What I don't know is how in the world I'll drum up interest once the time for that comes, beyond posting it on GitHub and linking it on hacker news. I don't have a big Twitter following or anything, I've never spoken at a conference, etc.
How did languages like Elm, Zig, and Crystal get the word out at the very beginning? Anyone with suggestions or firsthand experience? Is there a blog post I can read?
Then provide lots of source code comparisons showing how doing (whatever) in Whizzy New Language is more pleasant/concise/bug-resistant than Old and Busted Language.
The other poster's idea of publishing papers is a good one, if you're shooting for an academic language for intellectual pursuits. If you want a language that people will use in the Real World, though, you're gonna need some Real World code.
The problem is that there are a lot of new languages coming out all the time.
Whatever route you go, make sure you tell people exactly what the advantages of Whizzy New Language are right at the top (at the beginning of the github README, in the abstract of the paper, whatever), because no one is gonna grovel through a bunch of docs to learn that.
I'm interested in new languages, so I usually at least click on announcements, but it doesn't take long for me to lose interest if I don't see some plausible major advantage within the first paragraph or two of text. Too much new material, not enough time to read it all.
Or maybe you don't have to be "in that field" (have a name/reputation there). Maybe it would be enough to start building the language, and if there are interesting/novel aspects, write them up in a paper, and start submitting that paper to PL conferences.