Strategies of decoding actor-object-relations by aphasic patients.

By means of a sentence-picture matching task, 22 Broca-aphasics, 22 Wernicke-aphasics, 16 patients with damage to the right hemisphere and 16 neurologically normal subjects were tested for their ability to understand the actor-object-relation within a sentence. The 4 groups were approximately matched for sex, education, age and general severeness of disease. The stimulus sentences were constructed in such a way that the actor-object-relations were unambiguously clear by morphological means as e.g. case endings; thus, no error should be made by a subject operating on a purely algorithmic grammatical basis. One half of the sentences were irreversible, the other half reversible, i.e. the identification of the actor-object-relation was helped by a semantic cue in one half of the sentences, in the other half not. Crossed with this semantic factor was the syntatic factor "constituent order": one half of the sentences were presented in the normal order "actor-action-object", the other half in the topicalized form "object-action-actor" which is quite possible in German because of its free word order. Overall, the Broca-aphasics made less errors than the Wernicke-patients. The error pattern of the aphasics suggests that both aphasic groups remain responsive to semantic constraints as well as to constituent order: both groups made more errors in reversible than in irreversible sentences and both groups made more errors in topicalized than in normal sentences. There were no significant group differences in these two respects. The only group difference was that Broca-aphasics tended to neglect the syntactic aspect of constituent order if a semantic cue was given, while the Wernicke-aphasics continued to take the constituent order into account even in sentences where a semantic cue was present. In the light of the findings of the experiment, the hypothesis that Broca's aphasia is characterized by a supramodal blockade of syntax is disputed.