Forecasting language test performance with a back propagation neural network model

Abstract

Though researchers agree that test anxiety is one of the most debilitating factors that deteriorates the test performance of test-takers who are often anxious during examinations, it is generally neglected when calculating test scores by both classical and modern testing theories. This inevitably leads to the fact that most anxious test-takers' test performance has been seriously underestimated, which greatly undermines equality, one of the fundamental principles of language testing. In order to tackle this daunting problem in language testing arena, this study constructs a back propagation forecasting neural network model, aiming to make forecast on the performance of test-takers in large-scale English proficiency tests. It is expected that this study may pave the way for putting English test anxiety into consideration when calculating scores of test takers in high-stake English proficiency tests. Simulation result indicates that the back propagation neural network forecasting model in this study is accurate when used to predict normal scores, while not satisfactory for extreme scores. Therefore, it is suggested that more input variables should be included in future studies in order to improve forecast precision. Imperfect as this back propagation neural network forecasting model possibly is judged by neural networks professionals, the study is significant to language testing research since the employment of neural network technology is a bold attempt in language testing arena, which implies the possibility of solving some long-pending problems in this area with technology-enriched solutions that have long been adopted by science and engineering researchers. What's more, it also strengthens the necessity of enhancing collaboration between science and engineering researchers and language testing professionals so as to introduce neural networks theory to more and more challenging fields and bring language testing research onto a new stage.