No. A fluent Navajo speaker who had never been trained on the code could not read code talker traffic, and the codebook settles it in a single line.
In the Navajo Code Talkers' Dictionary revised 15 June 1945, the English word "that" is sent as Tazi-cha. The dictionary's own literal-translation column reads: turkey hat.
Turkey is the code word for the letter T. Hat is just a hat. Put them together and you have T plus hat, which is "that" in English. A Navajo speaker with no code training hears two nouns, one about poultry and one about headgear, and gets nothing. To recover the word you need Navajo, English, the letter table, and the knowledge that this particular entry is a pun rather than a metaphor.
Key Takeaways
- The 15 June 1945 revision holds 677 code words in three parts: 63 alphabet terms, 63 terms for named categories, and 551 general vocabulary entries.
- 29 vocabulary entries are English rebuses assembled from Navajo words. "When" is weasel hen. "Spray" is snake pray. "Farm" is fox arm.
- The alphabet grew from 26 terms to 44 to 63, because a Marine cryptographer warned that spelling words out was producing too many repeated letters.
- 282 of the 551 vocabulary entries are ordinary Navajo with no substitution at all, so the code and the language sit side by side in the same book.
- Philip Johnston proposed the idea. The 29 original Navajo recruits devised the code, alongside Marine communications personnel during their signals training.
What is actually inside the Navajo Code Talkers' Dictionary?
Three lists, doing three different jobs.
The declassified copy in the Naval History and Heritage Command's online reading room is headed "REVISED 15 JUNE 1945" and carries the line "DECLASSIFIED UNDER DEPARTMENT OF DEFENSE DIRECTIVE 5200.9". Counting its rows gives this shape:
| Part of the book |
Entries |
What it holds |
| Alphabet |
63 |
Navajo words standing in for the 26 English letters |
| Named categories |
63 |
Organisations, officers, countries, aircraft, ships, months |
| Vocabulary |
551 |
General terms, running alphabetically from "abandon" to "zone" |
| Total |
677 |
|
The dictionary is printed as a three-column table of 683 rows, six of which carry no entry: three are column headings and three are blank spacers between the parts. That leaves the 677 entry rows counted above, and they were counted twice, by two different methods that agreed, so anyone who opens the document can reproduce the figure.
The middle block is the one people rarely see, and it is the most revealing. Take the eight unit names. Corps is Din-neh-ih, "clan". Division is Ashih-hi, "salt". Regiment is Tabaha, "edge water". Battalion is Tacheene, "red soil". Company is Nakia, "Mexican", which is the name of a Navajo clan and not a nationality. Platoon is Has-clish-nih, "mud". Section is Yo-ih, "beads". Squad is Debeh-li-zini, "black sheep".
Most of those are not free-associated nature words but Navajo clan names, and the Marine Corps table of organisation has been laid over the Navajo clan system, with "corps" glossed as "clan" to announce the scheme. Five of the eight can be traced to a Navajo dictionary published in 1910, three decades before the war: the salt people, the tqabaha ("on the shores"), the debe' lizhini or black sheep people, the nakhai dine'e or Mexican clan, and the tqachini, which that earlier book glosses as "red water people" rather than red soil.
The officer ranks work differently again. A colonel is Atsah-besh-le-gai, "silver eagle". A lieutenant colonel is "silver oak leaf", a major "gold oak leaf", a captain "two silver bars", a lieutenant "one silver bar", a major general "two star", a brigadier general "one star". Nobody translated the rank. They described the insignia you could see on the collar.
And the months are not translated at all. January is Atsah-be-yaz, "small eagle". February is "squeeky voice", the book's spelling. September is "half". December is Yas-nil-tes, "crusted snow". That is the Navajo year, dropped straight into a military codebook with the Gregorian labels bolted on the front.
Why could a Navajo speaker not read the code?
Because a substantial part of the code is in English, spelled out in Navajo.
The famous entries are metaphors, and a Navajo speaker would at least follow the picture: a tank is Chay-da-gahi, "tortoise", and a tank destroyer is Chay-da-gahi-nail-tsaidi, "tortoise killer". You could guess at that under fire.
But 29 entries in the general vocabulary are built the other way round. The first element is one of the alphabet words, contributing its English letter, and the rest of the compound supplies the rest of the English word by sound. These are English puns wearing Navajo clothes, and they are unguessable unless you have learned the table.
| Code word |
The book's literal translation |
The English word it carries |
| Tazi-cha |
turkey hat |
that |
| Gloe-eh-na-ah-wo-hai |
weasel hen |
when |
| Gloe-ih-qui-ah |
weasel here |
where |
| Klesh-so-dilzin |
snake pray |
spray |
| Mai-be-he-ahgan |
fox arm |
farm |
| Be-bih-zihde |
deer liver |
deliver |
| Be-beh-sha |
deer tail |
detail |
| Tazi-na-eh-dil-kid |
turkey ask |
task |
| Wola-chi-a-moffa-gahn |
ant fight |
about |
| Ta-zi-kwa-i-be-ka-di |
turkey here after |
thereafter |
One row in that table goes a step further out, and it is worth naming because it breaks the rule just described. "About" is Wola-chi-a-moffa-gahn, glossed "ant fight". Ant supplies the A as usual, but "fight" does not sound like the rest of the word. It is the English synonym bout that does the work. That single entry needs English twice over, once for the letter and once for a synonym, which is as far from "just speaking Navajo" as a code word can get.
The prettiest case is a set of three. "Sit" is Tkin-cid-da-hi, "ice sitting". "Set" is Dzeh-cid-da-hi, "elk sitting". "Sat" is Bih-la-sana-cid-da-hi, "apple sitting". Ice is a code word for I, elk for E, apple for A. One shared Navajo element for the verb, and the vowel that distinguishes the three English tenses is carried by a completely unrelated noun.
Longer prefixes work the same way. "District" is Be-thin-ya-ni-che, "deer ice strict", which is D plus I plus "strict". "Dispatch" is "dog is patch", where D is carried by dog rather than by deer, because most letters had more than one word to choose from. That freedom is itself part of the design, and the next section is about where it came from.
This is why the distinction between speaking Navajo and reading the code is not a technicality. The Marine Corps History Division records that alternate terms were provided in the code for letters frequently repeated in English, and that "all code talkers had to memorize both the primary and alternate code terms", because although training material was printed, security precautions curtailed its use in combat. The book existed so that it could be learned by heart and then left behind.
Why does one letter get three code words and another only one?
Because the alphabet was patched twice during the war, and the second patch went past its own stated rule.
The National Archives, working from Marine Corps records, describes the original code assembled at Camp Elliott in 1942: 211 words, plus a system covering the 26 letters of the English alphabet. Then a cryptographer monitoring the transmissions concluded the code might be broken, because spelling out words that were not in the vocabulary "produced too many repetitions". This is the classic weakness of any letter-for-letter substitution, and it is the same weakness a cipher identifier leans on when it fingerprints an unknown message.
The fix was alternates. The alphabet grew from 26 terms to 44 by adding extra words for the most frequently repeated letters, which the National Archives lists as E, T, A, O, I, N, S, H, R, D, L and U. The vocabulary went from 211 terms to 411 at the same time.
By June 1945 the alphabet stood at 63. Seventeen letters carry three words each, three letters carry two, and six carry one. Here is where it stops matching the stated rule:
- The six letters still on a single word are Q, V, W, X, Y and Z.
- The eight letters that went to three words despite never appearing on that frequency list are B, C, F, G, J, K, M and P.
So J and K each ended the war with three alternates, while W and Y ended it with one word apiece, and none of those four letters was named in the expansion that was supposed to be about frequency. Neither the codebook nor the Marine Corps account gives a reason. The honest answer is that the record shows what was done and not why, and it is worth resisting the tidy explanation, because the tidy explanation does not survive contact with the table.
Two smaller oddities are worth having. The scheme needs an everyday English word that starts with the letter and also has an everyday Navajo equivalent, and it holds for every letter but two, which strain it in different ways. X is Al-na-as-dzoh, and the literal translation is "cross". There is almost no common English noun beginning with X to reach for, so that one entry encodes what the letter looks like rather than what it sounds like. V is "victor", which does start with the right letter but is a name rather than a thing, and it is the same word the NATO phonetic alphabet later settled on for V.
Half the vocabulary is not code at all
Words the code had no metaphor for were simply said in Navajo.
Of the 551 general vocabulary entries, 282 have a literal translation identical to the English headword. "Ambush" is Khac-da, ambush. "Wound" is Cah-da-khi, wound. "Zone" is Bih-na-has-dzoh, zone. There is no cipher operating on those at all. The remaining 269 are substitutions or descriptions: "demolition" is "blow up", "rocket" is "sand boil", "artillery" is "many big guns", "casualty" is "put out of action", "booby trap" is "man trap", "camouflage" is "hid".
That mix is the thing to hold on to. Roughly half the general vocabulary is plain language and roughly half is code, bound into one alphabetical list, with a spelling alphabet in front of it for anything the list did not cover. A message could move between all three modes inside a single sentence, and only a trained talker knew which mode any given word was in.
That is what makes this a different animal from the codes people usually compare it to. Morse code is one fixed mapping applied to everything, so knowing the mapping is knowing the system. Here there are three systems at once, and the book carries no marker saying which one is running for any given word.
Was Navajo really an unwritten language?
Not quite, and the men who proposed the code knew it.
The proposal Philip Johnston sent ahead of the February 1942 demonstration is summarised in the National Archives' account of the records. He stressed that Navajo remained mostly "unwritten" because an alphabet or other symbols of purely native origin did not exist, and then conceded the exception in the same breath: adaptations by American scholars, anthropologists, and Franciscan Fathers, who had compiled a Navajo dictionary.
That dictionary is real and it is readable today. An Ethnologic Dictionary of the Navaho Language was published by the Franciscan Fathers at Saint Michaels, Arizona, in 1910. It opens with a "Key to the Alphabet" that works through the consonants adapted from English, then the sounds peculiar to Navaho, digraph by digraph, describing the aspirated l as a sound "produced by taking and holding the position for pronouncing an ordinary l and pressing the air through the sides of the mouth". It runs to hundreds of pages and includes the clan list quoted earlier.
Its copyright page also says this: "Edition limited to two hundred copies."
That is the real security picture, and it is more interesting than the legend. Navajo had been written down, in English, with a phonetic key, more than thirty years before the war. The National Park Service records the Marine Corps estimate that fewer than 28 people outside the tribe could understand the language. Scarcity, not absence, is what the Marine Corps was banking on.
Which is exactly why a code was layered on top. If the protection had been the language alone, the whole system would have rested on a book you could find in a research library. Instead the protection was the language plus 677 arbitrary assignments held in the memories of a few hundred men, and that is a different and much harder problem for an interceptor.
Who wrote the Navajo Code Talkers' Dictionary?
Philip Johnston proposed the programme. He did not write the code.
The Marine Corps History Division is direct about it: at Camp Pendleton, while the course covered basic communications procedures and equipment, "the 29 Navajos comprising the first group recruited devised Navajo words for military terms which were not part of their language". The National Archives account puts it as the initial recruits designing the first code along with communications personnel.
Both readings agree on the load-bearing point. The metaphors, the clan mapping, the seasonal month names and the English rebuses are the work of Navajo Marines making a language do something it had never been asked to do, under a deadline, for a military vocabulary that did not exist in it. Johnston's contribution was the argument that it could be done at all.
The Marine Corps' own case for the code also inverts what people expect. Its record notes that tests in the Pacific showed a classified message could be translated into Navajo, transmitted, received and translated back into English faster than the same message could be run through conventional cryptographic equipment. The code was not just harder to break. It was quicker.
Frequently asked questions
Could a Navajo speaker who was not a code talker understand the code?
No. Roughly half of the 551 general vocabulary entries in the June 1945 dictionary are ordinary Navajo, so a speaker would follow those, but the rest are substitutions and 29 of them are English rebuses that only work if you know the letter table. "That" is sent as Tazi-cha, whose literal meaning is "turkey hat": turkey supplies the letter T and hat supplies the rest of the English word by sound. Without the codebook in your head that is two unrelated nouns, and a message assembled from dozens of them is unreadable.
How many words were in the Navajo code?
The revision dated 15 June 1945 contains 677 code words: 63 in the alphabet, 63 across the named categories for organisations, officers, countries, aircraft, ships and months, and 551 in the general vocabulary. That is the end state, not the starting point. The National Archives records the first code built at Camp Elliott in 1942 as 211 words plus a 26-letter alphabet, later expanded to 411 words and a 44-term alphabet.
Why did some letters have more than one Navajo word?
To stop repetition betraying the message. A Marine cryptographer monitoring the traffic concluded the code might be broken because spelling out words not in the vocabulary produced too many repetitions, so extra words were added for the letters most often repeated in English: E, T, A, O, I, N, S, H, R, D, L and U. By the 1945 revision, though, the pattern had drifted well past that list. J and K carry three alternates each while W and Y carry one, and none of those four letters was named in the frequency expansion. The surviving documents do not explain the final allocation.
Was Navajo a written language before the war?
It had been written down, though not by the Navajo themselves and not widely. The Franciscan Fathers published An Ethnologic Dictionary of the Navaho Language at Saint Michaels, Arizona, in 1910, complete with a key to the alphabet covering consonants, digraphs and the sounds with no English equivalent. Philip Johnston's own proposal acknowledged this work while arguing the language was effectively unwritten. The copyright page of the 1910 book records an edition limited to two hundred copies, which is the more accurate picture: documented, but very hard to get hold of.
Who created the Navajo code?
Philip Johnston, a missionary's son who had lived among the Navajo, proposed the programme to the Marine Corps early in 1942. The code itself was devised by the first 29 Navajo recruits, working with Marine communications personnel during signals training. The Marine Corps History Division states that this first group devised the Navajo words for military terms which were not part of their language, and the code kept growing after them, reaching 677 entries by the June 1945 revision.
If you want to feel the difference between a spelling alphabet and a code, read the 1945 letter table beside the NATO phonetic alphabet. Both give every letter a word. Only one of them also hides a second language inside the words it chose.
Sources
- Navajo Code Talkers' Dictionary, Revised 15 June 1945 (Declassified under Department of Defense Directive 5200.9), Naval History and Heritage Command Online Reading Room - retrieved 10 August 2026. The Command's own copy can be slow to answer; an archived copy of the same page is held by the Internet Archive.
- Marine Corps History Division, Navajo Code Talkers in World War II, Marine Corps University - retrieved 10 August 2026 from the Internet Archive's copy, the university's own server having refused the request.
- Adam Jevec, "Semper Fidelis, Code Talkers", Prologue, Winter 2001, National Archives and Records Administration - retrieved 10 August 2026. Footnoted throughout to Record Group 127, Records of the U.S. Marine Corps, National Archives at College Park.
- The Franciscan Fathers, An Ethnologic Dictionary of the Navaho Language, Saint Michaels, Arizona, 1910 - retrieved 10 August 2026. Locators are the book's printed page numbers, not the scan's image numbers: Key to the Alphabet beginning at page 19; clan-derived personal names at pages 123 to 125; clan list in the Sociology chapter at pages 424 to 431.
- National Park Service, Legacy of the Navajo Code Talkers - retrieved 10 August 2026. Text courtesy of the United States Marine Corps Historical Center.
