Re: [tlug] OT. Linguistic ?

["Stephen J. Turnbull" <stephen@example.com>]

>>>>> "Mauro" == Mauro Sauco <mauro@example.com> writes:

    Stephen> the problem is, how are you going to get people to
    Stephen> remember 20,000

    Mauro> Humans remember a lot more than that when learning any
    Mauro> language.

Not "humans": "baby humans".

    Stephen> If you've got a PDA, why not use its native input
    Stephen> capabilities, and just "spell it out"?  Allow
    Stephen> abbreviations (a la keitai messaging) and get a
    Stephen> predicting parser (see below).

    Mauro> 1. Inputing full lenth words using a keyboard

(1) I'm not talking full-length words; that's why prediction is
important.

(2) I'm also talking pen-based; that's why I specified a SL-C7xx or
SL-C8xx, the SL-5500 won't do that AFAIK.

    Mauro> 3. using less keys for writing is already proven to be
    Mauro> faster than an standard keyboard as you can see in
    Mauro> "taquigrafia" (sorry only know the word in Spanish), is the
    Mauro> typewriters they use in court that only has 14 keys.

Sure.  It's also true that people who can use stenotypes effectively
are a limited subset of adults.  Humans lose much of their language
acquisition ability between the ages of 5 and 12.  So even when you
say "children at the rehab center", they may already have lost a lot
of adaptability.

    Mauro> http://www.esteno.net/images/estenotipia.jpg
                                         ==========
Same word in English and Spanish, it seems.  :-)

    Mauro> 4. Using a glove will give you the freedom to use your
    Mauro> hands while the PDA can be into a pocket.

That's a killer reason for the glove.

    Mauro> Back to " how to remember 20,000": there is a big
    Mauro> difference between remembering numbers and moving your
    Mauro> fingers to convert thoughts into words, the motor-memory is
    Mauro> a very efficient system that does not required the same
    Mauro> type of memory (implicit non declarative memory) than
    Mauro> remember telephonic numbers (explicit declarative memory),

Sure.  The new pen-based systems (even Palm graffiti) depend on a
similar effect.  But think about sign language and kanji.  In both
cases, a limited number of radicals *which typically acquire semantic
content* are used in combinations.  In fact, in kanji there are 214
recognized radicals, and it's estimated that altogether even for all
the historic and mistaken kanji, only a total of about 2500 "stroke
configurations" (of similar content to the recognized radicals) are
required.  I don't know how big the "alphabet" is for sign language,
maybe a couple thousand?, but I rather doubt it's anywhere near as big
as the vocabulary of words that can be (reasonably) efficiently
expressed by combinations.  I know that there are a fair number of
"homonyms" in Japanese sign language, too, with semantics being
determined by context.

Both chess/go masters and musicians develop large "chunked"
vocabularies of such patterns.  But once again they get built out of
smaller parts, which already have semantic content.  So (eg) playing
12-tone music (Schoenberg) is considered to be _very_ hard work by
most musicians; they can't rely on the patterns that they know.  And
chess masters are no better than other people (on average) at
remembering random configurations of pieces.

There's also probably issues of some configuration transitions being
much slower than others; you'll need to avoid those somehow.  So, all
in all, I suspect that most people are going to run into a real
barrier somewhere around 1000 "words", long before you run out of
combinations.

In any case, there are going to be situations where they have to
"spell out" words whose "code" they don't know (eg, they look them up
in a dictionary).  Japanese is both phonetically and phonologically
very regular; I wouldn't be surprised if you could get away well under
200 total phonemes plus inflection operators (for accenting and
changing voice pitch.  That, combined with a dictionary for larger
units that people could learn as they need them, might also be an
interesting way to go.




-- 
School of Systems and Information Engineering http://turnbull.sk.tsukuba.ac.jp
University of Tsukuba                    Tennodai 1-1-1 Tsukuba 305-8573 JAPAN
               Ask not how you can "do" free software business;
              ask what your business can "do for" free software.