02BC treatment in PT9.5

Question

The treatment of 02BC in PT9.5 seems to have been changed. I am using 02BC as apostrophe for elision. It is defined as a word-medial character in Language - Settings.

However when running checks, it is treated as a diacritic by paratext  :

What is your advice ? What am I missing?
Thanks in advance for your help.

Comments

Can you verify that the 02BC does not occur in the Alphabetic Character settings also?

---

Thanks for responding Phil. I can conform that 02BC does not occur is the Alphabetic settings, but only in the "other character" tab:

---

In my experience with Paratext we have always used 02bc as a standalone glottal stop, and this has been recommended by SIL techs for years. Here's a clip from PT Help that was posted in this forum in 2022 Glottal as apostrophe?.

I have not yet noticed this new behaviour in PT 9.5, but in my experience it is new behaviour for SIL software in general to date. PT through v 9.4, Fieldworks, Bloom, etc, LibreOffice (can't speak for Word) and PTX Print (I think) have treated 02bc as a separate character, which is the linguistic status of glottal in the languages I have worked with, and not as a diacritic attached to another glyph. I have been assuming that adding it to the alphabetical inventory of a language marks it as a standalone character,  even though its Unicode classification is "Modifier letter".

This is a bit alarming to me as many languages have used 02bc in extensive libraries. Will this change how we use this character in spellchecking, Biblical Terms and other tools? Does it attach to the preceding or following glyph (in our case, to neither, and to assume either one would cause confusion for readers and editors alike).

Is there a way to force PT and other upcoming programs to classify it either way, depending on the writing system specification?

---

The above is actually a reply to this comment. Sorry I got it posted in the wrong place!

According to Unicode, 02bc is considered a diacritic so I think the current behavior is correct. I'm guessing the previous behavior was a bug. I'm not sure what we can do about this problem, but giving feedback is definitely the best way forward.

---

As far as I know, the impact of whatever change was made is in the way the character is displayed in the Characters Inventory. This should not impact how words are displayed in the Wordlist, Biblical Terms or other tools.

If you see any other impacts of the current checking - please report it.

---

Just for some added context and an additioan question below: With the coming of more in depth checks as of paratext 6, this meant that the use of u2019 for both elision (apostrophe) and closing citation quotes caused paratext to produce errors. On Paper and on screen we were fine of course ;-) . To get rid of the error messages in Paratext the advice was to start to start use u02BC for those needing it to mark elision. The use of u02BC for glottals I think is considered its "normal" use, although according to the unicode explanation uA78C is the actual normal alphabetic letter for a glottal.
We could in Paratext just approve all combinations of [C]+02BC as valid in the character inventory and not worry about it. But I am wondering if in Flex, a word word with a word medial 02BC would be "detected" as a new lexeme? In a language allowing with these kind of lexemes (in flex), it should of course detect it (via control-D).

So, would we need a new feature in Paratext so that paratext will be able to distinguish between languages using 02BC as a diacritic (rightly so, i.e. modifier) and languages using 02BC as a normal letter, either glottal or "replacing" a letter that is not pronouced, i.e. elision ? (A tickbox with : please indicate how you are using 02BC ?)

Answer 1

02BC is defined as a modifier letter in Unicode. I believe that what is happening is that even though you are placing it in word medial, it is still considered as a letter and shows up in combination with other letters.
Please use the Help > Give Feedback to report this as an issue to the developers.

Comments

Modifier letters in Unicode are always considered diacritics to Paratext. This has always been the case. What confuses me is that it sounds like there was a behavior change from 9.4 to 9.5. 

---

I believe this another one of those places where Paratext 9.5 is checking things that weren't necessarily checked in 9.4.

---

Foolrunning - I just tested PT8 and see that the 02BC is isolated. In 9.5 it is combined with the preceding character.

---

I dug a little deeper, and you're absolutely right. There was a change in 9.5 that I missed. 

According to Unicode, 02BC is considered a diacritic so I think the current behavior is correct. I'm guessing the previous behavior was a bug. I'm not sure what we can do about this problem, but giving feedback is definitely the best way forward.

---

Looking at the help in Flex concerning 02BC, we read the following:

02BC MODIFIER LETTER APOSTROPHE is defined as word-forming in the Unicode Standard so its use would be appropriate, .....

If you already have a special keyboard, adding 02BC probably would not be too hard. Otherwise, use 0027 APOSTROPHE. This does not have the rounded appearance, but it is available in all fonts and does not require anything special to enter it.

At the time we made the choice for 02BC, it was exactly the rounded appearance, which "tipped the balance" in favour for 02BC as compared to 0027. We wanted the printed Scriptures (as well as the app) have the hand-written curly-type apostrophe. We could go back to 0027, if Charis has an alternative curly glyph, but it would also require changing keyman files across the board. (My guess is at least for all West Africa, where the use of 02BC for elision, has become common practice).

Looking at the Glyph set of 0027, there is no alternative glyph for the "ugly" ' ... Looking at the image below of the Charis 7.0 documentation there seems to have been a change in the way the apostrophe is now implemented (although I cannot find a mention of it in the documentation)

If I compare Charis SIL 6.2 with Charis 7.0 there does not seem te be a difference:

Our readers have become used to the curly apostrophe, therefore 0027 is (still) not an option. I think for the time being the choices are 1: go back to Paratext 9.4 for the time being. 2: validate all combinations of [C]+02BC in the character inventory.
The questions remains whether the change from 9.4 to 9.5 affects the bridge function to Flex for words with a medial 02BC.