By age four or five, a typical child anywhere speaks in complex sentences, without formal lessons. How this happens is one of the great questions in the study of the mind. This level describes what children do, then the main theories about how they do it. The General Psychology course covers the wider history of behaviorism and cognitive science.
Milestones
Babies cry from birth and coo within a few months. Babbling, repeated syllables like "ba-ba-ba," usually begins around six months. Deaf babies exposed to sign language babble with their hands, which suggests babbling belongs to language, not just to the mouth. First words typically appear around the first birthday. Between about eighteen months and two years many children start combining two words: "more juice," "daddy go." The ranges are wide, and late starters are often entirely normal.
Mistakes that reveal rules
Some of the best evidence about learning comes from children's errors. Overregularization is when children apply a regular rule to exceptions, saying "goed" or "mouses." Often a child first says "went" correctly, apparently memorized as a single word, then switches to "goed," then later returns to "went." The middle stage shows the child has found the rule for the past tense and is applying it everywhere.
In a famous 1958 study, the psychologist Jean Berko Gleason showed children a drawing of a made-up creature called a wug, then a picture of two of them. Children given an invented word like "wug" readily produced the plural "wugs", a word they could never have heard. They had learned a rule, not just a list.
Chomsky's answer
Noam Chomsky, the most influential linguist of the twentieth century, argued that children must come equipped for this task. In his view, children are born with a universal grammar, an innate set of principles shared by all human languages. Experience sets the details: which words, which word order, which sounds.
A key argument is the poverty of the stimulus. The poverty of the stimulus is the argument that the speech children hear is too limited to explain the grammar they acquire without inborn guidance. Children hear messy speech, rarely get grammar corrected, and yet avoid whole classes of errors. For example, to make a question from "The man who is tall is happy," children move the second "is," not the first, though no one teaches them which one to move.
The critics
Many researchers reject the idea of an inborn grammar. Usage-based theorists, led by the psychologist Michael Tomasello, argue that children build grammar gradually from specific phrases, using general learning abilities and their skill at reading other people's intentions.
Other researchers emphasize how much infants can learn from patterns. In a 1996 study, Jenny Saffran, Richard Aslin and Elissa Newport played a stream of made-up syllables to babies. Eight-month-old infants tracked which syllables tended to follow each other, after only two minutes of listening, and used this to pick out "words." Critics of universal grammar also note how widely languages differ, and argue that few if any features are truly found in all of them. Chomsky's defenders reply that statistics alone cannot explain the abstract rules children reach, and the debate continues.
Timing matters
In 1967 the neurologist Eric Lenneberg proposed a critical period: a window in childhood after which a first language is much harder to acquire fully. The best evidence comes from deaf people who were not exposed to any sign language until later in life; on average, they never reach the fluency of those exposed from birth. For second languages the picture is softer, though early starters usually end up with a more native-like accent.
A language born in a schoolyard
In the late 1970s and 1980s, Nicaragua opened new schools for deaf children in Managua. The children, most of whom had used only simple home signs, began to sign with each other, and a new language emerged. The researcher Ann Senghas and her colleagues found that younger signers made the new language more systematic than the older signers who had started it. Both sides of the debate claim this case: children built structure without a model, but together, through use.
More than one language
Parents sometimes worry that two languages will confuse a child. Research has found the opposite. Children can learn two or more languages from birth without being confused or harmed. Mixing the two in one sentence is normal, and worldwide, growing up with several languages is common.
Worked example: a three-year-old's sentence
A three-year-old says: "I goed to the park and I seed two sheeps."
The errors are revealing. "Goed" and "seed" apply the regular past tense rule to irregular verbs. "Sheeps" applies the plural rule to a word whose plural is unchanged. The child has not heard these forms from adults. They are the child's own inventions, made by a rule.
If the parent replies, "You went to the park? You saw two sheep?" the child may well say, "Yes, I goed." Researchers have recorded many such exchanges. A Chomskyan sees here a mind following its own internal timetable. A usage-based theorist sees a child who will gradually notice the exceptions after hearing them many more times. Both agree that the child is not copying. The child is building a grammar.