The ability to differentiate speech sounds and detect abstract rules between them are fundamental for human language comprehension. However, these building blocks of human language abilities are not unique to humans, but also observed in other species. In my lecture, I will describe results from our studies in rodents, where speech perception and speech sound learning have been investigated. These studies are based on an electrophysiological marker of change detection, named mismatch negativity (MMN). Our findings suggest that, similarly to the human brain, the rodent brain can differentiate speech sounds, learn rules between them, and generalize these rules to new items. In addition, I will describe our results suggesting that mere passive exposure to (foreign) speech sounds can promote phonetic learning in adult humans and rats.