Advances in Cognition, Education, and Deafness

Constructive Processing in Skilled Deaf and Hearing Readers

Constructive Processing in Skilled Deaf and Hearing Readers Joan S. Pinhas Recent research has shown that inferential processing is an important cognitive skill in the acquisition of knowledge (Spiro 1980). The ability to draw inferences is crucial to recalling prose (Kintsch 1974; Paris & Carter 1973), adding structure to story meaning (Collins, Brown & Larkin 1980; Mandler & Johnson 1977), and understanding the reading comprehension skills of children (Kail et al. 1977; Wilson 1979; Wilcox & Tobin 1974; Gibson & Levin 1975). To date, little work has been carried out on deaf learners' inferential processing of written text, since inference, which requires abstract cognitive reasoning , had been considered too difficult for this population. The tendency to ignore inference is surprising, since it is generally accepted that inference is critical for their successful reading comprehension. The available data from Davey, LaSasso and Macready (1983), and from Pinhas (1984) suggest that relatively unskilled deaf readers do have greater difficulty processing inferential information than their hearing counterparts. This study examined the inferential processing abilities of skilled deaf and hearing readers in order to deThis work was supported in part by a grant from the SUNY Organized Research Initiatives Project and from the Nuala Drescher Affirmative Action Leave Program. I would like to thank the students at the National Technical Institute for the Deaf at Rochester Institute of Technology for their participation in the project and the faculty and staff in the Educational Research department for their assistance. Special thanks are extended to Keri McBride for subject recruitment and data collection. Thanks are also extended to David L. aids for reviewing the manuscript. 296 Constructive Processing in Skilled Deaf and Hearing Readers 297 termine what abilities they rely upon to process inferential information in written text. HYPOTHESIS According to constructive processing models of reading (Kintsch 1974), inferences are made while reading the text. We reasoned that skilled deaf readers, when presented with written text, may rely on alternative strategies, such as that identified by Wolk and Schildroth (1984), in which a question is matched in some way to information presented in the text. Such a strategy results in the inference being drawn at the time of questioning, rather than at the time of reading, as a constructive model of reading would predict. Thus, the deaf student reading written text was expected to process inferential information nonconstructively; the hearing reader was expected to process the same information constructively. More specifically, we expected the deaf reader who is processing written text to show longer response times for inferential questions than for literal questions in both the immediate and delayed question conditions. Additionally, inferential questions should take longer to process in the delayed condition as opposed to the immediate condition. We expected the hearing reader to show shorter response times for literal questions than for inferential questions immediately after the material was presented, but to show equivalent response times for both types of questions after a delay in which the surface cues in working memory had faded. Furthermore, the hearing reader was also predicted to respond to the inferential questions with equal latencies in both the immediate and delayed conditions. Response accuracy is not considered to be a sensitive measure of constructive processing. Previous research by Kintsch (1974) and Pinhas (1984) indicates that literal information is always retrieved more accurately than inferential information . Therefore, it was expected that for all subjects across both time intervals, literal questions would be responded to more accurately than inferential questions. METHODS Seventy-six deaf students who were reading at grade levels 9, 10, and 11, with a mean reading level of a 10.3 grade equivalent, were recruited from the National Technical Institute for the Deaf at the Rochester Institute of Technology. All students were identified as being prelingually deaf for unaided puretone averages of 500, 1000, and 2000 Hz, with a mean hearing loss of 94.48 decibels in the better ear. The mean age for the deaf subjects was 21.9 years. Fifty-three hearing students, who were selected according to reading level, served as the comparison group. The mean reading level was 10.4 grade equivalent , with a mean age of 16.19 years. The task consisted of reading four narrative stories and answering eight questions per story-four literal and four inferential questions. The questions were presented in two time intervals: 30 seconds and 20 minutes after story 298 Cognitive Processes presentation, respectively, to allow for determination of when the inference was...