Reading by the numbers: reconsidering numerical estimates of reading difficulty

Reading continues to arouse deep passions in educational and governmental circles. Recognition of its importance is almost always followed by acknowledgement of the difficulties that some people have in reading, or the difficulties that some texts cause for readers. Teachers, researchers and material developers then look for ways of more closely matching text to reader. This paper reports on the results of three, linked quantitative studies of numerical techniques that are used to guide teacher decisions as they estimate text difficulty and/or reader competence. Findings reveal that: (a) Readability formulae should be calibrated to particular groups of readers before they are used; (b) Cloze tests should be scored for exact replacement when group results are used as estimates of readability; (c) Choice of point of first deletion in a cloze test may affect its difficulty. Use of new technology may increase the use of teacher developed and adapted material. This would make the present study more significant as teachers and developers make predictions of readability. Such decisions will be more dependable if the basis of the numbers used is more clearly understoo