Bottlenecks & translation systems analysis (Nigeria)
- Teams rapidly draft, creating a slight backlog of drafts needing team check.
- Team check is fairly quick and not thorough enough to catch all the problems that should be found BEFORE a back translation for consultant check. Drafts then pile up waiting for back translation.
- Back translation is painful. Often the quickest and easiest way to do it is for a translator to give it. Sometimes back translators make it easier by copying out (consciously or not) an English translation they are familiar with. So consultants may still require interaction with translators or another community member to ascertain what the text is saying and how it is saying that. (cf Mark Gaddis 2013: Expediting the Translation Process)
- Consultant time is at a premium so team-checked drafts pile up (discouraging the teams). But with the back translations being unreliable, consultant preparation is difficult and face-to-face time is spent fixing many smaller issues that should have been fixed at team check. This forces consultants to choose between a superficial check or thorough but slow progress (exacerbating the checking backlog).
Note: % are of translator/team time spent doing this translation work, given the observation that many teams appear to take about half the time to team check as draft, and double the time to consultant check than draft, and back translating often takes longer than drafting. They mean nothing in absolute terms, but are mainly to help us compare with the next page.
How to mitigate the backlog?
- Improve team checks: more thorough — checklist + coaching by intern or consultant?
- Improve back translation: interactive to clarify uncertainties — taken by intern?
- Essentially we must push the work back up the chain, rather than expecting too much correction/improvement to happen at the consultant check point. Teams may need more coaching on team checking and here interns could cut their teeth for a time, but stepping back as soon as possible to coach other teams (unlike ‘advisors’ who would stay long-term with a team). Intern pre-checking should include collecting helpful oral back translations.
Note: Parallel pre-checking may be more fruitful than trying to do a final check in parallel (cf Mark’s experiment with Nyankpa and Duya).
Note 2: What if there is no team? Or if the ‘team’ consists of one ‘rock-star’ translator and then others who are considered much less able? Teamwork dynamics may need work before team checking is fruitful.
Finally, I actually think that a ‘spiral’ development path is more likely to be helpful:
Implementing such a system would take some careful thought and a change in approach since much of our current methodology is based on the idea of a waterfall or ladder, leading one step at a time to the target and never returning to earlier stages:
A spiral sees us iterate through the stages much more quickly, so that the first prototype (draft) is not a candidate for publishing, but just a suggestion that gets gradually refined as we go forward. This concept comes from software development, where practitioners have gradually realised that the ladder approach usually leads to unsatisfactory projects being delivered late.
At each iteration we may need to learn to live with a certain degree of uncertainty or incompleteness, knowing that we will complete those areas later. These might be areas, such as:
- Writing the language (we may do a first draft orally)
- Key biblical terms (we may use ‘safe’ or placeholder biblical terms on early iterations)
- Discourse (!)
- Dialect choice and language development issues
It may sound a little dangerous, but I’m not sure it’s worse than current practice.