Re: [tied] Lynx

From: Piotr Gasiorowski
Message: 45490
Date: 2006-07-24

On 2006-07-23 02:12, Grzegorz Jagodzinski wrote:

> Come on, give up with this. I have written correctly. In the clause
> "Lith. lú:s^is, Latv. lu~sis, OPruss. luysis (uncertain) show no
> traces of *n" the subject is "Lith. lú:s^is, Latv. lu~sis, OPruss.
> luysis (uncertain)". Can you see ANY traces of *n in these three
> recordings which are in the subject of my sentence?

Not in the forms you quote, but Baltic is more than the sum of standard
Modern Lithuanian and Latvian (plus what remains of Old Prussian data).
Sergei has already explained why the evidence of standard Lithuanian
can't be decisive. On the other hand, there is _positive_ evidence for a
nasal in Z^emaitian dialects and I see no reason to ignore it.

Incidentally Old Prussian luysis is _not_ a ghost word: it occurs in the
Elbing Vocabulary (#666 Luysis : Luchs).

> This is a well-known thing. If the Gmc word had long -u:-, I would
> say that it would allow us think about the possibility of *lunk-. But
> as no Gmc forms have long -u:-, I simply say that the evidence from
> Germanic contradicts your reconstruction.

I don't claim that Germanic *lux-a-, *lux-su/a- are stem cognates of
*lunk^-. They are just derived from the same root. If there are two root
cognates not reducible to the same stem prototype within Germanic, why
should we expect complete uniformity across Indo-European?

> Or, the same root could occur without any other suffixes (cf. Old
> Swedish ló). It is in a not-so-good concordance with your agent noun
> hypothesis.

Old Swedish ló is not *luk^- without any suffixes. It's *luk^- with a
thematic vowel, and thematisation is a way of forming adjectival
derivatives from root nouns, often with agentive meanings -- the type of
Skt. bHujá- 'something that bends (arm, coil, elephant's trunk)'.

> And according to your deduction, Greek and Samogitian (and Slavic, as
> you assume) had metathesis, Celtic and Germanic had not any *n, and
> only Armenian preserved the root without metathesis. Interesting,
> especially for an advocate of unexceptional sound rules.

I'm not a dogmatic defender of exceptionless sound changes. I can gladly
accept the replacement of *l by *r in Slavic *rysI as due to
contamination or folk-etymology. I have also pointed out many times that
metathesis is a type of change notorious for its tendency to occur
sporadically. At any rate, I don't derive all the 'lynx' words from the
same stem. At best, they are parallel derivatives of the same root.
Indeed, my etymology amounts to proposing a root like *leuk^- with the
probable approximate meaning 'hurt, attack fiercely'. The basis is still
shaky -- the 'lynx' etymon and Skt. rus'áti, but I hope to be back with
better stuff soon.

> And in
> lusunank', lus- stay for *luk^-, -un- stay for *-n- or even -nH-, and
> what does the second -n- stay for?
>
> Because of the two n's, I would say that the Armenian word is so
> aberrant that it can hardly count for any evidence.

As often in Armenian, "Though this be madness, yet there's method in
't." Miguel has already spared me the need to explain the details. The
nom.sg. <lusan>* of <lusanunk'> derives from a form like *leuk^-n.t-,
which supports my agent-noun hypothesis.

> 1) examples for *iR/*uR < *R. 2) examples for *íR/*úR < inherited
> *iR/*uR, if really any are present 3) examples for *íR/*úR < *RH
>
> Does it mean that 2) and 3) merged completely? Or, there is a way to
> separate one form another?

I think there _are_ some relevant examples, but please give me a little
time to prepare my case. I'm on vacation, which means, beside many
pleasant things, that the university library is out of my reach at the
moment.

Piotr