Spoken language is central to human communication, influencing cognition, learning, and social interactions. Despite its spontaneous nature, characterized by disfluencies, fillers, self-corrections and irregular syntax, it effectively serves its communicative purpose. Understanding how the brain processes natural language offers valuable insights into the neurobiology of language.Recent neuroscience advancements allow us to study neural processes in response to ongoing speech, requiring detailed, time-locked descriptions of speech material to capture the nuances of spoken language. While there are many speech-to-text tools available, obtaining a time-locked true verbatim transcript, reflecting everything that was uttered, requires additional effort to achieve an accurate representation.Our work outlines a semi-automatic pipeline for annotating natural speech, developed for German and Hebrew but adaptable to other languages, for creating temporally precise time-courses describing key linguistic features of continuous speech, which can be used to analyze their neural representation and level of processing. We discuss the methodological challenges and opportunities this presents, for improving our understanding of how the brain processes everyday language.

This study investigates how selective auditory attention influences the lexical speech segmentation process to phonemes and words in a two competing speaker scenario. Using EEG recordings from 20 participants, we applied temporal response function (TRF) analysis to distinguish attention-driven neural activity to phoneme and word onsets for the attended and ignored speech stream separately. Our results reveal distinct attention effects for phoneme and word onsets. Phoneme onsets elicited significant selective attention effects at an early (18-94 ms, P1), middle (186-252 ms, P2), and late time window (302-382 ms, N2). In contrast, word onsets showed attention effects only at a middle (192-280 ms, P2) and late (348-386 ms, N2) time window, occurring slightly later than phoneme-related effects. Prediction accuracy analyses demonstrated stronger model performance for the attended speech stream across all models, with notable improvements in prediction accuracy from word to phoneme and combined word & phoneme model. These findings are in accordance with both hierarchical and parallel processing frameworks, where selective attention enhances lexical segmentation for attended speech, improving prediction accuracy. Early attention effects observed for phoneme onsets underscore their role in low-level speech processing, while late attention effects for word onsets may reflect higher-level processing. This study highlights the importance of selective attention in neural speech tracking and provides insights into auditory processing mechanisms underlying speech comprehension in complex acoustic environments.