I personally think that text can, realistically, do a much, much better job adding character and nuance to events and conversations. With text, you can just throw tons of "he says, gingerly scratching his chin" kinds of characterization/contextualization lines everywhere all willy nilly, not to mention other behaviors and actions (like "he chugs the beer and slams the pint on the table, visibly annoyed" or whatever, I'm not a writer lol), even in the most minor conversations - it's really all up to the writer(s) how detailed they want to be.
With cinematic close-ups (assuming all conversations happen in the close-up perspective), you're gonna need unique animations for all of that sort of stuff, and that shit is downright expensive and time-consuming, so it's obviously not going to happen very frequently outside main story scenes and important conversations. There's also the expectation that all lines will be voiced, which might put a potential limit to the amount of different characters you can actually talk to.
I'd say I'm probably fine with either, personally, but I definitely see the strengths of just using text to do these.