Multilingual Metadata in Crossref

[ctgraham]

Describe the bug
PKP Sprint Copenhagen. Metadata exported to Crossref is currently mono-lingual (primary locale). Journals are collecting multilingual metadata for articles. Crossref can index this either as supplementary metadata (when there is multilingual metadata for a monolingual article), or as independent DOIs (for articles which have been translated). This facilitates discovery in the user’s language.

To Reproduce
Steps to reproduce the behavior:

1. Add translations of the title and abstract to a submission
2. Publish the submission
3. Export the Crossref XML for the submission
4. Only the primary locale will be represented

What application are you using?
OJS 3.3, 3.4

Additional information
Shayn Smulyan from Crossref writes:

the format you used would be fine for submitting multiple abstracts in multiple languages.

<jats:abstract xmlns:jats="http://www.ncbi.nlm.nih.gov/JATS1" xml:lang="en">
<jats:p>Abstract in English</jats:p>
</jats:abstract>
<jats:abstract xmlns:jats="http://www.ncbi.nlm.nih.gov/JATS1" xml:lang="es">
<jats:p>Abstract in Spanish</jats:p>
</jats:abstract>
<jats:abstract xmlns:jats="http://www.ncbi.nlm.nih.gov/JATS1" xml:lang="cs">
<jats:p>Abstract in Czech</jats:p>
</jats:abstract>

But, as far as the titles go, there are two options.

If the article you're registering is a translation of another article, you should use the <original_language_title> tag. In context that looks like this:

<journal_article language="en">
<titles>
<title>New Article</title>
<original_language_title language="es">Nuevo Artículo</original_langauge_title>
</titles>
.....

In this case, as well, we'd strongly recommend adding an isTranslationOf intra-work relationship to connecting to the DOI of the original.

If the article is not a translation, but you simply want the metadata expressed in multiple languages, you should not use <original_language_title> but instead add multiple <titles> sections, one for each language. The language attribute within the <journal_article> tag should reflect the language of the full text content.

So that would look like like this

<journal_article language="en">
<titles>
<title>New Article</title>
</titles>
<titles>
<title>Nuevo Artículo</title>
<titles>

---

Sprint contributors to this work include:

• Clinton Graham, University of Pittsburgh
• Radek Gomola, Masaryk University Press Czech Republic
• Susan Collins, Crossref
• Emma Uhl, Public Knowledge Project
• Ramana Fragola, National Library of Sweden
• Jyrki Heinonen, Federation of Finnish Learned Societies
• Esmee Klumpenaar, University of Groningen

---

PRs:
OJS

• Add support for multilingual titles and abstracts crossref-ojs#38
• Add support for multilingual titles and abstracts crossref-ojs#39 (stable-3_4_0)
• pkp/pkp-lib#9126 crossref submodule update ##bozana/9126-3_4_0## ojs#3960 (only submodule update stable-3_4_0)
• pkp/pkp-lib#9126: backport for multilingual crossref metadata ojs#3964 (stable-3_3_0)

OPS

• Add support for multilingual abstracts crossref-ops#32
• Add support for multilingual abstracts crossref-ops#31 (stable-3_4_0)
• pkp/pkp-lib#9126 crossref submodule update ##bozana/9126-3_4_0## ops#545 (only submodule update stable-3_4_0)

[asmecher]

@bozana, could you have a look at this when you're available? Thanks!

[bozana]

Just to double check: I would port this change to stable-3_4_0 but also to crossref-ops repository. Are there any objections or doubts, @ctgraham and @ajnyga?

[ctgraham]

I don't see a reason this wouldn't be appropriate to OPS as well. In the Sprint there was also interest in possible backporting to 3.3. Should I write something up for 3.3?

[bozana]

@ctgraham, great! Any contribution is very welcome! 🙏

[fgnievinski]

this is huge, many thanks for making it happen! it has the potential to unleash a ton of multilingual metadata that were never really exposed before (because of either #7272 or #699).

about multilingual metadata of monolingual content, last time I've checked with CrossRef, they've recommended "adding it [the language attribute] to both the journal-level and journal-article-level metadata", e.g.:

<journal_article language="en">
<titles language="en">
<title>New Article</title>
</titles>
<titles language="es">
<title>Nuevo Artículo</title>
<titles>

https://community.crossref.org/t/multi-language-support/3054/12

could you confirm if the language attribute is always set, even in the case of a single language?
pkp/crossref-ojs#21

maybe it's also a good time to reconsider the implications of #7569, to avoid problems downstream during deposit and indexing. as an editor/admin, I'd appreciate the option to follow CrossRef's recommended practices more strictly (i.e., adopting a language-agnostic journal title), without imposing the choice.

here are a couple of other related issues:

• Fix submission language/languages metadata #5000
• Multilingual articles (not just multilingual metadata) could receive multiple DOIs #7335
• Make reference list multilingual-capable #5303

[ctgraham]

Hi, @fgnievinski . In pkp/crossref-ojs#38, the journal_article language attribute is always set to the primary submission language, even in the case of a single language.
https://github.com/pkp/crossref-ojs/pull/38/files#diff-6ac12250187bc0d79809f02163656bd00203f38c8dc684057b883317800cc5c8R111

Note that Crossref does not (as of schema 5.3.1) support the language attribute on the titles tag.
https://data.crossref.org/reports/help/schema_doc/5.3.1/index.html

[bozana]

Regarding OPS:
Crossref schemas (4.4.0 (used in stable-3_3_0) and 5.3.1 (used in stable-3_4_0 )) only allow one titles element for posted_content.
In OPS stable-3_3_0 no abstract element is used.
Thus, only abstract element in stable-3_4_0 (and main) can be adapted to the changes here.