ELI5: What is statistical learning? And can you apply that explanation to language acquisition?

Statistical learning is your brain's natural ability to automatically notice hidden patterns and figure out how likely things are to happen just by experiencing everyday life.

Imagine you look outside every morning. If you see dark, gray clouds and then it rains, your brain takes a mental note. After seeing this happen fifty times, you automatically know that dark clouds mean you should grab an umbrella. Nobody had to teach you the math; your brain simply counted the clues over time.

This exact same super-power is how babies learn to understand and speak their very first language.

When grown-ups talk, they don't leave spaces between spoken words like we do in books; it sounds like one continuous stream of noise. A baby has to figure out where one word ends and the next one begins.

Here is how a baby uses statistical learning in everyday life:

• Spotting word boundaries: Imagine a parent frequently says "pretty baby" and "pretty dog." The baby's brain notices that the sound "pret-" is almost always followed by "-ty." But the sound "-ty" is followed by many different sounds, like "ba-" or "dog." Because "pret-" and "-ty" stick together so often, the baby figures out that "pretty" is a single word.
• Predicting word order: Over time, a child notices that the word "the" is almost always followed by a thing (like "the ball" or "the cat"), but never by an action (people do not say "the jump"). Without ever opening a grammar textbook, the child learns the rules of sentence structure.
• Matching words to the world: If every time the family pet walks into the room someone says "look at the kitty," the brain calculates the high probability that the sound "kitty" belongs to the furry animal with whiskers.