Re: [tied] Re: PIE i- and u-stems again

On Tue, 20 Feb 2007 22:06:26 +0100, Piotr Gasiorowski
<gpiotr@...> wrote:

>On 2007-02-19 21:31, Miguel Carrasquer Vidal wrote:
>
>> The question where the *i comes from aside, the G.sg. of
>> *kWis _is_ *kWesio (e.g. OCS c^I-to, c^eso) and the G.sg. of
>> *is _is_ *esio, just like *so has Gsg. *tosio, etc.
>
>Well, there's also Goth. gen.sg. þis

I think I have compared that in the past with Pol. tego, for
expected *togo :-).

>as well as instr.sg. neut. þe: 'the
>(+ comp.)', and I do't think it's evident that forms like *tosjo and
>*toh1 are primary. On the other hand, anaphoric *is may have a
>"thematic" counterpart (Hitt. a-).

The Hittite anaphoric pronoun is rather interesting. In Theo
van den Hout's lecture notes (Amsterdam, 1991) the paradigm
is given as:

N. asi (asis, enis, annis)
A. uni (unin)
n. eni (uni)
G. unias
D. edani
L. edi
Ab edez
pl.
N. unius (e-sta, enius)
n. eni
D. edas

The N.sg. form can go back to *o-s with a deictic element
-i. The A.sg. I can only explain as syllabic (enclitic?)
syllabic *m., with the same suffix -i added (but cf. kas,
kun; apas, apun: is the acc. there also *k^m., *obhm.?). The
neuter and oblique seem to be based on e-.

>> It is
>> undeniable that there were two different pronominal
>> declensions in PIE, and I find nothing surprising about the
>> fact that the two differ both in the strong cases (*-os,
>> *-om, *-od, *-ah2 vs. *-is, *-im, *-id, *-ih2) and in the
>> weak cases (*-o- vs. *-e-).
>
>But *-ah2 means *//-e-h2//, so there are at least some e-forms in the
>first set.

Of course: it's the thematic vowel, which gives /e/ before
voiceless /h2/.

I do not fully understand how you would explain all the
forms of the "i-pronouns", if they too contain the thematic
vowel. OK, I gather the NAsg. has a reduced thematic vowel
which gives /i/ (*is, *im; *id; *ih2) [but why only in the
strong cases?]. But if the oblique forms with /e/ contain
the thematic vowel, why does that not give /o/ where it
should? The /s/ in most of the oblique forms (*esyo,
*esmõi, *esmi, *esmoh1, *esmot) avoids the problem, and
perhaps you don't accept a *sm-less abl. *eot, and you would
reconstruct the *sm-less ins. sg. as *-h1 [*kWe-h1] (not
*-eh1), but that still leaves the plural *e-j-, the dual
obl. *e-ih1-, and the "absolutive" (endingless nominative)
*ei-, where a thematic vowel before *i/*j _should_ have
developed to /o/.

>Variation between the thematic vowel and *-i- is not
>restricted to pronouns: cf. free-standing *dwo- vs. compositional *dwi-,
>*wl.'kWo-/*wl.kWí-h2, *néwo-/*néwi-o-, etc.

I'm not convinced we're dealing with the thematic vowel
here. Normally, *-ih2 (devi:-type) is explicitly the
feminine marker of _athematic_ stems. The ending *-ih2-s in
"she-wolf" is a different formation, which may well be
originally bi-morphemic: *-i-ih2, where wl.kW-i-ih2 could be
parsed, right to left, as "she (*ih2) of (*i) wolf
(*wl.kW-)". In the other cases, we may also have this same
adjectival/genitival morpheme *-i- (*dw-i- "of two";
new-i-om "that of now" > "new", etc.). And this i-element
is not confined to thematic formations: we also find it in
compositional forms within the "Caland system".

>> ... Slavic *-i:- (circumflex) cannot come from *-éh1-,
>> *-h1-, *-h1jé- or *-éh1je-, which would have given -ê"-,
>> -0-, -jé- and -ê~-, respectively. And Baltic -i- (short)
>> cannot come from any of those either. In particular, a
>> present form in *-h1jé- would have merged with the jé-stems,
>> and this is precisely what did *not* happen in Baltic or
>> Slavic (although it happened in other branches).
>
>I mean, are there any problems whatsoever with the "thematic-derived"
>type (nove^je/o- etc.) if one derives it from *-e-h1-je/o-?

None (but neither for *-eh1-je/o-, and *-eh1i-e/o-).

>The present
>of the "athematic-derived" type for some reason fell together with
>whatever had become of the *-éje/o- causatives/iteratives. I admit I
>can't think of a convincing reason for that at the moment, so please
>give me a little time to think the problem over. It might be significant
>that there is a marginal overlap between the developments of *-(h1)je/o-
>and *-eje/o- (the 1sg. has *-joN in both types).

=======================
Miguel Carrasquer Vidal
miguelc@...