Numic Query (fwd)

[Koontz John E]

Before we go very far with the lost dialect theory of nasal dropping, it
might be worth noting the particular status of nasals in Hidatsa.  It's
true that Siouan languages typically have nasal vowels, but Crow and
Hidatsa do not.  In the modern languages the contrast between PS nasal and
oral vowels is neutralized in favor of orality.

In such a situation, a form like CVNCV would not contrast with CVCV or
might occur in free variation with it.  Perhaps the unnasalized Shoshone
forms represent the influence of Hidatsa on Sacagawea's speech.  Note that
there's a pretty strong tendency for VNC to be pronounced with an
intrusive nasal stop in at least Omaha-Ponca.  I'm not sure how far that
tendency occurs in other languages.

There are nasal segments in modern Crow and Hidatsa, but these are not
usually depicted as contrastive.  As far as I can recall the rules they
are, for Crow /w/ (to use Kaschube's notation), b when simple and
non-final, m when final and mm when geminate.  For Crow /r/ (Kaschube
again), d initially, l medially, n finally, and nn in geminates.  Of
course the change in the scheme of writing /w/ and /r/ are not the only
differences between Kaschube's system and the current popular orthography.
The popular scheme is close to the auditory quality of the sounds to
English ears.

The rules for Hidatsa, again by recollection, are that /w/ and /r/ are
pronounced (and often written) as m and n in initial position.

The distribution of nasality in Mandan, so closely associated today with
Hidatsa is somewhat different.  Kennard depicts something more or less
comparable to Dakota, but Hollow concluded that w and r have the
allophones m and n before nasal vowels and decided in the spirit of the
time to write w and r in those cases, too, since the nasality of m and n
was conditioned.  Kaufman argued in his unpublished work on Proto-Siouan
(which I know only from some summary sheets discussed in a class conducted
by David Rood) that nasal stops were also conditioned entirely by nasal
vowels in Tutelo and Winnebago.  I don't really know of any counter
arguments to either claim.  In most other Siouan languages the situation
is either that a few anomalous forms seem to have nasal stops where no
nasal vowel is present, e.g., mi 'sun' in IO (not sure of this), or not to
have nasal stops when one is, e.g., Da wiNyaN 'woman'.  Sometimes nasals
also arise regularly from other sources, e.g., OP has m and n for *W and
*R (not *w and *r) whether or not the following vowel is nasal, e.g.,
ine'gi 'uncle' cf. Teton Da leks^i' 'uncle'.

To the extent that Siouan languages lack nasal stops and have m n etc.
only where an adjacent nasal vowel acts upon w r etc. it is natural that
the loss of nasalization in vowels in Crow and Hidatsa would allow a new
distribution of nasality to unfold.

I believe that the distribution or occurence of nasal stops is anomalous
in Plains Algonquian, but I don't know the details.  It sounds like loss
of nasals might be "in the air" in the Plains, if it affects Comanche,
too.

In early transcriptions of Crow and Hidatsa b m etc. and d l n etc. often
seem to be in free alternations.  Washington Matthews more or less
systematized his Hidatsa usage, but he comments on the difficulties.

Randy Graczyk presented a paper a number of years ago in which he looked
at the freely varying forms in older transcriptions and concluded that -
for Crow, anyway - there was some tendency for #mV and #nV to occur where
the following vowel was nasal in cognates elsewhere in Siouan, and,
conversely, for #bV and #dV to occur where it was nasal.  He hypothesized
that vowel nasality was in the process of being lost in the early contact
period.