Re: [tied] Re: *H1wes-

From: proto-language
Message: 7950
Date: 2001-07-18

Dear Torsten and Cybalisters:
----- Original Message -----
From: Piotr Gasiorowski
Sent: Wednesday, July 18, 2001 11:30 AM
Subject: Re: [tied] Re: *H1wes-

*woh1-r- (as in Skt. va:r- 'water', Toch. B war, OE wo:r, etc.) and something like *woh1-s- could well be reduced to a common denominator even without a Grand Unification of all *w-initial water words. This would have the advantage of making the long vowels unproblematic. Furthermore, a laryngeal in *weh1-r- is independently needed to account for the post-PIE nil grade *u:r-, and since *weh1r- is not an acceptable root shape, its form implies root status for *weh1-, and partial unification begins to make sense. I do not see any good reason, though, why *h1wed- should be reconstructed instead of just *wed- in the "wet, water" root.
In my opinion, there are several good reasons why *H[2]ewed- should be preferred to *wed-:
1) Languages are built not invented. Complex roots and inflections are constructed of existing elements in a language (if not borrowed) not arbitrarily imagined. There is no verbum ex machina.
a) There are three other IE roots (at a minimum) that establish a connection between an element, *H[2]e, and 'water': *H[2]eb-, 'water'; *H[2]ep-, 'water'; and *H[2]epero-, 'bank'.
b) There are other indications that this element was in areal use for 'water': e.g. Sumerian a, 'water'.
2) Therefore, it comes as no great surprise to see Pokorny listing the root in question under *aw-, what some of us would indicate today as *H[2]ew-.
3)The association of this element with several IE words meaning 'water' suggests that the presumption of reconstructing this element for other words meaning 'water', the form of which permits, should be made.
4) Therefore, Pokorny, quite rightly, reconstructs *awed- (*H[2]ewed-) for the root in question.
a) The justifiable presumption is that we are dealing with two root-extensions of a root (*H[2]ew-) itself built from *H[2]e + *-w.
5. After the introduction of the IE *e/*o/*0 Ablaut, an [a]-quality was sufficient to differentiate a root from a root of the same consonantal form with Ablaut; therefore, in almost all cases of IE roots reconstructed with [a], we have attested variants: [a] and [a:]. The maintenance of the length of [a:] was not necessary to establish lexical differentiation. The length of the [a], so long as the vowel-quality is assured, is therefore secondary to reconstructing an earlier *H[2]e as the underlying source.
6. So we can visualize that IE *a:wC- became *auC- as in Old Indian o:dati: but also became *wa:, with the long vowel produced by the missing *a/a:, as in Hittite wa:tar, 'water'. It is also clear that in zero-grade, *auC could appear simply as *u and *u: as in Lithuanian údra.
7. Unless one wants to assume, which I believe would be wrong, that each IE word should be analyzed in an absolute vacuum, it seems certain that *wed- should be referred to *H[2]ewed-.
8. Let me add, though it is not of interest in the question here, that I believe the better reconstruction is *Haw-, with the "laryngeal" lengthening and preserving the pre-PIE vowel (*a:).
(501) 227-9947 * 9115 W. 34th St. Little Rock, AR 72204-4441 USA

"Veit ec at ec hecc, vindgá meiði a netr allar nío,
geiri vndaþr . . . a þeim meiþi, er mangi veit,
hvers hann af rótom renn." (Hávamál 138)