Group, with35students in each group. Thirty-two native speakers of Englishas a control group also participated in the present study. Each participant was asked tocomplete a working memory test at the beginning of the experiment, the purpose ofwhich was to provide data for the investigation of possible effect of working memorylevels on their online reading performance.The methods employed in the empirical study included both an offline and an onlinetask. In the offline task, the participants were asked to read sentences that had beenadapted from the online reading task and answer the comprehension questions after eachsentence. The purpose of the offline task was to make sure that all participants were ableto comprehend the grammatical structures under investigation. The online reading taskwas implemented three weeks after the offline task and the technique involved is calledSelf-Paced Reading (SPR), which collects both the online accuracy scores (for the onlinecomprehension questions) and the reading time of each segment of the experimentalsentences. The principal results of the empirical study are summarized as follows.(1) There are qualitative differences between L2learners and native speakers of English in terms of the parsing mechanism employed during the online reading task.Native speakers were sensitive to the syntactic information conveyed by certaingrammatical words and were able to make use of them in building the underlyingsyntactic representation. However, L2learners failed to employ syntactic gap analysislike the native speakers and mainly relied on the lexical/semantic information conveyedby the content words.(2) There are discrepancies between L2learners’ grammatical knowledge and theirparsing mechanism. The learners from the two Proficiency Groups achieved high scoresin answering the comprehension questions in both the offline questionnaire and theonline reading task, indicating that both groups possessed the necessary lexical andgrammatical knowledge to arrive at the correct interpretation of the experimentalsentences. However, unlike the native speakers, they were unable to make use of thesyntactic information conveyed by certain grammatical words to facilitate the full parsingof the experimental sentences.(3) The L2learners of the two proficiency groups performed similarly in theexperiment testing Intermediate Gap Effect in that they both failed to make use of theintermediate gap for filler integration. However, they differed in the experiment testingSubject Filled-Gap Effect in that learners from the High Proficiency Group showed asimilar pattern with native speakers while learners from the Low Proficiency Groupfailed to posit a subject gap.The main contributions of the present study are summarized as follows.(1) The present study sheds light on the development of new theories accounting forL2processing. The current theories, such as Input Processing Theory (Van Patten,1996,2000,2002,2004), Declarative/Procedural Model (Ullman,2001,2004,2006) andShallow Structure Hypothesis (Clahsen&Felser,2006a, b), have offered littleexplanation for the differences in processing difficulties between different grammaticalstructures during L2sentence processing. In light of the findings yielded from the presentstudy, the researcher has contributed to the further development of SLA theories by putting forward a tentative model “Gap Accessibility Hierarchy” hypothesis to accountfor the fact that learners from the High Proficiency Group adopted different parsingstrategies in processing the two kinds of syntactic gaps.(2) The present study has recorded some improvement in research methodology onL2sentence processing. Firstly, the present study has included Proficiency Levels as abetween-subjects factor in the experimental design. This departs from the pattern ofprevious studies that only selected subjects of similar proficiency level, thus paving theway for further research on the developmental pattern of L2learners’ parsing mechanism.Secondly, with the design of experimental sentences improved, the present study hasmade a successful distinction between gap processing and lexical/semantic association,thus more accurately revealing the picture of L2sentence gap processing.(3) It has pedagogical implications. In view of the discrepancies between L2learners’ grammatical knowledge and their parsing mechanism as found in the presentstudy, the researcher has posited a need to help the learners change their non-optimalparsing strategies through classroom instruction and explores the feasibility of“Processing Instruction” in China.