308 BELL SYSTEM TECHNICAL JOURNAL 



sation shown in this table form 2.4 per cent of the different nouns and 

 40 per cent of the total nouns; the verbs form 5.5 per cent of the 

 different verbs and 72 per cent of the total verbs; while the adjectives 

 and adverbs form 3.9 per cent of the different adjectives and adverbs, 

 but 48 per cent of the total. 



An examination of the origin of the words in Table V shows that the 

 influence of Latin on the frequently used words is largely confined to 

 nouns. Eleven of the first 25 nouns of telephone conversation, and 

 eight of the first 25 nouns of written English come from the Latin. 

 Among the first 25 telephone nouns, aside from the eight nouns men- 

 tioned above among the first 100 words, there are: "office," "market" 

 and "case"; among the first 25 nouns of written English the following 

 are of Latin origin: "people," "country," "part," "fact," "cent," 

 "power," "present" and "peace." Only one of the first 25 telephone 

 verbs comes from Latin: "try," and three of those in written English: 

 "use," "govern" and "save." Among the adjectives and adverbs 

 there are found in the telephone list: "just," "very" and "fine," as 

 above, and in the written English list the word "just" is added to 

 "very," which was in the first 100 words. 



Referring once more to the small number of different words found 

 it may be pointed out that this shows how difficult it would be to 

 estimate the size of vocabularies by recording spoken words. The 

 80,000 words of this study are equivalent to a complete record of seven 

 hours' conversation, taking a rate of 200 words per minute. As noted 

 before, the number of different words was only 2,240, even though the 

 conversations covered a wide range of topics by many different 

 speakers. To increase this number notably, the curves of Figure 1 

 indicate that the observations would need to be very extended, since 

 the rate at which new words appear has already become very low. 

 For example, if the conversations were to go on continuously for a 

 week at the above rate a total of 2,000,000 words might be expected. 

 By extrapolating the curves of Fig. 1, and using a similar curve for 

 adjectives and adverbs, which lies between the curves shown, it may 

 be estimated that only about 5,000 of these words would be different 

 words. Extrapolation is a rough tool, but even with its inaccuracies in 

 mind, the conclusion seems safe that to measure a vocabulary by 

 recording spoken words involves the risk of gross underestimation 

 unless the observations are exceedingly prolonged. 



It is suggested that teachers of languages may find the 737 words in 

 Tables lll-a and IH-^ to be of practical use in their profession. Pre- 

 sumably the progress of a student in speaking a foreign language would 

 be materially assisted by a thorough knowledge, early in his course, 



