Proper methodology (was: RE: [tied] Re: Mother of all IE languages)

From: elmeras2000
Message: 28444
Date: 2003-12-13

--- In cybalist@yahoogroups.com, "Glen Gordon" <glengordon01@...>
wrote:
> Richard:
> >[...] Miguel argues in [...] that 'thou' and 'you' are themselves
cognate);
>
> And if you believe that I have some property to sell you in Moose
Jaw.
> The words *tu and *yus are different, unrelated words in most, if
not
> all, published texts concerning IE. Miguel's theory is solely his
own.

No, not quite, I side with him on that point. It is arrived at
exactly by the reasoning formulated by Trask in the quote you give.
It is true that *tu and *yus do not *look* particularly related. But
the consonant of *tu/*te looks might close to that of the verbal
ending *-te of the 2pl. And the consonant of the 1pl ending *-me
also looks quite close to that of the 1sg acc. *me, actually more so
than to that of any forms of the 1pl pronouns (I do not accept a
PIE "*mes"). Now, in the dual, *yu looks related to *yus, and the
verbal ending *-tom has the same consonant as *-te. This combines
all the 2nd person verbal endings with the singular pronoun.

In the first person dual we have -w- in the verb, secondary ending *-
we. The pronoun has *we(:) which fits the 1pl *wey (vayam, weis and
all that). So again, the dual and the plural go together in the
pronoun, and have the company of the verb in the dual.

Now, the 1pl vb. *-me is expected to be somehow related to the forms
with /w/ of the 1du of the verb, since there is such an agreement in
the second person. The dual /w/'s are not too hard, for a sound
change "dual marker" + /m/ > /w/ can be based on more than one
observation: (1) Where the verb has *-me in the plural and *-m in
the singular, the dual has *-we, facts making one suspicious of a
change from /w/ to /m/ caused by whatever one marked the dual; (2)
Where the plural accusatives end in *-me, the dual accusatives end
in *-we, a fact confirming the suspicion that a dual morpheme has in
fact changed /m/ to /w/. Since the two cases are not related,
the /m/ being the first person marker in (1), but the accusative
marker in (2), the process can only be one of pure phonetics.

That leads to the positing of number markers in the preforms of the
verbal endings *-we, *-me, and also in the accusatives in *-we, *-
me. Parallel number marking must have been present in the 2nd person
forms in *-t- of the 2.du/pl. and the corresponding pronouns.

Now, the 1du/pl pronouns *we, *wey both alternate with *n-, as
enclit. *noH3, *nos. Are we to posit /w/, /n/ and /m/ as completely
separate units? Are we even to posit two different morphemes /w/,
one seen in the pronouns, one in the dual verbal ending? I am told
that the /y/ of the 2du/pl nom. pronouns have nothing to do with the
verbal endings with -t-, so in consistency with that principle I
should keep the /w/ of the pronouns *we/*wey completely detached
from the verbal ending *-we. That looks very unsatisfactory because
the diverse allomorphs are just one feature away from each other, a
fact that would be even more in need of an explanation if for some
reason phonetic change is rejected.

Now, the /n/'s are easy: The orthotone variants of *noH3, *nos are
*nH3we, *nsme; and if /w/ can proceed from /m/ in the position after
the dual marker, they can come from *nH3me, *nsme. What we expect is
*mH3me, *msme, so the actual *nH3me > *nH3we, *nsme can be explained
by simple dissimilation. One may note the same dissimilation assumed
for the gen.sg. *mene from *meme, and the assured assimilation
changing Indo-Iranian *mana into Sanskrit mama.

We still need a link between the *w- of *we/*wey and the 1st person
marker -m-. They are too close to be separated, yet not fully
identifiable just yet.

If the 1pl acc. is **m-s-me, it is highly unexpected that the 2pl
acc. is *usme, although this fits the enclitic *wos quite fine. What
in heaven's name is a -w- doing here in the second person? Well,
come to think of it there are quite many w's or w-like features
around where we don't want to see them. If -w- is an allomorph of
the 1st person marker, why not only *wos/*usme (and the dual
*woH3/*uH3we), but also 2sg acc. *twe and refl. acc. *swe? And if
the reduced forms *te, *se have the fuller forms *twe, *swe beside
them, why does *me not have a full variant *mwe? Well, the last
point is easy, for there is no IE initial *mw-, so apparently there
*earlier was* an interchange *mwe/*me, only in this case both
alternants later became *me and os obliterated the alternation. That
also tallies with the possessives of the sg., 2sg *two-, refl. *swo-
, but 1sg *mo-, and the particularly eloquent evidence of the vrddhi
variants which are *tewo-, *sewo-, but 1sg *emo- (not **mewo-, since
the derivation was only made after the loss of /w/ in the cluster
*mw-).

Now, that moved a *-w- into the (1/)2/refl.sg.acc., the 1.du/pl.nom.
and the 2.du/pl.acc. Now the vowel -u- is too close for comfort. Are
we really to assume that there is no connetion between the vowel /u/
of *tu(:) and the /w/ of *twe? And are we to assume the same for the
vowel /u/ of *yu(:)s and the /w/ of *wos? (and the dual *yu :
*woH3?). And why does the accusative singular end in *-we and not in
*-me as in the plural? It will seem that we need an element in the
acc.sg. capable of changing /m/ into /w/. That again issues the call
for an element /w/ this time also to the assistance of the
nominatives with vocalic /-u(:)-/.

Come this far we observe the need for an element /w/ or the like in
the following forms: 2sg.nom., (1/)2/3sg.acc., 1du/pl.nom.,
2du/pl.nom., 2du./pl.acc. Now, which forms do *not* demand a /w/?
Few actually, and none with any weight: The 1sg.nom. is not in the
game at all and so does not count. The 1du./pl.acc. have labials
anyway and so may or may not have the /w/ in addition. That means
that the element /w/ is demanded by most forms and excluded by none.

Now what could the function be of an element present in all personal
pronouns? I know too few other languages to be able to circumscribe
the full potentials, but I do know of the case oof Aleut where all
personal pronouns consist of a stem txi- (ti- before velar) followe
by person+number markers. In the case of IE, the "stem" would be
located in second position, i.e. after the person markers. The
functional idea wuld be the same.

Now, if there *is* an element *-w- lurking in the original forms of
all the personal pronouns, some changes will have to be assumed, but
they turn out to be quite consistent if one takes the trouble to
spell them out.

If the acc. is to end in *-me as in the pl., (and dual *-H3-we from
older *-H3-me), then the *-we of *twe, *swe must be from older *-w-
me. That would allow us to derive these from *tew-mé, *sew-mé via
*tewé, *sewé and reduction to *twe, *swe. The 1sg acc. *me would
then be from a parallel *mew-mé which then also changed wm to w,
only in this case the resulting *mewé > *mwe later also reduced *mw-
and so became *me.

The 1 du/pl would be *m-w- + number markers. The dual is marked
by /H3/, the plural by -s or -y. I suggest a common source of the
latter in something like a dental spirant, so I posit, in crude
terms, dual *-G(-), plural *-D(-). The forms *we, *wey obviously
have their vowels after the /w/ and immediately before the number
marker. That leads to *mweG, *mweD. We have already assumed *wm > w
or the earliest layer, so *mw > w will not be surprising: We arrive
at *weG, *weD as the expected output forms. We know from the noun
that the dual marker simply vanishes after *-e, and the nom.pl. *to-
y indicates that some vowels take the pl. marker in the shape *-y
and not *-s. I therefore find *wé, *wéy just as good as Gk. patére,
and toí.

Now, the accusatives corresponding to *we, *wey are *nsmé, *nH3wé.
The n- has already been explained by dissimilation, but the full
underlying form should also have the -w- in second position, That
leads to *m-w-G-, *m-w-D-. Now, if we separate the -w- from the
initial, we reak havoc, for we need a cluster to make the -w-
disappear. We therefore have to posit the vowel after the -w-, i.e.
in the same position as the nominatives. It will seem that
dissimilation should operate first: *mweG-me, *mweD-me -> *nweG-me,
*nweD-me, whence again by reduction (nw > n) *neGme > *neGwe and
*neDme, by the ablaut then *nH3we, *nsme (with the pl. marker
surfacing as -s- as its general form). Note that the immediate
prestage would have the reduced vocalism -o-, so that the preforms
*noG-we, *nos-me may immediately explain the enclitics *noH3, *nos
as simple shortened forms.

If the 1du/pl acc. were *mweG-me, *mweD-me, the corresponding 2du/pl
acc. forms should be from *tweG-me, *tweD-me. We find forms
reflecting this, only not having the *t-: PIE *uH3wé, *usmé. [I
leave out the minor problem of the 1du.acc. which may have lost oone
of its elements on its way to Indo-Iranian /uva-/ whcih can be from
either *uH3e or *uwe. The former can be handled by dissimilatory
loss of one /w/, the latter by analogical vowel shortening.] The
question is, can *tw- be reduced to *we-, just as *-w-me and *nwe-
were reduced to *-we and *ne-? The existing *twe has already been
traced to a fuller form *tew-me, so there is no internal conflict in
the system. If there *is* such a gross reduction as *twe- > *we-, we
may also get at the nominatives in *yu-.

The principle of vocalization is clear up to a point: The
accusatives end in *-mé added to the corresponding nominatives. The
nominatives end in a single consonant. That would explain the lack
of reduction in *tu as opposed to the reduced initials of *yu/*yus
and *uH3we/*usme: the original structures would be *téw,
*twéG/*twéD, *tweG-mé/*tweD-mé. It will seem that the vowel of the
2nd person is -u- when accented, i.e. at some point *túw,
*twúG/*twúD, but *tweG-mé/*tweD-mé. I ascribe this to assimilation,
and lack of it in unaccented position to blurring (bleaching). I can
now offer a principled account of the gradation of the initial
consonant: *t- is retained before a vowel, but before /w/ it is
reduced to something that survives as *y-. This is quite a jump, and
I have no evidence for the intermediate stages; the shortest way
appears to be *tw > *dw > *Dw- with a spirant (the first step could
also have a voiceless dental spirant). Another possibility could be
located in the vowel, if there is a change to [ü] and subsequent
fronting from /w/ to /y/. Many roads lead to Rome.

So, if *t- is retained before /u/, *tú(:) is no problem. And if it
is reduced to *y- before -wú-, then *ywúG, *ywúD look the perfect
preforms of *yú(:), *yú(:)s, demanding only loss of /w/ before the
vowel /u/. And before the unchromatic vocalism of the acc. *tweG-mé,
*tweD-mé reduction of the *t- all the way to zero would not be
shocking. Perhaps we have to assume *yw- here too, and then a
bleeding-order split into *ywu- > *yu- vs. *ywe- > *we- depending on
the vowel.

Now, of *twéG/*twéD took on u-vocalism, why, one might ask, did the
first persons not do the same? The forms have been posited as *mweG,
*mweD, and why is there no assimilation here? The obvious answer is
that there was dissimilation, or rather a dissimilatory ban on
assimilating too much. That is all it takes to save the system.
Conversely, it takes a theory labelling this impossible to rule out
the system and the pervasive regularity it offers.

With these details in the specific development one may then posit
the following original system of personal pronouns for IE:

1sg.acc. *mew-mé
2sg.nom. *téw > *túw
2sg.acc. *tew-mé > *tewé > *twé
1du.nom. *mwéG > *wéG > *wé
1du.acc. *mweG-mé > *nweGwé > *noGwé > *nH3wé (*noH3)
2du.nom. *twéG > *twúG > *DwuG > *DuG > *yuG > *yú
2du.acc. *tweG-mé > *DweGwé > *weGwé > *woGwé > *uH3wé (*woH3)
1pl.nom. *mweD > *wéD > *wéy
1pl.acc. *mweD-mé > *nweDmé > *nosmé > *nsmé (*nos)
2pl.nom. *twéD > *twúD > *DwúD > *yús
2pl.acc. *tweD-mé > *DweDmé > *weDmé > *wosmé > *usmé (*wos)

All other cases are formed by adding endings to the accusative
forms. The possessives are thematic derivatives based on the stem of
teh accusatives, in part with vrddhi. A substantivized form of the
vrddhied variant is used as genitive case of the pronoun.

The syntacic function of the number markers is elliptical: "I and
others" = "we".


>
> Anyone can play phonetic games and come up with convenient
> "dissimilatory processes" to get rid of the *t that is "supposed"
to be
> there because of a groundless, preconceived notion. "Supposed" by
> whom? Not by me. Nor many others much more qualified than me.

The underlying notion is not groundless. The system itself is
consistent in so many ways, yet apparently fails to show it some of
the way. But things can change, and sometimes the changes can be
uncovered.

I do not suspect you of supposing all this. I have no ambition of
convincing anybody as qualified as you.

Jens