60 



THE STUDY OF SPEECH CURVES. 



i 



F^ 



zz 



5 



S50 



450 



350 



In music the melody is supposed to proceed by the steps, the tone being 

 constant in pitch for the time of each note ; in figure 46 the melody indi- 

 cated by notation is given by the plot below it. What is the nature of the 



melody in speech? Is the melody 

 : of actual song like the melody indi- 

 ! cated by the notation? 



For a study of the melody of a 

 speech record the length of each 

 wave-group is measured in tenths of 

 a millimeter. When the waves are 

 long — for example, in records where 

 1mm. is less than 0.0010s. — it is suf- 

 ficient to use a millimeter scale and 

 estimate the tenths of a millimeter 

 by the eye. It is, in fact, preferable 

 to using a scale in tenths of a milli- 

 meter and reading by a magnifying glass, because after a little practice the 

 estimation of tenths is just as accurate as the reading of them, while the time 

 and fatigue involved are far less. When the waves are short — for example, 

 when 1mm. = more than 0.0010s. — it is desiraljle or necessary- to use a scale 

 finer than millimeters. In the results reported in this chapter the metal 

 scale graduated directly in tenths of a millimeter and read by a magnifying 

 glass has generally been used. Owing not only to the greatly increased 

 accuracy but also to the saving in fatigue, the writer now often uses the 

 coordinate measurer described alcove. The apparatus is placed over the 

 plate, the ocular scale is ])rought accurately over the waves by the two 

 micrometer screws, and the wave-lengths are read off and recorded. The 

 microscope is then moved axiaUy till a new set of waves is seen. 

 A record-book with the following columns is prepared : 



Fig. 46. — Note.s and their melody plot. 



The scale is placed parallel to the axis of the curve, but below or above 

 the highest points. Readings for the highest or lowest points of successive 

 waves are taken. The results are written in the second column of the record 

 book. The differences between successive results give the wave-lengths; 

 these are written in the third column. For the straight-hne portions with 

 no waves the lengths are inserted. Since the waves are smoothly rounded, 

 there will often be doubt as to where to locate the apex, but with running 



