Miguel:
> Does *kW give /p/ or /k/?
Both. To be clear, ITyr *kW gives Tyrrhenian *x, *k and *p depending
on the situation.
The overwhelming pattern is that *kW gives Tyrrhenian *x if it was
uvularized or *k if plain. Thus IT *kWatWan 'four' > Tyr *xota and
*kWere 'create' > Tyr *kere. (ITyr *a becomes Tyr *o neighbouring all
labial phonemes, btw, so *kW rounds the following *a to *o in Tyr.
This also happens in *nefota 'grandson' < *nepata) In EtruscoLemnian
this regularly yields *xuta and *kere respectively which predictably\
become /hutH/ and /cara/ in Etruscan since *e has a tendency to lower
to /a/ while *a disappears altogether. Wonderfully regular.
However, to account for /ipa/, one needs to either give up on its
etymology or to conclude that since so many other demonstratives are
clearly related to IE, that it too is related to its respective IE
counterpart *kWo-. In MIE, the interrogative is *kWa but OIE merged
final *-e and *-a to *-a as I said before, remember? The previous
contrast in final or unstressed vowels in open syllables was lost in
the early MIE period. MIE *kWa derives from OIE *kWe, identical to
its previous IndoTyrrhenian form (< ProtoSteppe *kWi, becoming
Boreal *ki as attested in Uralic).
With ITyr *kWe, we expect Tyrrhenian **i-ke under the normal rules but
this is homophonous with the demonstrative! We already can figure
out that Etruscan ica 'this' < Tyrrhenian *ike, with a later optional
prothetic deictic *i- for all demonstratives (cf. IE *i-). To add to
the complication, there is also the postclitic *kWe 'and' which must
surely become Tyrrhenian *-ke because we can clearly observe Etruscan
/-ca/, showing yet another example of expected Tyrrhenian *k from
plain ITyr *kW.
So what gives? I think what happened is that *kW became *p under
certain much rarer conditions but the question is what conditions?
Somehow *kWe would become *i-pe which would then yield our Etruscan /ipa/
nicely. I think I can formulate a rule but it has to be further tested.
The rule is *kW becomes *p before *e, in _medial_ position. This
naturally doesn't affect *kWere, regularly becoming *kere. It also
doesn't affect *kWe 'and' becoming *ke right on schedule. So an
earlier *ikWe then would _not_ become **ike, but rather *ipe as we
find.
The result of this rule is intriguing because it would imply that
verb stems of the form *CeQ- where Q is any labiovelar, for example,
should become *Cep- instead of the otherwise expected **Cek-. As I
said, the *kW > *p rule has to be further tested with other examples
so you can hate me if you like.
= gLeN