Changing model text for interlinearization

Question

From the beginning, we have been exporting our interlinear to a backtranslation project named A. We also made A the “model text for interlinearization”.

Then at some point, we made a second backtranslation project called B, but kept A as the model text for interlinearization.

Question 1: Is the above setup OK? As far as I know, it doesn’t seem to be causing any problems, but I’m not sure that it’s the optimal setup. In my mind, I think that it’d be better to make B the model text since it’s being continually updated with new data, but making such a change seems to turn previous glosses blue. Which leads me to the next question.

Question 2: Is there a way to change the model text to project B (or to an English Bible or a related national language Bible) without causing too many problems/work? Or what is the best way to change the model text as painlessly as possible?

Question 3: What is the principle for choosing a model text? I wish I’d asked this several years ago at the beginning of our project, but I’d still like to know in case someone asks me in the future.

Answer 1

Paratext stores the gloss data according to the language code of the model text. So if A and B have the same language code, I would expect all the glosses previously approved to stay black if you set the model text to B instead of A. If B has a different language code, I would expect all your glossing data to be lost when you change from A to B, not just glosses changing from black to blue.

If you decide to change the model text to one that has a different code, you can do some manual editing outside Paratext to transfer the gloss information to the new language ID (provided those glosses still make sense). We’ve come across this in migrating some projects to 8 from 7. This page is the writeup of how the gloss data is recorded:
[Link Removed]

Comments

Thank you.

A and B do not have the same language name but do have the same three-letter identifier when I look at Project Properties. But when I go to the Paratext registry, they have the same language name and the same identifier. What does this mean? Should I be able to change the model text for interlinearization from A to B without ill effects? If I try it and then things look messed up, would changing it back to A return the interlinearization back to the way it was?

---

I’m not sure why the project would have a different language name than the registry, but I think what is important for the interlinear data is the three letter code, plus whatever letters might have gotten added to it by filling in parts of the “Advanced” form (Script, region, or variant). If A and B have the same language code, the interlinear data should not be affected by changing the model from A to B.

---

OK. I’ll try it.

Going back to an earlier question, is it OK for me to keep the setup as is? The main possible disadvantage I see is that the model text (A, which is our previous backtranslation project) is incomplete since the interlinear is exporting to B. Is there a negative consequence to this setup? It doesn’t seem to be affecting the MTTs’ ability to do interlinearization, but could it cause any problems down the line.

I just thought of this. Could I keep A “updated” by restoring from B once in a while? As long as the language codes are the same, would there be no problems? Is this another possible solution?