UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

A psycholinguistic investigation of the cloze responses of grade eight students Cram, Ruby Victoria


This exploratory study investigated the role of exact and nonexact replacements of cloze responses in the assessment of reading comprehension. Two modes of discourse, narrative fiction and expository prose, were investigated. Two theoretical assumptions guided the study: from psycholinguistics, that reading involves responses to the graphophonemic, syntactic, and semantic cue systems of language (Goodman, 1976a); and from discourse analysis, that a "schema" or cognitive map directs the reader in the search for discourse cues (Winograd, 1977). Subjects were proficient and less proficient secondary school students at two levels of maturity. Attitude to reading was also examined. Operational definitions of discourse were: (i) narrative fiction or conventions of a story, and (ii) expository prose or coherent explanation of a topic. Subjects were entering grades nine (N = 107) and twelve (N = 100) in Lord Byng Secondary School, Vancouver, British Columbia. To examine exact replacements (E.R.'s) and attitudes, only subjects whose primary language was English were eligible. Exact and non-exact-replacements (N.E.R.'s) were examined for a random sample at each grade level of Good (N = 20) and Poor (N = 30) readers identified from scores on the comprehension subtest of the Iowa Silent Reading Test (1973). To verify how linguistic cues trigger responses, six subjects were randomly drawn from each proficiency group for retrospective verbalization interviews, which were taped and transcribed. Each subject (N = 207) completed the Estes Reading Attitude Scale and two cloze tasks: a narrative fiction and an expository prose, from the British Columbia Reading Assessment 1977, Grades 8 and 12. Responses were tested for exact match to the author's word (Bormuth, 1975). To evaluate N.E.R's, the investigator adapted the Cambourne Reading Assessment Procedure (1978), based on the Goodman Taxonomy of Reading Miscues (1969). Following two pilot studies, the classification scheme was made consistent with discourse theory and the coding simplified. A synonym replacement for the exact response was acceptable in three categories: syntax, semantics, and discourse. Statistical procedures included correlation, independent t-tests, and two-way analysis of variance. For the oral protocols, categories were induced from the transcriptions. Frequency of response was analyzed using the chi-square statistic, supported by qualitative description. Attitude to reading had a generally weak correlation with the selected indices of comprehension. For exact cloze scores, relationships with the standardized measure were significant, particularly with expository prose for poor grade twelve subjects. Narrative fiction scores exceeded expository prose scores. Good readers were differentiated from poor readers. The N.E.R. score discriminated between proficiency levels. At grade nine, narrative fiction scores exceeded expository prose, but at grade twelve, differences were not significant. The mean inter-rater agreement, calculated by the Arrington Formula (Feifel & Lorge, 1950), was 91.6 percent. The interviews demonstrated that three cue systems operated most frequently: syntax, semantics, and discourse; and two much less often: grammatical function and life experience. Significant differences in frequency were found between modes of discourse and proficiency levels. For combined exact scores plus synonyms, in grade nine, narrative fiction scores exceeded prose scores, but in grade twelve the reverse occurred. Discrimination between proficiency levels was noted; however good readers were significantly superior with expository prose. Data based conclusions were (1) attitude is not correlated with either proficiency or comprehension, (2) comprehension scores differed for modes of discourse: narrative fiction and expository prose, (3) exact cloze score discriminated between proficiency levels, (4) N.E.R. scores revealed differences in the use of cue systems by ability groups, (5) all readers used the same cue systems: syntax, semantics, and discourse, to gain meaning, but control of the set of cue systems, especially with expository prose, distinguished the good reader, and (6) the addition of synonym scores to exact cloze scores differentiated between proficiency levels, modes of discourse, and maturity levels.

Item Media

Item Citations and Data


For non-commercial purposes only, such as research, private study and education. Additional conditions apply, see Terms of Use https://open.library.ubc.ca/terms_of_use.