Revealing Secrets In Voynich MS With The Help of Word Endings Analysis

Do you want to start revealing secrets in Voynich MS through case endings? If you haven’t watched it already, visit our playlist where we share earlier findings. Here, you can also find a link to the transliteration video where you hear cool sounding gibberish, or visit https://arichichi.com/category/good-feed/voynich/ to have early access to our content in text format, link is in the description…

With regards to deciphering the Voynich Manuscript, a special care is necessary to address case endings. If it is true that Voynichese is a syllabic language, we have to admit that the only way we could differentiate sound is through how words end. I will raise a special notice before we carry on. One might want to know what are the odds that a particular word has a specific function in the sentence. In so doing, I suggest that we have to consider disyllabic constructs ending words. Let us call such a disyllabic structure C1VC2E. E would be the ending. Then, I propose that color coding transliteration can help identify which constructs C1VC2 are predominant.

One such construct that I identified is the -NER ending, which of course comes before the last vowel of a word. Just look at the green ending -NER in various words of the screen shot below. Do you notice that most of them end in -o. We are certainly lucky if we make a rule out of this, but I want to thoroughly discuss the matter. We will study in the coming minutes the dimensions of ending vowels with respect to the general case, and the -NER case.

First and foremost, let us talk about some basic statistics of the Manuscript. We will thoroughly study all the pages with commands that I have run on my computer. The total amount of words in the digitized version of the Manuscript is 11174 words. Of these, 38% end with -a, 35% end with -e, and 26% end with -o. The remainder 8% comprises both -i and -u endings, with -i endings amounting to 761 words, and -u endings to only 50 words. How much of a shock is this to you? Do you think my intuition was wrong? I surely was shocked to notice this myself.

Let us recap. Had we studied the matter with the VC representation offered by FSG, we wouldn’t have noticed that. Because then, we would need to take into account the last two characters of the word. Besides, FSG doesn’t compress data by relevance, so you would have to study a lot of combinations to get the same result as ours. And, had you done so, you would get very sparse data, which would limit your chances of success.

Let us closely examine the following tables. I assure you, revealing secrets In the Voynich MS through case endings is a long journey. And we are not there yet.

In the first table, you can see the statistical measurement of the number of occurrences, in column 2, of the ending pattern -nerV where V is one of the five possible values. You can also see the occurrences of all -V endings in column 3. In table 2, we just calculate the percentages over each column, or rather, the probabilities that the word ends with vowel -V given that -V is preceded by -ner, in column 2. In column 3, you can find the non-conditional probabilities of occurrence of vowel -V at the end of a word.

As you can see, the probability components are very different from case to case. The following is a summary of the observations and some conclusions pertaining to this.

We can see that the -i ending is not that frequent and that it shows up in the same way with -ner and without -ner. At the same time, roots such as -nera or -nere are way less frequent than their counterparts, namely, -a and -e. We also see that while the presence of -o endings is somewhat average, it’s way more likely to find words ending with -nero rather than not. So here, we tie -ner with -o. Such a large value has to be recognized as a token of our discovery of some functional property of the language. Remember that all this has been done while listening to our so-called transliteration. Ain’t that cool?

Here is a concluding note on our findings and hypotheses that we might explore in the future. Please take a closer look and discover with us what lies ahead.

As you can see, we think so far that the -i word ending signifies that a text portion has reached its limit. We won’t count on lines and paragraphs, since the language doesn’t have punctuation marks. We will tend to approach the text in chunks of words until an -i ending word. So far, let me let you know that I have reasons to think that -nero is a verb ending. If there is -neri instead, it would imply that the verb comes at the end of a thought, so it potentially means that the verb is intransitive.

For those of you who need proof of the dissimilarity that we stated earlier, take a closer look at the value of the cosine similarity function: it is 0,47. which is about one-half. It means that the angle between both five-dimensional vectors is greater than 30°. That is way better proof of dissimilarity than any other value… I’m joking. Just in case, they can’t be orthogonal so we’re half-way there.

Why? Because all values are bound to be positive so they all belong to the same quarter hyper-plane. It is like in 2-d space where if two vectors are on the same side from both axis, their components have the same sign, therefore they are more likely to be similar. Here, they are not. On this positive note, I let you in your peregrinations around the web. This is just a small step ahead towards revealing secrets in the Voynich MS through case endings.