ABSTRACT
Since the revival of cognitive psychology in the 1950s, significant contributions have been made to the knowledge base on the reading processes and on the processes involved in learning to read. From the work of Eleanor Gibson and her colleagues on the role of grapheme—phoneme correspondences in word recognition to the most recent work on phonemic awareness and decoding, much has been learned about different components of reading and learning to read. Nevertheless, many of the problems that drove the earliest studies of reading in Leipzig, Paris, and New York before the 1900s remain unresolved, including the features utilized in competent word or character recognition, the role of subvocalization in reading connected text, and the function of letter—sound or character—sound patterns in reading acquisition. Studies from a number of different orthography—language communities suggest that some components of reading and learning to read, such as the extraction and use of frequency information and automatization of basic processes, are universal, whereas others, such as the relationship of decoding ability to comprehension, are more language specific. One challenge remaining for researchers is to build more powerful models that can predict which orthography—language features contribute to which reading processes. A second challenge is to build such models from the more general concepts, strategies, and mechanisms of cognitive psychology, language development, and linguistics. For example, overgeneralization in letter—sound learning needs to be related to the processing models posited for overextension or overgeneralization in general language processing such as the learning of morphology; phonemic awareness ideas need to be related to the more general findings on the development of recoding of visual stimuli for storage in working memory, the growth of general processing speed, and the use of attentional mechanisms; and models for learning variant letter—sound patterns need to be built on mechanisms posited for other types of rule learning. The more we can base our models on general processing strategies rather than on mechanisms unique to our specific domain, the more compelling will be our claims. Our unified dream should be of a Nobel Prize for reading research, giving international recognition to breakthroughs that demonstrate how simple, general processing strategies can account for learning to read in many different orthographies and languages.
