Adults have the ability to differentiate where one word ends and the next begins, but the process of separating words from the speech stream requires complex mental processing. When someone speaks, the words come to the ear as an uninterrupted stream of sound that does not separate the spoken words. An infant uses statistical learning to understand the sounds of whole words. Studies show that eight-month-old infants discover word-like components supported by the likelihood that one syllable follows another. The scientists conducted an experiment in which babies listened to a series of computer-synthesized gibberish words made up of syllables, some of which appeared together more often than others. The infants were able to focus on the syllables that matched the nonsense language and identify likely words. This groundbreaking study demonstrated the statistical learning abilities of infants and presented a theory of language beyond the general idea that a child only learns through parental habit and the assertion that a word is correct or incorrect. Babies learn long before parents realize it. Additional studies shaped a crucial discovery that provided an imperative sign that the process of statistical learning does not only involve submissive listening. The infant brain does not appear to be a passive process for