The Nature of Speech and Its Interpretation' 



By HARVEY FLETCHER 



Introduction 



VARIOUS phases of this subject have received serious study by 

 phoneticians, otologists, and physicists. On account of its universal 

 interest, it has received attention from men in many branches of 

 science. In spite of the large amount of time devoted to the sub- 

 ject, the progress in understanding its fundamental aspects has been 

 rather slow. At the present time the physical properties which 

 differentiate the various fundamental speech sounds are understood 

 in only a very fragmentary way. Some very interesting and pains- 

 taking work has been done on the physical analysis of vowel sounds, 

 but the results to date are far from conclusive. Although several 

 theories have been advanced to explain the way in which the ear 

 interprets sound waves, they are still in the controversial stage. 



The material which is presented here is the result of an investi- 

 gation which has been carried on in the Research Laboratories of 

 the American Telephone and Telegraph Company and Western 

 Electric Company during the past few years. 



To make a quantitative study of speech and hearing it is necessary 

 to obtain the speech sounds at varying degrees of loudness and with 

 definitely known amounts of distortion. The main reason why so 

 few real results have been obtained in the investigation of speech 

 sounds is due to the fact that it is extremely difficult to change the 

 volume and distortion of these sounds by acoustic means. Due to 

 recent developments in the electrical transmission of speech it is 

 possible to produce the equivalent of these changes by electrical means. 

 For this purpose a telephone system was constructed which repro- 

 duced speech with practically no distortion. It was arranged so that 

 by means of distortionless attenuators the volume of reproduced 

 speech could be varied through a very wide range, and so that by 

 introducing various kinds of electrical apparatus the transmitted 

 speech wave could be distorted in definitely known ways. 



A method was developed for measuring quantitatively the ability 

 of the ear to interpret the transmitted speech sounds under different 

 conditions of distortion and loudness. By choosing these conditions 

 properly, considerable information was gained concerning both speech 

 and hearing. This indirect method of attack has a distinct advantage 



* Presented at a meeting of the Electrical Section of the Franklin Institute held 

 Thursday, March 30, 1922. Reprinted from the Journal of the Franklin Institute 

 for June 1922. 



129 



