Development - Write My Paper Today

Citation: Hoff, Erika. Language Development. Available from: MBS Direct, (5th Edition). Cengage Learning US, 2013https://mbsdirect.vitalsource.com/reader/books/9781285632896/pageid/48

Hoff (2014): Ch 1 (pp. 3-29 );

Chapter 1

Introduction to the Study of Language Development

Language and the Scientific Study of Language Development • A Definition of Language • A Chronological Overview of Language

Development • Reasons for the Scientific Study of Language

Development

The History of the Study of Language Development • Big Questions and Studies of Special Cases • Baby Biographies • Normative Studies • The Chomskyan Revolution • The Current Study of Language Development Major Issues in the Field of Language Development • What Are the Contributions of Nature and

Nurture to Language Acquisition? • Are the Mechanisms of Language Acquisition

Language-Specific or Domain General? • How Abstract Is Language?

Is There Continuity or Discontinuity in Language Development?
What Is the Relation Between Communication and Language?

Theories of Language Development Methods of Research in Language Development • Cross-Cultural and Cross-Linguistic Research • Research Designs and Procedures • Assessment of Productive Language from

Speech Samples • CHILDES—A Data Archive • Standardized Tests and Measures of Language

Development • Computational Modeling Sources for Research on Language Development • Journals • Indexes Summary Key Terms Review Questions

Somehow, in the span of just a few years, newborn infants who neither speak nor understand any language become young children who comment, question, and express their ideas in the language of their community. This change does not occur all at once. First, newborns’ cries give way to coos and babbles. Then, infants who coo and babble start to show signs of comprehension such as turning when they hear their name. Infants then become toddlers who say “bye-bye” and “all gone” and start to label the people and objects in their environment. As their vocabularies continue to grow, children start to combine words. Children’s first word combinations, such as all gone juice and read me, are short and are missing parts found in adults’ sentences. Gradually, children’s immature sentences are replaced by longer and more adultlike sentences. As children learn to talk, their comprehension abilities also develop, typically in advance of their productive speech. As children master language, they also become masters at using language to serve their needs. One-year-olds who can only point and fuss to request something become 2-year-olds who say “please”; later, they become 4-year-olds capable of the linguistic and communicative sophistication of the child who excused himself from a boring experiment by saying, “My mother says I have to go home now” (D. Keller- Cohen, January 1978, personal communication

This book is about these changes. It is about the what and when of language development—what changes take place and when they occur in the course of language development. It is also about the how and why. How do children learn to talk, and why is the development of language a universal feature of human development? In the following chapters, we will delve into these topics in detail. In this chapter, we begin with an overview of the field we are about to study.

Language and the Scientific Study of Language Development A Definition of Language

Language is the systematic and conventional use of sounds (or signs or written symbols) for the purpose of communication or self-expression (Crystal, 1995). This definition is short and simple, and, although true, it is misleading in its simplicity. Language is complex and multifaceted. The child who learns a language achieves the ability to recognize and produce a set of sounds and learns how these sounds can and cannot be combined into possible words. The child who learns English, for example, comes to know approximately 44 different consonants and vowels (Crystal, 1995) and that pling is a possible word but gnilp is not. By adulthood, the child who learns a language knows a vocabulary of tens of thousands of words. This vocabulary knowledge includes knowledge of each word’s mean- ing and its possibilities for combination with other words. Adult speakers of English know, for example, that give and donate are synonyms, that John gave a book to the library and John donated a book to the library are perfectly fine sentences, that John gave the library a book is also fine, but that John donated the library a book is not. The child who learns a language also comes to know the multiple ways in which pieces of the language can and cannot be systematically combined to form words and sentences. John kissed Mary and Mary kissed John are both fine sentences, albeit with different meanings; kissed is made up of kiss þ ed, and Mary þ ed John kiss just does not work. The child who learns a lan- guage also comes to know how to combine sentences into larger units of discourse—to tell a story or have a conversation. As they learn a language, children learn to use that lan- guage to communicate in socially appropriate ways. They acquire the means to share their thoughts and feelings with others and the skill to do so differently with their peers and their grandparents. In a literate society, children also learn to use language in its writ- ten form. They master both a complex set of correspondences between written symbols and meanings and a literate style of language use. Many children, perhaps most of the world’s children, hear and acquire more than one language (e.g., Grosjean, 2010), and there is no reason to think that monolingual development is more basic or natural for chil- dren than bilingual or multilingual development. One could argue that a text on language development should treat multilingualism as the norm and have one chapter on the special case of monolingual development. The history of the field, however, is that most of the research on language development has been conducted with children exposed to only one language. Studies of bilingual and multilingual development are fewer, although this is a rapidly growing research area. The organization of this text reflects the scientific literature in taking monolingual development as its focus and presenting research on bilingual devel- opment in a single chapter, Chapter 9.

Children develop knowledge in the different domains of language concurrently, and there are many ways in which knowledge in one domain is used in acquiring knowledge in another. It is useful, nonetheless, for researchers and for students of language develop- ment to make distinctions among the subcomponents of language. The sounds and sound system of a language constitute a language’s phonology. The words and associated

During children’s second year, the most obvious development is in the domain of vocab- ulary. Children typically begin this year by producing their first word, and by the end of the year, they have a productive vocabulary of about 300 words and are producing word combinations (Fenson, Dale, Reznick, & Bates, 1994). Their words do not sound quite adultlike. Both articulation abilities and underlying phonological representations undergo changes during this second year. Children are also becoming more communicative. Both the frequency and the conversational relevance of their communicative acts increase.Dring the third year of life, the most obvious development is children’s increasing mastery of the grammar of their language. Typically, children start this year producing two- and three-word affirmative, declarative sentences that lack grammatical endings (e.g., plural markers and past-tense markers) on nouns and verbs. By the end of the third year, children produce full sentences, including questions and negated forms with most grammatical devices in place. Vocabulary continues to grow, articulation of sounds improves, and children begin to develop an awareness of the phonological properties of their language—as evidenced, for example, in their appreciation of rhymes. Children’s conversational skills increase, and they begin to introduce short accounts of past events into their conversations.

The period from 3 to 4 years is largely one of refining and further developing the skills that are already in place. The most obvious new development occurs in the area of grammar, where children start to produce complex, multiclause sentences. Because there is nothing completely missing from the linguistic competence of most 4-year-old children, it is commonly said that language acquisition is completed during the first four years of life. Although there is some truth to that statement, language skills continue to grow in every domain after the age of 4 years. Articulation, vocabulary, sentence structure, and communicative skills all develop. There are also major transitions involved as children move from a home to a school environment and learn new ways of using language; literacy development is further associated with changes in language knowledge. We will return to each of these developments in future chapters.Reasons for the Scientific Study of Language Development

Language Development as a Basic Research Topic A child who has acquired language has acquired an incredibly complex and powerful system. If we understood how children accomplish this task, we would know something substantial about how the human mind works. The modern field of language development emerged in the 1950s when it became clear that language acquisition would serve as a test for rival theories of how change in human behavior occurs (H. Gardner, 1985; Pinker, 1984). In the 1950s, two psychological theories were pitted against each other: behaviorism and cognitivism.

Cognitivism asserts the opposite—that we cannot understand behavior without understanding what is going on inside the mind of the organism producing the behavior. From approximately 1930 to the early 1950s, behaviorism dominated American psychol- ogy. But in the 1950s, a “cognitive revolution” began (H. Gardner, 1985). During the next two decades, behaviorism came to be seen as inadequate, and the focus of the

uring children’s second year, the most obvious development is in the domain of vocab- ulary. Children typically begin this year by producing their first word, and by the end of the year, they have a productive vocabulary of about 300 words and are producing word combinations (Fenson, Dale, Reznick, & Bates, 1994). Their words do not sound quite adultlike. Both articulation abilities and underlying phonological representations undergo changes during this second year. Children are also becoming more communicative. Both the frequency and the conversational relevance of their communicative acts increase.

During the third year of life, the most obvious development is children’s increasing mastery of the grammar of their language. Typically, children start this year producing two- and three-word affirmative, declarative sentences that lack grammatical endings (e.g., plural markers and past-tense markers) on nouns and verbs. By the end of the third year, children produce full sentences, including questions and negated forms with most grammatical devices in place. Vocabulary continues to grow, articulation of sounds improves, and children begin to develop an awareness of the phonological properties of their language—as evidenced, for example, in their appreciation of rhymes. Children’s conversational skills increase, and they begin to introduce short accounts of past events into their conversations.

Reasons for the Scientific Study of Language Development

Behaviorism holds that change in behavior occurs in response to the consequences of prior behavior. Most readers are familiar with clear examples supporting this view. For instance, rats that initially do not press levers come to press levers after receiving food pellets for producing behaviors that increasingly approximate lever pressing. Radical behaviorism holds that all behavior can be accounted for in this way. A central tenet of behaviorism is that it is not necessary to discern what goes on in the mind of the rat in order to explain the change in the rat’s behavior; behavior can be fully accounted for in terms of things external to the mind.Cognitivism asserts the opposite—that we cannot understand behavior without understanding what is going on inside the mind of the organism producing the behavior. From approximately 1930 to the early 1950s, behaviorism dominated American psychol- ogy. But in the 1950s, a “cognitive revolution” began (H. Gardner, 1985). During the next two decades, behaviorism came to be seen as inadequate, and the focus of the

search for explanations of human behavior shifted to internal mental processes. Studies of language played a crucial role in the cognitive revolution. The ability to speak and understand language is incredibly complex, and children acquire that ability without receiving positive reinforcement for successive approximations to grammatical sentences. Simple theories that may well explain why rats push levers, why dogs salivate at the sight of the people who feed them, and why humans get tense when they sit in the dentist’s chair cannot explain how children learn to talk. When cognitivism displaced behavior- ism, theoretical dispute concerning how to understand human behavior did not end. In fact, a new interdisciplinary field called cognitive science emerged from the cognitive revolution.

Cognitive scientists now agree that it is necessary to understand how the mind works in order to explain human behavior, but they do not agree on how the mind works. The study of language acquisition plays a central role in the debate over how to characterize human cognition, for the same reason that language acquisition played a central role in the cognitive revolution. That is, it is so difficult to explain how language acquisition is possible that accounting for language acquisition is a test not likely to be passed by inac- curate cognitive theories. Language acquisition is the New York City of the field of cog- nitive science: If you can make it there, you can make it anywhere.

Language Development as an Applied Research Topic The goal for many researchers who study language development is perhaps less grandiose than discovering how the mind works, but it is more immediate. Success in modern industrialized society depends on having good verbal skills, and acquiring the verbal skills that society requires is problematic for some children. For example, some minority children and some chil- dren from lower socioeconomic strata enter school with language skills that differ from those that mainstream, middle-class teachers expect. Many children enter school with limited skills in the language of instruction because they or their parents are immigrants, and the language they have learned at home is not the language used in school. A sub- stantial area of research conducted by developmental psychologists, speech and commu- nication scientists, and educators is aimed at understanding the nature of the language skills that characterize children from diverse backgrounds and identifying the best approaches to educating them.

For some children, acquiring adequate language skills is problematic because of other conditions, including intellectual disability, hearing impairment, or brain injury. Some children have difficulty acquiring language in the apparent absence of any other sort of impairment. A substantial body of research focuses on trying to understand the nature of the problems that underlie such children’s difficulty and on finding techniques for help- ing these children acquire language skills.

The areas of basic and applied research in the study of language development are not wholly separate. There are important points of contact. For example, basic research on the process of normal language development is used to develop interventions to help children who have difficulty acquiring language (S. F. Warren &Reichle, 1992), and research on the processes involved in reading has provided the basis for successful read- ing interventions (Bus & van Ijzendoorn, 1999; Ehri et al., 2001; Lyytinen, Erskine, Aro, & Richardson, 2007). Sometimes work on language disorders also informs basic research. For example, evidence that children with autism acquire language structure even though they have severe communicative deficiencies suggests that learning language involves more than learning how to fulfill a need to communicate (Tager-Flusberg, 1994, 2007), and studies that find late talkers differ from typically developing children in other cogni- tive tasks suggest that multiple skills serve normal language development (Rescorla, 2009). There are also important points of contact among the various disciplines that

study language development. For example, anthropologists’ descriptions of cultures in which no one talks to babies is relevant to the work of developmental psychologists who study how mother–infant interactions contribute to language development (Hoff, 2006b; Lieven, 1994).

The History of the Study of Language Development

Although the modern study of language acquisition began in the 1960s, the linguistic capacity of children has been a source of fascination since ancient times. One can find examples in history of many of the motives that prompt current investigations of children’s language.

Big Questions and Studies of Special Cases

The Language in the Brain The first recorded language acquisition experiment was conducted by the ancient Egyptian King Psammetichus and described by the Greek his- torian Herodotus in the 4th century BC. The issue at hand concerned who among the peoples of the world represented the original human race. To resolve the issue, King Psammetichus ordered that two infants be raised in isolation by shepherds, who were never to speak in the children’s presence. The idea behind this experiment was that the babies would start to speak on their own, and whatever language they spoke would be the language of the “original” people. According to Herodotus’s account, one of the chil- dren said something like “becos” at the age of 2. Becos, as it turned out, was the Phrygian word for bread. In the face of this evidence, King Psammetichus abandoned his claim that the Egyptians were the oldest race of humans and concluded that they were second oldest, after the Phrygians.

Although the assumptions underlying that experiment seem slightly comical now, and the method of the experiment is certainly unethical, the idea of asking about the lan- guage the brain creates when it is not given an existing language to learn has not been discarded. Susan Goldin-Meadow has studied the gestural communication systems invented by deaf children born to hearing parents (Feldman, Goldin-Meadow, &Gleitman, 1978; Goldin-Meadow, 2003; Goldin-Meadow, Mylander, & Franklin, 2007). Because the children’s parents do not know any sign language (and have been instructed not to learn or use any sign language in these cases, in accordance with the oralist method of instruction for the deaf), these deaf children are just as isolated from a lan- guage model as were the infants in King Psammetichus’s experiment. Children in these circumstances invent “signs” and combine them in two- and three-sign sequences, sug- gesting that putting symbols together to communicate is something that naturally emerges in the course of human development. In Chapter 11, we will come back to the specifics of these findings and what they suggest.

“Wild Children” and the Nature of Humankind Occasionally, there are children who are not only linguistic isolates but also social isolates, and these unfortunate chil- dren afford science the opportunity to ask an even broader question: What is the intrin- sic nature of humankind? This question was hotly debated in the 18th century. On the one hand, there had been a long tradition of argument by philosophers such as René Descartes (1662) that human nature (including having an immortal soul) was an innate endowment. On the other hand, philosopher John Locke (1690) argued that at birth the human mind was like a sheet of blank paper and that humans become what they become as a result of society’s influence. What was needed to settle this question was a human raised outside of society. Such a human appeared in the winter of 1800.

That winter was an unusually cold one, and in January, a young boy who had been living wild in the woods near Aveyron, France, approached a tanner’s workshop on the edge of the forest (Lane, 1976). The child appeared to be about 12 years old. He was naked; he occasion- ally ran on all fours; he ate roots, acorns, and raw vegetables—but only after sniffing them first; and although he was capable of making sounds, he had no language. This “wild child” became the object of intense scientific interest because he provided an opportunity to exam- ine the nature of the human species in its natural state. The young boy’s muteness was prob- lematic for theories of innate knowledge for two reasons: (1) Language was held to be one of the defining characteristics of humanity, and (2) his muteness made him a difficult subject to interview to determine whether he had an innate idea of God (Lane, 1976). However, the boy’s muteness provided good support for the opposing idea that “man depends on society for all that he is and can be” (Lane, 1976, p. 5).

Victor, the wild boy of Aveyron, as he came to be called, was placed with young Dr. Jean-Marc Itard for training at the National Institute for Deaf-Mutes in Paris. The scientific community watched to see whether society could provide this child with the human characteristic of language. Although Dr. Itard was able to teach the boy some socially appropriate behaviors, the boy never learned more than a few words. Although we cannot be certain why the efforts to teach him failed, many of this wild child’s beha- viors suggested that he was autistic (Wolff, 2004). Thus, his outcome does not tell us about normal human development in the absence of society’s influence. Itard’s work did yield practical dividends. He later used the training methods he had devised for the wild boy of Aveyron in teaching the deaf, and some of the techniques for teaching letters that Itard invented are used in Montessori classrooms today (Lane, 1976).

Over the course of history, there have been other “wild children” who were discovered mute at an age when children in normal environments have learned to talk (see R. Brown, 1958a; Curtiss, 1989; L. R. Gleitman& H. Gleitman, 1991). The most famous modern case is that of a girl named “Genie,” who became known to the public in 1970. She was 13 years old and had been kept locked in a room by her mentally ill father since the age of approximately 18 months. Her language remediation was somewhat more suc- cessful than the boy of Aveyron’s, but Genie never acquired normal language (Curtiss, 1977; Rymer, 1993). To some, such cases suggest that there may be a critical period for some aspects of language acquisition, such that language acquisition begun after child- hood is never quite as successful as language acquisition begun earlier. This is also a topic to which we will return in Chapters 2 and 9.

Baby Biographies

Another approach to investigating “the nature of humankind” is simply to observe what emerges in the course of normal development. In this vein, several investigators in the late 1800s and early 1900s kept diaries of their own children’s development. The most famous of these “baby biographers” was Charles Darwin (better known for his theory of evolution), whose description of his son’s communicative development (Darwin, 1877) looks remarkably like that described in Figure 1.1. Darwin’s son said da at 5 months, and, before he was 1 year old, the young Darwin understood intonations, gestures, several words, and short sentences. At 1 year, the child communicated with gestures and invented his first word, mum, to mean food. Other well-known diaries include Clara and Wilhelm Stern’s Die Kindersprache (Stern & Stern, 1907) and Werner Leopold’s (1939–1949) four- volume account of his daughter Hildegard’s acquisition of English and German.Diary studies are not entirely a thing of the past. Child language researchers often have children of their own, and some researchers have kept detailed records of their chil- dren’s language development. Some of the data we will refer to in later chapters come

from such diaries (e.g., Bowerman, 1985, 1990; Dromi, 1987; Halliday, 1975; Mervis, Mervis, Johnson, & Bertrand, 1992; B. F. Robinson &Mervis, 1998; Sachs, 1983; Tomasello, 1992b). In addition, researchers have sometimes trained mothers to keep diaries so that detailed records of the early language development of several children could be studied (e.g., L. Bloom, 1993; A. Gopnik & Meltzoff, 1987; M. Harris, Barrett, Jones, & Brookes, 1988; Naigles, Hoff, &Vear, 2009; Nelson, 1973).

Normative Studies

In the period between the end of World War I and the 1950s, the goal of most research on language acquisition was to establish norms (Ingram, 1989). Toward that end, several large-scale studies were undertaken to provide data on when children articulate different sounds, the size of children’s vocabularies at different ages, and the length of their sen- tences at different ages. Consonant with the behaviorist orientation of the times, the goal was not to ask theoretical questions about either the nature of humankind or the nature of language development but simply to describe what could be observed. These older studies are still valuable as descriptions of normative development (e.g., McCarthy, 1930; Templin, 1957), and as new instruments for assessing children’s language are developed, new normative studies continue to be conducted (e.g., Fenson et al., 1994).

The Chomskyan Revolution

In the 1960s, the study of children’s language development changed radically. The cata- lyst for this change was the 1957 publication of a slim volume entitled Syntactic Struc- tures, written by Noam Chomsky, then a young linguist at the Massachusetts Institute of Technology. That piece, along with Chomsky’s subsequent prolific work, revolutionized the field of linguistics and, within a few years, the study of language development. Before Chomsky’s work, linguists concentrated on describing the regularities of languages. Linguists could study their own language or, better yet, a little-known language, but the job was the same: to find the patterns in what speakers do. Chomsky caused a revolution by saying that what speakers do is not as interesting as the mental grammar that under- lies what speakers do. Since Chomsky’s writings, the work of linguists consists of trying to describe what is in the minds of speakers that explains how speakers do what they do.

That new goal of linguistics raised a question about children. If adults have a mental grammar that explains what they do when they talk, then children must have a mental grammar that explains what children do. Children’s speech is different from adults’ speech; therefore, children’s mental grammars must be different. What are children’s grammars like, and how do children eventually achieve adult grammars? In 1962, Professor Roger Brown and his students at Harvard University began to study the grammatical development of two children given the pseudonyms Adam and Eve (R. Brown, 1973). Somewhat later, a third child, Sarah, was added to the study. Every week for Sarah, and every two weeks for Adam and Eve, graduate students visited these children in their homes and tape-recorded their spontaneous speech. Transcripts of the children’s speech were then analyzed with the goal of describing the grammatical knowledge that underlay the speech they produced. That project, begun by Brown, along with just a few other projects (L. Bloom, 1970; Braine, 1963; W. Miller & Ervin, 1964), marks the beginning of the Chomskyan era of studying children’s language. The graduate students who met with Roger Brown to discuss the analyses of Adam’s, Eve’s, and Sarah’s language—along with a few notable others who were not at Harvard that year—became the first generation of child language researchers. We will discuss some of these pioneering projects when we discuss grammatical development in Chapter 6 Chomsky focused on grammar (the structure of language), and the first new wave of research on language development in the 1960s was on children’s grammatical development.

Later, in part following theoretical trends in linguistics, child language researchers shifted their focus more toward semantics and the acquisition of word meanings. In the late 1970s, the domain of language development was further expanded. Again following developments in linguistics, language use was added to the field of inquiry, and child language researchers began to study pragmatic and sociolinguistic development. In the 1980s and 1990s, linguis- tics and language development returned to focus on syntax, but the other questions about the lexicon and pragmatics have not been abandoned (or solved). The study of phonology and phonological development has also continued throughout this period, and the study of phonological development is becoming increasingly central to the study of language acquisi- tion as evidence mounts that phonological development provides the underpinnings for other aspects of language and literacy development. This topic will come up again, particu- larly in Chapters 4 and 5. (Accounts of the early history of child language research can be found in Golinkoff& Gordon, 1983; Ingram, 1989.)The Current Study of Language Development

Current Topics The current study of language development includes a much wider range of topics and a much wider range of populations than it did at its inception. New methods have also become part of the enterprise of trying to understand the nature of the human language capacity. Researchers now search for the basis of language in images of brain activity during language processing and in maps of the human genome (see Kovelman, 2012; and Chapter 2). Hypotheses about processes of language development are tested in computer simulations (see the section on methods in this chapter). Cross-cultural and cross-linguistic research has grown and become central to the field. The relation of language to thought and of language development to cognitive development has become a major topic. The study of bilingual development and the study of literacy have burgeoned. Cur- rently, the study of language development is a multifaceted field that includes a variety of very different research questions and approaches (Bavin, 2009; Hoff, 2012).Current Approaches Research on language development has always been guided by views of what language is, and there are currently several such views. One can think of lan- guage development as the process of learning to communicate in the way that the adults in one’s social or cultural group do so. Language, in this view, is a social behavior, and language acquisition is really language socialization. The goals of language socialization research are to describe children’s language use and their underlying understandings of language as a vehicle for social interaction at different ages and to identify the factors that influence that developmental course. This work includes, for example, studies of gender differences and cultural differences in styles of language use, studies of how children recount stories, negoti- ate conflicts, and tell jokes (e.g., Slobin, Gerhardt, Kyratzis, & Guo, 1996) and studies of how children in bilingual environments learn which language to use when (Tare & Gelman, 2011). We will pursue these lines of work more fully in Chapters 7, 8, and 9.In addition to being a social behavior, language is also a complex system that maps sounds (for oral language) to meanings. If one thinks of language development as the acqui- sition of this system, the research question is, how does the child do it? That is, what is the mental capacity that underlies the human ability to learn to talk? This question can be con- ceptualized in the following manner: The human capacity for language is a device residing in the human brain that takes as its input certain information from the environment and produces as its output the ability to speak and understand a language. (This model is pre- sented in Figure 1.2.) Everything that is part of adults’ knowledge of language (i.e., the out- put of the device) must be in the input, be in the internal device, or somehow result from the way the device operates on the input it receives. Noam Chomsky (1965) termed this capacity the Language Acquisition Device (LAD). Not everyone uses this terminology,

Model for Studying the Nature of the Language Learning Capacity

Information from the environment

Language acquisition

Language-learning mechanism

because it is associated with a particular, Chomskyan, approach to the field, but everyone who is interested in how children acquire the language system is, in essence, asking the question: What is the nature of the human language acquisition capacity?

Researchers do not start out completely neutral with respect to an answer to this ques- tion. (Scientists must always start out with some ideas of how things work; the work of scientists is testing those ideas.) Current research on language development can be usefully organized in terms of four different approaches that researchers take—each motivated by a different premise regarding the nature of the LAD and the language development it pro- duces. The approaches are the biological, the linguistic, the social, and the domain-general cognitive approaches. We introduce them briefly here so that readers are familiar with them when they come up in more detailed discussions of particular domains of language development. The biological approach starts with the premise that the human capacity for language is best understood as a biological phenomenon, and language development is best understood as a biological process. This premise then leads to research that investigates the degree to which language and language development share the hallmark features of other biological processes. Research in this vein looks for universal features of language develop- ment, for a hereditary basis to language ability, for evidence of a biologically based timeta- ble for development, and more. In addition, biologically motivated research leads to the study of the structures and processes in the brain that underlie language development (see Friederici, 2009; Kovelman, 2012).

The generative linguistic approach to the study of language acquisition focuses on describ- ing the nature of the child’s innate linguistic knowledge. This approach works from the prem- ise that the LAD must contain some knowledge of the structure of language in order for language acquisition to be possible. That innate knowledge cannot be specific to any particu- lar language; thus, it is Universal Grammar (UG). This approach seeks to describe UG and how it interacts with language experience to produce linguistic knowledge as a result (see Deen, 2009; Goodluck, 2007; Lust, Foley, & Dye, 2009; de Villiers & Roeper, 2011).

Other approaches reject this nativist premise. The social approach starts from the pre- mises that language is essentially a social phenomenon and language development a social process, and seeks to describe the social processes that produce language acquisi- tion. Research in this vein focuses on the social aspects of interaction as the experience relevant to language acquisition and on the social cognitive abilities of the child as the relevant learning capacities (see Baldwin & Meyer, 2007; Tomasello, 2009). The domain-general cognitive approach starts from the premise that language acquisition is

learning problem no different from any other and that children solve it in the same way that they solve other learning problems. Research in this vein seeks an account of how language might be learned by the child’s application of domain-general cognitive processes to the information available in input (see J. R. Saffran& Thiessen, 2007; Thiessen, 2009). Other work, such as that which has identified relations between early attention and early speech perception on later vocabulary development, illustrates what is often termed a developmental systems approach (e.g., Colombo et al., 2009; Kuhl, 2009; Spencer et al., 2009). The premise of this research approach is that early develop- ments and/or genetically based characteristics in one domain provide the foundation for subsequent developments in other domains. Thus, language development reflects cascad- ing effects in which both the child’s language knowledge and the child’s language learn- ing capacity change with development, and the outcome reflects the complex interaction between capacity and experience over time (e.g., Colombo et al., 2009).

Finally, the dynamical systems approach rejects the premise that language is a static sys- tem of knowledge and that language development consists of acquiring that knowledge. According to dynamical systems theory (DST), language emerges as a result of the contin- uous interaction of the components of the system and the environment. This self-organizing process accounts for both change in the child’s language abilities over developmental time and the moment-to-moment processes that occur as the child assembles words and longer utterances (see Evans, 2007; Vihman, DePaolis, & Keren-Portnoy, 2009). Dynamical systems theory has its roots in the fields of complex, nonlinear, dynamic systems in physics and mathematics. It is best known within developmental psychology in the work of Esther Thelen who brought a dynamical systems approach to the study of early motor development (Thelen& Smith, 1998). At this point, it is not a comprehensive approach to understanding language development, but DST does direct attention to certain phenomena that are rela- tively ignored in other approaches. These phenomena include variability in children’s perfor- mance and the influence of transitory states, as opposed to stable states of knowledge, on children’s language performance. For example, it is standard in the field to take the words

BOX 1.2 Current Approaches to the Study of Language Development

children produce asfairly direct evidence of their stored word knowledge. However, Gershkoff-Stowe and colleagues (Gershkoff-Stowe, Thal, Smith, &Namy, 1997) found that children are particularly likely to make errors in naming familiar objects at points when their vocabularies are expanding rapidly, and the particular words they mistakenly use are likely to be words they recently said. Thus both the timing and the nature of the errors sug- gest that speaking is not a reflection of static knowledge but the reflection of an underlying dynamic system and retrieval processes at work during the act of speaking. These current approaches to the study of language development are summarized in Box 1.2.

There is another bit of terminology to introduce with respect to characterizing approaches to the study of language development. A distinction has been made between the learnability approach and the developmental approach (see, e.g., L. Bloom, 1991). The learnability approach focuses on explaining the fact that language is acquired (i.e., that language is learnable). The developmental approach focuses on explaining the course of language development. These approaches are not mutually exclusive, and few researchers focus on one goal and ignore the other. Rather, different lines of research may differ in emphasis.

Major Issues in the Field of Language

Development

Another way to organize current research and theory in the field of language develop- ment is in terms of the major issues that any account of language development must address. The major issues concern, among other things, the degree to which language is innate or learned from experience, the nature of the system that is learned, whether

hange over the course of development is continuous or discontinuous, and how the communicative functions of language are involved in the process of learning the language system. We elaborate these issues in the next sections.

What Are the Contributions of Nature and Nurture to Language Acquisition? Is the development of language in children the result of human’s innate endowment (like

the development of upright posture and bipedal locomotion), or is it the result of the cir- cumstances in which children are nurtured (like the development of table manners or the ability to do calculus, both of which depend on particular experiences)? This is the nature–nurture debate, and it predates not only the modern study of language development but also the emergence of psychology as a discipline. This was the ongoing debate when the wild boy of Aveyron left the woods in 1800. The extreme experience-based position, known as empiricism, asserts that the mind at birth is like a blank slate; all knowledge and reason come from experience (J. Locke, 1690). The alternative view, known as nativism, asserts that knowledge cannot come from experience alone. The mind must have some preexisting structure in order to organize and interpret experience (H. Gleitman, 1995; Pinker, 1994; see the works of Plato and Kant for the original arguments). The question of whether language acquisition depends on innate knowledge is a currently debated question.

The Nativist View Applied to the question of how children acquire language, nativ- ism is the view that language acquisition depends on innate knowledge of properties of language. There are nativist proposals with respect to phonological and lexical develop- ment, but the central arena of debate is on the acquisition of syntax. We have already raised this topic in discussing the linguistic approach to language development, and we will consider it again in subsequent chapters when we discuss phonological, lexical, and morphosyntactic development. Here we try to provide the flavor of the position, apart from any linguistic details. To proponents of nativism, there are three salient “facts” about language development: (1) Children acquire language rapidly, (2) children acquire language effortlessly, and (3) children acquire language without direct instruction. Rapid, effortless, untutored development seems more like maturation than like learning in the usual sense of the term. As Chomsky (1993) put it,

Language learning is not really something that the child does; it is something that hap- pens to the child placed in an appropriate environment, much as the child’s body grows and matures in a predetermined way when provided with appropriate nutrition and environmental stimulation. (p. 519)

There are several positions with respect to language acquisition that could be classi- fied as empiricist. One of them is behaviorism, and as mentioned earlier, behaviorism has not stood the test of time (or empirical evidence) as a theory of language acquisition. Behaviorist theories will be mentioned again in the following chapters, but primarily for historical completeness.

The Interactionist View In current debate, the alternative to nativism includes a variety of positions that describe language acquisition as resulting from the interaction of innate characteristics of the child’s mind and the child’s language experience. These interactionist views place a greater burden of accounting for language development on the nature of children’s language-learning experiences than the nativist position does. Research on the nature of the language input children receive and the relation of that input to the rate and course of development are relevant here (see, e.g., Gathercole & Hoff, 2007; Hoff, 2006). The position known as social interactionism holds that a

rucial aspect of language-learning experience is social interaction with another person. Interactionists contest the “facts” so salient to the nativists. As Catherine Snow put it,

We on the other side think that learning language is a long slog, which requires from the child a lot of work. And the child is working as hard as he can, fifteen, sixteen hours a day. We think it requires a relationship with an adult, and a whole set of cog- nitive abilities. (quoted in Rymer, 1993, p. 37)

Another term for this type of position is constructivism. Constructivism as a view of development was first argued with respect to cognitive development by Jean Piaget, and constructivism remains a term in current use. According to the constructivist view, lan- guage (or any form of knowledge) is constructed by the child using inborn mental equip- ment but operating on information provided by the environment. In 1975, Noam Chomsky and Jean Piaget debated their respective nativist and constructivist views of language development at the Abbaye de Royaumont near Paris. Nearly 200 years after the wild boy of Aveyron left his woods (and roughly 200 miles away), the debate about the essential nature of the human mind continued. In his foreword to the edited tran- script of that debate, Howard Gardner (1980) summarized the two views:

Piaget saw the human child—and his mind—as an active, constructive agent that slowly inches forward in a perpetual bootstrap operation, Chomsky viewed the mind as a set of essentially preprogrammed units, each equipped from the first to realize its full complement of rules and needing only the most modest environmental trigger to exhibit its intellectual wares. (p. xxiii)

Other views that fall under the broad heading of interactionism have guided and con- tinue to guide current research in child language. In some work, the term emergentism has been used to label the view that knowledge can arise from the interaction of that which is given by biology and that which is given by the environment (MacWhinney, 1999). This term tends to be used in the context of models of learning that are termed connectionist, parallel distributed processing, or neural network models (Bates & Goodman, 1999). Connectionism is a proposed mechanism of pattern learning, and it seeks to explain lan- guage acquisition as the learning of patterns among smaller elements of sound or meaning. The developmental systems approach (also termed the epigenetic approach), already discussed, argues that development is the result of interactions between genes and the environment and their mutual influences as they unfold over time (Colombo et al., 2009; Karmiloff-Smith, 2007).

The debate between nativist and constructivist positions continues (Ambridge & Lieven, 2011; Pérez-Pereira, 2011; Valian, 2009). The currently most prominent constructivist posi- tion argues that language emerges from the interaction of children’s desire to communicate, their ability to read speakers’ intentions, their ability to learn the patterns in their input, and the information provided in input (Lieven & Brandt, 2011; Tomasello, 2009). The alternative position is also currently argued: that there are principles and categories of grammar that are innate and do not need to be learned, in any usual sense of the meaning of “learning.”

Although interactionism encompasses a variety of possible positions regarding just what is innate and how big a role experience plays, to theoretical purists there is no mid- dle position, no compromise between nativism and empiricism. Any position—social interactionism, dynamic systems, constructivism—that claims to account for language acquisition without innate syntactic content is an empiricist position (Valian, 2009), and any position that allows innate knowledge as part of the child’s language learning equipment is a nativist position (Ambridge & Lieven, 2011). As Ambridge and Lieven put it, “…specifically linguistic knowledge is either present at birth or it is not. There can be no compromise position” (2011, p. 6)

Are the Mechanisms of Language Acquisition Language-Specific or Domain General? If language acquisition is possible only because humans possess innate knowledge of the

structure of language and the mechanism of acquisition is setting parameters in this innate University Grammar, then this is a learning mechanism that is good only for lan- guage. In contrast, if language acquisition is the result of children applying general abili- ties to understand the intentions of others and to learn patterns in the sounds they hear, then the learning mechanisms that yield language are domain general.

The issue of whether language acquisition is supported by a domain-specific mental module or by general cognitive processes is one question in a larger debate in the fields of cognition and cognitive development. The view that the human ability to acquire lan- guage is specific to language is part of a larger theory known as the modularity thesis (Fodor, 1983), according to which the mind consists of multiple special purpose abilities or modules, and many have their own innate content. The opposing view is that the mind contains learning and reasoning abilities that apply across domains, and specific sorts of knowledge result from the application of these general-purpose capacities to spe- cific domains of experience. The question of what sorts of core knowledge infants bring with them into the world and how those innate characteristics interact with experience in development is an open question and a focus of a great deal of current research in the field of developmental cognitive science (see Carey, 2009; Spelke&Kinzler, 2007).

How Abstract Is Language?

There is also a larger issue within the field of linguistics that shapes the nativist- empiricist debate within the study of child language. There are dramatically different the- ories of what it is that children acquire when they acquire language, and these different descriptions of what children learn make different demands of the learning mechanism. According to the usage-based theory of language, adults’ syntactic knowledge is not as abstract and removed from the data available in speech as Chomsky argues. Rather, the knowledge underlying the ability to produce language consists of a repertoire of con- structions or formulas for producing utterances (see Ambridge & Lieven, 2011; Ellis, 1997; Tomasello, 2009). In children, these formulas are based on particular lexical items, not abstract syntactic categories. For example, a child might have a formula for producing sentences with eat and another formula for producing sentences with run. Children do not, for a long time at least, have an abstract category, “verb.” In contrast, the position associated with Chomskyan generative linguistics holds that linguistic knowledge consists of a system of rules that operate over symbols. The symbols stand for abstract categories such as “noun” and “verb,” for example. The actual word that is the noun or verb can vary, but the rule applies just the same. We will return to this issue in Chapter 6 when we discuss children’s development of morphology and syntax.

Is There Continuity or Discontinuity in Language Development?

It is obvious that children’s language knowledge changes as they grow. The 1-year-old who knows five words becomes the 2-year-old who knows hundreds of words. If what 1-year-olds understand about their first five words is the same sort of understanding that 2-year-olds have of their larger vocabularies, then vocabulary development is continuous—the change involves acquiring more of the same kind of thing as was there at the beginning. If, on the other hand, the 1-year-old has a very limited understanding of his or her first words—he or she says bye-bye and night-night as part of social rituals and says up only when he or she wants to be picked up—but the 2-year-old knows his or her words refer to objects and events in the world, then knowledge has changed in ki

not just in amount. Changes in kind are discontinuities in development. We will review this issue as it plays out in language development in Chapters 4 through 6, on phonolog- ical, lexical, and morphosyntactic development. Like the nature–nurture issue, the issue regarding continuity is not unique to the study of language development. Researchers who study cognitive development ask whether children’s understandings of the world change qualitatively over the course of development. Readers who are familiar with the theory of Jean Piaget know one proposal regarding discontinuous or qualitative changes in development.

What Is the Relation Between Communication and Language?

Children use the language they learn to communicate. In fact, one might say that the value of language to the human species is the communicative power it affords. Although no one doubts that language is useful for communication, there are differing views on how important communication is to language and to language acquisition. The two extreme positions are (1) formalism, the view that the nature of language and its acqui- sition have nothing to do with the fact that language is used to communicate; and (2) functionalism, the view that both language itself and the process of language acqui- sition are shaped and supported by the communicative functions language serves.

Formalist Views A clear and strong statement of the formalist position comes from Chomsky (1991):

For unknown reasons, the human mind/brain developed the faculty of language, a computational-representation system. [This system] can be used … in specific language functions such as communication; [but] language is not intrinsically a system of com- munication. (pp. 50–51)

For the formalists, language is an autonomous, arbitrary system whose form is indepen- dent of its function. Another position asserts that language was shaped in the course of evolution by its communicative value, but that the nature of that form cannot be derived from the functions it serves (see Pinker & Bloom, 1990). From the point of view of language-learning children, this position asserts, as does the Chomskyan formalist view, that language is an external system that has to be figured out—or provided innately—and the use to which that system is put provides no clues to how the system is structured.

Functionalist Views The contrary view is that language “is not an arbitrary and autonomous system” (Budwig, 1995) but rather a system shaped by the communicative functions it serves. And, according to one view, because the form of language reflects the communicative functions to which it is put, children are led to discover the form of lan- guage in using the system to communicate. As MacWhinney, Bates, and Kliegl (1984) stated, “The forms of natural languages are created, governed, constrained, acquired and used in the service of communicative functions” (p. 128).A number of different functionalist views exist today, and some make stronger claims than others about the usefulness of communication to language acquisition. One claim is that the infant’s social capacities are the source out of which language emerges (Snow, 1999; and see Baldwin & Meyer, 2007). The key to language acquisition, according to this view, is very young children’s understandings that other people are trying to com- municate with them. A related claim states that the desire to communicate one’s thoughts and feelings to others is the motivation for language acquisition (L. Bloom, 1991). According to both these views, communication explains the why of language development, but not necessarily the how. A stronger claim has been made by Tomasello and colleagues (e.g., Carpenter, Nagell, &Tomasello, 1998; Tomasello, 1992a, 2001), who

argued that communication also provides the how of language development. According to Tomasello,

Children are not engaged in a reflective cognitive task in which they are attempting to make correct mappings of word to world based on adult input, but rather they are engaged in social interactions in which they are attempting to understand and interpret adult communicative intentions . . . children acquire linguistic symbols as a kind of byproduct of social interaction with adults, in much the same way they learn many other cultural conventions. (2001, p. 135)

We will return to this topic when we discuss the communicative foundations of language in Chapter 3.

Theories of Language Development

It should be clear from the foregoing discussion of approaches and issues that research on how children develop language does not neatly divide into one or two theoretical camps. Rather, current theories of language development have multiple roots in theories of language, theories of cognition, and theories of development. Also, as we shall see in subsequent chap- ters, current theories take somewhat different forms as they are applied to different aspects of language development. Nonetheless, people do speak of “generativist” or “social interaction- ist” theories of language as a shorthand form of reference, and these terms do have some meaning. Box 1.3 lists the major categories of theory, along with their antecedent theories and their major tenets. Some of the labels for theories are the same as labels for the Major Theories of Language Development

THEORY SOURCE OR ANTECEDENT BASIC TENETS

Generativist

Generative linguistic theory

Universal Grammar, which contains the universal properties of language, is innate. Language experience triggers innate knowledge and sets language-specific parameters. The language-learning mechanism is specific to language.

Social interactionist

Communicative approaches to language and social interaction- ist approaches to development

Language is a social phenomenon. Children acquire language because they want to communicate with others. Communica- tive interaction with others, not just language input, is crucial. Children’s social–cognitive abilities serve the language acquisi- tion process.

Usage-based

Constructivist approaches to development in conjunction with cognitive and construction– based approaches to grammar

Language is a set of formulas for constructing utterances that operate over categories ranging in their level of abstraction. Knowledge of these formulas and the necessary linguistic abstractions emerge from the child’s pattern learning abilities in conjunction with their social cognitive understandings of speakers’ intended meanings.

ConnectionistConnectionist theories of per- ception, learning, and cognition

Language is a system of patterns among smaller elements of sound or meaning. Repeated experience hearing examples of patterns results in children mentally representing an abstraction from those patterns, which is the basis of children’s language knowledge. This pattern-learning procedure is used in other domains of learning as well.

BehavioristBehaviorist theories of learning

Language is built up via positive reinforcement of successive approximations to correct productions. This theory is primarily of historical interest.

approaches discussed earlier. Usage-based theory is given its own entry because it is the most prominent constructivist approach. Connectionism is listed as a separate theory because it entails a very specific view of the mechanism of learning. These theories will be referred to again and considered in more detail in later chapters when we consider how to explain chil- dren’s phonological, lexical, and morphosyntactic development.

All of these theories are current contenders as accounts of how children acquire lan- guage, with the exception of behaviorism. There were attempts to account for children’s language development using behavioral learning mechanisms: children imitate what they hear, and they are reinforced when they get it right and are corrected—or at least not reinforced—when they get it wrong. This account is obviously inadequate because the adult language ability is not confined to repeating sentences that have previously been heard. Somewhat more sophisticated behaviorist accounts tried to handle the productiv- ity of language in terms of “grammatical habits,” or word-association chains in which each uttered word serves to elicit the next word in the sentence (Staats, 1971), but such attempts to account for children’s language development and adults’ language ability in behaviorist terms were fairly short lived and unsuccessful. In 1959, Noam Chomsky wrote a scathingly negative review of B. F. Skinner’s (1957) attempt to account for lan- guage in behaviorist terms, and he was successful in convincing the scientific community that adult language use cannot be adequately described in terms of sequences of beha- viors or responses (N. Chomsky, 1959). Because the behaviorists’ notion of the endpoint of development was wrong, the behaviorist theory of achieving that end-point is inade- quate as a theory of language acquisition.

Methods of Research in Language Development Cross-Cultural and Cross-Linguistic Research The modern study of language development began with investigations of the acquisition

of English by middle-class American children. Initially, this focus was not seen as a ter- rible limitation because, the thinking went, the processes underlying language acquisition are universal, and thus discovering how children in Cambridge, Massachusetts, acquire English is the same as discovering how all children acquire any language.

Currently, the study of language acquisition by children who live in other cultures and the study of the acquisition of languages other than English are considered crucial to dis- covering the universal processes of language acquisition. Different cultures and different languages provide a means of discovering what is universal and what varies in language acquisition. Different languages provide children with different language acquisition tasks (e.g., in some languages nouns take different forms depending on their role in the sentence and in others do not), and different cultures provide children with different language learning environments (e.g., in some cultures, mothers talk to their infants and in some do not). Researchers no longer assume that if you’ve seen one child acquire language, you’ve seen them all (Fernald & Marchman, 2011; Hoff, 2006b; Küntay, 2012).

Research Designs and ProceduresIn their search for answers to the question of how children learn to talk, child language researchers use many of the same kinds of research designs that other scientists use. They engage in longitudinal and cross-sectional observational studies to describe developmental changes in children’s language, and they analyze those patterns of development for clues to the process underlying that development. They do correlational studies in which they look for relations between different aspects of language development or between language devel- opment and other aspects of development or experience. They do experiments in which they provide children with different kinds of exposure to language and then look for

differences in what children have learned. Sometimes researchers use computer simulations to test whether a hypothesized model of language development could work in principle, and sometimes researchers do case studies of individuals whose unique circumstances or pattern of development promises to shed light on some issue. The focus of studies of chil- dren’s language development can be language production, language comprehension, or both. Researchers interested in comprehension have been very inventive in designing ways to get small children to reveal what they think a word or a sentence means. We will discuss the particulars of different methods in later chapters as the research is dis- cussed. An introduction to research methods in child language is available in Hoff (2012). One aspect of methodology in child language research is so often employed and so specific to this field that it is worth discussing by itself. The analysis of samples of spon- taneous speech is the method Roger Brown used in his pioneering study of Adam, Eve, and Sarah, and it is a method that is still widely used today.

Assessment of Productive Language from Speech Samples

Speech Sample Collection Recording and then analyzing samples of children’s spon- taneous speech is one route to assessing children’s language development (see Rowe, 2012). Typically, the researcher picks a setting in which children are likely to talk—a mealtime or toy play, for example—and then records interactions in that setting. The recording can be done in the children’s homes or in a laboratory playroom. The children can be talking to the researcher or to someone more familiar to them (usually their mother). The purpose of collecting such speech samples is to find out the nature of the language children produce. Thus, the speech sample collected should be representative of everything the children say. Achieving representativeness can be difficult because speech may be different in different contexts (Bacchini, Kuiken, &Schoonen, 1995; Hoff, 2010; Hoff-Ginsberg, 1991). Another concern is that the act of recording will alter children’s speech in some way. This is prob- ably more of a problem for recording the speech of adults than that of children, because children tend to be less self-conscious than adults; however, researchers typically spend some “warm-up” time with their subjects before turning on the recorder. One research project attempted to secure more representative samples of children’s speech by putting little vests with radio-controlled microphones on children (Wells, 1985). The children wore the vests all day, although the microphone was turned on only intermittently.

How much speech needs to be recorded in order to estimate characteristics of a child’s language? The answer depends on the research question being asked. There is some consensus that 50 utterances is the minimum acceptable speech sample size (Hutchins, Brannick, Bryant, & Silliman, 2005). If the focus of interest is some character- istic of language use not present in every utterance, then, of course, the sample would need to be larger (see Tomasello& Stahl, 2004, for additional discussion). Also, the size of the speech sample depends on what sort of claims the researcher wishes to make. It is very difficult to make claims that a child does not produce a particular structure based on the absence of that structure in a sample of the child’s speech. There are also no established guidelines for how often children need to be recorded (L. Bloom, 1991). Researchers select different intervals using the existing literature to make their best guess at what interval will reveal the sorts of developmental changes they are studying. If the goal is to plot the emergence of particular structures, then very long speech sam- ples collected very frequently, sometimes called “dense” sampling (Lieven & Behrens, 2012; Maslen, Theakston, Lieven, &Tomasello, 2004), is necessary. Sometimes the research focuses on a particular type of language use, such as narrative production. In this case, more directive techniques of elicited production can be used. Children can be asked to retell a story that was just read to them (Reese, Sparks, &Suggate, 2012), to tell the story of a cartoon they just viewed (e.g., Shiro, 2003), or to recount something that

happened to them in the past (Shiro, 2003). As we shall see in Chapter 10, an enormous body of research on children’s narrative development is based on studies using the same technique of asking children to tell a story using a book that has pictures but no words (Berman &Slobin, 1994).

Another approach to recording children’s spontaneous speech is not to sample at all, but to record everything (Naigles, 2012). In the Speechome project, one researcher has gone so far as to wire his house and outfit it with video cameras and microphones so that his child’s entire life inside the house was recorded starting when he came home from the hospital as a newborn (Roy, 2009; and see L. R. Naigles, 2012). A somewhat more practicable system, which is in wide use in current research is LENA (Language Environment Analysis). The LENA system includes a small digital recorder that the child wears in a pocket of clothing designed to hold the device. This device records every sound the child produces and every sound produced within 4–6 feet of the child for 16 continuous hours, providing a much larger database from which to estimate properties of speech than could otherwise be achieved. The recordings are processed by language analysis software that uses acoustic properties of the signal to identify speech “segments,” to identify the source of the segments (it can distinguish between different speakers’ voices and between voices and the television), and even to estimate the number of different words in each segment (a not-too-technical description of how it does this is available in Warren et al., 2010). This allows researchers to estimate how much talk children produce, how much talk they hear from different sources, and how many conversational turns (i.e., exchanges between two speakers) occur. Researchers using this system have discovered, for example, that children’s language devel- opment is related to the number of conversational turns they experience—not just the amount of talk that is addressed to them (Zimmerman et al., 2009).

Speech Sample Transcription Unless the researcher makes use of the measures produced by an automated system such as LENA (there is also an automated transcriber associated with the Speechome project), the work is not done when the speech samples have been recorded. The records have to be transcribed. Transcription consists of writing down what was recorded, but that is more difficult than it sounds because the children being recorded were not giving dictation but were engaging in conversation. In conver- sation, people do not speak in full sentences; they interrupt each other and even talk at the same time. Furthermore, especially if they are children, their pronunciation is less than clear, and their usage not quite adultlike. Creating a transcript that is a faithful record of what was on the tape is difficult and time consuming. It requires training to be able to transcribe, and then it takes a minimum of 5 hours and up to 20 hours (Tomasello& Stahl, 2004) to transcribe each hour of recorded speech.Transcript Coding and Analysis After the speech has been transcribed, the researcher has to code the transcripts. Coding varies, depending on what the researcher is studying. For example, if the research is attempting to chart the development of verb usage, then coding the transcripts might involve identifying every verb in the children’s speech. If the purpose of the research is to study children’s conversational skill, then cod- ing the transcripts might involve categorizing every utterance the child produces as related or unrelated to what was said before. Ultimately, for researchers to conduct the kinds of analyses that get reported in journal articles, the codes have to be turned into numbers. For example, a researcher might analyze changes in the number of different verbs in children’s spontaneous speech or changes in the proportion of children’s utter- ances that are related to prior speech.

When this sort of research started in the 1960s, transcripts were handwritten docu- ments with columns for different codes. In that era, graduate students in child language logged many hours poring over these transcripts, identifying verbs or whatever the

research called for and adding numbers in the code columns. The advent of computer programs for analyzing child language transcripts has considerably lightened that load. Although programs that directly analyze the acoustic signal, such as those associated with LENA and the Speechome project, can yield some measures of the speech recorded, for many purposes it still takes a human being to transcribe and code. It is now possible, however, to enter transcripts and codes into computer-based programs. Then, instead of the researcher counting all the codes, the computer can do it—and far more quickly and accurately. The most widely used programs for transcript analysis in the field of develop- mental psychology are those associated with the Child Language Data Exchange System (CHILDES) (Corrigan, 2012; MacWhinney, 1991). CHILDES provides a set of programs for analyzing language, and it also has tools that enable the transcriber to link the tran- script to the digitized audio or video recording. This makes transcription easier and more accurate. It also allows the researcher to later go back and find portions of the interaction that are of particular interest. CHILDES also has a program, Phon, that allows speech to be phonetically transcribed and links to another program, Praat, that conducts acoustic analysis of the speech signal.

Another transcript analysis program is SALT (Systematic Analysis of Language Tran- scripts) (J. F. Miller & Chapman, 1985; see www.languageanalysislab.com). SALT was developed specifically for researchers and clinicians in communicative disorders, but it is a flexible program that can be used for basic research as well. It can also be used with English and with Spanish language samples. PEPPER (Weston, Shriberg, & Miller, 1989) is a program that provides a font for phonetic transcription. Another program, Logical International Phonetics Program, or LIPP, allows the user to transcribe in the International Phonetic Alphabet and thus permits finegrained phonetic analysis (Oller& Delgado, 1999).

CHILDES—A Data Archive One benefit made possible by computer-based transcripts is widespread data sharing. Although researchers have always been able to share data with colleagues by photocopying their transcripts, sharing is easier when the transcripts are in computer-based files and when (as a necessary side effect of computerized transcription) the transcripts are in a standardized format. The CHILDES project has taken the concept of data sharing even further by estab- lishing an archive. In the early 1980s, the MacArthur Foundation funded a project, led by Brian MacWhinney and Catherine Snow, to establish an archive for transcripts of children’s speech. Roger Brown contributed his transcripts of Adam, Eve, and Sarah, and other researchers contributed transcripts they had collected. Since then, researchers around the world have added their transcripts so that the CHILDES database now has more than 230 corpora (i.e., speech samples) representing more than 32 different languages, including both monolingual and bilingual children. Some of the transcripts have linked audio and video files.

The availability of this archive allows a researcher who has a question that can be answered by looking at transcripts of spontaneous speech to address that question using a much bigger dataset than any one researcher could record for a single study. The database can be used to ask about properties of children’s speech and also to ask about properties of the speech children hear. As an example of asking about children’s speech, Gelman and colleagues made use of the speech samples of 8 children who were recorded between 14 and 203 times during the period from 2 to 4 years. Using this data- base of over 150,000 utterances, they were able to describe developmental changes in the frequency and content of children’s production of generic noun phrases (e.g., birds fly), and they were able to discover that children utter such expressions of general under- standings of the world from a very early age (S. A. Gelman, Goetz, Sarnecka, & Flukes, 2008). As an example of asking about the speech children hear, one study asked whether the word happen tends to be used with negative events more frequently than with

positive events, thus providing children with a clue to its negative connotation (Corrigan, 2004, 2012). (In speech to adults, happen is used more often in connection with negative events, as in the phrase shit happens. If you say “Birthdays happen,” you are likely to be talking to someone old. The question is whether input provides a basis for children to learn this). Combining all the U.S. corpora of adults talking to children in the CHILDES database resulted in a speech sample based on 151 adult–child dyads, which contained 116,909 adult utterances in total and 1,100 uses of happen. Analysis of this sample revealed that happen is increasingly used in association with negative events as children’s language become more advanced (Corrigan, 2004). The CHILDES database has also been used in computational modeling. The transcripts of the adults’ speech in the adult–child conversations provide the speech corpus that is the input to the acquisition model under test. An overview of CHILDES is available in Corrigan (2012). A full description of the archive and the corpora themselves are available at the CHILDES website, which can be accessed at www.psy.cmu.edu. The website also contains online tutorials and a bibliography of references in the field of child language.

Standardized Tests and Measures of Language Development

Sometimes researchers want to be able to describe a child’s language in terms that com- pare that child’s language to the language of other children of the same age. Child lan- guage researchers are typically interested in such measures for the purpose of describing the children they are studying, much the way researchers in cognitive development sometimes want to describe their samples in terms of intelligence quotient (IQ) or men- tal age. By far, the biggest users of standardized measures are practitioners in communi- cative disorders, who use such measures for diagnosis and for treatment evaluation.

There are essentially two ways to assess how a child’s language development compares with that of other same-age children. One is to collect a speech sample and code it using a coding system for which norms have been collected. For example, a child’s mean length of utterance (MLU) is a good index of a child’s level of grammatical development, and data that provide norms for MLU have been collected (Leadholm& Miller, 1992; J. F. Miller & Chapman, 1981; see Chapter 6 for more on this topic). The SALT program will calculate the MLU on an appropriately entered transcript and will indicate the child’s level of grammatical development. Age-referenced norms for phonological fea- tures of children’s speech are also available (Grunwell, 1981).

The second way of getting a norm-referenced measure of a child’s language level is to employ one of the many existing standardized instruments. These instruments estimate the child’s language proficiency either by asking caregivers to report their children’s language comprehension or production or by having an examiner test the child. An example of a caregiver report instrument is the MacArthur–Bates Communicative Development Inven- tories (CDIs) (Fenson et al., 1994; see www.sci.sdsu.edu/cdi/). There are three versions of the English MacArthur–Bates CDI: one for infants between 8 and 16 months of age; one for 16- to 30-month-old toddlers; and one for 30- to 36-month-old children. These inven- tories consist of checklists that caregivers fill out to report on the gestures, words, and word combinations that their children understand and produce. Data from nearly 2000 children have been collected using these inventories, providing a basis for evaluating an individual child’s level of development. Researchers in many countries have developed ver- sions of the CDI in many different languages. There are MacArthur-Bates inventories for over 50 languages. These inventories in other languages are not translations of the English version, but rather they have been developed independently based on evidence of the words children know in the population they are designed for. To illustrate, the Spanish language version includes madrina, cenar, and bigote, while the English language version does not contain their equivalents of godmother, to dine, and mustache

An example of an examiner-administered instrument is the Peabody Picture Vocabu- lary Test (PPVT), which is used to assess vocabulary knowledge in children from 3 years to adulthood. The examiner presents the child with a word (“Can you find boat?”) and asks the child to select from four pictures the one that corresponds to the word. An indi- vidual child’s performance is compared to a larger reference group that provided norms for this test. There are also a wide variety of examiner-administered tests of school-aged children’s oral language and reading proficiency (Pan, 2012).

Computational Modeling

Sometimes researchers use computational modeling, which is a means of testing hypoth- esized language acquisition processes by implementing them as computer programs and then seeing if the computer can accomplish some part of the language acquisition task (see MacWhinney, 2010 and Waterfall, Sandbank, Onnis, & Edelman, 2010). In the domain of phonology, for example, a proposal that children learn how to produce speech sounds by hearing the results of their babbling movements has been subjected to test by programming the proposed learning model into a computer, which is then given as input many pairs of descriptions of articulatory movements with descriptions of the correspond- ing acoustic signals. After this “training,” the computer is presented with a description of a new articulatory sequence, and the test is whether the computer can produce correct acoustic descriptions (Plaut&Kello, 1999). In the domain of word learning, the proposal that infants find words in input via statistical learning has been tested by programming a computer with an algorithm for counting co-occurrences of syllables. The program is then fed samples of child-directed speech, and the output of the program is inspected to see if the sequences the program extracted are actually words (Swingley, 2005). In the domain of syntax, Waterfall and colleagues programmed a learning algorithm into a computer, fed the computer sentences taken from real-world samples of child-directed speech, and tested the computer’s ability to produce grammatical sentences and reject ungrammatical sen- tences (Waterfall, Sandbank, Onnis, & Edelman, 2010).

Computational modeling has also been used to ask how infants accomplish other language-related feats. It is well established that infants will look at faces that are synchro- nized to voices rather than look at talking faces that do not match the sound signal, and this helps them focus their attention on a particular talker in a noisy environment (Hollich, Newman, &Jusczyk, 2005). It is not obvious, however, what information infants use to do this. It turns out that a computational model that matches a face to a sound source when changes in one co-occur with changes in the other does a relatively good job of mimicking the behavior of infants presented with the same stimuli (Hollich& Prince, 2009). Of course, these descriptions of computational models are hugely oversimplified. The actual programs and procedures for testing their adequacy are very complex. The basic idea is simple, how- ever. Computational modeling tests theories by implementing them as computer programs, feeding the computer program the same information the child gets, and then asking if the computer, thusly programmed, does what the child does.

Sources for Research on Language Development Journals One way students new to language development can get an idea of the range of topics,

issues, and research methods in the field is to scan journals that publish research on lan- guagedevelopment. The titles of the articles in these journals give an idea of the topics being studied. The list of journals that contain papers on language development is long and includes journals from a variety of disciplines. The major sources are listed in Box 1.4.

BOX 1.4 Major Journals That Publish Research on Language Development

Copyright 2013 Cengage Learning. All Rights Reserved. May not be copied, scanned, or duplicated, in whole or in part. Due to electronic rights, some third party content may be suppressed from the eBook and/or eChapter(s). Editorial review has deemed that any suppressed content does not materially affect the overall learning experience. Cengage Learning reserves the right to remove additional content at any time if subsequent rights restrictions require it.

Developmental psychology journals

Applied Developmental Psychology Child Development Developmental Psychology Infancy

Journal of Experimental Child Psychology

Cognitive psychology journals

Cognition Cognitive Psychology

Linguistics journals

Discourse Processes Language

Psycholinguistics journals

Applied Psycholinguistics

Language development journals

First Language Journal of Child Language Language Acquisition Language Learning and Development

Language disorders journals

American Journal of Speech-Language Pathology Journal of Communication Disorders Journal of

Multilingual Communication Disorders Journal of Speech, Language, and Hearing Research Language, Speech, and Hearing Services in Schools

Neuroscience journals

The Behavioral and Brain Sciences Brain and Language Developmental Neuropsychology Cognitive Neuropsychology

Second language learning journals

Applied Linguistics Language Learning Second Language Research Studies in Second Language Acquisition

Other specialized journals

American Journal on Intellectual and Developmental Disabilities Journal of Autism and Developmental Disorders Bilingualism: Language and Cognition International Journal of Bilingualism

International Journal of Bilingual Education and Bilingualism

ndexes

If you already have a particular interest in some topic, or if you find an interesting topic by scanning the journals, you may want to find other articles on the same topic. Indexes can help you track down everything that has been written on a particular topic in lan- guage development. Just as the index in the back of this book allows you to find all the places in this book that a particular topic is mentioned, these indexes allow you to find all the places a particular topic is mentioned in the set of journals they scan. At one time, these indexes were published as separate directories, but they are now all available online—probably through your library’s website. The most widely used index among researchers in psychology is PsycINFO, which covers nearly 2000 different journals in psychology and related fields and provides an index to material in those sources. It includes journal articles since the 1800s and books and book chapters since 1987. Lin- guistics and Language Behavior Abstracts provide an index to material in over 1500 journals in language and language-related fields. To use these databases, all you have to do is type in the subject you are interested in (such as “lexical development” or “sign language”), and you will get a list of all the articles on that topic that appear in all the sources covered by that indexing service.

Most students are quite expert in the use of search engines such as Google or Bing to find information, but these have the problem that nobody is watching out to ensure the quality of what these engines find. The indexes that search journals have the advantage that they find only material that has been published in scientific journals and books. Google Scholar is a search engine available to anyone online that does restrict its search to scholarly material—and many researchers use it a great deal. A particularly useful fea- ture of Google Scholar is that when you find an article through Google Scholar you will also find a list of other studies that have cited this article. That feature helps you follow a line of research up to the present. However, Google Scholar provides access only to abstracts of the many articles and chapters it finds. If you want the entire article—and you should never cite something in a paper if you have read only the abstract—you will need to pay. Because libraries purchase subscriptions to electronic journals, conducting your searches through your library’s indexes will provide you free access to material that would otherwise incur a charge.

Summary

Language development is a multidisciplinary field that has as its central question, how is language acquired? Because language is highly complex yet universally acquired, the answer to this question has profound impli- cations for understanding the essential nature of the human mind. Because language is a vehicle for social interaction and acquired in a social context, the answer to this question may also reveal how development is sup- ported and shaped by the social environment. The study of language development also has practical importance for education, for the treatment of communicative disor- ders, and for second language instruction.

Acquiring a language includes learning the sounds and sound patterns of the language (phonological devel- opment), learning the vocabulary of the language (lexical development), learning the structure of the language (grammatical, or morphosyntactic, development), and

learning how to use language to communicate (prag- matic and sociolinguistic development). The study of language development has a long history because ques- tions about how children’s language emerges have long been considered central to larger philosophical and sci- entific debates. These debates have concerned the intrin- sic nature of humankind and the role of experience in shaping human nature.

The modern study of language development began in the 1960s following the Chomskyan revolution in linguistics. Chomsky argued that the study of language is the study of the mind. In turn, the study of lan- guage development captured the interest of researchers interested in the study of the developing mind. Lan- guage development is a field divided along several fault lines. Some major points of disagreement are (1) whether language is largely innate in the child or

learned from experience; (2) whether the knowledge that underlies children’s language ability is highly abstract or more like a memorized repertoire of con- structions that serve communicative goals; (3) whether the mechanism that underlies language acquisition is specific to language or consists of general-purpose cog- nitive abilities applied to the task of learning language; and (4) whether the communicative functions that language serves (for children and adults) account for language acquisition, contribute to the process of acquisition, or are merely a benefit of language acqui- sition that must itself be explained in other terms. Child language researchers also debate whether the most useful approach to understanding language devel- opment is to focus on children and ask how they acquire language (the developmental approach) or to focus on language and ask how it is acquired by chil- dren (the learnability approach).

Language development researchers use a variety of research methods and designs. Central to a great deal of research is the collection of speech samples from children for the purpose of characterizing the children’s productive language. Collecting speech samples involves recording children as they talk and transcribing and coding the recorded speech. Computer programs help in that process. For some purposes, researchers may not need to collect new speech samples if their question can be addressed by examining the speech samples contained in the CHILDES archive. For descriptive and assessment purposes, a variety of norm-referenced tests and measures of language devel- opment are available. Because language development is a multidisciplinary field, articles and chapters on language development appear in widely diverse sources. Most of these are indexed in one of two computer-accessible data- bases: PsycINFO or Linguistics and Language Behavior Abstracts.

Key Terms

phonology, p. 4 lexicon, p. 5 morphology, p. 5 syntax, p. 5 pragmatics, p. 5 sociolinguistics, p. 5 literacy, p. 5 behaviorism, p. 7 cognitivism, p. 7 cognitive science, p. 8 language socialization, p. 12 Language Acquisition Device

(LAD), p. 12

Universal Grammar (UG), p. 13 dynamical systems theory (DST),

14 learnability approach, p. 15 developmental approach, p. 15 nature–nurture, p. 16 empiricism, p. 16 nativism, p. 16 interactionist views, p. 16 language input, p. 16 social interactionism, p. 16 constructivism, p. 17 emergentism, p. 17

connectionism, p. 17 developmental systems approach/

epigenetic approach, p. 17 modularity thesis, p. 18 formalism, p. 19 functionalism, p. 19

speech samples, p. 22 CHILDES, p. 24 computational modeling, p. 26 PsycINFO, p. 28

Linguistics and Language Behavior Abstracts, p. 28

Review Questions 1. Describe the role the study of language develop- 6.

ment plays in cognitive science and applied fields. 7. 2. Learning a language involves learning in several

separable domains. List and define the compo-

nents of language knowledge. 3. What questions can be addressed by studying 8.

children who grow up without exposure to

language? 4. What was the Chomskyan revolution, and how

did it affect the study of language development? 9. 5. Define and contrast nativism and empiricism as (a) explanations of the origin of knowledge and

(b) as approaches to explaining language development.

What makes interactionism a nativist approach? Why do generativists and usage-based theorists agree that there can be no compromise on the question of whether humans have innate lin- guistic knowledge?

What can be learned from studying language development in other cultures and other language groups that cannot be learned from studying the acquisition of one language in one culture? Imagine you had to explain to your skeptical family (or roommate, or somebody) why you are taking a whole course just on language develop- ment. How would you justify spending this much time on such a narrow topic?

Ch. 2 (pp. 31-37; 40-70)

CHAPTER 2

Biological Bases of Language Development

Language as a Human Universal

Language Creation • The Common Basis of Language Creation and

Acquisition

The Human Vocal Tract and Language The Human Brain and Language • Some Basic Neuroanatomy • Methods of Neurolinguistic Investigation • Localization of Language Functions in the

Brain

Brain Development and Language Development

An Early Left-Hemisphere Specialization for Language
The Basis of the Left-Hemisphere Specialization for Language
Neural Plasticity in Childhood The Critical Period Hypothesis • First Language Acquisition After Infancy
Second Language Acquisition • Processes Underlying Age Effects on Second

Language Acquisition

The Genetic Basis of Language Development

The Heritability of Individual Differences • The Genetics of Language Impairment Language and Other Species • The Natural Communication Systems of Other

Species • The Acquisition of Human Language by Other

Species

The Origin of the Human Capacity for Language • Language as an Evolved Capacity • Language as a By-Product of Evolution Summary

Key Terms Review Questions

Having outlined the basic facts and major issues with respect to language acquisition, we now turn our attention to the organism that accomplishes this feat. Clearly, some unique property of humans explains why all humans talk, but dogs and gerbils never do. In this chapter, we will look for that biologically given aspect of human nature. We start with evidence that language is not only a universal characteristic of humans but also a virtually inevitable feature of human society. Next, we discuss the biological structures that underlie language, and we introduce the brain imaging techniques that have made investigation of the neural bases of language a burgeoning area of research. Then we ask whether language development displays two hallmarks of biologically based abilities: a critical period for development and heritability. After exploring the biological bases of language in humans, we turn to studies of other species and ask whether language is uniquely human. Finally, we discuss where the human capacity for language came from in the evolutionary history of our species.

Language as a Human Universal

Wherever there are humans, there is language. Just as fish swim and birds (with a few exceptions) fly, humans talk. This fact—that language is a universal characteristic of the human species—depends on all human children (barring impairment) having the capac- ity to learn language. But when we compare language in humans to flying in birds or

swimming in fish, we really mean something more. We mean that language is not merely something that humans can do if exposed to the right conditions but that lan- guage is something that humans cannot help doing.

Language Creation

In the normal course of events, each generation learns to speak the language it hears spoken by others, and thus it could be that the universality of language reflects only the child’s learning capacity and the universality of language experience. However, some sources of evidence argue that there is a deeper linguistic capacity in humans. We have not just a language-learning capacity but a language-making capacity.

Pidgins Sometimes historical circumstance throws together people who share no common language. To communicate in this situation, people invent a language that typically uses the lexical items from one or more of the contact languages but which has its own, very primitive grammar. Such languages are called pidgins, and they have arisen many times in history (see Jourdan, 2006, for a review of the cultural circumstances that have given rise to pidgins). For example, Hawaiian Pidgin English arose on the sugarcane plantations in Hawaii during the early part of the 20th century when immigrant workers from Japan, Korea, and the Philippines came together; they shared no language with one another or with the English speakers for whom they worked (Bickerton, 1981, 1984). Another example is Russenorsk, which arose when Russian and Norwegian fishermen needed to communicate with each other (Todd, 1974). These are not isolated examples; more than 100 pidgin languages are currently in use (Romaine, 1988). Most pidgins are structurally simple, although if used over many genera- tions, they do evolve, as do all languages (Aitchison, 1983; Sankoff& Laberge, 1973).

Creoles Children born into a community in which a pidgin language is the common means of communication acquire that pidgin as their native language. When children learn a pidgin, they add to it, creating a new language that is grammatically more com- plex. A creole is a language that once was a pidgin but which subsequently became a native language for some speakers (Todd, 1974); “creolization” is a process that creates new languages. It is unknown how many of the world’s languages originated in this way because the evidence is lost to prehistory, but there are signs that Swahili may be the result of contact between Arabic and Bantu languages (Todd, 1974).

The process of creolization tells us something about the biological basis of language. First, creolization suggests that the existence of language in a community does not depend on someone importing a language for the community to learn. People can invent their own. (In pidgins, the vocabulary is borrowed from one of the contact languages, but the grammar is not, as the example in Box 2.1 shows.) Furthermore, when children acquire the language, they add some grammatical features that are universal characteristics of human languages. Creoles that arose independently in different places nonetheless have similar characteristics. The shared features of independently arising creoles suggest that the human mind tends to construct only certain kinds of languages (Gee, 1993; and see Todd, 1974).

Not everyone agrees with the foregoing view of the origin of pidgins and creoles. Some argue that the similarities among creoles do not necessarily result from properties of the human mind but from the common uses to which all languages are put (Jourdan, 1991) or from the fact that many creoles have been influenced by the same language (Muysken, 1988; and see Todd, 1974). For example, many creoles show the influence of either English or Portuguese because of Britain’s history as a colonial power and because the Portuguese historically were worldwide seafarers and traders. Furthermore, some argue that the creoli- zation process does not depend on children acquiring a pidgin as their native tongue. Rather, the processes of grammatization that one sees in creolization can also arise out o

BOX 2.1 Examples of Utterances in Hawaiian Pidgin English

Source: Bickerton, 1990.

language use among adults (Bybee, 2007). It is difficult to settle arguments about the origin of creoles as long as the processes under dispute occurred a long time ago. What would really be informative would be to watch a pidgin develop into a creole.

The Development of Nicaraguan Sign Language In the second half of the 20th century, circumstances provided researchers the opportunity to watch a language develop in Nicaragua. In 1978, the government of Nicaragua opened that nation’s first public schools for the deaf. The deaf children who came together in these schools initially had no shared language. They did have their own idiosyncratic manual systems that they had developed for communicating with their families, but no way of communicating with each other. (The manual systems that deaf children invent are termed home sign. In Chapter 11 we will discuss the nature of the home sign systems that children develop and how they, too, are evidence of the language-making capacity of the human mind.)

Once these children were in daily contact with each other, a new language, Nicaraguan Sign Language (NSL), began to develop. Ann Senghas studied the changes that occurred as the language evolved by comparing the signing of individuals who learned the language in its early years of evolution to the signing of individuals who learned the language more recently (Senghas& Coppola, 2001). She found that the language moved from a structurally simple language to a structurally more complex language and that the differences in structural com- plexity appeared primarily in the signing of those who began to learn the language at an early age. It appeared that the older learners were less able to master precisely those com- plexities that characterized the more evolved form of the language. For example, in devel- oped sign languages, such as American Sign Language (ASL), the location in which a sign is produced (slightly to the left or right of the signer) modulates the meaning of the sign. So, if a speaker produces the sign for “cup” and the sign for “tall” in the same location, that indicates that “tall” modifies “cup.” In its first version, NSL did not have these spatial modulations; 20 years later it did. In 1995, Senghas studied the signing of individuals who entered the school before 1983 and those who entered the school after 1983 and found dif- ferences between these two groups in their use of spatial modulations—the second group used them much more. Within that second group, however, it made a difference how old the individuals were when they were first exposed to NSL. The younger they were at first exposure, the more they used spatial modulations in their signing. These findings suggest that the changes that have occurred in NSL over time depend particularly on having young children acquiring the language, and thus the findings suggest that it is children, more than adults, who contain within them the engine that drives language creation.

The Common Basis of Language Creation and Acquisition

It has been argued that the capacity for language creation evidenced in phenomena such as creolization and the development of NSL is the same capacity that underlies language

acquisition. One argument holds that both processes are ones in which children take a little bit of material from what they experience, add to it their own internal knowledge of language, and produce a language system as a result. This is essentially the language bioprogram hypothesis proposed by the linguist Derek Bickerton (1981, 1984, 1988), who has argued that humans are endowed with an innate skeletal or “core” grammar that constitutes “part or all, of the human species–specific capacity for syntax” (Bicker- ton, 1984, p. 178). Normally, in the process of language acquisition, input in the target language causes the language-learning child to modify and add to this bioprogram. In the absence of a full-fledged target language, the bioprogram builds a language using the available input to fill out the core grammar. Evidence for Bickerton’s proposal con- sists of his analysis of a few creole languages and his claim of similarities between creoles and child language. Both types of evidence have been called into question (e.g., Aitchi- son, 1983; Corne, 1984; Goodman, 1984), but the proposal that the same language- specific inborn ability underlies both language acquisition and language creation is theoretically consistent with the nativist approach to language acquisition (Pinker, 1994; Senghas& Coppola, 2001).

Critics of nativism have proposed alternative accounts. It is possible to accept the pro- posal that creolization, language creation, and language acquisition all reflect the same process without accepting the idea that this process is language specific. Bates (1984) argued that both language creation and language acquisition result from nonlinguistic cognitive mechanisms seeking a solution to communicating. Meier (1984) also argued that general cognitive mechanisms could underlie creolization and language acquisition. One such mechanism would be the ability to find patterns or even impose patterns on noisy data. Adults who speak a pidgin show inconsistent use of many grammatical fea- tures. Children exposed to this inconsistent input end up acquiring a more regular sys- tem, which is the creole. This imposition of a system on variable data may be a feature more true of children’s learning than of adults’ learning. In artificial language-learning experiments, children who were exposed to unpredictable variation did not reproduce exactly what they heard; they made the language more consistent. Adults did not (Hudson Kam & Newport, 2005).

In sum, the evidence shows clearly that language is an intrinsic part of human nature. Humans are not just able to learn language; they create language. There is disagreement, however, over the extent to which this capacity is unique to children and specific to lan- guage. We will leave these questions for the moment and turn to a description of the anatomical structures that serve this capacity—the vocal tract and, more important, the brain.

The Human Vocal Tract and Language

The capacity to produce speech depends on the structure and the functioning of the human vocal tract, which is illustrated in Figure 2.1. Speech is produced when air from the lungs exits the larynx and is filtered by the vocal tract above the larynx.

We can change the pitch of the sound we produce by tightening or loosening the vocal folds in the larynx. We can further change the sound that comes out of our mouths by changing the shape of the vocal tract above the larynx (or, technically speaking, the supralaryngeal vocal tract). Although the structures in our vocal tract serve other purposes—biting, chewing, swallowing, inhaling—these structures have features that seem better suited for speaking than for their other functions. Human teeth are even and upright, which is not necessary for eating but is useful for producing certain sounds, such as [s] and [f]. The human lips and tongue also have properties that are useful for rapidly producing different sounds but are not particularly necessary for anything else

by examining whether these functions are carried out by different parts of the brain. To understand the research in this area, we need to begin with basic neuroanatomy.

Some Basic Neuroanatomy

The appearance of the brain as an undifferentiated mass is misleading. The brain actually has different parts that perform different functions. When you look at an intact brain, what you see is the outer layer—the cerebral cortex. Hidden underneath the large cortex are the subcortical parts of the brain. Roughly speaking, the cortex controls higher mental functions, such as reasoning and planning, and the subcortical structures control more primitive func- tions, such as eating and breathing. The cortex itself is divided into two cerebral hemi- spheres; in most individuals, the area of the cortex that sits over the ear (the temporal lobe) is larger in the left cerebral hemisphere than in the right (Geschwind&Levitsky, 1968). The left and right cerebral hemispheres are connected by a band of nerve fibers known as the corpus callosum. An interesting feature of the human nervous system, which will be important to understanding some of the research on the brain and language, is that each cerebral hemisphere is connected to the opposite side of the body. As a result of these contralateral connections, the right side of the brain controls the left side of the body and vice versa. Also, information coming into sense receptors (e.g., the ears, the skin) on the right side of the body goes directly to the left cerebral hemisphere and vice versa. (There are also same-side or ipsilateral connections, but these are not as strong as the contralateral connections.) The actual work of the brain is accomplished by neural circuits, which are made up of thousands of interconnected neurons (i.e., neural cells) that fire together when presented with a particular stimulus or when accomplishing a particular task (Shafer & Garrido-Nag, 2007). These, then, are the basic components of the organ that neurolinguists study: neural circuits, housed in two cerebral hemispheres connected to the rest of the body by contralateral fibers, connected to each other by the corpus callosum, and sitting on top of subcortical structures. The next topic is the study of what these different parts do.

Methods of Neurolinguistic Investigation

One way to study what different parts of the brain do is to study patients who have suf- fered injuries to different parts of their brains and determine what functions are impaired as a result. This technique is known as the lesion method—lesions being localized areas of damaged brain tissue. The goal of the lesion method is to correlate bits of missing brain with bits of missing psychological functioning (Damasio, 1988). Some individuals have a severed corpus callosum but an otherwise undamaged brain. These patients are called split-brain patients, and studying them provides a unique window on how each hemi- sphere functions. In the past, researchers have also studied healthy brains as they process language using the dichotic listening task. This task works by taking advantage of the fact that information presented to the right ear goes first to the left hemisphere. Thus, if two different stimuli are presented at the same time to the left and right ears, the subject will report hearing what is presented in the right ear, if the stimuli are processed in the left hemisphere. Measuring the right-ear advantage was used for many years as a way to ask whether the left hemisphere was doing the processing (Kimura, 1967).In the last 20 years, new techniques have developed, which provide researchers a way of watching the brain at work and asking what parts of the brain perform what functions. Collectively, these newer methods are known as functional brain imaging methods. They have revolutionized the field of neurolinguistics because they can be used with adults, children, and infants who have normal intact brains. All these methods allow researchers to present stimuli to the subjects in the study, and then the imaging technique provides data on where the brain is most active as it processes those stimuli. We briefly introduce

these techniques before turning to the evidence they have provided on the neurological underpinnings of language development. Photos of the equipment involved and the data produced are presented in Figure 2.3.

Electroencephalogram (EEG)/event-related potential (ERP) measures the electrical activity in the brain via electrodes placed on the scalp. (The infant on the first page of this chapter is wearing an electrode net, which holds electrodes to her scalp. The wires coming out of the net feed information from the electrodes to the computer.) The volt- age fluctuations associated with the presentation of particular stimuli or the performance of particular tasks are the event-related potentials. The location of ERPs associated with different mental activities is taken as a clue to the area of the brain responsible for those activities (Caplan, 1987; Neville, 1995). The ERP method has several advantages. It is

FIG-2-3 Neuroimag- ing Techniques

Neuroscience techniques used with infants

EEG/ERP: Electrical potential changes

Excellent temporal resolution • Studies cover the lifespan • Sensitive to movement • Noiseless

MEG: Magnetic field changes

Excellent temporal and spatial resolution • Studies on adults and young children • Head tracking for movement calibration • Noiseless

fMRI: Hemodynamic changes

Excellent spatial resolution • Studies on adults and a few on infants • Extremely sensitive to movement • Noise protectors needed

NIRS: Hemodynamic changes

Good spatial resolution • Studies on infants in the first 2 years • Sensitive to movement • Noiseless

+5μV Fz –5μV

Standard

200

Deviant

Vertex

1 6

2 7

3 8

4 9 10

Nasion

5 12 Inion

Tragus

Source: Kuhl & Rivera-Gaxiola (2008), Institute for Learning & Brain Sciences, University of Washington

noninvasive and relatively undemanding of the subject; the baby just has to wear a net or cap to hold the electrodes in place. As data, ERPs provide precise information about the timing of brain events. The major disadvantage is that ERPs provide poor information about the location of the event in the brain because all that is measured is what can be detected on the surface. Although localization may be imprecise, decades of research with adults have now established that certain ERPs are associated with particular tasks or types of processing (Friederici, 2005; Kovelman, 2012). For example, adults typically show negativity in the centro-parietal region 400 milliseconds (the N400 response) after hearing a semantically anomalous sentence (e.g., “We bake cookies at the zoo,” Holcomb, Coffey, & Neville, 1992) and positivity at approximately 600 milliseconds (the P600 response) after hearing an ungrammatical sentence (e.g., “My uncle watched about a movie my family,” Friederici, 2005). Researchers can then ask at what age children start to process language in the way adults do by asking when they show the same N400 and P600 responses to semantic and syntactic tasks. One response that infant researchers have made much use of is the mismatched negativity (MMN) response. This is an iden- tifiable change in surface voltage that reliably follows a change in the stimulus being pre- sented. For example, if adults hear /ba/ /ba/ /ba/ /ga/ /ba/ /ba/ /ba/, there will be an MMN response after the /ga/. The subject does not need to do anything—this is an auto- matic response. Looking for MMN responses, then, is a way researchers can ask what differences infants detect at an age when they are too young to tell you themselves.

Magnetoencephalography (MEG) detects magnetic field changes associated with the electric neural activity in the brain through sensors contained in a helmet that surrounds the participant’s head (Kovelman, 2012; Kuhl & Rivera-Gaxiola, 2008). The child sits under a device that looks a bit like the old hair dryers of beauty salons in the 1950s, and researchers get a picture of the magnetic field entering and leaving the child’s head. The procedure is safe, quiet, and provides good temporal and spatial resolution. Because the changes in magnetic field associated with language processing have to be detected against the background of the Earth’s magnetic field, the equipment has to be used in a magneti- cally shielded room. As you might imagine, this limits the portability of the equipment.

Another technique, functional magnetic resonance imaging (fMRI), provides images of activity in the brain by tracking blood flow, or the hemodynamic response to neuronal activity. When the brain works, it uses energy, and the body responds by increasing the flow of blood that carries oxygen and glucose to the working part of the brain. Because the oxygen-carrying blood has magnetic properties, that blood flow can be tracked using MRI. Many readers may be familiar with MRI as it is used for medical diagnosis. fMRI makes use of the same machine—a big, long tube the subject lies in, but in the case of fMRI, the subject inside the machine is doing something and the question being asked is what part of the brain is expending the most energy as the subject performs the task. fMRI provides excellent spatial location—it is much better than ERP at indicating where in the brain the processing is happening, but because blood flow lags behind activ- ity, fMRI does not provide as good temporal resolution as ERP does. fMRI also requires the participant to be quite still inside the MRI machine, and the machine is extremely expensive to operate—both these factors are obstacles to using fMRI in research.

Near-infrared spectroscopy (NIRS), also known as optical topography, is among the newest techniques for measuring brain activity. This technique takes advantage of the fact that the transmission of light through the tissue of the brain is affected by hemoglo- bin concentration changes. Light-emitting and light-detecting devices are placed on the scalp at particular landmarks, and measures of light transmission are taken as evidence of oxygenation of the blood and thus neural activity occurring between the emitter and the detector (Peña, Maki, Kovacic et al., 2003). NIRS has several advantages. Like the measurement of ERP, NIRS is noninvasive and relatively undemanding of the subject

also does not require a special room or room-sized equipment, which makes it portable.

This survey does not exhaust the list of brain imaging techniques that are used (see Kovelman, 2012). There are, for example, multimodal approaches in which two measure- ment techniques are used simultaneously to provide information about both where and when activity is occurring. The exploration of the neurological bases of language devel- opment is one line of research in the larger enterprise known as developmental cogni- tive neuroscience. This is a new approach to studying the brain and behavior, made possible by the advent of techniques for studying children’s brains at work (Kovelman, 2012).

Localization of Language Functions in the Brain

Language as a Left-Hemisphere Function In 1861, the French physician Paul Broca reported to the Anthropological Society of Paris on a patient known as “Tan” because that single syllable was all he could say. Tan survived with this condition of near-total mutism for more than 20 years, although he also developed paralysis on the right side of his body in his later years. When Broca examined Tan’s brain in autopsy, he found a lesion on the left side caused by a fluid-filled cyst. Broca subsequently reported on many more cases in which patients had lost “the faculty of articulate lan- guage” (as Broca’s words are often translated). All had left-hemisphere damage (see Caplan, 1987, for a full historical account).

The condition in which language functions are severely impaired is known as aphasia. Broca’s basic observation that loss of language is typically a result of brain injury to the left but not the right hemisphere still stands today (Caplan, 1987; Goodglass, 1993; Saffran& Schwartz, 2003). Damage to the right cerebral hemisphere tends to cause different pro- blems, particularly in processing visual–spatial information (Springer & Deutsch, 1981; Witelson, 1987). In broad terms, there seems to be a division of labor between the left and right hemispheres of the cerebral cortex: The left hemisphere is specialized for lan- guage (and some other things), and the right hemisphere is specialized for processing visual–spatial information. This state of affairs, in which one hemisphere is more impor- tant than the other for particular competencies, is known as functional asymmetry.

Although Broca concluded that “we speak with the left hemisphere,” we can now modify that conclusion to say that the left hemisphere processes language—regardless of whether it is spoken or not. Bellugi, Poizner, and Klima (1989) studied deaf signers who had suffered strokes that damaged portions of either their left or their right hemi- sphere. The researchers found that left-hemisphere damage resulted in aphasia for sign- ers just as it does for users of a spoken language. Furthermore, even though sign language uses a visual–spatial modality, signers with right-hemisphere damage were not aphasic. Since then, more evidence from signers who have suffered brain damage and findings from neuroimaging studies of deaf signers indicate that the left hemisphere is specialized for language, regardless of modality (Emmorey, 2003).

Another classic source of evidence that language is a left-hemisphere function is found in individuals with a severed corpus callosum. The corpus callosum may be sev- ered in a surgical procedure performed as a treatment for severe epilepsy that has not responded to other measures. Severing the connections between the two hemispheres seems to stop the spread of the electrical activity that accompanies seizures, thus reduc- ing the severity of the seizures. Patients who have had this surgery have an intact left hemisphere and an intact right hemisphere; thus, each hemisphere can still perform its normal functions. But because the corpus callosum is severed, there is no communica- tion between the two hemispheres. These split-brain individuals have little difficulty

functioning in daily life, but in experimental settings, they provide researchers with the opportunity to study what each hemisphere does alone.

Split-brain patients have been tested in experimental situations in which a picture is presented only in the left half of the visual field—that is, to the left of a center fixation point. Because the connections from the left half of each retina go to the right hemi- sphere, only the right hemisphere knows what is on the picture. The information avail- able in this situation to each cerebral hemisphere of a split-brain patient is illustrated in Figure 2.4. Split-brain patients in this situation typically cannot say what they saw. How- ever, they are able to draw with their left hand (the one connected to the right hemi- sphere) what they saw. Researchers believe that this inability of split-brain patients to indicate verbally what was presented to the left visual field means that the hemisphere that knows what’s out there, the right hemisphere, cannot talk; and the left hemisphere, which can talk, doesn’t know what’s out there.

Right-Hemisphere Contributions to Language The right cerebral hemisphere also makes some contribution to normal language functioning, as evidenced by language impairments associated with right-hemisphere damage. Right-hemisphere-lesion patients sometimes produce abnormal intonation contour when they speak, and they may have difficulty recognizing the emotional tone of an utterance (Caplan, 1987). Right- hemisphere-damaged patients have difficulty understanding jokes, understanding sarcasm, interpreting figurative language, and following indirect requests (Weylman, Brownell, & Gardner, 1988). These difficulties suggest that the right hemisphere may be involved in the pragmatic aspects of language use (Weylman et al., 1988), although not in the “core” psycholinguistic capacities (Caplan, 1987). Studies of ERPs in intact

FIG-2-4 Illustration of the Information Available to the Left and Right Hemispheres of a Split-Brain Patient

Fixation point

Stimulus in left visual field

Left eye

Left hemisphere

Optic nerve

Information delivered to left visual processing area

Stimulus in right visual field

Right eye

Right hemisphere

Severed corpus callosum

Information delivered to right visual processing area

patients show that the right hemisphere is activated by semantic processing, whereas the left hemisphere is activated primarily by syntax processing (Neville, Nicol, Barss, Forster, & Garrett, 1991). In sum, studies of what the right hemisphere contributes to language when the left hemisphere is intact indicate that the right hemisphere is involved in semantics and pragmatics but that syntax is the province of the left hemisphere.

Studies of patients who have had their left hemispheres removed (because of severe pathology) and studies of the language capacity of the right hemisphere in split-brain patients also suggest that the right hemisphere has some semantic and pragmatic compe- tence, but limited syntactic (and phonological) abilities (Baynes & Gazzaniga, 1988; Dennis, 1980; Gazzaniga, 1983; Zaidel, 1985). Such findings fit well with the findings from ERP studies that show more right-hemisphere involvement in language for indivi- duals who learned language late (Neville, 1995) and with the evidence that these late language learners are not quite as good at syntax as are those who learned language in infancy (J. Johnson & Newport, 1989; Newport, 1990). We will return to the question of the right hemisphere’s capacity for language when we discuss the case of Genie—a child who acquired what language she did relatively late in life and who seems to have done so with her right hemisphere.

Individual and Sex-Related Differences in Brain Organization Everybody’s brain is different. Although the standard description of the left hemisphere as the seat of language functions is true for most people (between 80 and 98 percent, depending on the source of the estimate), it is not true for everyone, and the degree of dominance is not always the same (Bryden, Hecaen, &DeAgostini, 1983; Caplan, 1987; Milner, 1974). People who are left-handed (but still left-hemisphere dominant) and even people who are right-handed but have left-handed family members may show more bilateral participation in language (Caplan, 1987). Also, women show more bilateral participation in language than men do (Bloom &Hynd, 2005; McGlone, 1980). Even within the left hemisphere, there are individual differences in where language functions are represented, although the causes and consequences of individual differences in brain organization are far from fully understood.

Other Neurological Divisions of Labor Language appears to be multifaceted, not only from the linguist’s point of view but also from the neurologist’s. The “real estate” of the brain appears to be more finely zoned into areas of specialization than merely the distinction between the left and right hemisphere. One source of evidence for the multi- faceted nature of language is the fact that there are many different types of aphasia. Some individuals with aphasia have difficulty producing speech, and the speech they do produce seems to lack grammatical structure. Instead, their speech tends to consist of short strings of content words—nouns and verbs—without grammatical morphemes. This syndrome is termed Broca’s aphasia. Other individuals with aphasia have no trou- ble producing speech, but the speech they produce makes no sense. This syndrome is termed Wernicke’s aphasia. When patients with Wernicke’s aphasia speak, either they use words that are wrong for the meaning they are trying to express or they use made- up, meaningless words. The speech of Wernicke’s aphasics has been described as “syn- tactically full but semantically empty” (Blumstein, 1988, p. 203). When one function can be disrupted without affecting another, we say these two functions are dissociable. In fact, there are many more different types of aphasia than just these two.

A great deal of research has been directed toward trying to relate the particular type of aphasia with the particular location of brain damage within the left hemisphere—to create a zoning map of the brain, in effect. For example, Broca’s aphasia is typically asso- ciated with damage to the front part of the left hemisphere, near the part of the cortex

FIG-2-5 Language Areas of the Brain

Motor cortex Wernicke’s area

Broca’s area Auditory cortex

that controls movement: an area known as Broca’s area. Wernicke’s aphasia is typically associated with damage to a region more posterior than Broca’s area, next to the primary auditory cortex: Wernicke’s area. These areas are mapped in Figure 2.5. This association between the location of left-hemisphere damage and the particular sort of resulting lan- guage impairment inspired hypotheses that Broca’s area was the seat of grammar and Wernicke’s area the seat of meaning. The truth is probably more complicated than that. There are actually many more than two patterns of aphasic symptoms (E. M. Saffran& Schwartz, 2003). Studies that make use of ERP and other brain-imaging techniques with intact adults also have provided evidence that different parts of the left hemisphere carry out different linguistic jobs (Neville et al., 1991; Stromswold, Caplan, Alpert, & Rauch, 1996). Mapping the brain is currently an active area of research. It is apparent that the correct account of how the brain is organized for the many functions that constitute lan- guage will be more complex than the classical model of a syntactic Broca’s area and a semantic Wernicke’s area (Poeppel& Hickok, 2004).

Brain Development and Language Development

At the same time as researchers are actively investigating how the adult brain is organized to serve language functions, researchers are also asking whether the child’s brain is orga- nized differently and how the adult state is achieved. The questions asked include where language functions are carried out in the brains of infants and children, what developmen- tal changes occur in the localization of function, and what factors cause the changes that occur. Another topic of particular interest in the study of brain development and language development is the ability of the young brain to recover language functions after injury.

An Early Left-Hemisphere Specialization for Language

Evidence from Neuroimaging Studies Studies using both ERP and fMRI— sometimes in combination—have found that young infants show greater activation in the left than right hemisphere when presented with a variety of acoustic signals, not only speech, and that the lateralization for speech becomes stronger as children’s lan- guage skills develop (Dehaene-Lambertz, Hertz-Pannier, & Dubois, 2006). Other researchers, using optical topography (NIRS) with newborns, have found that speech eli- cits greater left-hemisphere activation than does the same speech signal played in reverse (which removes some of the speech-like quality of the sound), although the properties of the speech signal that cause this effect are not clear (Peña et al., 2003).

Evidence from Childhood Aphasia At one time, it was thought that aphasia in children was equally likely to result from left- or right-hemisphere damage and that, unlike adults, children could recover completely from aphasia. Current research contra- dicts both those views. In children who have already acquired language, left-hemisphere damage is more likely to result in aphasia than is right-hemisphere damage, consistent with other data that suggest an early left-hemisphere specialization for language. Early reports of language loss in children following right-hemisphere injury might be based on cases that occurred before antibiotics were used, and the aphasias that were observed following right-hemisphere injury might actually have been caused by bacterial infections that affected the whole brain (Woods &Teuber, 1978). However, the type of aphasia that children experience following left-hemisphere damage is different from the aphasia seen in adults. Children are more likely to suffer nonfluent (Broca-type) aphasia, whereas Wernicke’s aphasia is more common in adults. This difference suggests that within the left hemisphere, developmental changes affect the way language functions are organized (see also Satz& Lewis, 1993; Stiles &Thal, 1993).

With respect to recovery from aphasia, current research supports the view that chil- dren recover more quickly and more fully than adults. However, studies also find that when children who appear to have fully recovered are tested and compared to normal controls they often show signs of residual impairment (Aram, Ekelman, Rose, & Whitaker, 1985; Vargha-Khadem, O’Gorman, & Watters, 1985; Woods & Carey, 1979).

Evidence from Cases of Brain Injury Prior to Language Studies that have fol- lowed the language development of children who suffered brain damage either in utero or in the first few months of life find a different relation of injury site to language impairment than do studies of older children or adults. A crude summary of the very complicated findings in this field is as follows: Brain damage in either the right or the left hemisphere prior to language acquisition can cause language delay, and the size of the lesion is more important than the location in predicting its effect. The relation of injury site to delay is different for different components of language development and differs for expressive language and language comprehension. Another factor that complicates identifying the effects of early lesions is that many children who suffer early brain damage also subsequently experience seizures, and it is difficult to untangle the effects of the initial brain damage from the effects of the seizures (Rowe, Levine, Fisher, & Goldin-Meadow, 2009). One thing that seems clear in the messy pattern of results is that the right hemisphere is more involved in language acquisition than it is in language functioning once language is acquired. That is, early right-hemisphere damage impairs language acquisition more than later right-hemisphere damage impairs language functioning.Like children who suffer brain injury after they have begun to talk, children with very early brain injury catch up to a substantial degree but also show some residual impairment. Studies find that language development is initially delayed, but, by age 5 to 7, children who experienced brain injury prior to language acquisition score within the normal range on standardized tests of language. However, when the children with early brain injury are compared to a control group of children matched to the children with brain injury in terms of age, sex, and socioeconomic status, their scores on standardized tests are lower (Feldman, Holland, Kemp, &Janosky, 1992; Levy, Amir, & Shalev, 1992; Rowe et al., 2009; Satz& Lewis, 1993; Stiles, Bates, Thal, Trauner, & Reilly, 1998).At one time, research on the development of the neural underpinnings of language was driven by two competing views regarding the left-hemisphere specialization for language that is seen in adults. One view, termed the equipotentiality hypothesis, held that at birth the left and right hemispheres have equal potential for acquiring language (Bishop, 1983, 1988;

Lenneberg, 1967). The opposite view, termed the invariance hypothesis, held that the left hemisphere has the adult specialization for language from birth and that nothing about lateralization changes with development (see Satz, Strauss, & Whitaker, 1990, for a review). The data from the neurological study of language in children suggest a position somewhere between equipotentiality and invariance. On the side of invariance, the data from neuroim- aging studies of intact children suggest some cortical specialization present from birth. On the side of equipotentiality (if not true equipotentiality, at least more nearly equal potential of both hemispheres), the degree of asymmetry in brain function increases with develop- ment, and the ability of the right hemisphere to take over language functions for a damaged left hemisphere is greater in children than in adults (see also summary in Conboy, 2010). This account leaves many questions unanswered: Why do newborns process speech in their left hemispheres? Why does the lateralization of language function seem to increase with development? Why do children recover from aphasia better than adults do?

The Basis of the Left-Hemisphere Specialization for Language

The rise of neuroimaging techniques has fueled an aggressive hunt for what properties of the left hemisphere make it the language hemisphere and what processes account for changes in the functional organization of the brain. One proposal for the initial and enduring left hemisphere role is that the speed with which neurons in the left hemi- sphere fire matches properties of language, and it is this temporal alignment that not only sends the speech signal to the left hemisphere for processing but also might even help to segment the speech stream (Morillon et al., 2010). The initial configuration, while biased to process language in the left hemisphere, is not the same as the final con- figuration of language processing in the brain. Changes may arise from maturational processes, such that some areas of the brain only become ready to serve their language functions at some later point in development.

Developmental changes in which hemisphere handles language may also arise from changes in how children process language as they gain expertise. An early finding related to this hypothesis came from a study that found that experienced musicians showed a right-ear (left-hemisphere) advantage for music, whereas naive listeners showed a left-ear (right-hemisphere) advantage for music (Bever &Chiarello, 1974). These researchers sug- gested that trained musicians process music in a more analytical way than naive listeners do, and that analytical processing engages the left hemisphere. Another related possibility is that the right hemisphere is better at processing novel stimuli, whereas the left is special- ized for executing well-practiced routines (see Mills, Coffey-Corina, & Neville, 1997). Language, for adults, is a well-practiced routine. In support of this latter view, Mills and colleagues found, in 20-month-olds, that processing familiar words was a more exclusively left-hemisphere function in children who had larger productive vocabularies. These researchers suggested that the development of left-hemisphere specialization for language may reflect the development of increasingly efficient language processing (Mills, Conboy, & Paton, 2005; Mills, Plunkett, Prat, & Schafer, 2005). (In Chapter 5, we will discuss evidence that children with larger vocabularies do process words more rapidly than children with smaller vocabularies [Fernald, Swingley, & Pinto, 2001; Marchman, Fernald, & Hurtado, 2010; Zangl, Klarman, Thal, Fernald, & Bates, 2005].)

Neural Plasticity in Childhood

The explanation of why children show quicker and more nearly complete recovery from aphasia is not likely to be unique to language. Younger brains show better recovery than older brains of all functions following injury, and the explanation has a great deal to do with the greater plasticity of children’s brains compared to adult brains. Plasticity is the

ability of parts of the brain to take over functions they ordinarily would not serve (Witel- son, 1987). For the most part, brain tissue does not regenerate once it is damaged. Thus, when children with left-hemisphere damage recover language function, other areas of the brain must be taking over the functions previously carried out by the damaged portions of the left hemisphere. Evidence of subtle syntactic impairments even after recovery from childhood aphasia suggests that the right hemisphere is never quite as good as the left at some aspects of language. Although the plasticity of the immature brain allows one part to take over the work of another, there remain telltale signs that the right hemisphere is doing a job for which the left hemisphere is better suited.

The source of the great plasticity of the immature brain is likely the initial redun- dancy in the neural architecture. Beginning during the fetal period, the brain grows syn- aptic connections, reaching a peak between the ages of 2 and 5, depending on the region of the brain. At this peak point, the brain has many more synaptic connections than it needs. Subsequent development consists primarily of losing connections (Clancy & Finlay, 2001; P. R. Huttenlocher, 1994). As connections are lost, redundancy is lost, and thus, particular functions come to be located in specifically dedicated areas rather than throughout the brain (Neville, 1995).

For this sort of developmental process to work, there must be a way to ensure that only redundant connections, and not needed ones, are lost. This is accomplished by the activity of the young brain, which influences which connections are lost and which remain. A variety of evidence suggests that connections that are used become fixed or stabilized, whereas unused connections are eliminated. Unlike the early empiricists’ notion of experience producing tracings on a blank slate, neurophysiological evidence shows that huge numbers of tracings are innately provided and that the absence of expe- rience structures the brain by erasing some of them. (For more complete discussions, see Bertenthal& Campos, 1987; Greenough, Black, & Wallace, 1987; P. R. Huttenlocher, 1994; Neville, 1991; Witelson, 1987.) Another metaphor used to represent the influence of experience on brain organization describes experience as a sculptor shaping the brain by removing the portions it doesn’t need (Kolb, 1989). Daily use of the left hemisphere for language appears to stabilize language in the left hemisphere and allows elimination of the redundant right-hemisphere capacity. If the left hemisphere is damaged early in life, the right hemisphere still has the capacity to take over language functions, but with age, that capacity declines. Biology appears to set a deadline by which connections must be used or they will be lost.

The Critical Period Hypothesis

The notion that a biologically determined period exists during which language acquisi- tion must occur, if it is to occur at all, is known as the critical period hypothesis. Nature provides many examples of biologically determined deadlines, the best known of which is probably the critical period for imprinting in birds. Some species of birds, including chickens and ducks, walk as soon as they hatch. They follow the first moving thing they see, and they keep on following it to maturity. Normally, that first moving thing is the chick’s or duckling’s mother, with the result that baby chicks and ducklings follow their mothers everywhere. The experience of following a moving object is necessary for imprinting, and it must happen within a few hours after hatching. The window of time during which the necessary experience must occur is the critical period. This require- ment is not always absolute. The terms sensitive period or optimal period are some- times used, and often the term critical period is used to mean sensitive or optimal period (Werker& Tees, 2005). There are also well-documented examples of critical per- iods in human development. For instance, some cells in the brain respond to input from

both eyes in the normal adult, but if these cells fail to receive input from two eyes during the first year or two of life, they lose this capacity. Thus, the features common to all examples of critical or sensitive periods are these: Some environmental input is necessary for normal development, but biology determines when the organism is responsive to that input. The critical period hypothesis with respect to language acquisition was originally proposed by Lenneberg (1967), who described language acquisition as an “age-limited potential” (p. 175) with the relevant age being puberty.

First Language Acquisition After Infancy

“Wild” Children The perfect experiment to test the critical period hypothesis would involve depriving children of exposure to language during the normal period of language development, providing that exposure later, and then examining the language develop- ment that occurs to see if it differs in any way from the language development that occurs when exposure begins at birth. Of course, such an unethical experiment could never be done deliberately. There are, however, unfortunate cases provided by history that do provide such a test of the effect of delayed exposure to language.

In addition to the most famous “wild child,” Victor of Aveyron, who was discussed in Chapter 1, there are other cases of children who suffered early social isolation, who were first exposed to language later than would be typical, and who were not successful in acquiring normal language. It is difficult to learn much from such cases, however, because they are so poorly documented. Furthermore, when such children fail to acquire language, we cannot be sure whether the failure was due to the late start or to some impairment the child might have had previously. Victor, for example, displayed many behavioral characteristics associated with autism (Wolff, 2004). Reviewing the evidence in 1967, Lenneberg came to the conclusion that “the only safe conclusions to be drawn from the multitude of reports is life in dark closets, wolves’ dens, forests, or sadistic parents’ backyards is not conducive to good health and normal development” (p. 142).

There is one success story among such children. In the 1930s, a 6-year-old child named Isabelle was discovered living hidden away in a dark room with only her deaf- mute mother for contact. After her discovery, Isabelle was trained intensively to speak, and she did learn to talk. Isabelle’s success makes it clear that she was cognitively normal, but her deprivation was also less extreme than that of the other cases of wild children. Furthermore, we do not have the sort of psycholinguistic details about Isabelle we would like. Although at age 8 she was described as having “a normal IQ” and “not easily distinguished from ordinary children of her age” (Brown, 1958b, p. 192), no one administered the tests of linguistic competence that would allow detailed comparisons of her language competence with that of children with normal experience. There is one modern case of a wild child who was discovered after linguis- tics and neurology were sufficiently advanced to allow us to ask questions that were not asked of the earlier cases.

The Case of Genie In 1970, a woman who is known to most of the world only as “Genie’s mother” was looking for the office of services for the blind in downtown Los Angeles. She was nearly blind, was seeking help for herself, and had only recently man- aged to escape virtual captivity by her mentally ill husband. By mistake, she entered the general social services office. She brought with her a 13-year-old daughter, Genie. The eligibility worker at the social services office noticed the small, frail-looking child with a strange gait and posture and called her supervisor, who, after questioning Genie’s mother, called the police. The police took Genie into custody and admitted her into the hospital for severe malnutrition (Curtiss, 1977; Rymer, 1993).

The story of Genie’s background that was eventually revealed was horrific. From the time Genie was 20 months old until her mother’s escape when she was 13 years, Genie spent her time alone, strapped to a potty chair in a small bedroom. She was fed hur- riedly, with minimal interaction and no talk. If Genie made any noise, her father would beat her with a large piece of wood he kept in the room for that purpose. Like the wild boy of Aveyron before her, Genie had no language when she was discovered. Also like Victor of Aveyron, Genie was immensely interesting to the scientific community. The story of Genie’s life and treatment both before and after her discovery has been described by Curtiss (1977) and by Rymer (1993). We shall confine ourselves here to the investiga- tion of Genie’s language development, described by Susan Curtiss in her dissertation and subsequent papers (Curtiss, 1977, 1985, 1988, 1989).

Genie did not talk at all when she was first discovered. Four years later, she scored in the range of a normal 5-year-old on standardized vocabulary tests. She combined words into complex utterances, and she could express meanings. However, her lan- guage was far from normal. As the examples of Genie’s speech in Box 2.2 show, her vocabulary and semantic skills far exceeded her syntactic skills. Her grammar was defi- cient in both production and comprehension. In production, her utterances were tele- graphic, lacking most grammatical morphemes. In comprehension tests, she failed to understand passive constructions and distinctions marked by tense, and she had other difficulties as well.

Another fact about Genie’s language might be related to her grammatical limitations. Dichotic listening tests showed that language was a right-hemisphere activity for Genie. In fact, the nature of her grammatical limitations has been compared to the grammatical deficiencies of patients who have recovered language after surgical removal of the left hemisphere. One possible explanation of this phenomenon is that Genie was exposed to language too late for the normal process of acquisition of language as a left-hemisphere function. She acquired language with the right hemisphere, and—as we have seen in the aphasia data—the right hemisphere is not as good at language as the left. Genie’s conver- sational competence was also extremely limited, and she often ignored the speech addressed to her. As Curtiss (1977) described it:

Verbal interaction with Genie consists mainly of someone’s asking Genie a question repeatedly until Genie answers, or of Genie’s making a comment and someone else’s responding to it in some way…. Except for those instances where Genie exerts control over the topic through repetition, verbal interaction with Genie is almost always con- trolled and/or “normalized” by the person talking to Genie, not by Genie. (p. 233)

Curtiss attributes this conversational incompetence to Genie’s lack of early socializing experience.

The study of Genie is certainly more informative than earlier reports on isolated chil- dren. Evidence that language was a right-hemisphere function for Genie suggests that, by

BOX 2.2 Examples of Genie’s Utterances

Source: Curtiss, 1977.

age 13, a left hemisphere that has never been used for language has lost that capacity. However, interpretation of Genie’s outcome is still hampered by the fact that we do not know with certainty that Genie was a normal child except for her experiences. Once, when Genie was seen by a doctor as an infant, she was diagnosed as mentally retarded. But there was never any follow-up to see whether that pediatrician’s impression was cor- rect, and even before Genie was totally isolated, she had something less than an ideal environment. Susan Curtiss, who worked most closely with Genie, vehemently disagrees with the possibility that Genie could be retarded (Rymer, 1993), but we simply do not know for sure.Late Acquisition of American Sign Language A better test of the critical period hypothesis is provided by individuals who have normal early experience except for being deprived of exposure to language. This has been the circumstance of many children born deaf to hearing parents. These children have no language input at home because they cannot hear the language their family speaks, and their parents do not know sign lan- guage. (Also, the parents have often been discouraged from learning sign to communi- cate with their children. We will discuss this again in Chapter 11, when we discuss language acquisition in children who are deaf.) Many of these deaf-of-hearing children, as they are termed, eventually are exposed to sign language when they meet other deaf children, some of whom have deaf parents and have been exposed to sign language from infancy. Comparing the sign language acquisition of children who learned sign from infancy to that of children who were first exposed to it later in childhood or in adult- hood provides a very nice test of the critical period hypothesis. If the young brain is bet- ter at language acquisition, deaf individuals who began to acquire sign language as older children should be less proficient than those who acquired it in infancy.

Newport (1990) studied the sign language proficiency of deaf adults who ranged in age from 35 to 70, who used ASL in their everyday communication, and who had done so for more than 30 years. Some of these adults had acquired ASL as infants from their deaf parents. Some had first been exposed to ASL when they entered a school for the deaf between the ages of 4 and 6; some had first been exposed only after the age of 12, when they entered the school as teenagers, or later, when they made friends with or mar- ried someone from that school. Newport administered a battery of comprehension and production tests to assess how well these deaf adults had mastered the grammar of ASL. She found that adults who were first exposed to ASL after early childhood did not per- form as well as those who had been exposed as infants, even after 30 years of using the language every day. Other studies have similarly found that early learners of sign achieve greater proficiency than do late learners (Mayberry &Eichen, 1991; and also see the summary in DeKeyser& Larson-Hall, 2005). This evidence suggests there is some benefit to being a young language learner.

Second First Language Acquisition in Internationally Adopted Children

Researchers have also studied the language development of internationally adopted chil- dren to ask whether a later start than normal affects language development. Children who are internationally adopted are exposed to one language from birth, but then, at the point of adoption, they stop hearing that first language and start hearing the new language of their adoptive parents. Thus the children start over again, in what has been termed second first language acquisition. Most international adoptees are adopted before the age of 2 (Hwa-Froelich, 2009), so their language outcomes are a test of whether even a slightly late start makes a difference.

Between 1998 and 2008, more than 200,000 children were adopted from foreign countries by families in the United States (Hwa-Froelich, 2009), and international

adoption occurs in other countries as well (Hyltenstam, Bylund, Abrahamsson, & Park, 2009). The outcomes for these children differ depending on the nature of their preadop- tive experience. For example, the children who were rescued from Romanian orphanages where the conditions were extremely poor often showed delays in multiple aspects of development and experienced long-term difficulty (Croft et al., 2007; Nelson, Furtado, Fox, &Zeanah, 2009; Scott, Roberts, &Glennen, 2011). In contrast, children adopted from China tend to have more successful outcomes (Roberts et al., 2005; Scott et al., 2011), probably because their early care was better—if still much less than optimal.

Studies of children adopted from China by families in North America have found that the children make rapid progress in acquiring their new language, and most score within the normal range for native speakers on standardized tests within 2 years after adoption (Roberts et al., 2005; Snedeker, Geren, & Shafto, 2007). This finding does not mean, however, that the international adoptees were unaffected by their late start. First, the normal range is wide and children could be delayed, but still be within that normal range. Second, adoptive families tend to be of higher socioeconomic status (SES) than average, and children adopted from China are almost exclusively girls. Both SES and gender affect language development.Two studies that have compared children adopted from China to control groups of children who were matched for socioeconomic status and gender found that the adopted children lagged behind the control children at 4 years of age (Cohen, Lojkasek, Zadeh, Pugliese, &Keifer, 2008; Gauthier & Genesee, 2011), while still scoring within the normal range on the tests that were employed. A follow-up of some of the adopted children at 7 years found significant differences on several language measures between the adoptees and control children matched on age, gender, and socioeconomic status. Although from a practical standpoint the children who were adopted as infants “caught up” to children whose language exposure began at birth, the small but measurable differences between the internationally adopted children and their matched controls hint at effects of their later start that are theoretically interesting.

Second Language Acquisition

A far more frequently occurring test of the effect of age on language acquisition occurs in the realm of second language acquisition. Imagine a family that immigrates from a non-English-speaking country to the United States. It is likely that any young children in the family will appear to acquire English quickly, and when they are adults, they will be difficult to distinguish from native speakers of English. In contrast, those who immi- grate as adults will master the new language slowly, with difficulty, and will never quite sound like native speakers. This apparent greater ease and success of second language acquisition in childhood is widely observable and is consistent with the hypothesis that children have a unique, biologically based ability to acquire language. When scientists carefully study second language acquisition at different ages, however, they find a slightly different story. They do find age of exposure effects such that the earlier one is exposed to a second language, the higher the level of proficiency one is likely to ultimately achieve. However, it takes children longer to acquire a second language than most people realize, and even children do not always achieve native-like proficiency. We start first with the evidence that young learners do have an advantage over older learners, and sec- ond, we turn to the evidence that qualifies that position. (Other aspects of second lan- guage acquisition in childhood are discussed in Chapter 9.)

Age of Exposure Effects on Second Language Acquisition In a study by Oyama (1976), the English speech of 60 Italian immigrants was tape-recorded, and two judges scored those recordings on a five-point scale ranging from no foreign accent to heavy

foreign accent. Oyama then analyzed the influence of two variables: (1) the age of the immigrants at arrival in the United States and (2) the number of years living in the United States. Oyama found a strong effect of age at arrival, with young arrivals showing less accent than older ones. The number of years had no effect. Oyama (1978) similarly found strong age-of-arrival effects on second language users’ ability to repeat English sentences presented to them under noisy conditions (i.e., static on the tape).

Other studies have similarly found that age of arrival affects the ability to speak a sec- ond language without an accent and that native-like performance depends on exposure beginning in early childhood (Flege, 1987; Flege& Fletcher, 1992). Sometimes, the benefit of youth to acquiring unaccented speech in a second language is explained as the effect of age on acquiring a motor skill. Speech production involves moving the lips, tongue, and mouth in ways particular to each language, and that may be what is difficult for a late learner. However, there is more to knowing the sound system of a language than just the motor skill, and Oyama’s (1978) finding of effects of age of arrival on comprehension sug- gests that not just the mouth, but also the brain, is involved in explaining age effects on the mastery of second language phonology. As we will discuss in Chapter 3, when we discuss the perceptual foundations of language development, there is evidence that the sounds of the language one hears in infancy tune the brain’s perception of speech sounds. The result of this tuning is that we become good at automatically detecting sound differences that matter in the language we learn as infants, but, in some cases, at the cost of a diminished ability to detect sound differences that do not matter. This may be a basis of the difficulty adult learners have with a new language (Kuhl, Conboy, Padden, Nelson, &Pruit, 2005).

Age-of-exposure effects have also been observed for measures of grammatical compe- tence. J. Johnson and Newport (1989) presented grammatical and ungrammatical English sentences to Chinese and Korean natives who were living in the United States and who had learned English as a second language. As a group, they did less well on identifying the ungrammatical utterances than a comparison group of native English speakers. The result that native speakers have an advantage over second language learners on gram- matical tasks has even been found in second language learners who are highly proficient, working as authors and college professors in their second language (Coppieters, 1987).

Evidence even more directly related to the critical period hypothesis came from an analysis of subgroups of the second language speakers in the Johnson and Newport (1989) study. Those who were between the ages of 3 and 7 when they arrived in the United States were not different from the native speakers of English. Those who were between the ages of 8 and 15 at arrival performed less well on the test than native speak- ers, but the younger they were at arrival, the more nearly they approximated native com- petence. Those who were 17 or older performed least well and did no better than those who were 30. As was the case for accent, differences were observed between those exposed as young children, under the age of 7, and those exposed as older, but still pre- pubescent, children. Particularly strong evidence that very early language exposure is necessary in order to achieve the competence of a native speaker came from an inge- nious study conducted in Sweden (Abrahamsson&Hyltenstam, 2009). The researchers started by advertising for people who spoke Swedish as their second language but who spoke it so well that they were usually taken to be native Swedish speakers. The subjects were recorded as they talked, and then those recordings were played for a panel of 10 native Swedish speakers who judged their accents. It turned out that not everyone who reported that they sounded native passed the judges’ scrutiny and that success was related to age of first exposure. The findings are presented in Figure 2.6. Among the par- ticipants whose exposure began before age 13, the majority were judged to sound native by at least 6 of the judges. Among the late learners, in contrast, only 5 out of 88 sounded native to at least 6 judges. A subset of the participants who sounded native were given

FIG-2-6 Effects of 10 Age of Exposure on

n = 20

Native speakers

Language Acquisition

9 8 7 6 5 4 3 2 1 0

n = 53

Early childhood AO <1–5

n = 54

n = 31

Late Adolescence childhood AO 12–17

AO 6–11 Age of onset group

n = 33

Early adulthood AO 18–23

n = 24

Later adulthood AO 24–47

Source: From Abrahamsson&Hyltenstam, 2009

other tests of their Swedish language proficiency, and on these too, better performance was associated with earlier age of first exposure.

Limitations on Second Language Acquisition in Childhood Although children clearly have multiple advantages over adults in acquiring a second language, they have limitations as well. Children are not faster at second language acquisition than adults. A study of the acquisition of Dutch as a second language by English-speaking families who moved to the Netherlands found that 12- to 15-year-olds showed greatest progress over the course of their first year. They learned more rapidly than adults and they also learned more rapidly than younger children (Snow &Hoefnagel-Höhle, 1978). Even the advantage of youth for ultimate achievement is sometimes overstated. Recall in Abrahamsson and Hytelstam’s (2009) study of second language learners of Swedish, not all of the learners who began before age 13 performed in the range of native speakers.

Processes Underlying Age Effects on Second Language Acquisition

When Lenneberg (1967) proposed the critical period hypothesis, he had a particular mechanism in mind. He argued that maturational changes in the brain at puberty end the special language acquisition ability of the left hemisphere, and to the degree that lan- guage can be learned after puberty, it is via other general-purpose learning mechanisms. Several avenues of research since Lenneberg have addressed both the timing of critical period effects and the processes that might underlie them.

The Timing of Age-of-Exposure Effects on Language Acquisition The evidence from the study of international adoptees suggests that even starting language acquisition at the age of 2 years is different from starting at birth. The tuning of phonetic categories that occurs in infancy provides a candidate mechanism for this very early critical period (Kuhl, Conboy, Padden, Nelson, &Pruit, 2005; and see Chapter 3). Other evidence suggests there are effects of age after puberty as well, such that 17-year-olds have an ad- vantage over 30-year-olds (Birdsong, 2005; Hakuta, Bialystok, & Wiley, 2003; Newport, Bavelier, & Neville, 2001). Evidence that different aspects of language display different age functions suggests there may be multiple critical periods, with some ending earlier and some later (Newport et al., 2001; Werker& Tees, 2005).

Age Effects on Mechanisms of Language Acquisition One question related to the critical period hypothesis is whether children and adults acquire language in different ways. One study found that among a sample of adult Hungarian-speaking immigrants to the United States, those who had arrived as children were more proficient than those who had arrived as adults and that a measure of verbal analytic ability predicted individ- ual differences in English proficiency among the adult arrivals, but not among those who arrived as children (DeKeyser, 2000). Other evidence, however, suggests that the abilities underlying first and second language acquisition overlap considerably (Dale, Harlaar, Haworth, & Plomin, 2010). Furthermore, process differences are not limited to differ- ences between child and adult learners. Some evidence suggests that international adop- tees who start over after the age of 3 follow a somewhat different course of language development than younger adoptees, perhaps because they are making use of their knowledge of their first language in acquiring the second (Snedeker, Geren, & Shafto, 2012). Thus, differences in the process of first and second language acquisition do not necessarily mean that the brain’s capacity for language acquisition changed. The different approach taken by later learners could simply mean that if you already know one lan- guage, you use it in learning a second.

Early Exposure Effects on a General Linguistic Ability Another piece of data to fit into the picture of how age of exposure affects language development is the finding that early exposure to a language may not only benefit the acquisition of that particular language but may also benefit the ability to acquire any language (Mayberry, Lock, & Kazmi, 2002). Two comparisons suggest this effect. One is the comparison of ASL skills between two groups of deaf adults who were first exposed to ASL after the age of 9. One group was of individuals who were born deaf, but who for a variety of reasons had no ASL exposure until the age of 9 or later. The other group was of individuals who were born hearing, acquired a spoken language, but then were exposed to ASL after becoming deaf. The adults who had early experience with a spoken language were better at ASL than the deaf adults who had no early language experience, suggesting that early knowl- edge of one language helps the acquisition of another language.

The second comparison in this study was of the English language skills of three groups of adults: (1) deaf adults who were late learners of English but who learned ASL in infancy, (2) deaf adults who were late learners of English and had no ASL expo- sure in infancy, and (3) hearing adults who were late learners of English as a second language. In this comparison, the deaf adults who had been exposed to ASL in infancy looked like the hearing second language learners (on this written task), and they were substantially better than the deaf adults who had not been exposed to ASL in infancy. These data are presented in Figure 2.7. Again, early exposure to any language seems to be better than no language exposure as far as later language acquisition is concerned. This finding has implications for understanding age effects on language

Chapter 2

FIG-2-7 Effects of Early Language Experience on Later 80 Language Learning

70 60 50 40 30 20 10

90 a

No early language

Early spoken language

Early signed language

Early spoken language

Source: From Mayberry, Lock, & Hazmi, 2002

acquisition—the benefit of early exposure to ASL for the later acquisition of English is not likely to be in the tuning of speech perception, and thus there may be other early- developed skills that support later language. These will be discussed more in Chapter 3. These findings also have relevance for deciding on the best approach to language expo- sure for children who are born deaf. They suggest that early exposure to ASL should not hurt deaf children’s ability to acquire a spoken language and may in fact help it.

Changes in Domain-General Learning Mechanisms Developmental changes in the span of short-term memory have been proposed as an explanation of the observed changes in language learning success. Children can remember less than adults, and the argument is that this may actually help them. According to the less-is-more hypothesis (Newport, 1991), it is easier to figure out the structure of language if you analyze small chunks than if you analyze long stretches of speech. Small chunks are all that children can extract from input and store in memory. Adults, in contrast, extract and store larger chunks, thereby giving themselves a more difficult analytical task. This hypothesis has received support from a computer simulation of language acquisition done within the connectionist approach, which found that the computer was more successful if fed shorter sequences as input (Elman, 1993, 2001), although that finding has been disputed (Rohde &Plaut, 1999).

Another proposal is that the age-related decline in language-learning ability reflects age-related declines in the capacity for sensorimotor learning (i.e., learning that involves a sense modality and a motor skill) and in procedural memory. Procedural memory is the memory system that underlies learning motor and cognitive skills, as opposed to learning information (Hernandez & Li, 2007). This argument is consistent with evidence that other domains that involve sensorimotor learning and procedural memory, in par- ticular music, also show age-of-exposure effects.

Age-Related Changes in Opportunities to Learn Language Part of why children are more successful than adults at learning a new language may be the different

opportunities for language exposure available to children and adults. Children attend school in the new language, whereas adults must do work that their limited language skills allow, thus limiting their exposure to the new language. Jia and Aaronson (2003) studied Chinese immigrants to the United States who were between 5 and 16 years old during their first year in their new country and found that the children under 9 years were exposed to a richer English language environment than the older children. These very young children watched more English language television, were exposed to more English language books, and had more English-speaking friends. The older children had richer Chinese language environments. Although all the children reported they were more comfortable using Chinese than English when they first arrived, by the end of the first year, the younger children, but not the older, reported being more comfortable using English. The younger children also scored higher than the older children on tests of English proficiency.

It is possible, then, that it is not age-related changes in the brain’s capacity to acquire language that account for the widely observed age-of-exposure effects. Rather, it may be that switching the dominant language is what causes the differences. To investigate this dominant language switch hypothesis, Jia, Aaronson, and Wu (2002) gave grammati- cality judgment tasks in English and Chinese to native Chinese who had come to the United States between the ages of 1 and 38. As did Johnson and Newport, they found that younger arrivals performed better on the test of English grammar. They also found that the younger arrivals performed worse on the test of Chinese grammar. In fact, scores on the two tests were negatively related—the better a subject did on the English test, the worse he or she did on the Chinese test. This finding supports the hypothesis that better second language acquisition by younger children occurs at least in part because they switch to the second language as their dominant language.

Social and Motivational Factors Also contrary to a biologically based account of the advantage of youth, there is evidence that social psychological variables play a role in second language acquisition. Johnson and Newport (1989) found two such variables that affected the performance in English of their Korean and Chinese participants: self- consciousness and American identification. The participants who were less self- conscious about making errors and those who identified themselves as American showed greater mastery of English. Both the characteristics of not being self-conscious and of identifying with the new country are more likely to be true of children than of adults, and thus these nonbiological factors may also contribute to the observed age difference in second language acquisition.

In sum, the data on the effects of age on first and second language acquisition make it clear that languages learned earlier are learned better. Not only do early and late learners perform differently on tests of language proficiency, early and late learners also show dif- ferent brain activity associated with some language tasks (Mayberry, Chen, Witcher, & Klein, 2011; Werker& Tees, 2005; Ylinen, Shestakova, Huotilainen, Alku, &Näätänen, 2006).

The data do not, however, support two widely held beliefs that are often thought of as part of the critical hypothesis. One myth is that second language acquisition is impossi- ble after puberty. Lenneberg never actually said that, but rather that “automatic exposure to a given language seems to disappear after this age” (p. 176). In fact, many late learners become highly proficient in a second language (Birdsong, 2005), and the difficulty of acquiring a new language as an adult may reflect, in part, the difficulty of obtaining the necessary experience with a new language.

A second myth not supported by the evidence is that up until puberty children have an almost magical ability to acquire as many languages as they are exposed to. The

research suggests, to the contrary, that even when language acquisition is begun in child- hood, language acquisition begun after infancy yields a different outcome than language acquisition begun at birth. For example, studies of adult Spanish-Catalan bilinguals who began exposure to one language at birth and the other in early childhood have found differences in the phonological competence between their first and early-acquired second language (Bosch, Costa, &Sebastián-Gallés, 2000; Pallier, Bosch, &Sebastián-Gallés, 1997). Among the second language speakers of Swedish studied by Abrahamsson and Hyltenstam (2009), only a very few of the early learners showed native-like performance on all tasks. These authors conclude that it is “a myth that [second language] learning that begins in childhood, easily, automatically, and inevitably results in nativelikeness” (p. 290). They suggest that early exposure is a necessary, but not a sufficient, condition of ultimate native-like achievement. When we focus on bilingual development and sec- ond language acquisition in childhood, in Chapter 9, we will also see that although chil- dren are very good acquiring language, their abilities are not infinite.

The Genetic Basis of Language Development The Heritability of Individual Differences Heritability is a hallmark characteristic of biologically based traits. Studies of the herita-

bility of language ask how much of the variation in children’s language skills has a genetic, and therefore biological, basis. The statistical methods used in the study of heri- tability (i.e., behavioral genetics) are complex, but they all rely on the logic that if a char- acteristic has a genetic basis, then people should look similar to each other to the degree that they share genes. Twin studies are particularly important in the study of heritability because all twin pairs are born at the same time to the same family, but they differ in how much of their genetic material they share. Monozygotic (identical) twins share all their genes, or nearly so (Stromswold, 2006); dizygotic (fraternal) twins and siblings share an average of 50 percent of their genes.

Based on a meta-analysis of data from almost 100 twin studies that looked at children over 3 years of age, Stromswold (2001) estimated that heritable factors account for between 25 and 50 percent of the variance in language abilities within the normal range and between 50 and 60 percent of variance in language abilities among children with impaired language abilities. Stromswold (2006) also argued that twin studies may under- estimate genetic contributions because even monozygotic twins may differ genetically and in unmeasured aspects of the prenatal environment. For younger children, heritabil- ity estimates range between 1 and 38 percent of the variance, depending on the measure (Reznick, Corley, & Robinson, 1997). One study of thousands of youth (14 to 22 years of age) found that scores on language tests in this age range also showed a moderate degree of heritability (Hart, Petrill, & Kamp Dush, 2010). Another study of nearly 4,000 twin pairs found that the importance of genetic influences on language variation changes with age. Genetic factors accounted for just under a quarter of the variance in children’s language skills from 2 to 4 years, and more than half the variance from 7 to 12 years (Hayiou-Thomas, Dale, & Plomin, 2012).

Heritability estimates seem to differ consistently depending on whether grammatical or lexical development is being assessed. One large-scale twin study found that the heri- tability of grammatical development was 39 percent and the heritability of lexical devel- opment was 25 percent (Dale, Dionne, Eley, & Plomin, 2000). Another found that genetic factors account for 26 percent of the variance in syntax and 5 percent of the var- iance in vocabulary among normally developing twins (Stromswold, 2006). Perhaps relatedly, studies of environmental influences on language development have found

larger environmental effects on lexical development than on grammatical development (Arriaga, Fenson, Cronan, &Pethick, 1998; Hoff-Ginsberg, 1998).

In sum, individual differences in children’s language skills are to a degree genetically based. The importance of genetic factors differs depending on the age of the child and the aspect of language skill under study. There are several possible reasons that the heri- tability of language might change with age and outcome measure. Some aspects of lan- guage development, such as vocabulary, likely depend on language exposure more than others, such as grammar. Some aspects of language, such as complex sentence structure, may be more variable in children’s experience (Vasilyeva, Waterfall, &Huttenlocher, 2008). Different genes relevant to language may be activated at different points in devel- opment, and schooling may reduce the environmental variation and thus increase the heritability of skills in older children (Hayiou-Thomas et al., 2012).

The Genetics of Language Impairment

Some children acquire language slowly and with difficulty. We will discuss the nature of language impairments more fully in Chapter 11. The topic arises in this chapter because the study of language impairment has provided the most detailed information about the genetic basis of language development. It has been known for a long time that language impairment runs in families. Children who are language impaired are far more likely than typically developing children to have family members who are also language impaired (Stromswold, 1998; Tomblin, 1989). Monozygotic twins are more likely to be concordant for language disorders than dizygotic twins (Eley et al., 1999). Adopted chil- dren with language-impaired biological relatives are more likely to be language impaired than adopted children with no language impairment among their biological relatives (Stromswold, 1998). Language impairment appears to be more heritable than individual differences in language development within the normal range (Eley et al., 1999; Stroms- wold, 2006).

A very strong case for the genetics of language impairment was made by the dis- covery of a family (known as the KE family) in which 16 of 30 family members were seriously language impaired (Gopnik, 1990; Gopnik &Crago, 1991). The affected fam- ily members had both poor language abilities and severe difficulties with the motor aspects of speech production. (A connection between language production and oral motor skills has also been found in typically developing children [Alcock& Krawczyk, 2010].) The inheritance pattern in this family suggested that a single dominant gene was at work. Study of this family’s DNA, combined with studies of a few unrelated individuals who had similar language difficulties has allowed researchers to identify the gene involved (see Fisher, 2006, for more details on the gene hunt). It appears that the KE family has a mutation that affects the encoding of a particular protein known as FOXP2. The unrelated individuals have different genetic stories, but in all cases, the same gene is involved (Fisher, 2006; Tomblin et al., 2009). The FOXP2 protein affects the formation of neural structures that are important for speech and language.

At the time of the initial discovery of the FOXP2 gene, there were reports in the pop- ular press that described the FOXP2 gene as the “gene for speech.” This description is inaccurate in several ways. First, genes do not specify particular behaviors; that’s just not how genes work. Rather, genes “make regulatory factors, signaling molecules, recep- tors, enzymes, and so on” (Fisher, 2006, p. 270). Second, language depends on other genetically based characteristics besides those supported by this one gene. This gene appears to play an important role in the human language capacity, but it is not the only genetically based characteristic involved. Both the human language capacity and

molecular biology are more complex than that. Furthermore, the disorder present in this family is not like most cases of language impairment, and in most cases of language impairment, the data suggest that the cause is multiple genes, in interaction with the environment (Stromswold, 1998). In sum, work on the genetics of language development makes a case that individual differences in the ease and speed of language development have a genetic basis—both within the normal range of variation and in cases of atypical development. However, the story of how that genetic basis yields language is complex and, as yet, not fully known.

Language and Other Species

The Natural Communication Systems of Other Species

Other animals besides humans communicate with each other. If we want to know whether language is uniquely human, we need to ask whether these other communication systems should be counted as languages, too. To answer that question, we need to define language and then ask whether the communication systems of other species meet that definition.

What Constitutes a Language? Human language is a vehicle for communication, but the fact that an activity is interpretable does not make it language. Several definitions of language have been proposed (e.g., Bradshaw, 1993; Hauser, Chomsky, & Fitch, 2002; Hockett, 1960). Most include as crucial features at least reference (there are symbols that stand for things), syntax (there is a productive system for combining symbols to express new meanings), and intentionality (the speakers use language for the purpose of commu- nicating with others).

Communication Among Primates One of the more sophisticated communication systems among primates that has been studied is the call system of the East African vervet monkey. These monkeys produce a distinct alarm call for each of three different predators; there is a “leopard” call, an “eagle” call, and a “snake” call. Vervet monkeys also respond differently to these distinct calls. They run into trees when they hear the “leopard” call, they look up or run into bushes when they hear the “eagle” call, and they stand on their hind legs and look around when they hear the snake call (Seyfarth & Cheney, 1993). Seyfarth and Cheney have argued that these calls are more than expressions of excitement; they also serve to denote things in the environment. In the absence of any decontextua- lized uses, though, we may not want to credit the monkeys with the capacity for reference. In terms of the other criteria for language, Seyfarth and Cheney do not argue for syntax, and they make it clear that there is no evidence that the monkeys produce their calls with the intent to modify the mental state of their listener.

Our closer relatives among primates, chimpanzees, also communicate via calls, facial expressions, and gestures (Goodall, 1986; Marler&Tenaza, 1977), but nobody has argued that there is a naturally occurring communication system in primates equivalent to a human language. The strongest argument made is for continuity. As anthropologist Richard Leakey put it, the continuity view holds that “spoken language [is] merely an extension and enhancement of cognitive capacities to be found among our ape relatives” (Leakey & Lewin, 1992, p. 240). If this is true, then our ape relatives should have at least a rudimentary form of some of the language capacities seen in humans. A great deal of research on both the communication systems apes use and their learning capacities addresses this question of what potentially language-related abilities apes have, and it provides support for the notion of continuity (see Oller&Griebel, 2004; Zuberbühler, 2005). Exactly what capacity allowed humans to move farther along this linguistic and communicative continuum is a topic of ongoing research (see discussions in Tomasello, 2007).

The Birds and the Bees A different perspective on the notion of what a complex system of communication requires comes from the study of communication systems in other species that are very distant relatives of humans. We tend not to talk about conti- nuity with respect to birds or insects, yet when we look at birds and insects, we find extremely complex communication systems. Bees do not communicate by making noises; they dance. After a bee finds a source of nectar, it returns to the hive and does a dance that communicates the location of the food source to the other bees. Different dances are used for nearby versus distant food sources; and if the source is distant, the dance also indicates the direction of the food source. Richer food sources cause the bee to dance longer and harder, which, in turn, more strongly arouses the other bees (Von Frisch, 1962). As effective and sophisticated as this system is, it fails to meet virtually every cri- terion for being a language. It does exceed the primate systems that have been documen- ted as having some limited productivity. A bee can communicate a new message that has never been produced before (such as distant food at a 65-degree angle from the Sun), but it can communicate only the location of food sources. Thus, it does not have the vast productivity that characterizes human language.

Some species of birds use their songs to communicate. The relevance of bird song to language development is not so much in the properties of the song (but see Snowdon, 1993) but in how the song is acquired (Goldstein &Schwade, 2010; Gros-Louis, West, & King, 2010; Marler, 1970; Nottebohm, 1970). Not all birds are songbirds, and not all songbirds show the same developmental pattern. But in many species of song- birds, the development of the songs that males produce to attract mates and maintain territories requires exposure to adult birds who model the song. There are further par- allels between the acquisition of song in birds and language in humans. Both develop- mental courses have early stages prior to the appearance of the adult form—babbling in humans and what is termed subsong in birds. Both birds and humans need to be able to hear their own early productions for normal development, and deafening after acquisition does not have the same deleterious effect as deafening before acquisition. For both birds and humans, there are sensitive periods during which the ability to learn is at its maximum, and for both birds and humans, the learning of song or language takes place within a social system and is guided by feedback from others. Finally, the production of both song and speech is lateralized in the left hemisphere of the brain.

In sum, research on the communication systems of other species has revealed more complex communication systems in a number of species than many would have thought. And it is certainly true that the study of animal communication has contributed to defin- ing the criterial attributes of human language; lists much longer than reference, syntax, and intentionality have been proposed (Hockett, 1960; and see Bradshaw, 1993). Some might claim that revising the definition of language while you are asking whether another species has it is not quite fair. It’s something like raising the high jump bar as soon as someone gets close to clearing it. However, another way of looking at it is to say that the study of animal communication reveals, by way of contrast, what is unique about human language and if there were nothing unique, comparison to animal systems would reveal that, too. Raising the bar is precisely the way to find out whether one high jumper has an ability the others do not. If we set the bar at reference, syntax, and inten- tionality, only humans can clear it successfully.

The Acquisition of Human Language by Other Species

Just because no other species has anything equal to human language doesn’t mean another animal couldn’t acquire language if it were exposed to language in the right sort of supportive environment. This is the logic behind a set of efforts, undertaken

several times, to teach language to a member of another species. Like many other areas of language research, these animal language experiments have been the source of great controversy. Unlike many controversies, which are confined to academic circles, the animal language controversy has played out in newspapers, magazines, and television. The ape experiments constitute the majority of animal language experiments, and they are the most nearly successful. The meaning of the carefully chosen words “nearly successful” should become apparent in the next few pages. (For discussion of the linguistic capacities of dolphins and parrots, see Kako, 1999; Premack, 1986; Roitblat, Herman, &Nachtigall, 1993.)

Efforts to Teach Chimpanzees to Speak The first efforts to teach human language to a chimpanzee used spoken English as the target language. In the 1930s, the Kelloggs raised an infant chimp, Gua, in their home, along with their infant son Donald. The chimp wore diapers, slept in a crib, and was in every way treated like a human child. The result was that although Gua’s motor development outpaced Donald’s, only Donald learned to talk. In the 1940s and 1950s, another intrepid couple, the Hayeses, raised an infant chimp named Viki in their home, but, unlike the Kelloggs, the Hayeses actively tried to teach Viki to produce words. After six years, Viki could approximate the sounds of mama, papa, cup, and up. These efforts to get a chimpanzee to talk were clearly fail- ures, but it is not clear that these efforts were fair tests of the linguistic capacity of the species. Chimpanzees have a vocal tract that makes speech production essentially impos- sible. But the question of interest in these studies is not whether chimps have the articu- latory apparatus for speech but whether they have the brain for language.

Signing Apes The next efforts avoided the problem of speech and capitalized on chimpanzees’ manual dexterity by employing ASL as the target language. In 1966, Beatrice and Allan Gardner, faculty members at the University of Nevada in Reno, acquired a wild-born infant female chimp (Wallman, 1992). The chimp was named Washoe, after the county in Nevada where the Gardners lived. Washoe lived in a trailer in the Gardners’ backyard, and she was cared for by the Gardners and by University of Nevada students. Everyone who interacted with Washoe was instructed to use only sign language, both with Washoe and among themselves in Washoe’s presence. In addition, Washoe was actively taught signs by physically molding her hands into the proper shape and by drilling her and rewarding her for correct usage. After four years of this sort of language experience and language training, Washoe had learned to produce 132 signs and had been observed to produce many sign combinations. (At that point, Washoe grew rather large to handle in a trailer, and she was moved to the Institute for Primate Studies at the University of Oklahoma.) Washoe could correctly label a variety of objects and could sign MORE FRUIT, WASHOE SORRY, and PLEASE TICKLE.

It seemed at the time that some great chasm had been bridged. Humans were not only talking to animals, animals were talking to humans. In 1972, Francine Patterson, a graduate student at Stanford University, began a similar sign language project with a lowland gorilla named Koko. A Nova television program was made about these signing apes, and it is hard not to be amazed and impressed by the phenomenon of an animal producing a sign in a human language. Certainly, the Gardners and Patterson were impressed. They both have claimed that their animals learned a human language. Patter- son claimed that Koko not only understands “everything that you say to her” (meaning in English) but she also communicates via her sign language skills “about the way ani- mals view the world” (Patterson, 1985, p. 1, cited in Wallman, 1992). Patterson was also very media savvy, and Koko appeared on major network television programs and in National Geographic magazine. She even graced the pages of the Weekly Reader, a widely

ead newsletter for elementary school children. It is not surprising, then, that the belief that chimpanzees and gorillas can learn a human language is widespread.

However, careful analysis of just what the apes did suggests that the linguistic abilities of even our closest relatives are quite limited. During the 1970s, when stories of talking apes filled the airways and impressed enough psychologists to be reported in introduc- tory textbooks, there were always some dissenting voices (see Seidenberg &Petitto, 1979). But the true unmasking of the supposed linguistic accomplishments of apes came in 1979 from a group of researchers who set out, as the Gardners did, to teach ASL to a chimpanzee.

The chimpanzee that was the focus of this ASL project was named NimChimpsky, an allusion to well-known linguist Noam Chomsky. Project “Nim” was started by Herbert Terrace at Columbia University in New York City. (Terrace and Nim are pictured in Figure 2.8.) No backyards in Manhattan could accommodate a trailer, so Nim spent the first 21 months of his life raised in a private home, sleeping in a hammock in the dining room. For two years after that, he lived in splendor on the northern edge of New York City in a mansion that had been bequeathed to Columbia. As with Washoe, Nim’s care- takers used sign language in interactions with him, and they also actively molded Nim’s signs. Like Washoe, Nim learned more than 100 signs and produced many sign combi- nations. A sample of these is presented in Box 2.3.

Close examination both of Nim’s “sentences” and of the circumstances under which they were produced revealed that Nim’s language acquisition was very different from a human child’s. The first problem with Nim’s language can be seen simply by looking at the length and nature of the multisign utterances Nim produced. From the time that Nim started regularly producing sign combinations until his departure two years later, his mean length of utterance (MLU) hovered between 1.1 and 1.6 signs. Unlike children, whose MLUs increase with development, Nim did not increase his mean utterance length. To compare Nim’s changes in MLU with those of hearing children acquiring

FIG-2-8 Chimpanzees have been successfully taught to use signs of American Sign Lan- guage to communicate with humans. However, the chimpanzees’ accomplishments always fall short of full acquisition of the lan- guage. Just what the differences are between chimpanzee and human linguistic abilities and what they mean is a matter of considerable debate.

BOX 2.3 Examples of Sign Combinations Produced by the Chimpanzee Nim

Source: Terrace, 1979.

English and deaf children acquiring ASL, see Figure 2.9. Also, even when Nim produced a long utterance—as he sometimes did—it was highly repetitious. Children’s utterances get longer because children express more content in each utterance, but Nim’s long utterances tended to say the same thing over and over. This can be seen in many of the four-sign combinations in Box 2.3 and is abundantly clear in Nim’s longest-ever utter- ance, GIVE ORANGE ME GIVE EAT ORANGE ME EAT ORANGE GIVE ME EAT ORANGE GIVE ME YOU.

The other problem with Nim’s language is that he didn’t produce his utterances by himself. Close inspection of the videotaped interaction between Nim and his teachers revealed that Nim’s utterances were very dependent on his teachers’ previous utterances. In fact, 90 percent of Nim’s utterances were imitations, reductions, or expansions of prior utterances produced by his human interlocutor. The extent to which Nim’s utter- ances depended on the teachers’ signing is suggested by an interaction that happened to be captured in a sequence of still shots taken by an automatically advancing camera. In this sequence, Nim produced the multisign utterance ME HUG CAT. In the frame where Nim is signing ME, the teacher is signing YOU. In the frame where Nim is signing CAT, the teacher is signing WHO (Terrace, 1979; Terrace, Petitto, Sanders, & Bever, 1979). Thus, it appears that Nim produced the appearance of signing combinations with inad- vertent support from his teachers.

Having found this problem with their own chimpanzee, Terrace and associates then analyzed the publicly available tapes of Washoe and Koko (one a film produced by the Gardners and the other a show produced for public television), and they found the same phenomenon. The sign combinations Washoe produced were always preceded by a sim- ilar utterance or by a prompt from her teacher; all of Koko’s signs were signed first by the teacher (Terrace et al., 1979).

The researchers who worked with Nimcame to the conclusion that chimpanzees can- not acquire a human language. Syntax was not the only way in which the chimps’ use of ASL differed from the language competence of children. Although Nim had 125 signs, he used only a few regularly, and these tended to occur only in particular contexts. Basi- cally, Nim signed to request food and other objects. Also, Nim’s conversational use of signing was inappropriate. Unlike children who master turn taking even before master- ing language, Nim frequently signed while his teachers were signing. Although the lim- ited use of signs might be attributed to chimps’ limited range of interests, and although their lack of conversational skill might not be a fatal flaw, syntax and reference are criterial features of language. The analyses of Terrace and associates make it clear tha

FIG-2-9 Age-Related Changes in Mean Hearing

Children:

Utterance Length for Two Hearing Children, Three Deaf Children (Learning ASL), and One Chimpanzee (Also “Learning” ASL)

“Eve” (Brown, 1973) “Sarah” (Brown, 1973)

Nim

Deaf “Ruth” (H. Schlesinger, undated) “Pola” (Klima &Bellugi, 1972) “Alice” (Hoffmeister, 1972)

4.8 4.6 4.4 4.2 4.0 3.8 3.6 3.4 3.2 3.0 2.8 2.6 2.4 2.2 2.0 1.8 1.6 1.4 1.2 1.0

18 20 22 24 26 28 30 32 34 36 38 40 42 44 46 48 50 52 Age (in months)

Nim, Washoe, and Koko never achieved syntax. Laura Petitto, one of Nim’s teachers, argued that chimps also lack reference. She wrote:

For Nim, meaning seemed to have no role outside of the specific association between a form and its referent that had been explicitly taught to him. I had not succeeded in bring- ing him to the water fountain as Annie Sullivan had done for Helen Keller. For Nim, signs did not refer; he did not have words—signs or names—for things. (1988, p. 189)

Artificial Language Projects Other attempts to teach language to chimpanzees have used experimenter-invented languages. David Premack taught the chimp Sarah a system that made use of metal chips on a magnetized board. Each chip had an arbitrary color, shape, and meaning, and there were rules for ordering the chips. After long and arduous training, Sarah learned to do things like request an apple by producing a sequence such as “Mary give apple Sarah” and to correctly respond to the instructions “Sarah insert banana pail apple dish” (meaning put the banana in the pail and the apple in the dish). Because Sarah was never exposed to full human language, it is not fair to judge her accomplishments against that criterion. However, it is clear that what she did accomplish is less than language, and Premack’s subsequent work focused on Sarah’s and other

chimps’ cognitive, as opposed to linguistic, skills. At the Yerkes Regional Primate Center in Atlanta, Georgia, chimpanzees have been taught a language that uses a vocabulary of abstract symbols, called lexigrams, that can be combined according to a grammar that operates over semantically based categories. Lana, the first chimp taught this language (dubbed Yerkish), learned to produce and respond to sequences of symbols, but even her trainers made very limited claims for her accomplishments. What Lana learned was essentially a repertoire of rote-learned sequences associated with different situations and rewards (Savage-Rumbaugh, Murphy, Sevcik, &Brakke, 1993; Wallman, 1992).

Language in a Bonobo The next major development in the animal language controversy came when researchers at the Yerkes center began to work with a different species of chimpanzee—the bonobo. Bonobos seem more similar to humans than do the common chim- panzees that had been the subjects of previous experiments. For example, bonobos engage in upright posture more frequently, and they use eye contact, gesture, and vocalization in com- munication more frequently than common chimps do (Savage-Rumbaugh et al., 1993).In 1981, Sue Savage-Rumbaugh began to teach the lexigrams of Yerkish to an adult bonobo chimpanzee named Matata. However, Matata was a complete failure at learning symbol use. The surprise development that reopened the debate about the linguistic capacity of apes came from Matata’s son. While Matata was being trained, her infant son Kanzi was allowed to tag along; and without anybody paying any attention to him, Kanzi not only learned the lexigrams his mother failed to master, he also acquired some ability to understand spoken English. Kanzi’s accomplishments are the basis of claims made by Savage-Rumbaugh and her colleagues at Yerkes that chimpanzees—at least bonobos—are capable of both reference and syntax. In support of the claim for referen- tiality, Savage-Rumbaugh, McDonald, Sevcik, Hopkins, and Rupert (1986) cite Kanzi’s performance on a vocabulary test and examples of his lexigram use in naturalistic exchanges. In the vocabulary test, Kanzi did a good job of matching lexigrams both to pictures of the objects that the lexigrams stand for and to spoken words. In interaction with his trainers, Kanzi used the lexigrams. Although Kanzi clearly could do some things with the lexigrams he knew, the question is how to characterize what he did and the nature of the underlying knowledge. To illustrate, Kanzi would use the lexigram for strawberry when he wanted to go to the place where strawberries are found, when he was asking for a strawberry to eat, and when shown a picture of strawberries. On the basis of this sort of variety in the contexts in which the lexigram strawberry was pro- duced, Savage-Rumbaugh and associates argued that Kanzi used strawberry to refer. Seidenberg and Petitto (1987) disagreed, pointing out that outside the testing context, all Kanzi’s uses of strawberry resulted in his getting to eat strawberries. According to Seidenberg and Petitto, lexigrams for Kanzi functioned as instruments for achieving goals rather than as symbols, and true language involves the use of symbols.The argument that bonobos are capable of syntax rests on evidence from tests of Kanzi’s comprehension of spoken English (Savage-Rumbaugh et al., 1993). In this case, Savage-Rumbaugh and colleagues explicitly compared Kanzi’s comprehension abilities to those of a 2-year-old human child. Comprehension was tested by presenting the subject with commands, such as “Give Sue the bubbles,” “Put the rubber band on the soap,” “Go to the oven and get the tomato,” and “Go get the carrot that’s outdoors.” Care was taken that many different objects were available so that correct responding depended on sen- tence understanding. Many of the sentences were presented in blind trials in which the researcher communicated to the subject from behind a one-way mirror and thus could not inadvertently cue the correct response. The results showed that Kanzi did about as well as the 2-year-old child, producing correct responses to the first presentation on 59 percent of trials, compared to thechild’s 54 percent correct.

Tapes of Kanzi’s performance have been shown at academic conferences and on pub- lic television, and it is difficult not to be impressed by the sight of a chimpanzee responding correctly to somewhat improbable commands. Again, however, the problem lies in what to make of this accomplishment. Kako (1999) argued that although Kanzi’s comprehension abilities showed understanding of some features of syntax, they fell short of demonstrating mastery of all the criterial features of syntax. Tomasello (1994) argued, even more strongly, that none of the sentences presented to Kanzi requires much syntac- tic competence to figure out. The comparison with the 2-year-old child doesn’t help this problem, because, by many accounts, a 2-year-old child’s syntactic knowledge is quite limited. Ultimately, the problem is that we know the child will go on to acquire language and would, if tested at maturity, do better than 54 percent correct. Kanzi was 8 years old when he was tested, and 59 percent correct is probably as high a score as he will ever achieve.

Why Can’t Chimpanzees Acquire Language? What do we humans have in our brains that makes us the only linguistic species? What do chimpanzees lack that makes language unattainable for them? This is perhaps the most interesting thing to be learned from efforts to teach language to chimpanzees. Although different and contradictory answers to this question have been proposed, there is one point of surprising agreement: The problem is not that chimpanzees lack intelligence. Both observations of chimps in the wild and laboratory experiments suggest to many observers that chimpanzees are highly intelligent. As Seidenberg and Petitto (1987) put it, “Apes present a paradox: Why should an animal so demonstrably intelligent exhibit such dismal linguistic abili- ties?” (p. 284).

One possible explanation of why apes fail at language is that language is the expres- sion of a domain-specific mental faculty that humans have and apes do not. But if humans’ language-specific capacity is the capacity for syntax (the Universal Grammar argument and the bioprogram hypothesis would be versions of such a proposal), then we are still left with a paradox. The absence of innate Universal Grammar would prevent any other animal from fully acquiring a human language, but that absence wouldn’t explain why chimpanzees, for example, don’t do more with what they have. Premack (1986) argued that chimps have the conceptual ability to support a semantically based grammar of the sort often attributed to 2-year-old children. And some interpretations of Kanzi’s accomplishments would grant him that level of linguistic achievement. So why don’t bonobos in the wild have a communication system that makes use of a semantically based grammar?

One answer proposed by Savage-Rumbaugh and associates (1993) is that maybe they do. Maybe, their argument goes, if we looked more closely than anyone has so far at the naturally occurring communication system among bonobos, we would find something like a semantically based grammar. But the researchers also suggest that chimpanzees may be limited in language by their limited production abilities. Chim- panzees cannot produce the number of discriminably different sounds that humans can (nor can they produce the finely articulated gestures of fully competent signers). According to Savage-Rumbaugh and associates (1993), “Kanzi’s ability to understand human speech suggests that, if apes could produce human-like sounds, they might well invent and utilize a language that would be similar to our own, although probably considerably simpler” (p. 107).

Seidenberg and Petitto (1987) offered a different explanation of what humans have that chimpanzees lack. Remember, Seidenberg and Petitto did not credit Kanzi with reference, and they claimed that it is the inability to achieve that naming insight that accounts for chimpanzees’ linguistic limitations. In this view, the human language-specific capacity is

not just syntactic; it also includes reference. Chimps are incapable of either learning a human language or creating their own because they are incapable of understanding that things have names.

Another proposal is that what chimpanzees lack is culture. By culture, we don’t mean museums and the ballet; we mean culture as socially transmitted behavior. It may seem odd to bring this up now, near the end of a chapter on the biological bases of language, but language in humans is also a cultural phenomenon. It is definitely a socially transmit- ted behavior. Human language acquisition depends on the human capacity to learn from other people. Even human language invention depends on more than one participant. As Shatz (1994a) pointed out, social isolates—such as Genie or Victor, the wild boy of Aveyron—did not invent languages, although linguistic isolates who have potential com- municative partners do. A great deal of research suggests that the chimpanzee’s ability to learn through interaction with others—that is, for the social transmission of behavior—is extremely limited (Tomasello, Call, Nagell, Olguin, & Carpenter, 1994; Wrangham, McGrew, de Waal, &Heltne, 1994). Chimpanzees certainly imitate behaviors they observe, and chimpanzees can learn from human instruction. What chimpanzees do not seem to do is figure things out in collaboration with others (chimpanzee or human).

By this account, what keeps language out of the reach of chimpanzees is neither a lack of general intelligence nor the absence of a language-specific mental capacity; it is the lack of the social/cognitive ability to learn through interaction with others. By some accounts, the crucial component of this social/cognitive ability to learn from others (and also the ability to teach, which chimpanzees also seem not to do) depends on the ability to attribute mental states to others. Relatedly, Tomasello (2007) noted that apes may use their gestures to try to directly influence the behavior of other apes, but apes do not gesture just to pass along information. What chimpanzees may lack that accounts

“We’ll start out by speaking in simple declarative sentences.”

for their limited communicative intentions, then, is a theory of mind (Cheney, 1995; Worden, 1998). Having a theory of mind both allows and impels the child to read the communicative intentions of others, which, according to some, is the basis of language acquisition (Tomasello, 2003). Premack (1986) similarly argued that language is only one difference between humans and chimpanzees and concluded that language exists “as an instrument for consummating unique human social dispositions” (p. 155). At the risk of infinite regression, one can then ask where these unique human social dis- positions come from. One suggestion, that experience may have a role, comes from the observation that the more linguistically successful apes were also ones with early experi- ence being reared by humans. Perhaps something about human rearing encourages the social disposition that contributes to language development. Evidence in support of this view comes from studies find that apes reared with humans show communicative abili- ties and inclinations, such as using points to refer, that wild apes do not (Bjorklund & Rosenberg, 2005; Call &Tomasello, 1996). A final sort of proposal is that what is unique to humans is the combination of social interest in communicating and the cognitive abil- ity to acquire a system as complex as language (Kuhl & Rivera-Gaxiola, 2008). We will explore the cognitive and social basis of language development further in Chapter 3.

The Origin of the Human Capacity for Language

If language is a biologically based characteristic of the human species, then it has an evo- lutionary history. Just as we can ask how the giraffe got its long neck and how humans came to walk on two feet, we can ask how humans came to have the capacity for lan- guage. As Cosmides and Tooby (1994) put it, “The human brain did not fall out of the sky, an inscrutable artifact of unknown origin, … [rather, it acquired its] particular func- tional organization through the process of evolution” (pp. 85–86). The idea that we can learn about aspects of human psychology and human development by considering their evolutionary origins is the approach taken by the field of evolutionary psychology (see Barkow, Cosmides, &Tooby, 1992; Buss, 2005; Platek, Keenan, & Shackelford, 2007).

In the following section, we will briefly outline the positions that have been taken on the way the capacity for language evolved. As we will see, this last topic touches on two major issues that have appeared repeatedly in discussions of how children learn to talk. One is the issue of whether acquiring language is one thing that humans do with their general cognitive abilities or whether it reflects a domain-specific, modular ability. The other concerns the relation of language development to the social communicative func- tions that language serves.

Language as an Evolved Capacity

Two basic tenets of evolutionary theory have been invoked in proposals regarding the evolution of the human capacity for language. The first is natural selection. That is, spe- cies currently have characteristics that in the past gave members of that species an advantage in passing their genes on to the next generation. Such characteristics are termed adaptations. Thus, fish have streamlined bodies as adaptations to living in water, and giraffes have long necks as adaptations to living in an environment in which other animals eat the lower leaves, and so on. Evolution is not purposeful, but it selects features because they serve a useful purpose. Applied to the human capacity for lan- guage, the adaptationist view holds that humans have language because having it gave some of our hominid ancestors an advantage in survival and reproduction over those who did not have language. The other tenet of evolutionary theory is descent with modification. Evolution does not start from scratch; rather, current characteristics of species came into being through the gradual modification, over generations, of earlier

characteristics. Applied to human language, this view holds that the human capacities that underlie language should have precursors in the capacities of our nonlinguistic ancestors. This view is consistent with the notion of continuity between nonhuman com- munication systems and human language.

Current proposals regarding the evolution of language differ in the attention they pay to these two principles, some focusing on language as an adaptation and others focusing on language as the result of descent with modification. Another way in which current proposals differ is in the degree to which they see the language capacity as a separate module or as relying on cognitive capacities that also serve nonlinguistic cognition. There are also proposals that language is not an evolved capacity in the usual sense of the term. We review these proposals next.

Language as a Module and an Adaptation According to one view, language is a complex, special-purpose mechanism, a modular language faculty, selected for in the course of evolution because it conferred an advantage on those members of the species who had it (L. Bloom, 1998; Pinker & Bloom, 1990). This view is in accord with the standard neo-Darwinian account of how other complex systems evolved. The eye evolved for vision, and the heart for pumping blood, so why not parts of the left cerebral hemi- sphere for language? This account of language as a special-purpose adaptation is also consistent with the evolutionary psychology approach, according to which all mental capacities are special-purpose faculties designed by evolution to serve specific functions. The mind, according to this view, is not some generally useful, all purposetool. Rather, the mind is like a Swiss army knife with many different special purpose tools. (For an argument as to why evolution would result in that sort of mental organization, see Barkow et al., 1992.)

Language as the Modification of Other Cognitive Capacities It is possible, however, to accept the insight from evolutionary psychology that human abilities have an evolutionary history without accepting that the result of evolution is necessarily a bundle of different special-purpose devices. Barring some huge mutation, evolution has to work with the material that is there. This is the principle of descent with modification (Marcus, 2006) or, as Elizabeth Bates put it, language is “a new machine built out of old parts” (Bates, Thal, & Marchman, 1991, p. 35). Even if language conferred an advantage on those who had it, and they survived to pass on their genes in greater numbers because of it, language did not necessarily emerge from this evolutionary process as a wholly domain-specific capacity. According to some versions of this view, there is no uniquely linguistic capacity. According to others, there is some mental capacity that serves only language, but language also depends on capacities that support other cognitive functions as well. It has been suggested that the human language faculty can be thought of in a narrow sense—as just the abstract grammar and the computational system necessary to employ it—or in a broad sense—as also including the conceptual system that provides the meanings, the intentional system that underlies communication, and the sensorimo- tor system involved in perception and production. In this view, the evolutionary story and the uniqueness of language to humans could be different for the narrow-sense and broad-sense components of the language faculty (Hauser, Chomsky, & Fitch, 2002). This may seem like a compromise proposal, but it is actually quite controversial because it involves a claim about what mental capacities are both necessary and sufficient for lan- guage (see Pinker &Jackendoff, 2005).

A separation of the lexical/semantic component from the syntactic component of lan- guage is consistent with some accounts of the evolutionary history of language. It has been suggested that a lexical semantic system appeared 200,000–300,000 years ago

which is when the fossil record provides evidence of enlargement of the temporal lobes. The mutation of the FOXP2 gene that supports the structures underlying syntax may be as recent as 10,000–100,000 years ago (Ardila, 2009). Both of these components of lan- guage may be more recent evolutions than the intentional system that language serves. That is, in order for language to have evolved at all, it must have conferred a survival and reproductive advantage to those who had it. Language is useful only to a species whose members are interested in communicating with one another. A system as complex as human language is more useful than calls and hoots only if the interacting members of the species are interested in exchanges of information more complex than food loca- tions and predator warnings. The readily observable facts that humans are extremely interested in talking about other humans and that managing interpersonal relations is at the core of managing human society may be what gives human language its particular characteristics. Thus, both the modular adaptationist accounts of the origin of language and the descent with modification accounts assign a central role to the social nature of our hominid ancestors (Hurford, Studdert-Kennedy, & Knight, 1998). (For an argument on the way the demands of human communication require the structural complexity of human language, see Pinker & Bloom, 1990.)

Language as Emergent from Other Cognitive and Social Capacities The idea that language emerged in the species when a variety of cognitive abilities and social incli- nations came together suggests a different way of posing the question about the origins of language in the species. Rather than asking what evolved to allow humans to have language, the question should be how language evolved to fit human cognitive capacities and social needs (Chater& Christiansen, 2010). According to this view, language is emergent in the social interactions among members of a species who share the same per- ceptual and cognitive abilities and limitations. Language evolution was shaped by those shared abilities and limitations. An interesting implication of this approach is that it makes explaining language evolution and explaining language acquisition the same task. The task is to specify the relevant human characteristics and the mechanism by which their interaction resulted in both outcomes—the emergence of language in the species and the acquisition of language by the child (Beckner et al., 2009).

Language as a By-Product of Evolution

The fact that we use some part of our brain or our anatomy to serve a particular func- tion doesn’t mean that that physical structure was selected for, in evolution, to serve that function. Such a view has been dubbed “Panglossian,” after Voltaire’s fictional character Dr. Pangloss, according to whom “Everything is made for the best purpose. Our noses were made to carry spectacles, so we have spectacles” (quote from Eldredge, 1995). The position that humans have language because language was selected for through evolu- tionary history has been criticized as Panglossian (Gould &Lewontin, 1979).

The alternative view claims that the capacity for language is not the result of evolution but rather is a by-product of other evolutionary changes. The capacities that underlie language were selected for other purposes and then recruited for language—much the way our nose has other reasons for its shape, even though it is useful for supporting eye- glasses. Similarly, there may have been a variety of selection pressures that operated to increase the size and power of the brain (a generally better brain being a generally useful thing to have). Then, having gotten so much larger in the service of general improve- ment in its old functions, the brain was also able to perform new functions. This view of language as a by-product of design for other purposes has been suggested by Gould and Lewontin (1979). Interestingly, Chomsky has suggested something similar. Although Chomsky certainly rejected the idea that language could be served by general cognitive

capacities, he did suggest that language may well be the result of changes that occurred for other reasons:

These skills may well have arisen concomitant to structural properties of the brain that developed for other reasons. Suppose there was selection for bigger brains, more cortical surface, hemispherical specialization for analytic processing, or many other structural properties that can be imagined. The brain that evolved might well have all sorts of special properties that are not individually selected; there would be no miracle in this, but only the normal workings of evolution. We have no idea, at present, how physical laws apply when 1010 neurons are placed in an object the size of a basketball, under the special conditions that arose during human evolution. (1982, p. 321)

Summary

This chapter considered language development as a biological phenomenon. The reasons for characterizing language development as a feature of human biology and some of what we know about the biological bases of language development include the following:

Language is a universal characteristic of the human species. Not only do all human societies have lan- guage, but in situations where there is no target language to learn, humans in interaction will spontaneously create language.
The capacity for language is served by physical structures (in the vocal tract and in the brain) that seem, to a certain extent at least, to be specifically dedicated to their linguistic functions. In most mature adults, language functions are carried out primarily by the left cerebral hemisphere.
The brain appears to be biased from birth to rep- resent language in the left hemisphere. In child- hood, however, the mapping of language function to brain location is different and more diffuse than it is in adults.
The development of new techniques for imaging children’s brains has led to the development of a new field, developmental cognitive neuroscience, and the study of the neural underpinnings of lan- guage development is central topic in this field.
Although many myths are associated with the critical period hypothesis, there is clear evidence

that language acquisition is more successful when it is begun early than when it is begun late, and age of exposure has effects even within childhood. The age boundaries for the sensitive period for language acquisition may differ for different components of language. There are probably multiple bases for the observed age-of-acquisition effects, including the declining plasticity of the brain, effects of early tuning of speech perception, and age-related dif- ferences in learning opportunities and motivation. Individual differences in language skill among typically developing children are, to a degree, genetically based, and language impairment has a strong genetic component. One gene that plays a role in language development, FOXP2, has been identified, but it is clear that multiple genetically based capacities contribute to language.

The human capacity for language is species spe- cific. The results of research on the naturally occurring communication systems of other animals and the results of experiments that have attempted to teach a human language to another primate suggest that language is uniquely human. Most interesting, the comparisons of the human’s capacity for language to the ape’s capacity for lan- guage begin to provide a basis for hypotheses about the nature and evolution of the uniquely human characteristics that account for language.

Key Terms

pidgins, p. 32 creole, p. 32 language bioprogram hypothesis,

34 supralaryngeal vocal tract, p. 34

functional architecture, p. 35 neurolinguistics, p. 36 cerebral cortex, p. 37 subcortical structures, p. 37 corpus callosum, p. 37

contralateral connections, p. 37 ipsilateral connections, p. 37 neural circuits, p. 37 neurons, p. 37

lesion method, p. 37