--- In
cybalist@yahoogroups.com, "tgpedersen" <tgpedersen@...> wrote:
>
> The *-i of the primary endings is not a hic-et-nunc particle, but
just
> a hic particle, ie. it is the locative case ending. That's because
> what we think of as finite verbs in the secondary endings are
actually
> verbal nouns; the secondary endings *-m, *-s, *-t are 'at
me', 'there'
> (< *so), 'there' (< *to), respectively (now we know why the two last
> ones are somtimes confused. The subordinate clauses, in which the
> secondary endings appear, are therefore non-finite clauses with
verbal
> nouns (cf. Finno-Ugric languages).
I am sure the verbal stems are indeed older verbal nouns, but not the
kind you seem to prefer. They are in my analysis rather plainly agent
nouns: *gWhen- does not mean 'murder', but it does mean 'killer',
Ved. vr.tra-hán- 'killer of vrtra'.
If one makes the stem of root presents or root aorists an agent noun,
one gets a relatively smooth syntax: *gWhén-m is then 'I (am) a
killer', and *gWhén-s (with *-s from word-final **-t) becomes 'you
(are) a killer'.
What is more, the 1du *gWhn-wé will then be *gWhen-G-mé-G (with a
velar spirant G (identical with /H3/) marking dual number), meaning
originally 'we two (are)two killers'. The loss of the dual marker
after vowel in word-final position is as with the animate dual marker
*-e; and the change of the underlying *-m- to *-w- in contact with the
old dual marker is as with the dual of pronouns where we have
*nH3wé 'us to' as opposed ot *nsmé 'we (pl.)'. One will have to assume
loss of the first -G- in the long verbal form, presumably at a stage
*gWhnGwéG. The corresponding plural (using D as plural marker, just to
call it something) would be *gWhen-D-mé-D; we find *gWhnmé, which
rather points to *gWhen-D-mé-G with a generalized dual marker; here
too one would have to assume loss of the number marker, presumably in
a preform *gWhnDméG, but with no change of the *-m- because it is not
contiguous with a /G/.
The second person dual should be *gWhen-G-té-G 'you two (are) two
killers', whence, via *gWhnGtéG, one expects PIE *gWhnté. The actual
form *gWhntóm has some added material, but is otherwise fully in
accordance with this. The 2. person plural would be *gWhen-D-té-D, and
the actual PIE form *gWhnté again points rather to *gWhen-D-té-G with
a generalized dual marker in the final segment.
The surprising dual marking in the final part of the endings seems to
have an exact match in Uralic. Finnish has 1.2.pl. luemme, luette 'we,
you read'. The protoforms of these used ot be posited as *luke-k-me-k,
*luke-k-te-k with a "present-forming -k-", but this has been abandoned
in favour of an analysis *luke-t-me-k, *luke-t-te-k with plural
markers on the verbal stem, and a different number marker on the
ending proper. I deed it close to obvious that an old dual morpheme is
at work in the endings of Uralic and IE alike, and that a common
prestage of the two branches changed a set *-G-me-G, *-G-te-G, *-D-me-
D, *-D-te-D into *-G-me-G, *-G-te-G, *-D-me-G, *-D-te-G. This would
constitute a shared innovation covering IE and Uralic.
The third persons rather appear to contain an overtly marked agent
noun in IE. Since the passive participles have *-to- or *-no-, the
agent nouns of which they are adjectival derivatives will be t-stems
or n-stems, and so they are: Ved. krtá- 'done' from *kWr-tó-s is
properly an adjective of belonging made from the stem seen in Ved. su-
kr.´t- 'gut handelnd', *kWr-tó- meaning properly 'belonging to the
doer, done'; likewise Ved. anná- 'eaten' from *H1(e)d-nó- is based on
the agent noun seen in Latin edo:, -o:nis 'Fresser' or OHG ezzo, acc.
ezzon 'eater', IE *H1(e)d-nó- meaning then, in parallel fashion,
originally 'belonging to the eater, eaten'. It seems obvious to
combine the two allomorphs -t- and -n- into a common *-nt-, which is
frequently seen as a variant of n-stems, and is of course used
productive as the active participle in *-(e)nt-. It will then also be
inviting to see a link to the 3. plural ending *-ent-, which may then
represent some form of this followed by an original plural marker.
Thus, the third persons will be overtly marked agent nouns in an
unmarked singular and a marked plural form.
Now, if the verbal noun involved is an agent noun rather than an
action noun, the marker *-i of the primary endings cannot well be a
locative: A killer who is active now is not "in a killer", he just is
a killer, and he is it 'now' or 'here'. There is no basic grammatical
difference in IE between place and time: Ved. dyávi 'in the sky',
dyávi-dyavi 'in each day, day by day'. The form *-i is in my analysis
simply the stem of the pronoun whose stem is either *e-/*o- or *H1e-
/**H1o-. The enclitic form of this would be *i or *H1i, cf enclitic
*im 'him', *id 'it' in a number of languages. Thus the *-i (or *-H1i)
may be itself just an endingless locative meaning 'in it, there'.
One thing I do not understand is why the accented variant, *é or *H1é,
is used as the augment in the meaning 'then' with distinct remote
deixis, while the presumed enclitic (and it never is accented) *-(H1)i
rather means 'now'. Are the two not identical? Or can the difference
in word order explain any part of this? Any ideas?
Jens