How big is the file? Can you remember any of its contents? If you're really desperate and it is indeed a raw text I guess you could load up a live Linux CD and try something like:
Code:
strings /dev/[disk] | grep "some string probably unique to the file" > results.log
That assumes that exFat stored the file contiguously. If there's junk in between chunks of the file, but they're still close together, then using "grep -A100 -B100" would give you the 100 readable strings before and after "some string probably unique to the file". Use the "-i" flag on grep for case insensitivity.
This is completely untested and just popped out of my head. So no warranty. Also it'll be slow as hell (well actually provided the grep string is unique enough it'll just be however long it takes to read the entire disk).