Empirical Reforms on New―CET4 Communicative Listening Test

[a] Yangtze University College of Arts and Science， Jingzhou 434020， Hubei， China.

Corresponding author.

Received 19 November 2012； accepted 2 February 2013

Abstract

Following An Empirical Study on Problems Involved in CET4 Communicative Listening Test and Teaching. This article also selects two representative classes of sophomores coming from Hubei University of Education as the sample to complete the empirical research on what necessary renovations should be taken place in New-CET4 communicative listening test in the methods of quantitative and qualitative comparative analysis with the help of the instrument SPSS （Statistics Planning of Social Science）. The research findings prove that such reformations should be employed in the traditional New-CET4 listening test as to enlarge the length of conversations step by step and involve conversational situations in Section A （Conversation Listening Part）， to add short audio-news and video-episodes as well as subjective questions to Section B （Passage Listening Part）， and to change traditional word and sentence gap filling into title filling， partial topic sentence filling and key words filling in Section C （Compound Dictation Part）.

Key word： Empirical reforms； CET-4 （College English Test Band 4）； Communicative listening test

LEI Min （2013）. Empirical Reforms on New-CET4 Communicative Listening Test. Canadian Social Science， 9（1）， -0. Available from： http：///index.php/css/article/view/j.css.1923669720130901.2201 DOI： http：///10.3968/j.css.1923669720130901.2201.

INTRODUCTION

In the article An Empirical Study on Problems Involved in CET4 Communicative Listening Test and Teaching， it has been proved that current CET4 listening test （New-CET4） is comprehension-centered rather than communication-centered and does not put the priority to the purpose of improving students’ authentic communicative competence （Lei， 2012）. As for the listening materials， they are less concerning English daily communicative situation， most of which belong to written English monologues with strange words and terminologies. As far as the listening question items are concerned， objective questions， namely closed questions always dominate the testing， whereas certain appropriate subjective questions are taken out of consideration， which result in confining students’ improvement of potential communicative competence. With regard to the variety of testing contents， it is very dull and plain， which is hard to draw students’ enthusiasm.

To reform above shortcomings involved in current CET4 listening test， and transfer the traditional comprehension-oriented test into communicative one， following empirical study has been conducted in the subject college--Hubei University of Education as well.

Empirical Reforms on New-CET4 Listening Test

Hypothesis and Purpose

In order to explore out the best listening test in favor of enhancing students’ real conversational competence in different communicative situations， the empirical reformed CET4 listening test and the simulated CET4 listening test have been conducted in study simultaneously after a term of communicative listening teaching practice， in the aid of paired samples test and independent samples test by SPSS.

The Hypothesis of this test research is that there is no difference between communication-oriented reformed CET4 listening test achievement and simulated New-CET4 listening test achievement in respect of mean values， which is called null hypothesis. If the null hypothesis is rejected by the result of research， it can be proved that there is a great difference between them， which identifies that the reformed CET4 listening test is superior to the current one in respects of testing and improving students’ communicative competence.

The purpose of this test is to justify the communication-oriented reformed CET4 listening test in the method of conducting quantitative analysis of comparative empirical study.

Subjects

Same as the subjects who have been enlisted in questionnaire and test research in the former study， An Empirical Study on Problems Involved in CET4 Communicative Listening Test and Teaching， the subjects in this research， except one absent in test， conclude 95 students in total with two different majors， one is engineering management， and the other one is primary education as well.

Qualification for Paired Samples Test and Independent Samples Test

Before conducting the empirical study of Paired Samples Test and Independent Samples Test， the distribution of normality and homogeneity of variance upon subjects’ achievements should be testified. That is the prerequisite for further empirical study.

Table 1 describes all the items compared or analyzed in this study： both reformed test and simulated current CET4 test comprise three sections： A， B and C； for the sake of clearly distinguish them， we name Section A， B， C in reformed test New-A， B， C， and employ Current-A， B， C to represent Section A， B， C in simulated current test； in addition， the total score of both are named New-total and Current-total respectively. They are applied throughout the following study. There are 95 subjects in total have taken part in both tests covering section A， B， C.

Table 2 shows the M-Estimators level， a very important reference to measure the normality of distribution of all the items such as New-A， B， C， New-Total and Current-A， B， C， and Current-Total. It is a principle that if the value of M-Estimators is much closer to its corresponding mean value， we can conclude that the data distribution belongs to normality distribution. Compared the mean value of each one of above eight items （shows in table 3）， all their M-Estimators values are very closer to their mean values， such as New-A’s M-Estimator value 10.7 is much closer to its mean value 10.58， New-B’s M value 5.41 approaches to its mean value 5.36， and so do that of New-C， Total， Current-A， B， C and Total. It identifies that the data distribution of above eight items belongs to normality. This is also testified by the Skewness level： all their value of Skewness （shows in table 3） are all far less than 1， （such as -.549 for New-A， -.225 for New-B， -.426 for New-C， -.200 for New-Total， -.412 for Current-A， .371 for Current-B， -.101for Current-C， -.223 for Current-Total， even though they have a light skewness， that can be ignored.） which strongly confirms us their normality of distribution as well.

The significance column of Table 4 illustrates that all the items being tested except for New-C accord with the prerequisite of the homogeneity of variance. It can be explained as follows： the significance levels of New-A， New-B， New-Total， and Current-A， Current-B， Current-C， and Current-Total arrive at the required significance level based on four great indexes such as Mean， Median， Median and with adjusted df and trimmed mean. The significance values of them are all far more than the required significance level of 0.05， even though the four indexes of Item New-C are so far less than 0.05， it does no matter in the whole study procedure.

Procedures of Comparison between the Simulated CET4 Listening Test and the Reformed One

The empirical study on comparison between the simulated CET4 listening test and the reformed one is on the basis of that the difficulty of the simulated test nearly equals to that of current CET4 listening test. According to the classification of item types， the Reformed CET4 listening test has been divided into four respects containing three separate sections A， B， C and the overall distribution of questions. Both tests contain 35 questions for answer and have 35 points in total. 95 student subjects have been assigned to finish both tests in succession； the exam time and situation are mostly same as that of the authentic CET4 test.

Comparison Between Sections-A

In simulated test， eight short conversations covering from 11 to 18 and two long conversations 19， 20 are same as that involved in current CET4 tests. According to analytical study， question 12 and 13 involves inferring meaning from conversation contents， which examines students’ logic thinking ability； Question 15 and 16 is concerned with the implied meaning which hide behind speaker’s words， which tests the depth of students’ listening comprehension ability as well as their quick response ability； Question 14 sets a interference which checks students’ distinguishing or judging competence； Question 17 and 11 examine students’ vocabulary commanding. From which we can get the conclusion that most of these conversation materials heavily focus on training and testing students’ obtaining and understanding of speaker’s intent or meaning rather than introducing the listeners into an authentic communicative situation， let alone encouraging their communicative willing and involving them into conversational situation. Moreover， the relevant situations of which are monotonous， not covering the situations close to daily life. As for two long conversations， the questions are more likely to examine listeners’ comprehension upon details appeared in conversation rather than cultivating their capability of discovering and solving the problems that will be involved in different communicative situations.

Compared with simulated test， the eight short conversations of reformed test are communicative oriented， by statistics， it covers eight different communicative situations which are much closer to our daily life， such as earphone trouble on plane in Question 1， booking a hotel room on telephone in Question 2， depositing money in bank in Question 4， going to an exhibition in Question 5， camping outside in Question 7， the economic and politic hot topic in Question 6 and 8 respectively etc. As regard to two long conversations， one is about visiting a patient in hospital； the other one involves a job interview， in which most of the questions are direct. Both short conversations and long conversations are concerned with how to inducting listeners into communication freely and naturally so as to stimulate their potential motivation for communication. In this case， listeners can give an answer freely. It looks like that they are undergoing speaking conversations. In addition， the length of conversation is enlarged gradually， that is to say ten conversations in this section are been designed into three scales： short （from one to six）， middle-length （from seven to eight） and long （from nine to ten）， in purpose of developing listeners’ communicative competence step by step.

Comparison Between Sections-B

So far as the materials of section B in simulated test is concerned， it consists of three monologue passages， first one concerns Britain justice， second one is about American history on Main Street in 1920， the last one involves natural being （bird’s nests）. Most of them belong to the category of western culture and histories， which aims at examining students’ knowledge upon English countries’ culture and history in a way of monologue input. By analysis， we can find the words of three passages are written English， of which there are so many long sentences， strange words and expressions that are not frequently used in daily life. It is more similar as reading comprehension， focusing on “comprehension” rather than “communication”. Even though the examining direction of western culture and history can’t be objected， we have doubt of whether this comprehension-oriented monologue can achieve the end goal of drilling students’ practical communicative competence in daily life. Obviously it fails to examine the students’ authentic English communicative competence.

Differ with the simulated CET4 test the reformed one attaches more attention to communicative conversation situation. The reformation can be reflected in following respects： Firstly， in question items renovation， the current short monologue is changed into short audio-news and video-episodes， which achieves the transfer from audition input to the combination of audition and vision input. In addition， the most important in this section lies in the reform that questions have been shifted from objective form to subjective one， which is intended to encourage students express their idea freely and briefly according to what they have heard. It is a bold idea in renovation against traditional Section B， aiming at constructing conversational room as larger as possible so as to bring listeners’ potential communicative competence into full-scale. Secondly， in terms of listening materials selection， its range has been extended to the field of media such as audio or video news and films which are the key channels to acquire a great number of information， from which listeners will get a full-scale acquaintance of our world. In detail， Audio Passage One is about a piece of latest news concerning Cross-Straits Airfares Cut； Video Passage Two covers three clips of hot Week News in terms of movie and television recreation， social odd facts and humor stories. Audio Passage Three is an episode of movie Across America， concerning a hospital situation. All of them have enriched the original monologue in variety. Thirdly， as to the speech of listening material， the spoken language often used in daily life apart from written form of terminologies has taken up the top position. That is to say， the reformed Section B affords authentic conversation situation and guarantees the timeliness of materials at the same time.

In summary， the reformed Section B has reformed the current one in respects of item type， materials and speech based on the purpose of developing students’ authentic conversational competence in different situations.

Comparison Between Sections-C

In simulated test， Section C consists of eight word blanks and three long sentence blanks. The former part of word gap filling are inclined to check listeners’ level of mastery of vocabulary， the latter part of sentence gap filling emphasizes on measuring students’ memory and shorthand level in stead of testing their ability of reprocessing or reproducing key information what they have heard.

Different from simulated test， the reformed one deals with the gap filling on the basis of testing students’ level of how to summarize or reorganize the information what they heard. The blanks involve title filling， partial topic sentence filling and key words filling. Even though the material selected has less difficulty degree than that of simulated one， it can effectively encourage listeners’ confidence and interest in practicing listening activities.

Comparison Between Overall Distributions

Based on the criterion of whether the listening question is comprehension-oriented or communication-oriented， we can classify all the questions both in simulated test and reformed test into three categories in terms of comprehension-oriented， communication-oriented and basic knowledge or others. The explicit classification and distribution have been reported in Table 5 and Table 6 which is contributed to find out an appropriate distribution of testing questions in CET4 listening test.

Table 5 and Table 6 indicate the Reformed test is more communication-oriented than the simulated one， in light of the different emphasis they have attached on examination. We can get this conclusion from above concrete data： At first， both tests have the equal number of questions 36， but there are great differences in distributions in respects of comprehension-oriented category and communication-oriented one. In detail， within the former respect， simulated test has 18 questions， three times more than reformed one has （5）； as to the latter respect， the communication-oriented distribution of reformed test are nearly three times that of simulated one. In order to make clear where the great difference lies in， a careful analysis of Table 5 and 6 has been shown as follows： In simulated test， only 7 communication-oriented questions， equals to 19.4% of total test， belong to long conversation in Section A， while there are 18 comprehension-oriented questions， takes up 50% of total test， centralize in short conversation of Section A and passage listening of Section B； Differently， within reformed test， on one hand， there exists 20 communication-oriented questions， takes up 55.6% of total test， which scatter in three Sections separately， of which， nines spread in Section A， tens locate in Section B and one is put in Section C. on the other hand， only 5 questions are comprehension-oriented， amounts to 13.9% of total test， which are set in short conversation of Section A.

Results and Discussion

Student subjects’ scores achieved both in simulated test and reformed test whose qualification for subsequent paired samples test and independent samples test have been proved， have been computed in SPSS system in the purpose of getting the results contributed to the communication-oriented listening test.

Results of Paired Samples Test and Analysis

The purpose of conducting Paired Samples Test in respects of threes pairs of Section A， B， C and one pairs of Total scores respectively are to testify whether the reformed test is superior to the simulated one. If the answer is positive， we can get the conclusion that every sections of reformed test are more beneficial to improve students’ communicative competence than simulated one， we should reform the current CET4 listening test with no hesitation.

Table 7 lists the pairs those are taking parting in Paired Samples Test， which consists of Pair1 （New-A against Current-A）， Pair2 （New-B against Current-B）， Pair3 （New-C against Current-C） and Pair4 （New-total against Current-Total）. Meanwhile， the values of their respective Mean， Standard Deviation and standard Error Mean have been reported explicitly. Compared four pairs of Means we can learn from which each mean of Section A， B， C and Total （10.58， 5.36， 5.81， 21.75） in reformed test are more than that （9.39， 3.75， 4.85， 17.98） involved in simulated one. It illustrates the fact that students have got higher scores in reformed test than what they have got in simulated test in above four respects of Section A， B， C and Total.

Table 8 informs us that all the four pairs of variances have the higher level of correlation （more than 0.3） such as .375 for Pair 1 （New-A against Current-A）， .269 for Pair 2 （New-B against Current-B） which is much closer to 0.3， .489 for Pair 3 （New-C against Current-C）， and .608 for Pair 4 （New-Total against Current-Total）. At the same time， all the four pairs of variances have the lower level of significance （less than 0.05） such as .000， .008， .000 and .000 listed in Significance column， which indicates that these pairs of variances are qualified for Paired Samples Test. Although the correlation of Pair 2 （.269） is not up to 0.3， it has the qualified lower significance so that it has no impact on T-test and still qualified for T-test.

Table 9 states that between the lower limit and the upper limit of the 95% Confidence Interval of The Difference， the values of four pairs do not include zero， which means that there is a great difference within each pair of variances； also the four values of Significance 2-tailed are far less than 0.05， which strongly reject the null hypothesis and reconfirms us that there exists a greater difference between each pairs.

In conclusion， student subjects got higher scores in reformed test than that they got in simulated one after a semester’s communicative listening teaching. Through analysis based on above statistics study and Table 10 below， we can get the results in detail： In terms of Section A， there are 5 comprehension-oriented questions， 9 communication-oriented questions and 1 other-oriented question existing in reformed test； while simulated test consists of 8 comprehension-oriented questions and seven communication-oriented ones. Judging from the material difficulty， Section A of reformed test is more difficult than that of simulated one. Despite the difficulty， students still got higher scores in reformed test than scores they have got in simulated one. As for Section B， all of the 10 questions belong to communication-oriented ones in reformed test； while in simulated test， all of them are comprehension-oriented. Judging from the material difficulty， Section B of reformed test is less difficult than that of simulated one in respect of listening comprehension because in simulated test， there are so many written speeches， strange words and materials which are scarcely used in daily life communication. However， judging from the effectiveness of measuring students’ authentic communicative competence， the communication-oriented Section B of reformed one is superior to the comprehension-oriented Section B of simulated test. In view of this case， students’ comparative higher scores in reformed test manifests that the Section B of reformed test can play an even more important role in examining students’ communicative listening competence， compared to that of simulated one. As to Section C， two tests are equal in difficulty； both of them tend to examine students’ basic knowledge and skill in respects of vocabulary and shorthand， the only different is that the reformed one even more focuses on inducting students how to capture the central idea effectively and skillfully. Therefore， the Blanks treatment is outline-oriented in the purpose of testing students’ ability of words reproducing. In this section， students still got higher scores in reformed test than scores they got in simulated one. It indicates that in some sense， Section C of reformed test is still the more beneficial one. As far as the Total Scores are concerned， the fact of higher scores in reformed test and comparative lower scores in simulated test strongly confirms us that the reformed test is superior to the simulated one in terms of building up students’ authentic communicative competence in different situations.

Results of Independent Samples Test and Analysis

The purpose of conducting Independent Samples Test upon two separate majors （one is engineering， the other is education） in respects of Section A， B， C and Total scores both in simulated and reformed test are to testify whether there are great differences between the two different classes of students’ achievements in above four sections. That is to say， it will be testified that whether the specialty has a significant impact on students’ achievement in listening test. If the answer is positive， we should pay more attention to analyze the potential causes from which the differences derives， for the sake of providing valuable instructions for communicative listening teaching activities. As we all known， testing and teaching are inseparable twins， they are interdependent， mutually beneficial. Effective testing can offer positive washback on teaching， and vice versa. Certainly， the results of testing can offer useful references to teaching strategies. For example， the exploring of different scores distribution within different majors’ same testing will benefit teachers take different measures according to the students’ practical circumstances in different specialties. It is necessary to conduct Independent Samples Test upon above two specialties of students.

In Table 11， 95 student subjects have been groped into two， an engineering class of 41 students and an education class of 54 students. All the scores they have got in Section A， B， C and Total both in simulated test and reformed test have been separated into two. The values of Mean， Std. Deviation and Std. Error Mean as well as student subjects’ number have been reported explicitly. From the column of Mean， we can find that， engineering class’ scores are less than education class’ in every section within both tests， which informs us that the former class have better achievements than the latter one.

Table 12 reports Levene’s Test for Equality of Variance， which reconfirms the equality of above data’ variances， and t-test for equality of Means， which point out two key reference indexes： the values of Sig. （2-tailed） and 95% confidence interval of the difference. The result of t-test for equality of Means is the item we must focus on in this part of research. Looking into all the values of Sig. （2-tailed）， all of them except Current-A（Simulated Test -Section A） are far less than 0.05， which indicates that two groups have a great difference in respects of New-A（ Reformed Test-Section A）， New-B， New-C， New-Total， Current-B， Current-C and Current-Total. Also， their corresponding values between lower and upper of 95% confidence interval of the difference do not include zero， which confirms us that the specialty has a notable impact on their listening achievements. It is a fact that the education class with majority of female students has higher achievement than the engineering one with majority of male students. From which we can learn the common sense that female student are generally superior to male students in language study， which had been proved by other researchers. Subsequently， the root causes of the achievement differences between two different specialties of classes will be taken into deep consideration. The columns of Mean Difference in Table 12 and Mean in Table 11 supply the valuable indexes for our analytical study on Mean differences between science and art class.

From the Mean difference appeared in Section A， we can get the root causes of the difference that took place in both tests respectively. From Table 12 and Table 11， we can discover that engineering class has 1.405 points less than education class in Reformed test， which at first attributes to its historical lower English listening level which can be testified by their Mean difference （-5.871 in CET4 2009） showed in Table 14 and Table 13， this Mean difference should also be given the credit to two classes’ original English listening level differences. Furthermore， the difficulty of reformed test in this section is more than that of simulated one， so the communication-oriented reformation may become another reason for their Mean differences. However， in simulated test， there is little differences （-.299） between two classes， because usually science class are better at Logical thinking than art class in terms of inference and judgment that are the focuses in simulated test. According to my teaching experience， science class’ quick response， creative thinking and good logical inference have sufficiently represented in daily classroom activities， all of this would be a good reason for the slight difference between them in simulated test. In other word， science class takes full advantages of their merits in logical thinking so that offsets their big gap in average listening level.

As far as Mean difference involved in Section B is concerned， the research of root causes can also be discussed in simulated test and reformed test. In reformed test， engineering class has .758 （Table 12） points less than education class does. As proved that the questions are more communication-oriented than that of simulated one. Moreover the knowledge examine item has been expanded to media report， current affaires and movie recreation etc. which belong to engineering class’ interests， so that they can make full use of their favorable factors in this section and got a considerable good results with little lag. However， in simulated test， the Mean gap is up to 1.057 points （Table 12） because of the hinder of comprehension-oriented questions and unfamiliar written English words and expressions that are just the weakness of engineering class. On the basis of my teaching experience， the male dominant science class such as the engineering class is usually lazier in respects of increasing vocabulary and extracurricular reading than the female dominant art class such as the education class. It is regarded as the root cause for engineering class’ lag in comprehension-oriented simulated test.

In terms of Mean difference occurred in Section C， the research of root causes can be analyzed based on the comparison between simulated test and reformed one. In Section C， there is a bigger gap between reformed test and simulated one. In reformed test， engineering class has only 0.696 points less than education class dose； however， at the same time it has 2.315 points less than education class in simulated test. Section C of reformed test is comprehension-reprocessingoriented， which is not a trouble for engineering class； but the simulated one is nearly memorization-oriented， which just hit the engineering class’ dead weakness. In addition， the reason can also lies in the fact that the gap filling sentences of reformed test are comparatively simpler and shorter than that of simulated one. In a word， the big gap between reformed and simulated test reconfirms us that engineering class are weaker in knowledge memorization but better in logical thinking.

Conclusion

Through relevant empirical studies， many effective renovations have been conducted and sufficiently justified in the reformation of New-CET4 listening test.

In Section A， the original eight short conversations concerning inference and judgment have been replaced by communication-oriented conversational situations that are much closer to our daily life communication. As regard to two long conversations， different from two traditional long conversations， daily conversational situations and direct questions for answer should be involved in it. In general， both short conversations and long conversations are concerned with how to induct listeners into communication freely and naturally so as to stimulate their potential motivation for communication. In addition， the length of conversation is enlarged gradually， that is to say ten conversations in this section are supposed to be designed into three scales： short， middle-length and long， in purpose of developing listeners’ communicative competence step by step.

In Section B， the justified reformation have been conducted in respects of question items， materials and language based on the purpose of developing students authentic conversational competence in different situations. Differ with the original three passages of monologues， the reformation can be reflected in following respects： Firstly， in question item renovation， the original short passage monologue should be changed into short audio-news and video-episodes. What’s more， objective questions are supposed to be added to test in favor of encouraging students express their idea freely and briefly according to what they have heard. Secondly， in terms of listening materials， it has been extended to the field of media such as audio or video news and films， the key channels to get listeners acquire a great number of information about the world. Thirdly， as to the language of listening material， the spoken language used in daily life instead of written form of terminologies has taken up the top position.

In Section C， the blanks involve title filling， partial topic sentence filling and key words filling. The reformation deals with the gap filling on the basis of testing students’ level of how to summarize or reorganize the information what they heard.

Despite so many valuable findings have been discovered in above empirical studies， there still have two respects of limitations occurred in this study： One is concerning the limitation of samples. Although 96 subjects with two different majors （one is science， the other is art） are representative within one university， it still does not arrive at the level of sufficient big. The other one lies in the similarity between the simulated CET4 listening tests adopted in empirical studies and the authentic New-CET4 listening test. In the course of empirical study， if the simulated CET4 listening test can be replaced by the first hand resources of 96 subjects’ scores in each Section（ A， B， and C） which they achieved in 2009 CET4 listening test， the authority of above findings got from the empirical studies could be strengthened definitely.

REFERENCES

Allwright， R. L. （2000）. Language Learning through Communication Practice. London： The British Council.

Bachmen， L. F. （1990）. Fundamental Considerations in Language Testing. Oxford： Oxford University Press.

Cao Qin. （2009）. Washback of Innovation in CET-4 Listening subtest on College English Listening Teaching and Learning. Southwest University， China.

David， Nunan. （2001）. Second Language Teaching and Learning.Beijing： Foreign Language Teaching and Research Press

Han Baocheng. （2000）. Statistics Methodology in Foreign Language Teaching Research. Beijing： Foreign Language Teaching and Research Press

Heaton， J. B. （1988）. Writing English Language Testing Tests. London： Longman.

Lei Min. （2012）. Development Status of Communicative Listening Test and Teaching， Studies in Literature and Language， 5（3）， 124-127.

Lei Min. （2012）. An Empirical Study on Problems Involved in CET-4 Communicative Listening Test and Teaching， Canadian Social Science， 8（6）， 170-178.

Liu Runqing. （2000）. Language Testing and Methods. Beijing： Foreign Language Teaching and Research Press.

Liu Xiao. （2008）. A Study on the Application of Language Test Authenticity in the Listening Comprehension Test of CET4. China University of Geosciences.

Qin Xiaoqing. （2003）. the Quantitative Date Analysis in Foreign Language Teaching Research. Wuhan： Huazhong University of Science and Technology Press.

Qin Xiaoqing. （2009）. Questionnaire Survey in Foreign Language Teaching. Beijing： Foreign Language Teaching and Research Press.

Sun Meng. （2007）. An Empirical Study of Using CLTA to Preparing IELTS. Wuhan University of Technology， China.

Wardhaugh Ronald. （2000）. An Introduction to Sociolinguistics. Beijing. Beijing： Foreign Language Teaching and Research Press.

Weir C. （1990）. Communicative Language Testing. New York： Prentice-Hall.

Yurui Zhang. （2007）. Cooperative Learning and College English Listening Teaching. Shanghai International Studies University， China.