This work adopts a thesis-by-publication format, collating three individual experiments that collectively examine attentional processing in simultaneous interpreting. Within this framework, the thesis presents an empirical investigation into the complex cognitive adaptation mechanisms that enable professional simultaneous interpreters to manage multi-task demands while operating near the limits of human processing capacity. Focused on attentional shifts (or orienting of attention) as a fundamental cognitive mechanism, we employ an integrated methodology combining the Visual World Paradigm (VWP) with pupillometry to provide real-time data informing attentional dynamics and cognitive load. In our data analysis, we rely on Generalised Additive Mixed Models (GAMMs) to capture the non-linear temporal dynamics and control for the autocorrelation inherent in high-resolution psycho-physiological data.

The thesis addresses three fundamental questions regarding cognitive adaptation in SI:

Experiment 1 (Chapter 2) investigates how complex predictive mechanisms adapt under sustained cognitive load. We found that cross-agent combinatory effects were significantly diminished or deferred in the early verb window of simultaneous interpreting. This indicates that in SI, high cognitive demands force a strategic trade-off in attention, though verb-driven selectional restrictions remain resilient.

Experiment 2 (Chapter 3) compares SI with comprehension and production tasks. We found that SI induces significantly higher and more sustained cognitive load compared to comprehension. Eye-movement data in SI revealed attenuated anticipatory gaze shifts, reflecting a pronounced verification bias and delayed processing strategies necessitated by continuous multi-task demands.

Experiment 3 (Chapter 4) examines how multimodal cues influence attentional allocation and cognitive efficiency. Here, we saw that the integration of congruent deictic gestures (pointing actions) facilitates quicker attentional shifts (early target fixation biases) and is associated with lower cognitive load. Importantly, auditory-only or neutral conditions impose the greatest ongoing processing burden, confirming that congruent visual input acts as an active resource conservation mechanism.

Collectively, these findings provide an empirical account of the attentional shifts and the features of cognitive load that define simultaneous interpreting. The thesis explores the psycholinguistic theories of incrementality and resource allocation by providing evidence that adaptation under pressure involves strategic simplification of prediction and reliance on multimodal input to mitigate overload. By doing so, we offer insights for enhancing multi-task resilience in interpreting practice and for informing the development of adaptive support technologies.