620 BEI.I. SYSTEM TECUXfCII. JOrRX.II. 



Examination first discloses a relatively easy separation of a given 

 record into a consonant and a vowel portion and, as might be expected, 

 a longer duration for the "voiced" consonants. In all the voiced 

 consonants a sufficient portion of the record is reproduced to illustrate 

 the voicing or fundamental of small amplitude in the early stages of 

 the record ; in the case of the unvoiced consonants of ( jroup X\'I this is 

 not necessar\\ In the case of both the voiced and unvoiced con- 

 sonants of Group X\'II, longer records are shown, the high frequency 

 component making this necessary, although the fundamental does not 

 appear in the early stages of the unvoiced consonants of this group. 

 The mean duration of the voiced consonants (b, d, g, dth) of Group 



XVI is 0.14 second ; of the unvoiced consonants {p, I, k, th) 0.05 second. 

 Aside from traces of the fundamental tone (and traces of its second 

 and third harmonics) there is nothing of interest in the early stages 

 of three of these four voiced consonants; in the case of dth there are 

 traces of a high frequency (4200 and 2(100 in the two records) in the 

 early parts of the fundamental cycle. The voicing for all four sounds 

 if uniformly of lower pitch than that used later in the records in speak- 

 ing the vowel sound. Leaving the early stages, the record then pro- 

 cet'tis to a transition point, lasting through from one to four cycles of 

 the fundamental, and culminating in the appearance of the vowel 

 sound. Before this transition point is reached, traces of high fre- 

 quency appear in most cases, sometimes suggesting a single transient 

 vil)ration. Aside from the lack of the fundamental vibration, there is a 

 further distinguishing characteristic of llu' "umoiced" sounds: a 

 tendency of the first transition cycle of the fundamental to appear 

 from 10 to 20 per cent shorter in duration than the mean of several 

 following cycles. With both \oiced and un\oiced sounds there is a 

 tendency for a moderately low frequency (500 to 700 cycles) to appear 

 during the transition; also a high frequency (of mean value 3225 cycles 

 for the IG records of this group) which latter may be due to the begin- 

 ning of the a sound. Some of the individual characteristics of the.se 

 records are given in Table \'I. 



The notable distinction between lliese sounds and tiie sounds of the 

 next Group (XVII) rests on duration factors, and of even more im- 

 portance, the pronounced high-frequency characteristics of the sounds 

 (jf the last group. The mean duration of the voiced sounds in Group 



XVII is 0.21 second; that of the unvoiced sounds, 0.18 second. Two 

 of the other characteristics are similar to those noted in the preceding 

 group; first the voicing, where it occurs, is of abnormalh- low frequency, 

 and second in the case of the unvoiced sounds, there is a marked short- 

 ness of the first fundamental cycle at the transition point. E.xcept 



