Listening to speech under adverse conditions such as background noise and competing talkers can demand higher effort and be challenging for hearingimpaired individuals. The higher effort can lead to adverse effects such as fatigue. The cognitive factors related to the higher demand are not considered in the traditional measurements of the performance of a listening task. Therefore, the concept of listening effort was defined and highly studied to improve the evaluation of hearing disability. Several methods and tools have been used to estimate correlates of listening effort, among which electroencephalography (EEG) has gained attraction due to being non-invasive and having high temporal resolution.

EEG is a method to record the brain’s electrical activity by electrodes placed on the scalp. The relationship between and within different regions of the brain can be measured by assessing the statistical dependencies between and within EEG signals through the functional connectivity (FC) analysis, which can provide insights into the cognitive functions of the listening effort.

The objective of this thesis was to utilize FC analysis in the EEG to estimate correlates of listening effort. We divided our FC analysis into globaland local-scale analyses based on the distance between EEG electrodes. A directional measure, conditional transfer entropy (TE), was selected regarding global-scale connectivity assessment in EEG signals.

We proposed a new estimator for conditional TE estimation and a new directed dependency measure based on conditional TE by which we could get higher accuracy in simulated data and real-world intracranial EEG data recorded during seizures. However, the estimation of correlates of listening effort by using global-scale connectivity analysis in real-world EEG data led to negative results. Regarding local connectivity assessment in response to speech in adverse conditions, an existing multivariate phase synchrony measure called circular omega complexity (COC) was selected.

The local connectivity was shown in this thesis to have the capability to estimate correlates of listening effort during a continuous long speech in noise task. Finally, we studied the effect of noise reduction (NR) schemes in hearing aid (HA) on local connectivity in EEG signals.