READING RATES 511 



beyond this and estimate information processing rates in the brain from 

 the performance of hghtning calculators, by dividing the performance 

 of the calculation into a sequence of tasks ecjuivalent to consulting 

 memorized multiplication tables and performing additions. It is hard to 

 interpret such a study clearly, for it is quite possible that there are many 

 sorts of mental acts which take different times to perform, just as multi- 

 plication and addition take different times in an electronic computer. 

 A tentative experiment we performed indicated something of the sort. 



Randomized lists were made up from \'ocabularies (a) of names of 

 common animals and vegetables in equal numbers, and (b) of common 

 men's and women's names in equal numbers. In reading these, a subject 

 w^as asked, not to read the word aloud, but merely to press one key with 

 his right and another key with his left hand; in (a) left-animal, right- 

 vegetable; in (b) left-man, right-woman. The same subject later read the 

 lists aloud. Pressing keys took 40 per cent longer than reading aloud. 

 (The additional time is not related to the keying operation itself; for a 

 2 word list, for example, keying speed is much faster than reading speed.) 

 Presumably an additional mental operation was involved, but it was not 

 one for which the time was equal to that for reading. This experiment 

 was not pursued further, partly because no clear conclusion could be 

 drawn from it. Had it been pursued and randomized lists of the same 

 words used repeatedly, the rate might have gone up. Conceivably, cow 

 and horse could become for a subject merely different ways of spelling 

 left, and lettuce and carrot variant spellings of right. In this case we 

 would end with a two-word reading experiment. 



Reading Rate as a Psychometric Datum 



It is interesting to speculate on the possible relationship of reading 

 rate to general intelligence or some other aptitude. Certainly, Fig. 5 

 indicates some such relationship. We might also ask in connection with 

 Fig. 3, does a rate which falls less rapidly with frequency of occurrence 

 indicate a larger vocabulary, and can we measure vocabulary by reading 

 speed tests? Certainly, measuring the speed of reading aloud is very 

 simple, and such tests might have some psychometric utility. 



The Channel Capacity Required for Satisfactory Communication 



In conclusion, we cannot help but wonder that the highest information 

 rate noted — 43 bits/second — is so much lower than the channel 

 capacity* of a telephone or a television circuit (around 50 thousand 



* This is the limiting channel capacity given by (2.1) of Appendix II. The prac- 

 tical rate at which binary digits can be sent over a telephone circuit with simple 

 equipment is less than 1(X)0 bits/second. 



