A story about cracking natural languages.
Let's focus on decyphering Linear B because:
- Champolion's story is retold over and over all the time.
- Decypherment of Mayan glyphs would focus too much on the graphics, it's not that relevant to crypto.
- Cuneiform? Don't know. Maybe.
- The story is relatively simple.
Some points to cover:
-
Number of signs hints at whether it's phoneme-based, syllabary or ideographic alphabet. But! Mayan glyphs. What looked like a limited ideographic script was in fact syllabary, but with art mixed in.
-
Now you read about dcypherment of Linear B. Guess what? There's also Linear A which is unsolved to this day. Also: Voynich.
-
Code talkers. That actually points at an interesting problem: When hearing a foreign language you can't split it into discrete phonemes. (Explain distinctive traits.) Therefore, it's hard to even start with the analysis. That in turn can, maybe, point to Shannon.
Distinctive traits:
- Japanese and Koreans don't distinguish L and R. Seoul may as well be Seour.
- Spanish does not distinguish B and V.
- Classic Arabic has only three vowels.
- Germans don't distinguish V and W.
- English, on the other hand doesn't distiguish open and closed E, palatalized consonants, aspirated consonants, emphatic consonants etc. Maybe N and N with tilde in spanish? Or maybe H and J in Spanish? Ask an English speaker whether they hear the difference.
Technically, the best way to explain the thing is showing the partitioning of wovel triangle in different languages, say Arabic, English, French.