Using Part Of Speech Structure Of Text In The Prediction Of Its Readability

Readability formulas predict the reading difficulty associated with text. They typically output a U.S. school grade level that indicates the reading ability required of a person in order for him to comprehend that text. Ability to predict text readability is useful because it helps educators select appropriate texts for students and authors write texts accessible to the audience they target. Existing readability formulas are based on countable aspects of the text such as average sentence length and average word length. We propose a new readability formula, the Readability Index, which is based on the part-of-speech structure of sentences in a text. We provide experimental results which show that the Readability Index makes better grade predictions than existing readability formulas.