What you will get from this article
- Why “children learn language because parents teach it” does not hold up
- What Chomsky’s “poverty of the stimulus” argument is
- How to check for yourself a “structure-dependent rule” using English questions
The answer first. Children know, before being taught, things that the examples parents provide do not settle. One is that a sentence is made of phrases, not a straight line of words. Pinker shows this with how questions are formed and with experiments on children, and explains that it is at the heart of Chomsky’s argument that the basic design of language is innate.
Background: Chomsky’s two facts (p. 22)
Pinker sums up Chomsky’s points as two.
virtually every sentence that a person utters or understands is a brand-new combination of words, appearing for the first time in the history of the universe.
So a language cannot be a repertoire of responses; the brain must contain a program (a mental grammar) that can build an unlimited set of sentences. Second, children develop complex grammars without instruction and give consistent interpretations to sentences they have never heard. From there the argument moves to the claim that children come equipped with a plan common to all languages (Universal Grammar).
The folk theory that “parents teach” (pp. 39–40)
Pinker first examines the idea that parents teach.
- No parent gives grammar lessons. But some child psychologists think mothers give implicit lessons in “Motherese.”
- In contemporary middle-class America, parenting is seen as a responsibility to keep the helpless infant from falling behind in the race of life, and that mentality feeds the belief that Motherese is essential.
- Other cultures differ. The !Kung San of the Kalahari believe children must be drilled to sit, stand, and walk, and prop infants up in sand. Every one of these children sits up on its own anyway: children develop the ability without teaching.
- In many communities parents do not speak to prelinguistic children at all, reasoning that it is pointless to talk to someone who cannot understand. Pinker quotes a South Carolina woman and notes that children in such communities, overhearing the talk around them, come to speak a fully grammatical dialect.
Pinker’s conclusion: Motherese is not a necessary condition of language acquisition.
The unicorn question (pp. 40–42)
What, then, do children know without teaching? Pinker takes Chomsky’s famous example.
Consider a simple procedure for turning a statement into a question.
- A unicorn is in the garden.
- Move the auxiliary is to the front → Is a unicorn in the garden?
It works. Now apply the procedure “scan left to right and move the first auxiliary you meet” to this sentence.
A unicorn that is eating a flower is in the garden.
Moving the first is gives the odd Is a unicorn that eating a flower is in the garden? The right result moves the second is: Is a unicorn that is eating a flower in the garden?
Why not the first is? Chomsky’s answer lies in the basic design of language.
- A sentence is not a linear string of words; words are grouped into phrases, and phrases into bigger phrases, a structure.
- The question rule does not look for “the first auxiliary” but for “the auxiliary right after the subject noun phrase.”
- The subject phrase “a unicorn that is eating a flower” behaves as a single unit; the first is is buried inside it and invisible to the rule.
This is called a structure-dependent rule.
Why is this evidence of “not taught”?
Chomsky’s reasoning (pp. 41–42):
- If the logic of language is wired into children, then the first time they meet a sentence with two auxiliaries they should be able to form the right question.
- That should hold even though the linear rule is simpler and presumably easier to learn.
- And it should hold even though sentences that would teach children the structure-sensitive rule is right (questions with a second auxiliary embedded in the subject) are so rare in Motherese as to be nonexistent.
Pinker calls this “the argument from the poverty of the input” (p. 42).
The experiment with children (p. 42)
The claim was tested, Pinker says. The psycholinguists Crain and Nakayama had children aged three, four, and five play a game.
- One experimenter holds a doll of Jabba the Hutt from Star Wars.
- The other tells the child: “Ask Jabba if the boy who is unhappy is watching Mickey Mouse.”
- Jabba looks at a picture and answers yes or no. The one being tested is the child, not Jabba.
The children cheerfully produced suitable questions. As Chomsky would have predicted, not one child produced the ungrammatical string the simple linear rule would yield, says Pinker.
“Aren’t they going by meaning?” (pp. 42–43)
One could object that children are just working out the subject from meaning. So Crain and Nakayama mixed in questions with a meaningless it: “Ask Jabba if it is raining in this picture.” The it of Is it raining? refers to nothing; it is a grammatical placeholder. The children handled these correctly too.
Pinker adds that the universal constraints cannot be explained as an outcome of usefulness. Many languages scattered over the globe move an auxiliary to the front to form questions, always in a structure-dependent way. But one could just as well have moved the leftmost auxiliary, or reversed word order (a trick the human mind is capable of). Languages do not. That would be as unlikely as isolated inventors independently coming up with identical standards for typewriter keyboards or Morse code (p. 43).
The “-s” agreement (pp. 43–45)
Another piece of evidence is the English third-person “-s” (He walks). In modern English it is nearly superfluous, a remnant of an older system. Yet to use it, a speaker must track four things in every sentence (p. 44).
| What to check | Example |
|---|---|
| Is the subject third person? | He walks / I walk |
| Singular or plural? | He walks / They walk |
| Present or past? | He walks / He walked |
| Habitual or going on now (aspect)? | He walks to school / He is walking to school |
And the child must pick out these four factors from all the information in the sentence (the number of syllables in the last word, whether the object of a preposition is natural or man-made, and so on).
Yet by three and a half, Pinker says, children use “-s” in over ninety percent of the sentences that require it and almost never where it is forbidden (p. 44). A child under the pseudonym Sarah spoke sentences such as Anybody knows how to scribble and It looks like a donkey face correctly (pp. 44–45).
Sarah also produced forms she could not have heard from her parents, such as He gots a scary one and She do’s what her mother tells her (p. 45). These are products of an unconscious agreement rule of her own making, not imitation. And if children were imitators, Pinker quips, why do they not imitate their parents’ habit of sitting quietly in airplanes? (p. 45)
Summary
- Without parents teaching, children acquire language (San, South Carolina).
- Children use structure-dependent rules without being taught (the question experiments).
- Children make rules themselves (forms like “gots” and “do’s” that they never heard).
- From this Pinker argues that the basic design of language is already on the child’s side.
Cautions
- The poverty-of-the-stimulus argument is still contested. There is a counter-view that input alone suffices (this series does not cover that side). This article is a presentation of Pinker’s argument.
- The experiments follow the book’s account; I have not checked the original papers.
Next
- Language that breaks alone, language that survives alone: evidence from brain and genes
- On the “tree” structure of sentences: A sentence is a tree, not a chain