False recognition responses of high and low SES kindergarten subjects to associatively and acoustically related words were measured. Acoustic attribute dominance for all subjects, and relatively greater acoustic attribute dominance for low SES subjects was predicted. Results indicated that subjects encoded on both attribute dimensions, with low SES subjects producing significantly more recognition errors. The importance of the results for the developmental shift hypothesis of attribute dominance was discussed as well as the importance of SES as a variable in future memory attribute research. (Author)