Hierarchical cluster analysis of data from the sorting of noun words was used to compare semantic structures in sixty-three profoundly deaf and sixty-three hearing adolescents. In the first study, performance differed only for a set of words referring to sounds, where deaf persons have no experience, and not for a set of common noun words and pictures. In the second study, differences between matched sets of high- and low-imagery words were comparable for sixty-three deaf and sixty-three hearing subjects. It is concluded that deaf subjects manifested abstract hierarchical relations and were not dependent on "visual mediators" or hindered by the absence of "acoustic mediators."