*V-stems (Re: verb suppletion)

[R. Rankin]

I think the real question I have about 'eat' is whether
the Dakotan 1st and 2nd person forms with the accented
/a/ have a plain /t/ or an aspirated /th/ in them.  I
know that yuta has a plain /t/, but 'chew, eat
physically' has /th/, and in fact has PSi *th (one of
very few forms in which *th isn't from *rh).  I showed
that in Hochunk and Chiwere *th and *t will have
identical outcomes, so those languages aren't
diagnostic.  If Dak. has wate, yate, then I think John
is right and some other explanation is in order.  It
occurs to me that we already know that the sequence *wu
(including wuN) is highly unstable in Siouan, and that
normally it dissimilates to either /ru/ or /wa/, i.e.,
either the vowel or consonant changes.  That being the
case, the 1st person of 'eat', *w-ute could give /wate/
regularly.  Then only the 2nd person requires
accounting for, and an analogical explanation, ad hoc
as they may seem, based on the 1st sg. plus the
existence of a near synonym, /yatha/ 'eat', is pretty
reasonable.

I don't think we've really ever figured out the precise
status of /?/ and/or /r/ and /w/ in the putative
vowel-initial verb stems.  The question is whether they
are organic or epenthetic, or both, and there are still
a lot of imponderables (several of which John points
out below).

Bob

> On Tue, 5 Apr 2005, R. Rankin wrote:
>> I think the pronouns in these 'eating' verbs are the
>> same conservative
>> pattern you get with 'be sitting' (ra:Nke) and 'be
>> lying' (ruNke), verbs
>> in which the initial glides may be epenthetic.  The
>> only difference is
>> nasality of the V.  I think John is expecting a
>> somewhat more innovative
>> pattern.

> That's true.  I was thinking of *raathe as taking an
> r-stem inflectional
> pattern, because it does in Dhegiha.  However, there
> are verbs that take a
> pattern of A1 wV..., A2 s^r-..., A3 r... as Bob says.
> Usually the stem
> initial vowel is nasalized and A2 s^- has been lost
> before r, so that
> after all the sound changes have worked out you get a
> pattern of A1 m...,
> A2 n..., A3 r... (or A3 w...).
>
> Anyway, with that inflectional pattern A1 *waathe, A2
> *s^raathe, A3
> *raathe might occur, and mixing with A1 *prute, A2
> *s^rute, A3 *rute you
> might get quite interesting things, especially if the
> language also merged
> *t and *th as *t (or *d) as Winnebago and IO do.  I
> hadn't allowed for
> that.  However, though it would be exciting to have
> an(other) oral
> instance of this pattern, I still tend to feel that
> this verb doesn't
> exhibit it, on the strength of the Dakotan forms,
> which don't have
> aspiration as far as I know.
>
> In regard to the *V-initial stems, I have the feeling
> that the m/n/(*r ~ w
> ~ 0) pattern they exhibit is essentially an outgrowth
> of the *?-stem
> pattern (m/*y/0).  I apologize for the mixed notation
> immediately
> preceding.  I think that the basic pattern for
> *?-initial (or maybe it's
> *V-initial) stems (and others) was
>
>   *V-stems      *Regulars      *r-stems
>
> A1 *w-V...       *wa-CV...       *p-rV... < *w-rV...
> A2 *y-V...       *ya-CV...      *s^-rV... < *y-rV...
> A3   *V...          *CV...         *rv...
>
> The *V-stems stems in question are mostly nasalized -
> though *o 'to wound'
> and *u 'to come (head this way)' seem to be oral -
> and mostly have some
> element before the inflectional slot that conditions
> an epenthetic *r or
> *w in the A3 form, e.g.,
>
> A1 *i-w- uNghe 'I question'
> A2 *i-y- uNghe 'you question'
> A3 *i-  ruNghe 'she questions'
>
> leading to paradigms like
>
>   Proto-Dhegiha    Omaha-Ponca
> A1 *imaNghe         imaNghe
> A2 *iz^aNghe        *iz^aNghe
> A2 *iraNghe         idhaNghe
>
> except that the second person appears instead as
> is^naNghe (later
> inaNghe), which amounts to substituting an A2 form
> from the *r-stem
> (dh-stem) paradigm, presumably by analogy with the
> apparent *r-stem
> (dh-stem) form in the third person.  Sometimes
> (across Dhegiha) you find
> the first person in dh-stem form, too, e.g.,
> ibdhaNghe, or the third
> person might have epenthetic w instead of epenthetic
> dh (*r), e.g.,
> iwaNghe.
>
>> As I recall, 'sit, stand' and 'eat' are among the
>> very few verbs with
>> the archaic (V-initial?) conjugation pattern.
>
> The pattern that appears with *i-(r)uNghe 'to
> question' also appears in
> Dhegiha with some of the positionals, e.g., dhiNkhe <
> *(r)iNk- 'SITTING
> ANIMATE' which inflects
>
> A1 miNkhe      I-the-sitting
> A2 (s^)niNkhe  you-the-sitting
> A3 dhiNkhe     she-the-sitting
>
> (And has the same pattern of inflection for the
> suppletive stem dhaNkha <
> *uNk- in the plural.)
>
> I regard the *?-stems as cases of this pattern, too,
> because I've noticed
> that the Dakotan and Winnebago patterns for those
> stems match this mixed
> *V-initial/*r-initial pattern, cf. Dakotan
>
>   Dakotan          Winnebago
> A1 muN              ha?uN
> A2 nuN < *s^-nuN    s^?uN < *s^-?uN
> A3 ?uN              ?uN
>
> (Winnebago forms from memory and I'm not sure about
> length.)
>
> In the same stem Omaha-Ponca has
>
> A1 maN
> A2 z^aN < *y-uN
> A3 aN
>
> OP z^ and Da n don't correspond (and neither does
> Winnebago s^?), but if
> the Da n is from s^n and that s^n is an analogical
> importation from
> *r-stems, then everything makes sense.  (I assume
> Winnebago has rebuilt
> things on the assumption of a root *?uN.)
>
> Of course, there's a rub, which is that I do have to
> assume that all
> *V-stem (or *?-stem) verbs in Dakotan were switched
> to the mixed
> *V-stem/*r-stem pattern, even when there was no
> initial element to
> condition epenthetic *r in the third person.  By
> contrast, in Dhegiha it
> seems that the switchover affected all stems with
> epenthetic *r, plus a
> few more (the positional auxiliaries), but not the
> glottal stop stems.
>
> If one is uncomfortable with different languages
> exhibiting different
> degrees of extension of the same analogy, then I
> think that to be
> consistant one has to recognize the following classes
> of verbs:
>
> I)   to question
> II)  auxiliaries with *r
> III) auxiliaries with *w
> IV)  *?-stems
>
> There is some potential for combining I-III, and it's
> pretty clear that
> the second persons of IV in the various languages
> don't correspond with
> each other, though some of them clearly do match
> second persons in the
> I-III classes.  In short, the purer you get, the more
> you sense that
> you're overlooking the obvious.  At least this was my
> progression to these
> conclusions:  careful insistence on regular
> correspondences => numerous
> implausibly similar mini-paradigms => an assumption
> of differential
> degrees of analogical leveling.
>
> My understanding of the "degrees of analogy" thing is
> that (a) the
> original IV (*V-stemor *?-stem)  pattern is quite odd
> relative to other
> patterns in each language - we'd expect Dakota to
> have A1 muN, A2 *c^huN,
> A3 uN, for example, by analogy with OP A1 maN, A2
> z^aN, A3 aN, and OP A2
> z^aN is quite a surprise as it is.  Hence, there's an
> obvious motive for
> analogizing pattern IV away.
>
> Then, (b) some verbs like *i-(r)uNghe 'to question'
> carry their own
> epenthesis conditioner with them.  Others, like the
> auxiliaries, acquire
> the conditioning only in situ as a positional
> enclitic following a
> suitably preceding noun or verb, e.g., *(r)iNk or
> *(w)uNk - think
> *s^uNka=r-iNk 'the sitting dog'.
>
> Still other verbs lack the environment at all, e.g,
> forms like *uN 'do',
> though some of them may also occur in contexts like
> *i-(r)uN 'do with,
> use' that condition it.  Result - transfers from
> *V-stems to
> *V-stem/*r-stem mixed-stems occur in different
> degrees in different
> environments in different areas of the
> Proto-Mississippi Valley dialect
> continuum and when the regional dialects become
> distinct branches of PMV
> they show different patterns of behavior with stems
> that occurred in
> different environments.  Eliminate random forms over
> a long period of time
> and you end up with the different patterns we see
> today.
>
>