this idea has been kicking around for a long time, and sounds nice, but is there any data to support it? A lot of the most visible misspelling seems designed to avoid spam filter detection.
I used to buy into the idea, but it increasingly grates my intuition as time goes on. I'm at a point that I believe much of the misspelling, poor grammar, etc are not intentional. If the same scammers were better at what they did, they'd snare more marks. I think these scammers are only capable of exploiting the bottom of the barrel when it comes to discerning audiences, though.
However the paper itself doesn't present any evidence around the scammer's intention. Rather it presents a mathematical model under which it would make sense for a scammer to intentionally exclude a large swathe of victims, and it posited that misspellings is a way to achieve it.