Think about listening to people speaking a foreign language. Even though we are expert communicators as adults, it’s very hard to pick up what people are saying in languages we don’t know. It’s hard not only to comprehend the words, but even to hear where the words begin and end. Fluent speakers talk so fluidly that the perceived spaces between words in a sentence are auditory illusions. When we listen to English, we hear where the breaks are because we know what the words are, rather than vice versa.
Speech doesn’t come pre-segmented for infants, so they can’t rely on speakers to provide a gap between every word (and in fact, it feels quite unnatural and challenging if we try to speak this way—try reading this sentence with a pause at each space). Instead, amazingly, they rely on statistics.
As young statisticians, babies pay attention to the probabilities of events in the world. As they accumulate more and more time spent hearing people talk, they begin to identify sounds that are more likely to occur together. Imagine hearing many instances of “baby” throughout the day. “Hello baby!” “Look at the baby.” “What a cute baby.” Each of these phrases helps the young listener learn which syllables reliably occur together, and which syllables are paired less often. These transitional probabilities—the likelihood that any given syllable will follow another—allow the baby to learn that the predictable pairing ba-by is probably a word unit, while a rarer succession such as lo-bay (from “Hello baby”) is less likely to be indicative of a meaningful association.
Babies attend not only to the sounds of speech, but also to when and where certain words tend to occur across situations.
Statistical cues also help babies figure out what words mean. If we don’t yet know the meaning of “song,” we may become reasonably sure that it’s a musical tune if someone bursts out singing every time “song” is mentioned. In this way, babies attend not only to the sounds of speech, but also to when and where they tend to occur across situations. One step is to pick out “baby” as a unit of speech, and another is to associate that “baby” tends to be used as an affectionate reference to the baby herself.
Social cues also help babies tune in to especially supportive learning opportunities. Child-directed speech (commonly known as “baby talk,” which is characteristically slower, higher-pitched, and more dynamic in register than speech to adults) helps signal to babies when communication is designed for them. Speakers can also use posture and gesture to share intentions: eye contact, gaze, pointing, and demonstration are all cues to helpfully guide babies’ attention. While laboratory studies demonstrate that babies can pick out words using statistical probabilities alone, they’re a whole lot faster at identifying words and their meanings when social cues are also available.
Once babies start to comprehend a few words, they can leverage their knowledge to learn more of them. Imagine seeing a ball and a funky new toy. We may not know the name of the new toy, but if we hear someone say, “Hand me the dax,” we can assume with some certainty that dax refers to the new toy. We reason that if the speaker had meant the ball, she would have said “ball.” Babies appear to use the same logic to learn new words in their environments, mapping new words to new meanings. The more words they do know, the more they can seek and apply labels to what they don’t know.
Building up an early vocabulary opens the floodgates for new learning opportunities. Knowledge of words helps young children communicate about intentions, ideas, desires, and memories. More experience with language also helps them develop a more nuanced understanding of how people share non-literal meaning, including sarcasm, metaphor, humor, and politeness. Breaking into the world of language is transformational to the life of a young child. Their remarkable decoding of streams of sound into shared communication allows them to fully join the social world that makes us human.
This article was originally published on