本研究旨在探讨读后续写任务中体裁差异对协同效应的影响。以往研究表明，读后续写任务具有显著的促学效果，学习者在续写时与前文的互动协同是其促学效果的主要来源，而这种协同效应会受到多种因素的影响，阅读材料的体裁就是其中之一（王初明，2010），但以往实证研究中采用的读后续写任务以记叙文为主，鲜有关注体裁差异。因此，本研究考察了读后续写任务中的体裁差异是否会对协同效应产生影响。

研究采用记叙文和议论文两个不同体裁的读后续写任务，围绕以下研究问题展开：（1）在记叙文和议论文续写任务中，学习者续写文本与阅读文本在单词和短语层面是否存在语言协同？若是，两个任务中的协同有何种程度的差异？若否，为什么？（2）两个任务中，学习者续写文本中的偏误率是否存在不同？若是，有何不同？若否，为什么？如果研究以北京某高中一个自然班的38名高一学生为被试，实验持续两周：第一周学生完成记叙文读后续写任务，并填写关于记叙文阅读材料语言难度、主题熟悉度和长度的问卷；第二周学生完成议论文读后续写任务，填写关于议论文阅读材料的问卷，随后五名随机抽取的学生完成关于两次读后续写任务的结构化访谈。问卷数据的分析结果显示，两个读后续写任务中的阅读材料在语言难度、主题熟悉度和长度上均无显著差异，排除了三个因素的潜在影响。为了回答第一组研究问题，单词协同和短语协同的分析分别借助Antconc 3.5.9的Keyword List和N-gram功能，为了细化对短语协同的分析，研究者还手动标注、对比了两个任务中形式协同和意义协同的短语数量；为了回答第二组研究问题，研究者标注了被试续写文本里中国英语学习者常犯的六类典型偏误，并将其分为了形式偏误和意义偏误。

数据分析结果表明：两种体裁的读后续写任务都能够产生协同效应，但议论文续写中协同的单词、短语均多于记叙文续写，被试在记叙文续写中与前文意义协同的短语显著多于议论文，而在议论文续写中与前文形式协同的短语显著多于记叙文。根据访谈数据，两种文体写作意图不同与被试对两种文体的熟悉度不同是导致差异的两个潜在原因。而且，被试在议论文续写中意义偏误的错误率显著低于记叙文续写。形式错误方面，除记叙文续写的时态错误率显著高于议论文续写外，其他的错误类型则无显著差异。总体而言，记叙文续写的语言产出量大，写作流利度高，而议论文续写的协同效应更强，写作准确度高。

本研究结果有助于外语教师和研究者明确不同体裁的读后续写任务对学习者的语言学习，特别是协同效应的影响，为读后续写的教学实践与相关研究提供了一定的借鉴和启示。为了验证本文的研究结果，扩大结论的适用范围，进一步的研究应考虑增加议论文对比续写，引入说明文、应用文等更多体裁，同时关注句法、语篇等其他层面的语言协同。

This study focuses on the effects of genre difference on alignment in reading-writing continuation tasks. As demonstrated in previous studies, the reading-writing continuation task is a valid tool to facilitate and promote language learning and use, and these positive results can be attributed to the interactive alignment between learners’ language output and the input text. The alignment might be influenced by various factors, including the genre of the reading text in a continuation task (Wang, 2010). However, researchers mainly exploited narrations in their empirical studies and only few of them paid attention to the genre difference. Therefore, the current study aims to investigate whether different genres of the reading texts will influence the alignment in reading-writing continuation tasks.

In this study, a narrative reading-writing continuation task and an argumentative one were exploited to explore two sets of research questions: (1) Do L2 learners’ writings show linguistic alignment with the input text at the one-word and phrasal levels in both the narrative and argumentative continuation tasks? If yes, to what extent are they different? If no, why? (2) Do the error frequencies of L2 learners’ writings differ in continuing the two input texts? If yes, how are they different? If no, why? The participants were 38 students in a Beijing senior high school and the data collection spanned two weeks: in the first week, the participants performed the narrative continuation task and accomplished the questionnaire on the narrative reading text; in the second week, they performed the argumentative task and accomplished the questionnaire on the argumentative reading text. Then, five of them were randomly selected for one-to-one interviews. The data analysis of the questionnaires on the two reading texts shows that there is no significant difference on topic familiarity, length and comprehension difficulty between the two reading texts, indicating that the overall levels of the two reading texts were similar and ensuring that students’ performance in two continuation tasks was not influenced by these factors. In response to the first set of research questions, the linguistic alignment at the one-word and phrasal levels has been analyzed with the help of the Keyword List tool and N-gram tool of Antconc 3.5.9, and for a further analysis of the phrasal alignment, the form-aligned and meaning-aligned lexical bundles, have been manually coded and counted. In response to the second set of research questions, the six categories of typical errors of Chinese EFL learners in participants’ compositions have been coded and divided into form-based errors and meaning-based errors.

The results show that the alignment manifests itself in both tasks, but the number of aligned keywords and the number of four-word phrases in the argumentative task are greater than those in the narrative task. It is also found that the form-aligned bundles in the argumentative task are significantly more than those in the narrative task while the meaning-aligned bundles in the narrative task are significantly more than those in the argumentative task. According to the interview data, two reasons including the different writing purposes of the two genres and participants’ different degrees of familiarity with the two genres may account for the difference. In addition, the frequency of meaning-based errors in the argumentative task is significantly lower than that in the narrative task, and as for the form-based errors, except the tense errors of which the frequency in the narrative task is significantly higher than in the argumentative task, there is no significant difference in other error types between the two tasks. In general, the narrative reading-writing continuation task can generate larger amount of language production and better writing fluency, while the argumentative one tends to bring stronger alignment and better writing accuracy.

The study provides language teachers with some pedagogic implications on the application of reading-writing continuation tasks. In order to further examine, consolidate or generalize the research findings, this study recommends that comparative argumentative continuations, other genres like expository and practical ones, and other levels of linguistic alignment such as the syntactic and discourse levels are taken into consideration in the future research.