Re: Language comparison

Tristan -

> Nevertheless, you've totally skipped the guts of your posts---we all
> know that all languages can express the same concepts (I'm not sure
> that Turing completion is a relevant concept to human languages,
> excepting ones like Lojban).
All languages that meet that minimum standard - being equivalent in
power to what a natural creole would have. I don't think all conlangs
meet this, not so much by fault of design as by completeness. I am
more interested in comparing fully fleshed-out languages, natural or
otherwise.

> *Why* are some languages better than
> others? *What* makes some languages better than others? *How* did this
> come about? *How* much variability is there in natural languages'
> goodness?
*nodnod* These would be the followup questions, if you accept the
premise that there is qualitative rather than merely arbitrary
difference. I'm not honestly sure that this *is* widely accepted.

> Otherwise all you've done is posted an ill-thought-out flame.
> (If you planned to use this as an opening to promote some
> auxlang---don't. We don't care.)
Heh. Hardly. I'm not an auxlang fan, although I have been very
interested in the concept of designing "ideal" language(s) - mainly to
push the envelope of what it's possible to do with language in various
ways.

E.g., from a recent discussion with a friend of mine: all written
language I know of is serialized, yet mind-native thought seems to be
non-serial in nature. Since spoken language must be serial, perhaps a
written language that is not particularly geared towards
serializability would be a useful representational tool to better
"think in" and record thoughts with. I've certainly had problems of
having thoughts that suffered from translation into language, which I
then had difficulty remembering or manipulating because I lacked a way
to formalize them into language without that loss.

> (BTW: I would be wary about describing this as a hypothesis. To my
> knowledge, hypotheses should be testable and falsifiable, but you've
> made a relative assertion. What you have here is an opinion/idea, which
> may be crafted into a series of hypotheses, which would then be tested
> and bring us to some conclusion about the validity of your opinion.)
It's provable, I think, inasmuch as it is "provable" that a computer
language that supports function spawning is necessarily superior to
one that does not. But to do this, one has to compare features... and
I've rarely if ever seen people try to do this with human languages.
Usually (IME), the comparisons are qualitative, or done just by
aesthetics or complexity/regularity of syntax.

So, if you *can* break up human languages in that manner - and say,
for example, that Language X supports  feature Y- then you would have
a basis for an actual proof, if you can also proove that feature Y
somehow improves the power of the language.

Granted, that part is difficult, and there are bound to be tradeoffs,
but I think it's okay as a theory.

> I think we all agree that not all conlangs are equal. Even assuming all
> words in Old Føtisk were created, you'd be hard pressed expressing
> yourself in much more complex sentences then "The dog ate the man's
> cat".
Then it probably wouldn't meet the minimum test...

> And even leaving that aside, if modern linguists flatly reject
> the concept (because it's false), I don't see how it's critical to a
> conlanger. If there is some alternate universe in which all bridges are
> equal, why should bridge-builders and engineers bother themselves with
> questions like: 'But what happens if a bridge made out of mobile phones
> is better than one made out of steel?!'. If, OTOH, modern linguists
> flatly reject the concept because political correctness prohibits them
> from discussing the issue, but, as a matter of fact, some languages are
> not on-par with Classical Latin, then it is an issue for *all*
> linguists amateur or professional, and not just those who are
> interested in constructing them.
*nod* Agreed. IMHO, it ought to be an issue for *all* linguists, but
admitting that there are *any* qualitative, value-judgable differences
between languages would be arguing against what the trend (as I've
seen it) has been - to defend quote-unquote "simple" or "primitive"
languages as being fully valid and equal in status to any other,
because they too have fully functional grammars, complex syntax, etc.

However, this feels like an argument of *sufficiency* rather than
*equivalency*. I would certainly agree that all natural languages are
*sufficient* for their users' purposes and are (with enough use)
grammatically etc. complete. But this does not answer my question of
whether some might be better than others.

Specifically for conlangers, the importance of this is that it would
imply a list of features - as you have in computer language design -
with various well-known tradeoffs or wins for each feature. E.g.,
something akin to "implement pointer arithmetic, and you introduce a
strong low-level control at the cost of bugs". I feel like the way it
is done now is more or less arbitrary; you know that you can do things
in a variety of ways, but the choice between them is arbitrary or
aesthetic.

Of course, this is only relevant for those like me who want a sort of
hacker's conlang. ;-)

> Also, please leave HTML where it belongs: In (parts of) files bearing
> the MIME-type text/html.
Oops. Sorry, habit. :-P

 - Sai

Replies