OK, why don't we look under the carpet and
examine a few of these "doublets"? Perhaps we could find an explanation for
their occurrence in PIE or at least speculate about it. I could imagine a few
reasons why such root constellations exist.
First possibility (already discussed ad
nauseam): purely accidental homophony or near-homophony. Hardly exciting but
common.
Second possibility:
onomatopoeia, phonetic iconicity and synaesthesia. It has been argued that some
articulatory gesture sequences may symbolise different kinds of movement. If P =
labial, T = coronal and K = dorsal, then T-K is often associated with touching
or pointing (Latin tango, digitus, French toucher < popular Latin *toccare,
Polish tykac', Greek deiknumi, ... -- all historically unrelated but
semantically close; cf. even that crazy Proto-World reconstruction *dik-
'finger', supported by lookalikes from many families); T-P has to do with
tapping, stamping, etc.; and K-P with restraining or holding something (PIE
*gHebH- [Latin "have", English "give"], *kap- [Latin "capture", English
"have"], various kup-, kub- etc. roots for 'container'). A warning is in order
here. Semantics is an extremely delicate business, and it's easy to get
entrapped in circular argumentation. English keep could be used as a strikingly
apt example of K-P, but its modern meaning is a recent affair. In OE and ME
ce:pan/keepen meant more or less 'care, heed, notice'. The Modern English
meaning has developed via 'observe, maintain' on the one hand, and 'seize,
avail oneself (of sth desired)' on the other -- a wide range of meanings
evolving simultaneously. Do we have a right to select only the one that fits our
theory and dismiss the rest as "peripheral"?
Third possibility: borrowing into PIE from
related but different non-IE sources at an early date. An attractive solution at
times, but always hard to justify unless the source(s) can be identified with
some confidence.
Fourth possibility: "real" PIE was no doubt
an ordinary messy language with a lot of dialectal variation and interdialectal
diffusion. There are in any language and at any time irregular processes
producing sporadic variation that cannot be captured in terms of sound laws.
Nothing to worry about as long as we don't admit a greater amount of such
irregularity than could be realistically expected. If there is too much of it,
we should start suspecting that we're missing a hidden generalisation (like the
laws of Grassmann and Verner).
Fifth and last possibility -- the most
exciting one, perhaps, from the point of view of the historical
phonologist. (Some of?) the doublets can be related to each other via internal
reconstruction. This would mean doing for PIE what Verner did for Proto-Germanic
and Grassmann for Greek and Indic. Their job was made easier because they could
verify their findings using data from the remaining IE branches (e.g. Vedic
accent for Verner's Law). It isn't clear what, if any, external
evidence is acceptable for PIE.
So what about a few doublets to analyse,
Miguel? do you happen to have a list?
Piotr
----- Original Message -----
Sent: Wednesday, January 24, 2001 1:20 AM
Subject: Re: [tied] Re: Ford -furta- fare
The "doublets", which, even if we don't "allow" them, are
still there [look under the carpet!], explain the curious fact that sometimes
both proposals overlap semantically, but not, strictly, phonetically. What
the significance, if any, is of this, I couldn't say. The only thing that
seems obvious is that if Nostratic exists, the exact sound correspondences (as
yet unknown) will hardly be as simple as, say, Bomhard's PIE ~ PAA *t ~ *t, *d ~
*t', *dh ~ *d. That just may be correct as a Grimmian first approximation,
but there's a whole lot of Verner still required.