Sometimes I wonder how many things we've "proven" because our studies aren't large enough to observe all the edge cases. I sometimes wish I could contact the researchers and volunteer myself as a counter-example.

I haven't hooked myself up to electrodes and measured nerve response, but as far as I can tell I don't subvocalise when I read. I can also turn off my internal monologue; I don't have to hear the words in my head as I read them though I typically do.

I tried this when reading Ready Player One (I counted to eight in my head to stop vocalizing). I did read it much faster (in fact, faster than my partner who is a native English speaker/reader). I cannot vouch for how much I remember of the book though. I think it's less, because I remember watching the movie thinking "oh, right, that's what happened" about a major plot part. That's never happened with other books that I first read, then watched the movie/series adaptation of.

It's hard to avoid subvocalizing when you're thinking about avoiding it. It's like not thinking about the pink elephant.

I feel like that experiment would be easy to contaminate..

Sometimes if I'm stuck in speech/subvocalizing mode it can take awhile to get out of it.

I think the easiest way to tell if I'm subvocalizing or not by checking my current reading speed. Subvocalizing can slow things down.

Also I think I'm more liable to subvocalize during fiction, with lots of dialogue involved. Acting out the characters in my head slightly.