Predicting Foreign Language Anxiety Using Machine Learning Algorithm

Abstract

The number of journeys people take to change their lives is rising, and in order for a person to be able to adapt to this openness and realize the value of his travels, it is crucial that he learn a language other than his native tongue. However, learning a new language can be very stressful, and this stress frequently prevents the person from continuing to learn the new foreign language. The numerous types of stress and anxiety are presently paying a lot of attention to this sort of anxiety. Foreign language anxiety can be detected using different scales and measures but this type of detection is not only difficulty for the teachers and time consuming but also its very traditional, in addition, there aren't many research that predict second language anxiety using contemporary technologies. To fill in this gap, we proposed a foreign language anxiety prediction model base on machine learning approach which enables prediction of the second language stress with an accuracy of 100%. In this study, 214 Somali students who are studying in Bangladesh were surveyed using the most famous and reliable Foreign Language Classroom Anxiety Scale (FLCAS) to find out if the student is dealing with foreign language anxiety. Six different machine learning classifiers (Naive Bayes NB, Support Vector Machine SVM, Decision Tree DT, Logistic Regression LR, Random Forest RF, and K-Nearest Neighbor KNN) were utilized in order to achieve the best result. Findings show that Logistic Regression perform better than the others with 100% accuracy.