The Impacts Of Lexical, Syntactic And Cohesive Features On The Quality Of EFL Writing

In order to study the impacts of lexical, syntactic, and cohesive features on the quality of English writing, we selected compositions from the "Chinese Learner English Corpus"(CLEC) as the research materials.This paper aims to answer the following questions:(1)What are the lexical, syntactic and cohesive features embodied in the compositions of CET4and CET6? Do CET4compositions and CET6compositions have any differences in the lexical, syntactic and cohesive features?(2)What are the impacts of lexical features, syntactic features and cohesive features on writing quality of CET4compositions and CET6compositions? What are the differences?This study includes the following steps:(1)CET4and6compositions scored from6points to15points were randomly selected from CLEC. In each score,20compositions were selected. If less than20, all compositions available were chosen. Finally, altogether173CET4and153CET6compositions were sampled.(2)The three computational tools, Lexical Frequency Profile, Coh-Metrix3.0and L2Syntactic Complexity Analyzer were used to collect the statistics of quantitative features of all the selected composition samples, including lexical features, syntactic features and cohesive features. (3)Finally, SPSS18.0was used to conduct descriptive statistic analysis, independent sample t-test, correlation analysis and multiple regression analysis.The major findings are as follows:Independent t-tests suggest that (1)Among the2indices of lexical features, U index in CET6compositions is significantly higher than that in CET4compositions.(2)Among the15indices of syntactic features,8indices in CET6compositions are higher than those in CET4compositions.(3)Among the11indices of cohesive features,3indices in CET6compositions are significantly higher than in CET4compositions. On the whole, CET6compositions are more lexically diverse, syntactically complex and textually cohesive than CET4compositions, displaying higher writing proficiency than CET4compositions do.Correlation analysis reveals that (1)In terms of lexical features, the scores of CET4have significantly positive correlation with the number of total words and U index; however, the scores of CET6just have significantly positive correlation with U index but no correlation with the number of total words.(2)In terms of syntactic features, the scores of CET4are significantly positive correlated with mean length of clause, complex nominals per clause and complex nominals per T-unit, and have significantly negative correlation with sentence complexity ratio and sentence coordination ratio. The scores have no correlation with other variables. However, the results of CET6are somewhat different. The scores of CET6are positively correlated to verb phrases per T-unit, showing no correlation with other variables.(3)In terms of cohesive features, the scores of CET4have significantly positive correlation with the frequency of means of explicit cohesion, such as argument overlap and stem overlap while the scores of CET6are positively correlated to the frequency of means of implicit cohesion, in particular, Latent Semantic Overlap.The multiple regression analysis shows that lexical, syntactic and cohesive features all have predictive power for CET4writing quality, respectively taking up8.1%,3.5%and25.8%, which can account for37.4%of differences in writing scores. However, for CET6writing quality,30.5%of differences of scores can be accounted for by6.9%of lexical features and23.6%of cohesive features respectively.The paper finally points out that (1)CET6compositions surpass CET4compositions with regard to lexical diversity, syntactic complexity and textual cohesion, exhibiting a significantly higher level of writing proficiency.(2)Lexical diversity and textual cohesion are directly conducive to quality of writing; hence it is suggested that students be encouraged to use a variety of vocabulary and pay attention to cohesion in context.(3)In CET4writings, learners can use stem overlap and argument overlap to enhance the textual cohesion thus improving quality of writing while in CET6writings, learners should lay more emphasis on the use of synonym, antonym and hypernym to achieve cohesion.(4)The quantitative features have a higher predictive power for quality of writing in CET4than in CET6, which indicates that as the writing requirements get higher, the assessment of the writing gets more complex and multi-faceted. Naturally the predictive power of the quantitative features for quality of writing will decrease while the qualitative features such as innovative ideas, logical argumentation, and in-depth reasoning may have a more overwhelming impact on quality of writing.