Towards a better measure of readability: Explanation of empirical performance results
Leslie A. Olsen, Rod Johnson · WORD · 1989
Towards a better measure of readability: Explanation of empirical performance results* I. Introduction.Systemic linguistics has been shown to be a useful analytical tool for text analysis and language training.However, a full, theoretically-informed analysis of a particular text requires a large amount of training and analysis time which may hinder its application in many situations.One possibility for expanding its impact is to narrow the focus of analysis to a smaller set of 'key concepts' which would trade the practical benefits of learning and applying a more compact system against the fullness of the theoretically complete analysis.We would like to report the results of such an approach, in which several concepts which will be familiar to systemic scholars were applied to a problem in the important real-world debate on readability.We hope to show that the "simplified" approach taken here produced useful insights without a full-blown systemic analysis.Even though currently used readability formulas are widely seen among scholars to have serious problems (Chall 1958, Selzer 1983, Duffy 1985), they are still being used extensively to "assess" the reading levels of texts in a wide variety of industrial and governmental settings (Duffy 1985).In these settings, significant resources are invested in lowering reading levels of texts as measured by the formulas, even to 6th grade levels (the assumed reading level of 12 year old children), by substituting simpler sentences and vocabulary.Thus, readability formulas have been examined to determine if texts at theoretically lower reading levels are in fact easier to process by readers as measured by increased reading speed, better comprehension, and better recall.Some empirical studies have found performance differences correlated with reading levels, sentence length, and word length (Flesch 1948;Klare 1976;Bormuth 1969;Carver 1976;Coleman and Miller 1968).However, many have not (Coleman 1962;