Lexical competition for production in a case of nonfluent aphasia: converging evidence from four different tasks.

We report an individual with Broca's aphasia (J.H.M.), who exhibited powerful lexical context effects in word production tasks. In an adjective-noun production task (Experiment 1), J.H.M.'s production accuracy decreased as the number of adjectives in the phrase increased (e.g., curly hair vs. long curly hair). In a picture pair naming task (Experiment 2), J.H.M.'s naming accuracy was high, but her naming latencies were abnormally delayed when pairs were semantically related (e.g., goat and pig). This pattern was not observed for older controls. In a computerized Stroop task (Experiment 3), J.H.M.'s naming latencies were abnormally prolonged in the conflict condition, relative to a baseline colour naming task. This effect was far in excess of that for controls. Finally, in a blocked cyclic naming task (Experiment 4), J.H.M.'s accuracy was poorer and her latencies slower when the pictures were semantically related than when they were unrelated, and this effect built up across successive presentation cycles. It was far in excess of that exhibited by older controls. We propose that J.H.M.'s pattern of impairment across these four very different tasks suggests an impairment to a lexical control mechanism, whose normal function is to modulate the flow of activation throughout the lexical network, so as to minimize the competitive effects of nontarget lexical items.