double inflection (Re: animate wa-)

[Koontz John E]

On Sat, 3 Jan 2004, REGINA PUSTET wrote:
> Double inflection of this type is extremely marginal in Lakota -- right
> now only one verb comes to my mind that behaves like the OP forms, i.e.
> iNyaNkA 'to run'. In the Boas/Deloria materials, and also in Buechel
> 1971, this verb is quoted as having wa-'iNmnaNkA for first person
> singular, which also is the standard Rosebud form today. My Rosebud
> speaker (about 80 years old) mentioned that Pine Ridge uses the
> (simplified, regularized) form wa-'iNyaNkA instead. My Pine Ridge
> speaker (about the same age) has confirmed this form, adding that in the
> 1930s, Pine Ridge had wa-'iNblaNkA for 'I run'.

There are essentially two kinds of double inflection in Siouan languages.
One (1) occurs when multiple elements of a stem are inflected, which is
the case with Dakotan i=yaNka in its full conservative glory.  Another
example of this type is (or was) hiyu, which is in Buechel as wahibu,
yahilu, (hiyu).  I think I've seen it given as wahiyu, yahiyu, hiyu in
modern, or at least Pine Ridge form.  Riggs gives wahibu ~ hibu and yahidu
~ hidu, showing an alternative pattern of simplification.  I apologize if
I managed to day that such formations are frequent.  I only meant to say
that they occur sporadically - perhaps enough to provide an example.

Dhegiha languages abound (relatively) in double inflections of this first
sort, some lexical, e.g., OP gaN=dha 'to want' (kkaN'=bdha A1,
s^kaN'=(s^)na A2, gaN'=dha=i A3, aNgaN=dha=i A12) and others grammatical,
e.g., =xti added to another verb requires the following supporting
auxiliary =maN 'I really ...', =z^aN 'you really ...'.  There are also a
lot of compound motion verbs (comparable to hiyu)  that are inflected on
both elements.

The other kind of double inflection (2.1) occurs when two pronominals,
usually regular + irregular, occur in sequence on a single stem, as in OP
a-t-taNbe 'I see it', etc., or Osage a-p-paN 'I call', dha-s^-paN 'you
call', etc.  This last is cited in LaFlesche, but I believe Carolyn told
me it is singly inflected regular today (a-paN, dha-paN, ...).  This
pattern is much more sporadic in Siouan than the "two elements separately
inflected" pattern, but does occur sporadically.  IOM has a lot of it with
r-stems, I believe, to the extent where it is almost the regular pattern:
ha-da- A1, ra-s-ra- A2, ra- A3, and so on.

A second varient (2.2) on this second kind of double inflection occurs
when the portmanteau for A1P2 is used regularly with the irregular (or
syncopating) A1 form, as in OP wi-b-dha- A1P2-A1-by_mouth.  Lakota does
this with c^hi, e.g., c^hi-b-la-, right?  In this case the context is
paradigmatically very limited, but also very common in terms of lexical
frequency.

A third version (2.3) of this occurs with datives and suus forms of
syncopating verbs in OP, where you can get sequences in which both the
dative prefix and the underlying stem are inflected, if the underlying
stem is syncopating:  eppaghe < a-(g)i-p-gaghe 'I made it for him'
A1-DAT-A1-make, etc.;  agippaghe < a-gi-p-gaghe 'I made it for myself'
A1-SUUS-A1-make, etc.

> One way of dealing with this is by hypothesizing that Lakota has moved
> beyond an earlier (pan-Siouan??) stage of using double inflection with
> many verbs to a point where erstwhile doubly inflecting paradigms have
> been completely regularized by eliminating the irregular (or let's say,
> less regular) part of the inflection, i.e. -mn-/-bl- in the case of
> wa-'iNmnaNkA/wa-'iNblaNkA, retaining only the canonical wa- '1SG.AG'
> marker.

Which is essentially what has happened with Osage paN, and I'd argue that
double inflection in OP daNbe and Os paN is a sort of route to
regularization - first bury the irregularity and then forget it.  And the
two inflected pieces can work the same way, as these Lakota data show,
though presumably here the double inflection is original.

> ... I can't imagine that individual Siouan languages have "invented"
> double inflection independently from each other. OP double inflection
> patterns would then occupy the centerpiece of the cycle, while
> structures like OP ttaNb'e 'I see' would be the historical point of
> departure.

I agree with this in general terms, but I think that particular cases of
double inflection of the first type are continually recreated via
different sorts of serial verb construction.  The system that creates the
forms is inherited, but the particular examples aren't always.  Still,
hi=yu matches a pattern of motion verb compound that OP has, though it
seems not to have the corresponding *thi=(dh)i, but only thi=dha and
thi=dhadha, matching hi=yaya (which can only occur reduplicated).  (I
believe hi=yaya is still doubly, actually triply, inflected in modern
Lakota, right?)  The second sort of pattern (or at least 2.1) comes about
entirely secondarily, as a kind of regularization.  Versions 2.2 and 2.3
are a bit harder to characterize.  We might call them something like
overgeneralization.

JEK