Hallo!

On Fri, 11 Dec 2009 11:52:41 -0800, Jesse Bangs wrote:

> This is related to the other thread about how to give reality to an a
> priori conlang. Just this morning I was working on adding a few
> thousand items to the lexicon for a new conlang, so I thought I'd go
> over my procedure for random vocab generation without sacrificing the
> individual character of the language.
> 
> I have two problem with pure random vocab generation: 1) you wind up
> with something without a naturalistic feel, and 2) you lack the quirks
> and subtleties of a real language's vocabulary. If you just use an
> English wordlist for your lexemes, you wind up with a relex, at least
> as far as vocab goes. Both of these make me unhappy with the typical
> results.

I don't use a random vocabulary generator, but I also found that
pulling lexemes out of thin air in Old Albic (or some other language)
doesn't always work well.

> My solution to (1) is to generate *ancestral* forms rather than
> contemporary forms. I already know what the sound changes from the
> ancestral to the contemporary language were, so when I randomly
> generate my words I generate the proto-forms, then use phonix to
> mangle them into the contemporary forms. This results in a
> distribution of sounds and syllables structures that has a more
> organic feel, and has some historical justification, too. (As a second
> step, I don't

Yes.  This is also what I do with Old Albic.  When I need a word in
Old Albic, I ask myself, "What is the form in Proto-Albic?", and when
I have found a satisfactory answer to that question, I apply the sound
changes to it, and there I am.  This of course raises the question,
"Where do I get my Proto-Albic forms from?".  What helps here is that
Albic is not entirely a priori but to a large part a posteriori.  My
assumption is that it is distantly related to Indo-European, and that
Albic or some close relatives to it have left loanwords in Celtic and
Germanic, also in Shelta.  So I use words with unknown etymologies in
those languages and re-create Proto-Albic cognates.  I am currently
in the process of systematically collecting Proto-Albic lexemes that
way.

> The solution to (2) is to use a language other than your L1. This
> morning I used Urdu: I found an online Urdu-English dictionary,
> massaged the data until it was in the right format, then generated
> lexemes for every Urdu definition. This gets me a rather different
> distribution of lexemes than if I had used an English-centric
> wordlist, which provides the variety that I'm looking for.

I do not use a specific foreign language in order to avoid relexing
English or my L1 German, but I sometimes look into what Insular Celtic
languages (which I fancy to be influenced by an Albic substratum) or
various other languages do, and often, I ask myself, "What would the
Elves derive that word from?".  This of course only works if you have
an idea of the mindset of your conlang speakers, which I have for the
British Elves.

... brought to you by the Weeping Elf
http://www.joerg-rhiemeier.de/Conlang/index.html