Auditory Speech Processing Efficiency: Functional Connectivity Graph Theory Analysis of Lexical Processing Effort

Network science models of real-time spoken word processing predict that word frequency and sublexical phonotactic probability directly influence verbal working memory. Behavioral accuracy and speed of processing suggest that high frequency words differ in cognitive processing effort as compared to low frequency words. Cognitive processing effort, assessed indirectly through changes in prefrontal oxygenated (HbO) and deoxygenated (HbR) hemoglobin concentrations, also suggest that high and low frequency words differ in cognitive effort. This novel study examines potential differences in linguistic processing effort and efficiency directly, using measures of functional connectivity (FC) strength, modularity, and global efficiency, derived from the listener's prefrontal cortex hemodynamic responses using a paradigmatic verbal working memory paradigm.

A total of 20 neurologically typical participants (age 18-21 years) completed an auditory n -back working memory task comparing performance for words differing in whole-word frequency and sub lexical phonotactic probability. Changes in HbO and HbR hemoglobin concentration were recorded with a continuous-wave, multi-channel fNIRS system (TechEn, Inc., Milford, MA) using a 20-channel optode montage across the prefrontal cortex. Partial correlation coefficients were calculated between each channel pair to produce 20 x 20 FC matrices for each participant. Prefrontal networks were constructed as a graph where nodes in the graph are the fNIRS measurements and edges are partial correlations between pairs of measurements.

Word frequency effects were evident in both in the behavioral and functional connectivity measures. Task performance, regardless of word frequency, was related to brain measures, such that higher prefrontal FC strength was negatively correlated with accuracy ( d' ) and high modularity was negatively correlated with response times for low frequency words. In keeping with the direction of the brain-behavior correlations, the highest performing participant had lower FC strength and higher efficiency whereas the lowest performing participant had higher modularity and lower global efficiency.

We identified network properties that are linked to lexical indices of cognitive processing effort in typical participants. These preliminary results point to strategies that could assess language function in atypical populations in future studies.