Re: context, a future archaeologist would have more examples of the language, just like we have lots of examples of Latin, Greek, and Sumerian which allow us to

Re: context, a future archaeologist would have more examples of the language, just like we have lots of examples of Latin, Greek, and Sumerian which allow us to learn the language with enough context to decipher the more damaged inscriptions. The trickiest part might be the name Joe Biden which isn't a part of English, per se. Correspondingly, there are ancient kings whose names we only have fragmented versions of for the same reason. Ithkuil would obviously also need a large corpus for future archaeologists to decipher, otherwise the runes are meaningless. Re: repetition - erasure coding (parity bits, basically) is far more efficient than duplication for data durability. How much this applies to natural language (where extraneous characters act sort of like parity bits) I am unsure, but there is nothing stopping you from just mapping your natural language to a binary code and using actual parity bits. Obviously, a future archaeologist will need to know what kind of erasure coding scheme you are using, but that is actually much simpler to convey than the corpus of the language itself. It's pretty simple math.