Pupil Size Reflects Increased Processing Load for Salient Variables

In linguistics, salience is used to point towards features that are e.g. more prominent or occur with more regularity. An exact and univocal definition of the concept of salience is currently lacking, but some have proposed that we might look into the possible relationship there might be with processing load (cf. Ellis, 2016). Technical advances in the fields of psycho and neurolinguistics provide us with opportunities for exploring this potential relationship between salience and cognition. While these techniques cannot provide us with an answer to the question of what salience is, the possibility of having a quantitative measure of salience can bring us closer to measuring its relationship with linguistic and social factors. One particularly useful technique in this respect is eye-tracking. Auditory salience may be related to dilation in pupil size, which in turn reflects cognitive load or mental effort (cf. Blumenthal-Drame et al., 2017; Ellis, 2016). The present study therefore examines participants’ pupil sizes while listening to stimuli in which various linguistic categories are manipulated to contain salient and non-salient equivalent variants (e.g. high intensity recordings vs. low intensity recordings, recordings with high-frequency words vs. low-frequency words). Using Generalized Additive Modeling (GAM), we found that pupil size significantly increased for three of the categories we hypothesized to be salient: Acoustic Prominence, Gender and Loudness. This points towards an increase in processing effort for these categories.