Re: Marking "affix" in Toolbox DB for FLEx import
I think John's right. I've seen both approaches used, perhaps mostly as a matter of preference. (Note: only full forms can be converted into something that could import as a subentry.) But when one linguist uses both approach, I think the full term tends to be used for a derivative, and the brief affix for an inflected form. (The former would typically deserve a full subentry in FLEx, and the latter could just be stuffed into a custom paradigm field.) That's guesswork, though.
I think you're right that lexical relation isn't quite the correct field here; I'm guessing that the label/value pair was a tempting thing to hijack. Also, be aware that there can be quite a different mentality in the Philippine-branch dictionaries, which sometime even leaks into dictionaries that are otherwise pretty standard MDF. Sometimes you have to look through a lot of entries to see the pattern. (Solid can help with that.)
BTW, a full-scale PLB SFM lexicon following the old manual would be extremely difficulty to import into FLEx without a lot of skewing. For the gory details, see this:
http://wiki.lingtransoft.info/doku.php?id=tutorials:toolbox#specific_techniques_for_specific_needs
In a nutshell, PLB SFM uses senses in a stem-based structure to do roughly what subentries would do in a root-based structure. So, if it were possible for FLEx to import subentries of senses (which it currently cannot), then that could probably line up fairly closely and reduce the skewing a lot. But it might be better anyway to slog through restructuring the thing.
