Kenneth L. Pike (19122000)
1938: In the January cold, four a.m., before the corn stalks in the fields could be calling for cutting by the language helper, we were at it. “What is a mountain?’
(Oh) (a) mountain is (just a) mountain! — yuku kú yuku.
And what is shrubbery? — Shrubbery is shrubbery — yukù kú yukù.
And what is (a) yoke?— Yoke is yoke — yukú kú yukú.
What a sameness! But what differences! The kú meaning 'is' was always as high or higher than anything else; the two syllables of yuku 'mountain' were both lower, with both syllables on that lower pitch; but for yukù 'shrub' the second syllable was lower still; and for yukú 'yoke' the second syllable was on the same level with the frame. Thus three stable, contrasting differences were in view; one pitch just as high as the unchanging frame tone (i.e., a high tone, which I have marked with an acute accent); one a bit lower; i.e., a mid tone (on which I have placed no mark here); and a third, lower still, i.e., a low tone (on which I have placed a grave accent mark). The first crucial component of the tone system had broken open — to be supported by similar data with hundreds of words in many different frames!
Kenneth L. Pike. "An autobiographical note on my experience with tone languages" p.23. In J. C. Sharma (ed.), From sound to discourse: A tagmemic approach to Indian languages, 21-31. Mysore: Central Institute of Indian Languages. 1992.