> And even if it encodes the sounds of a language, if that language utterly died out and is unknown to scholars today, then even if we know what the sounds were, we probably wouldn’t be able to determine much of their meaning.
I feel this conclusion rather weak.
As an counter example, consider how the decipherment of the mayan written language was performed [1]. Another one is the hieroglyphs.
Re-constructing how mayan words sounded is still on-going [2].
Also, this type of work is not really possible to "crack with computers" as some have suggested (when all you have is a hammer, everything looks like a nail).