How is linguistic memory accessed? A psychophysiological approach.

The role of "subvocalization" during language comprehension, especially reading, is examined. Four arguments against it having a role in accessing memory are erroneous because 1) its latency is much shorter than is conventionally stated; 2) rate of visual information processing is erroneously estimated by failing to distinguish between reading and scanning; 3) covert speech does not disappear in the competent language performer; and 4) the argument that subvocalization is an epiphenomenon is irrelevant. Rather, data support the generalization that covert speech is present during all cognitive functioning and that its specific topography is discriminatively related to the class of phoneme being processed. It is thus inferred that during cognition the speech musculature generates a phonetic code that may function to access linguistic memory. However, since there are also numerous other psychophysiologic events associated with covert speech, a multichannel processing system is hypothesized wherein speech, visual, and kinesthetic modalities interact with the brain. Illustrations are given of how this accessing model is compatible with existing holographic and feature analyzer models of memory. Data are presented that illustrate how phonetically encoded neuromuscular events can be directly measured through psychophysiologic methods. It is hypothesized that cognitive processes are generated when cybernetic neuromuscular circuits selectively interact. Consequently, all components of these neuromuscular circuits serve a function during cognition so that a role for "subvocalization" (a muscular component) cannot be ruled out in an apriori manner.