No Proto-Celtic?

[Thomas McFadden]

> To us it may look strange that the pronoun would have come after the verb and
> not before, but this may happen, and it is standard in modern Irish, isn't
> it?  E.g.: Scri'obhann SE' litir (he writes a letter).

but thats because Irish is VSO, i.e. subjects in general follow the
verb, whether they are pronouns or full NPs.  what you seem to be arguing
for here is that in PIE full NP subjects preceded the verb, while prnonoun
subjects followed it.  that's a very different thing, and would be rather
surprising, i think.  again, does anyone know of such a lanuage that is
actually attested?  pointing out (as you do below) that languages commonly
have postpositions on nouns is sort of orthogonal to the question.  of
course the fact remains that the personal endings
look a heck of a lot like the pronouns, so maybe the idea will turn out to
be right after all, i just think it should be regarded with
skepticism.  one alternative possibility (and im just making this up off
the top of my head.  i wont claim that its a priori more plausible than
the NP-V / V-PRO hypothesis, just that its another way to get out of the
problem) is that the agreement affixes started out as clitics on the
beginning of the verb, first became prefixes and then ended up as suffixes
because suffixes were generally preferred in the language.  just a
possibility.  it just seems to me that the key to the mismatch between
preverbal subjects and postverbal agreement affixes is more likely to have
arisen through the weirdnesses of affixal morphology (especially the
general preference for suffixes) than through an extremely marked
prehistoric word order.  or we can of course consider again the idea
that Pre-PIE was VSO at the stage when its agreement affixes
developed.

or its possible that PIE really was SOV, but had the type
of freedom in its word orders that we've been discussing for Latin
etc., and the affixes actually developed from those sentence types
where the subject followed the verb, again because suffixes are
preferred, so affixes couldn't develop in the unmarked orders.  this
is not as crazy as it might sound.  consider a language like German,
which arguably has a basic SOV order that is disrupted by
topicalization and V2 and other things (this argument works even if
you think that German is SVO based on the order in main clauses
instead of that in subordinate clauses).  in any case, you get a lot of
sentence types where an object or some other non-subject element precedes
the verb, and the subject follows it.  these orders have, arguably, had
their effect on some of the personal endings.  some have explained the
2sg. -st as being a resegmentation of forms like bistu as bist du < bis du
(although others have argued that the -st is from preterite-presents with
a dental final stem).  in the Bavarian dialects the 2pl. ending is -ts, as
opposed to -t in the standard language.  again this has two possible
explanations, one is that it is the preserved 2du. ending as in Goth. -ts
(wouldn't be surprising since in Bavarian the 2pl. pronoun is Nom. es/oes
Acc./Dat. enk, clearly the old dual).  the other possibility that has been
suggested is that the -ts is the inherited -t plus the cliticzed pronoun
's from es.  even if in both cases the other explanation turns out to be
right, it is clear that a variant that arises from cliticization in a
marked VS order in an otherwise SV language can be grammaticalized and
generalized to all orders.  again in Bavarian German (and a number of the
dialects and even to a certain extent in colloquial forms of the standard
language) the 1sg. pronoun in the nominative is mia, clitic ma both < mir
in place of standard wir.  this is transparently from the case where it
appears postverbally.  since the 1pl personal ending is -en, the following
wir assimilated to the nasal (and this pronunciation is common even in
fairly standard spoken forms when the pronoun is postverbal).  crucially,
this nasal-initial pronunciation was extended to all positions in the
relevant dialects, so that even sentence-initially this is what you get.

these are just a few possibilities.  it seems to me that the key to
the mismatch between preverbal subjects and postverbal agreement affixes
is more likely to have arisen through the weirdnesses of affixal
morphology (especially the general preference for suffixes) than through
an extremely marked prehistoric word order.  what we need is a language
where suffixal agreement markers are a more recent innovation, where maybe
we could actually trace their development.  any ideas?

Tom McFadden