I’m not impressed. The Universal Declaration of Rights is only about 1,800 words long, and most of the sentences have a markedly similar grammatical structure. As a rule, you need ten thousand or more words to draw statistically meaningful results from such computational linguistic analysis, and since the sentence structures analyzed as the base are so similar, the UDR isn’t going to provide a varied enough base. Now, maybe if you were analyzing an encoded treaty text, this method might yield meaningful results, but I doubt that’s what the Voynich manuscript is.
Why don’t they try an 1,800 word sample from the Bible (which is translated into as many languages) and see if they get the same result. I’ll bet anything such an analysis will say Voynich is written in a language other than Hebrew.