Adults have the ability to distinguish where one word ends and the next begins, but the process of separating words from the flow of speech requires complex mental processing. When someone speaks, the words reach the ear as an uninterrupted stream of sound that does not separate the words spoken. A child uses statistical learning to understand the sounds of whole words. Studies show that eight-month-olds discover word-like components supported by the likelihood that one syllable will follow another. Scientists conducted an experiment in which children listened to a series of computer-synthesized nonsense words made up of syllables, some of which occurred together more often than other syllables. The children were able to focus on the syllables that correspond to nonsense speech and identified probable words. This groundbreaking study demonstrated children's statistical learning abilities and presented a theory of language that goes beyond the general idea that a child learns only through parental habituation and assertion of whether a word is correct or incorrect. Children learn long before parents recognize this is happening. Further studies resulted in a crucial discovery that provided a crucial signal that the statistical learning process does not exclusively involve submissive listening. The infantile brain does not turn out to be a passive process