510 



THE BELL SYSTEM TECHNICAL JOURNAL, MARCH 1957 



Table V 



point at which the reader is uttering words continuous! j^ as fast as he 

 can. This is consistent with a strong subjective feehng that what hmits 

 the rate is the difficult}' of "recognizing" the word as one looks at it, 

 and that once the word is recognized one can utter it while recognizing 

 the next word. 



It would of course be wrong to conclude from this experiment that 

 multisyllable words are in general recognized as quickly as single syllable 

 words, for it would be possible to recognize one among a known group 

 of 16 multisyllable words without looking at the whole word. Indeed, 

 Fig. 3 indicates a substantial difference of reading rate between one- and 

 two-syllable words of like freciuency of occurrence. This was not ob- 

 served in reading the specialh' familiar lists of one- and two-syllable 

 words. 



Why is prose read faster than scrambled prose? It might be that some 

 short phrases are recognized as individual patterns. However, there is 

 another factor at work. A scrambled paragraph of prose is read slower 

 than the same paragraph in its natural word order but faster than 

 scrambled prose from a book or a long strefch of prose, as can be seen 

 from Table V. 



It should be noted that reading speed differs for different prose, and 

 that when comparisons among prose, scrambled paragraphs and scram- 

 bled prose are made, similar material should be used. 



The fact that a scrambled paragraph is read faster than scrambled 

 prose might be explained by saying that we expect, we are more readj^ 

 to recognize, words which are repetitions of earlier words or words which 

 are closely related in sense to earlier words than we are unrelated words. 

 Thus, the greater reading speed for prose than for scrambled prose seems 

 to be due only in part if at all to the recognition of phrases rather than 

 words as individual patterns. 



Rate of Mental Processes 



The rate at which information passes through a human channel in 

 reading experiments is indisputable. Quastler^ has attempted to go 



