The Language Instinct: How the Mind Creates Language (47 page)

ERROR THAT MIGHT TEMPT A CHILD
: He ate something.
Ate he something?

For virtually all of these patterns, she found
no
errors among the 66,000 sentences in which they could have occurred.

The three-year-old child is grammatically correct in quality, not just quantity. In earlier chapters we learned of experiments showing that children’s movement rules are structure-dependent (“Ask Jabba if the boy who is unhappy is watching Mickey Mouse”) and showing that their morphological systems are organized into layers of roots, stems, and inflections (“This monster likes to eat rats; what do you call him?”). Children also seem fully prepared for the Babel of languages they may face: they swiftly acquire free word order, SOV and VSO orders, rich systems of case and agreement, strings of agglutinated suffixes, ergative case marking, or whatever else their language throws at them, with no lag relative to their English-speaking counterparts. Languages with grammatical gender like French and German are the bane of the Berlitz student. In his essay “The Horrors of the German Language,” Mark Twain noted that “a tree is male, its buds are female, its leaves are neuter; horses are sexless, dogs are male, cats are female—tomcats included.” He translated a conversation in a German Sunday school book as follows:

But little children learning German (and other languages with gender) are not horrified; they acquire gender marking quickly, make few errors, and never use the association with maleness and femaleness as a false criterion. It is safe to say that except for constructions that are rare, used predominantly in written language, or mentally taxing even to an adult (like
The horse that the elephant tickled kissed the pig
), all languages are acquired, with equal ease, before the child turns four.

The errors children do make are rarely random garbage. Often the errors follow the logic of grammar so beautifully that the puzzle is not why the children make the errors, but why they sound like errors to adult ears at all. Let me give you two examples that I have studied in great detail.

Perhaps the most conspicuous childhood error is to overgeneralize—the child puts a regular suffix, like the plural -
s
or the past tense -
ed
, onto a word that forms its plural or its past tense in an irregular way. Thus the child says
tooths
and
mousse
and comes up with verb forms like these:

These forms sound wrong to us because English contains about 180 irregular verbs like
held, heard, cut
, and
went
—many inherited from Proto-Indo-European!—whose past-tense forms cannot be predicted by rule but have to be memorized by rote. Morphology is organized so that whenever a verb has an idiosyncratic form listed in the mental dictionary, the regular -
ed
rule is blocked:
goed
sounds ungrammatical because it is blocked by
went
. Elsewhere, the regular rule applies freely.

So why do children make this kind of error? There is a simple explanation. Since irregular forms have to be memorized and memory is fallible, any time the child tries to use a sentence in the past tense with an irregular verb but cannot summon its past-tense form from memory, the regular rule fills the vacuum. If the child wants to use the past tense of
hold
but cannot dredge up
held
, the regular rule, applying by default, marks it as
holded
. We know fallible memory is the cause of these errors because the irregular verbs that are used the least often by parents (
drank
and
knew
, for instance) are the ones their children err on the most; for the more common verbs, children are correct most of the time. The same thing happens to adults: lower-frequency, less-well-remembered irregular forms like
trod, strove, dwelt, rent, slew
, and
smote
sound odd to modern American ears and are likely to be regularized to
treaded, strived, dwelled, rended, stayed
, and
smited
. Since it’s we grownups who are forgetting the irregular past, we get to declare that the forms with -
ed
are not errors! Indeed, over the centuries many of these conversions have become permanent. Old English and Middle English had about twice as many irregular verbs as Modern English; if Chaucer were here today, he would tell you that the past tenses of
to chide, to geld, to abide
, and
to cleave
are
chid, gelt, abode
, and
clove
. As time passes, verbs can wane in popularity, and one can imagine a time when, say, the verb
to geld
had slipped so far that a majority of adults could have lived their lives seldom having heard its past-tense form
gelt
. When pressed, they would have used
gelded;
the verb had become regular for them and all subsequent generations. The psychological process is no different from what happens when a young child has lived his or her brief life seldom having heard the past-tense form
built
and, when pressed, comes up with
builded
. The only difference is that the child is surrounded by grownups who are still using
built
. As the child lives longer and hears
built
more and more times, the mental dictionary entry for
built
becomes stronger and it comes to mind more and more readily, turning off the “add -
ed
” rule each time it does.

Here is another lovely set of examples of childhood grammatical logic, discovered by the psychologist Melissa Bowerman:

These are examples of the causative rule, found in English and many other languages, which takes an intransitive verb meaning “to do something” and converts it to a transitive verb meaning “to cause to do something”:

The causative rule can apply to some verbs but not others; occasionally children apply it too zealously. But it is not easy, even for a linguist, to say why a ball can bounce or be bounced, and a horse can race or be raced, but a brother can only die, not be died, and a girl can only giggle, not be giggled. Only a few kinds of verbs can easily undergo the rule: verbs referring to a change of the physical state of an object, like
melt
and
break
, verbs referring to a manner of motion, like
bounce
and
slide
, and verbs referring to an accompanied locomotion, like
race
and
dance
. Other verbs, like
go
and
die
, refuse to undergo the rule in English, and verbs involving fully voluntary actions, like
cook
and
play
, refuse to undergo the rule in almost every language (and children rarely err on them). Most of children’s errors in English, in fact, would be grammatical in other languages. English-speaking adults, like their children, occasionally stretch the envelope of the rule:

So both children and adults stretch the language a bit to express causation; adults are just a tiny bit more fastidious in which verbs they stretch.

The three-year-old, then, is a grammatical genius—master of most constructions, obeying rules far more often than flouting them, respecting language universals, erring in sensible, adultlike ways, and avoiding many kinds of errors altogether. How do they do it? Children of this age are notably incompetent at most other activities. We won’t let them drive, vote, or go to school, and they can be flummoxed by no-brainer tasks like sorting beads in order of size, reasoning whether a person could be aware of an event that took place while the person was out of the room, and knowing that the volume of a liquid does not change when it is poured from a short, wide glass into a tall, narrow one. So they are not doing it by the sheer power of their overall acumen. Nor could they be imitating what they hear, or else they would never say
goed
or
Don’t giggle me
. It is plausible that the basic organization of grammar is wired into the child’s brain, but they still must reconstruct the nuances of English or Kivunjo or Ainu. So how does experience interact with wiring to give a three-year-old the grammar of a particular language?