Again, we didn't directly test the casting time, only the recast. The common consensus I've read is that casting reduction is 2x recast reduction, but for what it's worth I've never verified that myself, and can't say how reliable it really is. We tried for a bit to test spellcast reduction using a method I've used before to test rostrums/loquacious/warlocks involving interrupting long-casting spells by claiming the target of the spell from outside PT, but couldn't really find one with a sufficiently long cast time (I'd used AM1 in the past, obviously unavailable to whm/brd/smn). We then attempted to try to test by using a long cast self-buff and interrupting it, but for whatever reason due to her connection or PC specs or I don't know what, the numbers she'd get when the casting bar would halt had a precision of +/- 5% at least, compared to 1-2% for me.
As more people get Marduk body it may be worth testing directly, particularly if someone is a PS2 player or has access to one, I found when doing earlier Fast Cast tests it was more reliable than when I did similar tests later on PC. For PS2 the casting bar seems to halt at 75% on average with no fast cast (although you'd want to verify this anyway), although for my PC the average was around 72%, for reasons that escape my limited understanding of programming.
XI Wiki


