In the "long ago" I, a non-dancer, was a ballet cretin, and the Swan Lake/Giselle comparison was given to me so I could see the difference. Back then there were splendid versions of both ballets, readily available. But such was not the case with Sleeping Beauty. Of course, we've since lost Swan Lake too. (But we should all be confident that ABT's Mr. McKenzie will soon return Aurora to her proper environs.) With all the complaints about the state of ABT's aged Giselle sets, I'm sure we won't see that old tattered gem again. Will there be another choregrapher's name attached when Giselle comes back? Of course this is not a problem limited to what can be seen in America. The empirical luxury afforded me at my beginnings seems lost to the Genius of modern-day ADs.
...This moves me back to Paul Parish's post, which suggested that one that has to look at differences in specific movements.
For instance, if you take a particular dance movement -- port de bras, balances, cambres, rising on pointe or descending into plie, jumping, turning, or whatever -- is there something that Giselle should do ("romantic") that Odette/Odile does differently ("mixed") and Aurora even more differently ("classical")?
I thank Artist for beginning Paul Parish's project, her words resonate with real differences experienced in those early days. As empirical experience becomes less reliable, such inquiry becomes all the more important. This is a really difficult project, being akin to so many efforts at understanding emploi (put that in BT's Search and you'll find a book-length of material!) here at BT.
In the field of mathematical statistics there is something called variance explained: When comparing a given movement by dancers in two ballets each performed by a different company, I wonder which, the periods of the ballets or the styles (if any) of the companies, would contribute more to the variability we experience in that particular movement?