January 20, 192 1] 



NATURE 



665 



cess. Under the Carnegie Institution the work 

 was continued with an attempt to imitate more 

 closely the conditions in the living body. 

 Although the cheeks can be represented by water 

 resonators, the roof of the tongue is somewhat 

 more difficult ; the roof of the mouth is quite an 

 approach to a hard resonator. To imitate these 

 conditions, a resonator was made of a skull sup- 

 plied with cheeks and a tongue of gelatine. The 

 tongue was removable, so that models of different 

 forms could be inserted. The voice tone was ob- 

 tained from a vox humana organ-pipe. The ulti- 

 mate object was to find vowel resonators that 

 would respond with specific vowels to any tone. 

 These were then to be mounted oa a reed organ 

 as an extra register. All tones issuing from the 

 organ could be made to pass through one of the 

 vowel cavities, and the organ would thus sing the 

 vowels. In most singing the consonants are a 

 subordinate matter, and such an organ would aid 

 the singing of a choir or a congregation. The 

 beauty of such a vowel register in a large organ 

 in a cathedral is quite beyond imagination. This 

 investigation was also supported by the Hodgkins 

 fund, but was discontinued on account of the work 

 required for the study of speech curves. 



The Structure of Vowels. 



Tlie study of speech curves and the making 

 of vowel-producing instruments show that 

 two groups of elements are to be found in a vowel. 



The first is the voice tone. Three properties 

 of this tone are to be considered. The pitch of 

 the tone is given by the rapidity with which the 

 laryngeal puffs are repeated. If V indicates the 

 number of puffs per second, then \' = f{t) is the 

 general expression of the fact that the pitch of 

 the voice tone depends on the elapsed time. It 

 might be supposed that, in singing a vowel on 

 a given note, the pitch would remain constant. 

 .\ study of a record by the tenor Caruso (not 

 yet published) shows that he never keeps his voire 

 on a constant pitch during a vowel, but makes 

 continual small changes. .\ study of the vowel 

 "oh" used as an interjection shows that the pitch 

 of the voice tone changes to express the emotion 

 and the meaning of the interjection according as it 

 is spoken to express sadness, admiration, or doubt. 

 In fact, it is quite possible to obtain an equation 

 for V for each of the three cases. In ordinary 

 speech the pitch of the voice tone changes from 

 instant to instant. Fvery individual vowel has a 

 melody of its own. This melody varies with the 

 emotion, the meaning, and possibly other factors. 



The intensity of the voice tone depends on the 

 energy of the pufTs from the larynx. If I indi- 

 cates the puff energy, then I=/(0 expresses the 

 fact that the intensity varies from moment to 

 moment. The speech curves show changes of in- 

 tensity that express emotion and meaning. Even 

 in song — as Caruso's curves show — the intensity 

 is constantly varied in a way that makes his 

 song a production of art, and not a mechanical 

 performance. 



C^/S, VOL. 106] 



The third factor of this group is the musical 

 character of the voice tone. This depends on 

 the shape of the puff from the larynx. The 

 matter is of such importance that the following 

 statement seems to be needed. 



.\fter a tuning-fork has been struck its vibra- 

 tions slowly die away. Its curve is really not 

 that of a simple sinusoid, but that of a frictional 

 sinusoid 





(!) 



where y is the elongation at the moment t, a 

 the amplitude or maximum elongation, T the 

 period, e the basis of the natural system of 

 logarithms, and e the factor of friction or damp- 

 ing. The period T is affected by the factor of 

 friction, but the amount is so small that it can 

 bo neglected here. The effect is to cause a de- 

 crease in the amplitude. When the value of t 

 is great, the curve is that of a sharp puff. When 

 it is less, the puff is more gentle. If it were o, 

 the ordinary simple sinusoid would be obtained. 



The puff from the larynx njay be of a compli- 

 cated form that should be represented by the sum 

 of a series of frictional sinusoids. The complete 

 equation would be 



sin ?"/. 



(2) 



This comprises the whole of the vibration of a 

 single puff. It is a free, and not a forced, vibra- 

 tion. The musical or unmusical quality of 

 the voice depends solely on the presence or ab- 

 sence of the various members of (2). The quality 

 of the voice that distinguishes a Caruso from a 

 costermonger lies exclusively in the laryngeal 

 puff. This fact is of importance as contradicting 

 the almost universally accepted theory of vocal 

 training that is based on "tone-placing" by the 

 supposed action of the vocal cavities as resonators 

 to give the musical quality to the tone from the 

 larynx. 



The other group of elements comprises the 

 tones aroused in the vocal cavities. A puff strik- 

 ing a cavity arouses one or more vibrations of 

 the form of a frictional sinusoid as in (i). Each 

 cavity will have its own factor of friction and its 

 own j>eriod. As shown by the vowel siren, this 

 response will be a free vibration independent of 

 the voice tone and the periods that go to the 

 vibration that makes up the puff. Every change 

 in the sizes or openings of the cavities will alttr 

 the. periods of these vibrations. The possible corfi- 

 binations for the cavities of the chest, pharynx, 

 mouth, and nose provide for an almost endless 

 variety of vowels. 



The speech curves show quite unexpectedly that 

 there is no >uch thing as a constant vowel. The 

 vcwel "o" in "so" changes its specific vowel 

 character from beginning to end. The least 

 change is found in German ; more change appears 

 in .American. There is so much change in English 

 that an .American hears the vowel "o" as a sound 

 starting like "oh" and ending like "oo!" The 



