There are vowels

Upon reflection, this post was ill-considered, but it remains true that the appearance of vowels in the Voynich language is strong. - Oct 22

While the language in the Voynich ms. is far from being as straightforward as it might appear, certain things are as they seem. One of them is that there are vowels. The language looks like it has vowels, and it does. This is obvious enough but it has not been sufficiently underlined.

There are degrees of certainty, of course, and we cannot be absolutely certain about this, as with so much else, but the evidence that the text has vowels (and therefore consonants) and is written in a script with both is so compelling that it can be safely taken as a solid working assumption. That's what we need in this research. Otherwise, we are wading through a swamp. 

 There will be those who leap to their feet here and say, "Wait on! You haven't proven that there are vowels!" It is not a matter of true/false proof; it is a matter of probabilities. The weight of evidence says that what look like vowels and behave like vowels are most likely vowels.

The Voynich letters  a <a>, e <e> and o <o> are more or less identical to the corresponding letters in the Roman alphabet which, in itself, suggests they are intended to be those vowels. The reader is being led to think so, anyway. These vowels have familiar Latin forms. It is the consonants in the Voynich alphabet that are unfamiliar. The script might be seen as a Latinate script where the vowels are retained but all the consonants have been changed.

Along with y <y> one or more of these letters are found in the vast majority of words, and the longer the word the more of them there are. Indeed, words can be broken into syllables based on these letters. There are as many syllables in a word as there are these letters. All of this suggests they not only look like vowels but are vowels.


It is impressive that the EVA renders Voynichese entirely pronounceable. It is quite possible to pronounce EVA Voynichese because EVA's creators took a <a>, e <e>, o <o> and y <y> as vowels, with i <i> as a modifier and the rest as consonants. It was a natural division to make on appearances and it is probably right and wise as far as it goes.

There are also clear signs of rules governing what consonants can follow what vowels, which is to say phonotactic rules concerning the felicitous combination of sounds.

There will always be people who insist it is not a language at all, or that everything about it is deceptive, but the plainest evidence is that it is a language with vowels and therefore with a phonology.

There are also quite a few people who argue that the Voynich alphabet is an abjad without any vowels, perhaps modeled on Hebrew or Arabic. That is certainly a line of study worth exploring, but it must discount the strong evidence for there being vowels, and Latin-like vowels at that.

The late Stephen Bax suggested that the alphabets of 17 letters depicted on f57v may show us the Voynich alphabet as an abjad, i.e. with the vowels omitted. It is a useful suggestion. But in the running text, and the labels, vowels seem like vowels and words seem to have a CVC type of structure. 

* * * 

Just for fun here, I point out that it can be common in simple ciphers to leave the vowels of the plaintext unmolested. Many encryption strategies treat vowels and consonants differently. Let us take a quote from Goethe:

There is nothing more frightful than ignorance in action.

Now, we leave the vowels intact:

e e i o i o e i u a i o a e i a i o

Next apply any sort of encryption you like to the consonants. Let's say, remove word spaces, reverse order, with the combination 'Th' replaced by X. 

NTCNCNRNGNXLFTHGRFRMGNXNSRX

Now add the vowels back into the sequence:

NETEICNOCINRNGONEXLIFTHGURFARIMGONAXEINASRIOX

Finally, divide the string up into word-like pieces:

NETE ICNO CINR NGON EXLIFTH GURFAR IMGON AXEINA SRIOX

The person reading the cipher will know that the vowels reveal the bones of the plaintext and are the key to decryption and it only remains to work out the consonants.

I'm not suggesting anything like such a device in the Voynich, although if it is encrypted then it is likely to be encrypted by a series of simple methods rather than some super-code, and it is possible that the vowels were left alone just as their glyphs haven't been changed. That is, the vowels were set aside while the text was encrypted as an abjad. You can have an encrypted abjad with the original vowels restored.

R. B.