Diacritic error breaking the Interlinear

Question

I am having an issue with a project I work with. They have several diacritics but one in particular is not working properly in the interlinear. It breaks the words in the interlinear.

Here is what the word should look like:

But this is what it is showing:

At first I thought it was because I had not put the diacritic in the list of Non-Standard diacritics under Language Settings. So I added it there. But it would stay added.

Here I added the Diacritic:

And here after closing Language Settings and opening it up again the diacritic has morphed into something else.

Note: This only happens with one of the four non-standard diacritics we use. The three working diacritics are:
COMBINING CARON : hacek, V above U+030C
COMBINING CIRCUMFLEX ACCENT BELOW U+032D
COMBINING ALMOST EQUAL TO ABOVE U+034C

The only one that does not work correctly in the interlinear is:
GRANTHA SIGN NUKTA U+1133C

What am I doing wrong? Is there some setting I missed?

Answer 1

Generic.User,

I see that the one that is not working does not have the word “combined” as part of its identity so it perhaps needs special treatment.

I suggest that you list the characters which use it in the language settings alphabetic list as digraphs. Paratext help explains this, but for a simple example, in Roman Script, a digraph of kp would be listed as:

k/K

kp

l/L

You can do the same for your characters, in fact they are particularly helpful with complex scripts. Then Paratext will recognize it as part of a single character.

You should also see them properly listed as single characters in the Character Inventory when combined is ticked.

If you need help with this, you can reach out to me privately and I can walk you through it (or explore other options) via Skype or Zoom or other chats which allow screen-sharing.

Comments

Thank you so much.

I admit I was dreading having to enter in all the possible character combinations that use the diacritic because it could have been huge. But as soon as I entered in the most common one just to test if your solution would work, everything fixed it self.

Basically once I listed “த☐☐” (I presume these will just be boxes for you all, but the second box is the diacritic) in the list of Alphabetic Characters, the diacritic was accepted everywhere and the interlinear combined correctly.

The only thing I needed to add.

The Fixed Interlinear.

2021-04-28 08_14_35-Paratext 9.1746×264 6.47 KB

Answer 2

One other thing I note is that the three working items are all defined as combining characters. The character that doesn’t work (U+1133C) is not combining. The character it is transforming to is U+133c which is ጼ. It might be that there is a bug inserting a 5 character code in this dialog. You should use Help > Give Feedback to report this issue.

Comments

anon848905:

The character it is transforming to is U+133c which is ጼ. It might be that there is a bug inserting a 5 character code in this dialog

Good catch. I doubt we expect surrogate pairs (needed for any Unicode character above U+FFFF) to be put into those fields.

---

Oh I should have thought of that. Because when I was fiddling with autocorrect.txt to create a quick way of imputing U+1133c (just to test it out) I had to use the surrogate pair (\ud804\udf3c).

Once I switched to Keyman and modified an existing NLCI Keyboard I did not have to do that anymore since Paratext 8 and 9 seemed to work fine without having to use the surrogate pair.

---

So just to test it out I used the surrogate pair and once more the character appeared like it should have until I closed the dialogue box. However once I reopened Language Settings it had gone back to being the wrong character.

As @anon848905 suggested I just reported the issue.

Answer 3

Indic scripts (such this example) have many characters that combine (even consonants with consonants) but do not have the word COMBINING in the character name.

There is another nukta character that might be used (U+1133B COMBINING BINDU BELOW) but despite having the word COMBINING in its name there are still problems with some applications.