Updated the first post with the text of page two .
Here's a photo of my notes for those who are curious:
I've been working on translating the text off and on for about a week, ever since I finished reading the book. I literally turned the last page and thought to myself, "Okay, now I NEED to know what was on those sketch pages." I was actually surprised no one had cracked it yet -- the first thing I did was hit the forums, figuring someone had already worked it out. I started to get excited when I realized that I had the chance to be the first person to decipher this "lost text".
The very first thing I did was to scan and print the pages, so I could scribble all over them without defacing the actual book. (sacrilege!)
There were a few observations I made right off the bat:
There was enough of a pattern that these were probably words and not just pretty-looking soundwave sketches.
There seemed to be a mix of short lables and longer sentences.
Multiple lines tend to be left-justified, so I assumed the text was written left to right.
One particular sequence seemed to pop up all over the place. It looked sort of like this:
<,,|The tall
I shape seemed to mark the start of a formal sentence
For a while, I puzzled over the Double Eye diagram in the lower right of page 2, trying to link it up with the illustration inside the front cover. Turns out that was a blind alley -- but I was able to piece together various sources to positively identify half of the gems on the inside cover. Going clockwise from the upper right they are: Sapphire, Smokestone, Ruby, Diamond, Emerald. This follows the sequence in the table in the Ars Arcanum, so presumably the sequence continues all the way around. But I digress.
After exhausting any chance of a quick and easy correlation between the table of essences and the Double Eye diagram, I turned back to the "brute force" method.
To begin, I needed to determine how many graphemes the Alethi alphabet had so that I could construct a key. (A grapheme is the smallest unit in written language -- i.e. letters in the English alphabet, Chinese characters, individual symbols in a Glyphward, etc. It generally, but not always, corresponds to a single sound, or "phoneme".)
I picked a couple of the lines and practiced copying them down to get a feel for the letters. I noticed that the letters seemed to come in three heights, which I ingeniously labelled:
1, 2, and 3. I also identified three basic shapes:
left arrow, right arrow, diamond, hash, and fancy. The hashes were particularity tricky to differentiate, because it was pretty clear that sometimes multiple hashes made up a single grapheme.
Once I had a notation system worked out, I drew up a key of possible letters and converted a chunk of sentence-looking script. This yielded something like: 2L 113 3F 211 3L 2 1R 2L 3L etc.
I used this data to run a character frequency analysis, which is essentially counting how often each letter comes up.
Character frequency analysis is based on the fact that given a random chunk of text, some letters tend to be more common than others. (For example, there are way more E's in this post than Z's). Fortunately, there are online cryptography tools for doing this sort of thing, so I was able to feed in my typed notation and get a pretty chart.
I did this mainly to verify that the text was in English and not Native Alethi or indecipherable gibberish. I suspected it was English because of the lack of palindrome words and the fact that Shallan's pages were in English, but I didn't want to launch into a code-breaking session using all the wrong tricks.
Fortunately, the results of the distribution came out sufficiently close to English that I was encouraged to move forward.
Given a large enough body of text, you can use character frequency analysis to crack any simple cipher via the mighty power of statistics. I didn't feel like typing up all the notation that would require -- and it didn't have nearly the allure of poring over the beautiful illustrations themselves. I also wasn't 100% confident that I had properly identified the hash graphemes, which would throw off all the statistics -- potentially making such an exercise a complete waste of time.
I did get one useful thing out of the frequency analysis, because it told me which Alethi graphemes were most common. I highlighted those on my key, because they would probably end up corresponding to the English letters
etaoin shrdlu.
Blank key in hand, I went hunting for small words. Small words are great for codebreaking because they are simple and common -- words like
the, is, and, etc. They're an excellent place to start looking for patterns.
Right away, I focused on the sequence I had circled earlier: <,,| (3L 113)
The thing that caught my eye was that in addition to coming up often, it occasionally came at the start of a sentence, and it never came at the end of a sentence. It seemed extremely likely that it was the word
THE. The tricky part was that the Alethi word had four symbols, which I had initially grouped into two graphemes. I suspected that written Alethi might have a unique grapheme for the phoneme
TH, because of the way palindromes like
Alethela are formed. I also noticed word further down the page that included the string of symbols I had decided was THE, along with a few extra letters. Presumably it was something like THEIR or THERE, but I couldn't pin it down exactly because I was still trying to get my TH grapheme to work.
This threw me for a while, so I took a break and worked on the other page.
My second big break came from Page 2, where I noticed a couple of long words that all ended with the same symbol (3R). The only thing this could really be was S.
Armed with my key consisting of TH, S, and E, I attacked the list of small words that I had copied. I didn't make much headway, so I split the TH into separate T and H graphemes and gave that a try. This helped me narrow down the possibilities. "This letter is either I or A, and that's an N or T. But T is already taken... and so on."
The first word I actually translated (not just assumed the meaning of like THE) was "
SPREN".
The rest of the line was still a mess: ?HRPESPREN LRAPPET ?N ?ERAR?R -- but that was the point at which I knew I was on the right track. From there, everything swiftly fell into place.
It turns out that Alethi script is written partially phonetically, and the correspondence between English and Alethi phoneme/grapheme pairs is not 1:1. For example, the English C is written as either K or S depending on its pronunciation. The text I've posted here has been cleaned up somewhat for the ease of reading, and is not an exact transliteration. Sometimes it's hard to tell individual letters apart, because they share similar forms. The sequence
RI, for instance looks very similar to
LO. It makes translation somewhat of an art, and ambiguous words need to be verified by context -- which made the shorter labels tougher to work out than the long sentences.
((Note: the following speculative paragraph is... wrong, as is pointed out later in the thread. =P))
For instance, the English letter F is written in Alethi as PH, while The similar-looking Alethi letters for T and D seem to be largely interchangeable, while the letter H can have two different forms depending on neighboring consonants. Several letters (j, l, v) are written as another consonant followed by a vowel. For example, the word fabrial is technically spelled PHABRIARO. The Alethi alphabet is rather elegant in the way it organizes phonemes, and graphemes seem to be systematically derived from the sound properties of their corresponding phonemes, rather than simply being abstract symbols. Each grapheme has two elements that describe its sound properties: shape and size. The different shapes correspond to the location in the mouth the sounds are produced: hashes are vowels, left and right arrows are various alveolar consonants, diamonds are bilabial consonants, and fancies are velar consonants. Height relates to breath control - the amount of stress and voicing the sound recieves. The taller the line, the greater the stress.
Reading written Alethi script evokes speaking the word aloud in a much more direct fashion than English letters, with the sound rising and falling with the letters' curves.
I thoroughly enjoyed not only learning the intriguing revelations hidden on these secret pages -- but the entire process of deciphering them. My sincere thanks to everyone involved with including them in the book. This is the sort of attention to detail that makes a truly great work of art.
