AN-Biblio and ESfO-l-Biblio

Chaumont Devin devil at lava.net
Mon Oct 4 09:11:46 UTC 1999


Doug Cooper <doug at th.net> writes:

>At 21:24 2/10/99 -0500, Jeff Marck wrote:
>>Neither the Austronesianists nor the ESfO people have ever submitted
>>many new titles through the web sites and I am told this is a common
>>experience with other academic organizations.

>Yes, this inexplicable reluctance to engage in self-promotion contradicts
>most of what we know about human behavior.

If I am not mistaken, I have offered my manuscript on Panlingua (universal
subsurface language) on this mailing list, but not many orders resulted.
It comes in 13 e-mail messages, and covers everything from reasons for
believing in the existence of Panlingua, to assumptions and theorems, to
how to implement the results on a computer.  One of the exciting things
about this new theory for linguistics is that it provides the means to
forge a template against which all languages can be compared, which means
that language definition can be brought to a more precise and rigorous
level.  In other words, in everyday terms, a lot about Panlingua can be
deduced simply by observing the way languages behave, but there are
certain characteristics of language which resist this kind of analysis.
Some of these may yield before analyses involving larger numbers of
languages, whereas others may stubbornly remain obscure.  Whatever the
case, we need strong agreement wherever possible in order to create an
acceptable standard and definition for Panlingua, employing arbitrary
choices wherever necessary to fill in the less certain areas until more
advanced means can be found to discover precisely how the human brain is
doing things so that we will be able to discard these arbitrary
definitions in favor of certain knowledge.  Meantime, once a language has
been defined in terms of Panlingua, it will be possible to generate
sentences in that language automatically from a list of Panlingua
representations in order to prove that, at least insofar as this standard
list is concerned, the language has been defined correctly.

This theory has been implemented computationally for English, and works
very well.  We offer a computer program called Brainchild1 which can parse
English sentences, set up representations for them in a computer
implementation of Panlingua (you will understand why I dare make this
claim when you have read my manuscript), and generate English paraphrases
having meanings identical to the original input from these Panlingua
representations.

This computer program (Brainchild1) includes an ontology and lexicon
formerly marketed under the name, SEMLEX, which is still available
separately for linguistic researchers interested only in that.  It enables
one to define a language, not only in terms of dictionary definitions, but
in terms of the semantic relationships between words, thus capturing also
the cultural essence or foundation of the language (what native speakers
tend to believe about the world).  I am convinced that it is crucial that
we should make sure all disappearing languages are defined in this way,
because the problems of computational linguistics have clearly shown that
dictionaries and grammars--at least traditional ones--simply do not
capture enough information about a language to reconstruct an accurate
model.

The fact that Brainchild1 can encode meaning in a precise interlingua and
then generate accurate paraphrases from this representation offers
concrete hope that it will be possible to translate accurately between
languages using machines.  As a matter of fact, I have a strong personal
interest in this problem, having been involved in Bible translation in the
past, and I would be more than willing to provide advice to anyone
interested in attempting a pioneering effort using a Panlingua-based
system.

But language translation is not limited to academic or philanthropic
considerations.  The current population of Earth is about six billion, of
which there are probably about five billion who continually need
information of various descriptions available only in English, which they
are unable to understand.  A translation service fee amounting even to
just one cent per year from each of these five billion people would total
fifty million dollars per year.  So a market is definitely out there for
linguistic entrepreneurs.

My manual on Panlingua is available free of charge upon request.

Brainchild1 costs $200, but this price includes unlimited opportunity to
upgrade to latest version for %10 of current price at time of upgrade.  I
decided to make this offer because the English parser is clearly a
long-term project during which parsing capability will improve
dramatically over time.  In this way clients will not feel cheated when
they find that next year's parser is ten times better than the one they
bought, because they will be able to have that new parser for only %10 of
what new customers are paying.

SEMLEX is available for $100, which includes the right to upgrade to
Brainchild1 at any time for an additional 50% of current market price (an
additional $100 at this time).

My advice for persons interested in this approach to language is to begin
with SEMLEX, which will be valuable no matter whether you accept the
existence of Panlingua or not, because the idea of a lexicon and ontology
is commonly accepted everywhere.  Semlex enables you to set up your own
three-character mnemonics for all parts of speech, and lexical and
semantic link types.  Using SEMLEX, you will be able to define your own
parts of speech for any language.  The same applies for lexical link and
semantic link types.  And not only this, but any of these POS or link type
definitions can be added, deleted, or changed while the program is
running.  SEMLEX automatically does global upgrades on your data to
reflect all changes.  It will also enable you to move mnemonics to new
positions within any mnemonic list.

SEMLEX comes in two versions--a short version that allows only words and
semantic nodes having ordinal identifiers in the range 1-65535, and a long
version having a range that extends into many billions.  The advantage of
the short version is that less computational overhead is required for it
to run.  The advantage of the long version is that it is easily capable of
handling all of the words of any language on earth.  Both run on IBM PC
compatible machines having the now-standard 640kb of base memory.  Both
versions are very fast, with no noticeable delay, even for ontology
searches requiring many traverses of semantic linkages.

Please feel free to read my manual on Panlingua, which is 100% free of
charge, and will not ensnare you in any kind of advertising campaign.  I
make no attempt to hard-sell any of my computer products, assuming that
most people would prefer to buy something readily available instead of
rewriting everything from scratch, just like a construction man might
prefer to purchase a new home instead of building his own.  Not only this,
but if you should choose to write your own versions of the programs I have
offered above, I am prepared to go the second mile and offer free advice.
The only area in which I will not do this is in parsing technology,
because to do so would jeopardize the future of my own enterprise.
Everything else is open and free.

Thanks for your interest.

With best regards from Honolulu,

Chaumont Devin.



More information about the An-lang mailing list