Re: What is needed in an conlang classificatory system?

taliesin the storyteller <taliesin@...> writes:

> In the typologic survey I ask for languages to be classified according
> to Rick Harrison's system simply because I don't know of any better.
> The thing I like the most about it is its conciseness. A screenfull of
> data to classify a language is too much, as you can't get an overview at
> a glance.
This is a problem.  A screenfull of data might be too much, but anything
less has a hard time conjuring an image of any kind.  Languages are just
too complex and diverse to classify thoroughly with a few letters and
digits.

> I asked for others, and ideas for improvements, got some answers to that
> but decided to move the discussion to its own thread, so:
>
> - what is missing from Harrison's system to make it fit non-auxlangs?
> - if Harrison's isn't useable, what should be used instead?
Harrison's system works, of course, but with artlangs, it doesn't yield
particularly useful results, as can be seen by the great number of langs
that went into category 2.1.2.

I think a conlang classification scheme should start with the purpose
for which the language is created.  Most conlangs are euther auxlangs
(languages invented for the purpose of facilitating international
communication) or artlangs (languages invented for "artistic"
purposes).  A possible scheme could look like this:

A. Languages for practical communication
  A1. Auxiliary languages
    A1a. International auxlangs
    A1b. Regional auxlangs
    A1c. CETI languages (for communication with extraterrestrials)
  A2. Secret languages
  A3. Other (e.g. controlled languages in totalitarian states:
"Newspeak")
B. Languages for scientific, educational or similar purposes
  B1. Model languages for psycholinguistic and similar experiments
    B1a. Test cases for the Sapir-Whorf hypothesis
    B1b. Language acquisition research
    etc.
  B2. Demonstration languages (languages designed to demonstrate
linguistic
      concepts in linguistic education)
  B3. Philosophical languages
  B4. Magical languages
  B5. Reconstructed proto-languages
C. Languages for artistic purposes
  C1. Languages spoken in fictional worlds
  C2. Personal languages

There are of course borderline cases.  Is Enochian A2, B3, B4, C1 or
C2?  (Points could be made for any of these classifications.)  Or what
was the purpose of the Voynich Manuscript?  It could be A2, but just as
well C1.  Border cases between C1 and C2 are also abundant.  Quenya
started as C2, but soon became a C1 language as JRRT began building
Middle-Earth around it.

Fictional languages (category C1 in the scheme above) could be
classified as if they were natlangs, because that is what they are meant
to represent (unless they are fictional _conlangs_ in which case the
scheme could be applied again:
Orwell's Newspeak would be a C1:A3 language).  However, the diversity of
fictional languages is greater than the one found among natlangs, for
the following reasons:

1. Some fictional languages display combinations of features not
attested in natlangs.  For example, I have heard multiple times that no
active language has yet been found that marks nouns for case that way -
those that are known are all head-marking.  But several case-marking
active languages exist in the realm of fiction (e.g. Nur-ellen and I
think also Chevraqis).

2. Not all fictional languages are meant to be spoken by humans; some
are spoken by miscellaneous fantasy beings or by aliens.  This again
increases the number of feature patterns found in fictional languages as
they might even show designs that simply wouldn't work in human
languages.  Non-human languages might be unspeakable by humans, being
based on alien phonetics (there once was a discussion on this list about
glottal nasals - impossible for humans - in the troll language trQal;
Rikchik and Ilish don't even use the auditory channel), or on a really
wild grammar humans cannot cope with, at least not in real time (such as
the stack-based grammar in Fith).  A classification scheme for artlangs
should thus always address the issue what kind of beings speak the
language.  Here is a suggestion:

C1a. Fictional human natlangs
  C1aa. Alternative history
  C1ab. Future
  C1ac. Fictional past
  C1ad. Present*
  C1ae. Fantasy (not related to the real world)
C1b. Fictional non-human natlangs that could be spoken by humans
  C1ba. Spoken by fantasy races
  C1bb. Spoken by space aliens
  C1bc. Spoken by other kinds of beings
C1c. Fictional non-human natlangs that could not be spoken by humans
  C1ca. Spoken by fantasy races
  C1cb. Spoken by space aliens
  C1cc. Spoken by other kinds of beings
C1d. Fictional conlangs.  Apply the conlang classification scheme once
again, from the viewpoint of the fictional world.

*This is for languages which are assumed by the author to "exist in the
real world, but are yet undiscovered" or something like that.

Examples:

Brithenig C1aa
Adunaic C1ae
Khuzdul C1ba
Watakassi C1bb
Rikchik C1cb
Newspeak C1d:A3
Elet Anta either C1ad or C1d:A2 (uncertain from the background)

Again, several border cases.  Some worlds feature languages spoken by
both humans and non-humans, for instance, and it isn't even always clear
whether a certain language in that world is meant to be a natlang or a
conlang, as the Elet Anta example shows.

This scheme, I admit, is rather haphazard and could be more elegant.
But it isn't intended as an out-of-the-box solution, anyway, just as a
sketch what such a thing could perhaps look like.

There are certainly many ways to create subclasses of C1...

I think Luis Henrique has proposed some good ideas.

--
...brought to you by the Weeping Elf
"Bus, bi, bo, bum, bo; bi, borum, bis, bos, bis!"