Learning from the statistics of their native language, infants slowly build the inventories of sounds that are present in their native language. It may seem effortful for such young babies to keep track of all of this information, but scientists believe that statistical learning in children takes place automatically, without conscious effort.
Statistical learning is a useful strategy for learning about many aspects of language. Infants use it not only for tracking which sounds are part of their native language, but also for learning syllables, words, and even rhythmic patterns of language. One research study found that 8-month-old babies use statistical learning to figure out which sounds make up words in their language. As you are listening to this recording, think about what spoken sentences sound like. Even though we adults easily recognize words in a sentence, there are actually no pauses between the words in a sentence when you are speaking out loud. By using statistics about syllables in speech, children are able to determine which combinations of syllables might be words.
For instance, children can listen for syllables like ba, da, or ga, and track how often they occur in speech and also what syllables occur together. Syllables that frequently occur together in speech are likely to be words, like “bay” and “bee” (baby), while syllables that do not occur together often are not likely to be words, like “bee” and “bay.” Using statistical learning, 8-month-olds are able to pick out words from speech.
In summary, statistical learning is a powerful language learning technique. However, it is not the only strategy that babies use. In study after study, researchers have found that social interactions in the form of face-to-face exchanges with live humans are critical for language learning. In the next section, we will discuss the role that social interactions play in language learning.