Re: [tied] Olsen's Law [was: PIE Ploughs]

From: elmeras2000
Message: 29997
Date: 2004-01-25

--- In cybalist@yahoogroups.com, "P&G" <petegray@...> wrote:

> There seem to be a number of roots with final laryngeal that also
show forms
> with -i- [...]. They appear in
> Pokorny with -(i) at the end. Piotr has suggested a reduction
of an
> original diphthong, but that explains the variation -euH/-eH and -
eiH/-eH,
> not the sequence -Hi. Besides, it is an i that pops up so
unexpectedly in
> these words, not u.
> I don't know of any discussion of this. Any ideas?

Plenty. I wrote a lengthy treatment of the matter. It constitutes
chapter one of my "Studien zur Morphophonemik der indogermanischen
Grundsprache", Innsbruck 1989.

There is no regular coming and going of /w/ in such roots, only
of /y/. The basic structure is patently *CeHy-, i.e. roots ending in
laryngeal + /y/. The /y/ is an underlying part of the root, not a
suffix that got transferred from , say, present formations. The
basic rule is that, in the full grade, the /y/ is retained before a
vowel, but is lost before a tautosyllabic consonant (*seH1y-e/o-
'sift', *seH1-tlo- 'sieve'). Before CV there is metathesis
producing forms like *poyH2-men-, *poyH2-wa-H2 vs. *poH2y-u-, *poH2-
mn, *paH2-thlo-m. In the zero-grade, the /y/ is lost before /t/, but
retained before sonant + vowel; it also appears to be lost before
clusters.

Coming and going of /y/ and /w/ alike applies to the eru/ru:
problematics (next chapter of the same book): Roots ending in *-ERHw
and *-ERHy lose the entire segment /Hw/ and /Hy/ before
tautosyllabic consonants: *wel-mn : wel&w-e/o- 'envelop'. In the
zero-grade they metathesize to *-RuH/*-RiH (*wluH-to-). It seems to
work the same with all sonants and with all laryngeals.

Both sets have been treated by Kurylowicz who has a fine collection
of examples (Idg.Gr. II, 1968, 216f; already Apophonie, 1956, 123f),
but, strange to tell, Kurylowicz saw tru:-/tri:- as the zero-grade
of original *teru-/*teri-, thereby forgetting the laryngeal which is
crucial.

In both sets there is a lot of analogical restructuring, but enough
archaisms remain to make the original distribution stand out.

Jens