This paper uses a discussion of experiments with aphasics' use of verbally derived nouns to illustrate how one linguistic model may be superior to another in accounting for the facts of verbal behavior. The models involved are the transformational, which relates derived nominals to their source verb and lists only the verb in the lexicon, and the lexicalist, which lists both noun and verb together in the lexical entry. Subjects, sufferers from a type of aphasia in which the ability to use verbs is impaired to a greater extent than the ability to use nouns, were given a stimulus word (noun or verb) and asked to use the word in a sentence or explain its meaning. It was revealed that aphasics who have difficulty in using a verb can quite often produce the nominal derived from that verb with relative ease. The phenomenon was interpreted as suggesting that the lexical entry is coded in the brain in both its verbal and nominal forms and under the noun-facilitation circumstances, the nominal form is retrievable. It was felt that the lexicalist approach was able to reflect this phenomenon of brain function in a simple and elegant way, whereas the transformational model would be able to account for the phenomenon only by a complicated and implausible set of principles of brain function. (FWB)