The purpose of this paper is threefold: (I) to review important research on cloze procedure used as an instructional device; (2) to attempt to ascertain the most desirable method of using cloze exercises for improving reading comprehension; and (3) to propose a new-type cloze exercise, i.e. the cohesive cloze exercise, specifically for teaching long-ranged contextual units in reading texts.
Bloomer (1962), the first to use the cloze procedure as a remedial teaching technique for college students, showed that the cloze exercise group significantly increased total reading ability while the control group did not. On the other hand, no positive effect of cloze exercises was found by Schneyer (1965). In his exercises, however, the incorrect cloze responses the pupils made were corrected but not explained by the instructor, that is, reasoning reinforcement was not given to the pupils. Heitzman and Bloomer (1967) and Guice (1969) also reported a non-positive effect for cloze exercises, whereas Kennedy and Weener (1973), in a study of remedial third graders in which visual and auditory cloze training was carried out to improve reading comprehension, showed that visual cloze training improved the students' reading comprehension, as measured by the Durrell Listening-Reading Series reading comprehension subtest, and that there was some effect from visual cloze training upon the listening subtest scores as well. In their study, Kennedy and Weener discussed the effects of cloze training in light of the improvement in ability to use syntactic and semantic contextual constraints in reading texts.
The present paper takes the position that the conflicting results concerning the effects of cloze exercises obtained by previous studies procede from differences in the procedures and materials used. In particular, the way in which the students' incorrect responses are corrected plays an important role.
Beginning with one of the fundermental·questions of cloze functioning, the second part of this paper presents a new-type exercise, i.e. the cohesive cloze exercise. A previous study on cloze contextual ·ranges by Yamada (in press) indicated that 'cohesive cloze items', which account for about 15% of cloze items in an ordinary cloze test, measure the Ss' ability to read long constraints across sentences. Accordingly, it is claimed that cloze exercises consisting only of cohesive cloze items should be used for text-dyslexia-type students.
This paper is exploratory in character; rather than lay down definitive answers on all questions raised, it indicates to the reader a number of research questions which seem likely to yield useful results in further study.