Re: [tied] *g'(h)- > d as aberrant outcome

Piotr Gasiorowski wrote:

> I don't believe <ngjesh> can have
> anything to do with <dorë>; the standard derivation from
> *h1en-)joh3s- is impeccable [...]

do I understand you right here when I think you mean "ngjesh" <
"*h^en-)joh3s-"?
If yes, trough which kind of changes we will get "ghe" in Rom. in this
case? I ask it since I consider Albanian "ngjesh" is the same word as
Rom. "înghesui"

> [...], number 7 also looks impossible to me,
> especially because **g^Hos- 'guest' is a fake etymon. <gjymësë> is
> related neither to *g^em- nor to *sem- (BTW, two different IE

roots),

> whatever else it represents. Daco-Romanian jumãtate and Aromanian
> g^umitate are obviously related to it. Hamp reconstructs

pre-Albanian

> *jumitja: (shouldn't it be *ju:mitja: ?), but in fact the form seems
> to be hopelessly ambiguous.

"*ju:mitja:" looks somehow odd to me; the root for Rom. "jumãtate"
appears to be "jumã-" due the expression "half-half" which is
"juma-juma"( if in this expression is not a reduction of the word due
the lassyness of the speakers, thus a reduction of "jumatate-jumatate"
to "juma-juma"). Beside of this an "tj" here will have had the big
chances to have an "T" as outcome in Rom., thus "*jumiTa-" should have
been the output.
BTW, is the word "gjysmë" a newer form as "gjymësë"? In my dictionary
there is no trace of "gjymësë" but a lot of derivatives with "gjysmë-"

Alex