Re: Art is when someone says 'Now' -- or is it?

Jörg Rhiemeier wrote:

> Yes.  A realistic language grows with usage.  However, one can
> very much strive for stability in the *rules* of the language
> (i.e., in its phonology and morphosyntax) and a restriction of
> changes to its lexicon to the *addition* of new words, such that
> old texts are never invalidated by later changes to the language.
> I already have come to regret that there are Old Albic texts in
> public existence (a Babel text buried somewhere in this list's
> archive, contributions to several translation relays, and a few
> smaller bits) that are no longer valid in the current incarnation
> of the language.  Well, that's all work in progress, and the
> future new web site for Old Albic and its descendents will be
> declared "canonical" by me.
I've done relay translations in a number of different versions of
Tirelat and Minza. Lately I've been trying to bring some stability to
these languages, deciding whether to keep older features or newer ones,
and revising the existing texts to match. In the case of Tirelat, I've
taken the original alphabet I developed for it (Vlika) and given it to a
different language (Virelli), so I have to come up with a new alphabet
(possibly something like Tharkania or Ljörr).

> A conlang that has no "canon" is an unstable conlang.  If you
> change the inner workings of the language all the time, you
> arrive at a corpus of ungrammatical texts.  If you want to build
> up a *consistent* corpus, you must finalize at least some decisions
> in the design of the language.  Words that are already in your
> dictionary stay in; grammatical rules remain valid; and all that.
Before Tirelat, I would decide on a word or a feature of grammar, and it
would pretty much remain that way. One big difference is that I did
everything with pencil and paper in the old days. I got carried away
with revisions on different versions of Tirelat and ended up with a
mess. You'd think I would've learned not to do that with Minza, but it's
been through its share of revisions.

> A conlang is indeed something like a series: at the beginning of
> the series, it is not certain which course the events take, but
> once an episode has been released, it is part of the canon and
> later episodes ought to be consistent with it.  It is - or ought
> to be - the same with conlangs.  What has been released by the
> author and not designated "work in progress", becomes canon, and
> later canon ought to be consistent with it.  Sure, there are ways
> to "iron out" revisions.  Natlangs usually have synonyms or at
> least near-synonyms; they often have several alternative ways of
> expressing the same category.  In a naturalistic conlang, you
> can make use such synonymies - or dialect divisions - to
> retroactively canonize revisions: by declaring that *both* the
> old and the new version are valid.  However, there are limits
> to this.
It can take a while to come up with a consistent "character" for a new
language. The current version of Tirelat is much like the version in my
translation of "Tegla Jlána" from 2001.

http://www.io.com/~hmiller/lang/Tirelat/swallow-song.html

Even the old page http://www.io.com/~hmiller/lang/Tirehlat-old.html is
still recognizable as Tirelat. But if I'd stuck with that version, some
of the features that have become most characteristic of the language
(like the verbal morphology with fused evidential+tense suffixes) might
not have come about.

By the sixth Conlang relay, I was starting to add genders to nouns.

  Ispa ttehaban mi naji mikvidu myn metta fał.
  ("mi" and "naji" agree with "mikvidu", "metta" agrees with "fał")

But later I decided that I preferred the older version of the language
without noun genders, and the current version of the language has gone
back to genderless nouns.

  Ispa tətexaban my ñaĭ mikvidu myn metti fał.
  "All wells can indeed be cut into tiny pieces."

As much as I remember Tirelat having lots of revisions, the relay texts
actually aren't as far from the current version of the language as I'd
thought they might be. The same is true for the Minza relay texts. Most
of the crazy revisions must have gone on between relays.