Re: Conlanging as a personal thing

On Wednesday 12 March 2003 10:52 am, Roger Mills wrote:
> Mike Ellis wrote:
> > > Or to go even beyond that: that both writers may create new words,
BUT
> that
> > > any person who does so must also make three cognate or related
words
> (i.e.
> > > from the same root) or other words with concepts complementary /
related
> /
> > > opposite to the new word. That way, every new word would seem to
have
> > > related words or concepts in the language PLUS the lexicon would
grow
> that
> > > much faster.
>
> That accounts for the rapid growth of the Kash lexicon (and one of the
> reasons I have trouble remembering things)... For ex. every
adjective/verb
> has a potential 6 derivatives-- base, inchoative, causative,
potential,
> accidental (verb forms); patient/abstract noun, agent noun; to say
nothing
> of possible full or partial reduplications, idioms, colloq. variants.
Thus,
> according to the statistics, there are 1649 head entries in the
dictionary,
> but considerably more words.
>
> (Is there some way I could tell the computer to count all words in
> _boldface_? That would give a more accurate count.  Oho-- could I tell
it to
> count "< bold > " in the html version???)
>
>
I'm building my lexicon in XML, with a perl program that sucks in the
XML and
1) writes out a sorted xml file
2) writes out an xhtml version sorted by Term (the Jouevyaix or Imperial
morpheme)
3) writes out an xhtml version sorted by Class
4) writes out an xhtml version sorted by Keyword (providing an
English-to-Jouevyaix listing)
5) prints the numbers of terms on the command line (266 for Jouevyaix as
of last night: about 240 more than there were a couple of weeks ago,
though some of the older terms are several years old.)

The lexicon on my website only has the sections sorted by Term and
Class.

The perl tool is currently hardcoded for my development environment, but
if anyone is interested in the tool set, I can make a generalized
version available.

The XML dtd is currently very simple:
A <lexicon> contains multiple instances of <entry> tags.
An <entry> contains one <term>, one <gloss>, one <xsampa>, one or more
<class> tags, zero or more <keyword> tags and zero or more <glyphs>
tags.

Most of my XSAMPA tags are currently empty... my next version of the
perl tool will populate the empty ones by translating my Latin1
representation into XSAMPA. The resulting entries will need tweaking
(eventually) but should be better than nothing.

The <glyphs> tag is forward-looking to a time when I (1) develop the
con-orthographies to a point where I'm happy with the way they look,
(2) find a good way to display them on line and (3) research how to
deal with the directional problem in a mixed rendering environment. [*]

I'm planning to add <section> tags (paragraphs) within the glosses, and
eventually <usage> tags for usage examples. (All the existing corpus
has appeared in postings on this list in the past couple of weeks, so
there isn't much usage available yet. Actually... making a rule that
every word in the lexicon should be used in a sentence might help in
learning the language.)

Suggestions for additional tags that might be useful in a lexicon will
be gratefully accepted.

*
Imperial is written right to left, top to bottom.
Jouevyaix is formally written with beads, strings and crochet hooks, but
has a diagrammatic 2-dimensional representation that can use almost any
layout imaginable. If I ever do an example of spiral or labyrinthine
page layout, it will have to be a big jpeg, but even something as
conservative as 'vertical boustrophedon starting in the lower right
corner' presents challenges to standard web rendering engines.
--
Elyse Grasso