We have given these figures to show
what an enormous economy is achieved by having in human
language this "duality principle," as it has been called: first an
encoding into morphemes, and then a separate encoding of morphemes into oneormorephonemes each.
There is, however, a
very bad flaw in our figures: We have assumed that it is possible for phonemes
to occur in any mathematically possible sequence, such as (for English)
/ppppp/, /fstgk/, etc. But English of course does
not do this; like every language, it places very strict limitations on possible
sequences of phonemes. Nevertheless, even with the strictest sorts of limits,
the duality principle permits every language to form far more morpheme shapes
than it will ever use.