Re: (tangent thoughts arising from) Active-Ergative langs (discussion)

On Thu, 21 Sep 2000, Marcus Smith wrote:

> dirk elzinga wrote:
>
> >       e= same subject (<e> is a high, central, unrounded vowel)
> >       a= different subject
>
> What does the = mean in your glosses?
Clitic boundary. If the first word of a sentence happens to be a noun,
the clitic attaches to a noun. And so on. The switch reference clitics
are initial in the clause.

> >In the switch reference systems I've seen, there are usually a handful
> >each of SS and DS markers indicating various tense/aspect distinctions
> >and replacing the normal tense/aspect suffixes (there is usually some
> >homophony between members of the switch reference system and the t/a
> >system, though).
>
> Is that what the Numic languages do?
Yes. Shoshoni identical reference clauses are marked with suffixes
-ten 'while' and -kkanten 'after', both of which are also aspect
suffixes: -ten is 'continuous' or sometimes 'progressive'; -kkanten is
'stative'. IR can also be marked with -si(h) or -cci(h) in temporal
clauses when the action of the subordinated verb takes place prior to
the action of the main verb (-{s/cc}i(h) is not an aspect suffix):

        ne miakwa tuhupekkacci
        ne mia -kwa     tuhu -pekka -cci
        1s go  -thither mad  -got   -SS
        'I left because I got mad.'

There are two kinds of DS clauses: those in which the subject of the
embedded clause is inflected for objective case, and those in which
the subject of the embedded clause is inflected for possessive case.
The suffixes are:

        objective SR:
        -kkan 'when, after' (also resultative aspect suffix)
        -ku 'while'

        possessive SR:
        -na 'present' (also resultative aspect)
        -kkwai 'momentaneous' (also momentaneous aspect)
        -ppeh 'perfect' (also a nominalizer)
        -tu?ih 'future' (also future tense marker)
        -ih 'unmarked' (not a t/a marker)

By and large, there is a prohibition on the cooccurrence of t/a
suffixes and switch reference markers; makes sense--the SR system
seems to encode t/a distinctions and is historically derived from it.

> The only one I've looked at is
> Comanche, but the grammar did not characterize switch reference (hereafter:
> SR) like that at all.  That wouldn't be the only issue I've had with that
> grammar though.  (Never could figure out what final features are from that
> book.)
Was that Charney's grammar? I haven't looked closely at SR in
Comanche, but there are bound to be differences from Shoshoni, and
different people will characterize switch reference differently. (For
a real treat, trying reading Sapir on Southern Paiute switch
reference, or worse, Voegelin on Tubatulabal; what a mess.) Generally,
Charney's grammar is pretty decent, especially compared with what came
before. The little dictionary put out by Lila Wistrand Robinson is a
piece of crap. Jim Armagost did what he could to save it from complete
disaster but only partially succeeded.

I'm also not surprised you didn't get final features from Comanche (no
matter who described it). Final features in Comanche are completely
lexicalized and only make sense if you know the Shoshoni patterns;
Comanche is essentially bad Shoshoni spoken from the back of a horse.
(Paying attention, John?) If you want to know the final features
story, e-mail me privately; it was the subject of my dissertation.

> The Muskogean languages generally treat SR and t/a separately; SR is fused
> with the complementizers.  There is a same-subject and a different-subject
> marker for each set.  Generic is at/ka~; 'and' is na/cha; contrastive
> akoot/ako~; etc.  The homophony in these systems are between SR and case -
> which has led to all kinds of fantastic claims about Choctaw not have case
> but that all nouns are clauses, or that clauses are case marked.  All
> showing a lack of knowledge about the distribution and behavior of
case and SR.

Okay. Then Tepa is similar to Muskogean in this respect since I view
the SR markers as complementizers more than anything else. Interesting
that they came from case inflection in Muskogean, though.

> In Yuman languages, subordinate clauses aren't marked for tense, but they
> have switch reference markers.  Only a single set though, like Tepa.
And IIRC, Yuman switch reference markers are the same markers used in
person inflection, deixis, and t/a.

> In
> Hokan languages, SR also encodes temporal relationships between the two
> clauses: simultaneous, afterwards, earlier, and so forth.
This is a basic distinction in Numic subordination; simultaneity or
sequencing of verbal actions.

> In Eskimoan
> languages SR is part of the 3rd person agreement.
Makes sense; it's only in the 3rd person where this becomes
potentially ambiguous, no?

Dirk

--
Dirk Elzinga
dirk.elzinga@m.cc.utah.edu