Translation directionality and the Inhibitory Control Model: a machine learning approach to an eye-tracking study

Area of Science:

• Cognitive Science
• Translation Studies
• Computational Linguistics

Background:

• The Inhibitory Control Model suggests
• translation asymmetry
• due to directionality in cognitive loads during translation.
• Physiological data, such as pupillometry, can offer insights into cognitive processes.
• Novice translators' cognitive loads during L1 and L2 translation require further investigation.

Purpose of the Study:

• To empirically verify
• translation asymmetry
• in novice translators using pupillometry.
• To explore the application of machine learning in Cognitive Translation and Interpreting Studies.
• To investigate the influence of translation directionality on cognitive effort.

Main Methods:

• An eye-tracking experiment was conducted with 14 novice Chinese-English translators.
• Pupillometry data were collected during L1 and L2 textual translation tasks.
• The XGBoost machine-learning algorithm was employed to analyze pupillometric and demographic data.

Main Results:

• A Wilcoxon signed rank test confirmed
• translation asymmetry
• at a textual level, validating the Inhibitory Control Model's prediction.
• The XGBoost model accurately predicted translation directions using pupillometric and demographic data.
• Directionality was confirmed as a significant factor influencing cognitive load during translation.

Conclusions:

• Translation asymmetry is a valid phenomenon at the textual level for novice translators.
• Machine learning approaches show promise for advancing Cognitive Translation and Interpreting Studies.
• Pupillometry combined with machine learning offers a powerful tool for analyzing translation processes.