Funny W and R

[Rankin, Robert L]

I'll take this on when I return.  I'm in GA this week at my 50th high school reunion.  Bob

________________________________

From: owner-siouan at lists.colorado.edu on behalf of Rory M Larson
Sent: Wed 11/1/2006 9:24 PM
To: siouan at lists.colorado.edu
Subject: Funny W and R



Bob wrote:
> In favor of your idea of CN with a nasal release, I think you'll find in Riggs a notation that he heard Dakota (D-dialect) words /ob/ and comparable sequences as phonetically [obm] on occasion.  I included that in the paper I did at the Chicago SCLC 4 or 5 years ago.

That's very interesting!  Thanks for pointing that out.


> I don't think it matters one way or the other whether we postulate w/rN, Nr/w, or Hr/w, w/rH as long as there are no conflicting correspondence sets.  What I'm writing as N or H (=h/?) here are all equally just "features" right now.
[...]
> I still think the "solutions" are all pretty much notational variants and can't really agree that one feature is more "natural" (or economical . . . whatever) in these instances.

If we are only concerned about the existence of the particular correspondence sets, then W and R are sufficient.  But if we start postulating things like w/rN vs. Nw/r or w/rH vs. Hw/r, then we are talking about the mechanics of production, which is a perfectly valid question and one which has bearing on the possibilities for the historical development of correspondence sets as well.

We're a bit limited, and easily prejudiced, by the algebraic nature of our alphabetic representation system.  In the case of continuants like w and r combined with nasality or aitchiness, we actually have features occurring independently at at least two of three different locations, which can be in any temporal order, including simultaneous.  What we really need is a more analogue representation system that can be handled separately at each independent level.  Thus, (going off on a tangent here to set up a framework for discussions) we might try:

  Nasal       N     N   N                 N
  Oral         r   r    r    w   w    w   r
  Laryngeal                 H     H   H   H

which represents seven sample sounds: Nr, rN, nasal-r, Hw, wH, aspirated-w, and finally nasal-aspirated-r.

To make this analogue, however, we need to let the symbols continue arbitrarily long, so as to be able to show relative timing:

  Nasal          NNNNN
  Oral         rrrr
  Laryngeal   HHHHH

This would show an aitchy r, with the aitchiness starting just before the r, with nasalization added to the whole thing halfway through, and the nasalization continuing for some time after both the r and the aitchiness have ended together.

Liquid or semivowel type things, as well as vowels, are pretty independent of what goes on nasally and laryngeally.  Stops are not.  Making a stop implies that both nasal and oral passages are closed, and the stoppage of the air flow will interfere with laryngeal production as well if the stoppage is prolonged.

  Nasal          
  Oral         TTTT*eeee       TTTT*eeee    TT*eeee
  Laryngeal         VVVV     HH     VVVV      HHHVV

The above give possible representations of tte, hte, and tHe, respectively.  V in the laryngeal track stands for voicing, T in the oral track stands for full alveolar closure, and vowel is represented by the appropriate letter in the oral track.  Note that a stop is asymmetrical, in that it has a characteristic click only upon release.  I mark this click with *.  The click is the one element that cannot be extended.

This lets us write nasal consonants a little differently.

  Nasal        NNNNN
  Oral         TTTTiiii
  Laryngeal    VVVVVVVV

This is how we might represent ni, for 'water', in Omaha.  The first part has full alveolar closure, followed by i.  It's all voiced, and at least the alveolar closure part is all nasal, with nasality probably extending a bit beyond it.  If we hear it as niN, then the nasality extends just about all the way to the end of the vowel.

The shift from a nasal vowel to a stop requires both oral and nasal passages to close at about the same time.  Ideally, it should be:

  Nasal       NNNN
  Oral        aaaaPPP*aaaa
  Laryngeal   VVVVVVVVVVVV

to get aNba, where P in the oral track means full labial closure.  However, closing both nasal and oral passages at exactly the same instant is dicey, leading to a range of pronunciations between

  Nasal       NNN
  Oral        aaaaPPP*aaaa
  Laryngeal   VVVVVVVVVVVV

which we may hear as aba (since oral vowels seem to be more prominent than nasal ones), and

  Nasal       NNNNN
  Oral        aaaaPPP*aaaa
  Laryngeal   VVVVVVVVVVVV

which we will definitely hear as aNmba.  Hence, we will tend to get epenthetic mb/mp or nd/nt combinations anywhere that we have clearly defined nasal vowels immediately preceding a labial or alveolar stop.

Now let's look at what I mean by a nasally-released stop.

  Nasal             *NN
  Oral        aaaaPPPPaaaa
  Laryngeal   VVVV    VVVV

This is what we might write as apm^a (where m^ means voiceless m).  The key point is that the stop is released into the nasal passage rather than the oral one.  The click of the stop is in the same (nasal) place for any nasally-released stop.  This is entirely different from simply a stop followed by a random nasal, such as, say, igmu:

  Nasal              NNNNNN
  Oral        iiiiKK*.PPuuuu
  Laryngeal   VVVVVVVVVVVVVV

where K in the oral track means full velar closure and . in the oral track means a minimal unformed vowel or schwa.

Our standard alphabetic representation system can't easily distinguish between:

  Nasal            NNNN
  Oral        TTT*.TTTaaaa
  Laryngeal   VVVVVVVVVVVV

and

  Nasal          *NNNN
  Oral        TTTTTTTaaaa
  Laryngeal   VVVVVVVVVVV

, both of which we would probably write dna.  The former has an orally released stop followed by a nasal consonant in the same location, from which it is unavoidably separated by oral release-click and schwa before the tongue can return to full alveolar closure again.  The latter is a nasally-released stop in which full alveolar closure is never relaxed until after the nasal consonant has been expressed.

So if we suppose that *W and *R involved any sort of combination of nasal consonant and stop, we have three distinct options, not just two.  We can have nasal consonant preceding oral stop; we can have nasal consonant following oral stop; or we can have nasally-released stop.  The first two possibilities are clusters; the latter is a single phoneme.

I would imagine (this is not necessary to the argument) that *W and *R were originally tense and unvoiced, and that the various reflexes came about from weakening them in order to spit them out more easily.  So let's try a well-known word:

  Nasal           *NN
  Oral        TTTTTTaaaaKK*ooooTT*aaaa
  Laryngeal         VVVV  HHVVV   VVVV

which gives us RakHota, assuming *R is a tense, voiceless, nasally-released stop.  The shift from the initial stop to the following vowel is marked separately in the three different tracks.  In the nasal track, it is marked by briefly opening the nasal passage to allow the release-click.  In the oral track, it is marked by releasing full alveolar closure.  And in the larygeal track, it is marked by shifting from voiceless to voiced.  To reduce duration of the stop, we want to make these transitions happen sooner.

The first step in weakening the initial stop would be to reduce the length of time the stop is held:

  Nasal          *NN
  Oral        TTTTTaaaaKK*ooooTT*aaaa
  Laryngeal        VVVV  HHVVV   VVVV

This allows voicing to spread back from the vowel even to the full stop (optional):

  Nasal          *NN
  Oral        TTTTTaaaaKK*ooooTT*aaaa
  Laryngeal    VVVVVVVV  HHVVV   VVVV

The nasal release may take place sooner:

  Nasal         *NNN
  Oral        TTTTTaaaaKK*ooooTT*aaaa
  Laryngeal    VVVVVVVV  HHVVV   VVVV

Finally, given that there is no explosive release-click in the oral passage, the shift from full closure to vowel may be slow and sloppy.  The tongue might maintain only partial closure once the nasal release-click has occurred, which would mark the place of the full closure, but would no longer prevent speech breath from leaking through.  In this case, the most likely actualization of lax alveolar closure would be an l.

  Nasal         *NNN
  Oral        TTTTlaaaaKK*ooooTT*aaaa
  Laryngeal    VVVVVVVV  HHVVV   VVVV

At this point, the initial consonant would be a rapid glide: td*nl.  It would be rather unstable, and the likely spinoffs would be the four consonants t, d, n, and l.

One route would be to reduce the oral track from the front, and/or allow the nasal release sooner:

  Nasal        *NNN
  Oral        TTTlaaaaKK*ooooTT*aaaa
  Laryngeal   VVVVVVVV  HHVVV   VVVV

  Nasal       NNNN
  Oral        TTlaaaaKK*ooooTT*aaaa
  Laryngeal   VVVVVVV  HHVVV   VVVV

Going this route, opening the nasal passage eventually occurs before alveolar closure.  This eliminates the stop altogether, and we are left with an initial consonant that sounds pretty much like n.  The residual nasalized l will probably be dropped in analogy with pre-existent [n], and in the interest of making that consonant even shorter:

  Nasal       NNN
  Oral        TTaaaaKK*ooooTT*aaaa
  Laryngeal   VVVVVV  HHVVV   VVVV

This gives us nakHota.

Another route would be to bring the relaxing of the tongue forward in the oral track with respect to the nasal release.

  Nasal         *NNN
  Oral        TTTTlaaaaKK*ooooTT*aaaa
  Laryngeal    VVVVVVVV  HHVVV   VVVV

  Nasal         *NN
  Oral        TTTlaaaaKK*ooooTT*aaaa
  Laryngeal   VVVVVVVV  HHVVV   VVVV

  Nasal         *NN
  Oral        TTlaaaaKK*ooooTT*aaaa
  Laryngeal   VVVVVVV  HHVVV   VVVV

Following this route, oral release eventually catches up to nasal release.  If the vowel follows quickly after the lax l, the l will be blown apart by the new oral explosion that replaces the nasal release-click:

  Nasal          NN
  Oral        TT*aaaaKK*ooooTT*aaaa
  Laryngeal   VVVVVVV  HHVVV   VVVV

This renders the nasal release accoustically much less relevant, and it may be dropped:

  Nasal       
  Oral        TT*aaaaKK*ooooTT*aaaa
  Laryngeal   VVVVVVV  HHVVV   VVVV

This leaves us with dakHota.

Finally, the lax oral l may entirely replace full alveolar closure, while never being allowed to precede the nasal opening.  This would imply that most of the reduction takes place in front on the oral track:

  Nasal         *NNN
  Oral        TTTTlaaaaKK*ooooTT*aaaa
  Laryngeal    VVVVVVVV  HHVVV   VVVV

  Nasal        *NNN
  Oral        TTllaaaaKK*ooooTT*aaaa
  Laryngeal   VVVVVVVV  HHVVV   VVVV

  Nasal       NNN
  Oral        llaaaaKK*ooooTT*aaaa
  Laryngeal   VVVVVV  HHVVV   VVVV

At this point, the initial consonant is a nasalized l.  Nasalization might be dropped to give us:

  Nasal       
  Oral        llaaaaKK*ooooTT*aaaa
  Laryngeal   VVVVVV  HHVVV   VVVV

or lakHota.

In short, to reduce a nasally-released stop, we have three likely paths:

1. The nasal passage opens early, before the oral passage closes.  This gives us a nasal consonant (m, n).

2. The oral passage opens earlier with respect to opening of the nasal passage, leading to an oral release.  The result is an oral stop (p/b, t/d).

3. The oral closure weakens after nasal release.  The beginning of the stop is reduced up to the point of weakening, so that it is marked orally only by the semivowel or liquid that replaced its tail.  The result is a nasalized version of that weak consonant, in which the nasalization may go away: (w, l).

The reflexes we get repeatedly across Siouan, as I understand, are [p,b,m,w] <*W and [t,d,n,l] <*R.

Perhaps there are other possible phonotactic routes to explain the reflexes of *W and *R.  If so, I would propose trying to analyze them in the manner I've done here for the nasally-released stop hypothesis.  I'd also like to invite criticism of my own analysis from anyone who's actually made it this far!

Rory