THE sorxns or srnEcii 



TAHI.K \ 

 Speakers' Characleriitics, Semi-Vowel Sounds 



t>\7 



/ to m. Wv have here, then, a Kron]) "f doiihK' resonant sounds whoso 

 characteristic frequencies, whose amplitiuies, and general l)ehavior 

 arc such that the\- must be definitely related to the standard \o\vel 

 diagram. 



The amplitude frec|ueni\- rel.itioiis as ohlaiiied from a iiHchaniial 

 harmonic analysis, and corrected for the \ariation in sensitivity of the 

 ear are shown in Fig. 14. The process of mechanical harmonic anaKsis 

 has been outlined in connection witii liic \(iwel records, and the pro- 

 cedure was the siime here, except thai only the semi-vowel portion 

 of the records was taken as the unit for analysis. The record for 

 analysis was cut at the end of the last cycle before the transitif)n 

 point, anfl two profile copies of the semi-vowel wave were joined to- 

 gether in an endless belt which was passc^d through the analyzing 

 machine. 



Aside from the close resemblance between the frer|iienc\- spectra 

 of the four sounds the noteworth\- feature of F"ig. 14 is in the similarity 

 between the / spectrum and that for ee as previously given in line XIII 

 of Fig. 13. The essential differences arc a slight increase in the 

 importance of the low frequency characteristics, and the slight shift 

 of all the resonant regions toward low-er frequency, in passing from 

 e to /, and on through the sequence m?, «, nt. We may thus regard 

 the chart of Fig. 14 as a logical continuation of the generally accepted 

 chart of Fig. 13 and place the four semi-vowel sounds definitely in an 

 extended \owel diagram, following in regular order the sound long e. 



Sir Richard Paget has made the interesting statement that "all the 

 consonant soimds are as essentially musical as the vowels, i. e., they 

 depend on variations of resonance in the vocal cavity, and should be 

 capable of being imit.ited in the same wa\', if tiieir characteristic 



