pretty often lately I get annoyed at myself for not finding a way to get a phd in linguistics. I mean, I know that the academic and professional path I've taken gives me knowledge and insights I wouldn't have gotten otherwise, and I also know that I'm always going to be someone who makes things and not someone who studies things, but I really envy people with deep knowledge in linguistics
(this came up because I was making a plan for a poetry book and realized on the subway WAIT DISTRIBUTIONAL SEMANTICS ISN'T SEMANTICS AT ALL, IT DOESN'T TELL YOU ABOUT MEANING, JUST SYNTACTIC TENDENCIES only to google the whole thing and discover that yes, people have been having this debate for fifty years already. all this even though I *have* a BA in linguistics)
@aparrish huh, that way of putting it seems like it could also describe the origin of the Markov chain! that syntactic tendencies of text were divorced from meaning was precisely the point Markov wanted to make, i think?
{ can't find a great ref but here's a short one: https://www.crummy.com/2013/02/22/0 }
{{ my own syntax is tending now to the oulipo-stiff, apologies! }}
@oneiro @aparrish Something I didn't realize for an embarrassingly long time is that Chomsky's famous "colorless green ideas…" was specifically designed as refutation to the idea that *syntax* was distributional/probabilistic – it's constructed so every bigram is novel. Nowadays we'd say this refutes the idea that syntax could be a 1st order markov chain, but back then it wasn't clear that other probabilistic/distributional models were possible.