Abstract
The author takes up the suggestion made by Johnson (1981) that some of the assumptions made about cloze tests may be mis leading. Research which has now been going for thirty years indicates that cloze procedure is a criterion-referenced test which will identify whether an individual or a group possesses the basic skills in the target language. Cloze is a very blunt instrument for testing but within its limits it is valid and reliable. However, changing the variables such as deletion rate, scoring procedure or using a different text may well result in a radically different test each time. The raw data obtained from a cloze test cannot be used as continuous data, such as would be derived from most psychometric-type tests, but as an indicator of one of three levels: frustrational, instructional and independent. Finally, guidelines are given for the preparation of a cloze test.
Get full access to this article
View all access options for this article.
