Winnebago Vowel Length (Re: Helmbrecht Paper: Terminology 'modal prefix')

[Koontz John E]

On Mon, 8 Jul 2002, R. Rankin wrote:
> The verb 'eat' ordinarily implies 'eat SOMEthing', so I'd expect
> waruuc (or waaruc or whatever it is) ...

One of the little appreciated aspects of Winnebago vowel length is that it
is somewhat variable.

My understanding is that Winnebago lengthens monosyllables, so, whatever
the implicit length of ruc^, it would be ruu'c^ (or ru'uc^, as Miner would
position the accent).  I don't know if adding enclitics affects this, so I
don't actually know if it's ru'c^-s^aNnaN or ruu'c^-s^aNnaN, for example,
when you add the declarative.  I do know that the declarative is s^a(N)naN
after consonants, naN after vowels, which seems like a good candidate for
an unusual phonological rule, if anyone is collecting them.

This verb doesn't inflect regularly, so it's not possible to see if the
first person, say, is *haruu'c^ or *haru'c^, revealing whether the stem is
inherently long or not.  In fact, the first person is haa' c^, the second
person raa'c^, which will no doubt strike the Dakotanists as familiar
looking.  But, the inclusive is given by Miner as hiNnu'c^, and the
wa-form (or activity form, perhaps) is waru'c^ in the same source, so I
think we can take this as an underlying short-stem |[ru'c^]|, to the
extent that it has a stem ...  And, looking around, this is a common
pattern:  CVV'C : haCV'C.  I haven't yet noticed any CVV'C : haCVV'C.

Now, the next step is to look at the inflection of the wa-form, and in the
first and second persons Miner gives waha'c^ and wara'c^, not, for
example, *wahaa'c^ and *waraa'c^.  So it seems that the length of haa'c^
and raa'c^ is also due to monosyllabicity, rather than inherent length,
though these are certainly candidates for contracted forms.

A further step, is to look at wa+ha+ruc^, which, as expected by analogy
with Dhegiha is 'table' ('something to eat on').  This is waaru'c^, which
is what you expect if you start with Pre-Winnebago (a lot like Chiwere or
even Dhegiha, though the root is different there) *wa-a'-ruc^e and apply
first deletion of final e after simple stops, then Dorsey's Law (no
effect), and then Winnebago Accent Shift.

A point to make here is that everyone who works directly with Winnebago -
and Helmbrecht is no exception - takes the approach that morphemes like
the locatives ha, hi, ho, or the pronominals ha A1, hiN P1, hiN A12, etc.,
are h-initial, but lose these h's when something precedes them.  So wa-ha
is waa-, for example.  From a comparativist point of view, however, a more
likely scenario is that Winnebago just adds an epenthetic haitch to the
start of words that begin underlyingly with a vowel.  Notable exceptions
are verbs that we think of for comparative reasons as *?-initial, like
*?uN 'to do', which is uNuN in Winnebago.  (First person, incidentally, is
ha?uN' - glottal stop, short stem vowel.)  Another kind of exception seems
to be initial long vowels, like aagi' 'be ready', or at least
monosyllables like aa' 'arm'.

I suppose working with Winnebago speakers it must be clear that their
perception is that the h is organic.

There are some interesting wrinkles to that situation, however.  First,
the first person does lose its h in contractions usually, e.g., per
Lipkind ha < ha + ha.  But notice that now we would expect to have haa
with an epenthetic h before long a.  In fact, checking the inflection of
ha-verbs in Miner, I do find mostly haaCV'... first persons, e.g., hac^i'
'to live on' vs. haac^i' 'I live on', but you have to be careful, because
an underlying n-stem, for example, will behave differently, e.g.,
hanaNxguN' 'to listen' vs. hanaNaN'xguN 'I listen'.  There is normally a
shift of accent in an n-stem first person, e.g., naNaNs^e' 'to take away
from' vs. naNaN's^e 'I take away form'.  There is also haniN' 'to live'
and 'I live', but Miner gives 'I live' as haaniN' in his first IJAL
article on Winnebago accent, so the dictionary entry is probably a typo.

Then consider the verb 'to eat' mentioned above.  The wa-form had the
first person waha'c^, not waa'c^.  No loss of h here.  Perhaps of interest
here is that though Dhegiha simply lacks any initial in all the first
person pronouns, e.g., OP a A1, aN P1, aN A12, Ioway-Otoe, a close
relation of Winnebago, has ha A1, hiN P1, hiN A12.  Neither the Dhegiha
languages nor Ioway-Otoe have epenthetic haitch.  (In fact, the closest
case I can think of is in Shawnee.)  So perhaps h-initials in first person
pronominals have a sort of half-way status between organic and epenthetic.

Another interesting h-context is the causative, which is a suffixed ha
A1, ra A2, hi A3.  The inflection of 'to kill', for example, is t?ee'hi
'to kill' vs. t?ee'ha 'I killed'.  Note also t?ee'wahi < t?ee'-wa-ha-hi 'I
killed them'.  But both h's are lost after a consonant, e.g., ceebi' 'to
consume' vs. ceeba' 'I consume', but ceebwa'hi 'to consume something'.
I'm not sure why the two different accentual patterns for monosyllabic
root causatives.

JEK