SUO: Re: Monosemy, Semantics, and Natural Language
I received an offline note, which stated an extreme
version of a position that some people might actually
believe. Following is my response.
John Sowa
_____________________________________________________
Yes, there are certainly many legislated concepts, and
they are valuable for many purposes, but there are limits
to the possibility or desirability of legislation.
> I cannot imagine that there is anyone involved with
> ontology efforts, let alone any attempt to integrate theory
> representation without legislating (standardizing) concept
> meanings. The only question is what discrimination features
> will be used initially to bracket category granularity.
> Thereafter concept definitions will be defined much as
> Webster defined words, but without allowing 10-25 different
> meanings. Hereafter we will go to the ontology reference
> much as we went before to Webster.
Let us take the prototypical mathematical concept: Number.
Originally, it meant the natural numbers starting with 1.
Then the Indian mathematicians invented a brilliant
extension, which the Arabs adopted and popularized: 0.
Later, mathematicians extended the concept to include
negative numbers, imaginary numbers, complex numbers,
quaternions, etc.
If mathematics isn't sufficient to convince you, just go
to physics, another very precise subject, and look at the
way the meanings of atom, energy, heat, force, mass, etc.,
have evolved over the years. An atom was originally
defined as an indivisible (a-tom or not-cuttable) piece
of matter. But when electrons were discovered, people
realized that atoms were composite. Then they discovered
the proton and later quarks and gluons. Look at energy
and matter, which were originally distinct, but which are
now considered to be interchangeable. And now perhaps
people have discovered the Higgs boson, which is responsbile
for giving matter its mass.
If you want something more relevant to computers, just look
at the diversity of the meanings of operating systems and
their components, as they vary from IBM mainframes, to Unix,
to Mac, to Windows, and to the diversity of programs that
run all kinds of embedded systems. The definitions of files,
records, tasks, processes, threads, etc., are different for
all of these systems -- and they vary with every new release
or patch to every operating system.
The same is true of programming languages, which vary with
every release -- just ask anybody who has ever tried to write
a large system that would behave the same (or even run) when
it was compiled with multiple releases of the same compiler,
much less with Microsoft's C, IBM's C, Borland's C, Gnu gcc,
etc. Or look at the SQL standard, which ISO has legislated
as an International Standard, and then talk to anybody
who has tried to convert a large database running Oracle's
version of SQL to or from IBM's version or Microsoft's
version or many, many others.
You can also look at the Academie Francaise, which tried
to legislate the meanings of French words (not in anything
nearly as precise as needed for computers, but they tried).
The result was that they stifled the growth of the French
language, which resulted in large amounts of French slang
that never get into French dictionaries and a wholesale
borrowing of English terms to supplement the official
French. The Academie is busily trying to legislate new
terms, but they can't stem the tide.
Please note that none of this is the result of language
or notation. Meanings grow and change with every new
development of any kind. The only way to prevent meanings
from changing is to stop growth -- i.e., to destroy all life.
Summary: Your hopes of legislating the meanings of a
significant number of concepts are (1) hopelessly unrealistic,
(2) impossible to enforce, and (3) a disaster waiting in
the wings for anybody foolish enough to attempt it.
Growth and change are inevitable. I suggest that you learn
to live with it, accommodate it, take advantage of it, and
rejoice in it.