Salience is in the brain of the beholder: ERPs reflect acoustically salient variables
Jan 1, 2022·,
,·
1 min read
Boswijk, V.
Hilton, N.H.

Matt Coler
Loerts, H.
Abstract
A recent paper by Boswijk, Loerts & Hilton (Boswijk et al., 2020) in this journal discusses how technological advances allow us to explore the cognitive processing of so-called salient linguistic features, and how this could provide us with quantifiable measures of ‘salience’. The paper concludes that, although promising, the used measure of pupil dilation seems to be limited as a measure for linguistic salience, and therefore refers future research to other measures, specifically Event Related Potentials (ERPs). In this paper we therefore replicate the Boswijk et al. study using the ERP measure with the hypothesis that linguistic salience evokes distinct ERP components. We use the same materials that were used in the Boswijk et al. (2020) paper to observe changes in Dutch participants’ pupil sizes when listening to stimuli containing salient and non-salient variants of linguistic variables. Using Generalized Additive Mixed Modelling (GAMM), we find distinct responses for five of six stimuli categories. We consider our findings in light of the literature on linguistic salience and discuss how our findings relate to the Boswijk et al. (2020) study. We find that ERPs provide a more fine-grained measure of theoretically salient stimuli.
Type
Publication
Ampersand, 9, 100085
Publication Type: Journal Article
This study investigates how Event Related Potentials (ERPs) can be used to measure linguistic salience. The research replicates a previous study that used pupil dilation measures, finding that ERPs provide a more fine-grained measure of theoretically salient linguistic stimuli.
Highlights
- Six operationalizations of salience are investigated in relation to ERPs
- Linguistic salience resulted in an increased N400 for most categories
- Results indicate that words in the salient condition were harder to integrate with the pre-existing mental context
- The level of salience is strongly dependent on current context
The research employs Generalized Additive Mixed Modelling (GAMM) to analyze neural responses to salient and non-salient linguistic variants, contributing to our understanding of how the brain processes salient speech features.