A Study On The Construct Validity Of GRE Analytical Writing:From The Perspective Of Communicative Language Testing

Computer, which is regarded as the communicative medium, has been constantly changing the way to communicate with others turning itself into an indispensable assistive tool of the modern life. It results in the profound effect on the form of language testing and computer adaptive language testing comes into being. The development of testing form advances with time and improves continuously. Since ETS implemented GRE which is a computer adaptive language testing, 1993, it has been applied extensively throughout the world, making it a successful typical representative of the computer adaptive testing form for the past few years. On the other hand, the emphasis on communication leads the communicative language testing to the mainstay of language testing. With deeper studies, it’s acknowledged that reading and writing are productive communication thus the quality of communication is taken into consideration in the writing test designs. In this global background, computer adaptive language testing, as GRE has become a hot research issue concerning with its validity in terms of communicative language testing.The primary purpose of this research is to study the construct validity of the analytical writing in GRE, a representative of computer adaptive language testing in the light of communicative language testing. On this basis, the research departs from Bachman’s communicative language testing, aiming at making the following questions clear.1. How does the GRE analytical writing reflect the communicative language testing theory? To what extent does its embodiment of communicative language testing theory illustrate that this test type has the construct validity?2. What are the potential influences the computer adaptive language testing form generated on the construct validity of GRE analytical writing?The qualitative and quantitative methods are adopted in this research. The subjects are 100 test candidates attending the GRE test after 2011. The scores of the analytical writing and other sub tests are collected respectively and analyzed. The qualitative method is used to discuss how the scoring guide and the topics ofanalytical writing reflect communicative language theory. The correlation analysis is adopted to verify both the correlation of the sub tests and that between the sub tests and the whole test. By this way, it can verify whether the analytical writing measures what it is supposed to measure. The questionnaire checks students’ attitude toward computer and thus looks into the elements influencing the construct validity.The research demonstrates that the analytical writing reflects the communicative language theory by both the scoring guide and the topic contents. With the exception of the basic linguistic ability, the pragmatic and strategic ability as well as the psychophysical mechanism are examined in a comprehensive way. The correlation coefficient suggests that the analytical writing and other sub tests are correlated with and independent of each other, showing that different aspects of the same construct are measured. In addition, the coefficient between the analytical writing and the whole test is within the range of high correlation, thereby the analytical writing has the same underlying construct with the whole test, which is communicative language theory.The questionnaire helps gain three major possibilities influencing construct validity and they are the adaptability of measuring mode with a computer, the proficiency for the computer’s basic operating and the feeling of stress and anxiety brought by computer.The construct validity verification of GRE analytical writing may be conducive to a further understanding of this new test mode and contributes to a little more intuitive and profound comprehension of applying the communicative language tests to language writing test types. At the same time, a bit information or help could be provided for the theoretical development of computer adaptive language testing from some aspects. On the other hand, it’s beneficial a little for the developers to improve test modes as the study is to investigate some factors that may influence the test results. Testing modes could be adjusted and the reliability, validity and authenticity of the test could be improved. From the Angle of the subjects, this research may provide the possibility of progress for them as they know about these factors. In spite that some objective factors impacting test results such as the examination anxiety, the computer skills proficiency and any other external elements cannot be avoided completely, they could overcome these factors on purpose in the process of learning under the condition that the examinees know well about these factors. As a result, its degree of the impacts on the test results can be reduced in the real process of taking acomputer adaptive language testing. Form the perspective of language teaching, it’s beneficial for the teachers to form their acquaintance with the construct validity of computer adaptive language testing and thus gain a better understanding for clearing their teaching goals.