I also increase the font size (a lot) and add a slightly opaque black background to it -- again it helps to lower the amount of eye fixations.
For this (and obvious aesthetic) reasons, speech bubbles are a bad idea.
I've seen anime subtitles where characters are color-coded; I find to be distracting instead of helpful; the same with the italicized font when the line belongs to someone off screen. Though I'm sure one could argue it improves the accessibility for those having hearing disabilities.
For me it's dubbing that's absolutely awful, because I want to hear the real voices of the actors, plus I learned English and a little Italian and French by watching movies with subtitles. I don't understand countries where dubbing is the norm, I hope it never catches on in my country.
The human vision is drawn to quick changes & movement, when you have an otherwise immobile scene that's mostly just dialog subtitles will pop in & out, I can't help but focus on them just like I would if something else suddenly popped into the scene.
If you have words pop into your vision you also can't help but read them, at least I can't, subconsciously reading the translation and hearing the original version at the same time is really distracting.
I found the only way to deal with them was to position myself behind some sofa so I could only watch the top 3/4 of the screen, but of course you miss a big part of the movie that way.
And when they are useful to me, like an english translation... why doesn't anyone put the captions near the speaker's mouths, so I can read and see what's going on at the same time?