A Computational Analysis on Inter-Passage Semantic Intertextuality and Consistency of Text Difficulty in High School English Textbooks

Abstract

This study investigated inter-passage connectivity within a chapter of High school English textbooks in the four linguistic sections—listening, speaking, reading, and writing—in terms of semantic intertextuality and consistency of text difficulty. These two aspects of text such as descriptives (sentence count, word count and sentence length) and readability were investigated by the two recent computational linguistic tools, Latent Semantic Analysis (LSA) and Coh-Metrix. The results from LSA revealed that texts in each linguistic section had a weak semantic intertextuality in a chapter except for the writing section. Through Coh-Metrix analysis, passages in the entire linguistic sections were found to be relatively consistent on sentence length but had high variations in sentence count and word count. Overall the variations appear to differ depending on linguistic section and textbook. Another finding from Coh-Metrix analysis is that the textbooks showed different levels of text difficulty on a few measures even though their target learners are the same 9th grade high school students. This research is limited to partial indices of LSA and Coh-Metrix and quantitative representation of the passages. Pedagogical implications are suggested based on the findings from this study.