Message-Id: <m0qmlo8-00005XC@xiron.pc.helsinki.fi>
Date:         Mon, 19 Sep 1994 10:16:24 -0600
Reply-To:     Randall Holmes <holmes@DIAMOND.IDBSU.EDU>
Sender:       Lojban list <LOJBAN%CUVMB.bitnet@FINHUTC.hut.fi>
From:         Randall Holmes <holmes@DIAMOND.IDBSU.EDU>
Subject:      Re: Chief logician?
To:           Veijo Vilva <veion@XIRON.PC.HELSINKI.FI>
Content-Length: 3499
Lines: 69

Goodness, I decided against that post even going out; it smacks too
much of an invitation to political argument.  Betrayed by my mailer...

If Loglan or Lojban becomes a language really used, the decisions of
an academy will become about as relevant as those of the Academy which
imagines that it legislates for the French language, regardless of
what any individual thinks.  There is a proper function for an
advisory body in the area of correct logical usage, however.  My
suspicion is that the logical features of the language (either
language) would collapse under the pressure of widespread usage, and
will become closely analogous to "bad" NL usages in hard cases; the
_option_ of logical clarity will remain.  Machine parseability (and
maybe the logical usages to some extent) could continue to be enforced
if a major part of the speech community consisted of computer
programs.

A project which both languages should consider is the mechanization of
not only the grammar of the language but of the allowed logical
transformations; this would make it possible for interaction with
machines to enforce the logical usages.  In the limit, the
construction of a theorem prover with (possibly subset) Loglan/Lojban
as the input language should be considered.  I have indicated to TLI
that work of mine (not motivated by Loglan/Lojban) may make it
relatively easy to construct such a program in a few years; it is
probably already possible with "off-the-shelf" techinology using
existing theorem prover systems.

Note that I put "ready for use" in quotes (I think I did).  But, in
fact, the language (the despised TLI Loglan, that is) is ready for
use.  I have done enough translations into it to be fairly certain of
this.  I am not certain that it could be spoken correctly by anyone
without something approaching my own peculiar qualifications, and I
think that the same is probably true of Lojban; the pressure of the
usages of the natural languages is too powerful.  It is far easier to
learn to utter Loglan sentences which parse correctly than to learn
the logical (and philosophical!) background knowledge needed to use
the language(s) correctly.

I won't express an opinion of JCB's linguistics research; I'm not
competent to do so.  My "feel" for both languages is that they are too
similar to the native languages of the experimenters; see above.  If I
were designing a language from scratch, I would have adopted VSO or
even OSV word order (Polish or reverse Polish notation :-) ), for
example.  I don't think that the scientific or non-scientific nature
of JCB's method for contructing primitives, for example, is at all
relevant to the usability of the language.

Negation is fine in Loglan; except that I'm not sure I would have allowed
negation of arguments; this has no analogue in symbolic logic usage
and can lead to very misleading transformations.  Does anyone in the
Lojban community realize that logical connectives applied to arguments
produce problems of scope (usually handled implicitly in NL's)
precisely analogous to those connected with quantification?

Consider

John and James love Mary or Sally

versus

Mary or Sally is loved by John and James

In the second sentence, but not in the first, it is clear that John
and James love the same unspecified element of {Mary, Sally}; in the
first sentence, they may love different elements of the set.


                                        --Randall Holmes
                                        pet logician, TLI :-)