APPLICATION OF ARTICULATION TESTING 353 



or lower than normal, to maintain the desired average unless he is 

 continuously informed as to the amount by which he has failed to reach 

 his objective. 



In addition to providing a satisfactory means of control of the speech 

 intensity applied to the carbon transmitters in tests on a commercial 

 telephone system, the caller's control circuit can also be readily adapted 

 to the use of phonograph records of articulation lists instead of callers. 



Control of Carbon Button Transmitters 



Because the working of a carbon button transmitter depends to some 

 degree on the physical treatment it receives, steps have been taken to 

 subject the transmitters of the circuits tested to a cyclic routine 

 simulating that given them in an actual telephone conversation. The 

 insertion of the testing syllable in a carrier sentence is one step in this 

 direction since this subjects the transmitter to a conversational flow of 

 speech, rather than an abrupt monosyllabic impulse. The length of 

 the list, 66 syllables, is another, since this takes 3.9 minutes to call, 

 which is of the order of the duration of many telephone connections. 



A third step is prompted by the fact that in conversation the trans- 

 mitter at one end of the circuit is being agitated by speech while that 

 at the other end is being agitated by room noise, and as the conversa- 

 tion proceeds these agitations alternate: speech, room noise, speech, 

 room noise at one end; room noise, speech, room noise, speech at the 

 other. This is simulated in the present articulation testing routine by 

 having alternate test sentences called from opposite ends of the circuit, 

 while room noise is applied to both transmitters. 



The reversal of the direction of transmission is carried out auto- 

 matically every 3.4 seconds, which is about the average length of 

 utterance in a telephone conversation. 



In order to minimize the elifects of differences which are inherent in 

 any group of shop-product transmitters, a number of them, rather than 

 a single specimen, are used in each articulation test. These are 

 selected as typical from a group of 25 or 50, and are changed frequently 

 in testing, either between lists or between callers. 



When changing them it is desirable to agitate them in such a way as 

 to avoid using them under conditions of extreme variations in sensi- 

 tivity. In actual service this agitation is provided by the jar of the 

 switchhook when a deskstand is used, or by the movement of lifting if 

 the instrument is a handset, as well as by the introductory remarks 

 which usually start the conversation. To simulate this agitation and 

 to change the transmitters an automatic device has been developed. 



