An Analysis On The Word Factors Of The Readability Of Chinese Newspaper In Teaching Chinese As A Foreign Language

In the newspaper reading course in TCFL,it's necessary to choose some extracurricular materials from real newspapers as supplementary reading.Text readability is a conception used to measure the level of text difficulty.This thesis focuses on finding those word factors that have a marked impact on the text readability from an objective text perspective.Based on the significant factors,two algorithm models are established in order to make it more convenient for both teachers and students to pick appropriate and latest newspaper materials out of class.262 corpus samples in four different levels came from the four parts of Reading Newspaper and Learn Chinese,a newspaper reading textbook.Text samples were further processed in procedures like word segmentation and labeling.To pick the significant factors,firstly we listed fourteen related factors as well as their corresponding word lists or standards based on the journalistic language features and previous studies.Eight significant factors were later picked by ANOVA,multiple comparisons and Pearson correlation analysis.To have a look at their prediction effect,we took 70% texts as training samples and the rest as test samples,and used R to establish two models by Na?ve Bayes and Support Vector Machine to have a prediction test.A further questionnaire facing teachers and students was conducted to evaluate the validity.Finally we got a conclusion that the word factors that have a marked impact on the newspaper text readability include new words,polysemous words,TTR,idioms,common words,function words,disyllable words and average sentence length measured by word,arranged in a descending order.As for the two typical error predictions,subjective questionnaire results proved to be similar to those of the models.The general prediction accuracy rates of NB and SVM models are up to 75% to 82%.