Text difficulty and the measurement of reading growth

Margaret Forster, University of Melbourne

Abstract

This study examines the relationship between two variables which are assumed to influence a reader's ability to construct meaning from text. The first variable, 'text difficulty', is defined as the quantifiable features of text from which we ask readers to construct meaning. The second variable, 'task complexity', is defined as the skills readers need to employ in order to construct meaning from text. These variables are conceptualised as being independent. The study sought to explore and understand the way in which the two variables influence a reader's 'ability to construct meaning'—defined as a reader's ability to answer items (questions) on a reading test correctly. The study sought to explore and to understand the relationship between these two variables in the context of professionally developed reading tests: To what extent can the difficulty of reading items be explained by the difficulty of the text they address? To what extent can the difficulty of reading items be explained by the complexity of the reading skills that the items demand? Empirically, the study focused on answering a single research question: Which of the two variables, text difficulty or task complexity, best explains the difficulty of reading items? This question was explored in order to answer a more fundamental question about the nature of the underlying reading dimension: In conceptualising and mapping reading growth do we need to (and if so, can we) describe increasing reading ability in terms of a relationship between increasingly difficult text and increasingly complex task? The study was based on the proposition that a useful uni-dimensional reading variable must recognise and describe reading growth as a relationship between increasingly difficult text and increasingly complex task. The alternative to this position seemed to be one of two extreme positions. Either text difficulty did not influence item difficulty at all, or all items written for a given piece of text must be of the same difficulty. Both of these positions seemed untenable. Unexpectedly, the study found that text difficulty does not help to explain the difficulty of reading items and, therefore, that the underlying reading variable can be described entirely in terms of what readers do in order to make meaning of text. These findings are not only counter-intuitive; they stand in contrast with the findings of the Lexile research which reports that the difficulty of reading items in any data set is explained entirely by text difficulty and therefore that the underlying reading variable can be described in terms of text difficulty only.