Preferential changes are vexing gremlins, often confusing linguistic teams and disrupting the whole language QA process. On a typical language quality scorecard, there is not a category for “preferential” changes.
This is because a preference is a greater liking for one alternative over any other. If the change simply seems like a good idea — a certain fondness for one phrase over another or a quick judgment call — that’s because it is. Reviewers will not likely admit (or even be aware) that their changes are simply something they like better than what is there. So, these changes show up in other buckets, most often under style, terminology, accuracy, or language quality.
Many reviewers will still mark them even if they know they reflect personal preference. (“It’s hard to stop,” they will say. “I can’t help it!”) And so these gremlins pop up over and over again. When quality results come in, and after arbitrations have occurred, there they are. We can’t blame individuals for their preferences, but there is no room for them in an efficient localization quality process.
- They aren’t errors. Nothing is technically wrong. The reviewer may change the writing style, offer a synonym that is not in the termbase, and change word order … none of which improves the piece in any significant way.
- They blur the definition of quality. When errors are clearly defined and scoring methodologies are in place, quality really can be black and white. Preferential changes put that entire structure into question.
- They can be emotional. People stand by the changes they make and get fired up when someone challenges them. This causes arguments which in turn cause frustration and bad feelings. Bad feelings demotivate people. You get the picture.
- They slow down the process. Often the original translator will disagree that the change needs to be made and maybe even call it out as preferential. Then someone has to arbitrate between the two parties, and someone has to make a final call. Imagine doing this with dozens of changes, over dozens of languages.
- They can sometimes introduce errors. I once was on a project where a reviewer deleted sections and rewrote others completely. What he did was a major edit and the whole project came to a halt while the team had to back up and undo many of those changes.
- They can increase costs. We are sometimes asked to implement these preferences at no charge, for which we incur costs. If our client truly wants the preferences implemented, project costs may increase. (Of course, we fix any actual errors at zero charge to our customers.)
Can preferential changes ever be eliminated? We are talking about people after all, who take their work seriously and have pride in their qualifications and experience (as they should).
This is especially difficult for non-linguists to overcome … and somewhat unfair to expect in that case. A past blog discusses how to give those non-linguist, in-country reviewers guidelines. Training any reviewer on what these changes are and what they do is an important part of controlling and smoothing the review process.
Do you guide your in-country reviewers against logging preferential errors? Do you educate them as to what they are and how to avoid marking them? Share with us in the comments!