Re: Bootstrapping a cooperative conlang
|From:||Herman Miller <hmiller@...>|
|Date:||Tuesday, November 20, 2007, 1:48|
> In a message dated 11/19/2007 12:01:50 AM Central Standard Time,
> hmiller@IO.COM writes:
>> It's just not clear why these words have been singled out. At first
>> glance, some of them can be defined in terms of others (far = not near),
>> and yet other fundamental ideas are left out (left, right, turn,
>> straight, away). No doubt this is all explained in the expensive book
>> that was mentioned, but for conlanging purposes, I don't see the benefit
>> of such a minimal set of words. I don't know what the optimal number of
>> basic words for using in definitions might be, but I have the feeling
>> that most people who might otherwise be interested in a collaborative
>> project would have little patience for this kind of a system.
> One can think of the semantic primes as atoms to build molecules. Once you
> have built some very common small molecules, you can use those to build
> slightly bigger molecules, then still larger ones, until you get to the largest size
> you need. Done in small increments, the load for each word should not be too
> large. Explications of English words in NSM are so big because the words
> already have their complex meanings. Building the words from scratch would allow
> much smaller explications. Creating an equivalent of the word "mouse", e.g.,
> could be done in several stages, rather than all at once.
It's clear that you can use them as building blocks, but it's not at all
clear *how* to use them. Take the example of "left" and "right" -- how
can they be defined? "Turn" would be a kind of "move", but what exactly?
"Straight" describing motion could be defined as moving without turning,
but how to describe the shape of something? "Away" has something to do
with "far", but has to do with the direction of motion rather than location.
> The semantic primes are analogous to machine language. Using that machine
> language you can quickly build higher and higher-level languages that can
> accomplish a lot with relatively few statements.
It looks more like a Turing machine, if you ask me. Something that's of
theoretical interest, but not very practical. Machine languages already
have complex ideas like treating a group of bits as a representation of
a number and performing mathematical operations on them.