Hyphen in interlinear being treated differently?

[Jeff Heath]

I have two interlinear texts which include the same text: "al-awiin", but it is being analyzed two different ways. In one, it is correctly broken into two words:

The second one is just lumped together:

And before you ask, the hyphen is the same in both cases, the standard U+002D.

You may notice (and I just did), the breaking chain icon at the bottom of that second image. Hovering over it shows that it will "Break phrase into words". When I click that, it does just that, and I get what I want. But I know that I have never used that before, e.g. on that first analysis, which was analyzed some time ago.

So I'm guessing there has been a change sometime in interlinear processing. A hyphen used to always break up words, but people were having difficulties typing a hyphen that was word-forming (see https://groups.google.com/g/flex-list/c/RXIVmy_vxLY/m/DF1rV_nKBwAJ). So FLEx was changed to assume hyphen was word-forming, with a button added to allow you to break it up. Is that accurate?

Jeff

[Jeff Heath]

I just noticed a Bad Thing about this change. If I click on the chain icon to break the phrase into words, it appears in light blue: "A suggested analysis proposed by the FieldWorks program based on a previous user analysis for an identically spelled word", like this:

If I select another text, and then go back to this text, then I have lost the division of the phrase into individual words (i.e. it reverts to the second image above), and I have to Break the phrase into words all over again. That doesn't seem right. I shouldn't have to approve at least one of the words (which does seem to work) to have FLEx remember that the words have been divided, should I?