I have listened to numerous recordings of the Tango on the Internet. Given that their sound quality is poor, the fact that my speakers are inexpensive, and the high speed of the ornaments, I found it impossible to hear exactly when the LH G# in measure 3 was played in relation to the F#-A-G figure in the RH.
It seemed as though the LH A and RH F# were played together in one recording that was slower than average. This would make sense in that the F#-A-G figure is similar to a mordent, which is never played before the beat in my listening experience. In slower Romantic Period music I've seen, upper mordents written as two small notes can clearly be heard on the beat in the recordings.
In the Tango, any way of playing the m.3 left and right-hand ornaments in terms of timing produces some degree of dissonance. There is so much radical dissonance in the piece overall, and I have no prior experience with this composer. Therefore, I'm concerned about playing these combined ornaments incorrectly, i.e. how much dissonance is too much, or is too little dissonance wrong?
Overall, however, I prefer the less dissonant interpretation noted above.
If anyone knows of a musical edition that specifies how these ornaments are played, please let me know.
The isolated G# ornaments in the left hand do sound like they are played before the beat. I know from past experience, however, that extremely short notes can sound before the beat, when in reality, they are on the beat.
For example, in Mozart K 545, mm. 22-23, I've always played the short appoggiaturas before the beat, because that's what they sound like to me on recordings. Apparently my ears are mistaken, as I recently discovered a scholarly edition that indicates on-the-beat timing. Please comment on this issue, raised in my latest post.