On Sun, 4 Jan 2004, Rory M Larson wrote:
> In OP, the patient wa- refers to (I think only animate) 'them', in about
> the same distribution as Dakotan wicha-. It also refers to 'us', which I
> think in Dakotan is handled with the same uN(k)- that is used for 'we'.
> The latter has a quirk in that awa-, rather than wa-, is used for 'us'
> in the causatives. In Dakotan, wa- is not used as an animate patient
> affixed pronoun in either 'them' or 'us' cases. Is this all correct?
This is correct as I understand it. But note that wa as 'us' (P12) is
wa-a- with i and dative -(g)i-, since with them we find wea- < wa-(g)i-a-.
> What is the career of the animate patient affixed pronoun use of wa- ?
> Is it general in Dhegihan, Chiwere and Winnebago, in both 'them' and
> 'us' usages? Is it possible that it has been spreading into these
> contexts in the same way as we suppose Dakotan wicha did?
Da OP IO Wi
A12 uN(k)- aN(g)- hiN- hiN-
P12 uN(k)- wa-a- wa-wa waNaNg-a-
O3p wic^ha- wa- wa- wa-
P3ns wa- wa- wa- wa-/waz^aN=
Ps3ns wic^ha- wa- ??? ???
I think the bimorphemic P12 forms are all capable of having morphemes like
the i-locative fall in the middle. I may have length wrong in some of the
wa's in some of the forms. The gist of the current discussion is that
productive animate P3ns are actually wic^ha- in Dakotan, too. In general
O3p marking occurs only with animates anyway. Use of wa- in some of these
contexts may be more or less moribund in some of the languages. Some of
Constantine's examples looked like they might have wic^ha- S3ns with
active verbs. (A = agent (active + transitive subject), P = patient
(stative subject + transitive object), O = (transitive) object, Ps =
possessive, 12 = inclusive, 3p = third plural, ns = nonspecific)
> Judging from the Dakotan examples given recently, I'd like to propose an
> alternative hypothesis that wa- was originally not a noun marker at all,
> but that it acted to generalize the action of the verb.
This is essentially what I'm arguing, except that I believe it still works
this way in OP and Dakota (to the extent that wic^ha hasn't stuck its oar
in).
> My prediction is that where both versions exist and are transparently
> related to each other, the form without the wa- should imply a specific
> action, and the form with the wa- should imply that the action is
> normative. I would expect this rule to be standard in MVS and at least
> well fossilized in Dakotan.
I agree with this completely, though I'm calling it "acting on a specific
object" (third person object without wa) and "acting on a non-specific
object" (third person object with wa). The actual semantics might vary a
bit from language to language.
> Second, I would propose that generalized verbs were favored as noun
> constructions. ... Hence, we get many verb-derived nouns beginning with
> wa-. This also should be normal and productive in MVS.
Again, this is what I'd say, with different terminology.
> Third, as time passes and MVS splits into its daughter languages, new
> forms develop to indicate normative or habitual action, and wa- ceases
> to be used productively as a normativizer in the non-Dakotan languages.
> In ancestral OP at least, the loss of the original wa- rule forces a
> reinterpretation of wa- in both the noun and verb contexts. In the noun
> context, wa- is now understood as a head-marker for an entity
> normatively characterized by the action of the verb.
Here I disagree, though I think that linguists and speakers alike might
easily fall into the error of thinking this, based on (a) a tendency to
see things that translate English nouns as nouns, and (b) taking a sort of
rough and ready "this matches this" approach to cutting up the morphemes.
If we-action recurs in the form of translations for English nouns that are
conceptually action-er nouns, it is natural to assume at first cut that
we- is the analog of -er, and a little analysis of the morphology leads to
taking wa- in we- as the head marker. I just think this is the wrong
analysis.
Siouan languages are definitely not nounless, but, though they have
sizable numbers of words that are formally nouns, they also have many
words that refer like nouns, but are verbal in morphology. Naturally,
there is a tendency for such words to behave more and more like nouns in
terms of syntax and morphology over time, so we find anomalies in how
possessive is marked, etc. For example, the equivalent of 'my house'
might be rendered 'I live in it' or 'my he-lives-in-it' to cut across the
slippery but convenient slope of translational analysis.
I used to think I was maybe the only person with this perhaps somewhat
subversive view, but I've detected signs of late that others think
somewhat along these lines, too, though perhaps more clearly.
> Meanwhile, the wa- on the verb side is also reinterpreted from normative
> or habitual or repeated action, to imply multiple objects acted upon.
> Since the ancestral language perhaps had no patient pronouns for 'us'
> and 'them', wa- was then readapted to fill these slots.
I'm not sure that "acting on non-specific" and "acting on multiple" aren't
more or less interchangeable concepts in the Siouan languages, though this
idea may not hold water for Dakota. The distinction is clear enough in
English, but hard to justify in, say, Omaha-Ponca. However, what you
suggest is a plausible analysis of how wa- comes to be part of the
inflectional morphology of P12 and O3p in various languages and one that
I've definitely always used for wa O3p and recently been inclining to for
P12, too. It's encouraging to see you coming so readily to the same
conclusions.
Along these lines, Bob has argued that *(w)aNk- ~ *(w)uNk- as an inclusive
marker originates in an incorporated noun for 'man' (cf. French on < Latin
homin- used in this way in French), with something like IO waM<ny>e ~
waN<ng>e 'man' < *waNk-e fitting the formal requirements fairly nicely.
We also know that wa- is effectively a "generic incorporated noun" (a
pro-incorporand, or pro-noun, if you will, rather like Eskimo pi-, I
think). It could easily stand in for a generic reference of some sort in
the clause, and Siouan as a whole, with *wa/uNk- in the inclusive role,
and Dakotan, with wic^ha- in various animate third peson roles, show that
these categories are often handled with incorporated nouns. So, wa- can
easily stand in for one of these incorporands.
In terms of details, I'm not quite sure why two wa's in P12 in IO. I
think the IO and Wi hiN- are cognate to the OP dative contraction of
aN(g)- and gi-, which is iN(g)-, and I've argued the details of that
elsewhere. I suggest it's a sort of antipassive. The extra a in the
Dhegiha wa-a- and Wi waNaNg-a- may be a locative, and essentially the same
sort of thing - an oblique reference to the inclusive object. Maybe IO
wa-wa- conceals the IO cognate of this a- under some reanalysis.
It's interesting to note that Crow-Hidatsa lacks both of these uses of
wa-, along with the inclusive pronominal itself. They just pluralize the
first person. However, with stative verbs, according to Randy, Crow
substitutes for the first person plural a prefix balee which is paired, in
effect, with a third person singular verb as far as the morphology of the
rest of the form. I don't know if there is an etymology for balee-,
though as far as form goes it is essentially wa-ee with an epenthetic r
separating the two parts. It could also be an incorporated noun, I
suppose. In any event, the general form is very much like what Bob
considers we should expect for the historical underpinnings of the
inclusive, though I think it must be an independent development.
JEK