290 



BELL SYSTEM TECHNICAL JOURNAL 



Fig. 3 it may be seen that these, in conjunction with terse replies or 

 questions under one second in duration, constitute about a third of 

 the talkspurts. There were, however, a few very long talkspurts: 27 

 exceeded 30 seconds, and of these 2 were over 120 seconds long. Dur- 

 ing the longest talkspurt, which was 143.82 seconds, there were 62 

 resumptions following silent intervals ranging from 0.34 to 4.04 

 seconds. 



3 4 5 



NUMBER OF RESUMPTIONS 



Fig. 6 — Percentage of talkspurts containing a number of pauses equal to or less than 



a given number. 



5 6 7 8 9 10 II 12 13 

 LENGTH OF TAliK-SPURT IN SECONDS 



Fig. 7 — Lengths of talkspurts containing a given number of pauses. 



Figs. 6 and 7 show the results of analyses made to determine how 

 frequently pauses occur within talkspurts and how the number of 

 pauses varies with length of talkspurt. In Fig. 6 the percentage of 

 talkspurts having a number of resumptions equal to or less than a 

 given value is shown. It will be seen that about 60 per cent of the 

 talkspurts contain no pauses; these comprise all the monosyllabic 

 replies and about half the longer ones. A further analysis, shown in 



