More regarding "wa"

[Koontz John E]

At the moment I'm inclined to see wa prefixes in verbs as indefinite (or,
really, non-specific) patients, and in some languages as third person
plural object inflections (in OP not 3p subject inflections, even in
statives).

In nominalizations I think they play the same role(s), and are not subject
references unless the subject is encoded as a patient.  I'm arguing this
in terms of Omaha-Ponca, but I think that similar arguments apply in other
Siouan languages, modulo the wa vs. wic^ha complexity in Dakotan.

I'll take advantage of Rory's examples to play the devil's advocate, as I
think his analysis of wa as the subject marker in nominalizations is
essentially different, and requires that wa in nominalizations be regarded
as having a different pattern of functioning than wa in unnominalized
verbs.  Again, I have not yet done any examination of standard grammars to
verify this, but I think his approach is not without its advocates.  In
essence in his analysis wa is the reference to the head of the
nominalization, or it might be considered to be just the mark of
nominalization, since it doesn't contrast with another marker of
nominalization.  Rory already draws the necessary distinctions, so I'm
just running through his arguments in reverse, so to speak.

On Wed, 17 Dec 2003, Rory M Larson wrote:
> The issue of wa- prefixes in nouns that Tom and John are discussing has
> perplexed me too, particularly in parsing names for tools and other
> technical terms.  In my posting last week, I suggested that wa- might
> refer to the subject as well as to objects.  What I had in mind was this
> apparent use of wa- as a nominalizer:
>
>   wa-sabe = 'the one that is black'
>   wa-s^abe = 'the one that is dark'
>   ...

I agree that wa here is a reference to the subject, but also to the
patient, as the underlying stems here are stative.

> These are all stative verbs, but it looks as if active verbs can be used
> in the same way:
>
>   wa-nidhe = 'the one that heals'

Here I think the form is essentially 3pInd-(A3)-heals 'he heals them',
i.e., that wa refers to the ones healed (indefinite or actually
nonspecific third person patients), not to the healer (a specific, if
indefinite reference).

> And then there is the whole suite of implement terms that are built on
> the framework of
>
>   [NOM]-i-VERB
>
> where /-i-/ is the instrumental that implies that VERB is enacted by
> means of something.  Usually, if a noun sits in front:
>
>   NOUN-i-VERB
>
> then the noun is the object of the verb's action. Rarely, however, it
> seems that the noun can be the head of the derived noun phrase, and
> implies that the noun is used to perform the verbal action, rather than
> that it is the object of the verbal action.  I only have one example at
> the moment, and it's not as clear as I would like.
>
>   moNzezi-i-gattushi
>   brass  -i-  explode
>   'the brass thing that is used to explode'
>   = 'gun cap'
>
> As a caveat, it isn't certain that the internal -i- exists; it might
> just be

I'd agree that it could be there, "hidden," and missed in transcription.

>   moNzezi-gattushi
>   'exploding brass'

Another possibility here is that in this case ga functions to form a
stative of the sort invariably formed by the outer instrumental na= 'by
heat'.  In essence the inner instrumental ga- here is an oblique reference
'with violence' and the (patient) subject is governed by the underlying
stem ttus^i.  The clause structure is similar to

maN'ze na'=      z^ide
iron   with heat red
"red hot poker"

> Assuming that such constructions do exist, however,
> I'm inclined to think that the wa- in we- < *wa-i-
> nouns is the head of the derived noun phrase, and
> means 'that which is used to enact VERB'.

I'd argue that as constructions like

NOUN(instrument) i-VERB

are admittedly more the norm it would be more likely that wa was standing
in for an unspecified instrumental noun, though if nouns in other
capacities can occur we might want to admit that wa might also stand in
those capacities, too.  Whether we might want to allow wa to occur with
agents "bronze that causes an explosion" depends on a number of factors,
of course - whether this is the same wa that marks indefinite patients or
not, and whether we're really convinced that that wa is itself restricted
to patients.

> In fact, we can find up to three variants of the
> same i-VERB nominalization.
>
>   NOUN-i-VERB
>   moNkkoNsabe-i-dhittube
>   coffee     -i-   grind
>   'coffee-grinder'
>
> Here, 'coffee' is an object noun.
>
>   i-VERB
>   i-dhittube
>   i-grind
>   'coffee-grinder', literally 'grinder'
>
> Finally, we can get the same thing with a wa-:
>
>   wa-i-VERB
>   wedhittube
>   wa-i-grind
>   'coffee-grinder'
>
> But does this last construction mean
>
>   'thing used to grind (things)'
>
> or
>
>   '(thing) used to grind things' ?
>
> My gut feeling favors the first interpretation, and I think our speakers
> have also favored that, but it is really hard to find words that clearly
> distinguish the matter.

My gut feeling is the opposite, of course, though I really haven't
presented any general line of evidence in favor of it.  However, at a
minimum I feel it simplifies matters to have a single wa behaving in a
consistent way in several different contexts, rather than one indefinite
patient wa and one nominalization/head-marking wa, with overlapping but
different patterns of agreement.