Pers.pron.

[Jens Elmegaard Rasmussen]

[ moderator re-formatted ]

On Wed, 21 Apr 1999, Patrick C. Ryan wrote:

> [Personal pronouns in PIE; snip of a lot]

>>[... O]ne expects [*mwe beside *me]
>>in the light of the alternants *te/*twe and
>>*se/*swe, and that *mwe could not continue to live if it ever has.

> That is only if you assume that *te and *se developed from *twe and *swe,
> which I specifically deny. And, *te/*twe are not, IMHO, true "alternants";
> rather, they are the absolutive (*te) and w-inflection (*twe) of a
> pronominal root *te.

Where is the evidence for their differentiated use?
Does YOUR system not force YOU to assume that the "w-inflection" of *me
changed from *mwe to *me by the rule you won't have? That also answers the
following remark of yours:

> I specifically deny that [viz., that earlier *mwe is need under "my
> system"]. And doing the analysis your way is what I am, of
> course, questioning.

Everything should be questioned.

>>[...] IE itself combines the
>>nom. *-s with the thematic vowel to form *-os, but that of the gen. to form
>>*-es +-yo, which indicates that the two sibilants were not identical;

> I would be very interested to hear how you justify concluding that the "two
> sibilants were not identical". I find that surprising since we know only of
> one sibilant for IE. But if you believe IE had both <s>and <z>, why not say
> so?

I do say so, only in the variant, "IE had had both <s> and <z>", i.e. the
opposition belong to the prehistory of the protolanguage in which they
constitute two different morphophonemes for analyses of some depth. The
sibilant marking the nominative in IE had two properties not shared by
(most) other s's: (1) It causes lengthening of V in the environment
VC(C)s#, e.g. nom. *dye:w-s as opposed to 2sg *k^lew-s; (2) it causes a
preceding thematic vowel to take on the form /o/, while other s's take
/e/: nom. *yo-s 'he who' vs. 2sg *bhere-s (but *yo-m like 1sg inj.
*bhero-m and nom.pl. *to-y 'they' like opt. *bhero-yH1-t, etc.). The
lengthening is also exhibited by some s's that just happened to occur
root-finally (as all consonants could): *ters 'dryness' > *te:rs, reshaped
to *te:r-os (like *men-s -> *men-os 'thought'), whence Celtic *ti:ros >
OIr. ti:r 'land' (ntr. s-stem); also Ved. ma:s 'flesh' from *me:ms has
undergone the further lengthening leading to a form rhyming with m-stem
nominatives like Ved. ksa:s 'earth', Av. zii<ao> 'winter' - this must have
been different from the form of the acc.pl. which once must have ended in
stem + -m-s, in the o-stems *-o:ns, which therefore probably once had a
different sibilant from the one marking the nominative. Lengthening is
also seen in the s-aorist which then apparently had the lengthening s.
Now, the derivative present to go with the s-aorist appears too often to
be the type in -sk^e/o- for this to be fortuitous (the arrangement of Ved.
prcchati : aor. apra:ksam). Other derivative presents are formed with the
suffix *-ye/o- (*-e-H2-[ye/o-], stative aor. *-eH1- vs. stative prs.
*-H1-ye/o-, denom. in zero as opposed to denom. in *-ye/o- presumably
reflecting the two aspect stems, desid. or "future" in *-H1s- vs.
*-H1s-ye/o-, etc.), therefore one would like to derive *-sk^e/o- from
older *-s-ye/o-. A change of "lengthening s + /y/" to -sk^- may (by only
just MAY) be seen in the suffix *-isko-s "belonging to" which may simply
be an old izafet-like construction in *-is + *yos with the relative
pronoun tagged on to an old nom. in *-is (itself manifestly a reduced
variant of *-os): Thus a word like Danish would have meant "Dane who (is",
as indeed it does in "a Dane who is king" = "a Danish king". The (non-
lengthening) s of the desiderative (future) does not change a follwoing
/y/ into -k^-, cf. the cited desid. durative (Skt. future) in *-H1s-ye/o-.
It seems rather plain that IE /s/ has more than one source to it.

[...]

>>in addition, the gen. morpheme had a vowel (*-os) and formed a weak case,
>>while the nom.sg. ended in pure *-s (perhaps once voiced).

> I am suspicious of all "pure" items. It seems to me that reality is always
> slightly adulterated.

If so, they are adulterated in two different ways under the same
circumstances, and that spells opposition to an unbiased analyst, doesn't
it? If you say no to that, investigation stops, and discussion becomes
impossible. There is the open backdoor of "vanished variables". But one
should not refrain from saying out loud what one observes in places where
observations CAN be made, and that's all I'm doing.

>>Even so, however, we cannot exclude that they are _ultimately_ two different
>>variants of the same original entity.

> Yes, here we are in agreement. But for "cannot", I would subsititute "may
> not".

If you would state a reason for this will of yours I just might take it
seriously. BTW, the exact modality of "may not" escapes me in this
context.

>> - In the inflection of the IE pronouns the acc. plainly has a morpheme which
>>is absent in the nom. -

> And what plain morpheme is that if I may ask for clarification?

The accusative-forming segment *-me (dual variant *-we).

>>but all the weak cases are based on the accusative. In this, PIE has a system
>>differing from all other PIE inflection and looking more like Modern Indic or
>>Tocharian.

> Do not see this at all.

Then look at the inflection of the Vedic pers.prons. in your Grassmann.
They all go the same, and the simply add endings to the accusative - as it
is, or as you can esaily unravel it if you're willing.

>>This is not an ergative system, but there may or may not have been one
>>elsewhere in the morphology beside it.

> A quick look at  Lehmann's _Syntactic Typology_, one of his finest works,
> will inform you that there is no "magic bullet" for ergative systems, no
> universal pattern for marking. <0>for the absolutive is common but not
> diagnostic.

>>>Now I know some will quibble over whether or not the same logic should apply
>>>to pronouns, and yes, I am aware of what seems to be a more conservative
>>>retention of older inflections in the pronouns, but, based on the experience
>>>we have we languages around the world, there is, IMHO, absolutely no
>>>justification for separating nominal and pronominal developments absolutely.

>>In English you must: What's the "me-form" of _house_? What's the s-genitive
>>of _I_?

> I normally expand my view over more than English.

Then I misunderstood you - as you me: This was restricted to IE, and that
goes like English in the respect concerned.

>>> There is no necessity of reconstructing dual forms for earliest IE.

>>Then why are they there?

> You tell me.

What about this: To express dual number? Can't we accept the dual when we
SEE it?

>>Why does Old Germanic agree with Old Indic in this respect? By chance?

> You are conflating two issues. I maintain, with most IEists I presume, that
> dual inflection is later in time than singular and plural inflections, and
> *is built on them*.

That's not the way anything I know looks. On the contrary, the dual is
everywhere in the process of being eliminated, and it just beats me why
some scholars want that to reflect a stage of uncompleted creation. In
addition, the dual has some morphemes all its own, that's not at
all the way analogy works, but it IS the way archaisms look. You might as
well explain the irregular inflections of the verb "to be" in most IE
branches as due to new morphological trends that got cut off. That is very
definitely the wrong track, and I would not accept it even if there were
unanimity about it.

[...]

>>Since -mu is the enclitic of ammuk 'me' it has every likelihood of having
>>taken over the -u- from there; ammuk has it from the nom. /uk/, that in turn
>>from *tu (variant of *tu: which gave Anat. *ti: >Hitt. zi-k with -k from
>>'I').

> This is hopelessly muddled as far as I am concerned. Hittite ammuk is a
> stressed form for 'me' fairly certainly combining IE *e-, demonstrative +
> *me, 1st person + *-w, inflection (I believe its signficance is to mark
> topicality) + *g^-,  pronominal marker (I believe its original significance
> is to mark maleness).  It is backassward to suggest that -mu has "taken over
> the -u- from" ammuk. Rather, ammuk has incorporated mu into a fuller form
> expanded by a- and -k.

Nobody can _guess_ these things. But IF (1) Hittite is an IE language and
so prone to contain elements corresponding to those found in other IE
languages, (2) the vocalism /u/ is not found in these pronouns elsewhere
except in the nom.sg. 'thou', and (3) enclitic and orthotone forms are
sometimes found to influence each other, also in the direction I propose
(a well-known case among many is Germanic *mi:na-z 'my' from *meyn-o-s, a
reshaping of *emo-s based on the orthotone gen. *mene influenced by the
enclitic gen. *moy; also Lat. meu-s from *meyo-s has taken the /y/ from
the enclit. *moy), THEN you can link Hitt. to the rest of IE in a smooth
and unforced way if you assume that the /u/ has propagated from 'thou' to
'I' and from 'I' to 'me', and finally from orthotone 'me' to enclit. 'me'.
All these steps of spread then occur between closely associate forms. Is
that more "backassward" than an explanation based on a morphology the
language does not otherwise have? BTW, what is the basis for taking the
*-g^ to express topicality?

[...]
>>To strain your blood
>>pressure, I take the Skt. sma(:) 'verily' to be parallel with *nsme *usme,
>>only made from the reflexive plural, IE *sme from **sweD-me, through
>>invented stages like *sfeDme > *sfezme > *sfozme > *sphozme > *sphzme >
>>*sphme > acc. *sme 'the ones mentioned', of which it may be the instr.
>>*sme-H1 "per se".

> Invention is the mother of comedy.

Maybe so, but it is less amusing that some scholars - in this case
not you - claim that there is no way from A to B just because they have
not even tried to find one. I have shown how the IE system of pers.prons.
MAY be seen as underlyingly regular, you may have a different way of
achieving the same; but NOBODY should claim that there is NO conceivable
way this system could be originally regular.

My system, which was only a suggestion, may have its flaws, so don't stop
criticizing it, just because I have ready answers to the things I have
heard said about it over the years. This IS a dialectic business.

Jens