My interlinear gloss guesses seem to have been corrupted. Any thoughts?

Question

Any thoughts on why the interlinearizer is suddently generating unsual guesses with mixed upper/lower case red letters?

Here is are two examples from Gen 37.13:

2018-01-08_15-47-19.jpg1023×770 135 KB

Note the guess is fATHER and all the other possible guesses are also upper case.
And here is an example with a blue gloss:

2018-01-08_15-55-53.jpg872×606 100 KB

पुराना पोस्ट - इसकी मूल भाषा में दिखाया जा रहा है

Answer 1

I looked into your problem with the interlinear. There actually isn’t a bug in the Interlinear causing this. The problem is with the BTA project.

The problem is that the English language definition used by the BTA project got messed up and the character settings are not English settings (I’m not sure where they came from). They also define the lowercase characters as uppercase and vice versa. Because the language settings are shared between projects, this caused the settings for other English projects to also be messed up.

Fixing the characters in the English language definition (Project > Language Settings > Alphabetic Characters tab) fixed the interlinear issue.
Note that you may have to restart Paratext to see the fixed settings.

पुराना पोस्ट - इसकी मूल भाषा में दिखाया जा रहा है

Comments

That fixed it. Thanks so much anon291708. I will retract my help ticket with Paratext and try to learn how the language definition got changed.

पुराना पोस्ट - इसकी मूल भाषा में दिखाया जा रहा है

Answer 2

The Interlinear is making its guesses based on the “Model text for Interlinearization” that you have told it to use. Which project is that? What does its text look like?

Incidentally, our best practice approach for the Interlinearizer is to set it to use the back translation project (if you have one) as the model text.

पुराना पोस्ट - इसकी मूल भाषा में दिखाया जा रहा है

Comments

Dear dhigby,

Thank you for responding.

We’ve been using ESV as our interlinearizer model, which isn’t an all-caps translation as you know.

We have the same problem if we use the back translation project (BTA). It is generating all-caps or mixed-caps guesses and some of them don’t make any sense.

anon242106

2018-01-08_17-25-26.jpg1090×840 149 KB

2018-01-08_17-26-51.jpg1998×850 284 KB

पुराना पोस्ट - इसकी मूल भाषा में दिखाया जा रहा है

---

This is definitely a bug, please report it from the help menu.

⁣Sent from TypeApp ​

पुराना पोस्ट - इसकी मूल भाषा में दिखाया जा रहा है

---

dhigby:

This is definitely a bug, please report it from the help menu.

Thank you. I just reported it.

पुराना पोस्ट - इसकी मूल भाषा में दिखाया जा रहा है