I believe you’re incorrect about how well the text reflects other languages - from the skeptoid page:
The “complete nonsense” theory has one thing working against it. If it is nonsense, it’s very good nonsense. It’s almost too good to expect of an amateur. Computational analysis of the text has been run, exhaustively, many times by many different researchers, using many different techniques. This allows us not only to try and translate it (at which all attempts have met utter failure), but also to compare its metrics to those of actual languages. The letter frequency, word length, and word frequency are very similar to what we see in real languages. But they don’t quite match those of any real languages.
