Recent advances in brain imaging with positron emission tomography (PET) and functional magnetic resonance imaging (fMRI) coupled with studies of event-related potentials (ERPs) no permit us to examine safely the normal human brain while it functions. These studies provide a much clearer under-standing of the architecture of language systems in the brain and exciting new insights into how the normal human brain produces observable behavior. Neural-network models should yield important conceptual insights into the interpretation of the emerging data on functional brain imaging.