| In the recent decade, domestic searches based on learner corpus have been flourishing,and this trend has been reinforced by the compilation and publication of Chinese learnercorpora, such as CLEC, SWECCL, which serve as rich database. Literature review showsthat two kinds of learner corpus-based studies—computer-assisted error analysis (CEA)and cross-sectional comparison between native and non-native speakers—still playdominant roles in this realm. Second language development (SLD) exploration is still in itsinfancy, and the research results are not common. The existing developmental studies donein China are, in fact, cross-sectional, due to their corpus-collection methods, which featurethe comparison of students from different grades or with different language proficiency.Thus, the purposes of this research are to remedy this methodological shortcoming, and tooffer empirical contributions for the second language development study in China. Thestudy is based on a self-built learner corpus “The Diachronic Written Learner Corpus ofEnglish Majors”(DLC for short,107,907tokens). From the perspectives of Emergentismand Dynamic System Theory, the lexical and syntactic developments are exploredquantitatively and qualitatively, with a longitudinal span of3school years (6terms), byadopting institutional status as the criterion for advanced learners.The lexical development of the subjects in the corpus is measured from twoperspectives: lexical richness and lexical accuracy. Lexical richness consists of lexicaldensity (LD), lexical variation (LV) and lexical sophistication (LS). Lexical density (LD)measures the proportion of lexical words in all words. Lexical variation is computed bytype-token ratio (TTR). Lexical sophistication (LS) is revealed through “LexicalFrequency Profile”(LFP) which was designed specially for L2learners to work out thepercentage of2000high-frequency word families, Academic Word families (AWL) andword families beyond the two categories (Off-list). In order to eliminate the notoriousinterference of sample size, squared type divided by token is adopted for the fact that thiscalculation has been proved to allow the incorporation of length differences under timedconditions. Lexical accuracy is measured via the standardized frequencies of lexical errorsin each semester. Syntactic development is measured from three perspectives: syntacticcomplexity, accuracy and fluency (CAF). The two fluency indices are mean length per T-unit (MLT) and mean length per clause (MLC), while the four complexity indices areclauses per T-unit (C/T), dependent clauses per clause (DC/C), verb phrases per T-unit(VP/T) and complex nominals per T-unit (CN/T). Syntactic accuracy is also measured viathe standardized frequencies of syntactic errors in each semester.The major findings in this dissertation are summarized in the following four aspects:the overall developments in three school years, the stage features in the development, andthe challenges to previous studies and the relationships among different developmentaldimensions.First of all, the group averages in school years indicate encouraging progress: thecurves for LD, LV, AWL and Off-list families, together with MLT and MLC, show linearincreases, and that for2000high-frequency families shows descending linearity. Thestatistical tests prove that the changes are significant both between the first and last schoolyears, as well as among the adjacent years. These results can be interpreted to indicate theprogress in the use of low-frequency, more lexical and sophisticated words, the reductionin word repetition, as well as the increase in sentence length. The developments in lexicalrichness and syntactic fluency are verified. As far as the lexical and syntactic accuracy isconcerned, though the overall errors in the two aspects tend to decline, indicating theincrease in accuracy, the subcategories in errors move along complicated paths.The second major finding in the dissertation is the dynamic patterns in lexical andsyntactic developments. The shorter-spanned observation in each semester shows thenon-linearity and dynamicity of the indices in lexical richness, syntactic fluency andsyntactic complexity. This unveils the stage-like characteristics of the development: thefirst half (Terms1-3) is more fluctuated and the second part more linear (Terms4-6). Thestage features of lexical and syntactic development can serve as empirical evidence for theexistence of attractor state and repellor state in interlanguage. The developmental paths ofmost errors conform to this two-stage pattern in spite of their much more complicatedchanging trajectories.The three challenges or disagreements to previous studies at home or abroad featurethe third aspect of findings in the thesis.1) From the perspective of dynamic system theory,this finding is lexical development challenges the “plateau” hypothesis put forth bydomestic researches. The “plateau” phenomena are indicators of variabilities and non-linearity, which are regarded as the attractor states and repellor states from thedynamic point of view.2) In other researches on Chinese learners, the slowing-down orstandstill in C/T and DC/C is interpreted as the non-advance of learners’ syntacticcomplexity. Through the quantitative and qualitative analyses of the changes in VP/T andCN/T, the dissertation verifies the phrasal development orientation in learners’ syntacticcomplexity and this can serve as a support for the assumption held by researchers abroad,that learners adopt multiple strategies in syntactic evolution.3) C/T and DC/C are notproved to be valid indices for syntactic complexity for the fact that they cannotdiscriminate year or semester differences.Apart from the above-mentioned developmental dynamic characteristics andcontroversies, the dissertation analyzes the relationships among different measures viadynamic perspectives. There are positive correlations between low-frequency words (AWL,Off-list) and LV, LD. Thus they develop as connected growers, which means the mastery oflow-frequency vocabulary is the prerequisite of repetition avoidance and informationdensity. The syntactic multi-dimensional development is further empirically proven by thehigh and relatively high correlations between MLT and VP/T, MLT and CN/T, MLT andC/T, as well as MLT and DC/C.In spite of a number of insightful findings and tentative conclusions in secondlanguage development, the research, inevitably, is limited and imperfect in more than oneaspect. Learner corpus analysis has clearly only just begun, and there is still much room forimprovement in theory and technology. The data in this study are awaiting furtherverification or falsification due to the fact that they are extracted from only one type ofwritten task. It is hoped that the findings in this longitudinal research, together with thoseobtained from CLEC and WECCL, will help to chart more clearly the development of L2written proficiency within and beyond lexicon and syntax of Chinese English learners. |
PDF Full Text Request