On 2006-09-19 23:41, tgpedersen wrote:
> I just don't feel a rule of the type -Tn- > -TT- makes sense
> phonetically. -nT- > -TT- does.
It was *-bn-, *-dn-, *-gn- > *-bb-, *-dd-, *-gg- (devoiced later). Just
why doesn't it make sense phonetically? Cf. Pali nagga, soppa < OInd.
nagna-, svapna-. The fact that the Germanic assimilation operated before
an accented vowel suggests that it had to do with changing
syllabification in early Proto-Germanic. In a sequence like *-u.gná- the
stop was attracted into the preceding syllable (provided that it was
originally light, i.e. contained a short vowel), perhaps becoming
ambisyllabic/geminated as a result: *-ug.gná-; then the cluster was
simplified, producing *-ug.gá-.
> I'd prefer something like *Nd -> n.t (the stop is unvoiced, so
> the preceding nasal is unvoiced) -> nt/ht/tt (for Germanic
> substrate words). For that purpose I'm ready to steal the
> examples Kortlandt uses in his defense of preglottalisation
> of voiced unaspirated.
But Kluge's Law affects all original phonation types, i.e. *t, *d and
*dH alike, and remember that after a heavy nucleus the sequence -tn-
(with this order of segments!) survives.
Piotr