antipassives in Tokana (long...)

Inspired by yesterday's back-and-forth with Nik Taylor about antipassives
in Watya'i'sa (sp?), I have decided to add an antipassive construction
to Tokana.  (It makes sense, given that Tokana is supposed to be an
ergative language.)  In the process, I have dropped some other features
of the language, and changed my understanding of absolutive arguments
in Tokana.  So here's the state of the art:

In Tokana, agentive subjects are ergative while objects and non-agentive
subjects are generally absolutive (abstracting away from those verbs
which idiosyncratically assign dative case to their subject or object):

        Ani     kahte-ke
        she:Erg hit-you:Abs
        "she hit you"

        Ani     hostane
        she:Erg danced
        "she danced"

        Nai     tioke
        she:Abs died
        "she died"

Ergative arguments may be freely dropped, but absolutive arguments may
not be (* = ungrammatical):

        Ani     tsitspe-h     kopo
        she:Erg broke-the:Abs pot
        "she broke the pot"

        Te      kopo tsitspe
        the:Abs pot  broke
        "the pot broke"
        or "the pot was broken"

        * Ani     tsitspe
          she:Erg broke
        "she broke (something)"

        Hostane
        danced
        "(someone) danced"

        * Tioke
          died
        "(someone) died"

When the absolutive argument of a verb is unknown, arbitrary,
or unimportant, the antipassive prefix "u(k)-" is added to
the verb, and the absolutive argument drops.  The ergative subject
then becomes absolutive:

        Na      ikei kilhte-m
        the:Erg dog  bit-me:Abs
        "the dog bit me"

        Ne      ikei u-kilhte
        the:Abs dog  Antipass-bit
        "the dog bit (someone/people)"

If there was no ergative argument, then adding the antipassive prefix
gives you a subjectless sentence:

        Ne      iha   tioke
        the:Abs woman died
        "the woman died"

        U-tioke
        Antipass-died
        "someone died"
        or "there was dying going on"

When the verb is in the antipassive, the direct object may be
added back in as an optional dative case argument.  Thus we get
alternations like the following:

        Na      Tsion tiespe-h      katia
        the:Erg John  built-the:Abs house
        "John built the house"

        Ne      Tsion u-tiespe
        the:Abs John  Antipass-built
        "John built (something)"
        or "John did some building"

        Ne      Tsion u-tiespe-i             katia-i
        the:Abs John  Antipass-built-the:Dat house-Dat
        "John did some building on the house"

The antipassive construction is used in a number of contexts.  For
example, if the direct object is not affected (in the usual way) by
the action, then it will normally appear in the dative case with
the verb in the antipassive:

        Na      Tsion kahte-m
        the:Erg John  hit-me:Abs
        "John hit me"

        Ne      Tsion u-kahte-ma
        the:Abs John  Antipass-hit-me:Dat
        "John did some hitting at me"

The second sentence might be used if John took a swing at me and
missed, or if John hit me but it had no effect on me (e.g. I just
ignored it).

The antipassive is also used if the direct object is only partially
affected by the action.  Compare:

        Na      Tsion tiespe-h      katia
        the:Erg John  built-the:Abs house
        "John built the house"

        Ne      Tsion u-tiespe-i             katia-i
        the:Abs John  Antipass-built-the:Dat house-Dat
        "John did some building on the house"
        or "John built part of the house"

The antipassive construction can also be used to indicate imperfective
(or progressive) aspect.  Since "John is building the house" entails
"John has built part of the house", the semantic extension from partial
affectedness of the object to progressive aspect seems natural to me
(and has precedents in other languages):

        Ne      Tsion u-tiespe       katia-i
        the:Abs John  Antipass-built house-Dat
        "John is/was building a house"
        or "John is/was building houses"

Notice that this use of the antipassive can be translated using the
present progressive ("is building") or the past progressive ("was
building").  To disambiguate in favour of the present progressive
interpretation, the adverbial "kas" = "already, as of now" may be
added:

        Ne      Tsion kas       u-tiespe       katia-i
        the:Abs John  as:of:now Antipass-built house-Dat
        "John is building a house/houses (now)"

This last sentence could be more accurately translated "As of
now, John has partially built a house", which, under normal
circumstances, could be taken to imply that he is building the house
now.

The antipassive construction also feeds nominalisation.  Consider:

        iasa      "to eat"
        iasi      "the one who is eaten"

        ukiasa    "to do some eating"
        ukiasi    "the one who eats"

In the same series as the antipassive prefix "u(k)-" is the prefix
"uma(k)-", which forms reflexive/reciprocal predicates.  As with "u(k)-",
adding "uma(k)-" to the verb causes the underlying absolutive direct
object to drop, and the ergative subject to change into an absolutive
subject:

        Ani     kahte-m
        she:Erg hit-me:Abs
        "she hit me"

        Nai     uma-kahte
        she:Abs Refl-hit
        "she hit herself"

        Kim    uma-kahte
        we:Abs Refl-hit
        "we hit ourselves"
        or "we hit each other"

What do people think?

Matt.