Unwanted "Arabic" in epub metadata

[Don Mitchell]

I haven't been able to find the right search terms, I guess, because I can't believe I'm the only one seeing this issue. I'm using the 20.x series on MacOS Sequoia.

If I export an ID project as a reflowable epub, and then inspect the metadata using Calibre, I find that my epub is tagged Languages: "Arabic, English." 

There is no Arabic in the project, although in many cases there are words in other languages (Hawaiian, and another South Pacific language....) and diacritical marks such as a macron are there also. 

This causes a problem when I use Kindle Previewer 3 to create a .mobi file (which I often need to produce). Previewer won't export a mobi until I go back to Calibre and edit out "Arabic."

This isn't a big problem, because I expect it and know the work-around -- but I'd really like to know why this is happening. 

I should add that this has been happening for several years. It's nothing new. I'm just finally getting around to asking about it. 

[Joel Cherney]

There is no Arabic in the project, although in many cases there are words in other languages (Hawaiian, and another South Pacific language....) and diacritical marks such as a macron are there also. 

I don't do much epub work - and when I do, InDesign is rarely involved - so I honestly don't know that much about why this is happening. However, I'm in the last stage of a project where it was important for me to have the Marshallese, Palauan, Pohnpeian, Chuukese, and Kosraean interactive PDFs language-tagged. So if it's necessary, I can show you how to add language tags for e.g. Māori to InDesign.

I also know that getting the language tags correct for multilingual epub export is important, and I have just checked out the preferences under File -> File Info, and the EPUB Export preferences, and the File Information dialog, and the epub export settings that you can find when you right-click on an object and select "Object Export Settings." I didn't find anything language-related in any of those places. That is in line with my assumption that the only place you can affect language settings is in paragraph and character styles. If you have Hawaiian or Māori or whatever in your epub, and you don't want to go throught the process of adding those tags to InDesign, those words should probably be marked with character styles having "[No Language]".

(Or alternately, you could tag all your Māori as another language Romanian, and then edit your CSS after export? But I'm going to tag @James Gifford—NitroPress in to your thread, as he's the forums regular I know to be expert in epub export from InDesign, and he always says that post-export editing shouldn't be necessary. Honestly I wouldn't know, as I am constantly elbow-deep in raw text formats manually editing Things That Experts Agree Oughtn't Be Edited Manually. I'm ready to be told by James that I'm doing it wrong, for what it's worth.)

So: do you consistently have styles applied throughout your document to all text, without overrides? Have you specified the language in each one? Do you maybe have a spurious Arabic tag appended somewhere in your template or base styles? Do you (kinda reaching, here) have a Middle East/North African install of InDesign for doing right-to-left work? Are you willing to share a sample epub that is flagging as "Arabic, English" for me to rip open, contra James' advice, to find where spurious "Arabic" bits may be hiding?

[Don Mitchell]

Thanks! I use a lot of styles, and yesterday I checked each one in the most recent project, and they all declare "English." And I don't carry over styles from project to project. 

I did have a book project where there were some Arabic characters, but this problem appeared long before that. 

The other Pacific language (besides Hawaiian) is one from Bougainville Island, Papua New Guinea. And there, as in Hawaiian, the only thing unusual is what in Hawaiian is called the "okina," a backwards single quote within a word.

However I might not have been as clear as I should have been...the project I'm on now does have Hawaiian words but almost all of the others which give me the same problem don't -- most don't have anything but English words. 

[Joel Cherney]

but almost all of the others which give me the same problem don't -- most don't have anything but English words. 

Yep, I was absolutely operating under the false assumption that something weird and not in plain-vanilla Latin script - that is, something In My Wheelhouse - was causing your issue. 

That being said, one of the rarer InDesign bugs that I've kept an eye out for, the last decade or so, is the occasional case where someone with a US English install of InDesign has some Arabic stuff unintentionally turned on. It's usually not something the user did; the base us_EN install I assume you use has all of the Arabic tools under the hood, so you can use the same INDD files that I generate with my Arabic/English install, but you don't have to see all of the right-to-left tools in your interface.  So sometimes damaged preferences can cause issues similar to this one, where something Arabic gets turned on under the hood, but your interface doesn't have a UI element to turn it off.  Sometimes, the fix for this rare bug is to reset all of your preferences. In other cases, a one-line script (to e.g. fix a spurious right-to-left paragraph setting in the [Basic Paragraph Style]) can resolve the issue. 

Probably better to wait for other regulars, who might make epubs more frequently than myself, to sound off regarding your issue. I'm clearly still stuck In My Wheelhouse, saying "but maybe it's this other rare thing that only happens when complex scripts are involved!" (Alternately I still invite you to post a sample epub, and ideally the sample indd used to generate said epub, to see if we can track down a real spurious Arabic thing somewhere in your default settings.)

[Laura Brady]

This is buggy InDesign behaviour. It's very likely that the Arabic effects only some numerals in the ebook. The best way to get rid of it is to make sure that there is no Arabic declared in the book. You might check that the numbers haven't been transformed into Arabic numbers in the ebook in addition to removing the Arabic language declaration. 

[Don Mitchell]

Thanks to the posters.

It's really very strange. I took a project and began stripping out page after page...I got down to the final word (literally) and noticed the "+" sign after the paragraph style (which shouldn't have been there) and when I told ID to override everything and go with the para style, the Arabic tag disappeared. 

So I put the last section back in, cleared the para style...nope, back to Arabic. 

Some djinn out there hates me.

What I'm thinking now is that the issue might be somehow triggered by the incoming Word docs that I import. Some of them have styles, and some don't. It depends on how the author set them up. 

I'll experiment but, as I said in the beginning, the workaround isn't too onerous. 

[Abhishek Rao]

Hi @Don Mitchell,

Thanks for bringing this up. Could you share a short screen recording that captures the full workflow, starting from importing the Word file through to the EPUB export where the Arabic tag appears? That will help me understand the behavior more clearly. If possible, please also share a sample affected Word file, so I can test it with the team. Kindly confirm if you have tried this on the latest InDesign 20.5 version as well.

Looking forward to your update.

Abhishek

[Don Mitchell]

I'll do it when I can... the last troublesome import was 97,000 words, so re-doing it will be non-trivial. 

Also, I'll have to ask the publisher whether they're OK with my sharing (I'm the designer).

But I'll try to create something small and see if I can get it to misbehave.

And yes, I'm using 20.5 on Sequoia (but the problem long predates 20.5 and Sequoia).