From: Miguel Carrasquer
Message: 15674
Date: 2002-09-22
>What's irregular about *sebun- is only the early loss of *t, which however is hardly more remarkable that the deletion of /t/ in English <soften> or <cap'n>. The accentuation of Skt. saptá and Gk. heptá explains the Vernerian voicing in the cardinal; the original Germanic ordinal was *sebunda- (Anglian siofunda, OHG sibunto, ON sjaunde), also with Verner's Law (applying twice) < *sepuntó- < *sep(t)n.-tó-. Needless to add, Germanic *b couldn't continue pre-Germanic *b anyway.The loss of *t is what I was referring to, and we may compare the loss of *t in
> 0).But Winter's law is blocked when a resonant (at least *n, *r, *l) follows.
>Slavic *sedmI/*sedmU is a minor mystery, but I don't think it's the same mystery as Gk. hebdomos, ogdo[*w]os (or, for that matter, Lat. quadru- and -ginti:). For one thing, there's no trace of the voicing anywhere in Baltic; for another, the Slavic word shows no Winterian effect of *d (**se^dmU).
>Finally, old *-dm- was normally reduced to *-m- in Slavic (*e:dmi > *e^mI 'I eat', *waidmi > *ve^mI 'I know'), so who knows if *-tm- > *-dm- is not a "regular" development (of course one example cannot establish a rule, but are there any counterexamples?).None that I can think of with /tm/, although /tn/ > /n/ (and /tl/ > /l/, but not