Message-Id: <199510211049.GAA25567@locke.ccil.org>
Date:         Sat, 21 Oct 1995 06:44:47 -0400
Reply-To: Logical Language Group <lojbab@ACCESS.DIGEX.NET>
Sender: Lojban list <LOJBAN@CUVMB.BITNET>
From: Logical Language Group <lojbab@ACCESS.DIGEX.NET>
Subject:      Re: lojban, lip reading and cmavo
To: John Cowan <cowan@LOCKE.CCIL.ORG>
Status: OR

basically, we readily admit that Lojban has very low redundancy comapared to
most natural languages, perhaps too low for noisy channels.  But the problems
that you raise with cmavo are unsolvable for several reasons.  One is that
too many strings of Lojban words have some grammatical meaning - leave out
some cmavo, and it just changes the grammar rather than becomes an "error"
that one can recognize and intuitively correct.  Indeed, the whole scheme
of elidable terminators is based on the ideas of error correction to
resolve ungrammatical texts, and the wide range of elidability comes at the
expense of redundancy, because the only limit on elidability is that
the result have only one interpretation that can be recostructed by
reinserting elidables from left to right as needed.

The other problem is that even solving cmavo, you need only look at
lujvo to see that the problem is truly intractable.  Some 80% of rafsi have
meaning, and that means that over 60% of ALL possible 2-part lujvo are
theoretically meaningful, and missing a rafsi is a biot worse for understanding
than missing a cmavo.

It is the need to reintroduce redundancy into the language that led to our
philsophy that all possible lujvo forms for a tanru are considered the
same word, and hence you can minimize redundancy problems by using the
long forms of lujvo: CVCCy-CVCCV, etc.

AS for the specific problems you raised with lip-reading - I suspect that
Lojban has it far better than, say, Chinese.  And Lojban has SOME alternatives.
In addition, I think you have overstated the problem.

Specifically, if you are pronouncing the Lojban vowels properly, and
enunciating, I suspect that /e/ and /a/ are not particularly hard to tell
apart (though since I have never lip-read, I'm not sure I can do it).
I would be more worried about /a/ and /y/, especially unstressed /a/ as
spoken by English speakers.

The 5 basic Lojban vowels are the most distant from each other in formation
quality and hence in mouth position.  i and u are high vowels, e and o
are mid vowels, and a is an open/low vowel.  Any attempt to enunciate these
will thereby clearly lead to a difference in jaw opening for the letters
/e/ and /a/, comparable to the difference between /o/ and /a/.  To claim
that a lip-reader cannot read Lojban vowels effectively says that it is
impossible to read any language that has 5 or more vowels, because it is
impossible to have 5 that are more distinct in position.

As for the apostrophe - we have that one covered too.  The definition of
the apostrophe is that it is a "devoiced glide", NOT that it is an 'h'.
But just as Lojban can be perfectly understandable using an 'h' for the
apostrophe in all circumstances, you can also insert other glide-like
sounds and get the same effect.  The sound of English semi-vowels y and w
thus can serve in a pinch, the same function as 'h' does; indeed on of
the first Lojban students INSISTED on using 'w' instead of 'h' just to
push everyones' buttons.  WE do not teach this to English speakers,
because English speakers are already prone to diphthingize the LOjban vowels,
 which
are supposed to be pure vowels sounds (by this I mean that many people in
speaking Lojban will make their /o/ sound vaguely like it has a short /u/ or
/w/ on the end.  /h/ is a lot harder to cause diphthong problems.

But if I was having trouble communicating with a lip-reader, I might try
using 'w' for the apostrophe (keeping it short and unvoiced), and I suspect
that would show up better than /h/does.

Try this with your lip-reading friend.

lojbab