Zonai Script
from Nintendo's Tears of the Kingdom

This website is a collection of samples and notes on the use of the Zonai writing system in Tears of the Kingdom.

The game has many signs written in fictional alphabets: Gerudo, Hylian, and Sheikah. These are all known to be substitutions for the English alphabet, and were seen in Breath of the Wild.

Tears of the Kingdom introduces a new script for the Zonai (ゾナウ) people.

Structure

B C D H J L M N R S T U W Y

The text is usually written in vertical columns, but is sometimes written in rings. Less commonly, it is written in rows (often in rings wrapping around a column or ball).

I suspect that circular text is not written in a consistent direction, and I have annotated each sample with the direction I think it might be read in.

The text has a visual similarity to CJK logographs, especially oracle bone script / small seal script.

Zonai	Small Seal Script	Modern
W		笑
M		京

The game uses a handful of other symbols in the Zonai "language", such as those associated with the summonable champions, but they don't appear in larger passages. These are explicitly Japanese kanji. For example, the Fire Sage symbol is a variant of "炎", meaning "flames".

Previous installments in the Zelda franchise have created Hylian scripts which were simple ciphers for English or Japanese text. Recent games (including Breath of the Wild's Sheikah script) have been English ciphers (although some text was written in romanized Japanese using the English alphabet), but several factors point to this script rather being a cipher for Japanese:

• The script has a basic resemblance to old Chinese/Japanese writing
• Like classical Japanese, the script is written in columns
• 14 symbols is far too few to write intelligible English text
• Japanese is normally written in a syllabary of 45 symbols formed from 5 vowels and 9 consonants (plus several voicing marks)
  • Using voicing marks is a convention in modern Japanese; texts older than around 1850 wouldn't have used them regularly, making it appropriate that an "ancient" language based on Japanese wouldn't use them.

Observations

The text in the game is frequently reused in many different places. More interestingly, sometimes text has some slight variations to it.

For example, the string HNDBSLM appears in many places, notably on the columns surrounding the door to the inner chamber of a Shrine of Line.

This string also appears written bottom to top in the Water Temple, on the main pipes that dump water. However, of note, it is not a simple reversal, but a reversal in chunks of one or two letters:

This suggests some kind of grouping of the characters. If they encode romaji, possibly at the mora or kana level.

A similar, though quite different, kind of repetition with reversals can be found on the door exiting the Temple of Time.

Here, chunks of 1, 2, or 3 letters are reversed, or occassionally one letter is replaced.

Red Herrings

Zonai Pattern

There is a pattern which appears in many Zonai ruins throughout the world that looks something like this:

Although it often appears near Zonai text and looks like letters, this is not text. Every occurrence of it is exactly the same decorative sequence.

Ring Ruins Tablets

The tablets in the Ring Ruins are large corpuses of Zonai text. Unfortunately, they appear to be completely random: the distribution of letters is as-uniform-as-possible.

Therefore, it's not really possible that they form a substitution cipher for any language. I think it's best to ignore their contents until other samples are understood.

However, the structure of the Floating Ring Ruin is different. The letters aren't uniformly distributed, and the columns are spaced oddly.

Shrine of Light Voices

When a Shrine of Light "speaks" to Link, words appear on screen, with ghostly Zonai letters appearing behind them.

Those letters are randomly generated. They are not consistent between different players visiting the same shrines, and they are not consistent between shrines.

Alphabet Use

Many, many examples of Zonai text use only 7 of the 14 letters.

For example, the text that appears on the doors and walls of the Temple of Time and the Forgotten Temple use only BDHLMNS (they never use CJRTUWY).

This might suggest that these samples do not mean anything.

Trigram Frequency

Certain "phrases" are repeated between many different texts.

Try a cipher

Search for a Cipher

Technical Details

Japanese sample text

Japanese is normally written using two kinds of characters:

• kanji: There are around two thousand regularly used kanji.
  Their pronunciation changes completely depending on context.
  Each one generally has a "meaning" that is used in one or more words.
  For example, the kanji 食 means "food" or "eat":
  • 食taべbeるru: to eat
  • 食shoku堂dou: cafeteria
• kana: There are 50 kana symbols, in two forms (hiragana and katakana).
  Individual kana are purely phonetic and do not mean anything; they have essentially fixed pronunciations.
  Each kana represents a "syllable" combinining an initial consonant followed by a vowel.
  For example, the hiragana ほ represents the sound "ho".

I see no practical way to encode kanji using only 14 letters, so I presume any cipher is purely phonetic.

Japanese can be "spelled" using these 50 kana. Below is a sample of Japanese text taken from the Nintendo website, and its conversion to kana.

https://www.nintendo.co.jp/zelda/totk/character/index.html

城は宙へと浮かび上がり、空から謎の遺跡が降り注ぐ。 ハイラルは突如として、”天変地異”に見舞われた。 主人公リンクは、果てなき冒険に再び身を投じる。

かつてハイラルを救った勇者。 右手の力を駆使して、ハイラルの異変に立ち向かう。

聡明で好奇心旺盛なハイラル王国の姫君。 ハイラルを襲う天変地異に巻き込まれてしまう。

女性ばかりのゲルド族において、 百年に一度だけ生まれる男として生を受ける。 ハイラルを襲った天変地異と何か関係があるようだが・・・

しろはちゅうへとうかびあがり、そらからなぞのいせきがふりそそぐ。 ハイラルはとつじょとして、”てんぺんちい”にみまわれた。 しゅじんこうリンクは、はてなきぼうけんにふたたびみをとうじる。

かつてハイラルをすくったゆうしゃ。 みぎてのちからをくしして、ハイラルのいへんにたちむかう。

そうめいでこうきしんおうせいなハイラルおうこくのひめぎみ。 ハイラルをおそうてんぺんちいにまきこまれてしまう。

じょせいばかりのゲルドぞくにおいて、 ひゃくねんにいちどだけうまれるおとことしてせいをうける。 ハイラルをおそったてんぺんちいとなにかかんけいがあるようだが・・・

Because there are only 14 zonai symbols, they could not possibly encode the 50 kana, like what was done in the Ocarina of Time script.

Instead, I suspect the Zonai script is more likely based on a proper alphabet, such as the transliteration of Japanese into the Latin alphabet.

There are several ways to transliterate Japanese text into the Latin alphabet (referred to as "romaji", for Roman-characters). Here are the choices I made for the below transliteration:

• Grammatical particles are transliterated as they are written, not as they are pronounced
  (so を is "wo" not "o" and は is "ha" not "wa")
• Long vowels are transliterated as the two vowels
  (so おう is "ou" not "o" or "oh" or "ō")
• Irregularly pronounced consonants are spelled regularly
  (so ち is "ti" rather than "chi", つ is "tu" rather than "tu", しゅ is "syu" rather than "shu", ふ is "hu" rather than "fu")
• Small っ is written with a double consonant (so 救った is written "sukutta")

siro ha tyuu he to ukabiagari, sora kara nazo no iseki ga hurisosogu. hairaru ha totujo tosite, tenpentii ni mimawareta. syujinkou rinku ha, hatenaki bouken ni hutatabi mi wo toujiru.

katute hairaru wo sukutta yuusya. migite no tikara wo kusisite, hairaru no ihen ni tatimukau.

soumei de koukisin ousei na hairaru oukoku no himegimi. hairaru wo soutenpentii ni makikomaretesimau.

josei bakari no gerudozoku nioite, hyakunen ni itido dake umareru otokotosite sei wo ukeru. hairaru wo osotta tenpentii to nanika kankei ga aru you daga.

However, many of these features do not appear to be used by Zonai text.

• Double letters are extremely rare, so it's unlikely 救った would be spelled as "sukutta", because the small っ is very common in Japanese.
• Similarly, I removed double vowels like "ii" and "uu", but not "ou" or "ei".
• Irregular spellings are not used, because there aren't enough letters to include a "c" for "chi" or "f" for "fu"
• Voicing is probably not indicated, because there aren't enough letters to distinguish d and t, s and z and j, h and b and p, k and g, etc.
  • This is the case with the "Old Hylian" script from the Ocarina of Time.
• I chose to transliterate "small kana" Y- sounds without the preceding "i", and "small tsu" the same as a regular "tu", so a string like キャット ("kyatto") would become "kyatuto".

Below is the result of these operations. This text is what is used to generate the unigram and bigram tables for Japanese romaji above.