Re: Language superiority, improvement, etc.

I've been lurking around conlang for a long time, but with all the
discussion about language superiority, I thought I would put forth my own
thoughts. BTW, I heavily pruned the tree of responses in the following
quotes, partly because I lose track more than a couple of layers deep, and
partly because I dislike repetition ;-)

>Thus I was not saying that value judgments were bad, but only that
>they were not scientific. Science does not advance with opinions, but
>with ideas backed up by facts.  It's an attempt to be objective, not
>subjective.  There is nothing wrong with attributing value judgments
>to things, nothing inherently at least.  It's just that doing so cannot
>be backed up empirically, so one should not be lead into the delusion
>that there is actually any substance to them.
Actually, it can be argued that science only advances when it proves a
theory *wrong," not when it appears correct. It's a process of "creation by
negation" -- you observe something, you make the simplest possible theory
that fits the facts, and then you test the theory (read: you do your
darnedest to disprove it). If the first test fails to disprove it, then you
find some other test. It is precisely when you disprove the old theory and
come up with a new theory that science advances. A theory that cannot be
disproven is as worthless as an aria that can't be sung or a story that
can't be told. What we call scientific truth is just the combined total of
all the theories waiting to be disproved ;-)

>
>So, I'm not disagreeing with you that any tool can be improved, but
>what I am saying is just what constitutes an "improvement" is really
>not an easy issue to settle.  As I just alluded to, many people find the
>issue of cases irritating or difficult, because they are used to a language
>without them.  So, they say that cases are bad.  For the same reason,
>many people like cases, because they *are* used to them, and so
>advocate their use in a conlang.    There is no way to settle this issue,
>really, because syntax and cases are equally good at conveying the
>information, as that is, afterall, the ultimate point of language, to convey
>information and to communicate.
If the ultimate point of language is to convey information and communicate,
then the best language for an individual is the one that allows him, in the
finite span of his lifetime, to communicate and receive the maximum amount
of information. We can extend this to cover the entire human species: The
best human language is the one that allows the maximum rate of information
exchange among all people at each point in time (this is because the
individual componants of humanity have a small, finite lifespan, while the
species as a whole has an indefinite and hopefully very long lifespan). To
be fair, an equally valid definition would be: The best human language is
the one that maximizes the probability of survival of the human species at
each point in time.

Now, if we could only prove that the set of "swiftest communication" and
the set of "most survivable communication" have an overlapping subset --
that's the area *I* would claim contains the best language around.

Michael A. Rouse

PS Shifting slightly away from such lofty goals, making a language that is
both regular and takes into account how children acquire language, while at
the same time being as concise as possible for fluent native speakers of
the language, would go a long way to making an extremely good conlang, even
if not exactly perfect.