I think the point being made about the limitations of the
language analogy is valid.
In a written text, the accidental repetition of a word
or even full sentence produces a redundancy. The reader
may pause a moment, but the copy adds nothing to the
text.
If you add an extra pigment gene to a flower to enhance
the colour, you end up with white (colourless) flowers.
The reason for this (in violets if I remember correctly) has
been put down to something called RNA interference. This is
a kind of primitive 'immune' response in which the 'error'
is masked out. In the case above this also prevents the original pigment gene product to be masked out.
The copying of one gene produced a very marked change in
the phenotype.
I guess you could add it to your analogy as a destructive grammar
checker, where any grammatical errors are deleted from the text
(so both copies of the repetion are not expressed in print)