I have been reading (in bits and pieces, because the richness of ideas takes time to digest) Douglas Hofstadter's _Le Ton Beau de Marot: In Praise of the Music of Language_". The general topic of this book is how one translates verse from one language to another. This being Hofstadter, it spirals out into all sorts of related topics. It is, incidentally, highly recommended, as even partway through it, it has already given my brain lots to think about.
In recently-read chapters, he's been discussing the notion of whether it is actually possible to "translate" at all, verse or not. There are certainly arguments to be made that no translation can be perfect. One of the convincing ones is that no two languages share the exact same associational halos of meaning for any pair of words.
Naturally, Hofstadter spots the obvious reductio ad absurdum of that argument. Even within a single language, a given word does not call up identical associations with any two different *readers* of that language. So this argument would seem to imply that, not just translation, but *communication* is impossible!
The crux of the matter, of course, is "perfection". Communication and translation are akin to The Halting problem in computing. One can easily demonstrate that these things cannot possibly be perfect, in theory. In actual practice, however, one can *approach* perfection arbitrarily close. With care, you can usually get "close enough".
My insight today, is that this lack of perfection is why the concept of telepathy (as typically portrayed in SF) is so attractive. Telepathy allows *perfect* communication, unlike the clumsy tool of language. The Universal Translator is similarly appealing. Sadly, like FTL, neither of them actually hold up to logical scrutiny.