Archetype original language and translation policies

This comes from other topic, but I think it deserves its own discussion.

Talking about the translation of archetypes, and the requirement of using English as the original language.

I know, but nowhere in the specifications says that the original language must be English, which can certainly impose barriers to the international collaboration (i.e. local archetypes that could be promoted to the international community).

I can understand the implications, but here we face an equally unorthodox solution.

  • We can force non-English speaking modelers to use English as original language. Not being their mother tongue it can introduce errors to be corrected later, or it can require that a native speaker participates during the modelling activities. Afterwards, their mother tongue is added just as a translation.
  • We can allow that the original language is not English, but maybe ask that at least an English translation is always provided. It can be done by a native speaker/professional translator, not necessarily a clinician, once the modeling is finished. Then, it could be allowed that other translations are made, exceptionally, from the translated English.

In either case, we have forced to create an English translation, which is not the “original language of the modelers”. And of course, this applies to new archetypes. Changes or versioning of existing archetypes will always follow the first case.

I’m not defending any specific position here, just thinking out loud :grinning:

1 Like

The specs are very clear that there is no primacy of any language in archetypes, and none of the tech forces that, so this is very much just a (IMO sensible) CKM policy.

I can recall that just became quite difficult to work practically if the original language was not in English.

2 Likes

I am surprised to hear this myself. If some archetypes are authored originally in (say) Spanish and offered to CKM (and let’s assume they’re clinically high quality), then I can see that the international CKM group might say: well you guys need to also provide a reliable translation to English as well (@damoca 's second point above), since we are not expert Spanish speakers, and if you don’t, we are just going to ask you or someone else to do such a translation.

But that’s not the same as requiring the archetypes to be originally authored in English.

That’s not how this works in practice. We’ve had several examples of archetypes being originally authored in Norwegian, then translated into English. But for practical use in the international space, we then switch the languages around by a minor edit of the ADL.

That’s certainly technically easy - probably would be a useful function to add to the Archie library so that CKM could just provide a button for ‘make [XX] the original language’. @pieterbos , @sebastian.garde - thoughts?

1 Like

The CKM translation fuctionality only offers to translate from the original language. If that is Norwegian, most of you others would struggle. :grinning:
We’ve had the opportunity to choose to translate from another language instead, on our wish list for a long time. Or even better, choose to see multiple languages at the same time when translating, e.g. both English and Swedish.

2 Likes

David,

As you are no doubt well aware, there are many moving parts in clinical knowledge governance. The processes built into CKM are designed to support this. It is phenomenally complex and every time we solved one issue we discovered 10 more. Devising a process in which content publication was managed at the same time as technical versioning, plus add in the potential for simultaneous management of translation and terminology reviews… this is just the tip of the governance iceberg and it gets mighty convoluted real fast.

So the pragmatic decision is that the language of primacy in the international CKM is English, as is the UI etc. If we didn’t it would be like the Tower of Babel - multiple archetypes with multiple original languages and utter chaos. There would be even more complaints about how complex it is. Little do they know how much worse it could be.

The original language of the archetype is the one sent out for content peer review. It is also the one that is published. Other translations are intended to be relative to the original language and that’s why we are constantly advising people that translating the archetype prior to publication may require multiple re-translations as the content of the archetypes evolve through the publication process.

In the Norwegian CKM, the original language and language for review is set to Norwegian, for obvious reasons. So the capacity for multilingual CKMs is there, just like in the archetypes. The original language for a CKM is a setting.

If there was a Spanish CKM, the CKM tool can integrate a Spanish translation of the UI display and the archetypes could all be displayed in Spanish - a coherent and consistent tool for use by Spanish speakers, including archetypes with all variations of Spanish. If you ran a review it would be conducted in the ‘original language’ for the CKM and it would be the Spanish content version that would be published. The English translation of that same archetypes would be just a draft until a specific English translation review deemed it fit for publication in English. The same applies to any/all Spanish language variants.

For practical clinical governance purposes, we had to anchor one of the many moving parts. This is a significant one but it doesn’t prevent other CKMs from doing a similar thing in other languages, nor those archetypes being shared across multiple language CKMs.

Until we solve the issue of the multilingual federation of CKMs, which is still some way off (if ever), sharing archetypes between different language CKMs is absolutely possible, but it is the result of a deliberate strategy between Clinical Knowledge Administrators to collaborate with each other. This works well at present between the International CKM and the Norwegian CKM - @siljelb and her team are expert at managing the translations and their review of Norwegian translation in their CKM whilst the international community reviews the same one in English. The archetypes in the 2 CKMs are very closely aligned specifically because we choose to work in a way that prioritises convergence of the archetypes towards a coherent and shared library. This is the bit where people make active choices and the tooling is… well, just a tool.

Hope this helps provide context.

Best regards

Heather

4 Likes

@heather.leslie I completely agree with all your points. To be clear, I was not criticizing how it works now, I just wanted to clearly know the rules.

I think it is completely reasonable to ask that English is provided as the original language (or at least, as I said before, as a validated translation) in order to promote a local archetype to the international CKM. And probably there are some other requirements that we should consider and should be written somewhere. This is just another topic to add to the TO-DO list of archetype governance.

David

3 Likes

Not taken as a criticism. It is information not broadly known or understood to those who are not acting in a CKA role, so just trying to provide the context.
All good from my POV.

3 Likes

Yes please! :smiley:

I’m not sure having this just in the Archie library will help soon, as it does have a ADL 1.4 parser, but only an ADL 2 serializer. Replacing the original language should be fairly straightforward to implement with or in any library though - it is just moving some metadata/description fields around. The hardest part is probably deciding which fields go where in the description. If a description is available on how this should be done, implementing the functionality to swap is likely to be simple.
I cannot comment on any possible work related to the CKM, of course, and I do not know if you need any user input other than the button to swap original language with another language.

2 Likes