A COMPARISON OF THE TWO VERSIONS OF THE INTERACTIVE-COMPENSATORY MODEL OF READING
Abstract
The original time-locked version of Stanovich and West's interactive-compensatory reading model states that word recognition is basically a bottom-up process. If the stimulus and/or reader deficiencies slow bottom-up word recognition, then top-down (contextual) factors can compensate. However, sometimes context effects do not vary monotonically with word recognition latencies. For example, repeated words, which are recognized faster than nonrepeated words, are no less affected by context. Both inserting asterisks between a word's letters and reducing contrast slow word recognition, yet only the latter manipulation produces contextual compensation. In light of these findings, the strategy version of the model states that contextual compensation is observed when bottom-up processing is impaired, independent of recognition latencies. In the present examination, the problem findings for the original, time-locked version of the model were first replicated and extended. The lack of an interaction between repetition and context effects was found in both lexical decision and pronunciation tasks and for different stimulus quality conditions (clear, asterisk and mask). Also, the strategy version was strongly supported on examining how the effects of repetition and degradation combine. The second goal of the present experiments was to develop a method to predict when a degradation manipulation will require contextual compensation. Jacoby (1983) found that the perceptual identification of a brief, masked target during its second presentation was better if the word had been seen in an unrelated context (which involves more physical processing) than in a related context. It was predicted that any degradation manipulation which produces the same pattern of repetition effects should be relatively impoverished and require contextual compensation. However, further research is needed on this question. Finally, the nature of the strategy involved in contextual compensation was examined. In all three stimulus quality conditions there was a trend for expected but unrelated targets to be facilitated. The implications of these findings for the locus of expectancy effects was discussed.
Degree
Ph.D.
Subject Area
Educational psychology
Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server.