Generally speaking, translation of dead languages is done via means of bilingual texts. The Rosetta Stone is the most famous of these; for cuneiform, the key wa

Generally speaking, translation of dead languages is done via means of bilingual texts. The Rosetta Stone is the most famous of these; for cuneiform, the key was apparently a trilingual Old Persian/Elamite/Akkadian inscription (the ‘Behistun inscription’). In this case, Old Persian is an Indo–European language, and Akkadian is Semitic, so those languages have plenty of modern-day relatives with very similar structures, which helps decipherment. Once cuneiform was deciphered, Sumerian/Akkadian bilingual texts were sufficient to decipher Sumerian: again, the Semitic nature of Akkadian helped a lot here. It also helps that most languages use more or less similar techniques to express certain concepts and relations. The elaborate case system and verbal morphology, though ‘exotic’ for Europeans, aren’t necessarily all that different to those found in languages elsewhere. As for ambiguity… I’m not entirely sure how that particular problem is overcome, but the author mentioned duplicate manuscripts which together can remove some of the ambiguity. Beyond that, we just need to infer the missing pieces from the fact that languages are usually self-consistent to a large extent. (Disclaimer: I know very little about this area, though I do enjoy reading about linguistics!)