The fact that mathematics is not a perfect language was pointed first by Hume and then Russell about induction and this is in line with the answer of @Michael. Even for the construction of numbers, axiomatic does not prevent you from the final instantiation of your general abstract model in the reality. At some point, this instantiation is made possible by your experience of the reality, this brings you somehow to induction where the finiteness of your experience avoid any “perfection” (in statistics, this is called “no free lunch”).
I think this is all already contained in Heraclite’s words (and related fragments):
Mais bien que le Logos soit commun
La plupart vivent comme avec une pensée en propre.
(sorry I only have the french translation From Jean Paul Dumont, in english that could be
But even if the Logos is common to everyone
Most are like having their own thought. (Note that this Fragment is reported by Sextus Empiricus in “Against Mathematics” … :) )
At the end, this depends on the meaning you give to “perfect” in perfect language. I found the word perfect a bit of a problem here. I guess you make the distinction between a perfect language and a closed language. You might be able to create closed languages Boole algebra is closed. Is it what you mean by “perfect”.
Coming back to Wittgenstein, have you read “on certainty” ? Also “Différences et repetitions” from Deleuze, would fit here...