Why do some aphasics show an advantage on some tests of nonpropositional (automatic) speech?

Sixteen aphasic patients were given three pairs of tasks that compared the production of the same items in either propositional or nonpropositional contexts. A nonpropositional number production task involved counting from 1 to 10 while the propositional version of that task involved naming the Arabic numbers 1 to 10 in nonconsecutive order. A nonpropositional picture-naming task involved naming pictures with the aid of familiar phrase cues (e.g., Don't beat around the BUSH) while in the propositional version the cues were novel phrases (e.g., Don't dig behind the BUSH). Finally a nonpropositional phrase repetition task involved repeating well-known phrases while the propositional version involved repeating novel phrases. The group as a whole showed strong nonpropositional advantages for number production and picture naming with a somewhat weaker advantage for phrase repetition. Only 5 of the individual patients showed nonpropositional advantages on all three pairs of tasks: the remaining 11 patients showed a significant nonpropositional advantage on one or two of the pairs of tasks, but not on all three. All of the patients showed a nonpropositional advantage on at least one pair of tasks, and there were no examples of better performance on the propositional than on the nonpropositional version of any task. Contrasting patterns of performance seen in different patients was related to their performance on a battery of cognitive and linguistic tasks that was given to each patient.