sound symbolism

[Edmund.Grimley-Evans@cl.cam.ac.uk]

I was thinking last night that it might be amusing to try an experiment
concerning the "natural" allocation of meanings to sounds. The idea
would be to agree on a phonology and a list of senses to be covered and
to try to allocate the senses to words in as "natural" and mnemonic
manner as possible. Ideally, the result would be a word list that bears
no particular resemblance to an existing language but which is much
easier to learn than a random list and perhaps as easy as a list based
on Latin roots for a typical English speaker, say.

I would like to do this with a very simple phonology - open syllables,
10 consonants, 5 vowels, total of 50 syllables, say. There would then
be 50 monosyllabic words and 2500 bisyllabic words. One would definitely
not want to allocate all of them at the start, so a list of 1000
concepts to be allocated to the bisyllabic words would be a realistic
task. I would take the concepts from frequency lists; they would not
include structural elements as it would be too difficult to decide what
the structural elements should be, and some of them would be
monosyllabic.

Since cooperation in designing a conlang never works, it would be
better to organise this as a competition. Unlike conlangs themselves,
the results of this task would be directly comparable.

Does anyone else think this is a stupid idea? Has it been done before?
(Can anyone give references to a fairly complete word list created a
priori according to a principle of sound symbolism rather than
classification? The idea goes back to Comenius, so I'd be surprised if
no one has ever tried to apply it in practice in all those intervening
centuries, but I don't remember ever having seen such a list. Maybe, if
we were to attempt it, we would see why no one has ever completed the
task! Who knows?)

Edmundo