Prediction of Readability Formula Outputs with ANFIS: A Textbook Example of Turkish and English as a Foreign Language

This research aims to estimate the readability of Turkish and English textbooks by the ANFIS method. In the study conducted using the quantitative research method, the texts of A1–2 level were first subjected to the readability formula, and then the relevant results were analysed with ANFIS. According to the results obtained from the 3 models, the estimation of the readability coefficients of the texts was found to be insufficient. Numerical data showed an increase when the two text results were modelled together compared to model 1. The R2 values of the models are 0.3876, 0.5251, and 0.4213. Although there was an insufficient relation between the results of the readability formula outputs obtained from textbooks on Teaching English to Foreigners and the modelling, the ability to predict worked better compared to other models. ANFIS which is not designed to calculate the readability formula directly, contributes indirectly to the research. Although it was seen that both books did not exceed the target level as a result of the readability formula, ANFIS failed to predict the content structure. This result shows that improvement should be made in the consistency of the books. Recommendations for the research results were presented.