ARTICULATION TESTING METHODS 819 



the voiced sounds, woman's speech has only one half as many compo- 

 nents as man's, which also may cause greater difficulty in interpreting 

 the former. 



With respect to the vowel sounds, the crew members are instructed 

 in the correct manner of enunciation. Only those vowels which have 

 definite differences have been included in the testing lists, so that, 

 slight differences in enunciation do not seriously affect the observed 

 results. 



The object of the selection process is to determine in a broad but 

 definite way the normalcy in speech of prospective members, and to 

 eliminate those individuals who have speech characteristics which 

 are not readily reproducible should it be necessary to change the 

 testing personnel. The row averages of Table III show a typical 

 set of results for 8 callers who were selected in the above way, and 

 have had a year or more of experience in calling. 



The spread in results is of an order of 20 per cent, so that, if a crew 

 of 8 to 10 callers is used, a replacement may be made in the calling 

 personnel without causing a change in the average percentage of 

 more than 2 per cent. Owing to inherent differences in the voices 

 of men and women, they are equally represented on the testing crew. 

 Individuals who have the equivalent of a high school education, 

 and whose ages range from 18 to 23 years, are usually selected for 

 this work. 



Control of Haphazard Factors 



Haphazard factors arise from various sources, some of which can 

 be controlled reasonably well. The observers work in a sound- 

 proof room, so that extraneous noises will not affect the articulation 

 results. The calling is ordinarily done in a sound-proof booth that 

 has been especially treated with sound absorbing material so as to 

 reduce the reverberation time to an order of a few tenths of a second. 

 Ordinarily the crew does not test more than two to four hours during 

 the day, and the schedule is usually arranged so that this is not 

 done continuously. 



The intensity level of each caller is also measured during the test, 



as small variations in intensity level may cause rather large variations 



in articulation. Ordinarily the various callers are permitted to call 



at the intensity level most natural to them, although in some tests 



the callers all attempt, by watching an indicator, to call at the same 



level. Various instruments have been used for measuring the intensity 



levels during tests. The volume indicator "* has proven quite satis- 



^ This instrument depends for its readings, essentially upon the syllabic powers 

 of the vowel and semi-vowel sounds, so that the reading of the instrument is deter- 

 mined largely by the amplitudes in the frequency range from 100 to 2,000 cycles. 



