On the interpretation of word frequency as a variable affecting speed of recognition.

The duration for which a printed English word must be presented visually to a subject in order for him to recognize it is inversely correlated with the frequency of occurrence of the word in large samples of written English (2, 3, 4, 7, 8, 11). Since the former quantity (the duration threshold) is generally regarded as a perceptual variable and the latter (word frequency) as a response variable, this correlation offers a point of departure for the formulation of perceptual phenomena in behavioral concepts. The object of the present study is to test experimentally an assumption basic to one interpretation of this correlation. A mathematical formulation of the experimental data based upon this interpretation will be presented in a subsequent report. The interpretation to be considered here can be characterized as a responseemission theory. We may think of the momentary probability of a word (defined as the strength of S's tendency to emit that word in preference to any other) as a quantity that fluctuates widely from moment to moment in accordance with changes in innumerable environmental and organismic conditions that affect the emission of words. Over a time period of considerable length the average of these momentary probabilities will be a relatively stable statistic, which we shall call the base probability of the word. Visual exposure of a word to S for a brief length of time At is assumed to represent an environmental event tending to cause emission of the ex1 Present address: Wrightsville Beach, N. C. posed word. The momentary probability of a word following its exposure may therefore be analyzed into two components: a component due to the ordinary impulses to emission of the word, whose average value is the base probability; and a component due to the additional impulse of the word's visual exposure. Consequently, the average probability of a word following each of a number of exposures of given duration must be greater than the corresponding average base probability of the word. A given level of probability following exposure can result either from a relatively large component due to base probability plus a small additional component due to exposure or from a relatively small component due to base probability plus a large additional component due to exposure. It follows that the duration threshold of a word, which is defined as the duration of exposure for which 50% of S's reports following exposure are correct, will be lower for a word with high base probability than for a word with low base probability. In this interpretation of the experimental data, word frequency serves as an estimate of base probability. In the cited experiments word frequency was determined from the published tables of the Lorge magazine count (10). This count is based on a sample of 4.5 million words of text taken from issues of five popular magazines