A direct investigation into the grammatical abilities of Broca's and Wernicke's aphasics sought to obtain critical evidence for a revised model of the functional neuroanatomy of language. We examined aphasics' ability to make grammaticality judgments on a set of theoretically selected, highly complex syntactic structures that involve, most prominently fine violations of constraints on syntactic movement. Although both groups have been thought to possess intact abilities in this domain, we discovered severe deficits: Broca's and Wernicke's aphasics (whose performances differed) exhibited clear, delineated, and grammatically characterizable deficits - they follow from the Trace-Deletion Hypothesis, which is motivated by independent comprehension results. These conclusions have both linguistic and neurological implications: Linguistically, they show that the aphasic deficit interacts with more than one module of the grammar. Namely, it manifests not only when the thematic module is called for in interpretive tasks but also when constraints on syntactic movement are tapped in a study of judgment. Neurologically, the results support a view of receptive grammatical mechanisms in the left cortex, which is functionally more restrictive than currently assumed; neuroanatomically, however, it is more distributed.