Objections & Caveats

A few people (mostly consultant types) have raised the legitimate concern that Transcelerator pushes teams in the direction of relying on a "canned" set of comprehension checking questions, rather than doing the hard but valuable work of crafting questions that are likely to uncover confusion that might be specific to a particular culture or translation. To quote one such critic:

"...based on many years of translation and consultant experience, pre-prepared lists of questions make me very nervous. Good test questions are highly language specific. I have seen questions that translators have prepared using the old “Dallas” list and the result has not been good. I am convinced that the value of the process of preparing good questions that fit the Host Language context is higher than the time apparently saved by generating questions quickly."

We welcome such criticism and will continue to monitor the level of such “negative” feedback. One of the on-going challenges in language software development is that in theory we’re not trying to tell our users how best to do their jobs, but in practice it’s impossible to avoid the reality that if our tools make one path easier and another path harder (or in many cases the alternate path is merely ignored), then in one sense we’re sort of telling uses to go down path A, not path B. When the “better” way is not only inherently harder in and of itself but also harder to make easier using software, it’s easy to see how the loudest voices could come from the camp asking to make the easier way even easier still. This unfortunate bias is nearly impossible to eliminate.

Potential users of Transcelerator who are concerned about this should consider the following:

1. Transcelerator does allow translators to add to, delete, and modify the existing source questions as needed. True, the path of least resistance is to translate the canned questions as they are, but we do recognize the inherent weakness in that approach and have made provision for customization. It’s even theoretically possible to derive some benefit from Transcelerator when creating a completely custom list of questions, but this is a somewhat dubious use case, and it will be surprising if anyone ever uses it that way. We are considering possible future enhancements to help users know which questions have been deemed most useful by other teams in similar projects.
2. The target users for Transcelerator are teams which have been hitherto incapable of writing useful questions from scratch. Teams that are successfully writing good checking questions from scratch and getting good results should not switch to Transcelerator merely to make their lives easier.
3. While some customization is almost certainly going to improve the checking process, it's not clear whether the maximum quality can be achieved by most teams writing questions completely from scratch. Obviously some of the built-in questions in Transcelerator could benefit from a careful review and re-write, but there a lot of pretty good solid questions that seem pretty likely to be useful for checking the comprehensibility of a translation in pretty much any language. Maybe the ideal approach would be for the team to read through the list of canned questions first, and then – before translating any of them and without referring back to the list – work through the section they’re trying to check and write out some questions from scratch. Then use this list to supplement the canned list in Transcelerator.
4. There is still very much a need for savvy consultants, translation facilitators, and translation team members to be alert for comprehension issues that are not adequately addressed by the canned questions in Transcelerator.