Re: Language comparison

Sai replied privately to the mail which I sent just to him in error
instead of to the list; I subsequently sent the mail to the whole list. I
hope Sai does not mind, but I think as my mail was intended for the list,
I'll reply to Sai's reply publically on the list (if you see what I mean).

On Friday, January 7, 2005, at 09:16 , Sai Emrys wrote:

>> Quite frankly, I am a little disturbed by your example - but I will
>> assume
>> it was not meant as flame-bait.
>
> See the separate reply about it.
OK - yes, I appreciate I misunderstood your example.

[snip]
>> But computer languages have been developed for very much more restricted
>> purposes. Conlangs are developed for all sorts of purposes. Whether the
>> choices made by designers of computer languages or conlangs are arbitrary
>> or not is debatable.
>
> ... in fact, it's *this* debate. :-P
Right - hence the confusion. So we are talking about 'languages'
specifically designed by humans (formal languages), rather than about
natural languages, are we?

So your hypothesis is that some _formal_ languages are better than others,
  is it. If this is so, I would not disagree - particularly as regards
computer languages. Evaluating certain types of conlangs is probably more
problematic.

>> Yes, but to evaluate the implementations one must know the goals -
>> otherwise the evaluation is fairly meaningless and is itself arbitrary.
>> What I was trying to say is simply that the goals of computer languages,
>> of conlangs, and of natural languages are simply not the same. To apply
>> to
>> a natural language the type of evaluation one might give to a computer
>> language is IMHO inappropriate and misleading.
>
> Where did I say that we should? (Perhaps, next time, I would do well
> to also leave out any references without giving extensive diffs about
> what parts of the implied analogy I am using...)
>
> While I think that some criteria for computer languages would be
> useful as applied to human ones,
Very limited, I think.

> I have consistently maintained that I
> believe the choice of criteria themselves to be axiomatic - and thus
> not something to debate.
Are these the criteria for evaluating computer languages, or for
evaluating language in general. If the latter, then it is something one
could debate.

>> Sounds like an 'engelang'. I suggest concentrating on engineering your
>> conlang. Not only does it keep things on topic, it avoids flames    :)
>
> This wasn't meant to be about that particular project,
Maybe not - what I meant is that the approach you appear to be taking
would, if translated into a conlang project, seem to suggest an engelang.
I was suggesting, in fact, getting 'on topic'    :)

> though, but about the larger point of whether it is *possible* to make
> objective
> decisions about language design.
Of course you can. How else can you go about designing an _engineered
language_, for goodness sake?

> Unless you can say that one decision
> is better than another - and hence, that the resulting language (all
> things being equal) is better than what would have resulted with the
> another decision - then there is no real basis for decisionmaking.
I do not think anyone would disagree with this.

But your initial hypothesis was: "some human languages are better than
others".

Decisions made in engelang may have very little to do with anything in
natural human languages. In artlang there will inevitably be some, often a
large, subjective element and therefore some arbitrariness in choice; but
even so the language will usually impose its own constraints so that not
all choices are arbitrary. With an a_posteriori auxlang one will make
choices dependent upon features in natural languages; one will choose such
features as seem to the designer better to the designer. In other words,
the designer will find _certain features_ of some natural languages to be
better for his/her purpose than those of other natural languages. S/he
will not however find one natural language itself better than another. The
designer of an a_priori conlang is likely to be guided more by logic and
the taxonomies of science and to be work in a very objective way, more
akin to the designer of an engelang.

>>> I should clarify (again) something that I think you mistook about what
>>> I was saying; I am not - necessarily - saying that there exist natural
>>> languages that are better than each other (though this is a
>>> possibility).
>>
>> Yes - but in what wat "better"?
>
> For any given [axiomatic] set of goals/criteria and ranking thereof,
> "better" is that which maximally fulfills those goals. If the goals
> are in conflict, then the ratings will imply the desired balance-point
> between them.
Yes, yes - this works fine for formal languages - the where people like
Mach, I & some others disagree strongly is the idea that natural languages
can be successfully evaluated in such a way.

>> Well, there are surprising oddities like Pirahà - but it does very much
>> stand out as an oddity and probably more research needs to be done.
>
> Never heard of it; elaborate?
It would take quite a bit to elaborate, and threads do get a bit tedious
if unduly prolonged. I did download a very interesting article by Daniel L.
  Everett but unfortunately I do not have the web address. Do a Coogle
search for "Cultural Constraints on Grammar and Cognition in Piraha" - it
is 52 pages long, and I think you will find it interesting.
>
>> Quite so! The goals of computer languages are, as far as I can see, not
>> the same as those of human languages.
>
> You're implying that the goals of computer languages are all the same
WHAT!!! Where have I said that?

All I said was that the goals (please note the plural) of a subset of
formal languages (i.e. computer languages) are different from those of
natural languages. I did _not_ say that the goals of all computer
languages are the same. Even I noticed the the goals of Assembly languages,
  of BASIC (in all its multifarious dialects), of Pascal, of C, of Ada, of
C++, of Java, of Prolog, of LISP and of the various scripting languages
are not all identical. I did actually get my Master's degree in computer
science and did best in programming and next best in compiler writing - so
give me some credit, please!

> - something you were railing against when applied to human ones. I
> think that the same standard should apply. ;-)
They cannot. The primary purpose of computer languages is to give the
human a means whereby s/he can instruct the machine, using (ultimately)
binary code. That is not by any stretch of even the wildest (tho still
sane) imagination the primary purpose of natural language.

The differences in computer languages are well known and I will not bore
the list by rehearsing them here. But the goals are known, generally
agreed and can be evaluated.

This is simply not the case with natural languages.

>> What are the goals of natural human language and in what ways odo,
>> apparently, some fail to meet these goals.
>
> You just contradicted it again - by implying that there is a single
> set of goals.
Sorry - it was _you_, not I, who wrote: "some human languages are better
than others". It is you who maintained above that the same standard should
apply both to formal computer languages and to natural human languages.
Therefore, I was asking you to clarify what are the sets of goals by which
natural human languages are evaluated.

But in view of what you write below, perhaps it is best not to pursue this
and time to bring this thread to a close.

>> Eh? While I understand why 'aesthetically pleasing' may well be the goal
>> of some artlangs, I fail to see why this is a goal of natural languages.
>> In any case, what some find pleasing others don't. Some people, for
>> example, find Welsh beautiful, others find it ugly.
>
> I said nothing about "natural languages" as distinct from constructed
> ones, for that. I was talking about the whole domain of human-use
> languages (though I believe the *natural* ones to be mostly equal to
> each other).
This thread is becoming a little tedious. It is quite clear that there has
been misunderstanding. You have made it quite explicit above that you you
were talking about the whole domain of human-use language, i.e. both
formal languages (created by humans) and natural language (evolved through
human usage).

There seems in fact to be quite an area of agreement between us:
- we both agree that computer languages can be evaluated and that some are
better than others;
- we are probably largely in agreement that the same applies to conlangs.
As I see it: as far as engelangs are concerned, (nearly) all criteria are
objective and can therefore be evaluated; with artlangs there will be more
subjectivity, and therefore evaluation will have certain subjective
element; I certainly agree that auxlang can be evaluated but I also know
from personal experience that doing so sadly but inevitably provokes
flames.
- as regards natural languages, you have written that you believe them to
be mostly equal to each other. That is good enough for me.

[snip]
> Perhaps so. I mean language to be any system designed to
> explicitly/intentionally convey information.
As I have said, the confusion has arisen because we have not been using
the term 'language' in the same way. If you are in broad agreement with my
summary above, I suggest we bring this thread to a close.

Ray
===============================================
http://home.freeuk.com/ray.brown
ray.brown@freeuk.com
===============================================
Anything is possible in the fabulous Celtic twilight,
which is not so much a twilight of the gods
as of the reason."      [JRRT, "English and Welsh" ]