MECHANICS OF THE HUMAN VOICE 247 



of men. Some approach to the truth has been reached by lexicog- 

 raphy in respect to the long vowels, but its results are largely erro- 

 neous in respect to the short ones. Both phoneticians and lexicog- 

 raphers are misled by spelling, and, by a well known psychological 

 prejudice, vowels that are physically quite different appear to the 

 ear as the same, and vice versa. For example, most phoneticians 

 distinguish two forms of the so called indefinite vowel. Sweet 

 (Axford) has lately recorded seventeen forms. In my curves I find 

 several hitherto unrecognized forms, physically quite distinct, for 

 nearly every vowel. 



5. To study the differences of speech among various dialects. 

 The phonetic survey of France (Guilleron) consisted in recording 

 by ear the pronunciation of certain typical words in each district of 

 France. The results were published as a series of maps, each one 

 containing in phonetic type the pronunciation of the word at each 

 place. The ear loses most of the differences in pronunciation ; the 

 type loses still more ; the results are vague. The proper method 

 would be to take records by the phonograph or the gramophone (as 

 recently done by the Vienna Academy of Sciences for certain Bra- 

 zilian dialects), and then to trace the curves by my method for 

 accurate study. I find that phonograph records could be made 

 with indestructible metal matrices and cylinders cast and delivered 

 at a smaller cost per word than the atlas of Guilleron. Instead of 

 maps with phonetic type, the student would have a collection of the 

 speaking records. Various disappearing and changing languages 

 (American Indian, for example) might thus be collected and pre- 

 served. The records might then at any time be traced off and 

 studied by the methods used in these researches. 



6. To determine the differences due to the speaker or singer him- 

 self, and to investigate the influence of training on these differences. 



7. To investigate the influence of the emotions on the voice. 

 These last two lines of investigation would lay the foundations 



for the psychology and physiology of vocal expression. It would 

 lead to data concerning the laws of expression in vocal music and in 

 oratory ; also possibly to the use of vocal records for determination 

 of mental conditions in health and disease. 



Of the work on these and other problems only the results for the 

 first three will be considered in this report. 



