Agrammatism as evidence about grammar.

A variety of experimental paradigms has yielded surprisingly fine-grained evidence about the kinds of syntactic information to which agrammatic aphasics are sensitive. This paper contrasts three accounts of agrammatism which draw quite different conclusions about the implications of this disorder for normal function: the chain-disruption, trade-off, and mapping hypotheses. Counterarguments to the chain disruption and trade-off hypotheses are presented, and it is argued that agrammatism provides considerable support for the modularity of syntax but provides no evidence more specific than that regarding the psychological reality of government binding theory vis-à-vis other current theories of grammar.