He does admit it's controversial.

I also wonder if he based his observations on Modern Welsh or Old or Middle Welsh. It's quite possible that the periphrastic do formation entered Welsh from English. (The thing to do would be to speak with a couple of Celticists or Welsh philologists.)

It seems to me that do not V for V not is something that started to happened around early Modern English (16th century CE). The more I ponder it, the less seriously I take it.

Ceci n'est pas un seing.