Awahawi / Amahami

[VOORHIS at BrandonU.CA]

John Koontz wrote:
> A number of the Siouan languages have alternations of w with m and r/l/d
> with n conditioned by the nasality of the following vowel, or come close
> to a situation like this.
> Crow and Hidatsa are unique in Siouan in having
> (in the modern languages at any rate) no constrastive nasals, either as
> vowels or resonants.  However, both languages have conditioned contexts in
> which their resonants normally assume nasal values, and informal
> transcriptions of both often have nasal resonants in other contexts.  I
> believe the conditioned environment for nasalization in Hidatsa is initial
> position.  In Crow it is when two resonants occur adjacent to each other.
> I believe that the explanation for nasals being transcribed in other
> contexts arises from the non-contrastive status of nasality in these
> languages.  Speakers can produce a degree of nasality in all resonants
> without perceiving it as a mispronunciation as there is no need to keep r
> or l and n, for example, apart.  The degree of nasalization in a resonant
> might well vary with the speaker or even with instances of a particular
> word from a particular speaker.
>
> John Koontz

This situation, where nasal consonants appear as allophones of phonemes
which also have oral allophones, is a bit uncommon but not unique.  In
at least some varieties of Fukienese Chinese I believe [b] and [m]
belong to one phoneme, as do [l] and [n], with the nasal consonants
preceding nasal vowels and the oral consonants preceding oral vowels.
Oral and nasal vowels contrast in Fukienese after other consonants.
Furthermore, in some varieties of Cantonese, [l] occurs only at the
beginning of syllables and [n] only at the end -- no reference to
nasalisation elsewhere here, but just to position as in Crow and
Hidatsa.

According to Gregores and Suarez, Guarani has [m], [n], and [ng] before nasal
vowels and [mb], [nd] and [ngg] before oral vowels.

Closer to home geographically, Rood reports that Wichita has [n] and [r] as
allophones of one phoneme distributed by position again like Cantonese and Crow
and Hidatsa.

And closer to home genetically, Catawba, just before it ran out of speakers
altogether, seems to have been in the process of merging [m] with [b] and [n]
with [d] into one phoneme for each pair, with nasal allophones before nasal
vowels and oral ones before oral vowels.  Surely [d] was already present from
other sources, and [b] may have been too; in any case [d] and [b] appeared
before oral vowels only, whereas [m] and [n] must have originally preceded
vowels of either sort.  Whether the pre-oral allophones of [m] and [n] ever got
beyond [mb] and [nd] in the mouths of Catawba speakers or just in the ears of
transcribers is unclear to me at this point.  (Thanks to Blair Rudes for
confirming my  thinking about Catawba on this and providing additional details.
I trust he will filter out any errors here if I have mucked up the
explanation.)

Going back to the Chinese for a moment, in Cantonese dialects with no
initial [n], it has been lost through merger with [l] from other
sources, a process involving denasalization.  And the [b]~[m] phoneme in
Fukienese goes back to [m] in Ancient Chinese -- denasalization again to
produce the [b] allophone.  On the other hand, I think the [l]~[n] phoneme
derives from both [l] and [n] which contrast in Ancient Chinese, so here we
have both denasalization of original [n] before oral vowels and nasalization of
original [l] before nasal vowels.

The phonetics of Guarani suggest a denasalization to me, inasmuch as there is a
nasal element in both allophones, but an oral element only in the pre-oral.

Reading Chafe on Caddoan, we find that the Wichita [n]~[r] phoneme derives from
a merger of contrastive Proto-Caddoan *n and *r, hence there is both
nasalization and denasalization again, though note that in Pawnee a similar
merger yields only [r] with no nasal allophone according to Parks, so there is
only denasalization in this case.

The Catawba case is clearly a matter of a denasalization of originally wholly
nasalized segments before oral consonants.

In each of these admittedly rather few examples of nasal~oral consonantal
allophones, the recoverable history suggests denasalization of original nasal
segments always plays a role, while the opposite also happens in only some of
the cases.  Furthermore, all these unusual phonologies derive from more
conventional ones with nasal consonants that have only nasal allophones.

In history at least, nasal allophones seem to be prior, and in that sense more
basic.  Should the phonemes then be written as /m/ and /n/ rather than /w/ or
/b/ and /l/, /r/ or /d/?  Should this be considered in the transcription of
Proto-Siouan?

More complete references on request if anyone needs them.

	Paul