Algonquian Parallel? Muskogean Parallel?

[Pamela Munro]

Hi, John and Siouanists,

I hope we aren't getting to parochially Muskogean here....

It's debatable how completely segmentable the dative element is from the
"III" paradigm, for two reasons (A and B below; consider also C).

A. In all but first person singular, the forms are purely "II" class + dative
im- (with the nasal assimilating to a following stop and coalescing with the
preceding vowel before a fricative). All the II prefixes end with a vowel,
but it is completely regular for an i to delete following a vowel in this
environment.

The problem is first person singular. The II prefix is sa-; the III + dative
prefix is am- (i.e. a- + im-), not sam- (sa- + im-). Immediately this looks
as though we have a different series here, which is the same in all persons
but 1sg.

The situation is a bit more complicated in a few languages, such as
Chickasaw, which provide evidence that 1sg III am- really should be seen as
underlying /sam-/ at some level: after certain preceding prefixes (I class
2sg. ish-, I class 2pl hash-, or "hypothetical" ik-) the 1sg III + dative
prefix is sam-:

Am-pilachi. 'He sends it to me'
1sIII.dat-send

Ik-sam-pila'ch-o. 'He doesn't send it to me'
hyp-1sIII.dat-send-neg

Is-sam-pilachi. 'You send it to me'
2sI-1sIII.dat-send
(just like 1sII sa-, 1sIII.dat sam- triggers a change of 2sI ish- to is-)

So we might like to say this really was the 1sII prefix sa- and that the
initial s- just drops in initial position for some reason.

However, this would have to be a truly ad hoc rule: there is no other
invironment in which initial s- drops. (In particular, initial s- never drops
from the 1sII prefix, though see C below.)

B. Another difference between the II class prefixes and the III+dative
prefixes is positional/morphophonological. To me these arguments seem fairly
clearly to show that the prefixes that appear with the dative cannot be
synchronically the same as the II prefixes. (Some of this argumentation may
appear in Charles Ulrich's dissertation on Choctaw.)

i. In Chickasaw (I won't try here to review the complex facts for other
languages) the III+dative prefixes precede the segmentable a- at the
beginning of verbs; the II prefixes follow this a- (which below I will gloss
simply as a; complex issue):

Am-a-pila. 'He helps her for me'
1sIII.dat-a-help

A-sa-pila. 'He helps me'
a-1sII-help

ii. The III+dative is phonologically a clitic, or at least is outside the
domain of the rhythmic lengthening rule; the II prefix is within the domain
of rhythmic lengthening. In these examples I'll put a # following lengthened
vowels:

Amapi#la.

Asa#pila.

cf. Api#la. 'He helps her'

This rule (described extensively elsewhere) lengthens short nonfinal vowel in
even numbered open syllables within a morphologically specified domain.

C. Despite the fact that we don't see a- as a 1sg prefix except in the
III+dative markers within the strict domain of agreement, there is one other
environment where this appears -- fossilized vocatives of a few kinship
terms, such as a-ppo'si' 'granny'. (Otherwise we'd expect II here, e.g.
sa-ppo'si' 'my grandmother'.)

All in all, I don't regard the "III" concept as "notional"! But not everyone
would agree with me....

Pam



Koontz John E wrote:

> On Wed, 2 Oct 2002, Pamela Munro wrote:
> > I'm very interested indeed by these thoughts about the Siouan "III"
> > series. I had not thought about this this way, and it's interesting.
>
> The three-way contrast of pronominals presumably works for Muskogean
> because the marking of dative is wholely confined to the object
> pronominal.  I seem to recall that it has something to do with the nasal,
> but it has been a while since I read anything relevant.  If the marking
> can be fairly transparently factored off and the remaining pronominal
> element matches the II series, then maybe the III series is still somewhat
> notional.  But I don't recall if this works.
>
> In OP, dative marking affects the wa-locative-pronominal string in
> idiosyncratic ways, depending on the locative (or absence thereof), but,
> in the absence of a locative, it affects both pronominals (agent and
> patient, or I and II).  So, while I tended to think in terms of agent,
> patient, and dative pronominals initially, I ended up resolving not to,
> feeling that in the OP case that approach obscured matters mroe than it
> helped them.
>
> The situation in Dakotan is materially simpler, but complicated in a
> different way (from my point of view) by the apparent swapping of the
> dative and suus paradigms (relative to Dhegiha).  Still, I think it makes
> more sense there, too, to think in terms of dative as separate from the
> pronominals, though I'd have to review matters to be sure!