Redintegrative memory for novel shapes (remembering the whole shape when only a part is presented) was studied as a function of whether or not S had learned names for the shapes. Naming markedly facilitated short-term redintegrative recognition, but had no effect on simple recognition. The results support the hypothesis that naming facilitates the formation of an integrated representation in which information about the different contours of the shape is more closely linked than in a nonverbal representation. Spiker (1956), Ranken (1963), and others have obtained facilitative effects of stimulus naming in a variety of memory tasks. Ss' comments during post-experimental questioning suggest that with unfamiliar shapes one effect of names is to provide stable, integrated representations of the shapes: Ss often report that the names give them "something concrete" to remember (Ranken, 1963). An integrated representation of a complex shape may be thought of as one in which the representations of the various parts or contours of the shape are closely linked, so that the occurrence of one tends to evoke the others. On this interpretation, the degree of integration of S's representation would be reflected in a redintegrative memory task, in which he is shown part of a shape and asked to remember the rest. In the present experiment, the hypothesis that naming facilitates the formation of integrated representations of unfamiliar shapes was tested by determining the effect of naming on performance in a redintegrative recognition task.
METHOD
StimuliThe stimulus shapes ( Fig. I) had irregular top and bottom contours and straight sides. Thus each shape could be divided by a horizontal line into a top and bottom half, each half containing one of the two contours which defined the shape. The shapes, I in. wide, were shown on a rear-projection screen at a viewing distance of about 20 in.
DesignTwo training conditions, Named and Unnamed, and two test tasks, Redintegration and Recognition, were used in a 2 by 2 design. Two replications were run, approximately a month apart, with four Ss (introductory psychology students) in each cell in each replication.During training, Ss in the Named condition learned names for the shapes, while Ss in the Unnamed condition had equal practice in discriminating both top and bottom contours of the shapes in a delayed matching-ta-sample task, but were not given names. In the Redintegration test task, in essence, S was shown a top contour, followed by a bottom contour, and was asked to indicate whether the second contour came from the same shape as the first. The task may be thought of as a recognition test of redintegrative memory, in which S's ability to remember the correct bottom contour of a shape, given the top contour, is tested by his ability to recognize the presented bottom contour as correct or reject it as incorrect.The Recognition test task served as a control for the effect of naming on memory for specific contours. In essence, S was shown a top contour, followed by another top contour, ...