Speech segmentation is one of the first steps of language acquisition. Understanding how this cognitive process unfolds in early infancy is important for better defining the origin of later linguistic deficits often observed in children born pre-term or in children with language learning deficits. In this presentation, I will show the results of a longitudinal study conducted in infants, in which electrophysiological data were collected at birth, at 6, and at 10 months while infants were presented with both flat and melodically enriched streams of artificial syllables. Results show enhanced brain responses to melodically enriched compared to flat speech streams across the three ages. Taken together, these results provide direct neural evidence of the benefit of prosody in the first steps of language acquisition.