Importing from Movable Type – wrong authors

I’m migrating a blog for the organisation I work for from a University based Movable Type platform (3.2) to a self-hosted WordPress blog, with which I’m familiar from hosting my own blog.

We’re having a couple of serious issues in importing all old posts and comments, one of which being that the exported text file has thousands of ??? throughout, which have apparently replaced a whole raft of different punctuation characters somewhat randomly. The movable type export function has no settings or options, so we’re stuck with it as it is, and I’m left going through the text file changing most of them manually. Woe is me.

Anyway, another problem that I’ve encountered is that half of the posts get attributed to the wrong author.

I’ll go through my process:

1. After receiving the exported text file from the University’s IT staff, I split it into two (because it is larger than 2MB)

2. Ran the mt.php importer on each split half.
-After the first, it asked what to do with the unknown authors (all were unknown at the time) so for each unique author from the MT system I created a new author in the WordPress system (though not of the same exact spelling – the reason being that they were previously university account IDs and not very semantic).
-After the second import, it asked again what to do with the authors, this time I had the option to select the author that I had created during the first import from a drop-down menu.

3. Upon checking posts, I found that the first import worked well, while the second had misattributed every post to one of the other authors, that is, all of john’s posts were attributed to mary, whose posts were attributed to eggbert, whose posts were attributed to morag, whose posts are attributed to john. There appears to be no order to this, as when I deleted all posts and started again, the misattributions were different, but still there. It’s as though all of an author’s posts are being given to a randomly selected remaining author.

I haven’t tried this, but I’ll have a crack at uploading the unsplit file to the server itself, and running the import that way. Hopefully the problem lies in reconciling posts with authors that already exist, and doing it this way will mean all authors are created from scratch.

Any other ideas? And any ideas on how I can overcome this ??? issue?