A test of the multiple connections model of reading acquisition

Within the framework of Society of Mind Theory (Minsky, 1986), learning to read is conceptualized as a process of creating new communication links or neural connections between an existing visual society and an existing linguistic society. Four visual-linguistic connections may become functional: letter-phonemic code, whole word-semantic code, whole word-name code, letter sequence-aural syllabic code. The hypothesis was tested that more than one of these visual-linguistic connections must be taken into account in predicting reading achievement. Results showed that the combination of the composite letter-phoneme variable and the composite whole word-semantic code variable accounted for significantly more variance in oral reading than did either single variable at the end of the first grade. Groups with large absolute discrepancy (1 or more standard scores) or small absolute discrepancy (1/3 standard score or less) on corresponding visual and linguistic skills differed significantly in both oral (whole word-semantic code composite) and silent reading (whole word-semantic code and letter sequence-aural syllabic code composites). There was a relationship between the number of large discrepancies and reading achievement. Results are discussed in reference to neuropsychological models of connectionism (Rumelhart & McClelland, 1986) and working brain systems (Luria, 1973).