Bayes model for assessing the reading difficulty of English text for English education in Jordan

Abstract

Predicting the reading difficulty level of English texts is a critical process for second language education and assessment. Reading difficulty level is concerned with the problem of matching a reader’s proficiency and the appropriate text. The reading difficulty level or readability assessment is the process for predicting the reading grade level required from an input text or document, which corresponds to the reader and to the materials. Students in Jordan at their academic levels find obstacles in finding relevant readable data for any subject at their levels. This paper is intended to introduce a model that foretells the reading difficulty level of a given text in terms of a student's ability to read and understand English as a non-native English speaker in Jordanian schools. In this paper, Jordanian students were classified into four categories according to their knowledge of English. The prediction of the reading difficulty level is achieved by using a modern statistical model that is situated on the Bayes model. The model compares the given text with some standard predefined text that strongly reflects the ability to read and understand English text. The accuracy of the proposed model was tested using the hold-out method. The overall prediction accuracy was 75.9%.