Predicting Human Translation Difficulty with Neural Machine Translation
This work addresses the problem of understanding cognitive processes in translation for researchers and practitioners, representing the most comprehensive investigation to date, though it is incremental as it builds on existing methods.
The study tackled the problem of predicting human translation difficulty by evaluating how well surprisal and attentional features from a Neural Machine Translation model account for reading and production times, finding that surprisal is the most successful predictor of production duration based on data from hundreds of translators across 13 language pairs.
Human translators linger on some words and phrases more than others, and predicting this variation is a step towards explaining the underlying cognitive processes. Using data from the CRITT Translation Process Research Database, we evaluate the extent to which surprisal and attentional features derived from a Neural Machine Translation (NMT) model account for reading and production times of human translators. We find that surprisal and attention are complementary predictors of translation difficulty, and that surprisal derived from a NMT model is the single most successful predictor of production duration. Our analyses draw on data from hundreds of translators operating across 13 language pairs, and represent the most comprehensive investigation of human translation difficulty to date.