Thanks, Tessa!  That should have worked.  But... I'm not getting good results yet.  I made the change and ran it on two documents.  One I got the exact same result as reported above (the three target lines smooshed together); on the second document, I got the entire OCR grab, instead of just the pattern match result only.

The Pattern Matching grabs the right details:

But the Field Token Editor does not:

I believe part of my problem is that the test run is fine, but when you actually do the QF run on the documents, the field token is not always taking just the pattern match, it's taking the entire OCR results (which is a lot more!).  Any ideas what to do about that?

And on the other one, probably what happened is it actually returned the right answer, just not separated again, because even though I told it to return each line as a multi-value item, it has not done that, it has returned all three lines as one value with no separation.

Hmm. Hard to say exactly what's happening without the documents and session in front of me but I recommend focusing on one document at a time and taking a look at the output pane and seeing if you can find any clues there. 

When you see it giving you the entire OCR results instead of just the pattern, that probably means that a matching pattern wasn't found in the text, and you have this option checked:

So playing with those options might also help you troubleshoot. 

That's right, that's what I have it on.