Differentiation and integration in human language Reply to Marslen-Wilson and Tyler

N400-type effect [7]. This evidence, that the irregular-stem mapping in the intact system is no more semantic than the regular-stem mapping, leads us to interpret the co-occurrence of semantic deficits and of disrupted access to irregular past tense forms as accidental rather than causal in nature. This interpretation is supported by the report of an anomic patient with a deficit for the irregular past tense but no semantic deficit [8]. The second problematic aspect of M&P's model is that it seeks to explain poor performance with the regular past tense purely in terms of general phonological processing deficits, and rejects the possibility of a deficit specific to morphological or morpho-phonological factors. This generates clear predictions, which we have falsified in two recent studies. Our experiments use a speeded same – different judgment task, where participants are asked to detect differences between the past tense and stem of regular (played/play)and irregular (taught/teach) past-tense verbs, matched pseudo-regular and irregular word pairs (trade/tray; port/peach), and matched sets of non-words. In one study [5], patients with documented difficulties with regular inflection performed consistently worse on the regular past-tense pairs than on the phono-logically matched pseudo-regular and non-word pairs. Furthermore, performance on the task did not correlate with the patients' phonological processing difficulties, which ranged from very mild to severe. Preliminary results from a second study, using fMRI to examine activation patterns in the intact brain for the same experimental contrasts, showed differential activation for regular pairs in brain areas that overlap with regions that are damaged in the patients, and where purely phonological factors can again be excluded. In summary, although we remain agnostic as to the types of mental computation implicated by these results, we do not believe that connectionist models of the type proposed by M&P represent a promising direction, either for resolving the past-tense dispute, or for capturing the specific functional and neural architecture of the human language system. Associations and dissociations in the processing of regular and irregular verbs: electrophysiological evidence. The work of Marslen-Wilson and Tyler contributes importantly to our understanding of the neural basis of language processing. The arguments given in their letter [1], however, do not refute our view [2– 5] that both regular and irregular verbs are processed in the same integrated system, and that performance on regular verbs is more affected by a disruption of phonological processes whereas performance on irregulars is more affected by …