Lexical compounds, syntactic compounds, and truncation

[rwd0002 at unt.edu]

Quoting "R. Rankin" <rankin at ku.edu>:

> I think it's often impossible to distinguish so-called syntactic from
> lexical incorporation in Siouan because, for many verbs, the two processes
give homophonous results.  Skata/+ACQ-kata does have two forms, distinguished by
> accent, but the fact that both have phonological truncation strongly suggests
that there is not a hard-and-fast line between the two types of incorporation,
and, that,in fact, what we have is a continuum between what were earlier and
later lexicalizations of particular compounds.  Given the lexeme-by-lexeme
nature of compound formation, this would not be surprising.
>
> You gonna give us a preview of your solution, Wim?  :-)


Hi Siouanists, Bob, Violet:

Sure, Bob.  (By the way, I just sent you a copy of the Serial verbs paper,
comments most welcome, there is still time for revision).

Following Chambers (1978), I view lexical and syntactic compounding (or
incorporation) as a matter of stress placement and reduction: there is only one
stress (on the second syllable)in lexical compounding, there are two stresses
in syntactic compounding, and the second one gets reduced.  In stripping there
are no stress changes at all.  How does lexical vs. syntactic vs. stripping
match up with truncation?

In noun incorporation of the lexical kind, there is usually truncation,
sometimes coalescence (de Reuse 1994:204-205).  In noun incorporation of the
syntactic kind, there is never truncation (although coalescence might be
possible). In noun incorporation of the stripping kind, there is neither
truncation nor coalescence.

In verb compounding (or incorporation, or, as I prefer to call it: verb
serialization) the situation with regard to truncation is quite different.  In
verb compounding of the lexical kind there is always truncation (unless of
course the first verb is not truncatable by its phonological nature, e.g.
because it is monosyllabic, or the last syllable is untruncatable).  In verb
compounding of the syntactic kind, there is always truncation as well (again if
phonology allows it).  And in a third kind of serialization: verb stripping
(i.e. two verbs in a construction without any stress reduction), there is
truncation as well!  So truncation has something to do with serialization, but
not with the lexical vs. syntactic vs. stripping distinction.

(A difference between lexical and syntactic serialization on the one hand, and
verb stripping serialization on the other hand, is that in lexical and
syntactic noun stripping ablaut occurs (unless of course truncation occurred),
but in verb stripping serialization ablaut never occurs).

But, back to the John Kyle's original examples:

(1) shka'l-oma'wani I travel playing
and
(2) shkal-o'mawani I go about to play.  These are in Boas and Deloria (74, 84).

Can someone recheck on these?  Violet? The theory of meaning I develop in my
paper (no time to explain here) predicts that (1) is not a syntactic compound
at all, but a case of verb stripping, hence the unreduced stress.

And does (2), apparently a lexical compound, have the correct stress pattern?
(Disturbingly to me), my theory predicts a syntactic compound: *shkal(primary
stress)-oma(reduced stress)wani.

I generally agree with Bob's view of lexicalization of compounds on an
individual basis, and if my theory is wrong, I will have to account in that way
for shkal'o'mawani.  I have been trying hard to make generalizations where form
and semantics can be predicted from each other to some extent, even though the
semantics of Lakota serial verbs are subtle, and the formal variables are many:
stress pattern (i.e. the difference between lexical compounds, syntactic
compounds, and stripping), truncation (yes or no), and ablaut (yes or no).

Best,

Willem J. de Reuse