摘 要

听力测试作为英语专业八级考试的一个重要组成部分，其构念效度受到越来越多的关注。本研究以南京某高校英语专业大四学生为主要对象，从专八听力部分的对话和新闻两个测试项目探究听力测试的构念效度。因子分析表明：专八听力测试主要考查监听听力材料大意，监听听力材料主要思想，抓取关键信息以及解释总结概括共四项听力技能。回归分析表明：抓取关键信息的技能对专八听力测试成绩的影响最为显着。本研究对今后专业八级听力测试备考乃至听力教学均有一定的借鉴意义。

关键词：听力测试；英语专业八级；构念效度；因子分析；回归分析

Contents

1. Introduction 1
2. Literature review 1
   1. The Definitions of Construct Validity 1
   2. External research 2
   3. Domestic research 3
3. Methodology 4
   1. Research Questions 4
   2. Participants 4
   3. Data Collection 4
   4. Data Analysis 4
4. Results 5
   1. Analysis of Sub-skills in Listening Test through Factor Analysis 6
   2. Comparisons of Significances of four listening skills 7
5. Discussions 10
   1. Possible answer to research questions 10
6. Conclusion 11

Bibliography 13

A Study of the Construct Validity of Listening Test in TEM-8

Introduction

With the rapid development of modern society, communication and cooperation among countries is becoming more and more intimate. English as an official language inevitably plays an indispensable role of bridge. As for English majors, they will have more access to strive for employment in an interview if they obtain a TEM-8 certificate. TEM-8 is a large-scale and high-risk test in the field of higher English education in China. It is widely recognized by society and schools for its authority and impartiality.

Listening test is a part of TEM-8 as well as a part of many language tests, which can assess the listening abilities of English majors. Only if listening test papers has a high degree of the construct validity will it reflect more real language abilities of test-takers. However, English teaching in our country has been affected by traditional grammar teaching and the concept of translation teaching. What is worse, listening has never been put much emphasis and was slighted as Cinderella (Zuo, 2001). Therefore, developing listening ability as an effective method to improve listening is of great importance and value.

The ultimate goal of listening is to have a good knowledge of the comprehensive understanding of sentences and chapters and so on. Nevertheless, it is difficult to improve listening ability without some listening skills being possessed. This thesis aims to clarify specific construct validity and its components in listening test of TEM-8.

Literature Review

• 1. The Definitions of Construct Validity

Zou (2005) claims that to some extent validity is equivalent to construct validity, therefore, a study of validity on TEM-8 should mainly concentrate on its construct validity. Kelly (1927) points out that validity of a test refers to whether the test

measures what it aims to examine. In Hughes’ view (1989), a test is said to be valid if it measures accurately what it is intended to measure. Henning (2001) believes if the test checks what it is supposed to measure, then it is said that the test is valid.

In a word, construct is not directly measured, so the term “construct” can be acknowledged as the term “ability” to refer to the underlying ability required for successful task performance.

If a language test has construct validity, it can measure the testee’s specific language abilities. In accordance with Zou, if a listening comprehension test is in lack of high construct validity, the scores of the test takers could not demonstrate the listening ability, so the scores will lose its effect (1998). It is widely accepted that construct validity is essential for inferences, interpretations or judgments made based on test scores.

• 1. External research

As listening test has played an important role in foreign language testing, listening is actually a positive and active processing of listening material. As for the validity of listening test, various scholars and linguistics have done massive researches. Some scholars describe listening abilities from diverse perspectives, for example, Buck (2001) proposes the construct of listening construct ability from the following three dimensions.

• • 1. process extended samples of realistic spoken language, automatically and in real time; 2) understand the linguistic information that is unequivocally included in the text; and 3) make whatever inferences are unambiguously implicated by the content of the passage. Buck (2001) thinks that construct validity is the most important property of a test, and the basic idea of construct validity is the extent to which the test measures the right constructs.

This definition indicates that listening ability is made up of the ability to comprehend explicitly and implicitly stated information. In this sense, only if we know the listening specific constructs can we measure whether the test is well examined. Therefore, listening abilities include some basic skills, which suggest that

the listening constructs can refer to certain abilities.

• 1. Domestic research

Peng (2011) addressed a specific construct validity issue of TEM-4 listening comprehension. To this end, the author analyzed 120 items by a set of text variables, text-by-item overlap variables and item variables. In conclusion, the pattern of these results was interpreted as favoring the construct validity of TEM-4 listening comprehension. He suggested that teachers need to strengthen the voice training of students and enhance the sense of language. On the other hand, teachers should train students to capture textual cohesion in the process of listening comprehension and analyze its role in the discourse. What the most important is that students must acquire the skill of listening for the implicit information and make inferences and deductions.

Zhang (2015) tried to explore the construct validity of TEM-4 listening test by employing the combination of the qualitative and quantitative analysis. Quantitative analysis is mainly employed to construct hypothetical models by structural equation modeling software Amos 18 to investigate the construct validity of TEM4 listening. It is found that dialogue, essays and news share higher correlation coefficients between each other, which indicate that there is higher-level common factor for the above factors.

Based on Buck’s Default Listening Construct, a model of second language listening ability was developed in Lu’s study. The model was used to investigate the construct validity of the dialogue items in the listening section of Hubei Matriculation English Test. The internal consistency reliability analysis and exploratory factor analysis are employed to provide empirical evidence in support of the hypothesized model. She defined listening ability as the ability to understand explicitly stated information and the ability to understand implicit information to make inferences (2013).