Re: Lojban and case theory{ - response to Bruce and others

[And Rosta <cbmvax!uunet!UCL.AC.UK!pucc.PRINCETON.EDU!ucleaar>]

[problems with my mail system mean that I can't read the most recent
conlang postings. So apologies if I am saying what has already been
said or appear to be ignoring others' contributions.]

(1) CASE GRAMMAR. Anyone doing a library trawl for studies on semantic
case will find much information under its various aliases:
	semantic roles
	thematic roles
	thematic relations
	theta roles
	[greek lowercase theta]-roles
The terms are given in order of increasing theoretical specificity.
'Semantic role' is the neutral term.

The recent discussion on this list has been concerned not with *case*,
which, strictly speaking, is a morphosyntactic feature (though Fillmore,
John Anderson & Starosta use it additionally as a semantic feature), but
with the way the grammar should represent *semantic roles*.

(2) COLLABORATING ON A CONLANG. An a posteriori language, being defined
in terms of its differences from existing languages, can be invented in
two minutes (though one can obviously take a lot longer). An a priori
language equivalent to a natural language in expressive scope could take,
at a guess, thirty years: I have been working (on my own) on Sta for
15 years & still have no proper lexicon.

(3) REPRESENTING THE TYPE OF SEMANTIC ARGUMENTS IN LOJBAN.
I certainly agree that given our present state of knowledge we cannot
say that languages involve a small or enumerable number of semantic
roles. Therefore, if BAI selma'o (subclass of words) is a *closed*
category (like prepositions in English) then either:

(a) BAI cmavo must be given prototype definitions (meaning features
that are sufficient but not necessary), and the BAI cmavo must be
lexically specified by the brivla ('content word'); or

(b) BAI cmavo are not the appropriate means to represent semantic
roles. An open class of predicate words (gismu or lujvo) should be
used instead. Lojban already has some gismu that express semantic
roles: _gasnu_ ('do, be the actor of') is one example.

Lojbab writes that:
> - the construction of Lojban tanru and lujvo allows for a hypothetical
> lujvo to have an infinite number of rafsi components and hence places,
> and hence no case system could exist in one-to-one correspondence to the
> set of places of such an abstract predicate;
But this is merely an (inconclusive) argument against using a closed
class of words to represent semantic roles.

My overall response to Lojbab's various arguments is that Lojban already
represents the lexically-specific and lexically-specified semantic roles
of a predicate *by word order*. Basically the argument comes down to
whether expressing the semantic roles lexically makes the language easier
to use. I think it does: it is easier to remember the word for marking
goals or agents than to remember the place on some particular predicate
where the goal or agent role is encoded.

You would have to learn more words, but you wouldn't have to learn the
place-structure of each predicate.

The fact that one might wish to use the same role-marker for recipients
and destinations does not necessarily impose a metaphysical or English
bias on the language, providing that one realizes that the precise meaning
of the role marker depends on the meaning of the predicate (e.g. giving
versus going). However, I do incline to the view that the present
Lojban approach is less metaphysically biased but much harder to use.

Furthermore, does one necessarily wish to avoid metaphysical bias?
I would always wish to be able to say that something is at the "back"
or "front" of my mind, or that I am in "high" or "low" spirits. I
might wish to avoid distinguishing recipients from destinations and
treat them as the same thing; I might want to treat possession as a
kind of location, say.

---
And