Lexical and syntactic compounding in Dakota

[Koontz John E]

On Sat, 29 May 2004, Kyle, John H wrote:
> While reviewing the data on lexical vs. syntactic compounds in Dakota I
> came across a form which has me scratching my head.  Lexical compounds
> are formed by the concatenation of roots and contain one stressed
> syllable (peninitial).  Syntactic compounds concatenate 'words', each
> with their own stress (the second stressed syllable is reduced)  One of
> the classic examples given to show the difference between the two types
> of compounds is:
>
> c^he'Ga zi'   (yellow kettle)(syntactic compound)
> c^hexzi'     (brass kettle) (lexical compound)
>
> In the L compound, the c# root (c^ex  or c^eG) directly attaches to zi.
> In the S compound the root undergoes stem formation (epenthetic final
> vowel and initial stress) before concatenating.  My question concerns
> the following classic example:
>
> s^ka'l o ma'ni (he goes playing about) (syntactic compound)and
> s^kal o' mani   (he goes about in order to play) (lexical compound)
>
> What is the nature and behavior of s^kal ?  Why doesn't it appear as
> s^ka'la in the syntactic compound.  Is it considered a C# root?  And
> does anyone have examples of its use in other constructions?  Thanks.

Well, the underlying stem is s^ka't# right?  As a non-Dakotanist I am
helped here by this being a Mississippi Valley-wide set, s^ka'ta being
comparable to OP s^ka'de.  So the problem is that one expects a full word
form of s^ka'ta in a syntactic compound?

You mentioned that Santee has c^eg^azi (accent?) for 'brass kettle'.  I
see that Riggs does give c^he'g^aska for 'tin kettle'.  Maybe exceptions
occur in the opposite direction, too, with less than full word forms in
some syntactic compounds.  In that case, the critical factor in
distinguishing the two kinds of compound is the accentuation.

I always think of C-final forms in Dakotan as being in "subordinate" or
compounding form, at least in Teton and Santee.  I think that with certain
morphemes, e.g., =lu ~ =l or =ka ~ =c^a ~ =l, Dakotan is perfectly capable
of deleting an organic final vowel to achieve this C-final "subordinate"
form, but perhaps historically the opposition comes about through the
alternation of "complete" C-final forms and "independent" forms in which a
C-final root has been rendered "independent" by adding a suffix or
enclitic.  This is implicit in calling -a or -A "stem forming."

In some non-Siouan languages this sort of suffix is called an absolute
marker when it occurs with nouns.  (Not to be confused with absolutive as
used in characterizing ergative case systems.)  Absolutive markers are not
always final vowels added to C-final stems, of course, but often there are
some nouns that require absolutive markers and some that don't.  This sort
of things is pretty common in Uto-Aztecan, and it also occurs in Caddoan,
though it is not restricted to North America.

For Uta-Aztecan, Langacker (Overview of Uto-Aztecan Grammar, 77):
"Absolutive suffixes are one of the more distinctive and characteristic
features of UA grammar.  An absolutive suffix, in UA terms, is an ending
with no apparent semantic value that appears on nouns in citation forms
but may drop when a noun is subjected to various morphological processes,
such as affixation, compounding, or reduplication.  The P-UA absolutive
suffix was *-t<barred i>, deriving ultimately from *t<barred i> 'be' ...
[it] was retained before the accusative *-a, and possibly before the
plural *-m<barred i>, but lost when any other morphological elements were
attached."

For Caddoan, Parks (Pawnee Grammar, p. 97):  "An absolutive suffix -u
occurs with the majority of noun stems when they stand independently,
i.e., when they are not incorporated or otherwise in composition with a
following stem or suffix.  This elements seems to serve no other purpose
than classifying the noun as a noun and rendering it absolute."

Or, Rood (Wichita Grammar, p. 7):  "-?a.  This suffix occurs on the
citation form of a majority of nouns and may even be part of other
suffixes listed below.  Although its meaning seems to be 'nominalized', it
is not used productively to form new nouns.  ...  Occasionally two nouns
are differentiated solely by the presence or absence of the suffix, but
the suffixed form should not be taken as a derivative of the non-suffixed
form."  I mention this last because the examples show that simple nouns
occur without the absolutive, e.g., he:c 'fat', but he:c?a 'fire'.  (Hmm.
Reminds me of PS *hpet 'fire'!)

Of course, absolute (or absolutive) markers are characteristic of nouns,
not verbs, but in Dakotan and other Siouan languages ablaut vowels, and
the intimately related stem-formant vowels, and so, Dakotan C-final stems,
are at least as prominent an issue with verbs as with nouns, and, of
course, John's conundrum was posed in connection with verbs.  However,
citation forms of verbs are often effectively verbal nouns or adjectives,
and subordinate incorporated verbs are generally of the same character.
In addition, citation forms of nouns, as the Uto-Aztecna case shows, may
be effectively predicative forms.  So, the boundary between the two
classes is somewhat murky.