[Fwd: Re: [tied] Timing of ablaut]

On Fri, 26 Sep 2003, Miguel Carrasquer wrote:

> >> >
> >> >If stéar (stéa:r) is Proto-Greek /stá:(j)ar/ or /sté:(j)ar/ (thus
> Frisk), then it does not show anything about the quality of the
> laryngeal: IE *stá:yH3-r or *sté:yH3-r would be expected to
> produce this result. The lengthened grade would be as in *yé:kW-r
> 'liver'.
> >>
> >> I was thinking of *steh2yr. > *sta:yar, which explains the length
> and
> the
> >> a-quality in a straightforward way.
> >
> >Is *yé:kW-r not straightforward?
>
> In my analysis, not really.  As you know, I consider the
> straightforward lengthened grade of the pre-PIE vowels **a and **u to
> give PIE *o, so that
> *yé:kWr can only be a straightforward lengthened grade if the original
> vocalism was **i.  In the case of *yé:kWr, the **i vocalism is
> confirmed by
> the alternation *l ~ *y (Germanic liver, Armenian leard), pointing to
> palatalized *l^ due precisely to *i: in the strong root.  [The weak
> root, e.g. G. *li:pu-án-a:s should have given *(l/y)ikWnós, preserved
> in Irish and Slavic in the meaning "fish roe", but otherwise replaced
> by
> analogical
> *yekWnós].

A form *likW-n-ós (supposing there was an /l/, or even a separate
phoneme /l^/) would be the one to be secondary. It is commonplace for
Narten forms to be replaced by normal ablaut (short/zero). The
alternation /e:/ : /e/ does not point to /i/ by any rules for which I
know any evidence.

So, to me, *yékW-r (or was it *lyé:kW-r or even *{l^}(y)é:kW-r?) is
quite unproblematic in its vocalism and is easily parallelled.

> >And is there a noun-forming suffix *-yer-?
>
> I would analyze it as *stah2i- (probably *stah2- + *-i) with the noun
> forming suffix *-(V)n-: *stah2i-an > *sta:yr.

How did the -i- part get into the picture? Are we back to the old school
of long diphthong analysis where a /y/ just cut loose from somewhere and
went on the rampage through the morphology? I have written a book
against that, so don't expect me to accept it. Except for cases of
analogical spread of a type, long diphthong roots have the long variant
underlyingly, meaning that the y-forms are not extensions of shorter
forms, on the contrary the short alternants are reductions of the full
forms.

>
> >> Apart from Greek stéar, there is no
> >> other indication about the nature of the laryngeal in this word.
> All things considered, I'd say Skt. stya:yate: "gerinnt, wird hart"
> is
> more
> >> likely to be an *a:y-formation (denominative from *stih2ah2 "etwas
> dichtes") than a causative. The connection with *steh2-, as
> suggested
> in
> >> EIEC, seems appropriate for the semantics of the cognate group.
> >
> >If IE *stói-no-s/-na-H2 seen in Goth. stains 'stone' and Slavic
> ste^na
> (b)
> >'wall' is "something made of compact stuff", the lost laryngeal will have
> to have been rounded, since the derivative has reduced *-mn- to *-n-,
> not *-m-. This law is not completely flawless, but it comes very, very
> close
> to
> >being just that.
>
> But not all -no stems come from earlier -mno-.  If I may quote "The
> noun in
> Biblical Armenian". p. 834: "A suffix *-no- has at least two potential
> sources: it may represent the thematicization of a basic stem or word
> form
> ending in *-n (origial n-stem or heteroclitic), or it may be a
> secondary derivative of a *-men- stem, i.e. *-mno- > *-no-".  Now if
> stéa:r is from a
> heteroclitic **stah2i-(a)n-, as I argued above, then **stah2i-nos/
> *stah2inah2 is a thematized heteroclitic.

A protoform *staH2ynV- should be metathecized to *stayH2nV- and so
produce a Balto-Slavic acute, but Slavic ste^na has circumflex; it also
reflects radical accent just as a regular o-infix form *stóy-na-H2 with
regular loss of the laryngeal should.

>
> >It is also part of the evidence
> >that the same root forms Slavic te^sto and OIr. taes, Welsh toes
> 'dough' pointing to *taisto- which may be a perfect superlative of an
> adjective, IE *táy(H)-isto- 'most compact' showing that any a-timbre
> seen in forms
> of
> >this root does not have to come from the laryngeal, for the root had
> /a/ itself. -
>
> I would argue that the /a/ timbre comes from *h2 (*stah2i-t >
> *staih2t-), cf. also Greek staís/staîs, staitós "dough" (Boisacq also
> connects OHG deismo, OE thae:sma as *tais-tm-).

You obviously would, but there is no need for that, as I have shown. I
have suggested that the Greek form is the corresponding comparative
*stáyH3-is '(something) more compact, relatively compact', a formation
nice to see alongside the superlative.

> >I do not see how *stih2ah2 can be derived from *steh2-.
>
> **stah2-y-áh2 should have given *sth2i(y)ah2, but perhaps also, with
> metathesis, *stih2ah2.

If this is from 'stand', I really do not find the deletion of aspiration
in a protoform *stH2iaH2- probable, especially since it is not
phonetically regular and all other forms of 'stand' have generalized the
aspiration on the basis of the antevocalic zero-grade. Also Skt. chyáti
'severs' shows aspiration, pointing to absence of metathesis in
*sk^H2i-é-ti, a structure of precisely the kind you are dealing with. So
I think you could have only "sth(i)ya:-" from the morphological
structure you are envisaging.

> >>
> >> >The root 'to swell' is posited as *tewH2- in LIV, albeit on quite
> slender basis (sáos 'safe' < *twawo- < *tuH2-ewo- as per Peters).
> >>
> >> If it's the etymon of so:ma, it must be *tewh3-. I would consider
> that stronger evidence than sáos.
> >
> >You would of course, or else your favourite idea falls on its face.
>
> Well, what about Arm. t`oyn-k` "poison" (*teuH-no-) in the light of
> the mn-rule you mentioned above?

I don't follow: What is wrong with PIE *tów-no- from older *tOwH-mn-o-?
At least /w/ is rounded, even if the laryngeal is not, the result should
have -n- in any case. It may be surprising that this possibility has
been overlooked in Olsen's Noun; she just didn't dare to use it at the
time she wrote that part.

None of this of course *proves* that my analysis of the verbal
stem stya:ya- as *stiH3-oye- is correct. It could also be a stative
*stiH3-e-H1-yé- (via *-o-H1-yé-?) from an adjective *stiH3-o- if that
exists. What we need is of course an example that *does* reflect a
structure *-H3-éye- (root-final laryngeal three + suffix *-éye-), so
that we can see what *does* come out of pre-PIE *-H3-e- in Sanskrit.
Both anas- and apas- can have syllables closed by laryngeals, and avi-
is apparently an example of the wrong type.

Then I see no valid positive evidence for a distinction between /H3e/
and /o/ in the way Brugmann's law works in Indo-Iranian.

Jens