Hallo conlangers!

On Sat, 21 May 2011 12:53:20 -0400, Jan van Steenbergen wrote:

> Oh dear, there we go. ;) Well, let me explain a bit what I had in mind with 
> this 
> hexagon.
> 
> Conlang classification has kind of been my hobby-horse for years now. I've 
> come across many different models and what struck me was that none of 
> them were really satisfying:
> * either they use categories that refer to completely different criteria 
> (like 
> f.ex. IALs vs. fictional diachronic languages vs. superset languages vs. 
> logical 
> languages)

Which is of course bogus.  You get gaps and overlaps that way.

> * or they are sort of biased towards auxlangs (for example: a priori vs. a 
> posteriori - somehow I can't stomach the idea of Wenedyk being a close 
> neighbour of Esperanto in any conlang classification)

Yes.  Few linguists concern themselves with conlangs; and of those
who do, most only concern themselves with auxlangs, and come up with
classification schemes which break down when applied to artlangs.
Clearly, Wenedyk is an a posteriori language - but it is a posteriori
in a way very different from Esperanto!

> In my opinion, the most relevant criterion is PURPOSE. And from that point 
> of view, the Gnoli Triangle is definitely the best solution I have ever 
> seen. It does have the great advantage that a language can be placed 
> anywhere on or in the triangle, thus making it possible to quantify the 
> level of auxlanginess, artlanginess and engelanginess on one chart. However, 
> as I see it, it has two disadvantages. 

Sure.  The most relevant criterion for conlang classification is the
purpose the language was created for, not least because the purpose
determines the criteria by which one has to judge whether the conlang
was a good or a bad design.  The Gnoli triangle is based on just that,
though one can argue whether three categories suffice or not.

> First of all, I've noticed that this works for some 95% of all conlangs, but 
> there are several examples of languages that wouldn't really fit anywhere on 
> the triangle. Take Reconstructed Proto-Indo-European as an example: 
> obviously, PIE itself is not a conlang, but any reconstruction of it
> undoubtedly is - just look at the vast differences between various 
> reconstructions!

Indeed.  Any linguist worth his stripes will maintain that such a
reconstruction is just a *model* of a past reality which captures
some facets of that reality well and others not.

> And 
> well, unless a reconstruction was explicitly meant to be used as an auxlang 
> or something, you can't really place it anywhere in the triangle. 

Sure.  The standard model of PIE, for instance, is not an auxlang
("Modern Indo-European" as proposed by those people from Spain is,
but that is not really Standard PIE itself but a language *based*
on it but modified); it is not really an artlang either, and also
not really an engelang.

> Secondly, "engineered" refers more to HOW the language was created than to 
> the reason why it was created, or the purpose it was created for. There is 
> absolutely no contradiction between being a loglang and an auxlang, and 
> being both doesn't make the language any less of either. 

Indeed.  I perceive the term "engelang" more or less as an antonym
of "naturalistic conlang", where "naturalistic" is meant in the
artlang sense (auxlangers and artlangers use the word with somewhat
different meanings!).  Volapük, for instance, is clearly a language
that was designed the way an engineer designs a machine, i.e., an
engelang.  But by *purpose*, it is clearly an auxlang.  Lojban is
perhaps the canonical example of an engelang, yet the LLG (or at
least, several people in the core of the Lojbanist movement) did make
the suggestion of using it as an auxlang, even if that was not the
primary purpose of it.  (Third-party proposals such as using
Quenya as an IAL are not really relevant here; Tolkien never even
thought of that, and we all agree that Quenya is a prototypical
artlang, and a naturalistic one at that.)

> Of course, you can redefine the three categories. For example:
> A) Languages intended to be spoken by as many people as possible (auxlangs)
> B) Languages intended to be spoken by a select group of people (ritual, 
> stealth, micronational, dialect standarisation etc.)
> C) Languages intended to be spoken by fictional people only or not at all 
> (artlangs, experilangs)

This at least gives a useful metric, ranging from languages
proposed to be used by 100% of the world population (world IALs)
via languages to be spoken by a large but defined subgroup of
humanity (zonal languages) and languages for use in a small circle
(stealth languages etc.) to languages not meant to be put into
practice at all (most artlangs).

> Or otherwise:
> A) Languages to be used for communication
> B) Languages to be used for illustration

This way, (A) would be all sorts of international, regional or zonal
auxlangs, natlang reform projects etc.; (B) encompasses fictional
languages (which illustrate the language of a fictional ethnic group),
reconstructed proto-languages (which illustrate a language spoken in
a distant past), logical languages (which illustrate how a language
based on formal logic could work), etc.

> ... but such a thing doesn't really give satisfactory results either.

Indeed, the category (B) is a very wide one; there are basically
three modes of "illustration":

B1) Artistic illustration: artlangs
B2) Scholarly illustration: reconstructed proto-languages,
                            scientific speculations about alien languages
B3) Technical illustration: loglangs and similar schemes

At least the first two have in common that they are not meant to be
used in practice, but serve only to illustrate a hypothetic language.
But the boundaries are sometimes blurry.

>       In addition 
> to that, "purpose" is not always easy to establish anyway. Some languages 
> clearly serve more than one purpose, and it's not always easy to tell which 
> one is the primary.

Yes.  Most "philosophical" languages, for instance, are such a border
case.  They had two purposed: to illustrate how a language based on
a universal taxonomy (or whatever) could work out (purpose B3), and
to overcome language barriers (purpose A).

>       Besides, I'm pretty sure a lot of auxlang authors only use it 
> as a means to justify their artlanging by providing it with some higher 
> goal. 

Sure.  While auxlangers usually claim to follow "objective" criteria
in their design, they are in fact usually biassed by either things
they are used to from their ethno-lingiuistic background (an often
cited example is the Slavic flavour of the Esperanto phonology with
its profusion of sibilants), personal preferences, or both.

> And Rosta once suggested making a distinction between PURPOSE and 
> MOTIVATION, but this leaves us with a practical problem: if there is any 
> difference at all here, only the author himself knows, and unless he writes 
> explicitly about it, we can only make an educated guess here. Not much of a 
> base for a reliable classification scheme either, methinks.

Indeed not.  The motivation of the author is often unknown.  All
we know is the purpose statement of the author.

> So I decided to go with the triangle and just look for a solution for
> languages that don't really fit there. The result was the hexagon. 
> Obviously, there is a critical difference between hexagon and triangle, 
> namely, in a triangle every group is the direct neighbour of the two other 
> groups. In a hexagon, there is not much point in putting a language anywhere 
> else than on one of the ribs, and therefore the effect of "60% auxlang, 30% 
> engelang, 10% artlang" is lost. 

Indeed; and that is, as far as I can tell, the main point raised
in this discussion.

> Personally, I don't see that as a disadvantage at all though: there is a 
> difference between CLASSIFYING conlangs and QUALIFYING them. The 
> hexagon is basically my proposal for a simple classification scheme, but I 
> am well aware that it is not suitable for qualifying them at all.

Yes.  That is a difference.  The Gnoli triangle is a qualifying
scheme which is *meant* to put languages somewhere in between
(even if in practice, most conlangs will end up near one of the
apices), while your hexagon is more of the top level of a
classification in which each conlang is placed in one and only one
category.  Your article, from which the Hexagon is taken, makes
this quite clear.

> I should add that there is grey area between all the six categories. 
> Experimental languages belong to the category of engelangs, but they clearly 
> have a lot in common with artlangs as well. Micronational languages can 
> easily  be treated as "constructed languages for special uses", while toy 
> languages ("language games") like Pig Latin have often been used as stealth 
> languages. 

Sure.  There are overlaps.  Not to speak of the cases where a
conlang is put to a use not foreseen by the creator, though such
cases are not really relevant to the classification of the conlang.

> Reconstructed languages have a purpose very similar to that of languages 
> created for educational purposes, while creating a standard for a number of 
> closely related dialects (like Rumantsch Grischun) is closely related to 
> creating a zonal auxiliary language.

Yes.  As I have observed before, the difference between an engineered
standard language and a zonal auxlang is analogous to the difference
between a dialect continuum and a language family - it is a gradual
one.

--
... brought to you by the Weeping Elf
http://www.joerg-rhiemeier.de/Conlang/index.html
"Bêsel asa Êm, a Êm atha cvanthal a cvanth atha Êmel." - SiM 1:1