A Study Of The Lexical Richness Of TEM8 Compositions

This thesis aims to study the lexical richness of the compositions in Test for English Majors Band-8 (TEM8). With the development of English teaching and learning, writing has become an indispensable part in English test as first language (L1) and second language (L2). Generally speaking, the lexical richness of writing is regarded as an important predictor of learners’language proficiency and has been receiving great attention by second language researchers in recent decades. However, there is no consensus on which parameters should be applied to measure lexical richness, and what the formulas of these parameters are. Besides, these findings on lexical richness are different.Based on Read’s theory (2000), this thesis is to have a tentative study of the relationship between the writing quality of TEM8 compositions and their lexical richness in terms of five dimensions: lexical variation (LV), lexical density (LD), lexical sophistication (LS), lexical error (LE) and synonym (Syn). Three major questions to be studied are listed as follows:(1) What are the general features of lexical richness? (2) What are the differences among three groups in terms of five parameters? (3) Which parameters are the main factors influencing writing quality? The samples consist of 72 examination compositions in TEM8, which are collected by the Corpus of English Writing (CEW). Firstly, according to their test scores, these 72 samples are divided into three groups. Those whose scores ranging from 14 to 16 belong to high group (HG), those whose scores between 11 and 13 are intermediate group (IG), and those whose score ranging from 7 to 10 are low group (LG). Secondly, the software Range 32 is used to process the samples, which produces the lexical frequency profile, the numbers of content and function words; the number of lexical errors and synonyms are manually counted; the values of five parameters are produced by computer based on their formula respectively. At last, statistical analysis is carried out by the computer software SPSS 17.0.The research yields the following findings:firstly, broadly speaking, compared to native speakers, these students overuse high-frequency words; besides, LV and LS cannot differentiate the writing quality so well, LE and Syn do well in distinguishing the language proficiency level. Secondly, lexical variation, lexical density and lexical sophistication in three groups are not significantly different, while, lexical error and synonym in these groups are significantly different. Thirdly, in HG, two parameters:lexical sophistication and lexical error together determine 46.6% change of the final score. In IG, lexical error and lexical density predict the change of writing quality significantly, it is 31.7%, however, there is no parameter can predict the writing quality significantly in LG, five parameters can only explain 12.8% change of writing quality. Among five parameters, lexical error has a negative effect on writing quality of three groups; at last, there are interactions among the five parameters. The above results have shed light on the lexical richness in TEM8 compositions. It tells us lexical richness cannot explain all the change of the writing quality, other factors influence the writing quality to a certain degree.Due to limited time and length, the collected essays are of a limited number; there is no cross-sectional or longitudinal study. Hence, the further studies should make an improvement in these respects.