That was the intention, though I didn't comment on it, due to the usual US-centric egotism ;) and the fact that the vowel & consonant arrays aren't broken out to be easily configured. But internationalization was on my mind when I wrote it. (In fact, most of my passwords are based on Hangeul - Korean - words, since that language and culture is of interest to me.)

I've briefly reviewed the third-order code on Tom Van Vleck's site - it's definitely an improvement. If I had bothered to research this at all before implementing it, I probably would have either integrated that idea, or decided the problem had already been solved and was becoming more complicated than I felt like dealing with ;)

Actually, while I did not know whether it had been done before, I was not so bold as to suppose noone had ever had the idea. (I should have mentioned that in my post. "While I dreamed this program up without knowingly copying someone else's idea, it may have been done before, somewhere.") I just didn't research it at all. (Possibly I was afraid I'd find that some monopoly had a current patent on the method, and I'd be opening myself up to a lawsuit or worse, a DMCA violation and imprisonment, by posting it here ;)

Thanks for the link. Maybe some day I'll modify the program to pay attention to frequent neighbors for each letter. Definitely a good idea.