2.2.1 Previous Research Aboard
A large amount of research has been conducted to investigate the effects of frequency on the production of L2 learning; however, only has a few of them are about L2 writing. Most of them focus on speaking skills (Xie Mi, 2009; Zhou Dandan, 2006; Zhou Weijing, 2005).

Perin (2002) investigated the relationship between task frequency and writing. 141 American adult L2 learners participated in the experiment. They were asked to do the same writing twice in two tests respectively. According to the study, their performance on the particular writing improved significantly. Changes in performance were found on four of the five variables across the two rounds of writing task, while there was no improvement in the accuracy of the writings.
Some previous studies, for example, Zhou Dandan (2006) investigated the effects of input & output frequency on story retelling of Chinese students. In Zhou’s (2006) study, 16 English majors were randomly assigned into four groups. The students listened to a story in different input and output frequencies respectively. The results showed that the overall performance achieved an optimal development at the third round of the task. Perin’s (2002) experiment has only been through 2 rounds. Based on Zhou’s (2006) study, this study will set a proper threshold when designing the experiment.

2.2.2 Previous Research in China
In China, Bei Xiaoyue (2009) studied the effects of frequency on L2 writing. The study reported the effects of task repetition, teacher feedback, and proficiency levels on the quality and fluency of L2 writing. The eight-week investigation found that generally repetition of writing tasks contributed to the improvement of writing quality and fluency significantly. However, probably due to the eight-week span and the ways of administrating feedback, the effects of teacher feedback on writing quality and fluency were very weak.

Unfortunately, Bei’s study only analyzed the overall quality and fluency of L2 writing. The changes in some specific aspects of writing such as grammar, lexis, and content, have not been discussed further. Moreover, the subjects of this experiment were junior high school students, the result of which only stood for the writing situation of elementary L2 learners and could not be used as a reference to improve the writing proficiency of L2 learners in the long run, although they were pided into two groups (the high-proficiency group and the low-proficiency group).        

Then, Zhou Dandan (2011) selected 20 English majors. These 20 subjects did the writing task on the same topic for three times during a period of 9 weeks. The results showed that the subjects made changes in their writing including the content, organization, and accuracy. The writing quality also improved as the writing frequency increased. The advantage of Zhou’s (2011) study over other previous studies is that it valued the subjects’ self-proofreading more and teacher feedback was involved in it. However, in many of the previous studies (Chandler, 2003; Chen & Li, 2009; Ferris, 2004; Ferris & Roberts, 2001), “revising” was defined as correcting based on teacher feedback or peer revision. Zhou’s (2011) study will inspire future research, that is, during the repetition tasks of writing, the revision should depends more on writers themselves rather than the feedback of others. Just as what Charles (1990) proposed, revising composition is not just receiving others’ remark and feedback, but the writer’s self-monitoring and reflecting to the writing process.

2.3 Measures of Lexical Richness
Laufer and Nation (1995) introduced some methods for measuring lexical richness, including lexical originality, lexical density, lexical sophistication, and lexical variation. Engber’s (1995) method included lexical variation, error-free lexical variation, lexical error and lexical density. Read’s (2000) measurement consists of 4 indicators: lexical variation, lexical sophistication, lexical density, and a low number of errors. In this paper, Laufer’s taxonomy was employed. Lexical richness in the present study was measured from two aspects: lexical sophistication and lexical variation. 练习频次对写作词汇丰富性的影响(3):http://www.751com.cn/yingyu/lunwen_5174.html