Most social species have systems of communication that allow them to transmit messages to each other. Honeybees communicate the location of food sources by means of a “waggle dance” that indicates both the direction and distance of the food source from the hive (Kirchner & Towne, 1994; Von Frisch, 1974). Vervet monkeys have three different warning calls that uniquely signal the presence of their main predators: a leopard, an eagle, and a snake (Cheney & Seyfarth, 1990). A leopard call provokes them to climb higher into a tree; an eagle call makes them look up into the sky. Each different warning call conveys a particular meaning and functions like a word in a simple language.

Language is a system for communicating with others using signals that are combined according to rules of grammar and convey meaning. Grammar is a set of rules that specify how the units of language can be combined to produce meaningful messages. Language allows individuals to exchange information about the world, coordinate group action, and form strong social bonds.

Human language may have evolved from signalling systems used by other species. However, three striking differences distinguish human language from vervet monkey yelps, for example. First, the complex structure of human language distinguishes it from simpler signalling systems. Most humans can express a wide range of ideas and concepts as well as generate an essentially infinite number of novel sentences. Second, humans use words to refer to intangible things, such as unicorn or democracy. These words could not have originated as simple alarm calls. Third, we use language to name, categorize, and describe things to ourselves when we think, which influences how knowledge is organized in our brains. It is doubtful that honeybees consciously think, “I will fly north today to find more honey so the queen will be impressed!”

In this section, we will examine the elements of human language that contribute to its complex structure, the ease with which we acquire language despite this complexity, and how both biological and environmental influences shape language acquisition and use. We will also look at startling disorders that reveal how language is organized in the brain and at researchers’ attempts to teach apes human language. Finally, we will consider the long-standing puzzle of how language and thought are related.

353

Compared with other forms of communication, human language is a relatively recent evolutionary phenomenon, emerging as a spoken system no more than 1 to 3 million years ago and as a written system as little as 6000 years ago. There are approximately 4000 human languages, which linguists have grouped into about 50 language families (Nadasdy, 1995). Despite their differences, all of these languages share a basic structure involving a set of sounds and rules for combining those sounds to produce meanings.

The smallest units of sound that are recognizable as speech rather than as random noise are phonemes. These building blocks of spoken language differ in how they are produced. For example, when you say ba, your vocal cords start to vibrate as soon as you begin the sound, but when you say pa, there is a 60-millisecond lag between the time you start the p sound and the time your vocal cords start to vibrate. B and p are classified as separate phonemes in English because they differ in the way they are produced by the human speaker.

Every language has phonological rules that indicate how phonemes can be combined to produce speech sounds. For example, the initial sound ts is acceptable in German but not in English. Typically, people learn these phonological rules without instruction, and if the rules are violated, the resulting speech sounds so odd that we describe it as speaking with an accent.

Phonemes are combined to make morphemes, the smallest meaningful units of language (see FIGURE 9.1). For example, your brain recognizes the p sound you make at the beginning of pat as a speech sound, but it carries no particular meaning. The morpheme pat, on the other hand, is recognized as an element of speech that carries meaning.

All languages have grammar rules that generally fall into two categories: rules of morphology and rules of syntax. Morphological rules indicate how morphemes can be combined to form words. Two classes of morphemes—content morphemes and function morphemes—can be distinguished. Content morphemes refer to things and events (e.g., cat, dog, take) or are affixes that, when added to a base word, change its meaning or part of speech. Examples of affixes that are content morphemes include re-, as in replay and redo, since re- changes the meaning of the base word (e.g., play, do). Similarly, adding the affix -able, as in playable and doable, changes the meaning of the base word (changing them from verbs to adjectives). Function morphemes serve grammatical functions, such as tying sentences together (and, or, but) or indicating time (when). Function morphemes also include affixes that indicate the grammatical role of a base word, such as those required for plurals and for verb agreement. For example, the -s on the end of cats and the -ed or -ing on the end of walked or walking are function morphemes. It is the function morphemes that make human language grammatically complex enough to permit us to express abstract ideas rather than simply to point verbally to real objects in the here and now.

354

Content and function morphemes can be combined and recombined to form an infinite number of new sentences, which are governed by syntax. Syntactical rules indicate how words can be combined to form phrases and sentences. A simple syntactical rule in English is that every sentence must contain one or more nouns, which may be combined with adjectives or articles to create noun phrases (see FIGURE 9.2). A sentence also must contain one or more verbs, which may be combined with adverbs or articles to create verb phrases. So, the utterance “dogs bark” is a full sentence, but “the big grey dog over by the building” is not.

Sounds and rules are critical ingredients of human language that allow us to convey meaning. A sentence can be constructed in a way that obeys syntactical and other rules, yet be entirely lacking in meaning or semantics, as in the famous example provided by linguist Noam Chomsky (1957, p. 15), “Colourless green ideas sleep furiously.” Though we would not be breaking any grammatical rules by uttering such a sentence, we could expect to elicit head scratches and strange looks from any nearby listeners. Language usually conveys meaning quite well, but everyday experience shows us that misunderstandings can occur. These errors sometimes result from differences between the deep structure of sentences and their surface structure (Chomsky, 1957). Deep structure refers to the meaning of a sentence. Surface structure refers to how a sentence is worded. “The dog chased the cat” and “The cat was chased by the dog” mean the same thing (they have the same deep structure) even though on the surface their structures are different.

355

To generate a sentence, you begin with a deep structure (the meaning of the sentence) and create a surface structure (the particular words) to convey that meaning. When you comprehend a sentence, you do the reverse, processing the surface structure in order to extract the deep structure. After the deep structure is extracted, the surface structure is usually forgotten (Jarvella, 1970, 1971). In one study, researchers played tape-recorded stories to volunteers and then asked them to pick the sentences they had heard (Sachs, 1967). Participants frequently confused sentences they heard with sentences that had the same deep structure but a different surface structure. For example, if they heard the sentence “He struck John on the shoulder,” they often mistakenly claimed they had heard “John was struck on the shoulder by him.” In contrast, they rarely misidentified “John struck him on the shoulder” because this sentence has a different deep structure from the original sentence.

Language is a complex cognitive skill, yet we can carry on complex conversations with playmates and family before we begin school. Three characteristics of language development are worth bearing in mind. First, children learn language at an astonishingly rapid rate. The average 1-year-old has a vocabulary of 10 words. This tiny vocabulary expands to over 10 000 words in the next 4 years, requiring the child to learn, on average, about 6 or 7 new words every day. Second, children make few errors while learning to speak and, as we will see shortly, the errors they do make usually result from applying, but overgeneralizing, grammatical rules they have learned. This is an extraordinary feat. There are over 3 million ways to rearrange the words in any 10-word sentence, but only a few of these arrangements will be both grammatically correct and meaningful (Bickerton, 1990). Third, children’s passive mastery of language develops faster than their active mastery. At every stage of language development, children understand language better than they speak.

At birth, infants can distinguish among all of the contrasting sounds that occur in all human languages. Within the first 6 months of life, they lose this ability, and, like their parents, can only distinguish among the contrasting sounds in the language they hear being spoken around them. For example, two distinct sounds in English are the l sound and the r sound, as in lead and read. These sounds are not distinguished in Japanese; instead, the l and r sounds fall within the same phoneme. Japanese adults cannot hear the difference between these two sounds, but American adults can distinguish between them easily—and so can Japanese infants under 6 months or so of age.

In one study of 1 to 4-month-old infants, researchers constructed a tape of a voice saying “la-la-la” or “ra-ra-ra” repeatedly (Eimas et al., 1971). They rigged a pacifier so that whenever an infant sucked on it, a tape player that broadcast the la-la tape was activated. When the la-la sound began playing in response to their sucking, the infants were delighted and kept sucking on the pacifier to keep the la-la sound playing. After a while, they began to lose interest, and sucking frequency declined to about half of its initial rate. At this point, the experimenters switched the tape so that ra-ra was repeatedly broadcast. The Japanese infants began sucking again with vigour, indicating that they could hear the difference between the old, boring la sound and the new, interesting ra sound.

356

Infants can distinguish among speech sounds, but they cannot produce them reliably, relying mostly on cooing (i.e., simple vowel-like sounds, such as ah-ah), cries, laughs, and other vocalizations to communicate. Between the ages of about 4 and 6 months, they begin to babble speech sounds. Babbling involves combinations of vowels and consonants that sound like real syllables but are meaningless. Regardless of the language they hear spoken, all infants go through the same babbling sequence. For example, d and t appear in infant babbling before m and n. Even deaf infants babble sounds they have never heard, and they do so in the same order as hearing infants do (Ollers & Eilers, 1988). This is evidence that infants are not simply imitating the sounds they hear and suggests that babbling is a natural part of the language development process. Recent research has shown that babbling serves as a signal that the infant is in a state of focused attention and ready to learn (Goldstein et al., 2010). Deaf infants do not babble as much, however, and their babbling is delayed relative to hearing infants (11 months rather than 6 months).

In order for vocal babbling to continue, however, infants must be able to hear themselves. In fact, delayed babbling or the cessation of babbling merits testing for possible hearing difficulties. Babbling problems can lead to speech impairments, but they do not necessarily prevent language acquisition. Deaf infants whose parents communicate using American Sign Language (ASL) begin to babble with their hands at the same age that hearing children begin to babble vocally—between 4 and 6 months (Petitto & Marentette, 1991). Their babbling consists of sign language syllables that are the fundamental components of ASL.

At about 10 to 12 months of age, infants begin to utter (or sign) their first words. By 18 months, they can say about 50 words and can understand several times more than that. Toddlers generally learn nouns before verbs, and the nouns they learn first are names for everyday, concrete objects (e.g., chair, table, milk) (see TABLE 9.1 below). At about this time, their vocabularies undergo explosive growth. By the time the average child begins school, a vocabulary of 10 000 words is not unusual. By fifth grade, the average child knows the meanings of 40 000 words. By university, the average student’s vocabulary is about 200 000 words. Fast mapping, in which children map a word onto an underlying concept after only a single exposure, enables them to learn at this rapid pace (Kan & Kohnert, 2008; Mervis & Bertrand, 1994). This astonishingly easy process contrasts dramatically with the effort required later to learn other concepts and skills, such as arithmetic or writing.

Around 24 months, children begin to form two-word sentences, such as “more milk” or “throw ball.” Such sentences are referred to as telegraphic speech because they are devoid of function morphemes and consist mostly of content words. Yet despite the absence of function words, such as prepositions or articles, these two-word sentences tend to be grammatical; the words are ordered in a manner consistent with the syntactical rules of the language children are learning to speak. So, for example, toddlers will say “throw ball” rather than “ball throw” when they want you to throw the ball to them, and “more milk” rather than “milk more” when they want you to give them more milk. With these seemingly primitive expressions, 2-year-olds show that they have already acquired an appreciation of the syntactical rules of the language they are learning.

357

Evidence of the ease with which children acquire grammatical rules comes from some interesting errors that children make while forming sentences. If you listen to average 2- or 3-year-old children speaking, you may notice that they use the correct past-tense versions of common verbs, as in the expressions “I ran” and “you ate.” By the age of 4 or 5, the same children will be using incorrect forms of these verbs, saying such things as “I runned” or “you eated,” forms most children are unlikely to have ever heard (Prasada & Pinker, 1993). The reason is that very young children memorize the particular sounds (i.e., words) that express what they want to communicate. But as children acquire the grammatical rules of their language, they tend to overgeneralize. For example, if a child overgeneralizes the rule that past tense is indicated by -ed, then run becomes runned or even ranned instead of ran.

These errors show that language acquisition is not simply a matter of imitating adult speech. Instead, children acquire grammatical rules by listening to the speech around them and using the rules to create verbal forms they have never heard. They manage this without explicit awareness of the grammatical rules they have learned. In fact, few children or adults can articulate these rules of their native language, yet the speech they produce obeys these rules.

By about 3 years of age, children begin to generate complete simple sentences that include function words (e.g., “Give me the ball” and “That belongs to me”). The sentences increase in complexity over the next 2 years. By 4 to 5 years of age, many aspects of the language acquisition process are complete. As children continue to mature, their language skills become more refined, with added appreciation of subtler communicative uses of language, such as humour, sarcasm, or irony.

358

Language development typically unfolds as a sequence of steps in which one milestone is achieved before moving on to the next. Nearly all infants begin with one-word utterances before moving on to telegraphic speech and then to simple sentences that include function morphemes. It is hard to find solid evidence of infants launching immediately into speaking in sentences—even though you may occasionally hear reports of such feats from proud parents, including possibly your own! This orderly progression could result from general cognitive development that is unrelated to experience with a specific language (Shore, 1986; Wexler, 1999). For example, perhaps infants begin with one- and then two-word utterances because their short-term memories are so limited that initially they can only hold in mind a word or two; additional cognitive development might be necessary before they have the capacity to put together a simple sentence. By contrast, the orderly progression might depend on experience with a specific language, reflecting a child’s emerging knowledge of that language (Bates & Goodman, 1997; Gillette et al., 1999).

These two possibilities are difficult to tease apart, but recent research has begun to do so using a novel strategy: examining the acquisition of English by internationally adopted children who did not know any English prior to adoption (Snedeker, Geren, & Shafto, 2007, 2012). According to government statistics, there were just over 8600 international adoptions to the United States in 2012 (U.S. Department of State, 2013). Although most of those adoptees are infants or toddlers, a significant proportion is composed of preschoolers. Studying the acquisition of English in such an older population provides a unique opportunity to explore the relationship between language development and cognitive development. If the orderly sequence of milestones that characterizes the acquisition of English by infants is a by-product of general cognitive development, then different patterns should be observed in older internationally adopted children, who are more advanced cognitively than infants. However, if the milestones of language development are critically dependent on experience with a specific language—English—then language learning in older adopted children should show the same orderly progression as seen in infants.

Snedeker and her colleagues (2007) examined preschoolers ranging from 2½ to 5½ years old, 3 to 18 months after they were adopted from China. They did so by mailing materials to parents, who periodically recorded language samples in their homes and also completed questionnaires concerning specific features of language observed in their children. These data were compared to similar data obtained from monolingual infants. The main result was clear-cut: Language acquisition in preschool-aged adopted children showed the same orderly progression of milestones that characterizes infants. These children began with one-word utterances before moving on to simple word combinations. Furthermore, their vocabulary, just like that of infants, was initially dominated by nouns and they produced few function morphemes.

These results indicate that some of the key milestones of language development depend on experience with English. However, the adopted children did add new words to their vocabularies more quickly than infants did, perhaps reflecting an influence of general cognitive development. Overall, though, the main message from this study is that observed shifts in early language development reflect specific characteristics of language learning rather than general limitations of cognitive development. A later study provided additional support for this general conclusion, but also produced new evidence for a role of cognitive development in specific aspects of language (Snedeker et al., 2012). For example, adopted preschoolers acquire words that refer to the past or the future, such as tomorrow, yesterday, before, or after, much more quickly than do infants, perhaps reflecting that infants have difficulty representing these abstract concepts and therefore take longer to learn the words than more cognitively sophisticated preschoolers.

359

We know a good deal about how language develops, but what underlies the process? The language acquisition process has been the subject of considerable controversy and (at times) angry exchanges among scientists coming from three different approaches: behaviourist, nativist, and interactionist.

According to B. F. Skinner’s behaviourist explanation of language learning, we learn to talk in the same way we learn any other skill: through reinforcement, shaping, extinction, and the other basic principles of operant conditioning that you learned about in the Learning chapter (Skinner, 1957). As infants mature, they begin to vocalize. Those vocalizations that are not reinforced gradually diminish, and those that are reinforced remain in the developing child’s repertoire. So, for example, when an infant gurgles “prah,” most parents are pretty indifferent. However, a sound that even remotely resembles “da-da” is likely to be reinforced with smiles, whoops, and cackles of “Goooood baaaaaby!” by doting parents. Maturing children also imitate the speech patterns they hear. Then parents or other adults shape those speech patterns by reinforcing those that are grammatical and ignoring or punishing those that are ungrammatical. “I no want milk” is likely to be squelched by parental clucks and titters, whereas “No milk for me, thanks” will probably be reinforced.

The behavioural explanation is attractive because it offers a simple account of language development, but the theory cannot account for many fundamental characteristics of language development (Chomsky, 1986; Pinker, 1994; Pinker & Bloom, 1990).

360

The study of language and cognition underwent an enormous change in the 1950s, when linguist Noam Chomsky (1957, 1959) published a blistering reply to the behaviourist approach. According to Chomsky, language-learning capacities are built into the brain, which is specialized to acquire language rapidly through simple exposure to speech. Chomsky and others have argued that humans have a particular ability for language that is separate from general intelligence. This nativist theory holds that language development is best explained as an innate, biological capacity. According to Chomsky, the human brain is equipped with a language acquisition device (LAD), a collection of processes that facilitate language learning. Language processes naturally emerge as the infant matures, provided the infant receives adequate input to maintain the acquisition process.

Christopher’s story is consistent with the nativist view of language development: His genius for language acquisition, despite his low overall intelligence, indicates that language capacity can be distinct from other mental capacities. Other individuals show the opposite pattern: People with normal or nearly normal intelligence can find certain aspects of human language difficult or impossible to learn. This condition is known as genetic dysphasia, a syndrome characterized by an inability to learn the grammatical structure of language despite having otherwise normal intelligence. Genetic dysphasia tends to run in families, and a single dominant gene has been implicated in its transmission (Gontier, 2008; Gopnik, 1990a, 1990b; Vargha-Khadem et al., 2005). Consider some sentences generated by children with the disorder:

She remembered when she hurts herself the other day.

Carol is cry in the church.

Notice that the ideas these children are trying to communicate are intelligent. Their problems with grammatical rules persist even if they receive special language training. When asked to describe what she did over the weekend, one child wrote, “On Saturday I watch TV.” Her teacher corrected the sentence to “On Saturday, I watched TV,” drawing attention to the -ed rule for describing past events. The following week, the child was asked to write another account of what she did over the weekend. She wrote, “On Saturday I wash myself and I watched TV and I went to bed.” Notice that although she had memorized the past-tense forms watched and went, she could not generalize the rule to form the past tense of another word (washed).

As predicted by the nativist view, studies of people with genetic dysphasia suggest that normal children learn the grammatical rules of human language with ease in part because they are “wired” to do so. This biological predisposition to acquire language explains why newborn infants can make contrasts among phonemes that occur in all human languages—even phonemes they have never heard spoken. If we learned language through imitation, as behaviourists theorized, infants would only distinguish the phonemes they would actually heard. The nativist theory also explains why deaf infants babble speech sounds they have never heard and why the pattern of language development is similar in children throughout the world. These characteristics of language development are just what would be expected if our biological heritage provided us with the broad mechanics of human language.

Also consistent with the nativist view is evidence that language can be acquired only during a restricted period of development, as has been observed with songbirds. If young songbirds are prevented from hearing adult birds sing during a particular period in their early lives, they do not learn to sing. A similar mechanism seems to affect human language learning, as illustrated by the tragic case of Genie (Curtiss, 1977). At the age of 20 months, Genie was tied to a chair by her parents and kept in virtual isolation. Her father forbade Genie’s mother and brother to speak to her, and he himself only growled and barked at her. She remained in this brutal state until the age of 13. Genie’s life improved substantially and she received years of language instruction, but it was too late. Her language skills remained extremely primitive. She developed a basic vocabulary and could communicate her ideas, but she could not grasp the grammatical rules of English.

361

Similar cases have been reported, with a common theme: Once puberty is reached, acquiring language becomes extremely difficult (Brown, 1958). Data from studies of language acquisition in immigrants support this conclusion. In one American study, researchers found that the proficiency with which immigrants spoke English depended not on how long they would lived in the United States, but on their age at immigration (Johnson & Newport, 1989). Those who arrived as children were the most proficient, whereas those who immigrated after puberty, showed a significant decline in proficiency regardless of the number of years in their new country. More recent work using fMRI shows that acquiring a second language early in childhood (between 1 and 5 years of age) results in very different representation of that language in the brain than does acquiring that language much later (after 9 years of age) (Bloch et al., 2009).

Nativist theories are often criticized because they do not explain how language develops; they merely explain why. A complete theory of language acquisition requires an explanation of the processes by which the innate, biological capacity for language combines with environmental experience. The interactionist approach is that although infants are born with an innate ability to acquire language, social interactions play a crucial role in language. Interactionists point out that parents tailor their verbal interactions with children in ways that simplify the language acquisition process: They speak slowly, enunciate clearly, and use simpler sentences than they do when speaking with adults (Bruner, 1983; Farrar, 1990).

Further evidence of the interaction of biology and experience comes from a fascinating study of deaf children’s creation of a new language (Senghas, Kita, & Ozyurek, 2004). Prior to about 1980, deaf children in Nicaragua stayed at home and usually had little contact with other deaf individuals. In 1981, some deaf children began to attend a new vocational school. At first, the school did not teach a formal sign language, and none of the children had learned to sign at home, but the children gradually began to communicate using hand signals that they invented.

Over the past 30 years, their sign language has developed considerably (Pyers et al., 2010), and researchers have studied this new language for the telltale characteristics of languages that have evolved over much longer periods. For instance, mature languages typically break down experience into separate components. When we describe something in motion, such as a rock rolling down a hill, our language separates the type of movement (rolling) and the direction of movement (down). If we simply made a gesture, however, we would use a single continuous downward movement to indicate this motion. This is exactly what the first children to develop the Nicaraguan sign language did. But younger groups of children, who have developed the sign language further, use separate signs to describe the direction and the type of movement—a defining characteristic of mature languages. That the younger children did not merely copy the signs from the older users suggests that a predisposition exists to use language to dissect our experiences. Thus, their acts of creation nicely illustrate the interplay of nativism (the predisposition to use language) and experience (growing up in an insulated deaf culture).

362