Phonetic notation experiment

[oshfdk]

If anyone is interested, here's how this cipher could have been broken very quickly with the tools and techniques available in the XV century (I have no idea how they actually would have approached a problem this, but nothing in the method below requires any modern math or cryptology).

If you know that the underlying language is English (even somewhat phonetically mangled), you can look for some easy cribs. Most useful cribs in English are combinations of some very short words with repeated letters. My favorite and the easiest is the following combination:

AB AC (two two-letter words one after another with the first letter being the same)

especially if you can also find:

AB AC D (same two words followed by a one letter word with a letter that is not present in the first two)

Overwhelmingly these will be "it is" and "it is a". There are other possibilities, like "in it" or "at an", but these are usually much rarer.

In the original ciphertext "hm hc" appears a few times, at least once as "hm hc b". After you substitute h => I, m => T, c => S, b => A, the result looks plausible (when I do exercises like this I normally keep the ciphertext in lowercase and the plaintext in uppercase to tell them apart easily and make the automated replacement work without any problems).

Then there are a number of short words with some of the letters already substituted, the most common are lA and I@. While lA is not immediately clear, I@ is most likely IN, since IT and IS are already taken.

The following is the full transcript of replacements I made. I had to adjust and remap a few times, but overall it was an easy process and I think one that a person from the XV century would be able to follow easily too.

hm hc b => IT IS A
h => I
m => T
c => S
b => A
common short words: ft, lA. Also, Tr, lI, I@ (in?)
@ => N
Very common lA and also lIS => the/this? let's make it phonetic Z
l => Z
IT IS A inANT ZAT eANI => that many?
e => M
AN% => and
% => D
A => E (better phonetic fit)
SIpSTI SEZENTI SENTIMITEwS => Z is V (phonetic TH & V), w is R, p is K
Z => V
w => R
p => K
DISK$VERED IN VE MEDITERENI#EN z$SIN => $ is A, z is B, # is J?


At this point the text is almost fully readable.

The exercises like this are the reason I'm almost absolutely certain that the Voynich Manuscript cannot be a simple substitution cipher over any plaintext language. Even for obscure languages certain grammatical structures would be very easy to identify and reverse engineer. These structures seem absent in the Voynich MS, mostly the text of the MS looks like a randomized combination of chunks of various lengths.