Re: 3rd. person *-s(V)

--- In cybalist@yahoogroups.com, Miguel Carrasquer <mcv@...> wrote:

> On Mon, 16 Feb 2004 01:36:48 +0000, elmeras2000 <jer@...> wrote:
>
> >I find it strange that the IE reflexive is
> >not addressed in Greenberg's IE and its closest Relatives, nor,

as

> >far as I can see, in Bomhard/Kerns. It may have been treated
> >elsewhere, there is a lot of literature about Nostratic I have

not

> >seen. I am beginning to wonder if one should really be sure that

IE

> >*se is the same as the 3rd person pronoun of Uralic. The Uralic

3rd

> >person possessive is structured quite differently from the 1st

and

> >2nd person. In Eskimo too, the 3rd person has a structure all its
> >own, but the reflexive is completely parallel with the 1st and

2nd

> >person, much as IE *emó-, *tewó-, *sewó- are different from *(H1)
> >ésyo, but quite like each other. To an amateur like me that still
> >looks like a good indication that the reflexive function is the
> >original one with the IE refl.pron.
>
> By the same token, that can be taken as evidence that the original

function

> was third person, matching the formal parallelism to the

functional one (1,

> 2, 3 belong together more than 1, 2, R).

Then why is that not what we find? If I may add a completely
impressionistic idea I would suggest that the IE counterpart of the
extra-IE sibilant 3rd person pronoun is the pronoun *i-/*e-. That
could also make the thematic vowel a marking of 3rd person
belonging. I think that would accomodate something.

> A brief review of the evidence:

[...]

[...]

> Uralic
> As Jens mentioned, the third person /s/-morpheme has a different

structure

> in Uralic than the 1/2 person morphemes. [...] Not all Uralic

languages show the pattern, but enough to

> think it goes back to Proto-Uralic. We have:
>
> Finn.
> 1. mi-nä
> 2. si-nä
> 3. hä-n (*sa-n)
>
> The Baltic Finnic possessive is reconstructed as:
> 1. *-mi
> 2. *-ti ~ -di
> 3. *-sen ~ *-zen
>
> The other possessive (and verbal transitive) suffixes showing the
> difference are:
>
> Mari Mordvin Mansi Magyar PSamoyed
> 1 -m 1. -m 1. -m 1. -m 1. *-m&
> 2 -t 2. -t 2. -n 2. -d 2. *-r&
> 3 -s^&/-z^& 3. -zo, -zE 3. -t& 3. -0 < *-sa 3. *-ta
>
> The third person suffixes seem to go back to *-sa (*-za), with a

different

> vowel than the one in 1/2 sg. *-mi, *-ti (*-di). Another

possibility is

> that perhaps the 3rd. person suffixes were agglutinated later than

the 1/2

> suffixes.

[...]

> Eskimo-Aleut
>
> In Uwe Seefloth's original theory, the Eskimo facts are interpreted
> similarly to the Samoyed situation. We have (Yupik):
>
> stative/ trans./poss. trans./poss.
> intrans. sg.poss./obj. pl.poss./obj.
> 1 -Na -ka -nka
> 2 -ten -n -ten
> 3 -(q) -a -i
> pl.
> 1 -kut -pu-t -(p)pu-t
> 2 -ci -ci -(c)ci
> 3 -t -a-t -i-t
>
> The idea is that 3sgxpl -i comes from *-i-a < **-i-sa (sort of like
> Hungarian), and that the sg.xpl. paradigm was once something like:
>
> 1. *-d-m > *-n > -n+ka
> 2. *-d-(&)t > *-ten > -ten
> 3. *-j-sa > *-ia > -i,
>
> quite like the (pre-)Proto-Samoyed paradigm.
>
> The weak point is that little lexical evidence exists for an

equation

> Uralic /s/ ~ Eskimo zero. Jens suggested that the cognate of

Uralic *sa

> rather was the Eskimo reflexive pronoun -c- (in final position *-ñ
> -ni).
> I don't remember now if lexical evidence was offered for Uralic -s-

~

> Eskimo -c-.

I do not think I have expressed an opinion on the Uralic 3rd person.
If I have it should not be taken seriously. But the Eskimo //c// is
quite safe, really. It surfaces in the refl.dual and the refl.plural
which end in -z&k/-t&k and -z&ng/-t&ng respectively in Central
Yupik. They have been somewhat conflated with the 2du and 2pl which
are -z&k/-t&k and -zi/-ci in the same language. Seefloth
reconstructs the four as 2du -t&-g, refl.du -c&-g, 2pl -t&-d,
refl.pl -c&-d (perhaps, however, in part under my bad influence). I
would not put in the anaptyxis in the 2pl where the two dentals *-t-
d apparently developed into *-ci before the time of supporting
vowels. But the rest looks fine, so at least one other pair of eyes
is ready to accept a sibilant-like consonant here. Seefloth's
handout from his presentation at the ICHL in August 2003 in
Copenhagen also contains "-c" as the reconstruction of refl.sg.
possessive /-ni/, just as I have it (except that I would of course
push back *-c into a prestage of the relevant protolanguage). The
analysis of the personal markers as *-m, *-t, *-c is Bergsland's,
who just wrote that "a reasonably careful analysis" leads to that,
only till his death he never wrote how. I can show how for *-t and *-
c, while for the first person I can arrive only at "some labial" (*-
k of the singular).

> The only grammatical evidence I'm aware of, Eskimo 2pl. -ci, rather
> suggests to me that Eskimo /c/ somehow reflects 2nd. person *-t,

at least

> in the special circumstances of the 2pl. ending, where we would

expect

> something like intransitive *-d-t-k (plural - 2nd person - stative

*k),

No, no, the enclitic personal pronouns have their stem material
*before* the person + number markings. They are plainly identical
with the Aleut free pronouns reduced to their final syllables. Thus
Esk. -ci equals Aleut txici (txicix by influence from the dual); the
2sg ending in Esk. is variously -t&n or -k&n, matching Aleut txin.
Seefloth sees an "intransitive" /k/ in the 1sg possessives, but that
is only the personal marker which is /k/ here.

> transitive sg.obj. *-t-d (2nd person - plural),

You mean "2pl possessive of a sg. noun (inergative)". Yes, that is *-
ci (not causing gemination) from //-t-d//, lenited to *-zi after a
vowel.

> transitive pl. obj.
> *-dj-t-d (plural oblique - 2nd person - plural),

You mean "2pl possessive of a plural noun (inergative)". That is
invariant *-ci (causing gemination, not itself lenited) from //-d-t-
d//.

> which all three merged as
> -ci.

Yes, more or less.

> In any case, I don't see an /s/ there.

Right, this is underlying //t//.

> If the Eskimo reflexive is from *-ti or something similar, it

bears a

> striking resemblance to the Anatolian reflexive *-ti(a).

The Esk. refl. is from //-c// within its own set of rules. The main
surface form of that element is [s], but some dialects have a
palatal assibilated dental affricate (I suppose Serbo-Croatian <c´>
with one stroke or Polish <ci> before a vowel will do). It is turned
into a palatal nasal when word-final, just as 2sg *-t becomes -n,
and Aleut has 1sg -ng from *-k. The actual reflex of the resulting *-
ñ is Proto-Eskimo-Aleut *-ni, but it works morphophonemically like a
(single) consonant in causing gemination.

Now, I observe underlying *-k/-m *-t, *-c in Esk.-Al., and *-m-, *-t-
, *-s- in IE, for 1st person, 2nd person and reflexive, all quite
parallel and differing structurally from the 3.person. Moreover, I
am being lectured to learn about a structural inequality with the
3rd person in Uralic as well. I find it very hard to disregard this,
and it turns me distinctly towards the identification of the IE
reflexive as an old reflexive, not demonstrative or anaphoric.

Jens