greek letter β is shown as B in bibliography

Mojo Dodo

Hello,

I am experiencing some strange behaviour with my Zotero. My situation is that I have to cite a paper which title contains the greek letter β. Although the title is correctly saved with the unicode character in my zotero library, I get a "B" in the bibliography when citing it.

The strange thing is:
* it doesn't work (i.e. B appears instead of β) on my computer in Windows 7, with Firefox 36.0.4, Zotero 4.0.26.2 neighter with the MS-Word-Plugin 3.1.19, nor with the LibreOffice Plugin 3.5.9. So it's very likely not a matter of the office software.
* it works just fine on my Ubuntu Linux 14.04, with with Firefox 36.0.1, Zotero 4.0.26.2 with the LibreOffice-Plugin 3.5.9.
* it also works on my laptop with Windows 7, Firefox 36.0.4, Zotero 4.0.26.2. So it also doesn't seem to be a Microsoft Windows specific problem.

I am using the exact same reference on each computer with the Zotero-Sync function.

Has anybody an idea what could cause the β to change to a B in the bibliography? I really don't know where to start here.
Any help is greatly appreciated!

adamsmith

Sounds like an issue with the Word font on the computer where it fails. Zotero has no problem with greek letters (or any other utf-8 characters).

aurimas

Can you copy paste the text containing the beta character from Word into the forums here?

Mojo Dodo

@adamsmith: I know that Zotero can handle UTF-8 characters; it works well on 2 of my 3 systems. I have really no idea what causes this strange behaviour.
It is certainly not a font issue as the font I've used for testing was Calibri and Times New Roman. It it for sure a "real B" in my bibliography. Besides that I have the same problem in LibreOffice as well on that computer.

@aurimas
Paulsson, Marie, Petr Dejmek, and Ton Van Vliet. 1990. “Rheological Properties of Heat-Induced Β-Lactoglobulin Gels.” Journal of Dairy Science 73 (1): 45–53

adamsmith

how about if you "Create Bibliography from Item" --> Clipboard and paste it into the Word document? Same effect?

Is this with the Word add-on? If so, might be worthwhile to look at the field codes on a computer where it works and where it doesn't.

Any chance this could be a different item (e.g. stored in the document)? I really see no way how Zotero could change the character content of a citation.

aurimas

What you pasted above is actually a capital beta, not a capital Latin letter b. Do you expect it to be displayed different somehow? Not capitalized? Can you post a screenshot of what you expect it to look like? (upload it to Dropbox, Imgur, or something similar and link to it here)

aurimas

Btw, you can see both lower and upper case betas here http://en.wikipedia.org/wiki/Beta

adamsmith

nvm my questions then. Good call on aurimas's part. This is likely Zotero title casing at work. Not sure what to do about that, though.

aurimas

Still odd that this would behave differently on different computers. Is it possible that you have different (modified) versions of the same style installed on different computers? One that title-cases titles and one that does not? Otherwise, maybe you have set the export.bibliographyLocale hidden preference set to some non en-US locale? Or maybe one of your Zotero clients is not set to use en-US locale? Maybe your system locale settings differ.

Mojo Dodo

Yes, the problem also appeares if I choose the "Create Bibliography from Item" option and paste that into MS Word.
When I directly copy the title out of the Zotero "Title" field of that item in Firefox (where I can see the β) the β is correctly shown in MS Word.

Yes I'm using the Word Addon (in both cases Version 3.1.19). The field code of the bibliography is on both computers
ADDIN ZOTERO_BIBL {"custom":[]} CSL_BIBLIOGRAPHY
Interestingly, in both documents (i.e. also in the one where the "B" appeares" the title in the field code of the in-text citation (which contains, title, the whole abstract, ...) is shown correctly with the β.

It is certainly the same item, as I'm using the Zotero Sync funtion. In both in-text citation field codes it says:
http://zotero.org/users/NUMBER/items/X8NZXBW9


Another interesting thing I've just tried:
I transfered the "correct" file to the computer where the β was incorrectly shown as B and vice versa. When opening the file the entry of course shows up like it was saved, but when refreshing the bibliography with the button of the MS Word Plugin, the letter turn be correct (β) or wrong (B) respectively.
So it certainly does have to do something with the computer and it is not a pure MS Word issue (because on that computer I got the same problem with LibreOffice).
What I did not mention in my previous posts: the Windows 7 on which I am having trouble is actually a virtual machine (using Oracle VM VirtualBox on Ubuntu 14.04 as the host system) - I don't know if that could have anything to do with it.

Mojo Dodo

Ok, thank you very much for your constructive help! I see, I have really two experts here on my issue :D

First of all: it is really not a "B" what appeared to be incorrect, but a "Β". This looks alike in my MS Office (with Calibri and Time New Roman) and it also does in this forum (check the title I have posted via copy & paste in my second post).
So it is actually a capital beta what appeared to be a capital b!!

I think I have indeed different local settings on those Windows systems - the one showing the beta as a capital beta it set to en-US as far as I know, whereas the otherone should be de-DE (I'll check on that later).

But still, I don't quite get why the en-US setting makes the beta a capital one. I was using the "Chicago Manual of Style 16th edition" on both systems and there are no modifications as far as I know.

Mojo Dodo

A quick update: changing Zotero's language from en-US to de-DE causes to show a small beta in the bibliography.
en-US leads to a capital beta.
Is this a wanted behaviour or a bug?

What can I do about it? I need the language to be set to "en", because my citations need to be like "Johnsons and Smith" instead of "Johnson und Smith" for example.

adamsmith

as a workaround, just entering "ignore" (or really pretty much anything) into the language field of that item will work.

What's going on is that the Chicago Manual of style (and a number of other styles) require English titles to start every word with a capital letter and greek letters are currently, it appears, included in how Zotero does that. We'll look at adding an exemption, not sure if that raises other tricky issues.

Mojo Dodo

Okay, thank you! Setting the language field to "ignore" seems to work for me so far. I would have expected that Zotero would then use my system's setting which is in that case German, but it seems to stay English, except for the fact that the small beta is no longer capitalized. I hope that this will not raise any problems during my work now.

For anybody who doesn't know how to change the language field:
type: about:config into the firefox' URL field and then change the field extensions.zotero.export.bibliographyLocale

adamsmith

No that's not what I meant and it could cause problems. I'd change it back. I meant literally typing "ignore" into the field language of that item in zotero.

Mojo Dodo

Okay, there really was a misunderstanding. If I got it right, you meant changing the Zotero-item in the Zotero menu (where other information like title, author, year, etc.) is specified for each paper that I'm citing.

That works as well for me so far. Thank you!

adamsmith

correct, that's it.

fbennett

If we assume that title-casing should only be applied to English words in a title, it should be possible to tune the processor in such a way that words beginning with a non-English character are excluded from capitalization.

With a simple tweak to avoid capitalization of words that have a non-English character (i.e. [^a-zA-Z]) at first position, we get a test result like this, without any adjustment to the item content:

https://bitbucket.org/bdarcus/citeproc-test/src/tip/processor-tests/humans/textcase_NonEnglishChars.txt?at=default

Note that with this minimal fix, an English word that follows a hyphen join will be capitalized IFF the item is set with English as its Language (or if the item has no language setting and is being rendered into an English locale). Maybe that is acceptable - thoughts?

adamsmith

Thanks Frank. I'm very happy with the suggested fix. I tried to come up with a situation where we'd want non-Latin characters affected by title case and couldn't think of with anything real.
The one thing we should check on is that we don't (poorly) use text-case="title" on any terms en lieu of text-case="capitalize-first" -- but if we do it at all, it's not going to be common and it can be easily fixed in the styles. So I say go ahead.