Re: Bootstrapping a cooperative conlang

--- Herman Miller <hmiller@...> wrote:

<snip>

>
> There's not enough context in one picture of a mouse to infer a meaning
> of "pest". A picture of a mouse infestation in a grain warehouse might
> be a better clue, but for a word like "pest", you'll probably need a
> whole set of pictures including cockroaches, zebra mussels, etc. to get
> the idea.
>
> "Animal" is another word that a single illustration isn't enough for.
<snip>

I'm beginning to think that pictures are not the best route for categories like
"animal" (as opposed to specific instances of the category like "zebra"). As it
happens, "animal" is in the current incarnation of my 122-word core vocabulary,
so does not need to be defined. "Pest" could be defined recursively by first
defining "crops", "goods", "damage", "annoyance", "disease", etc. Armed with
whatever precursor definitions make life easier, defining "pest" could probably
be reduced to something like "small animal that annoys people, damages crops or
stored goods, or spreads disease." The NSM method of "reductive paraphrase"
becomes the process of factoring a word into successively simpler constituent
concepts, and then defining those simpler concepts before returning to the task
of defining the original target word. In the process of defining "mouse" we
might have  as a by-product, definitions for such new words as "mammal",
"rodent", "pest", etc.

--gary