A Comparative Study Of EFL Writers' Performance On Writing Tasks Specified In Terms Of Words And Sentences

The paper reports on an empirical study which sets out to find out the effect of Length on the EFL students' composing processes and their performance by comparing two different writing tasks, task specified in terms of words and task specified in terms of sentences.In this study we first designed two tasks specified in terms of words and sentences respectively, and then we did the pilot study. Afterwards we invited 219 senior II middle school students in Guangdong province to participate in this study. Among them, 107 were asked to write compositions which required the students to express ideas within certain words, and 112 were asked to write compositions which required the students to express ideas within certain sentences. After the students finished the tasks, four from each group were asked to tell their composing processes.In analyzing the students' compositions we opted for an approach to discourse analysis that involved relatively objective coding of characteristics of the students' written compositions. We first identified relevant indicators of discourse features for analyses, then selected and coded sample compositions, and last analyze similarities and differences in students' performance across the two types of writing tasks.The analysis of the students' performance showed that there were statistically significant differences at text length, lexical sophistication and syntactic levels of discourse. But there were no statistically significant differences on accuracy and coherence. The analysis of the verbal report showed that the students who completed writing task specified in terms of sentences realized their inadequacy in using English, more specifically, in using complex sentences and devices that connect sentences. However, the students who completed writing task specified in terms of words did not realize these. The writing task that specified in terms of sentences was designed for the new National Matriculation English Test (NMET) Guangdong version, 2007 and NMET has strong washback effect. Washback refers to the influence of test on the teaching and learning of English (Bailey, 1996). Since the NMET has strong washback effect, and the analysis showed that the new task type can increase the students' awareness of their inadequacy in using English, the new task type will help the students master English well.