"Extensions" are supposedly former
derivational suffixes that became non-productive and coalesced with the root
producing a new root, unanalysable in PIE but similar in form and meaning to the
shorter one. For example, we have *ser- and *sreu- 'flow', *dreu-, *drem- and
*dreh2- 'run' (here only the extended variants survive and there's no *der-
'run'). The analysis of PIE tri- or quadriconsonantal roots in terms of
"primordial" roots with extensions is easier to accept if the semantics
is unproblematic. The 'bend' --> 'know, see' link seems rather
arbitrary to me, but *g^Heu- (well-attested especially in Greek, Armanian and
Indo-Iranian) and *g^Heud- (a "western IE" root) have the same central meaning
and the "extension" analysis is accordingly uncontroversial.
Piotr
----- Original Message -----
Sent: Saturday, October 05, 2002 2:15 AM
Subject: [tied] Verb extensions
Piotr:
>[...] presumably from the PGmc. root *geut-
< *g^Heud-, i.e. PIE *g^Heu-
>with the extension *-d-,
What's
up with this extension *-d- business? I've
seen it proposed that *weid- "to
know" is actually
an extension of *wei- "to bend" (erh, as in "to
bend the
truth a little"...). I'm just wondering how
secure this extension *-d- is. My
hypothesis is that
either there is no such extension or its occurence
is
vastly exaggerated.