The Right Tool for the Job: Relating Diverse Translation Technologies to Diverse Project Settings

Much excitement has gathered around Artificial Intelligence (AI) and Natural Language Processing (NLP) approaches to assist Bible translation. However, when evaluating new approaches, it is essential to place them in context with approaches that already exist, recognizing the different benefits and costs that each involves. Similarly, the settings of the translation projects that may benefit from assistance are not all the same. This paper surveys the landscape and situates each approach relative to the contexts in which it may be helpful.

The paper first explores specific factors that are relevant when evaluating technological approaches for a specific translation project. For the translation project, factors may relate to the existing reference translations, the state of progress of the new translation, and the level (or desired level) of linguistic knowledge of the language in question. For each technological approach, it is important to be aware of the effort required to start and to continue, the effect of the approach on drafters’ behavior, and side benefits to the language community as a result of applying that approach.

Next, this paper describes several computer-assisted approaches, from straightforward encoding conversion, through rule-based approaches, to newer AI efforts. All have the potential to offer specific quality improvements to translations. Finally, it includes tables summarizing the relationship of the factors to the approaches, to assist groups making decisions about applying computer-assisted approaches to translation projects.