本研究采用定量和定性相结合的方法,旨在探究听力测试中，采用单项选择题的图片选项这种形式对具备不同语言水平的学生产生怎样的影响。北京八中初中分校150名初二年级的学生参加了此次测试。在定量研究中，学生首先对图片型听力试卷进行作答，为避免记忆效应和水平提升的影响，三个星期后对他们实施文字型试卷的测验。这两次测验的分数录入电脑后，用SPSS 15.0 社科软件进行分析，描述性数据和单样本T检验的数据显示，图片型选项的单项选择题对中低水平的学生的听力理解产生阻碍作用，对高水平的学生则没有显著性影响。通过分别对那些对听力理解产生帮助作用和阻碍作用的图片型单项选择题的特征进行分析，结合每道题的难度和所涉及的话题及交际功能特点，结果发现难度系数越高的题对听者听力行为的阻碍作用越明显，这表明听力任务的设计必须反映听力文本的结构难度并且应与听者的语言水平相适合。在定性研究中，通过对学生同步有声思维、追溯回忆及访谈所搜集到的数据用winMax 98 分析发现，单项选择题的图片选项对中低水平的学生回顾对话中的关键词及概念有阻碍作用，对高水平学生无任何影响，这与从T检验中得出的结果是一致的。研究还发现高水平的学生比其他学生对建立声音感知和图片之间的联系更敏感。此外研究表明虽然图片大大减轻了学生的阅读负担，但并没有减轻学生在听力过程中的认知负担，低水平学生甚至认为图片增加了他们的认知负担，使他们在听力过程中不知应该专注于听力内容还是观察图片。由于在听力过程中，高水平学生不需要图片选项来帮助理解，而中低水平的考生又会被图片选项所干扰，本研究对试题开发者的启示是，不要单纯为了卷面效度而使用图片型题型，而应该首先弄清楚听力考试要考查的构念，再决定使用哪种题型来真正达到考查到预先想考查的听力构念。关键字：图片型选项；文字型选项；影响；语言水平

This study investigates the use of visual response options in Multiple Choice listening test items with an aim to test whether presenting the audio text with visual response options or verbal response options generates different results among L2 listeners at different proficiency levels. The study involved 150 junior 8 homogenous students in branch of Beijing No. 8 Middle School. Both quantitative and qualitative research methods are adopted. The quantitative study employs within-group design, with test-retest method adopted, descriptive and t-test statistics are analyzed. While think-aloud protocols, immediate recall protocols, and interview were collected in the qualitative study. Three major findings provide answers to the three research questions. First, it was found that the presence of visualized response options have debilitating effect on low and intermediate-level students, whereas no statistical significance is found for the high level students. Second, visuals don’t play an important role in supporting high level students’ recall of notions and key words, but have slight debilitating effect for low and intermediate level students’ recall of test content. Third, based on think-aloud protocols and follow-up interview, advanced students are able to establish link between sound perception and visuals more easily.Additional finding suggests that while visuals help to reduce reading burden for all the students, they don’t actually reduce the cognitive burden for the high-intermediate group and they could prove to add more cognitive burdens for low level students. Since the study indicates that debilitating effects become more obvious with those test items of high difficulty index, thus it implies that listening tasks should reflect the complexity of audio text and it should be appropriate for students’ proficiency level. The implication for test developers is that it is essential to clearly define the listening construct that one wishes to measure before determining what would be the best test type for the intended test construct.Key words: Visualized response option; verbal response option; effect; language proficiency