Abstract
Sixty-seven patients who had suffered from aphasia for less than 15 days were investigated with a shortened version of the Aachen Aphasia Test and clinical neurological examination. Another 30 acute aphasics were excluded because their assessment could not be completed in one session. The data were interpreted by means of cluster, discriminant and regression analysis. The clusters reflected mainly degree of impairment. In addition, mainly phonology, repetition and Token Test were used for cluster assignment. The clusters proved to be unstable with respect to long-term development. The best predictor for degree of chronic impairment was the ability to communicate.