Truncated Stems (was Re: affixes vs. enclitics)

[Koontz John E]

On Tue, 9 Nov 2004, Rory M Larson wrote:
> However, even with the clarification, I'm not sure that I didn't really
> mean 'enclitic'.  In OP at least, we seem to have two or three common
> "endings" for nouns and stative verbs: -ga and -ge (both < -*ka ?);
> and -de < -*te.  They seem to be tightly bound on individual words,
> but in old compounds of such words these affixes seem to be dropped
> in preceding position.  I don't think this is just the result of
> abbreviated speech dropping syllables.
>
>   iNgdhaN'ga    - 'cat'
>
>   siNde'        - 'tail'
>
>   sne'de        - 'long'
>
> But the word for 'puma' or 'mountain lion', "long-tailed cat", is
>
>   iNgdhaN'siNsne'de
>
> where the -ga in iNgdhaN'(ga) and the -de in siN(de)' are absent,
> leaving only the -de in sne'de at the end of this NP.
>
> From this and other cases, I would suppose an earlier compounding
> grammar in which the NP or stative verb phrase could be closed by
> one of these affixes, such that only the final one was allowed.

Because the syllables that come and go here are fixed parts of the
cognates elsewhere, and are more or less unpredictable, I'd argue that
what is happening here is that an originally organic part of the stem is
being deleted in compounding.  Putting it the other way, the bound stem is
truncated.  I take these truncated forms as a Dhegiha version of Dakotan
C-final stem form.  That is, iNgdhaN- and siN- in iNgdhaN-siN-snede 'puma'
are what happens to the historical bound alternants *iNkraNk- and *siNt-
of the corresponding free forms *iNkraNk-e 'cat' and *siNt-e 'tail'.
Since 'long' comes last, it takes the free form *sret-e, so the whole
presupposed pattern something like *iNkraNk-siNt-sret-e.

Obviously, in the Dhegiha context the free forms are produced by adding e,
or the bound forms are produced by substracting it.  I just went over the
options on that for PMV in another connection, so I won't repeat it here.
What I think I can fairly add here is that the loss of not only the "final
e" but also the preceding consonant, so that iNgdhaNge reduces to iNgdhaN,
not iNgdhaNg or something like that is consistent with Dhegiha's stringent
restrictions on consonant clusters and also on consonant-finals.  I think
this explains truncation in Dhegiha vs. simple loss of final vowels in
Dakotan:  Dhegiha doesn't like Cs clusters (and reduces *ps and *ks to s)
and Dhegiha doesn't like consonant finals.

I don't want to suggest that iNgdhaNsiNsnede or other particular similar
forms like waz^iNttu or s^aNttaNga, etc., are necessarily reconstructable
for PMV or PDh, but only that the pattern of converting *CVC- bound forms
to *CV- bound forms is, and that this pattern of truncated bound forms has
been retained in at least some OP compounds, whether these are actually
inherited or simply formulated on an observable conservative pattern.

One possible hint that this pattern might be preserved in more recent
forms by analogy is that I don't think that waz^iNga '(small) bird' would
normally be expected to follow this pattern, though it does in waz^iNttu
'bluebird'.  Perhaps in that case the truncation pattern has been over
generalized?