Using trio writing and pair writing to improve bilingual content

Using trio writing and pair writing to improve bilingual content

I recently worked with Natural Resources Wales to help them improve their content, processes and content toolkit. It was a joint effort with Crocstar, Basis and dxw.

During the project we identified opportunities to rethink how bilingual content was being planned and created. I worked with some of the team to experiment with both pair writing and trio writing. In this article I’ll share the reasons a different approach to bilingual content was needed, how we did pair writing and trio writing, who was involved and the benefits and outcomes of this new way of collaborating.

For the purpose of this article I’ll be focusing on content that was required in both English and Welsh. 

If you’re looking for a summary of pair writing, I recently wrote about that including what it is, how to do it, benefits and risks in my article: Pair writing to create user-centered content

A traditional approach to translation, and its problems

It’s common practice for translation to be a stage towards the end, or sometimes the final stage before publishing, in a content workflow. While this may work well in some instances, in my experience and through my own observations, this can cause problems for those working on the content, the content itself, and for the users.

Though a little simplified for the purpose of this article, here is how translation fits into many content workflows:

An example content workflow with six linear stages of outline and research, write draft, review and edit, approve, translate and publish.i

Regardless of the precise number and names of workflow stages, here are some problems and challenges when content is translated late in the process, or even at the very last minute.

Both languages are not treated equally

One language gets a lot more care and attention than the other language. Let’s call them the first and second language for ease, though it isn’t intended to suggest one is inferior to the other.

The first language, English, is written based on user needs and research insights. Content design good practices are applied to make sure the content is written in clear language, is well structured and is inclusive and accessible. Lots of time is spent on getting the English version right. It is tested with users, iterated based on feedback and eventually approved. 

That researched, designed and approved content is then sent to be translated into the second language, Welsh.

No time to iterate and improve the second language

At this point, there is no time for further amends or to take the Welsh version through the same process. It has to be translated as is with no changes based on differences in meaning, nuances in the lexicon or word length and tone.

The expectation is a like-for-like translation of the English version and this doesn’t mean the Welsh version will be error free or retain all its meaning and context. The English version is deemed the best it can be at this moment in time but the Welsh version can only be as good as possible.

A substandard experience for users

This approach doesn’t give Welsh language users the same priority as English language users. It is assumed the content still meets user needs even though things may quite literally have been lost in translation.

Language and its cultural meanings can be complex and sometimes there are multiple ways of saying things but the message conveyed can be different depending on which way you choose. By translation approved English content, rather than content designing the Welsh version, the context of the translated language isn’t considered. Though the user may still be able to achieve their task, there is a chance they won’t be able to or they will have to work harder to do so. We put a burden on the user because of our processes. That’s not fair.

Places a burden on the translator

This way of working also puts a burden on the translator. They are sent ‘approved’ content that they have to translate. Often they get very little time to do so. Translators are busy people and get stuff added to their to-do list without consideration for their priorities and workload. It puts them under unnecessary pressure and if they do spot anything with the content, there is little or no process for sharing that feedback and actually being able to improve the content accordingly.

This can cause friction between roles and teams when really, the relationship with a translator is one that should be nurtured, respected and valued. They should be seen as partners and are content experts in their own rights too.

Identifying a need to improve bilingual content processes

At Natural Resources Wales the translators were very much valued. A lot of the content being translated was high quality and working for users and the organisation. But during the recent project we were working with quite a technical and specific subject matter. 

It was March 2022 and we had worked hard on the English language content that was focused on legal processes, funding, permitting and licencing. It had to be factually correct and up to date.

There was collaboration with subject experts and several rounds of refinement. Once the content was deemed to be ‘approved’ or close to, it was translated into Welsh. Then, both versions were tested with users in their preferred language. We had an amazing user researcher, Gwenno Edwards, who I worked with to create the discussion guide for the research and testing. 

Gwenno played back the findings of the testing and there were plenty of good insights and things to iterate on before we published the content. But with the Welsh version there was a different theme and layer to the feedback in that some of the content wasn’t making complete sense and the Welsh language users had more questions about the meaning and intentions of the content. 

It was clear that we needed a different approach to creating and testing bilingual content. Fairly quickly we decided to involve the translator earlier in the process. This was happening alongside other changes to the workflow and so the new process we defined and would test in practice through doing, looked like this:

An example of an improved workflow with five stages of outline and research, pair write, fact check, approve and publish.

That pair writing stage (which started as trio writing, more on that shortly) is where the content creation took place as so by comparing the two workflow diagrams in this article, we can see how much earlier the translator became involved. In the first instance though we actually tried trio writing. 

Using trio writing to improve bilingual content

I had done pair writing before but never considered adding another person, not just because of the very definition of ‘pair’ but also because when too many people are involved the content can suffer. In this case we were willing to try new things and had the space, time and support to do so. It was quickly decided we would write together and the initial trio included myself as the content designer and then the user researcher and a translator.

We each had a specific role in the process based on our skills, knowledge and expertise:

  • Content designer - clear language, structured content, user needs, business needs, subject expertise (after working with subject expert) and perspectives from internal stakeholders

  • User researcher - insights from content testing, verbatim quotes from user, deep understanding of users and their needs

  • Translator - expert in both languages, writing and in this case had personal experience with the topic of the content too

This was a case of flexing an established process to suit our needs on the project and that was always the intention with the project because of the bilingual element. But for Manon, the translator, this was the first time she had ever been involved with content in this way.

For that reason we spent more time upfront updating Manon on the work done to date, the outcomes of the research and the reasons we wanted to try trio writing.

We were all working remotely so on a Zoom call I talked about the content we wanted to create based on the template I use for creating user-focused content. I outlined the user needs, purpose of the content, goal for the content, necessary call to actions and all of the work done to date. Gwenno shared the insights from the research and examples of where the translated content hadn’t quite hit the mark with Welsh speaking users. Manon had the chance to ask questions before starting any content which was very valuable for all involved.

In this case it meant we had existing content to work from. It didn’t make sense to start from scratch when the content had broadly tested well in both languages and a lot of time had been spent to get to this point. For future content we actually wrote together in both languages from scratch.

I shared my screen and on one side I had the English version of the content and on the other side I had a blank Google doc. Manon started to write the Welsh version and asked questions as she progressed. Gwenno is also a fluent Welsh speaker so could write too. I know some Welsh but am by no means fluent so my role was to answer questions, fill in the blanks and unexpectedly, to improve the English content as this process gave us insights to improve both versions of the content.

The content we were writing was a few hundred words so it didn’t take long before we had a newly written Welsh version and an improved English version. The trio writing process was productive, surprising and successful. 

Considerations and examples when translating content

There were some translation challenges with specific parts of the content and language:

  • Mutations
  • Abbreviations
  • The name of things

This is where the nuance of language became so apparent. Meaning was lost or things didn’t read or sound as natural in Welsh. While the name of things was fixed (such as what a certain grant and funding stream was called) we could rethink how we framed that name and also if we could say it in a more meaningful way.

In the context of the content we were trio writing, here are some examples of the questions, details and variations we experienced and had to consider:

  • ‘Verified’ didn’t really translate into Welsh but the process being talked about was a verification check and people becoming verified. We did determine this was internal language and so we opted for ‘checked’ in the Welsh version. We then decided to change it to ‘checked’ in the English version too as this is more clear than saying ‘verified.’ Context was important here as Manon added that depending on the situation it may be better to say ‘confirmed’ rather than ‘checked.’ Again, nuances that impact on meaning.
  • The English content included ‘You will need the following’ and Manon asked why it couldn’t say ‘You will need’ and it absolutely could and should have been that. So by collaborating on the Welsh content and with fresh eyes, we actually improved the English version too so both languages benefited and that was something I never expected or contemplated before we started trio writing. For me, that alone sells the benefit of working this way.
  • We had ‘If that’s not you’ in the English version and though it added a nice tone, when discussing the Welsh content we agreed it wasn’t really needed so removed it entirely. If we had translated that it would have been 6 longer words in Welsh.
  • The English content talked about ‘risks’ and Manon asked what sort of risks (business, environmental, other) so we removed it altogether as there was enough context in the content without it and it also made it a shorter sentence. Plus we were retesting the content afterwards too.
  • A phrase that was more specific and technical, and hard to alter, was ‘Management control’ but in Welsh they would be the same word. In that case we didn’t do a like for like translation but rather we wrote a version in Welsh that made sense for that language.’
  • A similar example was ‘eligibility’ in the English content which was proving tricky to translate so we added ‘criteria’ to it as that made more sense in the Welsh version.

We had to decide where it was necessary for the versions to be the same and where differences were fine too. As Manon said during the session, ‘it is more important for people to read more naturally in their own language so the versions don’t need to be like for like.’ The benefit of this way of working is that Manon had the context to make informed decisions about which version in Welsh made most sense based on the meaning in English. Manon was also able to explain why some things can't be translated literally from one language to another so context is shared in both directions.

Trio writing also makes the process better for the translator. Manon said:

‘Trio writing gives us an opportunity to gain an insight into the content design process, and an in-depth understanding of the content itself. Working together during the initial phase enables us to give both languages the same amount of thought, which hopefully results in better, more user friendly content in both languages.’

- Manon Williams, Translator

Challenges with trio writing and how to mitigate them

Trio writing does take time, especially at the start when more time is needed to talk through the process, reasons for doing it and the content itself. But it could save time long term and even if it doesn’t the users are benefiting from it which means the organisation will benefit too.

The biggest challenge we found was scaling the process. It was an ideal way of working but certainly not the standard way of working. This was a tough realisation as we all had a taste of how things could be improved but we also knew that it wouldn’t be easy or quick to apply this process to all content.

There were three ways we mitigated this:

  • Pair writing - this was a small mitigation but we reduced the number of people involved as did pair writing between a content designer and a translator. Still not a solution to scale but as the content designer writing with the same translator each time, we noticed it was taking less time to create the content the more we did pair writing
  • We flexed the process again so I would talk through English language content with Manon and there was the same scene setting and time for questions but then Manon would write the Welsh content in her own time. This was a hybrid between the old and new processes but it did mean the Welsh language content was created with more context
  • Prioritising what content was written using pair writing. We just had to accept not all content could be given this treatment and so we prioritised it and for some content the old process was still followed.

A reminder that the content translated at the end of the workflow wasn’t bad or wrong but when we did trio or pair writing the Welsh version was more user-centred. Sometimes you have to work within fixed boundaries even if you want to approach things differently. This applies to other areas too such as not being able to test every piece of content that is published.

Rolling out trio writing and pair writing is something that can happen long term. We also introduced pair writing at other stages of the workflow too such as between a content designer and subject expert.

Outcomes of trio writing and pair writing on bilingual content

There were positive outcomes for the people involved and for the content itself, including:

  • Welsh content that is as user centred as the English by considering user needs and acceptance criteria during creation
  • English content was changed and improved through feedback and suggestions from the translator
  • A style guide started to be created for Welsh language content explaining why some terms in Welsh have to be used
  • A more collaborative process that elevates the importance of translation
  • Opportunity to share with others around an organisation the process and value of content design
'User researchers don’t always need to be involved in the writing process, but this was a great opportunity to work as a trio. I could share insights and make suggestions live in the writing session and we could work iteratively. With the added mental load of writing in two languages, it would be difficult to rely on the content designers remembering everything from my analysis briefing notes. The readability of the content improved significantly after the trio writing, and the content tested much better after the translation tweaks.'

- Gwenno Edwards, User Researcher

Trialling new ways of working is always beneficial. Sometimes processes take time to be refined but the effort is worthwhile because everyone involved stands to gain from improved collaboration - the team involved with the content, the organisation they work for and the audience they create the content for. Content in all languages should be useful and usable and pair writing and trio writing can help achieve that.

More about improving bilingual content

There’s some incredible work happening in the Welsh public sector to improve services and bilingual content. The Centre for Digital Public Services are exploring new ways of working and sharing their experiences. They tried trio writing and wrote about it in their article Producing bilingual content through trio writing.

I talked about trio writing on the Words in Progress podcast that I co-host with Christine Cawthorne from Crocstar. It was working with Christine that led us to experiment with trio writing in the first place. Listen to that special episode about trio writing.

The Canadian Digital Service presented for The Centre for Digital Public Service's community of practice. Their insightful talk is on YouTube and was about building bilingual services.

I also wrote about ...