I saw you moved this from keldaz. Torp says "Stamm keldiz, keldaz" [i.e. z stem], but that is obviously not a very up-to-date source, so I was a little iffy about it (especially with descendants only in the English branch), but I don't know about this either. What about the umlaut (e>i)? Also, is the change -lþ- > -ld- regular? Then I wonder whether this might perhaps come from *kindą instead (as it seems to have replaced that word), but with influence from *kilþį̄, *kulþaz. – Krun (talk) 18:48, 22 September 2012 (UTC)
I'm not actually quite sure. I found the reference to it being an a-stem on etymonline.com, but we'd need an Old English editor to confirm it. The change lþ > ld is regular for West Germanic, there are other words like gold where it happened too. I'm not actually sure about the stem vowel, but if kull is a cognate, the stem vowel must have been -e- unless it was umlauted to -i-. It could only have been -i- if the stem originally had -y- in it, i.e. e-grade -ey- (> -ī-), zero grade -i-. But kull must be a zero-grade kulþ-, so the e-grade must be kelþ-. lþ is also the only possible etymology for West Germanic ld paired with North Germanic ll, because a Germanic ld would have become ld in North Germanic also. And then there is Gothic kilþei (kilþei).
I did a quick search of our Old English noun category and I seem to have found at least one parallel for the e>i change in an a-stem: knehtaz > cniht, although this also has variants (cneht, cneoht); this seems to show the possibility, at least. I would still be very curious to see any theories as to the absence of *cind (< *kindą) in Old English and its connection to cild. – Krun (talk) 00:03, 23 September 2012 (UTC)
I think many people have wondered that, it is certainly curious. It's quite possible that the two words were mixed up in the early languages, and that cild took its z-stem inflection from *cind. But that doesn't explain the i, really. For eht > iht it's a regular change, but there is no parallel for elþ > ild. The closest we can get is gieldan from Template:termx, but that's -ie-, not -i-.