--- In
cybalist@yahoogroups.com, Piotr Gasiorowski <gpiotr@...>
wrote:
>
> On 2006-03-31 17:59, alexandru_mg3 wrote:
>
> > It was *melh1- with h1
>
> I wouldn't say that you can't be right, but I just wonder what
makes you
> so certain. I can't see much evidence either way. The Gk. word
family
> built around malak- 'soft' seems to indicate *h2 _if_ it's the
same
> root. I suppose that's why most (though not all) IEists prefer the
> reconstruction *melh2-. Some just refrain from specifying the
index.
>
> Piotr
>
0. "malak- 'soft' seems to indicate *h2 _if_ it's the same
root."
-> well said: 'IF' => but seems that you doubt (as I doubt too)
about 'that semantic link'.
1. my rejection of *h2 in *melh- is based on Romanian mãlai that
means 'flour, or a <specific composition> made by flour'
For the meaning please see:
http://dexonline.ro/search.php?
cuv=malai&source=
Romanian mãlái < PAlb/Dacian? *mallája: < PAlb./Dacian?
*malwája: /syll.-by-syll.: *ma-lwá-ja:/ < PIE *mlh-wó-yo/eh2 < PIE
*melh- 'flour'
So the link: Romanian mãlai 'flour' to the PIE *melh- 'flour'
appears to be obvious (at least for me).
Note: PAlb *lw > PAlb *ll (> later Rom l) explains also the lack of
rothacism in Romanian mãlai (see a similar case in Romanian mal >
Dacian Mal-wa)
Why we don't have h2 in melh- ?
-----------------------------------------------
1. First we need to consider for mãlai that we have in the root the
nil-grade *mlh- and not the full-grade *melh- > because in mãlái
the stress pattern is not on the root: it was *mlh-wó-yo/eh2 (and
not *melH-wó-yo/eh2)
2. Next 'the supposed' *melh2- (see Beekes etc..) would give in
this particular case as nil-grade *mlh2-wó-yo/eh2
But *lh2 , *rh2 gave in PAlb *ul, *ur :
see as examples:
1. *g^rh2-no/eh2 > PAlb *gruna: (see Alb grurë)
2. *gWerh2- 'mountain'> PIE *gWrh2-i- > PAlb *gur-i- 'stone'
(see Alb. gur <-> Rom grui)
3. *gwrih2- 'neck' > PIE *gwrh2-mo-dy-o > PAlb/Dacian *grumadya
PAlb/Dacian *grumadza (see Rom. grumadz 'both parts on the back of
the neck' <-> Alb gurmaz 'id'
(I suspect that *grumadza was reshaped from a dual-form similar with
PAlb. *budya: < Rom budza <-> Alb. buza)
I know that you 'don't trust' this PAlb rule: PIE *rh2, *lh2 > PAlb
*ul, *ur and as results the above derivations too, but 'they are too
many' and finally they have one big advantage : they 'arrived to be
exaplained on a regular basis'.
NOTE also that (if the rule above is true, and I think so): we have
*h2 in *gWerh2- 'mountain' and *h2 in *gwrih2- 'neck'
So applying PIE *lh2 > PAlb/Dacian? *ul to a supposed root *melh2- >
*mlh2- we would obtain:
PIE *mlh2-wó-yo/eh2 > PAlb *mulwája: > Romanian *muláj that is not
the case: the Romanian form is mãlái
In conclusion: we cannot have *h2 in *melh-
3. Next a *h1 in 'a possible' *mlh1-wó-yo/eh2 form give us the
correct form: PAlb/Dacian? malwája: > Romanian mãlái
see for PIE *rh1, *lh1 > PAlb *ar, *al:
*bHrh1g'-o/eh2 > PAlb bárdza:
*(s)krh1-p-o/eh2 > PAlb kárpa:
(Please remember that you didn't answer to my question: 'why not o-
grades in place of the above forms?' I really need to know :))
4. But *h3 cannot be excluded too based only on Romanian 'mãlái':
Because *mlh3-wó-yo/eh2 gave us the correct form too : PAlb/Dacian?
malwája > Romanian mãlái
For *rh3, *lh3 > PAlb *ar, *al:
*prh3-wo/eh2 > PAlb pára: (Alb. parë)
5. Finally, viewing that I need to find out 'somehow' if it was h1
or h3, I have tried to find-out 'other' PIE <<flour, to grind>>
roots to see if there was h1 or h3 in that roots too:
And I found:
*h2elh1- `grind' =>
Greek aleo:
see also:
PIE *h2leh1-ur. > Arm. alwer 'flour' (Beekes)
Of course 'maybe there is no direct link' between *h2elh1- 'flour,
to grind' and *melh1- 'flour, to grind' however (don't ask me
why), 'I prefer to see a link' => so the 'final' reconstruction is
*melh1-
Marius
P>S> Do you have some other arguments "on why not melh3-"? I saw
that nobody proposed *melh3- however I'm not aware of the reason
behind (but my argument no. 5 that is not a 'direct' one)