Hi, rainer, and thanks for the comments as always.
At the very beginning, the first two bars make a statement, and the next two bars answer it by using identical material an octave lower, differing only in the last two chords of bar 4. You play these two chords subito f. I'll let you off if your edition specifically marks them so, but I really don't like them that way.
You're correct, and my score (the Alfred for Volume 2) has it that way too, though I must confess I deliberately played it that way. It seemed to me as though those chords should be played a bit more assertively, as more of a surprise as you say, since they are indeed a bit unexpected in the reprise of the first figure. You're certainly right that I could experiment with doing it a bit less obtrusively. One overall thing to keep in mind, perhaps, is that I'm not sure the urtext of this particular sonata (I have been playing with a different edition recently since the second volume of my old Universal urtext needs mending) has any actual dynamic markings (if I'm remembering correctly), which may allow for greater artistic liberties to be taken.
Something jumped out at me in the 16th bar from the end (at approx 3'57" in the recording), namely the execution of the turn. A similar problem occurs two bars later (but was less obvious, I nearly failed to notice it). It sounds as though the fourth note of each turn is missing. I think this is just finger-trouble (note failing to sound) rather than a matter of interpretation (if it were interpretation, I would disagree with it). You do play all four notes at the corresponding places in bars 9 and 11.
Quite possible. Listening back, I'm not sure I hear a missing note, but that part does sound a bit weaker to me than the parallel figure in the exposition (most likely because of the greater leap in the left hand)