Subjects named or categorized a picture preceded sometime earlier by itself or by its verbal label as well as a word preceded by itself or a pictorial counterpart. The extent to which the second item benefited from prior processing was a function of the mode of the second item, the mode of the previous occurrence of the concepts, and the task. The results of the naming task of Experiment 1 suggested that words are more generic, in that they tended to facilitate processing of subsequent occurrences of the concept regardless of the mode of the second occurrence, whereas pictures tended to facilitate only subsequent pictures. In fact reading a word was not faster after having just named a pictorial representation of that concept. Experiment 2 explored, and ruled out, the possibility that this failure to find facilitation was the result of an ingibitory mechanism. In general, the data are consistent with the idea that pictures and words consult the same semantic system for their interpretation but that words activate a more generic set of semantic information than do pictures.