QUALITATIVE ANALYSIS. 47 



whisper: This would probably be the action corresponding to Seelmann's 

 view of the nature of the Greek spiritus asper and lenis as being the 

 strong and weak breathy beginnings of vowels. 



The speech curves suggest another view. This is that the glottal hps 

 continue to vibrate during the intervocalic [h]with no disturbance, the glot- 

 tis remaining closed as in the adjacent vowels, and that the [h] is produced 

 by narrowing the air passage either by bringing the ventricular bands 

 together or b}^ partially closing the epiglottis down over the larjaax. One 

 can sing a breathy [a], [e], etc., indefinitel}^ long with some closure that 

 is behind and below the tongue. According to this view the sonant [h] 

 would be a sonant fricative of the same class as [j] in North German 

 "Jager" or [j] in North German "Sage," etc., with the passage narrowed 

 in the mouth. It thus differs radically from the ordinary surd [h]. This 

 view is in conflict with a fundamental psychological principle that I 

 have felt obliged to assume as the basis of all deductions concerning 

 sound change, namely, that where a person varies from one form to 

 another — for example, surd to sonant [h] — in the same coml^inations of 

 sounds used on different occasions without perceiving any difference, the 

 two forms must be merely variations of the same articulation. In records 

 from a person speaking continuously we find surd and sonant [h] used 

 indiscriminately between vowels; exactly the same words are spoken on 

 one occasion with a sonant [h] and on the next with a surd one. That 

 in one the breathy sound should be produced by the ventricular bands 

 and in the other at the glottis seems — from the psychological principle 

 of performing the same act in the same way with a variation around 

 a mean — to be improbable. The final decision must be left till definite 

 experimental knowledge can be obtained. It is hardly profitable to go 

 into discussion of the possible relations of the sonant [li] to the Arabic 

 "ain." The physiology of neither sound has yet been estabhshed; what 

 speculation can do is shown by Sweet's amusing supposition of a vibration 

 occurring below the glottis (that is, in a wide open tube with walls of 

 cartilage) for the Arabic " ain." 



The [i] in line 9 is short and weak, being almost lost before the [m], 

 which is much longer. The vibrations of [m], like those of all sounds 

 where the mouth is closed, are of much smaller amplitude than those of 

 the vowels. The vibrations for the [d] are still weaker than those for 

 the [m]; there is no break between the two, the sounds being run together. 

 The explosion of the [d], line 10, and the large irregular vibrations as the 

 [d] passes into the following vowel extend over an unusually long space. 

 The curve for [a] — which may be made to include the irregular vibrations 



