13-12-2025, 01:03 PM
Thanks everybody! I didn't actually expect to get any response within 24h, I'm amazed how active the forum is!
And thanks for all the input. Turns out I did have the Currier separation wrong. I'm super happy I posted the base stats first before basing the rest of the analysis on this data.
So below are the recomputed plots with
- RF1b-er transcription
- Currier AB indication directly from the transcript
Thanks for all the hints!
It does change the plots meaningfully and resolves some of my confusions from the initial post.
1. The word length plots largely stay the same.
![[Image: word_length_stats_corrected_currier.png?raw=true]](https://github.com/Marvel4U/Voynich_semantic_exploration/blob/master/plots/word_length_stats_corrected_currier.png?raw=true)
2. The Zipf distribution actually changes quite a bit! Currier B is now much more clearly an outlier.
![[Image: Zipf_stats_corrected_currier.png?raw=true]](https://github.com/Marvel4U/Voynich_semantic_exploration/blob/master/plots/Zipf_stats_corrected_currier.png?raw=true)
3. The bigram heatmap also visibly changes. Mainly the pairs "ed" and "dy", but the two patterns now look less similar. However, still much more similar than the reference languages to each other.
![[Image: bigram_heatmap_a_b_corrected_currier.png?raw=true]](https://github.com/Marvel4U/Voynich_semantic_exploration/blob/master/plots/bigram_heatmap_a_b_corrected_currier.png?raw=true)
4. Word end trigram: This is the big one, the edy ending is now virtually exclusive to the Currier B. So yes, I had a pretty big mismatch. Great that you helped me to find it early!
![[Image: word_end_trigrams_a_b_corrected_currier.png]](https://github.com/Marvel4U/Voynich_semantic_exploration/raw/master/plots/word_end_trigrams_a_b_corrected_currier.png)
I fixed the changes on the You are not allowed to view links. Register or Login to view. but I'll leave them in the initial post. (I added a hint that they are incorrect.)
This does change the semantic analysis a bit, I'm doing my best to make the results visually intuitive to understand. I'll post them in a new thread when I'm ready.
And thanks for all the input. Turns out I did have the Currier separation wrong. I'm super happy I posted the base stats first before basing the rest of the analysis on this data.
So below are the recomputed plots with
- RF1b-er transcription
- Currier AB indication directly from the transcript
Thanks for all the hints!
It does change the plots meaningfully and resolves some of my confusions from the initial post.
1. The word length plots largely stay the same.
2. The Zipf distribution actually changes quite a bit! Currier B is now much more clearly an outlier.
3. The bigram heatmap also visibly changes. Mainly the pairs "ed" and "dy", but the two patterns now look less similar. However, still much more similar than the reference languages to each other.
4. Word end trigram: This is the big one, the edy ending is now virtually exclusive to the Currier B. So yes, I had a pretty big mismatch. Great that you helped me to find it early!
I fixed the changes on the You are not allowed to view links. Register or Login to view. but I'll leave them in the initial post. (I added a hint that they are incorrect.)
This does change the semantic analysis a bit, I'm doing my best to make the results visually intuitive to understand. I'll post them in a new thread when I'm ready.
![[+]](https://www.voynich.ninja/images/collapse_collapsed.png)