Re: Language superiority, improvement, etc.
From: | David G. Durand <dgd@...> |
Date: | Friday, October 16, 1998, 2:26 |
>On Thu, 15 Oct 1998, charles <catty@...> wrote:
>We will get nowhere with these inappropriate analogies.
Frankly, I think this discussion is getting nowhere. We can probably
optimize a lanaguage for any definable and testable property that a
language can have, but are unlikely to secure general agreement as to what
forms of "optimization" are worth carrying out.
Some perennial topics of this sort on CONLANG are self-segregating
morphology (trade off of coding efficiency and word-form rigidity against
machine parsing ease, and arguably human parsing ease), compression
(huffman-style minimal encodings can increase the bandwidth of speech, but
trade off redundancy and intelligibility in noisy environments), semantic
relations (philosophical-style taxonomic vocabularies make finding the word
for a thing easy -- as long as we can agree that the ontology is correct,
and it doesn't need to be revised), and derivational flexibility (starting
with verbs (or nouns) one can derive many vocabulary items by means of
regular derivational functions).
These are all attractive to some and useless or foolish to others. None has
secured universal acclaim from all. Some aspects of language in use are
fairly limited however: The complexity of possible human thoughts probably
has an upper bound, and that bound seems to be met by language in most
cases (possible/probably exceptions in the arts and mystical experience are
noted). The complexity of human brains (and thus the languages they can
understand and produce actively) are limited.
The fact that languages all satisfy the design criteria of being usable by
humans, and adequate to express human thoughts seems to make comparisons
silly, if not invidious. Human communication all has recourse to
meta-linguistic functions that let us define new words, use analogies, and
so forth, thus extending our linguistic systems on the fly.
>Language when seen as a tool is obviously improvable,
>and some features are better or worse, and overall utility
>is more or less, according to the user's priorities.
Right. Superior and inferior imply a universal standard. I won't claim that
such a thing couldn't exist in some world, but I see no evidence that even
any one linguistic micro-point of potential evaluation has secured any
agreement even in this small group, never mind the kind of global
evaluation that would be required to judge whole languages against one
another.
>A better analogy is a car or programming language.
>Sure, you can do almost anything in Fortran or a Turing
>machine, but it would be silly to say all such
>languages are equally good.
Depends on the application. For proving certain kinds of theorem, Turing
machines are ideal, because they are so simple (and yet powerful). For
numerical programming, Fortran has had practical advantages for decades
because social effects reinforced a situation where the best numerical
libraries and optimizing compilers for numerical programming were both
Fortran related. Similar tradeoffs explain why so many other programming
languages exist.
In Goedel, Escher, Bach, Doug Hofstadter defines a programming language
that generates only the recursive functions (not Turing complete). He notes
how _hard_ it was to define an even remotely useful-seeming language that
was _not_ Turing complete.
> And you can ride a donkey,
>but horses are demonstrably *better*.
Not on a mountain trail, where a donkey's smaller size, toughhess, and
balance make it more effective.
>As for natlangs, over time they must tend to a common,
>rough equality of overall usefulness, for reasons far
>beyond the scope of my laziness to even attempt to detail.
>Lexicon can expand at will. Grammar can be a mess though,
>and if one were to search for the "best" inter-language,
>some would serve better than others depending on criteria.
>It is unlikely that we or anybody else will agree on
>any criteria, though. So all our languages are safe ...
In the absence of any evidence that such criteria even exist, talk of
"optimizations" is a waste of time, without a clear definition of the basis
of measurement.
Personally, I think we may have reached closure in the sense that no
opinions are being changed, and the positions are very clear.
-- David
_________________________________________
David Durand dgd@cs.bu.edu \ david@dynamicDiagrams.com
Boston University Computer Science \ Sr. Analyst
http://www.cs.bu.edu/students/grads/dgd/ \ Dynamic Diagrams
--------------------------------------------\ http://www.dynamicDiagrams.com/
MAPA: mapping for the WWW \__________________________