Siouan root constraints

[Koontz John E]

On Thu, 3 Feb 2005 cstelfer at ucalgary.ca wrote:
>    I was wondering if anybody has ever taken a look at the root
> constraints on stops in any of the Siouan languages.  What I want to
> know is: are there very many roots with two voiceless unaspirated
> stops?  The nomenclature for this type of root might be something like
> TVT(V).  It seems to me that a lot of roots have one aspirated and one
> unaspirated stop.  Is this accurate?

Apologies in advance if I'm doing violence to the notation. I'm using Th
for both *hT and the much rarer *Th forms.  These contrast in Dhegiha, but
fall together as Th in Dakotan (except that *Th comes out h with one
exception, *matho 'grizzly').  In Ioway-Otoe and Winnebago the *Th falls
together with *T as T while *hT becomes Th.  (There's some subsequent
shuffling in Winnebago to arrive at a T vs. D contrast.)

With that in mind:

ThVT(V) is relatively common.   Lots of examples.
TVT(V) is relatively rare.      See below.
T(h)VThV is vanishingly rare.   Off hand I'm no thinking of any examples,
                                as roots, anyway.

Of course there are other patterns of roots.  The statements above apply
with little modification to more general forms like these:

ThVC(V) ~ CCVC(V) is relatively common.
TVC(V) is relatively rare.
C(C)VThV is vanishingly rare.

Again there are other classes of roots, e.g., various V and CV forms,
CVC(V) with initial fricatives (voiceless, sometimes voiced), forms with
ejectives or glottalized fricatives, rare trisyllabic patterns, etc.  The
standard sources for odd forms are numerals and loan words, and there's a
sporting chance that the relevant numerals are just really old loans.

Actually, forms with initial T (unaspirated stops) are distributionally
limited in general.

- They occur in demonstratives, e.g., *ka 'yon', maybe *te 'this' (but the
cognate sets suggest *re and *Re, too).

- They occur in enclitics, e.g., *tu 'in' seems to account for Dakotan -l
~ tu and OP -di.  Or Winnebago has -gi in comparable contexts, from *-ki.

They occur with instrumentals of the shape TV.  These involve at least two
patterns:

- the instrumental forms with PS/PMV *p, i.e., *pa 'by pushing', *pi 'by
pressing', which seem originally to have been syncopating (A1 *p-pa-, A2
*s^-pa-, A3 *pa-), and

- the instrumental forms with *k, PMV *ka- 'by striking', which in
Mississippi Valley often involves traces of k-loss or lenition (e.g., OP
A1 a-(a)'-, A2 dha-(a)'-, A3 ga-), most likely because the original form
is *(r)aka- (inflected A1 *w-aka-, A2 *r-aka-, A3 *(i)raka-) and there has
been a lot of analogical repair in the paradigms.

- The dative and suus particles often include reflexes of a prefix *ki-.
Lately I've been thinking that the dative might have been *(r)iki,
parallel with *ka.  It would explain a lot.

Apart from this there are certain verbs with initial *p, *t, *k.  The
standard Dhegiha set is something like:

- *p-instrumental verbs
- *taNpe 'see, look at'
- *kaghe 'to make' (originally 'to make marks')
- *kaNdha 'to donate' (not sure I have this form right by memory)
- *kaNze 'to immitate, demonstrate'
- *kaNyiNka 'not to want to'
- *kaN=ra 'to want'

For 'see', Ioway-Otoe and Winnebago substitute:

- *a...ta 'to see'

These verbs are usually syncopating in Dhegiha.  Most are syncopating
Ioway-Otoe and Winnebago.  Many have trace irregularities due to
syncopation in Dakotan.

There are traces of others in this class, e.g., *paN 'to call out to',
maybe *ta 'to get up', and so on.

Nouns with initial T are also rare.  The one I remember easily is

- Da paha 'hill'

I remember it because the set here is irregular, since OP, etc., ppahe
suggests *hpah-, while the Dakotan form suggests *pah-.

> Also, would it be fair to say that ejectives are relatively rare in the
> Siouan languages that have them?

Ejectives and glottalized fricatives are rare.  I don't think you ever get
more than one in a root except by reduplication.  Mandan seems to have
*T?V as TV?  Glottalized obstruents are about as rare as T-initials.  They
are aways root initial, and the roots tend to have the form T?V (or S?V),
though you find forms like Omaha iNs^?age (*s^?ak-) or s?adhe (*s?a(r)-).