Re: Bootstrapping a cooperative conlang

Gary Shannon wrote:

> Long-winded definitions can probably can be avoided by building the vocabulary
> in small incremental steps. For example, "hunt = find animal, then kill for
> food", "fly = travel in air", "bird = animal that can fly", "swim = crawl on
> water", "duck = bird that can swim." and finally "duck hunt = hunt for ducks".
> Even if mouse had to be defined without recourse to a picture, I'm sure it
> could be a very short definition consisting of other defined words that
> introduced intermediate concepts such as "mammal", "vermin", ...
I think those kind of phrases are more "clues" than definitions. Bats
and insects are also flying animals, and with this broad a definition of
"fly", flying squirrels, flying frogs, and flying fish would also be
included as birds. (But ostriches, penguins, and other flightless birds
would be excluded.) Birds that can swim would certainly include loons
and coots (which might be mistaken for ducks), and other kinds of birds
such as puffins.

For a word like "mouse", you could start with something like "small
furry animal with gnawing teeth, round dark eyes, and a tail", which is
somewhat imprecise and could include voles or gerbils. This assumes
you've already defined words like "furry", "gnawing", "teeth" and so on.
Words like "vermin" are not really useful in distinguishing mice from
non-mice (and may vary from one culture or geographical region to another).

> As for core vocabulary, I started with the Swadesh list, added the NSM semantic
> primes, trimmed out the duplicates, and then started trimming words that could
> be easily defined by a picture, and words that could be easily defined by two
> or three other words in the list. I then added some words that made it easier
> to define other words already in the list so that they, in turn, could be
> dropped. Even the semantic prime "now" can be dispensed with by adding the very
> handy word "time", so that "now" = "this time", etc., and "time" itself is now
> available for use in defining other temporal concepts.
>
> I've gradually pruned the list to the point where I can't find any more words
> that are easily dispensed with. Currently that leaves 122 words, each of which
> has one very specific definition (no polysemous words allowed). I imagine the
> list might grow or shrink a bit as I play around with defining more "complex"
> words in terms of these 122 core words.  Eventually I should end up with a
> practical list of core words, and a nice list of secondary words defined only
> in terms of the core words.
>
> --gary
That definitely sounds like a useful goal.