--- In
cybalist@yahoogroups.com, Miguel Carrasquer <mcv@...> wrote:
> On p. 112 explicitly to exemplify *-n.d, an analysis that
> demands initial stress (*wé:dn.d).
Yes, but is that more than just a point a view? Also /-énd/ (with or
without a lost vowel after it) gives Hitt. /-and/. The form is not
*attested* just because Melchert imagines it. The line of reasoning
seems to be this: Word-final /-nt/ is reduced to /-n/, as in the
neuter of the participle, but final *-d is retained after vowel, as
kui-t, apa:-t, and, he says, in general "after a syllabic", quoting
the two instrumentals kissarta/-at and wedanda as representing *-r.-
d without further comments. To be of use in our context this ought
to imply that the instrumental ending /-d/ would be lost when it
follows a consonantal nasal. But there are no instrumental forms
without /-d/, so the dental can just as well have been restored
where it is found. That is what Melchert assumes (along with most
everybody) for other cases of retained -Ct. Sure, if there have been
no restoration, no ablaut levelling and no accent shift in wedanda,
it represents *-n.-d leaving only the root with full grade and thus
implicitly accented. However, in a general sense, the language only
preserves these details undisturbed in relatively exceptional cases.
That disqualifies Melchert's quotation "we:danda" as evidence for
initial accent on weak-case weden-/wedan- vis-à-vis loc.-dat.
wede:ni. The diachronic implications using the presumed attestation
as a support must now stand without that support. I do not think
that is impossible, it is just not demanded by this evidence.
Jens