Message-Id: <199510202357.TAA02850@locke.ccil.org>
Date:         Fri, 20 Oct 1995 16:55:02 -0700
Reply-To: "Peter L. Schuerman" <plschuerman@UCDAVIS.EDU>
Sender: Lojban list <LOJBAN@CUVMB.BITNET>
From: "Peter L. Schuerman" <plschuerman@UCDAVIS.EDU>
Subject:      lojban, lip reading and cmavo
To: John Cowan <cowan@LOCKE.CCIL.ORG>
Status: OR

coi

I was showing some Lojban instructional materials to a friend the other
day, and she brought up an interesting point.  She's an instructor of deaf
children, and she pointed out that, unfortunately, a good deal of the
vocabulary would be almost impossible for someone to lip-read.  I thought
I'd pass along her comments for discussion.

Three problems she saw were 1) V1'V1 constructions, 2) use of /a/ and /e/
to distinguish words that are otherwise similar, and 3) the use of "a'i"
vs. "ai" (and "e'i" vs "ei", etc.) to distinguish words.

According to her, word pairs like those shown below would be extremely
hard to distinguish because the sound /h/ doesn't "show up" very strongly,
if at all, as a movement in vocal articulation.  Spoken V1'V1 *looks*
like spoken V1.

a (A)           a'a (UI1)
sumti "or"      attentive

ba (PU)         ba'a (UI2)
after           I anticipate

mu (PA1)        mu'u (BAI)
digit 5         exemplified by

no (PA1)        no'o (PA5)
digit 0         typical value

Also, the pronunciations of words like a, e, a'a, a'e, e'a and e'e are
hard to tell apart in the same way, as are such pairs as di'a/di'e,
fa'a/fa'e, ja'e/je'e, to'i/toi and pa'i/pai.  We did some practice with me
reading and with her watching my lips (with and without her hearing aid
turned on), and she was really only able to tell the difference when she
could hear me.

All of this got me thinking about how natural languages have constructions
that are essentially designed to ensure communication...  like double
negation, for instance (e.g. "I don't have no money").  Logically, two
negatives make a positive but every English speaker *knows* that a double
negation is still negative... and a listener who might miss "don't" or
"no" would still get the message.  Another method is to have "little
words" with a great deal of morphological variety.  In English, the
"little words" (e.g. a, an, the, this, that, which, who, what, where) are
distinct both typographically and phonetically, and even where there are
similarities, the words have a functional relationship (the "th*" group
all used to designate something, the "wh*" group used to ask questions).
To me, this (and other evidence) suggests that natural language tries to
compensate for possible miscommunications.  Shouldn't this be a feature
of an artificial language as well?

There was once an artificial language (I can't recall the name now) that
categorized all of the members of the universe, and reflected that
categorization in the word morphology.  As a (conceptual, not factual)
example, all sky-phenomena started with "kala."  A comet would be a
"kalat", while the moon would be "kalad" and the sun "kalam" and stars
"kalag" and so on.  The problem here is that many things which could
easily be confused with each other are also very similar (and confusable)
in their pronunciation.  A man pointing at the sky at night would have to
be certain to over-enunciate so that listeners would know whether he was
drawing attention to the moon, a nearby planet, a star, a constellation,
or even the sky itself.

The connection that all of this has with Lojban is this: I think it should
be a priority to examine each cmavo with the following considerations in
mind.

1) Does the cmavo have more meaning than its size warrants?  Small words
are simply easier to "lose" in communication, and that is one reason why
people use larger (but still understandable) words when they want to
ensure that their message gets across.  They also use redundancy (such as
double negation, verb conjugation, simple repetition, etc.), but I don't
see how these features could be implemented in Lojban, so the imperative
should be that a cmavo, if mistakenly missed, should be conspicuous by its
absence.

2) Does the cmavo sound similar to words which are, in meaning, similar?
Ideally, homonyms or near-homonyms should have radically different
meanings, so that if a listener mistakes one word for another, the
mistake is obvious from context.

I know no one is interested in changing Lojban at this point, but I
thought I'd bring up these thoughts anyway... I found them interesting
from a linguistic, if not necessarily a lojbanic, viewpoint.

co'o mi'e pitr.                                 plschuerman@ucdavis.edu