ARTICULATION TESTING METHODS 



843 



to a single index, or for a more comprehensive picture, to five indices 

 corresponding to the five groups of speech sounds. In order to 

 compare the indices obtained by a given crew with those of a reference 

 crew, it is necessary to correct the data in accordance with Eq. 2 

 for the effects of practice. To do this, as previously discussed, 

 articulation tests are made upon one or more of the reference circuits 

 by the crew in question. If /' is the syllable index so obtained, the 

 practice factor for the crew is given by the relation 



log (1 - I'^ 

 log (1 - /) 



(13) 



The practice factors for the other indices may be obtained also, 

 by substituting the appropriate indices for the syllable index in 

 Eq. 13. In Table XV the reference values for the various indices 

 are given for the reference circuits that were previously described. 



TABLE XV 



Reference Values 



Circuit 



Master Reference System 



Auxiliary Circuit of Master Ref. Sys. 



Carbon Transmitter Circuit 



Master Ref. Sys. plus 3,750" L.P.F. . 



" 750- H. P.P. . 



" " 1,500- L.P.F. . 

 " " 1,500- H. P.P. . 



Vi 



98.5 

 95.0 

 97.0 

 99.0 

 96.0 

 93.5 

 85.0 



98.9 

 95.0 

 97.0 

 99.5 

 92.5 

 86.5 

 82.5 



Cn 



99.6 

 96.5 

 96.5 

 99.5 

 99.0 

 90.5 

 96.0 



Cs 



99.2 

 88.5 

 93.5 

 99.0 

 99.5 

 76.0 

 97.5 



Q 



66.5 

 82.0 

 86.5 

 98.5 

 52.5 

 97.0 



99.3 

 90.0 

 94.0 

 97.6 

 97.1 

 80.2 

 92.0 



If the values for the sound index be compared with the sound 

 articulation values based on uniform weighting, that were given under 

 the section on practice effects, it will be seen that for these circuits 

 there is very little difference, between the two sets of values. In 

 other words, the average sound articulation is very nearly equal to 

 the average that is obtained when the individual sound articulations 

 are weighted according to the frequencies of occurrence of the sounds 

 in English. 



Similar comparisons have been made for a large number of other 

 transmission systems. They showed similar small differences between 

 the weighted and unweighted averages. For this reason we consider 

 it unnecessary to use the weighted average when great accuracy is 

 not required, for example, in a great deal of our routine work where 

 comparisons are being made between circuits which have similar 

 characteristics. This means that when testing an unknown circuit, 



