Siouan Long Vowls

[Koontz John E]

On Mon, 19 Mar 2001, Rankin, Robert L wrote:
> I think we are ignoring some basic tenets of phonology. It would suffice
> even if all long vowels were traceable to polymorphemic sequences. It isn't
> legitimate to consider the morphological environments in determining what is
> distinctive in the phonology. For example, many instances of 'locative a-
> prefix' are derivational and have little or no semantic content.

But if all long vowels were also accented, whether or not we could discern
that they came about from morphological sequences, we would have to omit
length from the picture, which is the Dakotan situation.  So unless length
is in some degree independent of accent, we are overdifferentiating to
write it, however nicely we can predict the location of accent by assuming
it.

> Looked at across Siouan, it is notable that Crow, Hidatsa and Tutelo and
> Ofo, at opposite geographical ends of the family, had length transcribed in
> the same (i.e., cognate) lexemes as early as the late 19th and early 20th
> century. Miner's Winnebago clearly shows that this persisted into
> Mississippi Valley Siouan.

Awkwardly, if you undo accent shift and Dorsey's Law and certain V1 + V2
=> VV sequences at morpheme boundaries, you have more or less exact
correspondence of accent and length along the lines of Dhegiha, or, to be
precise, along the lines of Ioway-Otoe (cf. Ken Miner's joke that the way
to predict accent in Winnebago was to study IO).  I've spent a lot of time
looking at this in Winnebago and it's pretty much just the accent shift
and Dorsey's law that make accent and length seem so independent in
Winnebago.  That is, they do make them independent, but only secondarily,
since the synchronic accent can end up on, or one or two syllables after,
the long vowel it was once associated with, depending on the structure of
the word.

> These languages have length in both accented and unaccented syllables.

I guess we need a set of standard examples of these, for the individual
languages and diachronically, too, to help us out here.

JEK