Interesting Constituency Example

[Koontz John E]

I encountered this example in the Dorsey OP texts:

JOD 90:117.18-19

"e=da'=daN               wanitta ttaNbe=kki,
 what(ever) (particular) animal   I see when

it?e'=adhe=       kki=z^i,
I kill it with it when

bdhathe= hnaN= maN  a'=daN    abdhiN"    a'=bi=ama
I eat it HABIT I do therefore I have it  he said

He said, "I have it because whenever I see an animal, then I kill it with
this and eat it."

The context is that the speaker has been asked why he carries a gun, a
device unfamiliar to the asker.

What I noticed first was that the applicative i- is, in effect, outside
the infixing (incorporating) causative construction t?e=dhe 'kill' ('cause
to die').  The constituency of it?e=...dhe is i-[t?e=...dhe].

I thought this was a nice example in serveral ways.  For one thing, it
shows layering or hierarchy in word morphology.  For another it shows that
inflection doesn't necessarily always occur at the highest level, though
there may be forms that do extract inflection.  Other examples:

u'-mu=s^te 'to be left from shooting' (u'- cf. Dakota wo'-)
u-mu'=xpadhe 'make fall (at a place) by shooting'

I also believe it would be correct to say that the habitual or exclusive
enclitic s^naN ~ hnaN ~ naN applies to the whole structure before it, not
just to the last verb, i.e., habitually (see + kill + eat), or maybe see +
habitually (kill + eat), but not not see + kill + habitually (eat).  Note
also that the a'=daN conjunction clearly applies to the whole aggregation.

Probably most of those reading this last will have a reaction like "Well,
duh!" but I think there may be some merit in pointing out that clause
boundaries can occur in mid word.  This is something that Randy deals with
in his Crow grammar.

Something else of interest about this example is the use of the 'when'
form =kki-z^i in which =z^i ia clearly not the negative, or, at leat, does
not have negative force here.  From the context, perhaps it means "only
when, in the case that."