Three approaches used in the analysis of semantic differential data to determine similarity in meaning between profile ratings on concept pairs are discussed. These approaches are: (1) to compute mean scores on each scale for each concept and to compare concepts on a scale-by-scale basis; (2) to form mean scores on each dimension for each concept and to compare concepts on a dimensional basis; and (3) to compute the D statistic between ratings on two concepts and use this as a multidimensional measure for comparison purposes. A figure and tables illustrate the three approaches. (DB)