Mechanisms of word identification.

Abstract

Three word-identification experiments suggest that a model derived from experiments with pseudowords can be applied successfully to word identification. The first experiment used a technique known to disrupt subjects' ability to use the familiarity of higher order approximations to English without impairing single-character identification. Word identification paralleled the size of the familiarity effect in comparable pseudoword work, a result that suggests both word identification and the familiarity effect are mediated by common mechanisms. The latter are thought to include a postidentification character buffer and a sequential scan operator responsible for converting the character string to a verbal-temporal representation. The conversion includes a parsing function which breaks the character string into verbal units. Two subsequent experiments illustrated predictions of the account by disabling the scan completely or by forcing it to deal with whole words and with syllable and nonsyllable subsets of a word. Taken together, the data confirm the role of higher order verbal units in word identification and suggest the structural components of a verbal-mediation theory of reading.To what extent should theories and data based on identification of pseudowords predict word identification ? Some theorists (e.g., Kolers, 1972) suggest words are read as visual symbols, a position which implies a direct visual route to meaning. Clearly, such views do not permit prediction of word identification from performance on nonsense materials. Pseudowords lack meaning, and thus, in this view, performance on such materials could not reflect reading processes in any di-The research was supported by National Research Council of Canada Grant APA-318 to D. J. K. Mewhort. We thank M. J. Withey and Pat Crawford for assistance in conducting the experiments. We also thank R. von Konigslow, A. J. Campbell, E. E. Smith, and K. T. Spoehr for critical comments on various drafts of the work.

Processing spatial information and the selective-masking effect

“…Building from a series of free-recall studies using pseudowords, Mewhort and Beal (1977) have suggested a model for word identification which requires a postidentification character buffer. The model, like that of Smith and Spoehr (1974), implicates a scanparse operation which reads the buffer, forms a series of verbal-temporal units, and passes the units into short-term memory.…”

Time course of visual information utilization during fixations in reading.

“…Recent evidence (Mewhort, 1974;Mewhort & Beal, 1977) has lead some current theorists to propose that this scanning mechanism deals with syllables rather than letters (Adams, 1981;Mewhort & Campbell 1981). McConkie (1979) suggests that the region from which information is acquired shifts from left to right within a fixation in reading.…”

Scanning, holistic encoding, and the word-superiority effect

“…Although subjects in that study were not told to consider the entire letter array, the appearance of different terminal letters with the postexposure mask may have induced the subjects to scan across the array from left to right to discover which terminal letter would complete each exposure. Since subjects self-initiated each presentation, the scan might have been accomplished with an eye movement, or it could have been an internal scan of a posticonic buffer (Mewhort & Beal, 1977;Mewhort, Merikle, & Bryden, 1969). Due to the subjects' experiences in reading, scanning across the array in this manner may have induced them to attempt to encode the entire letter string holistically.…”