Sixty-eight variables, consisting of the results of 25 subtests of the “Standard Language Test of Aphasia” (SLTA) and 43 speech symptom variables evaluated from conversational speech, were classified by the cluster analysis method. The relatively high similarities among the subtests within SLTA, and the relatively low similarities between SLTA-subtests and the other speech symptoms and among the speech symptoms themselves are shown in the dendrogram. These results suggest thst many aspects of the speech symptoms of aphasia cannot be fully grasped by the SLTA used as a psychometric test-battery.