Matt & John seemed pretty skeptical. But as long-time subscribers to
this list may have noticed, I qua artlanger-cum-linguist have a bee in
my bonnet on this subject.

> Hm.  I haven't the maths but I suspect that isn't any better
> than chance.  I'll ask someone.  But probably there isn't
> enough data, that's what they generally say...
> "Murud=brown" seems to have struck a chord, but apart from that
> just like the Tokana results, one person's yellow is another
> person's red...

Oliver Cromm already pointed out some perhaps significant patterns in
the data, but the test was not really "fair". There were too many
permutations to detect slight phonesthetic influences. Also, it might
be that the Tokana & Elet Anta colour words just aren't optimal for
receiving intersubjective agreement.

> Sorry, 100% wrong!  :-)
> Just goes to show you that sound-symbolism is in the eye
> (or ear) of the beholder...

It is easy to disprove this. Here's my ill-recollected version of an
experiment proposed by some august American linguist (I forget who).
     [Brad Coon? Are you listening? Does that ring a bell?]
One of the following words means "a small object with a jagged spiny
exterior", and the other means "a large amorphous cloudlike shape".


     shmoozh    (i.e. [Smu:Z])

-- which is which? The answer is so obvious that one needn't actually
conduct the experiment to know that answers would give vastly better than
chance results (though in classes when I have conducted it I get 100%

Besides this experiment, look through the OED & see how many words
are said to have an "imitative" etymology. Or consider some of the
putative global (or very widespread) etymologies some longrangers or
megalocomparativists turn up: for many of these onomatopoeia is a
better explanation than either cognacy or chance.