The perils of combining translation tasks and judgment tasks

Abstract

The combination translation/judgment task involves a translation task immediately fol- lowed by a judgment task, where the sentence to be judged is a minimally altered version of the sen- tence just volunteered in the translation task. While this composite task can be an efficient way of gen- erating semantic data, it also has some significant vulnerabilities. In this paper, I examine transcribed dialogues from elicitation with four Kwak’wala language consultants and discuss how the structure of the combination translation/judgment task can give rise to an implication that the consultant’s vol- unteered sentences are deviant. This implication, in turn, can undermine consultants’ confidence in their own native speaker intuitions, which can change the way they respond to judgment tasks. To prevent this issue, careful attention must be paid to how the combination translation/judgment task is explained to and understood by consultants.