The human brain is regarded as an ensemble of dynamic systems in which communication between neural centres is very important. In order to perceive sounds many different cortical and subcortical brain areas have to coordinate their activity. After hearing loss, the connections and information pathways between these areas may rearrange and this may be one of the reasons for unsatisfactory speech perception after cochlear implantation (CI). It remains unclear how the brain connectivity and its re-organisation contribute to this, and this provides the motivation for the current study. The brain organisation can be quantified by connectivity measures, which may give the strength, direction, and timing information on the connections between brain areas. This research project aims to assess different methods of brain connectivity and response detection in the Electroencephalogram (EEG) and use these to investigate brain responses during tone, word, and sentence perception in normal hearing adults and CI users. The initial focus of this project was Dynamic Causal Modelling connectivity method, but early results raised questions regarding its reliability. DCM was then replaced by simpler and more established linear multi-variate auto regressive (MVAR) based models such as Coherence, Directed Transfer Function (DTF), and Partial Directed Coherence (PDC), as well as classical non-parametric power-spectral and coherence analysis. Both latter approaches could find changes in the brain activity in different time-frequency windows after the stimulus onset, which depended on the stimulus type and electrode positions. MVAR-based models were then employed and showed promising results when applied on synthetic data but on recorded data only model-based coherence (both pair-wise and multi-channel models) was able to detect connectivity changes in response to the stimuli presented to normal hearing participants; DTF and PDC appeared insufficiently sensitive. Different artefact rejection methods were also employed to remove the CI artefact from the EEG, prior to performing connectivity analyses. While connectivity changes could be identified, results need to be interpreted with caution, due to remaining uncertainty about the removal of all CI artefacts. This work is original in analysing and detecting changes in connectivity following repeated stimulation with words and sentences. Finding these changes proved challenging, with many pitfalls with established methods, but the current results and methodological approaches are promising for the continuing study of higher level responses to speech stimulation.