On 2006-01-12 14:29, alexandru_mg3 wrote:
> Opposed to this, the 'Niger' by its nature is a man (and usually not
> every known men, by its nature, were dark) so they have constructed
> *negW-ro in this case, trying to remember that that person has a
> property 'dark' but is 'a man' and not 'a kind of dark'
Is it really possible that you have not yet grasped the simple fact that
Lat. niger is an ordinary adjective meaning 'black, dark, dusky', and is
_not_ applied specifically to humans? If you "doubt" that, as you say, I
wonder on what basis. The attested use of the word in Latin (with plenty
of examples) should lay your doubts to rest.
> It is a reason : The PIE Roots are contructed in an uniform way: CVC
> (using 'a more general' definition for C and V) so we cannot have a
> PIE root *negWt- *nekWt- (see also Derksen's decomposition *nokW-t-is
> etc..)
No, Marius, there are many PIE roots ending in two consonants, including
cases when both consonants are obstruents. Even if you treat the final
consonant as a kind of old extension, that doesn't make it extraneous to
the root in PIE. A verb root like *nekWt- doesn't violate any IE
morpheme-structure constraints that I know of. True, many authors
hyphenate *nokW-t-s in the way Derksen does, but this merely reflects
their belief that the root *nekW- 'get dark' really exists, which is
questionable i.m.o.; and the *-t- extension is blithely left unaccounted
for, as if such an athematic formation were attested for PIE.
> So the PAlb word c^wa:ra: 'crow', usually derived as *kwers-no or
> kwers-n-eh2, didn't contain any -n-.
>
> In this case the single (and the logical solution) that remains
> for 'crow' is *kwers-ro (or *kwers-r-eh2).
In this way you lose touch with potential cognates elsewhere (and thus
with reality). The _colour_ adjective *kWr.s-nó- (of which *kWers-nah2
would be a vr.ddhied substantivisation, like *bHerh1g^-o- 'birch' from
*bHr.h1g^-o-) is independently attested with the meaning 'black' (Slavic
*c^IrnU, Skr. kr.s.n.a-), while a variant with *-ró- isn't.
The objection that Romanian preserves Latin /-rn-/ while Albanian
doesn't is not an obstacle. Romanian isn't simply a continuation of
Dacian or Proto-Albanian. It's a Romance dialect that borrowed many
items from its Albanoid substrate, where inherited *-rn- could already
have become emphatic *-R-. If Latin borrowings continued to undergo the
change _in Albanian_, that doesn't mean that the same would have
happened automatically _in Romanian_, even if bilingualism was
widespread at the time. If you speak two languages, some phonotactic
transfer is likely but not inevitable.
> Viewing this, I remembered your suposition that *bhrh1g'-ró- could be
> the right source of Romanian bardza => and I say "yes" : it should be
> a uniform way 'to construct' such 'similar-
> to' 'objects,properties,...,man,birds' (thanks for you correction
> related to my English, in Romanian is 'similar cu') : *nig-ro,
> *bardz-ra, *cwa:r-ra
>
> Also, next, I said: why to consider 'a PIE reduction' and not
> a 'PRomanian reduction' *bardzra > bardza (knowing the evolution:
> Latin fra:ter > PRom fratre (attested) > Rom. frate)
>
> In this case the Romanian bardza 'storck' is from *bhrh1g'-ro and the
> Albanian bardhë 'white' is 'simple' from *bhrh1g'-o (or *bhrh1g'-eh2
> if the stems wasn't changed in PAlb, as I suppose) => that has Sense
> for a Semantical point of view.
That's a may-or-may-not-be story not backed by any solid evidence.
There's no attestation of *bHr.h1g^-ró- (as distinct from *bHr.h1g^-ó-)
with the meaning 'a kind of white creature'). In fact, the only such
substantivisation that is well-attested shows no *-r-; it's the word
*bHérh1g^-o-s 'birch' (coll. *bHr.h1g^-ah2). Also, if Proto-Albanian had
had a word derived directly from *bHr.h1g^ró-, the *g^ would have not
produced an affricate, since Albanian shows dispalatalised reflexes of
the *k^ series before liquids.
Piotr