Text complexity predictors: methods and approaches for assessment

Abstract

. An ever-increasing need for quality textbooks and objective linguistic expertise encourages more intensive research into complexity of academic discourse. The current research focuses on lexical density viewed as an effective complexity predictor and defined as the ratio of content words per number of words in a text. Being predominantly quantitative, the study also examines dynamics of Flesh Kincaid grade levels and ratios of parts of speech across 12 Science and Social Studies textbooks taught in Grades 7 - 12 of American schools. The analysis shows a consistent pattern of strong positive growth of nouns and adjectives across grade levels, while lexical verbal elements slightly decrease across the textbooks. The total adverb count changes slightly, and its movement vector depends on the discourse: it rises in Social Studies textbooks and is stable in Science textbooks. This multidirectional movement of components in Lexical density structure explains its marginal increase across the grades in Science and Social Studies discourse. The findings indicate discourse sophistication increase realized predominantly in text nominalization. We also discuss challenges which nominalization presents for comprehension of academic texts by readers and suggest that provided with reference values of text complexity features, educators receive a reliable tool to select reading texts and assess their suitability for target learner groups. The findings can be beneficial for textbooks authors, exam material developers and discourse researchers.