Autonymy, cohort, and logogen models of lexical access: When and how does interactive processing occur?
Edward C. T. Walker · The Journal of the Acoustical Society of America · 1981
The stages of “autonymy” models of lexical processing accept as input the output of a prior stage and operate without active constraint from other stages of processing or sources of knowledge. Outputs from different stages may be compared by a general purpose processor in making response decisions, but processing interactions among stages are prohibited. By contrast, different types of processing proceed in parallel in “interactive” models, with current results from different processes contraining one another. In “cohort” models words are individual active processors which operate on full or partial output from all types of processing. In “logogen” models recognition occurs passively when a threshold level of activation has been reached; these thresholds being raised or lowered by results from a variety of processes. Thus, the argument turns not on whether, but on how and when interaction occurs. There is some evidence that ambiguity and frequency effects persist in the face of constraining contexts, but results from a variety of on-line experiments indicate that the interaction takes place very early in the process of word identification and not exclusively in the process of making response decisions following identification. This evidence also supports the view that interaction occurs in active recognition processes having access to multiple sources of information.