27-11-03 20:22, Miguel Carrasquer wrote:
> On Thu, 27 Nov 2003 18:05:15 +0000, Glen Gordon <glengordon01@...>
>>I'm still thinking about the supposed accent on *u in the variant
>>*kúwo:n. Yet with *dúwo:-, it becomes a little more suspect as
>>evidence for Piotr's case. Is it possible that this is just a rule
>>affecting nouns starting in Cw- where C is any stop?
>
> As long as they are monosyllabic. *dwo: and *duwo: are called "Lindeman
> variants". I'm not sure if the phenomenon is known as "Lindeman's Law",
> but, in any case, you have rediscovered it.
Lindeman's "law", which is a subcase of generalised Sievers-Edgerton's
Law, describes the variable treatment of initial CR clusters, where R is
any liquid, nasal or semivowel. Typical examples include *d[i]je:us,
*gW[n.]na:h2, *h1[i]jonts and *g^H[i]jo:m . Let's note, however, that
Lindeman variants normally show unshifted stress, and that they are very
rare when R = *w. Apart from *k^uwo:n and *duwo:, the only clear example
I can only think of is the Sanskrit treatment of pronominal forms
(<tva:M> ~ <tuva:M>). Other examples (like <svar-> ~ <suvar-> 'sun')
involve roots with laryngeals (*sh2wel-) that affected their
syllabification. There are no Lindeman variants of such well-known roots
as *dHwor- 'door' or *g^Hwer- 'wild animal', so in my opinion *k^uwo:n
and *duwo: really stand apart.
Piotr