Abstract
No studies have investigated whether the writing of certain transformations can predict comprehension of prose written in them. In studying this possibility, 90 subjects, 30 from each grade level, were asked to read three major prose forms, each written in three transformed versions, and to respond to each immediately on Cloze tests. Every subject also rewrote a simple sentence paragraph using his own sentence combining techniques. The Cloze test scores were then compared to sentence-lengthening criteria in a multiple regression analysis. Cloze tests did not satisfactorily predict sentence length. Yet there was strong evidence to suggest that Cloze tests are less valid measures of advanced reading comprehension and that the higher Cloze scores predicted lower sentence-lengthening criteria.
