The Constructed Language Hypothesis

[nablator]

Now, in the frame of my question about a word-by-word substitution:
*if* a word-by-word translation can be ruled out on some basis (that we do not yet have)

But we do, in the case of a one-to-one substitution. You are not allowed to view links. Register or Login to view., which I replicated, show abnormally low values for a European natural language: vord pairs do not "work" together at any distance except distance 1 where there is a small but measurable effect. It is possible to write such a (short, like Cato's Distichs) text when the author is careful to never reuse the same pair of words at the same distance (>1) at a frequency superior to what would be found in a random shuffle, but it is extremely unlikely to have a long meaningful text like this.

In the article by Andrew Caruana, Colin Layfield and John Abela You are not allowed to view links. Register or Login to view., presented at the Malta Voynich conference 2022, they used a similar metric on word pairs, but less sensitive because of an arbitrary threshold for deciding which pairs are 'skewed'.