ordering of person markers

[Koontz John E]

On Mon, 30 Sep 2002, Shannon West wrote:
> I know that this is the diachronic explanation, but that doesn't offer a
> synchronic explanation. The speakers of the language don't know the
> historical background, and can't use that to figure out which positions
> these affixes appear.

I'm wondering why they can't use the learned output of the historical
process to decide what order things go in?  Eric Hamp is said to be
exceptional in this regard, but I'm pretty sure most of us just learn how
it works from the way it is in our own lifetimes.  Including literary
exposure to earlier models, of course.  The models are somewhat fuzzy, of
course, because we are exposed to a population of varying examples, not
one uniform model.  I know for a fact that my modifier order parameter
tends to flicker among several states.

> Your diachronic explanation would have to be fitted with a templatic
> account. I don't so much mind templates, but they're not exactly
> popular right now, if you know what I mean.

Templates don't work very well with Siouan languages, I think.
Omaha-Ponca may be a bit fuller of nasty subcases and exceptions than
Dakota, but it isn't too different in general nature.  I think it's more a
a situation with a set of morphosyntactic rules for adding given affixes,
some of which conflict.  In a general sort of way the derivational rules
go before the pronominalization rules, but I know of a horrendous class of
exceptions involving datives in Omaha-Ponca.  (I mentioned this in passing
this weekend.)  The data in Dakota are substantially simpler, but my
recollection is that comparable cases occur there.  There are at least
cases of ki merging with multiple pronouns, if not with pronouns a slot or
two away from it.

> Right now, the general feeling is that there should be rules that the
> learner can use to figure out where things go rather than templates to
> learn.

Templates are certainly a kind of rule, but maybe you mean rules of a more
fashionable sort?  Or are we using template in different ways?  I'm
thinking in terms of templates as lists of position classes:  knee-bones
go before thigh bones go before pelvises, etc.

The problem with this approach in Omaha-Ponca is that while first and
second person pronouns generally go after outer instrumentals and
similarly located "preverbs"  they also go before reflexive/reciprocals
and reflexive-possessives, and these are perfectly capable of being placed
in front of preverbs, with the result that you get PRO(1,2) > REFL >
PREVERB instead of PREVERB > PRO(1, 2), even though combining rules like
PREVERB > PRO(1,2) and PRO(1,2) > REFL might lead you to expect PREVERB >
PRO(1.2) > REFL.  Of course, you actually get both orders, depending on
the stem.

Another kind of problem:  all pronouns of the form V (reg A1, reg P1, A12)
pop out of position and plop down in from of an initial wa, if they aren't
too deeply buried in the stem, e.g., a-wa-naN?aN, but wa-z^u=a-he (? from
memory). (And wa-dha-naN?aN and wa-b-dhathe.) There's certainly a general,
rule-based principle involved, but it's a mix of phonology and
morphosyntax and has no (direct) cross-linguistic generality.

Yet another kind of problem:  aN A12 goes before the locatives a and u
(*o) and after i, while wa P3p goes before a and i, but after u (not
counting fossilized u'- < *wa-o).  And wa- ~ wa...a- P12 goes before a and
u and around i (and becomes a-wa- with the causative).  This mostly makes
historical sense - not all of it - and doesn't seem to depend on any
general principle at all, except possibly that the perversity of OP
morphology tends to a maximum.