Personal Pronouns

[Jens Elmegaard Rasmussen]

On Tue, 27 Apr 1999, Patrick C. Ryan wrote:

> [...]
> (JER:)
> >Even so, the non-occurrence of the very FORM *mwe (in any
> >function) does look as strong evidence for a rule *mw- > *m-.

> Not strong! Not evidence! This seems completely illogical to me.

But you can't deny the existence of *te and *se along with *twe and *swe,
a variation for which there seems to be no tangible reason. Nor can one
deny that *me *te *se look parallel (and inflect in very parallel
fashion). Is the non-occurrence of a **mwe to join the w-forms *two *swe
then not a thing to be noted and explained? What if we do have rules to
explain it - isn't it then worth talking about?

[On Gk. mo:^mar : amu:'mo:n as reflecting *mwoH-/*muH-:]

> Irrelevant to the question of the pronominal form *mwe but, in any case, why
> not *mouH-/*muH?

Because the Gk. full-grade form is not **mow(V)-, bot /mo:-/.

[On the "thematic vowel" e/o:]

> >>I am also not aware of any IE who has successfully asserted that apophony
> >>is governed by specific consonants.
> >This part of it is, whether assertion to this effect proves successful or
> >not.

> Unless "assertion to this effect (does) proves successful", I feel
> comfortable in not considering it a viable factor.

It belongs to the generally accepted descriptive facts of IE that the
"thematic vowel" alternations in its own fashion. Saussure recognized it
and suggested a rule, Hirt spent half a lifetime working on it, Kurylowicz
suggested several explanations. At least in the verb, if a stem-final
vowel is followed by /m, nt, y, r, w/, it has the shape /o/, whereas if
/t, s, H2/ or zero follows, it turns up as /e/. In pronouns we find the
same: *to-m, *to-y, *to-r 'there', likewise before /d/ in *to-d 'that' and
before a vowel in *to-e > *to: 'those two', but /e/ before /s/ in *te-syo,
*te-smo:y and before /H2/ in *te-H2 > *ta-H2 'those things'. In the noun
we mostly find *-o- generalized, but here too we have remains of regular
/e/ before zero in voc. *-e and fem./coll. *-e-H2 > *-a-H2. - It is often
claimed that the "thematic vowel" is a latecomer to IE wordforms, since
thematiuc forms show more than one full vowel in a word, but that simply
cannot be the reason, for, if it were, OLD vowels would also have been hit
by the alternation rules that plainly work _only_ in the thematic vowels.
Therefore, despite the immense productivity of thematic stems, their
original form must go very far back in the prehistory of PIE - back to a
time when the "thematic vowel" had a phonetic shape that could exempt it
from the usual havoc played by the accent and instead make it susceptible
to influence from the following segment which other vowels are not. - BTW,
if it matter to you if such an assertion were successful, you can MAKE it
successful by taking it seriously - I trust you will once you really check
with the facts. Look at Greek, Germanic and Celtic, then you can't miss
the rule - and from the basis you thus define you can then derive any of
the lesser clear languages with no force at all.

> >[... Further on the length of nominatives:]
> >The nominative lengthening also works on stops: *ne'po:t-s,
> *wo:'{kw}-s.

> Why can you not accept that *ne'po:t-s is the result of  *nepoH-t-s?

Because other forms, esp. derivatives, show that the suffix was -Vt-, not
-VHt-, e.g. the fem. Skt. napti:-, Lat. neptis. And even if it were
*-oHt-s, then other words behaving just the same certainly had no
laryngeal; what about 'foot'?

> >But what is the probability that morphological differentiation not based
> >in phonetic change gets to LOOK so much like the result of phonetic change
> >that its variation can be stated in terms of consistent phonetic rules?

> That is a tough one. I have not ready answer.

I'd say that is decisive, for such is the situation at hand. And pointing
out such case of morphological variation LOOKING exactly like the results
of consistent phonetic change by specifiable rules is ALL we can do in
internal reconstruction. Our task is here to specify how the rules ought
to look in case they constitute the reason for the variation - for then we
have something to check next time we find a relevant piece of the puzzle.
May I add that most, if not all of my pre-PIE rules were originally
formulated on the basis of only a subset of the observations for which I
have later found them to supply regularity.

> Why do you not give us your best
> arguments for proposing a non-Hittite <m/w> alternation *outside of the
> pronoun series*?

I am not sure there was such an alternation elsewhere. I have found two
cases where in-depth analysis leads me to postulate *-G-m- (G being the
dual marker, I suggest a voiced velar fricative, but do not insist on it)
as an older form of what I find surfacing as *-w- or *-H3w-. In one case
the *-m- is the 1st person marker, in the other it is the marker of the
accusative. Since I cannot believe that the 1st person and the accusative
was one semantic entity, the homonymy must be accidental, so that the
covariation can only be due to real phonetic change, i.e. a sound law
*-Gm- > *-Gw- (~ *-w-). Both of these involve pronouns where we get to
look into some very distant older periods of the language. Of course one
would then have liked to find the same change of m to w when a
suffix-initial /m/ is added to a root-final /H3/, but we find no such
thing, e.g. Lat. no:men. I suspect the answer to this lies in the
chronology: the two sets of observations are ages apart.

[...JER:]
> >I was saying that, in the personal pronouns, the oblique cases are all
> >built on the acc.: Skt. dat. asma-bhyam, abl. asma-t, loc. asme /asma-y/
> >[...]

> Well, does this not suggest the primacy of the accusative? before the
> addition of -m to designate animate accusatives?

It spells primacy of the acc. over other non-nominative cases in the
system implemented by IE for the personal pronouns.

> >>All I was saying is that most IEists do not believe (and I agree) that the
> >>dual is as old as the singular and plural in IE.

> >[JER: but it is not formed by productive components]

> [PCR: ...]the dual forms
> incorporate -y, which is not specifically dual but only differentiating.
> [...]

Your task is to demonstrate the youth of the dual. You could do this by
showing us that its forms are derived according to rules of younger
periods than those of the plural. In that case the dual should be more
directly transparent than the plural. If anything, the dual is more
_opaque_ than the plural. If the y's of some dual forms are there to
differentiate, then demonstrate that such is their business elsewhere and
that they have been implemented by the pertinent rules.

[...]

> >>I do not assume that the basal form is *tu(:), and so cannot justify
> >>migrating <u>'s.

> >Then why not change your assumption about 'thou' and get the benefits?

> If this were a friendly drinking bout, I would be accomodating and agree.
> But are we not both trying to approximate the truth as closely as we are
> able?

I would be serious even over a beer. The "benefits" I'm talking about
comprise the possibility to explain more in a coherent and principled way,
in general experience no bad measure for closeness to the truth, if not
without its pitfalls.

[...]

Jens