What if Voynich plaintext is very heavily shortened?

[MarcoP]

Hi Rafal, in addition to what Koen and Nablator wrote, I'll add that You are not allowed to view links. Register or Login to view. has been discussing similar ideas for many years:

...Voynichese combines scribal abbreviation with strong elements of verbose cipher, along with other cipher tricks...

A verbose cipher accounts for the low character entropy and abbreviation accounts for the resulting insufficient average word length. One of the other tricks that Nick considers is the presence of nulls (which was indeed common in ciphers from that time). The gradual shift between Currier A and Currier B could suggest that the encoder(s) had some freedom in mapping plain-text words to cipher words (the idea of nulls could fit here). I think this implies that a single plain-text word can be encoded by different cipher words (e.g. Herbal A 'chey' could be equivalent to Herbal B 'chedy') with abbreviation also implying that a single cipher word can encode different plain text words (Sicyonio, Sicyonium -> scn, in your example).
If one assumes that, in a small enough subset of the text, each plain-text word is encoded consistently, it should still be possible to detect some grammatical patterns, like function words; this does not seem to be the case (Gaskell and Bowern, You are not allowed to view links. Register or Login to view.).

A notable feature of the VMS that has to our knowledge only been discussed by one other publication [20] is positive autocorrelation of word lengths. Word lengths in most meaningful texts are negatively autocorrelated: that is, long words tend to be interspersed with short words (long-short-long-short). By contrast, the VMS exhibits positive autocorrelation (long-long-short-short). Positive autocorrelation is only observed in a limited number of natural languages, but is common in gibberish (Figure 3).

The negative auto-correlation in a language like Latin is of course due to how shorter function words tend to alternate with longer content words.

De Sicyonio oleo. E praedictis licet et Sicyonium sic parare .

Word-length autocorrelation is only a special case of a more general Voynichese phenomenon: the fact that spatially close word tokens tend to be similar. Though cases like You are not allowed to view links. Register or Login to view. 37-39 are clearly extreme, the general phenomenon is a very prominent feature of Voynichese. It appears in both computer-generated (Timm and Schinner) and human-generated (Gaskell and Bowern) gibberish, but explaining it under the assumption of a word-to-word mapping with a natural language like Latin is hard.

<f75r.37,+P0>    qokedy dy sheety qokedy qokeedy qokechdy lol
<f75r.38,+P0>    qokeedy qokeedy qokedy qokedy qokeedy ldy
<f75r.39,+P0>    yshedy qokeedy qokeedy olkeedy otey koldy