Saffran, aslin, and newport (1996) presented infants with a continuous speech stream made up of four different three-syllable nonsense words, for a total of two minutes. then, they tested whether infants had picked up on "word boundaries" by presenting these nonsense words and new sequences. the infants showed a preference for demonstrating that they

The infants showed a preference for "listening to the new sequences", demonstrating that they "had picked up on the word boundaries of the original nonsense words".

This investigation demonstrates that a basic work of language acquisition, division of words from fluent discourse, can be accomplished by 8-month-old newborn children based exclusively with respect to the statistical connections between neighboring discourse sounds. Additionally, this word division depended on statistical learning from just 2 minutes of presentation, proposing that newborn children approach a ground-breaking instrument for the calculation of statistical properties of the language input.

can you explain it better

well to answer ur question to be completly honest