Just saw the movie. Is it actually true that during the Roosevelt administration, some marines were instructed to murder Navajo marines if they were in danger of being captured by the Japanese?
Apparently so, at least according to a speech entered into theCongressional record in 2000.
They weren’t talking code, they were just speaking Navajo, which the Japanese couldn’t translate. You probably know this, but saying they were code-talkers may be misleading to at least a few that read this thread.
Still, the language itself served as a code, because the Japanese didn’t have access to anyone who could understand it.
Nope, they were in fact talking in “code”. See from this description:
The code talkers story has always fascinated me.
But holy moly, that was a TERRIBLE movie.
Most sources say that the story that there was a guard that was supposed to kill the codetalker if he was in danger of being captured was nonsense. Among other things, many of the surviving Marines say that there is no way that a Marine would obey an order to kill another Marine. It’s possible that this story is a confused version of the orders that the Marines in a codetalker’s unit were given to not let the codetalker wander away from the unit. They were afraid that he might be found by another Marine unit who didn’t know about the codetalkers.
Remember, the Marines and all other American military units were segregated at that point. The only black Marines were in all-black units. The only Asian Marines were in all-Asian units (and were in Europe, I believe). It’s entirely possible that another Marine unit would think that someone of American Indian ancestry was Asian. In particular, they might think that the codetalker was a Japanese spy who spoke pretty good English who had killed a Marine and stolen his uniform.
That doesn’t sound particularly unbreakable to me. Sure, the Japanese wouldn’t have known Navajo, but you don’t need to know Navajo to deduce that the code word “be-la-sana” means the letter “a”. Pure letter-substitution ciphers are the stuff of newspaper puzzles done by amateurs with no formal training. Admittedly, the bit about having multiple code words for each letter would make things more difficult, but it seems to me that there would have to be a standard list of only a few code words per letter, known to the code talkers. Otherwise, you’d have the risk of one translating “tse-nill” as “hatchet”, instead of “axe”, and thinking it was meant to be an “h”. Did the Japanese just not realize that it was that simple, or was there more to it?
There’s a lot more to it. As I understand it, the grammar is unique. Changing the context of words can completely change their meaning. On the other hand, several words can easily sound like homonyms to non-native speakers. Unless you understand the context of a Navajo converstion, you will have no clue as to its meaning. The only way to learn the context is to go and live with the Navajo for a couple of years - an option not open to the Japanese during WWII.
I know this is bad description, but there is a much better one in The Code Book by Simon Singh. I don’t have it with me right now.
From what I understand, they didn’t even get that far. Navajo sounded like such incomprehensible gibberish to them at first many thought that it was just a smokescreen and wasn’t even a code at all. They couldn’t do an adequate letter substitution because they couldn’t even transcribe it. When they did capture a non-code talking Navajo late in the game and attempted to get him to translate, he couldn’t make heads or tails of what the code talkers were saying and thought they were drunk.
Aha, my geek bookshelf comes in handy…
This spelling was still susceptible to brute-force interpretation - the most-repeated sound likely to represent ‘E’. So other words were used as replacements, such as be-la-sans (axe) and *tse-nihl (apple) for A.
Except that grammar wouldn’t apply to a string of words like ‘pig, ant, cat, ice, fox, cat’, and there’s likewise no real context there. If they were speaking full sentences in Navajo, then yes, I can see how that would be difficult to crack. But if you’re just spelling out code words for English letters, then I don’t see the point.
That’s why they switched to a multitude of words to represent each English letter.
…plus, it was only the awkward words that were spelt this way. “Setting sail” or “Many dead” could still be conveyed in pure Navajo, which nobody has claimed is ‘crackable’.
And the multiple substitutions weren’t done in a systematic way. They were more or less ad hoc, which produces something reasonably close to a one-time pad without any actual pads. Since the code was produced in real time, there was just no way to break into it in a time frame that would have any meaning.
Chronos writes:
> That doesn’t sound particularly unbreakable to me.
It wasn’t unbreakable. Anything written about the codetalkers that describes the code as unbreakable has only a superficial knowledge of it. It was merely good enough for the circumstances in which it was used.
It was a preposterously difficult code to break under the circumstances. Most of these messages were apparently short and time sensitive.
Even had the US announced that the Navajo were using 450 code words and spelling the rest with the first letter of the English equivalent, it won’t have done much good. The message would have to be translated quickly on the scene, meaning the person would have to be fluent in Navajo and in English. I don’t know how many Japanese military men were fluent in English but I’m sure my guess of the other is pretty close – zero.
About the best that could have been done was teach someone in the units some of the code words. If they heard a message with “airplane”, that generally would have been of little use–are they calling in planes, describing where our planes or theirs are, saying we have no planes, etc etc etc.
As much as I deal in infosec and crypto, I’m woefully lacking in knowledge of the Navajo Code Talkers, so I have to ask - for any given Navajo talker, was there a corresponding Navajo listener, or was the “code talk” received at the other end of a radio wave by non-Navajo ears?
Reason I ask is that almost every language has sounds that are indecipherable to foreign ears - the Chinese confusion of English “l” and “r” sounds being a prime example. In addition to speaking a rare language, were the Navajo using sounds that Japanese ears simply couldn’t comprehend?
So far as I can tell by looking at some of the above romanized translations of Navajo words, there are several sounds that have no close approximations in Japanese.
From this partial list,
I see the phonemes “tse,” “gli,” and “tsah,” which I could not begin to write in Hiragana. My assumption is that a native speaker of Japanese who knew no Navajo would not catch these sounds properly the first time.
Yes there was, for the reasons you said.