Analysis of the Practice of Human-Machine Collaborative Translation from the Perspective of Cognitive Load

Abstract

Artificial intelligence technology has gradually expanded the scope of applications for translation, and at the same time, the cognitive features of translation activities have also changed. Translators no longer complete the translations alone but work with machines. Cognitive Load Theory can be used to study the cognitive processing process of humans and machines in collaboration. Based on the above theory, this paper explores the characteristics of cognitive load for translators in human-machine collaborative translation, traces the application of this theory in translation research, and compares the cognitive differences under different collaboration modes. This paper takes Chen Lei's (2025) eye-tracking data as its reference, shows the gradual changes in cognitive load over time, and explores how translation direction and task difficulty affect cognitive load. Based on the above data, the fixation time and the number of regressions for the human-machine collaboration group were both lower than those of the manual translation group. However, this advantage is due to the quality of the machine's initial output and text difficulty. From the perspective of translation direction, the reduction effect of AI assistance in Chinese-to-English translation is relatively more pronounced than that in English-to-Chinese translation. Based on the above findings, this paper proposes the following three improvements: interface Design, skill training and task management.