Published online by Cambridge University Press: 24 December 2009
In my paper ‘On the Altaic numerals’ I made the point that Turkic, like Mongolian, segregated 6–9 by suffixes, and that Turkic further divided these into two morphological groups, 6–7 and 8–9. My good friend Robert Underhill has pointed out to me that I neglected an additional piece of evidence for these groupings, in the formation of the decades; that is, I overlooked the sets represented in Osmanli Turkish by altmiş, yetmiş ‘60, 70’ and seksen, doksan ‘80, 90’. The latter pair is clearly from *sekiz+on and *dokuz+on, and this shows up clearly in Old Turkish and Chuvash. I am grateful to Underhill for having pointed out this additional argument, and it leads me in turn to some further considerations which I take this opportunity to present.
1 Jakobson, R. and Kawamoto, S. (ed.), Studies in general and Oriental linguistics presented to Shirô Hattori, Tokyo, 1970, 188–97Google Scholar.
2 It should be noted in this connexion that I wrongly segmented the numerals ‘6’ and ‘7’; instead of a suffix +ti, we should attribute to them simply —i. I also now reject (see below) the suggested comparison of the initials of ‘6’ and ‘7’ with Tungus ‘3’ and ‘4’.
3 cf. the Danish decades, where ‘20’has been deleted.
4 Perhaps -mä ? cf. Chagatay yigirmä. Yakut sürbä presumably shows m > b after consonant.
7 See my remarks in Jakobson, and Kawamoto, (ed.), Studies…presented to Shirô Hattori, p. 194Google Scholar, § 7, where however, in terms of the present discussion, I now reject the suggestion of a direct Altaic relation for Turkic ‘7’ with a pre-form in *d-.