As I said, this actually takes quite a bit of time if you do it correctly. But I ran your picture/passage through the AI as a feasibility study using my AI trainings—(this, too, is NOT a solution to the VMS)—just to show how something like this works.
f75r.37 qokedy dy sheety qokedy qokeedy qokechdy lol
f75r.38 qokeedy qokeedy qokedy qokedy qokeedy ldy
f75r.39 yshedy qokeedy qokeedy olkeedy
The actual sequences produced are:
.37 endeinendenerehrtje
.38 nenendendenerha
.39 benenerrettete
The mapping remains consistent throughout:
EVA Core Consonants
ked nd
eet, keed n
kechd r
o hrtj
d rh
shed b
lkeed rr
te tt
old t
Bridge Vowel
y.qo, y.d, y.l, l.qo, y.o, y.k e
y.sh i
y.y a
It’s constructed as a sentence fragment:
[…]end einen, den er ehrt – jenen endenden Erhabenen – errettete.
roughly translated
…saved one whom he honors—that exalted one who has come to an end.
The phrasing is clearly contrived and melodramatic, and the grammar is somewhat questionable. AI isn’t really good at this; getting an LLM to think this way is very difficult.
[For example, it's impossible for an “o” to always be a “hrtj”...]
But it shows that it’s possible to produce phrases and sentences—even in such difficult passages.
----
So, in German, clusters like “nd” are actually very common, but so are “st,” “sch,” “ch,” etc.
Once you have passages that contain, for example, many “and”s, the same or similar structures appear over and over again. And if the texts are “structured”—that is, if they follow a specific pattern, such as in the description of plants or other topics—these patterns should repeat themselves.
The best example from German pharmacopoeias:
ist heiz unde trucken und gud / (it is hot and dry and good)
“Seich es durch ein tuech” = filter through a cloth /
“und leg es” and place it
“so gib im” (so give) etc.
This means that in the VBM, it’s not the tokens that repeat, but the clusters / vowels, and that leads to such similar structures.
And so these effects can be explained, at least in theory.
f75r.37 qokedy dy sheety qokedy qokeedy qokechdy lol
f75r.38 qokeedy qokeedy qokedy qokedy qokeedy ldy
f75r.39 yshedy qokeedy qokeedy olkeedy
The actual sequences produced are:
.37 endeinendenerehrtje
.38 nenendendenerha
.39 benenerrettete
The mapping remains consistent throughout:
EVA Core Consonants
ked nd
eet, keed n
kechd r
o hrtj
d rh
shed b
lkeed rr
te tt
old t
Bridge Vowel
y.qo, y.d, y.l, l.qo, y.o, y.k e
y.sh i
y.y a
It’s constructed as a sentence fragment:
[…]end einen, den er ehrt – jenen endenden Erhabenen – errettete.
roughly translated
…saved one whom he honors—that exalted one who has come to an end.
The phrasing is clearly contrived and melodramatic, and the grammar is somewhat questionable. AI isn’t really good at this; getting an LLM to think this way is very difficult.
[For example, it's impossible for an “o” to always be a “hrtj”...]But it shows that it’s possible to produce phrases and sentences—even in such difficult passages.
----
So, in German, clusters like “nd” are actually very common, but so are “st,” “sch,” “ch,” etc.
Once you have passages that contain, for example, many “and”s, the same or similar structures appear over and over again. And if the texts are “structured”—that is, if they follow a specific pattern, such as in the description of plants or other topics—these patterns should repeat themselves.
The best example from German pharmacopoeias:
ist heiz unde trucken und gud / (it is hot and dry and good)
“Seich es durch ein tuech” = filter through a cloth /
“und leg es” and place it
“so gib im” (so give) etc.
This means that in the VBM, it’s not the tokens that repeat, but the clusters / vowels, and that leads to such similar structures.
And so these effects can be explained, at least in theory.