Just to clarify something that keeps getting lost in this discussion: I am not proposing a substitution cipher! That has been repeatedly ruled out, and I agree with that conclusion. What I am proposing is something fundamentally different—an absorption cipher, in which function words, articles, and prefixes are integrated into adjacent words rather than being encoded separately. This changes everything about how one would statistically expect the text to behave.
The second point worth mentioning: The cipher appears to be poorly designed. Not because the author was unintelligent, but because this looks like a personal shorthand—something someone develops for their own use without the intention of creating a robust cryptographic system. Personal shorthands lose information. They are inherently ambiguous. It’s a point that many overlook, and one that can explain some statistical properties. Including repetitions, because the text is probably not a well-written recipe book.
A poor cipher is actually harder to decrypt than a good one, precisely because the ambiguity is unintentional and throws off any normal attempt at decryption. It wasn't designed so that someone else could decrypt it
As for earlier attempts related to the German language: Most of them focused on Standard High German and assumed a clean substitution. Very few specifically sought a phonetic-based spelling system for a abbreviated Bavarian dialect (but only that is consistent with the observed word lengths), and even fewer considered the possibility that the coding might be imprecise rather than systematic. These are different approaches to the search. I am not claiming that this has been proven. I am presenting a method and building my argument step by step.
The second point worth mentioning: The cipher appears to be poorly designed. Not because the author was unintelligent, but because this looks like a personal shorthand—something someone develops for their own use without the intention of creating a robust cryptographic system. Personal shorthands lose information. They are inherently ambiguous. It’s a point that many overlook, and one that can explain some statistical properties. Including repetitions, because the text is probably not a well-written recipe book.
A poor cipher is actually harder to decrypt than a good one, precisely because the ambiguity is unintentional and throws off any normal attempt at decryption. It wasn't designed so that someone else could decrypt it
As for earlier attempts related to the German language: Most of them focused on Standard High German and assumed a clean substitution. Very few specifically sought a phonetic-based spelling system for a abbreviated Bavarian dialect (but only that is consistent with the observed word lengths), and even fewer considered the possibility that the coding might be imprecise rather than systematic. These are different approaches to the search. I am not claiming that this has been proven. I am presenting a method and building my argument step by step.