NATURE OF LANGUAGE JONES 



493 



as obtained from a large number of observations with six different 

 speakers. 



Speech energy extends from Si frequency of 60 cycles to above 

 6,000, with a maximum at about 200 cycles. The vowel sounds 

 carry most of the energy of speech and their important frequencies 

 lie below 3,000 cycles. The consonants are the characteristic quips 

 and quirks with which the syllables begin and end. They are weak 

 in energy, but very important to good intelligibility. In general, 

 they are rather high in frequency, some of them involving vibra- 

 tions going up to a frequency of 6,000 cycles or even higher. The 

 speech energy output of the normal voice has been found to be at 

 the rate of about 125 ergs per second. 



Fig. 3. — "Acoustic Spectrum " of English 



In other terms, simple computations show that if we could have 

 a million persons talking steadily and convert the energy of the 

 voice vibrations into heat, they would have to talk for an hour and 

 a half to produce enough heat to make a cup of tea. This merely 

 serves to illustrate that in terms of power or energy human speech 

 is exceedingly weak. Furthermore, most of this energy is carried 

 by the vowel sounds. On the other hand, the consonants, as will 

 be shown, are more important to perception and interpretation by 

 the ear, so that energy per se is not so much the primary require- 

 ment of speech reproduction, but rather its distribution, and par- 

 ticularly its distribution among the higher frequencies. 



The human hearing mechanism is usually considered to have 

 three parts. The outer ear includes the lobe, the ear canal, and 

 thfi drum. The middle ear is a small hollow space containing the 

 chain of small bones (malleus, incus, and stapes) which comprise 



