How is meaningful and related information from sentences combined and comprehended across time by the human brain? This project uses electroencephalography (EEG) to examine the neural signatures for language discourse processing. Across two experiments, we are measuring neural activity associated with discourse processing as human listeners listen to related vs. unrelated sequences of sentences. Preliminary results show that neural activity more closely tracks speech signals when sentences form a coherent discourse as compared to sentences that not meaningfully related. This research will have implications for understanding the neural instantiation of language processing and natural intelligence in the human brain and may provide useful algorithms for enhancing large language models used in AI. It also provides an empirical test of the assemblies hypothesis, which proposes an intermediate level of organization of brain computation, on a scale far larger than that of individual neurons and synapses, and realized by highly interconnected sets of assemblies.