Re: Language-generating software (was Re: Replies to my Introduction)

Roger Mills skrzypszy:

>All that may be true, but don't be too hard on yourself.
I wasn't being hard on myself; I thought it was an honest assessment of the
current state of the program.  I think it was great for 1997 when it came
out and okay in 1998.  But it hasn't been updated in 35 dog years (Windows
is a dog, hence the use of dog years for measuring the lifespan of Windows
software<g>).

I always wanted to update it, but never had the time, between work and
family.  And then when I did go to update it (early 2000), I was doing
32-bit development, and I discovered that the company I licensed the
spreadsheet from had changed the terms of the license to require royalties.
So to do 32-bit development would require rewriting LangMaker to use another
grid or spreadsheet control.  And that just requires more effort than I have
time for.


>It's helpful e.g. in the Aidan exercises, when I need a new word for, say,
>"bristle"-- I can scroll through the list until I see a form that seems
>"bristly" to my warped mind.  :-)
That's a neat way to do it, and not one made easy by the program.  I think I
had assumed people would tinker with their phonologies more and would
constantly regenerate their lexicons.


Sarah Marie Parker-Allen wrote:

>stogy when it comes to Windows applications (my number one problem with the
>entire OS is that it's too rigid in its "user friendliness" -- well, and
>it's also massively unstable...), doesn't mean I lack imagination.
Windows 2000 is very stable - the most stable of the Windows versions,
including XP.  I think it has crashed on me twice in the years I've been
using it.


>More management functions, the ability to generate tables of declensions
and
>conjugations, the ability to instantly apply conjugation and declension
>rules to root words (and apply other rules like that), perhaps the ability
>to translate basic constructions (i.e. a command that would allow me to
type
>in "the dog runs to the store" and see how, based on the various rules for
>sentence formation that I've defined, it'll look -- maybe even with all the
>variations that are allowed grammatically, the way that some annagram
>software will work).  In short, a skeleton translation software.
You'd probably have to settle for a function to look up a group of words at
once: a translation assistant.  What you describe is pretty difficult to
program.

I had wanted to write a simple translator *from* one of my languages, but
the words are polysemous, which makes it difficult to translate back to
English.  For instance, in this language, the color words have many
meanings, such as the word for "green", which also means "unripe, immature,
new".  So it would have to know enough contexts to translate "green leaf",
"unripe apple", "immature boy", "new tool".  Eke!


John Cowan scripsit:

>Hey, I just started to use Lotus Agenda, a DOS program whose last version
>came out in 1991.  The interface is clunky as hell (it looks like
>warmed-over Lotus 1-2-3) but what's inside is just brilliant.
Yeah, that was marketed so well it was known as "Hidden Agenda".  :-)
Similar to that devotion to Agenda, consultants I know still swear by Lotus
Manuscript.



Muke Tever wrote:

>I would like a do-everything program: lexicon, phonological changes
(synchronic
>and diachronic), make keeping track of cognates easy, create and manage
>translations, interlinears, keep track of phonemes and orthography,
language
>families, oh, and automatically update all those things if I alter a sound
or
>spelling change somewhere.
Kura does much of this:
http://www.xs4all.nl/~bsarempt/linguistics/


Roberto Suarez Soto wrote:

>        My own, "wordpuker". It's a Perl program based in morphological
>syllabic constraints. Or something alike, at least :-)
Perl's always been too Byzantine for me.  I coded an early forms processor
in Perl, and a Tolkien character quiz, but that's about it.  At work I'd
rather use Python on the server, but Perl's rich collection of modules makes
it tough to beat.


Mike Ellis jarhe:

>I don't like getting error messages that I don't understand. Subscript out
>of range? It gives this when I close the program, as if it were a reason it
>can't save my changes. But the changes are saved anyway. Odd.
I never could track this bug down.  "Subscript out of range" means I'm
trying to access an array past its upper or lower limit.  For instance, if I
have Filename(1) to FileName(9), FileName(10) would cause the error.  I
think its related to the list of recent file names, but never could
re-create it on my system. Sorry.


>5. What would you all like to see in such software?
>Also, with a language like Rhean, there are few phonological constraints.
>With what combinations ARE allowed outnumbering by far those which are NOT
>allowed, I'd like to be able to tell it that "everything is allowed
>except ..." rather than having to give all the permitted combinations.
Can you give me an example?  I looked at the Rhean RPG site - very
impressive!  But I couldn't find the language phonotactics.

Best regards,

Jeffrey