11.5 The Nature of Language and Children's Early Linguistic Abilities

Of all the things that people can do, none seems more complex than understanding and speaking a language. Thousands of words and countless subtle grammatical rules for modifying and combining words must be learned. Yet nearly all people master their native language during childhood; in fact, most are quite competent with language by the time they are 3 or 4 years old. How can children too young to tie their shoes or to understand that 2 plus 2 equals 4 succeed at such a complex task? Most developmentalists agree that language learning requires innate mechanisms that predispose children to it, coupled with an environment that provides adequate models and opportunity to practice. In this section, we will first consider briefly the question of what language is and then chart the normal course of its development. Then, in the next section, we will explore the innate and environmental requirements for language development.

Universal Characteristics of Human Language

Just what is it that children learn when they learn a language? Linguists estimate that at least 6,000 separate languages exist in the world today, all distinct enough that the speakers of one cannot understand those of another (Grimes, 2000). Yet these languages are all so fundamentally similar to one another that we can speak of human language in the singular (Pinker & Bloom, 1992).

All Languages Use Symbols (Morphemes) That Are Arbitrary and Discrete

Every language has a vocabulary consisting of a set of symbols, entities that represent other entities. The symbols in a language are called morphemes, defined as the smallest meaningful units of a language—that is, the smallest units that stand for objects, events, ideas, characteristics, or relationships. In all languages except the sign languages used by the deaf, morphemes take the form of pronounceable sounds. Most morphemes are words, but others are prefixes or suffixes used in consistent ways to modify words. Thus, in English, dog is both a word and a morpheme, -s is a morpheme but not a word, and dogs is a word consisting of two morphemes (dog and -s). The word antidisestablishmentarianism contains six morphemes (anti-dis-establishment-arian-ism), each of which has a separate entry in an English dictionary.

Morphemes in any language are both arbitrary and discrete. A morpheme is arbitrary in that no similarity need exist between its physical structure and that of the object or concept for which it stands. Nothing about the English morpheme dog, or the Spanish morpheme perro, or the French morpheme chien, naturally links it to the four-legged, barking creature it represents. This is in contrast to nonverbal signals (discussed in Chapter 3) that typically develop from and bear physical resemblance to such actions as fighting or fleeing; such signals communicate intentions to engage in the actions that they resemble. Because morphemes are arbitrary, new ones can be invented whenever needed to stand for newly discovered objects or ideas, or to express newly important shades of meaning. This characteristic gives language great flexibility. A morpheme is discrete in that it cannot be changed in a graded way to express gradations in meaning. For example, you cannot say that one thing is bigger than another by changing the morpheme big. Rather, you must add a new morpheme to it (such as -er) or replace it with a different morpheme (such as huge).

All Languages Are Hierarchically Structured in a Similar Way

In addition to commonalities in their symbol systems, all languages share a particular hierarchical structure of units (see Figure 11.13). The top level (largest unit) is the sentence, which can be broken down into phrases, which can be broken down into words or morphemes, which can be broken down into elementary vowel and consonant sounds called phonemes. The power of this four-level organization is that the relatively few phonemes (anywhere from 15 to 80 occur in any given language) can be arranged in different ways to produce an enormous number of possible words, which themselves can be arranged in different ways to produce a limitless number of possible phrases and sentences.

Every language has rules—collectively referred to as the grammar of the language—that specify permissible ways to arrange units at one level to produce the next higher level in the hierarchy. Grammar includes rules of phonology, which specify how phonemes can be arranged to produce morphemes; rules of morphology, which specify how morphemes can be combined to form words; and rules of syntax, which specify how words can be arranged to produce phrases and sentences. These rules differ from language to language, but every language has them, and similarities exist across languages in the fundamental nature of the rules (Pinker & Bloom, 1992; Pinker & Jackendoff, 2005).

Grammatical Rules Are Usually Learned Implicitly, Not Explicitly

People often think of grammar as something they learned (or tried to learn) in elementary school (which used to be called grammar school, perhaps for that very reason). But grammar is learned implicitly, without conscious effort, long before formal schooling. The fact that 4-year-olds can carry on meaningful conversations with adults, producing and understanding new and unique sentences, indicates that by age 4 children have already acquired much of the essential grammar of their native language. Four-year-olds can’t name or describe the rules of grammar (nor can most adults), yet they use them every day. Indeed, even professional linguists have yet to describe the full grammar of any of the world’s languages (Jackendoff, 2003). Grammatical rules in this sense are like the rules that underlie the sequence and timing of specific muscle movements in walking or running; both sets of rules are generally encoded in implicit rather than explicit memory. We generally can’t state them, but we use them when we walk, run, or carry on a conversation.

People’s implicit knowledge of grammar is demonstrated in their ability to distinguish acceptable from unacceptable sentences. Nearly every English speaker can identify The mouse crawled under the cabinet as a grammatical sentence and The crawled cabinet mouse the under as nongrammatical, although few can explain exactly why. The ability to distinguish grammatical from nongrammatical sentences is not based simply on meaning. As the linguist Noam Chomsky (1957) pointed out, English speakers recognize Colorless green ideas sleep furiously as grammatically correct but absurd.

The Course of Language Development

In a remarkably short time, infants progress from cries, coos, and babbles to uttering words, sentences, and narratives, becoming “linguistic geniuses” over the course of just a few years. Children across the world, although learning different languages, achieve this impressive cognitive and communicative feat in very similar ways.

Early Perception of Speech Sounds

Infants seem to treat speech as something special as soon as they are born, and maybe even before (Werker & Gervain, 2013). In experiments in which new-borns, just 1 to 4 days old, could produce sounds by sucking on a nipple, the babies sucked more vigorously to produce the sound of a human voice than to produce any other sounds that were tested (Butterfield & Siperstein, 1974; Vouloumanos & Werker, 2007).

The ability of very young infants to hear the differences among speech phonemes has been demonstrated in many experiments. One technique is to allow an infant to suck on a pacifier that is wired to trigger the playing of a particular sound each time a sucking response occurs. When the baby becomes bored with a sound, as indicated by a reduced rate of sucking, the sound is changed (maybe from pa to ba). Typically, the rate of sucking increases immediately thereafter, which indicates that the infant hears the new sound as different from the previous one. Another method, which can be used with infants 5 months old and older, involves rewarding the baby with an interesting sight for turning his or her head when the sound changes. The baby soon learns to look to the right—where the interesting visual display appears—each time the sound changes. This response serves as an index that the baby distinguishes the new sound from the old one.

The results of such experiments suggest that babies younger than 6 months old hear the difference between any two sounds that are classed as different phonemes in any of the world’s languages (Saffran et al., 2006; Tsao et al., 2004). At about 6 months of age, however, two kinds of changes begin to occur in their ability to discriminate between similar speech sounds: They become better at discriminating between sounds that represent different phonemes in their native language, and they become worse at discriminating between sounds that are classed as the same phoneme in their native language (Kuhl et al., 2008). For example, infants growing up in English-speaking cultures gradually become better than they were before at distinguishing between the English /l/ and /r/, which are distinct phonemes in English but not in Japanese, and they gradually lose the ability to distinguish among the subtly different /t/ sounds that constitute different phonemes in Hindi but not in English (Kuhl et al., 2008; Werker & Tees, 1999). In contrast, Japanese infants gradually lose the ability to distinguish among /l/ and /r/ and Indian infants in Hindi-speaking homes become better at distinguishing among the /t/ sounds relevant to their language.

Cooing and Babbling

Beginning at birth, infants can cry and produce various other vocal signs of distress, but at about 2 months they begin to produce a new, more speech-like category of sounds called cooing, which consists of repeated drawn-out vowels (oooh-oooh, eeeh-eeeh). At about 6 months, cooing changes gradually to babbling, which consists of repeated consonant-and-vowel sounds such as paa-paa-paa or tooda-tooda (Masataka, 2003). Cooing and babbling occur most often when the infant is happy. They seem to be forms of vocal play that have evolved to help the infant exercise and refine the complex muscle movements needed to produce coherent speech.

Coos and the earliest babbles do not depend on the infant’s hearing spoken sounds. Deaf infants coo and begin to babble at about the same age and in the same manner as hearing infants (Lenneberg, 1969), and early babbles are as likely to contain foreign-language sounds as native-language sounds (Locke, 1983). By 8 months of age, however, hearing infants begin to babble in ways that mimic the rhythm and pitch patterns of the language they hear around them; the babbling of a French baby becomes recognizably French, and that of a British baby becomes recognizably British (de Boysson-Bardies, 1999). Beginning at about 10 months of age, hearing infants produce babbled sounds that increasingly resemble syllables and words of their native language (de Boysson-Bardies, 1999; Locke, 1983). Also at this age, deaf babies who are exposed to a sign language begin to babble with their hands—repeating over and over hand movements that are similar in form and rhythm to those of the language they see around them (Petitto et al., 2001). Eventually, recognizable words appear in the hearing infant’s vocal babbling and the deaf infant’s manual babbling.

Word Comprehension Precedes Word Production

During the babbling phase of life, before the first production of recognizable words, infants begin to show evidence that they understand some words and phrases that they hear regularly. In one experiment, 6-month-olds were shown side-by-side videos of their parents. When they heard the word “Mommy” or “Daddy,” they looked reliably more at the video of the named parent than at the unnamed parent (Tincoff & Juscyk, 1999). Other experiments have revealed that 9-month-olds can respond to a number of common words by looking at the appropriate object when it is named (Balaban & Waxman, 1997) and can follow simple verbal commands, such as “Get the ball” (Benedict, 1979). By the time that they say their first word, at about 10 to 12 months of age, infants may already know the meaning of dozens of words (Swingley, 2008).

Naming and Rapid Vocabulary Development

Babies’ first words are most often produced in a playful spirit. The child at first uses words to point things out, or simply to name them for fun, not generally to ask for them (Bloom & Lahey, 1978). For instance, my (Peter Gray’s) son used the word ba joyfully to name his bottle, after he received it, weeks before he began using it to ask for his bottle when he wanted it.

New words come slowly at first, but then, typically at about 15 to 20 months of age, the rate begins to accelerate. Between the ages of 2 and 17 years, the typical person learns about 60,000 words, an average of about 11 new words per day (Bloom, 2001). Relatively few of these are explicitly taught; most often, the child must infer the meaning of a new word from the context in which others use it. How do young children, early in the language-learning process, draw these inferences?

Most of the earliest words learned, in any language culture, are nouns that refer to categories of objects in the child’s environment (Golinkoff et al., 1994). Young children’s tendency to look at whatever an older person is looking at (shared attention, discussed earlier) no doubt helps them identify objects that the older person is referring to when speaking. Infants are especially likely to follow an adult’s gaze when the adult is labeling an object in the environment (Baldwin, 2000). Researchers have found that those infants who show the most reliable gaze following, when tested at 10 and 11 months of age, show the greatest gains in vocabulary over the next several months (Brooks & Meltzoff, 2008).

In addition, young children seem to have a number of cognitive biases, or built-in assumptions, that help them narrow down the likely referent to a new word they hear (Golinkoff et al., 1994; Regier, 2003). One of these is a strong tendency to link new words with objects for which they do not already know a name. Other things being equal, young children assume that a new word is not a synonym for a word they already know but a label for something whose name they don’t know. In one experiment, 3- and 4-year-olds were presented with toy animals that they could name (a pig, a sheep, and a cow) plus one that they could not name (a tapir). When they heard the novel word gombe in the presence of these objects, all the children applied it to the novel animal (Clark, 1987). Other research indicates that toddlers begin to manifest this bias at about the same time—in their second year of life—at which their rate of vocabulary learning begins to increase rapidly (Mervis & Bertrand, 1994). Although the bias leads to some mistakes, it is apparently more helpful than harmful to their acquisition of words.

By the time they can understand multiword sentences, young children are able to use their tacit knowledge of grammar to help them infer the meaning of new words, including verbs and other parts of speech as well as nouns (Bernal et al., 2007; Fisher, 2000). Thus, if they are shown a videotaped scene and told, “The duck and bunny are biffing,” 2-year-olds—who understand duck and bunny and know implicitly that words ending in -ing refer to actions—infer that biffing means whatever the duck and bunny are both doing (Naigles, 1990). If told, instead, “The duck is biffing the bunny,” they infer that biffing means whatever the duck is doing to the bunny. As another example, 2-year-olds who heard “Mommy feeds the ferret” inferred that a ferret is an animal, not an inanimate object (Goodman et al., 1998).

Extending Words to Fit Appropriate Categories

In addition to linking a new word to its immediate referent, children must learn how to extend it to new referents. Common nouns such as ball refer to categories of objects, and a full understanding is demonstrated when a child applies the word to all members of the category and not to nonmembers. Researchers have found that young children, including even infants as young as 12 months, behave as though they assume that a newly heard label applies not just to the specific object that has been labeled but also to other objects that are perceptually like the original one (Golinkoff et al., 1995; Waxman & Markow, 1995). That is, infants are biased toward assuming that labels are common nouns, not proper nouns. The bias leads to some mistakes, as when a child refers to all men as Daddy, but is useful overall because the vast majority of nouns to be learned are common nouns.

By the time they are 2 years old, children can use the grammatical context of a sentence to discern whether a name for an object is a proper noun or a common noun. In one experiment, 24-month-old children were shown a stuffed animal and were told either “This is a ZAV” or “This is ZAV.” Those in the first group subsequently applied the label to other stuffed animals that looked like the original one, and those in the second group did not (Hall et al., 2001).

Children sometimes overextend common nouns, using them more broadly than adult usage would allow. On the basis of an analysis of a large collection of examples, Eve Clark (1973) proposed that overextension results when a child implicitly defines a new word in terms of just one or a few of the prominent features of the original referent object. Thus a child who hears ball in association with a specific ball might take the object’s most prominent feature—its roundness—as the defining characteristic and subsequently refer to all round objects, including oranges and full moons, as balls. In other cases, overextensions may not represent errors at all but simply derive from children’s attempts to communicate about objects that they have not yet learned to name (Clark, 1995). A toddler who says, “Look doggie,” while pointing to a cat, may in essence be saying, “Look at that thing that is something like a doggie.” The same child, when asked which animals in a set of pictures are dogs, may pick the actual dogs and not the cats (Thomson & Chapman, 1977).

The remarkable fact, however, is not that children make errors in their extensions but that they usually extend new words appropriately (Bloom, 2001). Somehow they figure out quickly the categories that are referred to by the words they hear adults and older children using.

Using Grammatical Rules

All children go through a relatively prolonged period during which each of their utterances is only one word long. When they do begin to put words together, typically at about 18 to 24 months of age, they at first use content words almost exclusively, especially nouns and verbs, and usually arrange them in the grammatically correct sequence for simple, active sentences (Brown, 1973). For an English-speaking child, this means that subjects are placed before verbs, and verbs before objects. A child at the two-word stage will say “Billy kick” to mean that Billy is kicking something, and “Kick Billy” to mean that someone is kicking Billy.

When children acquire a new grammatical rule, such as adding -ed to the end of a verb to create the past tense, they almost invariably overgeneralize it at first (Kuczaj, 1977; Marcus et al., 1992). The 3-year-old who says “kicked,” “played,” and “laughed” also says “goed,” “thinked,” and “swimmed.” Similarly, children who have just learned to add -s to pluralize nouns will talk about many mouses, sheeps, and childs. This overgeneralization, called overregulorization, confirms that children really know the rule. If they followed the rule only when adults did, their usage might be attributed to simple imitation. As further evidence suggesting that their grammar is based on rules, young children have been shown to use the rules with made-up words that they had never heard before, as illustrated in Figure 11.14.

Children are not taught the rules of grammar explicitly; nobody sits a 2-year-old down and tries to explain how to create infinitives, possessives, or past-tense verbs. Some parents correct their children’s grammar, but even this is rare (Brown & Hanlon, 1970), and long-term experiments in preschools have shown that deliberate programs of correcting grammar have little effect on rule acquisition (de Villiers & de Villiers, 1979). Through their own devices, children actively (and mostly unconsciously) infer grammatical rules from examples of rule-based language spoken around them and to them.