Bipartite structure

[Koontz John E]

On Mon, 7 Jan 2002 Zylogy at aol.com wrote:
> Thank you, John. Just looking at what you wrote about the semantic ranges of
> the outer and inner instrumental sets (though a rather short list, so hard to
> prove in any statistical sense) leads me, impressionistically, to posit that
> perhaps the inner set is more "about" direct actions of body versus the
> mediated or body-external actions encoded prototypically by the outer set
> ...

Yes, there does seem to be a semantic difference, but I think this arises
naturally from the probable syntax of the morphologized constructions.
Of course, this is one of those situations in which in deductive terms
there is a circularity in the reasoning.  I think, however, that the
reasoning involved is inductive, or, putting it another way, that a model
can be proposed based on the whole situation and validated by repeated
successful application here and elsewhere.  It would also help, of course,
if cognates of the outer instrumentals were available.  My recollection is
that this is not the case as yet.

Note that the basis of the terminology inner vs. outer instrumentals (my
own coinage) is simply whether the instrumental morpheme is inside or
outside the pronominal slot(s) relative to the underlying stem.  The outer
instrumentals behave (morpho)syntactically like various other strictly
"outer" preverbal morphemes in Siouan.  These can also be termed preverbs,
though that also covers the locatives, which are movable preverbs (movable
relative to the pronominal slot) as opposed to fixed preverbs.  Another
possible term would be proclitics, or, as I said, the main verb stem and
its inflection can be seen as enclitic to the preverbal "base."

Other fixed preverbs or proclitics are various incorporated nouns and
demonstratives, and also the governed verb in causatives.  There are also
preverbs that are simply fixed parts of certain stems and don't have any
clear etymology, though many of these must be old incorporated nouns.

We probably want to call the syntactic (in the primitive sense of
ordering, etc.) behavior of preverbs morphosyntactic, but I think that the
description of Mississippi Valley Siouan morphosyntaxes can be
considerably simplified if one treats preverbs, incorporation,
compounding, causatives and other dependent verbal constructions,
enclitics, etc., at a higher level than matters like the relative ordering
of pronominals, locatives, reflexives, reciprocals, datives, reflexive
possessives, and inner instrumentals.  These latter are all much simpler
to deal with, in terms of mutual ordering and pronominalization (pronoun
placement), if we consider them only within the component units of the
compositions produced by compounding, adding preverbs, etc.

So, essentially, there are two levels of morphosyntax:  a fairly simple
one that accounts for the insides of inflected verb stems and noun stems,
and manipulates pronominals, locatives, inner instrumentals, reflexives,
etc., and a more elaborate one that manipulates preverbs, compounding,
enclitics, etc.  This latter system generates words, but not all of these
words are lexical entities, though some are.  In addition this latter
system generates trees that ramify on both the right and left, i.e., they
have a constituent structure of sorts, whereas the inner level structure
ramifies only to the left - it simply adds new morphemes to the front of
the whole, if we overlook the issue of the locatives.

Looking at things this way does mean considering that the two classes of
instrumental cannot be regarded as a single kind of thing, at least
morphosyntactically.

One other thing to note is that inner istrumentals typically have complex
morphophonemic interactions with preceding pronominals, whereas outer
instrumentals do not.

> In Yahgan, the general trend is for instrument prefixes to be themselves
> breakable into a generic initial (gross force source name- strike, squeeze,
> pull, tread, weight, blow, throw, etc.) followed by a specifier (a spatial
> distributive- into multiple bits, short or long lengths, slabs, spheres,
> etc.). I suppose one might call these latter incorporated shape classifiers
> or some such- they are apparently state-naming.

The closest thing to this in Siouan that I can think of are the
Mississippi Valley aorist/inceptive/iterative auxiliaries (moribund in
Dakotan) - the "suddenly and repeatedly" auxiliaries - which are compounds
of motion verbs (path elements) and positional verbs (shape elements).
There are large sets of these in Omaha-Ponca, Ioway-Otoe, and Winnebago,
it appears.  They carry the overall notion that the action occurs suddenly
or starts to occur, but the path and shape elements have to harmonize
appropriately with the action and its patient.

> I don't get any direct sense that there is anything like an inner/outer
> instrumental split -is there some relation having to do with control??

I hope I've clarified the basis of the distinction.

> Reanalysis of nom-postp stem to nom postp-stem feels right- does this
> possibly mean that the parent language was originally dep-marked?

Maybe not.  I may have overstated this if I called it reanalysis.  Maybe
it would be more appropriate to simply call it analysis.  Think of the
element that we've called the postposition as a directional adverb.  If it
fused to the noun, it can be characterized as an adposition.  If it fuses
to the verb, it can be characterized as a locative.  Probably it will be
closer to the verb than the nominal argument, and on the same side,
depending on whether the language tends to place sattelites before or
after the verb.

> ... since I've been promoting the notion (yet to be proven) that these
> constructions originate in something like serialization. More, please!

I suspect everyone is off at the SSILA/LSA meetings, or somebody with a
better memory would have cited the paper that provides Catawba verbal
cognates for the Siouan (inner) instrumentals.

At a previous SSILA/LSA I provided a serial verb explanation of the
Mississippi Valley reciprocal/in the middle prefixes in terms of a
commitative coverb.  Several of the MV languages still have cognate or
non-cognate commitative coverbs.  Crow, Hidatsa, and Mandan all have
benefactive coverbs based on 'give'.  All of the MV languages have
extensive compounding of verbs of motions.

JEK