Distinguishing Natural Language Processes on the Basis of fMRI-Measured Brain Activation