Multi-language support

mkoc · October 24, 2022, 7:56am

Hello, we are sending data to crossref via xml schema. We have 2 language options in our articles, Turkish and English. We want to display in crossref according to the first language of the article. We send the first language of the article via original_title_language. But the notation in crossref is referencing the tag. How can we include the <original_language_title> tag in the display? Or how can we do multi-language support? Thank you.

ifarley · October 24, 2022, 9:33pm

Hi @mkoc ,

Thanks for your message, and welcome to the community forum.

DOIs are really citation identifiers, so we advise that members register the journal-level metadata as well as your journal-article-level metadata as you believe those works will be cited. Do you believe that the content is more likely to be cited in English or Turkish? If the former, register the journal title in English. If the latter, register the journal title in Turkish.

We recommend registering the metadata in the language that matches the language of the content itself. For illustration, if the article (or, content) was written and published in Turkish, then the article would most likely be cited in Turkish, so we’d recommend that the metadata registered with us also be in Turkish.

Keep in mind that <original_language_title> is meant for use with translations only. It’s not a way to insert extra metadata in another language. If you want to submit multiple titles for a given article, to have the metadata in two languages, then you can register two <titles> tags.

Please let me know if you have any additional questions,
Isaac

mkoc · October 25, 2022, 5:45am

thank you very much for answer. I needed this so much.

fgnievinski · December 23, 2022, 4:49am

Dear @ifarley,

I was reading the following example:

Could you confirm which option below would be correct when the multilingual content is not a translation, please?

Option A:

<titles>
<title>When your best metadata isn't good enough: working with an imperfect specification</title>
<title language="fr">Quand vos meilleures métadonnées ne suffisent pas: travailler avec une spécification imparfaite</title>
</titles>

Option B:

<titles>
<title>When your best metadata isn't good enough: working with an imperfect specification</title>
</titles>
<titles>
<title language="fr">Quand vos meilleures métadonnées ne suffisent pas: travailler avec une spécification imparfaite</title>
</titles>

Option C:

<titles>
<title>When your best metadata isn't good enough: working with an imperfect specification</title>
</titles>
<titles language="fr">
<title>Quand vos meilleures métadonnées ne suffisent pas: travailler avec une spécification imparfaite</title>
</titles>

I tried reading the schema but couldn’t confirm the issue above:
https://data.crossref.org/reports/help/schema_doc/5.3.1/crossref5_3_1_xsd.html#titles

Thank you,
-Felipe.

ifarley · December 30, 2022, 10:06pm

Hi @fgnievinski ,

Yes. This is the correct format for multilingual content that is not [Note: I was wrong on this recommendation; our documentation, at the time of the original post, was not as clear as it could be and I misinterpreted the recommendation - <original_language_title> is only for content that is a translation.] a translation:

<titles>
<title>When your best metadata isn't good enough: working with an imperfect specification</title>
<original_language_title language="fr">Quand vos meilleures métadonnées ne suffisent pas: travailler avec une spécification imparfaite</original_language_title>
</titles>

Warm regards,
Isaac

fgnievinski · December 31, 2022, 8:15pm

Hi Isaac, now I’m confused:

initially we had “Keep in mind that <original_language_title> is meant for use with translations only.”
later we had “This is the correct format for multilingual content that is not a translation … <original_language_title>”

The two cases seem contradictory, unless I’m misinterpreting them?

Kind regards
Felipe

ifarley · January 5, 2023, 2:00pm

Hi @fgnievinski ,

My apologies for the added confusion. I think the documentation could be clearer, so I’m going to update it.

I was wrong in my previous response. Sorry about that. <original_language_title> is only for content that is a translation. Like I said above, that could be clearer in the documentation too, so I’ll get that updated. It will probably be next week before that change is live on https://www.crossref.org/documentation/schema-library/markup-guide-metadata-segments/multi-language/.

The annotation in the schema for <original_language_title> is pretty legible so I don’t think we need an update there:

<xsd:documentation> The title of an entity in its original language if the registration is for a translation of a work. When providing the original language of a title, you should set the language attribute.</xsd:documentation>

In your examples above, option C is the one that works for multi-lingual non-translated content, and the schema.

My best,
Isaac

fgnievinski · January 11, 2023, 2:51am

Thanks for the clarification, Isaac – now I can follow it. Just to recap:

1st) for single-language translated content, do:

<titles>
<title>When your best metadata isn't good enough: working with an imperfect specification</title>
<original_language_title language="fr">Quand vos meilleures métadonnées ne suffisent pas: travailler avec une spécification imparfaite</original_language_title>
</titles>

2nd) for multi-language content, do:

<titles>
<title>When your best metadata isn't good enough: working with an imperfect specification</title>
</titles>
<titles language="fr">
<title>Quand vos meilleures métadonnées ne suffisent pas: travailler avec une spécification imparfaite</title>
</titles>

How about a third common case: multi-language metadata for a mostly single-language non-translated content, where only the title (and abstract) are provided in a secondary language? For example, the full text is in Portuguese (primary language) but the title is also provided in English, for broader dissemination. Could the metadata for this third case be deposited as in the second case above? My intention is to eventually expose multi-language metadata already setup in PKP/OJS.

Thank you!
-Felipe.

ifarley · January 18, 2023, 10:03pm

Hi Felipe,

Adding the article titles and the abstracts in different languages should be fine. Yes, this additional example can follow the 2nd option.

My best,
Isaac

fgnievinski · January 20, 2023, 2:20pm

Thanks once again for the confirmation, Isaac.

I hope you don’t mind a follow-up question: what would be the best level for setting a non-English primary language attribute – in the journal_metadata, journal_article, or titles tag?

Journal level:

<journal>
<journal_metadata language="pt">
<full_title>A Revista</full_title>
</journal_metadata>
<journal_article publication_type="full_text">
<titles>
<title>Quando mesmo os seus melhores metadados não são o suficiente: trabalhando com uma especificação imperfeita</title>
</titles>
<titles language="en">
<title>When your best metadata isn't good enough: working with an imperfect specification</title>
</titles>
(...)
</journal>

Article level:

<journal>
<journal_metadata>
<full_title>A Revista</full_title>
</journal_metadata>
<journal_article publication_type="full_text" language="pt">
<titles>
<title>Quando mesmo os seus melhores metadados não são o suficiente: trabalhando com uma especificação imperfeita</title>
</titles>
<titles language="en">
<title>When your best metadata isn't good enough: working with an imperfect specification</title>
</titles>
(...)
</journal>

Titles level:

<journal>
<journal_metadata>
<full_title>A Revista</full_title>
</journal_metadata>
<journal_article publication_type="full_text">
<titles language="pt">
<title>Quando mesmo os seus melhores metadados não são o suficiente: trabalhando com uma especificação imperfeita</title>
</titles>
<titles language="en">
<title>When your best metadata isn't good enough: working with an imperfect specification</title>
</titles>
(...)
</journal>

I think I remember reading that journal_metadata would be the recommended level, but I’m thinking the journal_article might be more appropriate.

Thanks,
-Felipe.

ifarley · January 30, 2023, 10:44pm

Hi @fgnievinski ,

In your use case here, I’d recommend adding it to both the journal-level and journal-article-level metadata. Why not both?! I think it makes sense for it to be included at both levels.

-Isaac

samogot · June 3, 2023, 1:08pm

I tried to use the mentioned format for multilingual metadata for conference proceedings, but it doesn’t work.

What I tried:

<conference>
	<!-- ... -->
	<proceedings_metadata language="uk">
		<proceedings_title>КОНСТРУКТИВНА РЕФЛЕКСІЯ КОНФРОНТАЦІЇ І КООПЕРАЦІЇ: ПСИХОЛОГІЧНІ РИЗИКИ І РЕСУРСИ ВІЙНИ. Матеріали міжнародної міжгалузевої конференції</proceedings_title>
		<!-- ... -->
	</proceedings_metadata>
	<conference_paper publication_type="full_text" language="uk">
		<titles><title>МIЖНАРОДНИЙ ТУРИЗМ ТА ПУБЛIЧНА ДИПЛОМАТIЯ</title></titles>
		<titles language="en"><title>INTERNATIONAL TOURISM AND PUBLIC DIPLOMACY</title></titles>
		<!-- ... -->
	</conference_paper>
</conference>

The error I got:

Error: cvc-complex-type.2.4.a: Invalid content was found starting with element 'titles'. One of '{"http://www.ncbi.nlm.nih.gov/JATS1":abstract, "http://www.crossref.org/schema/5.3.0":publication_date, "http://www.crossref.org/schema/5.3.0":acceptance_date, "http://www.crossref.org/schema/5.3.0":pages, "http://www.crossref.org/schema/5.3.0":publisher_item, "http://www.crossref.org/schema/5.3.0":crossmark, "http://www.crossref.org/fundref.xsd":program, "http://www.crossref.org/AccessIndicators.xsd":program, "http://www.crossref.org/clinicaltrials.xsd":program, "http://www.crossref.org/relations.xsd":program, "http://www.crossref.org/schema/5.3.0":archive_locations, "http://www.crossref.org/schema/5.3.0":scn_policies, "http://www.crossref.org/schema/5.3.0":doi_data}' is expected.
Error: cvc-complex-type.3.2.2: Attribute 'language' is not allowed to appear in element 'titles'.

Is it not supported for conference proceedings, or I am doing something wrong? Schema indeed doesn’t mention language attribute on neither titles nor title elements, but it is the same for titles in journal_article as well.

I would also like to enter english version of the title of the proceedings book itself if it would be supported.

ifarley · June 20, 2023, 9:59pm

Hi @samogot ,

Thanks for posting this example, and apologies for the delayed response. Do you happen to have the submission ID that is producing that error message? If so, can you share here so I can take a closer look.

-Isaac

fgnievinski · July 10, 2023, 7:14am

Dear @ifarley, could you confirm if the citation and citation_list elements support the language attribute, please? Or do they only inherit the language attribute from the parent journal_article element?
Thanks,
-FGN.

fgnievinski · July 10, 2023, 1:56pm

Hi again Isaac. I’m afraid my constructed XML snippets might have mislead @samogot. That’s because the language attribute remains undefined for titles or title elements – it has no <xsd:attributeGroup ref="language.atts"/>, which is only defined for element original_language_title:
Schema documentation for crossref5.3.1.xsd

This affects mainly the “third common case” discussed above (multilingual metadata of monolingual content), e.g.:

<journal>
<journal_metadata language="pt">
<full_title>A Revista</full_title>
</journal_metadata>
(...)
<journal_article language="pt">
<titles language="pt">
<title>Quando mesmo os seus melhores metadados não são o suficiente: trabalhando com uma especificação imperfeita</title>
</titles>
<titles language="en">
<title>When your best metadata isn't good enough: working with an imperfect specification</title>
</titles>
(...)
</journal>

In the current schema version, the secondary title’s language would have to be left unspecified. Could CrossRef consider supporting the language attribute for titles in the future, please? Otherwise, the value of providing multiple versions of titles seems more limited.

Thanks,
-Felipe.