From pycyn@aol.com Sun Mar 04 15:44:20 2001 Return-Path: X-Sender: Pycyn@aol.com X-Apparently-To: lojban@yahoogroups.com Received: (EGP: mail-7_0_4); 4 Mar 2001 23:44:20 -0000 Received: (qmail 36668 invoked from network); 4 Mar 2001 23:44:19 -0000 Received: from unknown (10.1.10.26) by l10.egroups.com with QMQP; 4 Mar 2001 23:44:19 -0000 Received: from unknown (HELO imo-m09.mx.aol.com) (64.12.136.164) by mta1 with SMTP; 4 Mar 2001 23:44:19 -0000 Received: from Pycyn@aol.com by imo-m09.mx.aol.com (mail_out_v29.5.) id r.5f.11b58db1 (3997) for ; Sun, 4 Mar 2001 18:44:12 -0500 (EST) Message-ID: <5f.11b58db1.27d42d4c@aol.com> Date: Sun, 4 Mar 2001 18:44:12 EST Subject: Re: meaningless language. To: lojban@yahoogroups.com MIME-Version: 1.0 Content-Type: multipart/alternative; boundary="part1_5f.11b58db1.27d42d4c_boundary" Content-Disposition: Inline X-Mailer: AOL 6.0 for Windows US sub 10501 From: pycyn@aol.com --part1_5f.11b58db1.27d42d4c_boundary Content-Type: text/plain; charset="US-ASCII" Content-Transfer-Encoding: 7bit ko'a ru'a satci bangu .i ko'e ru'a traci bangu .i ro da poi jufra bau ko'a zo'u go da selsmuni gi le xe fanva be da bei ko'e be'o selsmuni i ku'i ma bangu ka satci i ma poi xe fanva da ko'e cu se cipra fi leka selsmuni. Enough! (to show my lojbanic incompetence as well as to get started on the point). xod's proposal looks like only the Logically Perfect Language part of the usual testafiability criterion and so may avoid the problem of self-application (though maybe not). So we have s sentence in English, say, that we suspect of being meaningless. So we translate it into LPL (NOT Lojban, I hope) and check. But how do we translate it into LPL? Maybe an algorithm that always gives a unique and accurate traslation? We know that is impossible for moving from a natural language to LPL, since a natural language sentence is almost always ambiguous and fuzzy. Okay, then maybe an algorithm that, applied to a given situation, gives a unique and accurate translation of the sentence as applied to that situation? Well, maybe indeed, and we can guarantee uniqueness then. But what about accuracy? Well, a translation is accurate, we may suppose without too much cavilage, just in case it means the same thing as the original. But what if the original has no meaning? Will it be possible to have meaningless sentences in LPL and, if so, how is it perfect? Typically, this question gives rise to one or the other of: well there is one meaningless expression in LPL into which all meaningless sentences translate and which, of course plays no further role in the language. Or meaningless sentences don't translate into LPL at all and so the right side in the equation is false for containing a nondenoting expression {le xe fanva be da bei ko'a be'o} (or we can fiddle the equation in some insignificant way, using bound variables rather than descriptions). But all this assumes, to be effective, that we have agreed that the algorithm is correct (this gets worse, of course, without the algorithm but relying on mere skill in translating). And, as soon as one's favorite claim is shown to be meaningless on this test, one withdraws one's agreement, pointing to the result just mentioned as evidence (adequate evidence, yet) that the algorithm is not correct, since it gives here an inaccurate result. So we try another algorithm that works in this case and the cycle repeats itself. But we might have objective tests for accuracy, at least the extensional one that the original and the translation are true in exactly the same cases. Way too weak, since it allows all true sentences of mathematics, say, have the same translation and similarly all the "laws of nature." But more importantly in a practical sense, the person whose sentence was declared meaningless presumably thinks that it is sometimes true and so will know that no translation which makes it meaningless (and thus never true) can be right. We could move on to saying that accuracy means being the same in every possible situation, but -- quite aside from xod's determination not to allow such things (as I understand) -- this notion has little explanatory power (ya gets out what ya puts in), so the problem of rejecting accuracy remains unresolved. And that doesn't even touch the issue of an exact (or LP) langauge. --part1_5f.11b58db1.27d42d4c_boundary Content-Type: text/html; charset="US-ASCII" Content-Transfer-Encoding: 7bit ko'a ru'a satci bangu .i ko'e ru'a traci bangu .i ro da poi jufra bau ko'a
zo'u go da selsmuni gi le xe fanva be da bei ko'e be'o selsmuni i ku'i ma
bangu ka satci i ma poi xe fanva da ko'e cu se cipra fi leka selsmuni.

Enough! (to show my lojbanic incompetence as well as to get started on the
point).  xod's proposal looks like only the Logically Perfect Language part
of the usual testafiability criterion and so may avoid the problem of
self-application (though maybe not).  So we have s sentence in English, say,
that we suspect of being meaningless.  So we translate it into LPL (NOT
Lojban, I hope) and check.  But how do we translate it into LPL?  Maybe an
algorithm that always gives a unique and accurate traslation? We know that is
impossible for moving from a natural language to LPL, since a natural
language sentence is almost always ambiguous and fuzzy.  Okay, then maybe an
algorithm that, applied to a given situation, gives a unique and accurate
translation of the sentence as applied to that situation?  Well, maybe
indeed, and we can guarantee uniqueness then.  But what about accuracy?  
Well, a translation is accurate, we may suppose without too much cavilage,
just in case it means the same thing as the original.  But what if the
original has no meaning?  Will it be possible to have meaningless sentences
in LPL and, if so, how is it perfect?  Typically, this question gives rise to
one or the other of: well there is one meaningless expression in LPL into
which all meaningless sentences translate and which, of course plays no
further role in the language. Or meaningless sentences don't translate into
LPL at all and so the right side in the equation is false for containing a
nondenoting expression {le xe fanva be da bei ko'a be'o} (or we can fiddle
the equation in some insignificant way, using bound variables rather than
descriptions).
But all this assumes, to be effective, that we have agreed that the algorithm
is correct (this gets worse, of course, without the algorithm but relying on
mere skill in translating).  And, as soon as one's favorite claim is shown to
be meaningless on this test, one withdraws one's agreement, pointing to the
result just mentioned as evidence (adequate evidence, yet) that the algorithm
is not correct, since it gives here an inaccurate result.  So we try another
algorithm that works in this case and the cycle repeats itself.
But we might have objective tests for accuracy, at least the extensional one
that the original and the translation are true in exactly the same cases.  
Way too weak, since it allows all true sentences of mathematics, say, have
the same translation and similarly all the "laws of nature."  But more
importantly in a practical sense, the person whose sentence was declared
meaningless presumably thinks that it is sometimes true and so will know that
no translation which makes it meaningless (and thus never true) can be right.
We could move on to saying that accuracy means being the same in every
possible situation, but -- quite aside from xod's determination not to allow
such things (as I understand) -- this notion has little explanatory power (ya
gets out what ya puts in), so the problem of rejecting accuracy remains
unresolved.  
And that doesn't even touch the issue of an exact (or LP) langauge. --part1_5f.11b58db1.27d42d4c_boundary--