*G^EN-

[Stanley Friesen]

At 09:07 AM 6/8/01 -0500, proto-language wrote:

>Methodologically, if it (and some others) can be derived from *both* *g^en-
>and *g^enH-, I see no reason to assume the non-existence of *g^en- just
>because all forms can be derived from *g^enH-

Actually, there is a methodological reason to at least not reconstruct a
*g^en- (even if one cannot positively say it did not exist).  It is called
Occam's Razor - do not multiply entities beyond necessity. In a case like
this there is no *need* to introduce three elements: a root, and extension,
and an extended root, so it should not be done.  Only if there is some
specific evidence requiring the existence of the "non-extended" root should
it be reconstructed as such.

Now, if Skt 'janati' does in fact point to *g^en instead of *g^enH, this
could be such evidence.  I am not really qualified to judge, but I am
somewhat skeptical.  For one thing, it is generally held that a form needs
to be found in two or three branches of IE before it can reconstructed for
PIE, so if Skt is the only source seeming to require bare *g^en, there is a
very real possibility of innovation in Indic.

>. My preference would be to assume that a root, *g^en- (which I would consider
>to be *basic*) and a root with extension, *g^enH- have been conflated into a
>single paradigm.

I would apply Occam, and only introduce this entity if I could reconstruct
it as such.

--------------
May the peace of God be with you.         sarima at friesen.net