We have a sample encoding of MND annotations, and I worked today with MT to figure out how best to map it onto the ISE XML that has to be generated from it. I've ended up with some conversion XSLT, along with a set of recommendations emailed to the team; if everyone agrees, I'll be able to a) fix the encoding done by DJ and SW, b) harden-up the TEI schema to help with this type of encoding, and c) convert it to what the ISE machinery needs.