Re: A Conlang, created by the group?
From: | Charles <catty@...> |
Date: | Saturday, October 10, 1998, 15:23 |
Mathias M. Lassailly wrote:
> This is my ONLY message regarding cases :
>
> 4.1. Nouns as sole verb roots : what does this imply ?
That section alone was worth the price of admission
to this list. You *must* either post further references
or elucidate further the 10 nominal-ergative (?) cases.
> 5. Vocabulary :
>
> Yes. That's a good idea. Who volunteers ? [:-{
>
> Why not Charles ? He's stuck his finger in the hinge, now he must pay to get free :-)
I was only asking, not telling!
> I only have my essential vocabulary list of 1150 roots partly drawn from
> Japanese kanjis. But there are no verbs (one verb is 'represented' by only
> one agent, patient, result or instrument) and no opposites. Maybe it could
> help. Maybe not.
What I am using now is the English word frequency
data from http://info.ox.ac.uk/bnc/what/index.html ...
It is broken down into CLAWS grammatical categories.
Example of raw data:
100106029 !!WHOLE_CORPUS !!ANY 4124
6187267 the at0 4120
2941444 of prf 4108
2682863 and cjc 4120
2126369 a at0 4113
1812609 in prp 4109
1620850 to to0 4115
1089186 it pnp 4097
998389 is vbz 4097
923948 was vbd 4005
917579 to prp 4099
884599 i pnp 3746
Columns are: count, word, category, files.
This shows that the most frequent words are
grammatical particles and pronouns etc.
See also "Zipf's Law". Further down are
the nouns, adjectives, and verbs:
6222 substantial aj0 1699
6222 funds nn2 1389
6219 northern np0 1120
6209 reasonable aj0 1836
6208 onto prp 1520
6204 learn vvi 1871
6200 aircraft nn0 731
6197 games nn2 1222
6170 background nn1 1956
6164 officials nn2 1111
6158 strategy nn1 1316
6157 works vvz 2078
6154 prepared vvn 2110
What I'd like would be a sort of ontological
tree as in WordNet, grouping the near-infinite
nouns into a tree of sub-categories, a taxonomy.
The higher-level words would be the essentials
of a "reasonable" vocabulary. Maybe just 2000
or so well-chosen words is enough? Anyway,
it would be a good list to re-lex from.