On the Influence of Pretraining on Recognition Thresholds for English Words

Abstract

Four experiments assessed the influence of a pretraining session on the recognition of English words. While pretraining with artificial words establishes a word-frequency effect, pretraining with the English words reduced the effect. When pretraining frequency was held constant, recognition thresholds for words of low natural frequency were reduced by pretraining while those for words of high natural frequency were unchanged. When natural-language frequency was held constant, differential pretraining frequency affected thresholds within lists of words of low but not of high (AA) natural frequency. Two experimental paradigms have been dominant in the literature relating word frequency to recognition thresholds. Howes and Solomon (1951) required their subjects to recognize English words presented tachistoscopically; no pretraining was involved, and no hints (or 'sets') were given that any particular subset of words might be presented. Solomon and Postman (1952) varied the frequency with which their subjects experienced artificial words in pretraining, and then they obtained tachistoscopic recognition thresholds for these words. In both experimental paradigms, an inverse relationship between frequency and thresholds was demonstrated: high-frequency words could be recognized with shorter presentation times than could low-frequency words. Although similar results were obtained in the two situations, it is not clear that the paradigms are comparable or that the same processes are involved in both.