Construction Of Primary English Text Readability Evaluation Model Based On Linear Regression Model

Reading is an important way of language learning.The main task of text readability research is to predict the difficulty of text reading,and recommend reading materials suitable for their reading ability to target readers,so as to improve the learning quality and learning effect of language learners.The study of text readability began in 1921,mainly by asking experienced experts or teachers to subjectively evaluate the difficulty of English texts.This method is highly subjective.At present,the widely used text readability research mainly uses natural language processing related technologies to quantitatively and automatically analyze the reading difficulty of the text;and the language of text readability research has also expanded from the original English to Chinese,Spanish,and Bengali.,Hindi,Arabic,Portuguese,Russian,Japanese and other nearly 20 languages.English is currently the most studied language for text readability evaluation.Its research mainly includes readability research based on the first language learning,readability research based on the second language learning,and the second language readability research is divided into ESL(English as a Second Language,a study on the readability of learners whose native language is not English and and who use English as a second language)and EFL(English as a Foreign Language,for learners whose native language is not English and who use English as a foreign language),the study of English readability in China belongs to EFL.At present,the study of English readability in China mainly includes the results of direct use of the first language readability,such as the Flesch formula;readability studies for senior students,such as Liu Jiangli’s design of the English readability formula for junior high school students.Therefore,this paper is mainly based on primary school English textbooks,using linear regression methods to design suitable English readability formulas for primary school students.The main work of this paper has three aspects:The first is the construction of the text feature system of English readability in China.There are more than 170 readability features.It is divided into traditional features,language-specific features,and learning object-related features.In view of the current situation of English learning in China and previous research,we finally selected 13 features at the vocabulary level,sentence level,and text level as the basis for the construction of the text readability formula.They are the general characteristics(number of words,average word length,TTR,proportion of nouns,proportion of verbs,average sentence length)and EFL-specific characteristics(proportion of common words,coverage of common words,proportion of clauses,average clause length,Proportion of personal pronouns,proportion of foreign named entities,proportion of abbreviations).The second is to design and test the readability formula based on 240 texts of primary school English textbooks published by People’s Education Edition.The readability evaluation formula is constructed based on the linear regression model and 13 features extracted from 160 primary school English texts published by the People’s Education Press:Y=-22.044+8.462*X1+504.718*X2+45.416*X3.Among them,Y represents the readability score of the text,Xi is the average sentence length,X2 is the coverage rate of common words,and X3 is the proportion of foreign named entities.Using this formula to test 80 texts in primary school English textbooks published by the People’s Education Press,the accuracy of the formula is 82%;the readability evaluation of 50 texts in《Kid’s Box》shows that the formula can better evaluate the grade level of Chinese primary school students compared with the first language,ESL and EFL formula.The third is to use the web development technology to build the English readability evaluation platform,which can realize the first language evaluation(Flesch-Kincaid,Gunning Fog,Coleman-Liau,ARI,SMOG)and evaluation based on the design formula of this paper,giving the evaluation results of each formula and related readability feature information.