Bipartite structure

[Koontz John E]

On Sun, 6 Jan 2002 Zylogy at aol.com wrote:
> Hi, folks. Happy 2002. Just bubbling up from lurker status to ask whether
> anyone has any ideas about the historical origins of the Siouan instrument
> and location affixes. ...

> I'm wondering whether Siouan represents some sort of transitional phase,
> given that the instr. is prefixing and the loc. terms are so broad and few in
> number and also come immediately before the lexical root (so perhaps are the
> last gasps of an older system). ...


The structures I'm aware of are:

nominals PRO1-LOC-PRO2-LLS
nominals OuterInstrumental-PRO3-LLS
nominals PRO4-InnerInstrumental-LLS

where LLS is "lower level stem" (in derivational terms), and PRO is
pronominal with the numbers appended for convenience below.  LOC is
locative (which includes the concept 'by means of; using').

PRO1 is generally the inclusive and any third personal plural or
detransitivizer, though the Dakotan third person plural follows the
LOC slot, and the inclusive follows some of the LOC forms in Dhegiha.  For
that matter, all pronominals of the form V precede wa in Dhegiha, if a wa
(third person plural or detransitivizer) occurs, except, e.g., with
causatives.

PRO2 is the first person and second person.  I usually imagine the
zero-form third person being here, too, but ... well, it's a zero form!

PRO3 and PRO4 are all pronominals in the patterns in question.  Outer and
inner instrumentals are defined in terms of the location of the
pronominals.  Outer instrumentals behave as proclitics, or, rather, the
inflected stem behaves as an enclitic (postclitic) of the outer
instrumental.

Some outer instrumental stems behave as a single chunk
OuterInstrumental-LLS when further derived (e.g., with a reflexive) in at
least Omaha-Ponca in Dhegiha.

I'd guess that locatives are historically a case of

nominal postposition stem

being handled in word formation as

nominal postposition-stem

instead of

nominal-postposition stem

I think this is probably a general tendency in head-marking languages.
Rankin argues that the inclusive and some third person plurals remain
before the locative because they originate in incorporated nominals
("man", "person") at a date subsequent to the fossilization of the
locative construction.

I'd guess that most of the outer instrumental stems are cases of

generic-nominal stem

being reformulated as

NOM-stem

That is, of noun incorporation.  The inflectional slot remains where it
was, following the stem. The outer instrumentals usually have the semantic
domains 'heat', 'spontaneous' (derived from 'heat'?), 'by shooting', and
'with a blade/by cutting'.

The explanation that has been offered with the inner instrumentals, based
on evidence from Catawba (I'm blanking the name of the author of the
paper) is that they are old verb stems, implying that the constructions
were originally something like:

nominals instrumental-verb resulting-state-verb

Both verbs were inflected, or perhaps the first was a sort of participle.
However, the modern constructions have the inflection only on the
instrumental component, or, putting it another way, preceding the stem
composed of the instrumental prefix and lower level root.

The semantic domains of the inner instrumentals are 'pushing', 'pressing',
'with the hand/pulling', 'with the mouth', 'with the foot', 'by striking'.
The latter is usually identical to 'by wind or water' in Mississippi
Valley Siouan.

Dakotan converts 'by foot' into an outer instrumental, apparently because
it became homophonous with 'by heat/spontaneously'.

I should say that the set of instrumentals is essentially fixed across
Siouan.  There is a particular set that seems to exist in all the
languages.  Extensions to it are vanishingly rare, as are omissions.  I
seem to recall one extension in Crow, ha-, but I forget what it does.
Ioway-Otoe seems to have split *ru- 'by hand' into ru- and ri-, which,
are, I think - but check - 'by hand' and 'by pulling'.  What does happen
commonly is that instrumentals may be more or less productive, and they
may extend their meanings in subtle ways.

The stems that can take instrumentals are generally roots.  Locatives can
precede stems of more general form, or, putting it another way, locatives
stay outside the lower level stem, or can't be submerged in lower level
stems.  There are some compound locatives, cf. Dakotan iyo- and iya-,
which have parallels throughout Mississippi Valley Siouan.  Winnebago
seems to have variants roo- and hiro- for the first.  The inflection of
these can be rather complex.  Dhegiha seems to have -iu- and udhu-
variants of the first, depending on the person.  The developments of
*i-r-a- are even odder.