Abstract
Purpose. Comprehension in continuous discourse is widely viewed as a product of processing meaning across sentences. Across-sentence comprehension is considered to be of major importance to reading comprehension and must be taken into consideration when measuring reading comprehension. The purpose of this study was to determine the validity of the standard cloze procedure (every-fifth-word) deletion pattern as a measure of across-sentence reading comprehension.Method. The method of investigation was to randomly disorder the sentences of cohesively written passages to determine if sentence disordering would affect cloze scores. If cloze is a valid measure of across-sentence comprehension it would be expected that passages containing sequentially ordered sentences would yield higher scores than passages containing disordered sentences.The study was conducted on the University of Miami campus using 272 students enrolled in undergraduate courses. Four cloze treatments were applied to four expository passages creating sixteen cloze treatments. One passage was randomly assigned to each subject and a twenty minute time limit was set for completion of the cloze tests.Findings. The results of an analysis of variance of cloze treatments and expository passages indicated no significant main effect for treatment, i.e. cloze scores on disordered passages were not statistically significantly different from cloze scores on sequentially ordered passages. This implies that the standard cloze procedure does not measure across-sentence comprehension.Although there was a significant interaction between treatments and passages, variation accounted for by treatments was only 1% to 2.7% of the total variation in cloze scores. Passage variation accounted for 14.% of the variation in cloze scores. The simple effects of treatments were not the same across all of the passages.Conclusions and Recommendations. Within the limitations of this study, the standard every-fifth-word cloze procedure does not appear to have a strong empirical basis as a measure of across-sentence reading comprehension. Test publishers and test users should consider the lack of evidence to support the use of the standard cloze procedure as a measure of comprehension of continuous, cohesively written discourse, or across-sentence comprehension.