What Kind of Quality? Differentiating Three Categories of Computer-Assisted Translation

In the current climate of pursuing innovative technical approaches to Bible translation, the goals are often phrased in terms of “speed” and “accuracy”. However, “accuracy” is a multi-faceted, ambiguous term, and there are other metrics that are more important than speed. This paper begins with the assumption that new Bible translation projects are undertaken because there are some “barriers” that prevent existing translations from being used by the target audience. Whether starting a traditional or computer-assisted translation, it is crucial to understand the goals of the target project regarding the various features that could be considered “barriers to cross” or “features to retain” relative to existing translations. This paper differentiates “barriers” that are easy for computers to cross (e.g., orthography, writing system, morphology) versus those that are hard (e.g., world view differences, reason-result order, level of vocabulary), on the assumption that if the hard barriers have already been crossed in a reference translation, then presumably retaining those features and crossing other easier barriers could quickly multiply the audience for that translation, leveraging the hard-won victories achieved. It groups computer-assisted approaches into those that are specially suited to retain the “features to be retained” versus those that provide effective help when those desired features don’t exist in the available reference translations. In addition, it discusses those that require some material already translated in the target language versus those that do not.