Just started the game and what struck me first as a rather large inconvenience was the dialogue representation in the game.
PoE apparently chooses to go the way Planescape: Torment did (as opposed to BG series) and along the regular dialogue replicas, it gives you also some fluff information around, describing the general circumstances (i.e. that you interlocutor shrugged or that he is looking at you very strangely or some such).
Now I admit I am not a huge fan of this, however I respect this design choice. However I think it is implemented rather poorly. I beleive the fluff text and dialogue text should be distinguished much more than by quotation marks and a very slight degree of brightness. The distinction should be either by different font colour (e.g. white and light blue) or different font type (italics / non-italics).
I am aware that e.g. in PsT, there was no such distinction and it did not pose any problems, however first of all this game was much more text heavy and, more importantly - there were comparably much fewer *voiced dialogues*. Now this is where the real problem enters the scene in PoE. On one hand, you have the text dialogue that gives you the dialogue replicas *and* the fluff text of various lengths, on the other, you have the voiced dialogue that gives you the dialogue replicas only.
You are forced to follow the text dialogue because it gives you more information in the fluff (some of the fluff descriptions are quite lengthly) but you are disturbed by the voice dialogue that *does not* follow the text you are reading at all and you get desynced very fast:
Text - Sound
You either need to stop reading the text and just listen to what they say (at least for me it is difficult to read fluff text and try to perceive what the character is saying aloud) or frantically try to find the text the character is saying (can you imagine the experience of people with not that good English?).
For the future, I would propose to forgo entirely the fluff text in all voiced dialogues. The information provided by the fluff text should be conveyed in a different way, i.e. combination of the contents of the dialogue and voice acting (or an actual animation if you feel daring), however this is obviously impossible for the existing dialogues, where I would go for a decent colour distinction. This should not be too hard to patch in since most of the speech is already pre-parsed very clearly by quotation marks.