360 SECTIONAL TRANSACTIONS.—A. 
10, Sir Ricnarp Pacer, Bart.—The Nature and Artificial Production of 
Speech Sounds. 
The resonances heard in the lecturer’s own whispered vowel sounds were compared 
with those obtained by instrumental methods by Dr. D. C. Miller (‘ Science of Musical 
Sounds,’ 1916, Lecture VII, page 235 et seq.). 
The double resonances obtained by Miller for the vowel sounds mat, met, mate, 
meet, were shown to be comparable with the lecturer’s own resonances & (hat), e (men), 
ei (hay), and i (eat), allowing for variation of pitch due to differences of pronunciation. 
For the vowel sounds ma, maw, mow, moo, for which Miller finds only a single 
resonance, observations by ear showed double resonances, one resonance of which in 
each case was fairly comparable with Miller’s single resonance. 
It was demonstrated that the English vowel sounds a (calm), e (men), i (eat), 
9 (all), u (who), could be recognisably reproduced by means of an organ reed attached 
to a single cylindrical resonator of cardboard divided into two resonators of proper 
pitch by means of a central perforated diaphragm or stop, the actual resonances being 
a, 1084/683 ; e, 1722/406; i, 2298/342; 0, 861/483; u, 812/322. The cardboard 
resonator gave similar vowel sounds to those of plasticene metal or glass. 
The same organ reed was attached in succession to two single resonators tuned 
respectively to Miller’s whispered frequencies 781 (9) and 383 (u). 
It was shown that though there was a trace of the 9 sound in the case of a single 
resonator of 781, there was no trace of u sound in the case of a single resonator of 383. 
Reference was also made to the recent work of Crandall and Sacia (Bell System 
Telephone Journal, April 1924, vol. 3, No. 2, pp. 232-7). These results of vowel 
sound analysismade by a system of electrical filters showed clear double resonances 
for the vowel sounds a, » (not), ou (no), u, U (put), the vowel sound 0 (all) being 
shown as having a single extended region of resonance coinciding with that of the 
two separate resonances heard in the lecturer’s voice. 
Crandall and Sacia found certain additional resonances. It was suggested that 
some of these might be distinctive of the American accent. In the lecturer’s own voice, 
when producing the vowel sound & (hat) with the American accent, an additional 
resonance of the order of 2,700 was audible. This resonance appeared to be due to 
a constriction of the pharynx, as though the back resonator were divided into two 
resonating cavities. 
The experiment was shown of constricting a rubber tube model. It appeared 
that by constriction of the rear portion of the back resonator an appreciable ‘ nasal ’ 
quality was added to the vowel sound. 
Experiment was shown of slowing down the speed of a dictaphone record of the 
vowel sound a (as in up) which gave in succession sounds very similar to a (calm), 
» (not), 9 (all), and u (who) by successive reduction of the speed of the driving 
mechanism. 
A sound similar to U (put) was also produced by slowing down a record of the 
vowel sound (hat). 
The cumulative effect of these experiments appeared to indicate that the vowel 
sounds a (calm), » (not), 0 (all), u (who) and U (put) were all in effect due to double 
resonance. 
The artificial production of consonants by means of resonators was shown, con- 
sonants m and n being reproduced by a double resonator of frequency 2169/342 giving 
the vowel sound 1 (it) and having a third (nasal) resonator of frequency 704 connected 
to the back resonator. With this model, closure and release of the mouth during 
blowing produced mi; closure and release of the central orifice gave ni. The two 
operations in succession produced the name ‘ Minnie.’ 
The production of the consonants p, t, k, v, z, dh, r (untrilled) and r (guttural) was 
shown by means of a flexible double resonator made of a l-in. diameter rubber tube 
actuated by an air-blown reed. It was shown that the different consonants were pro- 
duced by the varying resonance effects of complete or partial closure and release at 
different points in the length of the tube. The untrilled r sound was shown to be due 
to a double restriction. 
The consonant sound ‘1’ was proved to be due to a rapid change of resonance, the 
‘1’ sound being lost if the same changes were made at a slow speed. 
The general conclusion was that the consonant sounds are as essentially musical 
as the vowels, and are produced in a similar manner, viz.—by resonance. The 
principal differences were that, whereas the vowel sounds represent the effect of 
