THE POWER OF EUNDAMENTAE SPEECH SOUNDS 403 



prevails here, there are considerable differences throughout which may 

 well be expected since the ear is used in making the balance. The 

 frequency response characteristic of the ear is the complicating factor 

 in this case. The ratio of maximum to minimum here is of the order of 

 one thousand or about one-half the absolute power ratio found in the 

 two preceding tables. 



A more orderly comparison between power and "relative attenu- 

 ation" exists in the case of the vowels alone as shown in the chart of 

 Fig. 2. Thus the peak power and "relative attenuation" most nearly 

 correspond at the ends of the chart (especially the left) where there is 

 resonance of lower frequency in the vowels. The vowel "o" again 

 shows a peculiarity in that the two trends — as shown by the envelopes 

 — intersect here. Peak power predominates over "relative attenua- 

 tion" in the three successive vowels "a," "a," "e," which have strong 

 resonance in the region from 600 to 1200 cycles. The vowel "i" gives 

 the only erratic turn in this comparison, differing considerably from 

 the two adjacent vowels. 



As for loudness in the ordinary sense, let us note a phenomenon of 

 rather common occurrence in these days of good quality sound repro- 

 ducing apparatus. One may be listening to well reproduced speech at 

 ordinary volume when suddenly a slightly accented syllable containing 

 "a" (tap) comes through with noticeable overload distortion and its 

 accompanying disagreeable effect upon the ear. Although the listener 

 does not judge this sound to be any louder than numerous accented 

 sounds preceding and following it, still the fact remains that there has 

 been considerable overload due to the peaks of the wave being cut off 

 by the amplifier. Where do we look for the explanation? As noted 

 in the earlier paper this vowel has the highest peak factor, and we have 

 already seen in Table III that it normally contains the greatest peak 

 power. In spite of this therefore, it would seem that the loudness of 

 this sound does not predominate over the loudness of the sounds in the 

 first half of the chart, as does the peak power. This phenomenon can 

 also be demonstrated, for the vowel "e" (teem) and to a lesser degree 

 even for the vowels which intervene between these two in the tables 

 and chart of the vowel sounds. 



