This study tests the hypothesis that agrammatic comprehension is due to a computational rather than a structural language deficit. It is claimed that grammatic Broca's aphasics do not meet the temporal constraints in the activation of different types of linguistic information necessary for normal parsing. These temporal constraints are investigated in two experiments using a crossmodal syntactic priming paradigm. Each experiment tests the effect on recognition of grammatical versus ungrammatical links between an auditory sentence fragment (the prime) and a visually presented word (the target). The experiments differ in the interstimulus interval (ISI i.e., the amount of time provided between the offset of the auditory prime and the onset of the visuai target. Experiment 1, with an ISI of 0 msec, reveals that Broca patients — much like the control groups — show a grammaticality effect. In contrast to normal students and age matched controls, however, these patients are in general much slower in making lexical decisions when auditory context is present as compared to when visual targets are presented in isolation. Experiment 2 with an ISI of 200 msec demonstrates that when additional time is given to process the syntactic context, Broca's aphasics in contrast to age matched controls show faster decision times on the target than when the ISI is 0 msec. The results are in accord with the view that agrammatic Broca's aphasics have not lost their syntactic knowledge, but that their a grammatic comprehension behavior is due to the inability to process the linguistic information within a given time frame. More generally, the findings suggest that language processing can break down once activation of different types of linguistic information does not follow the normal time pattern.