In two fMRI experiments, participants named pictures with superimposed distractors that were high or low in frequency or varied in terms of age of acquisition. Pictures superimposed with low-frequency words were named more slowly than those superimposed with high-frequency words, and late-acquired words interfered with picture naming to a greater extent than early-acquired words. The distractor frequency effect (Experiment 1) was associated with increased activity in left premotor and posterior superior temporal cortices, consistent with the operation of an articulatory response buffer and verbal self-monitoring system. Conversely, the distractor age-of-acquisition effect (Experiment 2) was associated with increased activity in the left middle and posterior middle temporal cortex, consistent with the operation of lexical level processes such as lemma and phonological word form retrieval. The spatially dissociated patterns of activity across the two experiments indicate that distractor effects in picture–word interference may occur at lexical or postlexical levels of processing in speech production.

We report on two brain-damaged patients who show contrasting patterns of deficits in memory and language functioning. One patient (AW) suffers from a lexical retrieval deficit and failed to produce many irregularly inflected words such as spun, forgotten , and mice , but demonstrated intact production of regularly inflected words such as walked and rats . She also had preserved declarative memory for facts and events. The other patient (VP) presented with a severe declarative memory deficit but showed no signs of impairment in producing either regular or irregular inflections. These patterns of deficits reveal that the retrieval of irregular inflections proceeds relatively autonomously with respect to declarative memory. We interpret these deficits with reference to three current theories of lexical structure: (a) Pinker's “words and rules” account, which assumes distinct mechanisms for processing regular and irregular inflections and proposes that lexical and semantic processing are subserved by distinct but interacting cognitive systems; (b) Ullman's “declarative/procedural” model, which assumes that mechanisms for the retrieval of irregular inflections are part of declarative memory; (c) Joanisse and Seidenberg's connectionist model, in which semantic information is critical for the retrieval of irregular inflections.

We report the case of an Italian anomic subject who was invariably able to provide the auxiliary of verbs he failed to produce in oral naming tasks. This pattern of performance contrasts with that of another Italian-speaking patient documented by Miceli and Caramazza (1988) who showed a selective impairment in accessing syntactic features of words, but not their phonological forms. This double dissociation suggests that syntactic and phonological information in the lexicon are accessed independently and represented in distinct neural structures.