The Sounds of Speech 



By IRVING B. CRANDALL 



XOTE: As professor of \iical plusiology, Alexander (jraliaiii Bell did 

 pioneer research in "devising methods of exhibiting the vibrations of sounds 

 optically." In 1873, he became familiar with the phonautograph, de- 

 veloped by Scott and Koenig in 1859, and with the manometric capsule, 

 developed by Koenig in 1862. (ireatly impressed by the success of these 

 instruments "to reproduce to the eye those details of sound vibration that 

 produce in our ears the sensation we term timbre, or quality of sound" 

 Bell used an improved form of the phonautograph haxing a stylus of wood 

 al)out a foot long. He obtained "large and very beautiful tracings of the 

 vibrations of the air of vowel sounds " upon a smoked glass. 



In describing his early attempts to improve the methods and apparatus 

 for making speech waves visible and to interpret wave form. Bell wrote: 



"I then s;ing the same vowels, in the same way, into the mouth-i)iece 

 of the manometric capsule, and compared the tracings of the phonauto- 

 graph with the flame-undulations visible in the mirror. The shapes of 

 the vibrations obtained in the two ways were not exactly identical, and I 

 came to the conclusion that the phonautograph would require considerable 

 modification to be adapted to my purpose. The membrane was loaded by 

 being attached to a long lever, and the bristle, too, at the end of the lever, 

 seeme<l to have a definite rate of vibration of its own. These facts led 

 me to imagine that the true form of vibration characteristic of the sounds 

 of speech had been distorted in the phonautograph by the instrumentalities 

 employed. I therefore made many experiments to improve the con.struc- 

 tion of the instrument. I constructed, at home, quite a number of different 

 forms of phonautographs, using membranes of dilTerent diameters and 

 thicknesses, and of different niaterials, and changing the shape of the 

 attached lever and bristle." 



Struck by the likeness of the phonautograph and the mechanism of the 

 human ear, Bell conceived the idea of making an instrument modeled after 

 the pattern of the ear, thinking it would probably produce more accurate 

 tracings of speech vibrations. In 1874, he consulted a distinguished 

 aurist. Dr. Clarence Blake of Bostim, who suggested that instead of trving 

 to make an instrument modeled after the human ear, the human ear itself 

 be used. Dr. Blake prepared a specimen containing the membrane of 

 tympanum with two bones attached, the malleus and incus. The other 

 bone, the stapes, was removed and a stylus of wheat straw about one inch 

 long was substituted. A sort of speaking tube was arranged to take the 

 place of the outer ear. "When a person sang or spoke to this ear, I was 

 delighted to observe the vibrations of all the parts and the style of hay 

 vibrated with such amplitude as to enable me to obtain tracings of the 

 vibrations on smoked glass." 



In the accompanying paper. Dr. I. B. Crandall describes modern methods 

 whereby with the most refined apparatus, highly accurate speech wave forms 

 have been produced. The analysis and interpretation of both vowel and 

 consonant sounds made possible by these records, are the realization of an 

 objective sought by Bell a half century ago. 



This article is the result of an extended study of 160 gra|)hical records 

 of vowel and consonant sounds, of which a few are reproduced in the present 

 publication. One htmdred and four of these records are of vowel sounds 

 an<l formed the basis of the " Dynamical Study of the Vowel Sounds," by 

 I. U. Crandall and C. K. Sacia which was published in this Journal in .April, 

 1924. The purpose of the present article is to describe all of the records 

 in sufficient detail, including in'one discussion the outstanding character- 

 istics of vowel, semi-vowel and consonant sounds; it is hoped shortly to 

 Hupplement lhi.s with a reproduction of a larger group of records from the 

 coni()lete colled ion. — Editor. 



586 



