C'ali update: Split-S cross-referencing, agentive pivot

Greetings all.  Now that I finally have some time, I'd like
to provide a little update on some of the developments that
have been going on in C'ali morphosyntax in the last year.
The primary changes have been the fleshing out of the cross-
referencing system on verbs, which operates according to a
Split-S system.  In terms of syntax, I will also describe
the agentive pivot in C'ali.

CROSS-REFERENCING

The C'ali verb cross-references the person, number and
gender features of the agent, patient and goal NPs in
three distinct morpheme classes.  Because C'ali is a split-S
language, its class of intransitive verbs is subdivided into
two classes which take the agent and patient markers,
respectively, to refer to their single NP. Grammatical gender
is distinctively marked only in the third person, where a
great deal of homophony ensures that there is no one-to-one
correspondence between gender and gender-marker:

(1)
               AGENT           PATIENT           GOAL
            Sg        Pl      Sg      Pl      Sg       Pl
1st       -ta       -?V [1] -nu      -kas    ?V-      -tse
2nd       -(i)m     -qwo    -thæ     -thæ    twe-     -(l)la

3rd I     -(a)n     -t|on   -këi     -xela   me-      -nar
    II    -(e)ssa   -t|on   -këi     -xela   me-      -nar
    III   -ku       -möra   -xela    -parsu  kwe-     -nar
    IV    -pha(ma)  -ku [2] -ni      -yö     tšis-    -kwe [3]
    V     -ku       -phai   -ni      -yö     swi-     -mi

[1] "V" here represents a [-hi] vowel equal in quality to
    last root V
[2] Archaic. Colloquially, -phai usually Gender IV ag. pl.
[3] Archaic. Colloquially, -nar usually Gender IV go. pl.

So, for example, a transitive verb root _tal-_ "look at"
might be inflected to mean "we look at them(IV)" as tal?ayö;
"we give him it", _t'o-_ "X give Y Z": met'o?oni.
Intransitives use essentially the same verbal inflectional
for the vast majority of verbs:

(2) ?os-këi=wi    saxmë-tsi
    die-PAT.I=RP  man-PAT3
    "The man just died"

(3) t|-anat-essa=luN      olma-qa
    thither-run-AGT.II=DP woman-AGT1
    "The woman ran off long ago"

(Glosses like "AGT1" or "PAT3" refer to particular nominal
declensions.)  Speakers have no choice about the selection
of paradigms for these verbs;  they simply subcategorize
for patientive or agentive affixes.  The interesting fact
about C'ali syntax is that this sensitivity to agentivity
is deeply embedded, as we will now see.

AGENTIVE PIVOT

A syntactic pivot is a relation of form (such as agentive
subjects of transitive and intransitive) that is coreferntial
with an another relation (such patientive subjects of transitive
and intransitive verbs) and involved in syntactic rules for
coordination, complementation, relativization, etc.  In the
following English sentence, the relation of transitive and
intransitive NP-s are the pivot (namely, the subject):

  (4) The man saw the animal and [X=the man] ran off.
  (5) *The man saw the animal and [X=the animal] ran off.

In both (4) and (5), the argument of the second verb has been
elided.  (5) is ungrammatical because it violates the pivot
that licenses the elision of X as the transitive subject.
We may construct sentences in C'ali which must be licensed by
a differnt kind of pivot:  the interpretation only holds if the
verbs share in common some agentive or patientive marking:

(6)
t|-anat-essa=luN      olma-qa    som t|e-qwöl-essa-xela
thither-run-AGT.II=DP woman-AGT1 and thi.-shoot.at-3SgA.II-3SgP.III

[x= woman] æntwes-si
            small.game-PAT4

"The woman ran off long ago and shot at the small game"

Now compare with a patientive intransitive verb, where it makes
the construction ungrammatical:

(7)
 * ?os-këi=luN   olma-s     som t|e-qwöl-essa-xela
   die-PAT.II=DP woman-PAT1 and thi.-shoot.at-3SgA.II-3SgP.III

   [x= woman] æntwes-si
              small.game-PAT4
   *"The woman died and [she] shot at the small game."

(8)
   ?os-këi=luN   olma-s     som t|e-qwöl-aN-këi
   die-PAT.II=DP woman-PAT1 and thi.-shoot.at-3SgA.I-3SgP.II

   saxmë-thei   [x= woman]
   man-AGT3
   "The woman died and the man shot at [her]."

Here, the woman is a patient and only argument of the first verb,
while the patient of the second is the small game. Because the
woman in the second clause is an agent, there can be no elision:
the change to agentivity must be explicitly expressed. In (8),
the argument for "woman" may be elided because it shares the same
role, patient, and is coreferential with the same word in the
preceding clause. To show that this is truly Split-S agentive
pivot, you also need to show that two intransitive verbs of
differing categories, one patientive and one agentive, may not
allow such elision; this is the case, as in (9):

(9) *t|-anat-essa          olma-qa    som  ?os-këi        [X]
    thither-run-AGT.II=DP woman-AGT1 and  die-3SgPAT.II
    *"The woman ran off amd [she] died"

Instead, you must explicitly mention again the coreferential
argument:

(10) t|-anat-essa          olma-qa    som  ?os-këi       olma-s
     thither-run-AGT.II=DP woman-AGT1 and  die-3SgPAT.II woman-PAT1
     "The woman ran off and [she] died"

I plan to work on voice-operations, such as antipassivization
and passivization, which would get around this problem. Any questions,
comments or criticisms?

 =========================================================================
Thomas Wier            "I find it useful to meet my subjects personally,
Dept. of Linguistics    because our secret police don't get it right
University of Chicago   half the time." -- octogenarian Sheikh Zayed of
1010 E. 59th Street     Abu Dhabi, to a French reporter.
Chicago, IL 60637