Re: Burushaski

--- In cybalist@yahoogroups.com, Piotr Gasiorowski <gpiotr@...> wrote:

>
> On 2008-02-20 03:11, stlatos wrote:
>
> > Changes s- > h- ( > 0 before C)
> >
> > *septm' > *hatwu > *hltu > tHalo '7'
>
> This does _not_ illustrate the change *s- > h- but something much more
> complex in which *s- > h- is just a hypothetical intermediate step.

No correspondence has a difference of only one sound change. As I add more, I'll put
them in this category and some will be clearer than others.

> You
> make an extra assumption practically every time you propose a match. Why
> is the final *-m lost here but retained in *dek^m.?

As said, it's possible stressed m>mu not m>um as in others. Also, as the endings of
the numerals vary so much (from cont. with each other), this particular criticism would
never be a problem.

>
> > *smi:x > mi-, Arm mi
> >
> > tsundo '5'
> > mis^indo '1+5 > 6'
>
> You are mixing up the Burushaski dialects. If you use Yasin tHaló, you
> should use also <bis'índu>, even if it complicates your derivation.
> Anyway, is there any other evidence for *(s)mi- 'one' in Burushaski?

Some show older stages. As a preliminary description, I'm not going to worry about
m>b (when already seen in Muz^a-, Pru:s^ava, Burus^o).

> > *pc^li > c^api 'pliers'
>
> With so much metathetic freedom plus random deletion and epenthesis any
> pair of words can magically become a match.

Look, all the surrounding IE languages lose final -a(s) and create odd clusters.
Sometimes there's met. or other changes. Even if areal, I'd explain it the same way.

> > m>w after (all?) C
> >
> > *septm' > *hatwu > *hltu > t(h)alo '7'
> >
> > *piixYmYn, > *firwan > iran 'cream'...
>
> Where's the /w/? And this particular kind of rhotacism has not been
> mentioned before. Any other examples of *h1 > r?

How do I know what rw gives compared to stop+w? I don't even know if all m>w
happened at the same time (that is, compared to w>gW). If I do more work and can't
explain it with more ev., then I'll worry.

>
> > *kYmtom > *c^umtam > *suta > *sta > tHa '100'
> >
> > *wiikYmt- > *dwikYumt- > *ltirut- > *altr > altar '20'
>
> <áltar> looks obviously related to <altán>.

Just like two/twenty.

> The rest is guesswork, and
> rather complicated guesswork at that. It seems *k^m.(t) can develop any
> odd way, yielding -rum, -r, or tH-, depending on expedience.

Completely ridiculous. There's no -t in 'ten', like in most. The reason why doesn't
matter, even -t > 0 beforehand would work, but a change umt > uwt > ut is just like I
already gave for Slavic, and a regular *ltiruta > *altirura or sim. could be expected to
undergo haplology.

> I know you
> regard this variety as conditioned, but this conditioning is far too ad
> hoc for my taste.
>
> > *kYswekYs? > *c^swac^s '6'
>
> Unattested, but what the hell.

I gave that to show where the initial in '5' comes from, as often happens.

> I'm not in principle against odd or weakly supported sound changes,
> occasional irregularities, etc. But the acceptance of such messy facts
> of life is a luxury we can afford if a relationship has been established
> on a solid basis (numerous and strongly supported regular
> correspondences in the lexicon and inflectional morphology). If
> _everything_ is odd, tentative and half-regular, it's just a house of cards.

This is not a description of what I've done. In only a few days I've seen almost every Br.
word fit into the rules. If, for example, p>h is not conditioned, why do my examples all
show this? The correspondences of exact meaning are many, simple differences like
boy/son are hardly important. If these correspondences weren't real, I couldn't keep the
same rules for all of them. That is, if Ks > ts didn't occur, why are both tsel and tsendo
derived from such? How can all those I derive from e/o-i > a-i > e have matching
cognates? Shouldn't Br. e be independent of IE cognates if not derived from IE?