THE SOUNDS OF SPEECH h2i 



souiul, tluTC is, in addition t<i the pri-sinirc of the fundamental tone, 

 a breaking: up of the eharacterislic hiy;h fref|uen(y wave-train intf) 

 discrete units correspond in^j to the fundament.il tone, where.is in the 

 un\-oiced sound the hi^h fre(|uency characteristic is continuous, though 

 irrei;ular. Thus noting tiiat the characteristic frefiuency is of 3000 to 

 HUM) cycles the outstandinij phenomena of zh sh are well defined. In 

 adilition to frequencies of 2048-324!) noted by Paj;et (91), p. KY.l) he 

 gives a "pronoimced middle resonance of U)2')-2048." This latter 

 observation of Paget's may correspond to the 1800-2000 fref|uency in 

 the records of MB (Plates 154, 15G) in the transition region, but this 

 component does not seem to be prominent in the records. 



Z/S. — (Plates 157-160). The general properties of these sounds 

 can be inferred from the discussion of the preceding pair (zli/sh), add- 

 ing only the fact that their principal characteristic is of much higher 

 fretjuency. From Table V'll we note a range of 4200-8000 cycles; 

 P.iget (9b, p. 162) gives "a characteristic upper resonance of 5790- 

 6886." Paget also gives "a middle resonance of 1084-2298." The 

 records do not show as low a range of characteristic frequencies unless it 

 be the frequency range 2200-2800 (see Note 1, Table VII), within 

 which fall certain \ibrations occurring in the early parts of the funda- 

 tnental cycles of the voiced sounds sh and s. The true s sound is, 

 as Paget has stated, "a relati\'ely complex hiss" and this is true of sh 

 as well. And to complete the record, we must observe that sh and z 

 are even more complex, if possible, and thus not inappropriate ex- 

 amples of the sounds of speech with which to conclude this sur\'ey. 



To summarize, we have considered some of the more outstanding 

 features of the w-ave forms of speech sounds which have been re- 

 corded. Many more detailed properties of these records deserve 

 further study. The progressive change in wave form from cycle to 

 cycle of the fundamental, particularly at the beginning of a sound, 

 is undoubtedly an important factor in determining the character of 

 speech sounds; it becomes most important, as we have seen, in the 

 study of the more impulsive consonant sounds. There is material in 

 these records for extended studies of this kind, which require a har- 

 monic analyzer of a large number of components. \Vc have not dealt 

 with the question of the inherent power in speech sounds, another very 

 characteristic property; these important data are accurately given 

 in a paper by C. F. Sacia in this issue of the Journal. The relative 

 power in consonant and vowel sounds can also be determined from 

 those records in which vowels and consonants appear in combination, 

 and it is hoped to carry this study further. Many other investigations 



