Re: 'case' and Lojban

[cbmvax!uunet!math.ucla.edu!jimc]

I think lojbab has neatly knocked into a cocked hat the Institute
initiative on case tags.  (We'll deal with diklujvo later.)  I sent
off a reply to a message from Art Protin but on re-reading lojbab's
message I thought it might be of interest to the whole membership.

My feeling on cases is this:  There are equivalence classes of "places"
such as the actor, the victim of a transitive relation, and the event
(abstract sumti) place of a word like "want".  Each of these has many
members and would be instantly recognized as a "case" by one inclined
to do so.  But there are also very specialized gismu and, thus, "cases"
with only a few members.  The "reference frame" and "culture" cases
come to mind, at least for -gua!spi.  I don't want to go through the
word list right now, but I expect that there are several tens of gismu
with places which cannot be fit credibly into any set -- that is, they
are literally unique cases.  It's this kind of case that drives case
grammarians into a screaming frenzy, and wrecks attempts to enumerate
a specific list of case tags.  Not that it bothers me.

In any case, neither Lojban nor -gua!spi contemplate abandoning the
numbered places, nor ordering those places in any particular way cued
by cases.  (Yes, the actor is usually first if there is one.  Usually.)
The closest we come to case tags is sumti tcita (modal operators), for
adding a case to a word that doesn't ordinarily have it.  Note that
with fi'o <bridi> /fe'u/, any selbri potentially can contribute a new
case in this way.

Thus Lojban turns its back on the traditional baggage of case grammars.
Nonetheless, I find "case" to be very useful as an organizing principle.
But Lojbab wants to avoid useless arguments with grammarians, and hence
avoids using the term "case" in the loose sense that I do.

		-- jimc