Meaningfulness and pronounceability in the coding of visually presented verbal materials.

Experiment I was a replication of a previous experiment by Gibson, Bishop, Schiff, and Smith concerning the perceptual threshold for identifying meaningful, pronounceable, and nonsense trigrams. The thresholds were found to depend upon the relative frequencies of the item types in the stimulus set, indicating that the earlier report was artifactual. In two additional experiments, 5s were asked to give partial reports of displays of each of the item types. In Exp. II, immediate recall was highly correlated with meaningfulness. In Exp. Ill, 5s were given practice and were timed at pronouncing the items aloud. The pronounceable items were recalled best, and several qualitative differences in short-term retention of meaningful and pronounceable items indicated that coding of pronounceable items is more related to overt response processes than is coding of meaningful items. A fundamental problem in the study of human information processing concerns the means by which physical stimulation is initially coded and represented in the nervous system. Gibson, Bishop, Schiff, and Smith (1964) have reported the results of an experiment which would appear to be an important contribution to the problem. Their study examined the extent to which pronounceability and meaningfulness served as grouping principles in the perception and retention of visually presented verbal materials. The stimuli used were trigrams in which the same letters could be arranged into a meaningful, pronounceable, or nonsense permutation. The pronounceable items were identified at a lower threshold luminance level, as determined by an ascending method of limits. The authors concluded that pronounceability provided "a better grouping principle for reading or coding to speech units [p. 173]." Thus, the results of the Gibson et al. (1964) study indicate that the perception of visually presented verbal material is accomplished, at least in part, by means of a translation into a speech code. Their data add support to the proposition that much visual