Neurological Distribution of Processing Resources Underlying Language Comprehension David Swinney University of California, San Diego Edgar Zurif Brandeis University and Boston University School of Medicine Penny Prather Boston University School of Medicine and BVAMC Tracy Love University of California, San Diego Abstract Using a cross-modal lexical priming technique we provide an on-line examination of the ability of aphasic patients to construct syntactically licensed depend- encies in real time. We show a distinct difference be- tween Wernicke's and Broca's aphasic patients with respect to this form of syntactic processing: the Wernicke's patients link the elements of dependency relations in the same manner as do neurologically intact individuals; the Broca's patients show no evidence of such linkage. These findings indicate that the cerebral INTRODUCTION The work presented in this paper seeks to advance our understanding of both sentence processing and func- tional neuroanatomy. In what follows, we link a syntactic comprehension failure to a neurologically localizable and elemental processing disruption. This work provides a basis for inferring how different brain regions normally serve aspects of sentence comprehension. A relevant fact here is that the contrasting syndromes of Broca's aphasia and Wernicke's aphasia can be distin- guished with respect to lesion site. The brain area asso- ciated with Broca's aphasia often involves the foot of the third frontal convolution (Broca's area) as well as adja- cent and deeper areas (Alexander, Naeser, & Palumbo, 1990; Benson, 1985; Dronkers, Shapiro, Redfern, & Knight, 1992; Mohr, 1976; Naeser, Palumbo, Helm- Estabrooks, Stiassny-Eder, & Albert, 1989; Vignolo, 1988). Yet, notwithstanding the extent of this area, the modal © 1996 Massachusetts Institute of Technology tissue implicated in Wernicke's aphasia is not crucial for recovering syntactically licensed structural depend- encies, while that implicated in Broca's aphasia is. More- over, additional considerations suggest that the latter region is not the locus of syntactic representations per se, but rather provides the resources that sustain the normal operating characteristics of the lexical process- ing system-characteristics that are, in turn, necessary for building syntactic representations in real time. lesion site for Broca's aphasia remains distinguishable from that for Wernicke's aphasia, where the associated lesion is typically inferior to the Sylvian fissure and mostly involves the superior temporal gyrus (Wernicke's area) (Benson, 1985; Vignolo, 1988). The speech (output) patterns exhibited in these two syndromes are noticeably different from each other. Broca's patients present with nonfluent and telegraphic output; Wernicke's patients present with fluent, relatively empty speech (Goodglass & Kaplan, 1972). The matter is more complex with respect to compre- hension, however. Although both groups show a compre- hension problem at the sentence level, particularly for complex sentences, the problem appears more focused on structural processing-more restricted to syntax- for Broca's than for Wernicke's patients. To be sure, there are claimed exceptions to this generalization. At least one reported Broca's patient demonstrates a comprehension pattern that appears to resist any kind of structural Journal of Cognitive Neuroscience 8:2, pp. 174-184