Search

Data are presented from a neurological patient (M. P.) with an acquired deficit for naming words with atypical spelling-sound correspondences. In Experiment 1, the degree of consistency within neighborhoods of orthographically similar words had a parallel impact on M. P.'s pronunciations of regular and irregular words and nonwords. This result is more compatible with models in which the same basic procedure, sensitive in a graded fashion to both frequency and consistency, computes pronunciations for all types of letter strings than it is with models postulating separate lexical and nonlexical mechanisms. In Experiment 2, both correct and regularized pronunciations of exception words yielded response times significantly modulated by word frequency, a finding not predicted by any current model. Neuropsychological observations provide an important additional source of evidence regarding models of cognitive function.

Three patients with semantic dementia, involving progressive deterioration of semantic memory, performed immediate serial recall of short sequences of familiar words. On the basis of their performance in other tasks of word comprehension and production, the stimuli were selected individually for each patient as either known or unknown words. All patients showed a marked advantage in recall of known as compared to familiar but now unknown words. Errors consisted primarily of incorrect combinations of correct phoneme sequences in the stimulus string, with a large number of errors preserving onset/rime syllable structure (e.g., mint, rug reproduced as 'rint, mug'). Discussion focuses on the implication of these errors for the structure of phonological representations, and in particular on a hypothesis that meaning plays a crucial role in binding the elements of phonological word forms.

It is generally assumed that access to phonology for words written in logographic Japanese Kanji must be mediated by access to their meaning. This proposal was examined in a semantic categorization task with homophones. If the assumption about Kanji processing were true, then homophony should have no effect on semantic judgments. However, there was a significant homophone effect: Reaction times (RTs) were longer and more errors occurred to homophone foils than to control foils. There was also a significant effect of visual similarity: Incorrect target words that were visually similar to correct exemplars of the category names yielded longer RTs and higher error rates. The effects of both visual similarity and homophony were obtained even under conditions of pattern masking (though only on errors). The data suggest that the reading of Kanji is characterized by parallel access to semantics from orthographic and phonological representations.

Two experiments on oral reading of single words compared naming performance in pure blocks of nonwords or exception words with performance in blocks of randomly mixed nonwords and exception words. Ss named exception words faster and made fewer regularization errors when they were not also prepared for nonwords. These data suggest Ss inhibit or ignore the computation of assembled phonology when only exception words are expected. Ss named nonwords faster, but no more accurately, when low-frequency exception words were not also anticipated. Thus, Ss' readiness to execute assembled phonology appears to be adjusted in relation to the likely time course of retrieval of learned pronunciations, when the latter must be attended to. This evidence for strategic dissociation between sublexical and lexical translation is discussed in relation to current models.