I'd like to clarify some things here. In certain usage scenarios, you absolutely do want to use XML validation rather than enabling validation inside the disassembler.
Let's say you have a flat file with 6 messages. 5 of them are compliant with your body schema, and 1 is not. You want to enable failed message routing, but you don't want the entire transaction to fail.
If you enable validation through the FF Disassembler along with Recovery Interchange Processing, it will work... but only partially! If message #4 is the corrupt one, only messages 1,2, and 3 will successfully route. 4,5,and 6 will be routed as one huge failed message.
HOWEVER if you define the Head/Body schemas in the disassembler, disable validation, disable RIP, and add an XML Validator that uses the body schema with RIP, then messages 1,2,3,5, and 6 will successfully route. Only message #4 will be routed as a failed message.
EDIT: Here is where you can also find this information in perhaps a more digestible format.
Thanks again to @boatseller for the clarification, and here's a very real-world situation in which I'm having to use XML validators in this pipeline.