Re: making up words

At 3:51 pm -0700 19/3/02, Sean M. Burke wrote:
>A question to all language constructors:  once you've settled on the
>phonology and phonotaxis of your language-in-progress, how do you go about
>making up the phonological forms of new words (as opposed to their meaning)?
>
>Do you use a random number generator?
>
>Or just play it by ear?
>
It really depends partly on your own preferences and also on the language
itself.  If you are creating, e.g. a Romance conlang (practically every
does at some time or other :)  then obviously a random word generator is no
good: you adapt your words from the Romance stock.   Similar considerations
will apply to any a_posteriori conlang.

In designing Classical Yiklamu, Mark Line was quite happy to generate all
his vocabulary by programming a (pseudo)random word generator with  the CY
phonotactic rules and mapping the results to WordNet - no playing by ear.
Forming vocab items was of little interest to him; he was more interested
in other aspects of CY.

I tried the same approach, on a far more modest scale, to see if it might
be a good idea for my BrSc* ; it worked, but I did not like the results -
some of the words didn't seem to fit the meanings  :)

I've decided to play it by ear and use a (vaguely) a_posteriori approach;
but as root words must conform to the CVC pattern they will be
'hyperSchleyerized'.

Tolkien clearly didn't use a random word generator & played it by ear.  One
reason, of course, is that the computer technology that makes it so easy
for us to use a random word generator was not available for him; but I'm
pretty sure that even if it had been, he wouldn't have used it because
word-creation was surely for him part of the overall creation process.
That is exactly how some conlangers feel: "I'm not having a dumb machine
make up my words."

Basically, in the end it's your own decision and depends to some extent on
why you are creating a particular conlang.
------------------------------------------------------------------

At 10:09 am -0700 20/3/02, Dirk Elzinga wrote:
[snip]
>
>I have generated all of the possible roots for Tepa/Miapimoquitch
>according to the phonotactic patterns I think I want (thanks, John!).
>I then check off forms from the list as they are made. This does two
>things: 1) it ensures that I'm using phonotactically legal forms for
>the lexicon (though the phonotactic patterns themselves are subject
>to change), and 2) it ensures that I don't create homophones.
This is a good idea, which computer technology makes eminently possible.
Of course, the words aren't generated randomly - there's no point; indeed,
'twould make it harder to search for a particular form.  Just generate the
complete list in order, and then 'fill in the blanks' as and when you need.


It also makes it easier to look for the sort of sounds you think should be
associated with a particular idea, if that's what you want.

My immediate task now is to program coding that will generate all the
possible root words in BrSc   :)

Ray.

*BrSc is a name coined by some fellow conlangers for a 'briefscript'
project that's been pre-occupying me for 40+ years and still hasn't got a
name!




=========================================
A mind which thinks at its own expense
will always interfere with language.
                     [J.G. Hamann 1760]
=========================================

Replies