同伴反馈活动被广泛用于写作批改中，关于如何分组的研究存在差异也存在争议。是教师任意把写作文稿分给学生还是按照学生英语水平的高低进行分组才能使同伴反馈的写作者和反馈者最大方面的受益是本文研究的焦点。

本研究目的是继续探讨组内同伴语言水平对同伴反馈的影响。本研究一共有44个高三的学生参与研究，教师根据学生的语言水平能力，将其分为三组，即高高水平组，高低水平组，低高水平组。学生被要求在6周内完成6篇英语作文，每周写一篇作文，然后进行同伴反馈。研究者采用一般线性模型中的单变量分析方法对学生的反馈建议类型情况和结果进行研究，从而得出不同英语水平组对反馈结果的影响。研究者还通过访谈得出写作者和反馈者对自己和同伴语言水平的理解和认可如何影响同伴反馈建议的采用率和建议类型并分析了各水平组提出和采纳不同建议类型的原因。

研究结果表明：高高水平组给予低水平写作者更多的修改意见。但是，不管参与者的英语水平的高低，所有的写作者都采纳了反馈者提供的修改建议，其中较多关注的是词汇的增加，删除和增加从句或增加新的句子到修改稿里的意义相关的变化；单词的拼写，名词单复数，时态，主谓一致等的形式变化为次之，而以词语的选择，主动句到被动句子的变化为主的保留意义的变化机制的修改数量较少。研究还发现，写作者和反馈者对同伴之间的相互信任，影响到了建议类型的数量和采用率。该研究结果在理论上有助于加深教师对同伴反馈活动的理解，思考如何更好的利用学生的同伴反馈提升学生的写作水平。

Peer feedback has been widely used in the teaching of writing. However, there is some controversy about the way of grouping. Some teachers randomly group the students when correcting the students' essays. The focus of the present study is on how to group the students based on the students’ English proficiency to benefit the writers and reviewers in the peer feedback.

The aim of the research is to explore the effects of peers’ language level on the peer feedback. 44 Senior students participated in the research. Students were divided into 3 groups based on their language proficiency , that is, high and high group, high and low group, low and high group. Students were asked to finish the 6 writing tasks within 6 weeks and then participated in the peer feedback. Uni-variate analysis method in general linear model has been adopted to get the results of the different language proficiency of respondents and writers on the peer feedback. Besides, students were interviewed to explore whether the perceptions and recognition of writers and reviewers’ language level influenced the adopted suggestions and analyzed the causes why different language proficiency writers and respondents came up with and adopted different kinds of feedback suggestions .

The following are the research’s results: high level respondents give more suggestions to the low level writers. High and high level , high and low level or low and high level proficiency students all adopt the suggestions given by the reviewers. Most reviewers mainly focus on the meaning-related changes ( the deletion or addition of a word, and adding the new content by adding clauses) , formal changes (grammar, tense, spelling,words and punctuation) but few the meaning- preserving changes. Participant’s perception and recognition of their own or peers’ proficiency has influenced the suggestions and revisions made by the writer. The trust between the peers has affected the type and the number of the revisions. In theory, the results deepen the teachers’ understanding of the peer feedback , and let teachers think how to use peer feedback improve students’ writing ability.