Homework answers / question archive / California’s Best Practices for YOUNG DUAL LANGUAGE LEARNERS Research Overview Papers GOVERNOR’S STATE ADVISORY COUNCIL ON EARLY LEARNING AND CARE ? SACRAMENTO, 2013 California’s Best Practices for Young Dual Language Learners Research Overview Papers Governor’s State Advisory Council on Early Learning and Care Sacramento, 2013 Publishing Information California’s Best Practices for Young Dual Language Learners: Research Overview Papers was prepared under the direction of the Child Development Division, California Department of Education (CDE), for the State Advisory Council on Early Learning and Care

California’s Best Practices for YOUNG DUAL LANGUAGE LEARNERS Research Overview Papers GOVERNOR’S STATE ADVISORY COUNCIL ON EARLY LEARNING AND CARE ? SACRAMENTO, 2013 California’s Best Practices for Young Dual Language Learners Research Overview Papers Governor’s State Advisory Council on Early Learning and Care Sacramento, 2013 Publishing Information California’s Best Practices for Young Dual Language Learners: Research Overview Papers was prepared under the direction of the Child Development Division, California Department of Education (CDE), for the State Advisory Council on Early Learning and Care

Writing

Share With

California’s Best Practices for YOUNG DUAL LANGUAGE LEARNERS Research Overview Papers GOVERNOR’S STATE ADVISORY COUNCIL ON EARLY LEARNING AND CARE ? SACRAMENTO, 2013 California’s Best Practices for Young Dual Language Learners Research Overview Papers Governor’s State Advisory Council on Early Learning and Care Sacramento, 2013 Publishing Information California’s Best Practices for Young Dual Language Learners: Research Overview Papers was prepared under the direction of the Child Development Division, California Department of Education (CDE), for the State Advisory Council on Early Learning and Care. This publication was edited by Faye Ong and John McLean, working in cooperation with Cecelia Fisher-Dahms, Education Administrator I, Quality Improvement Office, Child Development Division. It was designed and prepared by the staff of CDE Press for online posting, with the cover created by Tuyet Truong. The document was published by the Department of Education, 1430 N Street, Sacramento, CA 95814. It was distributed under the provisions of the Library Distribution Act and Government Code Section 11096. © 2013 by the State Advisory Council on Early Learning and Care All rights reserved ISBN 978-0-8011-1742-8 CDE Publications and Educational Resources For information about publications and educational resources available from the California Department of Education, visit http://www.cde.ca.gov/re/pn/rc/ or call the CDE Press sales office at 1-800-995-4099. Notice The guidance in California’s Best Practices for Young Dual Language Learners: Research Overview Papers is not binding on local educational agencies or other entities. Except for the statutes, regulations, and court decisions that are referenced herein, the document is exemplary, and compliance with it is not mandatory. (See Education Code Section 33308.5.) ii Contents A Message from the State Advisory Council Co-Chairs ................................................................... Acknowledgments.............................................................................................................................. Paper 1. Neuroscience Research: How Experience with One or More Languages Affects the Developing Brain ......................................................................................................................... Barbara Conboy Paper 2. Cognitive Consequences of Dual Language Learning: Cognitive Function, Language and Literacy, Science and Mathematics, and Social–Emotional Development........................... Catherine Sandhofer and Yuuko Uchikoshi Paper 3. Program Elements and Teaching Practices to Support Young Dual Language Learners .. Claude Goldenberg, Karen Nemeth, Judy Hicks, Marlene Zepeda, and Luz Marina Cardona Paper 4. Family Engagement in Early Childhood Programs: Serving Families of Dual Language Learners........................................................................................................................................ Linda Halgunseth, Gisela Jia, and Oscar Barbarin Paper 5. Assessment of Young Dual Language Learners in Preschool ............................................ Linda Espinosa and Vera Gutiérrez-Clellen Paper 6. Early Intervention and Young Dual Language Learners with Special Needs .................... Deborah Chen and Vera Gutiérrez-Clellen iii A Message from the State Advisory Council Co-Chairs We are pleased to present California’s Best Practices for Young Dual Language Learners: Research Overview Papers, a publication we believe will provide early childhood educators with valuable information on the most current research on the development of young dual language learners. This series of research overviews spans the disciplines of neuroscience, cognitive science, developmental psychology, assessment, educational research, family engagement, and special needs. Insights from the reviews informed the creation of the forthcoming California Preschool Program Guidelines, a publication that addresses how to provide high-quality, developmentally and individually appropriate preschool services for young children. In particular, information from the research reviews guided the formulation of best practices for supporting the learning and development of California’s young dual language learners. The first two research overviews focus on different aspects of dual language development. Paper 1 (“Neuroscience Research: How Experience with One or More Languages Affects the Developing Brain”) reviews basic neuroscience and developmental cognitive neuroscience research relevant to language development. The paper discusses the general process of learning language in the early years, whether learning one language or two or more languages. The paper also addresses differences in how the brain processes language when learning two or more languages, and concludes that these differences do not reflect delays or deficits, but rather are adaptations to the unique circumstances of learning two or more languages—which, in turn, can lead to developmental advantages when two or more languages are supported through enriched learning opportunities. Paper 2 (“Cognitive Consequences of Dual Language Learning: Cognitive Function, Language and Literacy, Science and Mathematics, and Social–Emotional Development”) summarizes the current research on the cognitive consequences of dual language development. The paper highlights how dual language learning affects general cognitive functioning, including executive control and memory, as well as areas of learning that have a strong cognitive component, such as language and literacy, mathematics, science, and social–emotional development. The next two papers focus on the preschool program, addressing programmatic elements, teaching practices, and collaboration with families. Paper 3 (“Program Elements and Teaching Practices to Support Young Dual Language Learners”) summarizes research on program iv elements and strategies that effectively support the learning and development of young dual language learners in preschool. The authors describe the elements of high-quality preschool that benefit all children and identify additional practices that specifically enhance the learning and development of young dual language learners. In particular, the paper discusses the importance of providing continuing support for children’s home language as they learn English, as home language proficiency is foundational for learning and development across all domains including English-language development. Paper 4 (“Family Engagement in Early Childhood Programs: Serving Families of Dual Language Learners”) reviews the many positive developmental child outcomes that are associated with family engagement. The authors underscore the importance of strengthening family engagement in preschool for children’s future learning and families’ continued participation in educational settings. The paper pays particular attention to how to foster family engagement with families of young dual language learners. Topics include addressing the bilingual and bicultural needs of families, developing warm and mutually respectful relationships, engaging in regular two-way communication, and approaching families with a strength-based perspective. The final two papers examine assessment, early intervention, and young dual language learners with special needs. Paper 5 (“Assessment of Young Dual Language Learners in Preschool”) focuses on the importance of accurate and valid assessment of young dual language learners’ development and achievement. The paper discusses the need to take into account linguistic, cultural, and background considerations when assessing young dual language learners. Two specific purposes of assessment are addressed: (1) observational assessment for instructional decision making and improvement, and (2) assessment for screening and referral of children who may have special needs. The paper also makes clear that assessments must be valid, reliable, and linguistically and culturally appropriate. It closes with guidance for practitioners on assessing young dual language learners, including a flowchart and assessment matrix with specific questions and suggestions that can guide teachers’ decisions. Paper 6 (“Early Intervention and Young Dual Language Learners with Special Needs”) addresses both the language development of young dual language learners with special needs and key considerations when choosing the language for intervention. The overview states that children with a range of special needs can learn more than one language. In fact, children with language disorders can apply their home language skills when learning a second language, which v in many cases results in a greater rate of learning of the second language. Of particular note, the authors found that the use of the home language in intervention does not slow the acquisition of the second language. As a set, these six research overviews reflect the most current research related to the learning and development of young dual language learners. They provide insight into how young dual language learners learn two languages, and also how they learn and develop in other domains. At the same time, the research summaries provide guidance to early childhood educators on how to support the learning and development of young dual language learners in preschool programs. We hope that these research overviews will be an invaluable resource for supporting the learning and development of young dual language learners in California’s preschool programs. PETE CERVINKA LUPITA CORTEZ ALCALÁ Co-Chair, State Advisory Council Co-Chair, State Advisory Council California Department of Social Services California Department of Education vi Acknowledgments California’s Best Practices for Young Dual Language Learners: Research Overview Papers was developed with the support of the California Department of Education and the leadership of WestEd project directors Peter Mangione and Ann-Marie Wiese, as well as project manager Katie Monahan, in collaboration with Linda Espinosa, the lead researcher for the project. This research project of the California State Advisory Council on Early Learning and Care was made possible through American Recovery and Reinvestment Act (ARRA) funds granted to California under the Improving Head Start for School Readiness Act of 2007, with the California Department of Education as the lead agency. The following individuals are gratefully acknowledged for their expertise and contributions to this project: Eugene E. Garcia, Arizona State University (reviewer for Paper 2 and Paper 3) Gisela Jia, Lehman College (contribution to Paper 1) Marlene Zepeda, California State University, Los Angeles (contribution to Paper 2) Advisory Group: California’s Best Practices for Young Dual Language Learners Linda Espinosa, Co-Principal Investigator, Center for Early Care and Education Research— Dual Language Learners, Frank Porter Graham Child Development Institute, University of North Carolina, Chapel Hill Oscar Barbarin, Tulane University Luz Marina Cardona, Cabrillo Community College Deborah Chen, California State University, Northridge Barbara Conboy, University of Redlands Claude Goldenberg, Stanford University Vera Gutiérrez-Clellen, San Diego State University Linda Halgunseth, University of Connecticut Gisela Jia, Lehman College Karen Nemeth, Language Castle LLC Catherine Sandhofer, University of California, Los Angeles Yuuko Uchikoshi, University of California, Davis Marlene Zepeda, California State University, Los Angeles vii Advisory Group—Field Representatives Karen Gonzales, SETA Head Start Antonia Lopez, National Council of La Raza Ofelia Medina, Alliance for a Better Community California Department of Education Camille Maben, Former Director, Child Development Division Cecelia Fisher-Dahms, Administrator, Child Development Division Karen Cadiero-Kaplan, Director, English Learner Support Division Meredith Cathcart, Special Education Division WestEd Peter Mangione, Project Co-Director Ann-Marie Wiese, Project Co-Director Laurel Stever, Director of Operations Amy Schustz, Program Associate Gina Morimoto, Project Coordinator Teresa Ragsdale, Program Assistant __________ Note: The names, titles, and affiliations of the individuals named in these acknowledgments were current at the time this publication was developed. viii Paper 1 Neuroscience Research: How Experience with One or More Languages Affects the Developing Brain Barbara T. Conboy, PhD, Associate Professor Department of Communicative Disorders University of Redlands 1 Over the past several decades, improvements in technology have increasingly allowed researchers to address questions regarding human brain and cognitive development. One type of research question concerns whether early experiences alter perception and informationprocessing systems in ways that impact children’s early learning and future outcomes. This paper focuses on the question of whether growing up with two or more languages, instead of only one, has small effects on brain function and, subsequently, on future language learning. A promise of developmental cognitive neuroscience research is that it can help inform best practices in education. However, a pitfall is that neuroscience research findings are sometimes misinterpreted by the public and by members of the educational community, and such misinterpretations may be used to justify questionable educational practices. The goal of this paper is to review some key findings from basic neuroscience and developmental cognitive neuroscience research that are relevant to the preschool education of young dual language learners (DLLs), and to provide the reader with a conceptual framework for understanding those findings. The research reviewed includes a growing body of studies that use measures of brain activity, which tap into the organization and functioning of language-relevant neural systems in ways that measures of young children’s behaviors cannot. Together with studies of children’s behaviors, including performance on tests and other structured tasks, the brain-imaging studies have led to the following conclusions: 1. Language experience affects the organization of the neural systems involved in learning, storing, processing, and producing language (i.e., there is evidence of structural and functional differences between the brains of monolingual learners and DLLs). 2. Dual language learning and use involve some different cognitive processes than single language learning and use. 3. The effects of language learning experiences on the brain facilitate and constrain further learning, and these experiential effects may be what are often referred to as “critical period” effects in second language acquisition. In addition, this paper will attempt to provide a realistic view of how practitioners and policymakers can translate evidence from neuroscience research to best practices in the education of young DLLs. 2 A Conceptual Framework: Neuroconstructivism and Language Learning The neuroconstructivist approach to cognitive development (Karmiloff-Smith 2008; Mareschal et al. 2007; Westermann et al. 2007) provides a theoretical framework for thinking about relationships between children’s early language experiences (such as dual language learning), children’s skills and behaviors in each language, and brain development. In contrast with frameworks that pose questions regarding the relative roles of “nature” and “nurture” in various aspects of development, neuroconstructivism assumes that a dynamic interplay between nature and nurture influences both the structure and the functional organization of the brain throughout development (for similar approaches, see Diamond 2009; Gottlieb 2007; Johnson 2000). Moreover, neuroconstructivist and related approaches do not assume that there is a unidirectional causal path from genes, to the maturation (i.e., physical growth) of particular brain areas, to the use of those brain areas for cognitive behaviors such as language (see Paper 2, Development Across Domains, for more information about the relationship between dual language development and cognition). Such approaches do not assume that there is a biological clock that determines when the brain will be ready for particular aspects of language learning to take place or when a “critical period” for learning aspects of language ends. Instead, these frameworks assume that specialized systems for complex cognitive behaviors such as language emerge as a product of experience, and that there are bidirectional influences between genes, structural brain changes, and cognitive functions. According to this view, humans inherit a genetic blueprint that guides certain aspects of development, but the environment can influence which genes are expressed (this process is known as probabilistic epigenesis); genetic influences on brain maturation can also influence how much a child can take in, and thus learn, from his or her environment at various points in development; and previous learning can influence further learning by changing the brain’s structure and function (Gottlieb 2007). In sum, neuroconstructivism views brain development as being dynamic throughout the life span rather than ending after a particular end-state of maturation is reached. Cognition is thought to be constructed in a progressive manner, with new cognitive abilities based on previous, simpler ones. That is, development itself changes the ways that further development occurs (Mareschal et al. 2007). 3 This paper will review how the neuroconstructivist framework might be applied to thinking about the following dynamic aspects of dual language learning and use that have been noted in the scientific literature: • differences in brain structure and function between bilingual individuals (who control two different vocabularies, sets of speech sounds, and systems of grammatical rules) and monolingual individuals (who control only one language); • changes in brain structure and function that occur within individuals as they learn another language, or as they become more or less proficient in one of the languages that they already know due to increases or decreases in use over time; • different patterns of brain activity for processing each of the same individuals’ languages, with which they have different experiences and levels of expertise, even when the two languages are acquired in early childhood; • the apparent ease of learning a second language during early childhood compared with later childhood or adulthood (i.e., what is commonly known as “critical period” effects in second-language acquisition), as well as cases of individuals who reach native-like proficiency in a language that they begin to learn at later ages. As discussed by Mareschal and others (2007), the goal of neuroconstructivism is not to reduce complex cognitive behaviors or functions to descriptions at the neural level or to map functions to specific regions of the brain. Instead, the goal is to develop models of how aspects of cognition, such as language, might emerge in a brain that is embodied within a developing human, which in turn is embedded in a physical and social environment. Thus, cognitive behaviors, such as language knowledge and use, are thought to be influenced by external physical and social experiences. A central tenet of the neuroconstructivist approach is that knowledge becomes represented in the brain through the progressive elaboration of cortical structures. Thus, in order to consider how the neuroconstructivist framework might be applied to dual language learning and use, the reader needs to consider some key aspects of early brain development and how they might be influenced by the contexts in which many preschool DLLs develop. These are summarized in the next section. 4 Key Events in Brain Development Early brain development has been characterized as the product of progressive (i.e., additive) and regressive (i.e., eliminative) events that work together to shape the brain and that are essential for normal development. These events may be predetermined by genetics, may occur spontaneously, or may be influenced by events in the environment (referred to as “experience-dependent” aspects of brain development; for example, see Greenough, Black, and Wallace 2002). An organism’s experiences are mediated by neural activity, and this activity influences how some neural structures are added, eliminated, or become active (Mareschal et al. 2007). Examples of progressive events in brain development are neurogenesis, synaptogenesis, and myelination; regressive events include programmed cell death and synaptic pruning (see below, for further explanation). Another key event in brain development that may also be explained by neuroconstructivism is the specialization of networks of neurons for particular sensory, perceptual, motor, and cognitive functions (Johnson 2000). Practitioners should be aware of what is known, as well as what is not known, about how these events relate to particular experiences of young DLLs, such as the timing and contexts of first- and second-language learning. Practitioners should also be aware of the dynamic nature of brain development throughout the life span, and its implications for language learning. This section provides a brief overview of several key aspects of development in the cerebral cortex, the outermost layers of the brain most important for language and cognitive functions, and a discussion of the implications for dual language learning. (For more comprehensive reviews of the literature on human brain development, see Brown and Jernigan 2012; Clancy and Finlay 2001; Lenroot and Giedd 2006; Sowell, Thompson, and Toga 2004.) During brain development, neurons (nerve cells) are generated in regions called proliferative zones (neurogenesis); then migrate to particular destinations in the brain; differentiate into particular subtypes; and project extensions from the cell body called axons (which carry neural activity to other neurons by transmitting chemoelectrical signals), and branch-like dendrites (which receive activity from other axons). The gap where the axon of one neuron forms a connection with either the dendrite or cell body of another neuron (and where the firing of one neuron triggers the firing of another neuron) is called a synapse; synaptogenesis refers to the formation of such connections. Glial cells are supporting cells for neurons; some glial cells form fatty sheaths around axons called myelin, which increase the speed of transmission of signals 5 between neurons. The formation of myelin is known as myelination. Much of the process of neurogenesis and synaptogenesis occurs prenatally. Neurogenesis begins from the earliest moments of embryonic development, though it can continue throughout an individual’s lifetime (see below). During the second trimester of gestation, connections between the cerebral cortex and the thalamus (a region of the brain below the cortex that receives input from peripheral visual and auditory systems) are established (Clancy and Finlay 2001). A regressive event called programmed cell death also occurs, beginning in the second trimester, and serves to refine the populations of neurons that work together (Clancy and Finlay 2001). It has been estimated that approximately half of all neurons generated during embryonic development are later eliminated, although the factors that determine which neurons survive are not completely understood (Lenroot and Giedd 2006; Mareschal et al. 2007). During the third trimester, connections between neurons in higher and lower processing areas within the cortex are formed, and the process of myelination begins in frequently used pathways (Clancy and Finlay 2001; Lenroot and Giedd 2006). Although myelination is not needed for the transmission of activity between neurons, it makes transmission more efficient. Toward the end of gestation (the eighth to ninth month), acceleration in synaptogenesis, a massive formation of new connections, begins. Many of these developments take place spontaneously, that is, without the influence of learning or experience from the external environment, but there is evidence that by the third trimester, the brain has begun to develop in response to sensory inputs, a form of “talking back to the body” (Clancy and Finlay 2001). In neuroconstructivist terms, the developing human brain can be thought of as being “embodied” within the fetus (Mareschal et al. 2007), which by the third trimester has already begun interacting with its environment by sensing sounds and other inputs. By the time an infant is born, his or her brain possesses most of the neurons it will ever have, and major sensory pathways have already become organized to process input and are thus ready to learn from the external environment. Yet the brain is far from fully formed at birth. Brain development continues throughout childhood and into adulthood (Casey, Giedd, and Thomas 2000; Casey et al. 2005; Lebel and Beaulieu 2011; Shaw et al. 2008; Sowell, Thompson, and Toga 2004). Neurogenesis has been shown to occur throughout the lifetime in mammals in at least two areas of the brain, and one of these areas (the hippocampus, in the medial temporal cortex) is known to be important for certain types of memory formation (e.g., 6 Eriksson et al. 1998; Manganas et al. 2007; for a review, see Deng, Aimone, and Gage 2010). There is also evidence that new neurons are born and become functional in response to challenging learning situations in adulthood (van Praag, Kempermann, and Gage 2000; Shors et al. 2012; Zhao, Deng, and Gage 2008). The extent to which new neurons are created and become functional across different areas of the cerebral cortex of human adults remains a matter of scientific debate. Moreover, the extent to which new neuron formation and functionality are influenced by experience and learning is also not resolved (Deng, Aimone, and Gage 2010). Synaptogenesis also continues throughout the lifetime (for reviews, see Casey et al. 2005; Lenroot and Giedd 2006; van Praag, Kempermann, and Gage 2000). Experience-dependent synaptogenesis has been reported in studies of rodents reared in relatively “enriched” (as opposed to deprived) cages, and studies showing that multiple synapses form in adult animals in response to learning situations (Greenough, Black, and Wallace 2002; van Praag, Kempermann, and Gage 2000). Although it is obviously not possible to conduct studies that involve deprivation with humans, many of the causal links between experience and synaptogenesis that have been observed in other mammals are thought to also exist in humans. However, not all of synaptogenesis can be described as experience-dependent. For example, around the time of birth, there is an explosion in synaptogenesis across brain regions that does not seem to be caused by experience or learning. This overproduction of synapses leads to synaptic densities that exceed adult values within the first several years of a child’s life, and synaptic pruning (the gradual elimination of unnecessary synapses) is needed to allow the synaptic densities of these different areas to decline to their more “mature” or optimal levels, for the brain to function efficiently (Clancy and Finlay 2001). Synaptogenesis and synaptic pruning work together to enhance the functioning of the brain throughout childhood and even into young adulthood (Lenroot and Giedd 2006; Petanjek et al. 2011). Some practitioners may believe that having more synapses is necessarily “better” for language learning than having fewer synapses, and that for language learning to be optimal, it needs to occur during the time in development when the most synaptic connections are present, before synaptic pruning occurs. This belief may be, in part, based on an overly broad interpretation of the “use it or lose it” principle that has been demonstrated in neurobiological research (e.g., Shors et al. 2012). In fact, efficiency in language processing, and other aspects of cognitive development, is associated with the specialization and refinement of connections 7 between particular groups of neurons (see the next section for further details). Understanding how synapses are formed and strengthened is essential for understanding how the brain learns and represents information, such as speech sounds, words, and the rules of grammar. Synaptic Pruning • Most neurons are produced prior to birth, and connections between these neurons (synapses) begin to be formed prenatally. Soon after birth and during the first couple of years of life, an explosion in synaptogenesis (production of synapses) across brain regions leads to synaptic densities that exceed adult values. • Through synaptic pruning (the gradual elimination of unnecessary synapses that occurs throughout childhood), synaptic densities decline to adult levels. • Contrary to popular belief, there is no evidence that having more synapses is “better” for language learning than having fewer synapses. Efficiency in language processing is associated with the refinement of connections between particular groups of neurons that become specialized to work together. It is known that connections between neurons at the synaptic juncture are strengthened with repeated activity; this process, called Hebbian learning (Hebb 1949), is one way that areas of the brain come to work with each other, and it is thought to be how knowledge is stored in the brain (Clancy and Finlay 2001). The principles of Hebbian learning and competition between neurons for connectivity are sometimes expressed in the phrases “‘Cells that fire together, wire together’ and ‘those that don’t, won’t’” (e.g., Penn and Shatz 1999). Though few studies have directly linked measures of brain connectivity to aspects of cognitive development in humans (Casey et al. 2005; Fair et al. 2009), and though the actual relationships between the progressive and regressive events of early brain development and language development milestones are not yet understood, it is clear that much of early brain development takes place during the same time frame when many important developmental milestones in first-language acquisition occur (the beginnings of speech perception and production, learning of the first words, and the beginning use of sentences; see Centers for Disease Control and Prevention 2012, http://www.cdc.gov/ncbddd/actearly/milestones/index.html). First-language experience during the first few years of life is necessary for setting up a brain that can process a language efficiently (Boudreault and Mayberry 2006; Mayberry, Lock, and Kazmi 2002). However, there 8 is no definitive scientific evidence that second or subsequent languages also need to be learned during this same period in order for the brain to process those languages efficiently; this point is discussed in a later section of this paper. Many areas of the brain are mature in structure and metabolic activity by the time of birth; this is known as “absolute functionality” (Clancy and Finlay 2001). For example, the auditory cortex, in the temporal lobe of the brain, can process speech and other complex sounds, even prior to birth, and late-term fetuses can remember some auditory information (Kisilevsky et al. 2009). Thus, it might be said that this region of the brain is equipped to process language from birth. However, it is important to recognize that several factors can prevent areas that have absolute functionality from being recruited for tasks in the same way that mature adult brains use those areas for the same tasks (i.e., “task-specific functionality”; Clancy and Finlay 2001). The task-specific functionality of the auditory cortex may be limited by how this region communicates with other brain regions, which may be limited by immaturity of those other brain areas. The rate of synaptogenesis peaks and then declines earlier in posterior regions of the cortex than in frontal regions; thus the more posterior regions, including the auditory cortex, are said to “mature” earlier than the frontal regions (Huttenlocher and Dabholkar 1997). Additionally, myelination, whereby regularly used pathways become surrounded by fatty sheaths (myelin) that assist in the transmission of impulses between neurons, is a protracted process that continues into adulthood, and occurs at different rates across regions of the brain (Casey et. al. 2005). There is evidence that the efficiency of language processing relies on the myelination of axons that connect language areas in the brain (Aslin and Schlagger 2006; Nagy et al. 2004; Pujol et al. 2006; Sowell, Thompson, and Toga 2004). Task-specific functionality may also be limited by the brain’s chemical properties. Different areas of the brain contain unique combinations of chemical substances necessary for synaptic functioning (neurotransmitters); information transfer in the brain occurs when these chemical substances are released from a neuron, across the synapse and onto the receptors of another neuron (Clancy and Finlay 2001). Several neurotransmitters have been linked to learning and developmental disorders (e.g., Hasselmo 2006; Marshall and Born 2007; Wise 2004). The distributions of neurotransmitters change with brain maturation, and, although direct mappings have not yet been discovered, it has been hypothesized that developmental changes in the amounts and types of neurotransmitters available for synaptic functioning within particular brain regions may affect the timing of 9 language acquisition (Clancy and Finlay 2001). Differences in maturation across regions may therefore limit how populations of neurons in those regions work together. For example, research using brain imaging has suggested that left-hemisphere language areas (“Broca’s area,” in the inferior region of the frontal lobe, and “Wernicke’s area,” in the superior region of the temporal lobe) begin to become functionally connected for speech perception between six and 12 months of age, as infants gain experience listening to and producing speech sounds (Imada et al. 2006; see below for further description). Another factor that can limit the task-specific functionality of brain areas is the physical environment. Studies of fetuses and pre-term infants have shed light on this issue, because they show to what degree the fetal environment versus brain maturation limits learning. For example, by the third trimester of gestation, the auditory sensory and perceptual areas of the fetal brain are sufficiently developed to detect fine-grained distinctions between speech sounds (e.g., the difference between the vowel sounds “ah” and “ee”), yet fetuses cannot hear all of these differences in speech sounds due to the presence of amniotic fluid in the uterus, which filters out sounds with frequencies higher than 500 Hz (hertz, or cycles per second) (Kisilevsky et al. 2009). Studies of infants born pre-term have shown that the brain can encode distinctions between speech sounds during the late gestational period (i.e., 36 to 40 weeks gestation), but in full-term infants who are still in the uterus at this gestational age, the ability to encode speechsound distinctions is limited by the fetal environment, which filters out important acoustic information. Cheour and colleagues (1998) used an auditory “oddball” mismatch experiment to test newborn pre-term infants’ discrimination of two speech sounds: one served as a standard stimulus (repeated at constant intervals) and the other as a deviant stimulus (randomly presented 10 percent of the time within the stream of standards). Infants born preterm (30–35 weeks conceptual age at the time of testing) showed a discriminatory response to the deviant sounds in the form of larger amplitude voltage to deviants versus standards from 200 to 500 milliseconds (ms) after the onset of the syllable. This response was similar, though not identical, to that shown by full-term newborns, three-month-old infants, and adults using the same sets of sounds. It is important to note that the acoustic difference between the speech sounds used in this research was in the high frequencies (above 1700 Hz). Typically developing 30- to 35-week-old fetuses would not be exposed to frequencies above 500 Hz (i.e., the frequencies contained in many speech sounds) in utero because of the presence of amniotic fluid, but the fetuses’ brains 10 are already able to process the full range of speech sounds. Thus, even before infants are born they are ready to perceive fine-grained distinctions between speech sounds, but the physical environment can have an important effect on how development proceeds. Experience Affects the Organization of the Neural Systems Involved in Learning, Storing, Processing, and Producing Language Given both the experience-dependent and experience-independent (i.e., maturational) aspects of brain development summarized in the previous section, differences across children in brain structure and/or function are expected, and such differences are also expected to lead to variability in future learning. A genetic factor that alters some aspect of early brain development (for example, one that causes a language learning disability) could diminish an individual’s readiness to learn from the environment and thus affect subsequent experience-dependent aspects of brain development, further diminishing learning and widening the gap between the affected individual and peers. At the same time, even a typically developing brain could be altered given small variations in experience, such as amounts or types of language input. Thus, neuroconstructivism predicts that, in any group of learners (monolingual, bilingual, or multilingual), differences in experience will affect language functioning and subsequent learning. When children are learning two languages, differences in experience with each language could lead to slightly different patterns of activity for the functioning of each language, affecting subsequent learning in each language. Other differences in early childhood experiences that are relevant to understanding the brain development of DLLs are those cultural influences that shape cognitive processing, and the effects of poverty. Neuroconstructivism predicts that culturally determined patterns of language input to children could lead to different patterns of brain activity and hence language learning. For example, Rogoff and colleagues (2003) have reported differences in attentional abilities and nonverbal communication between children from indigenous communities and children from Western communities, and they have also documented that parent educational level modulates these differences. Neuroconstructivism also predicts that any environmentally caused disruptions in normal brain development may hinder language learning, and subsequently affect the ways in which language is processed by the brain. Children living in poverty are more likely than children from middle-income homes to experience a range of risk factors that can affect brain development (Dilworth-Bart and Moore 2006; Farah et al. 2006; Hackman and Farah 2009; 11 Huston and Bentley 2009; Raizada and Kishiyama 2010). Raizada et al. (2008) found that when five-year-old children performed a rhyming task, the left inferior frontal gyrus (IFG)—an area known to be important for language processing in the adult brain—was active. However, there were differences in the activity levels across children of lower and higher socioeconomic status (SES). Children with higher SES were more likely to show specialization of function limited to the left IFG than were children with lower SES, who showed engagement of both the left and right hemisphere IFGs. Functional specialization to more narrow brain regions is typically viewed as a sign of brain efficiency. Thus, the functional specialization of the higher SES children was interpreted as reflecting the relatively stronger language skills that higher SES children tend to have, which are associated with more efficiency in how the brain processes language. Moreover, the study showed SES-related differences in the structure of the brain, both in white matter (myelin) and gray matter (neurons) in the left IFG of individual children. The neuroconstructivist framework thus accounts for many of the experience-dependent aspects of brain and language development that have been documented in infants and young children, and leads to additional predictions regarding the brain development of DLLs. The next section provides a brief description of the brain measures that are used with infants and young children, and reviews how those techniques have been used to study how certain experiences with language affect the organization of the neural systems involved in language processing. Overview of Techniques Used to Measure Brain Function Much of the research on brain development in humans described in the previous section used noninvasive brain-imaging techniques, which are challenging to use with young children because they require study participants to remain fairly still and quiet during recording sessions. Nonetheless, functional brain measures are important complements to behavioral measures because they can detect learning without requiring an overt response from the study participant. Moreover, brain measures can discover subtle differences in processing that behavioral measures may not detect. In this section, some widely used techniques are described, and the application of those techniques to DLLs is discussed. Functional magnetic resonance imaging (fMRI) and event-related potentials (ERP) are the techniques used most widely in research on language processing (see Dehaene-Lambertz, HertzPannier, and Dubois 2006). The fMRI technique can be used to address the “where” question, in 12 that it has excellent spatial resolution and can pinpoint, on the order of millimeters, which regions are activated during particular aspects of language processing. In fMRI, hemodynamic (blood flow) responses of brain regions that occur several seconds after the firing of neurons indirectly measure brain activity. Because of this lapse, fMRI is not a direct measure of brain activity, and it does not measure brain activity on the order of milliseconds; however, it is remarkably good at localizing primary regions of activation (Heeger and Ress 2002). In contrast to fMRI, the ERP technique can be used to address the “when” question, in that it has excellent temporal resolution and can pinpoint, on the order of milliseconds, stages in the time course of processing units of language when particular aspects of language knowledge (e.g., sound, meaning, grammar) are accessed. ERPs are direct records of neural responses to stimuli, obtained by placing electrodes on the scalp to detect the electrical activity produced by neurons in response to particular stimuli. ERPs are obtained by averaging together numerous epochs (like snapshots) of electrical brain activity that are time-locked to the onset of a stimulus (e.g., the beginning of a sound or word); thus they yield information about language processing on a millisecond-by-millisecond basis. For example, as an individual listens to the word dogs, the electrical brain activity that is involved in processing the word’s sound structure (phonology), meaning (semantics), and grammatical properties (e.g., the plural s) can be recorded using ERPs. Because of the spreading of electrical potentials through the skull before they are recorded at the scalp, ERPs do not offer precise information about where processing occurs, though the location of sources may be inferred using mathematical algorithms (see Luck 2005, for discussion). Two other imaging techniques that are used with young children are magnetoencephalography (MEG), which measures magnetic fields created by neural activity at the scalp, and functional near infrared spectroscopy (fNIRS), which measures neural activity indirectly by assessing changes in hemoglobin levels using infrared light that is directed at the skull (see Mareschal et al. 2007). Like ERPs, MEG is ideally suited for studying language processing because it has the same high temporal sensitivity as ERPs, but at the same time, it provides better spatial sensitivity given that magnetic fields are not distorted before they reach the scalp (Hari, Levänen, and Raij 2000). However, MEG has high operating costs. Functional NIRS allows researchers to determine in which part of the brain activation occurs, but, like fMRI, it has poor temporal resolution. MEG and fNIRS are just beginning to be used in developmental studies (e.g., Aslin and Mehler 2005; Imada et al. 2006; Petitto et al. 2011). 13 To understand how the brains of DLLs process language, it is necessary to consider evidence about how the brain processes language in adults and in monolingual children, because few neural-imaging studies with young DLLs have been published. The next sections highlight some of the most relevant available research on these populations, focusing on the role of experience in setting up language-processing systems and considering what we know and do not know about how this process unfolds in DLLs. Experience and the Functional Specialization of the Brain during Language Learning As discussed above, experience with language and other forms of sensory input are critical for setting up the brain’s language-processing systems. In addition to affecting the formation and elimination of neurons and synapses, experience influences how brain areas become organized for language functions in specialized networks. The idea that there are “functional specializations for language” does not mean that activation occurs only in one area or that separate brain areas operate in isolation of one another. Brain areas are interconnected; they cannot be studied in isolation and should not be thought of as a collection of boxes (Mareschal et al. 2007, 50). Studies of adults with language deficits due to focal brain injury such as stroke (called aphasia) and functional imaging studies with healthy adults are important for understanding how different regions of the brain are involved in language functioning (for a review, see Hickok and Poeppel 2004). Studies have shown that symptoms of language problems do not always map onto lesion sites in a clear, one-to-one fashion, which suggests that language processing relies on wide networks across the brain, rather than on discrete regions (Dronkers et. and al. 2004). Although left-hemisphere specialization for language is the most typical pattern in adults, it is not observed in all adults. Moreover, right-hemisphere areas have been shown to be important for certain aspects of language processing and language acquisition, and appear to be used to a greater extent in bilingual adults who acquired their second language during early childhood compared with those who were later second-language learners (Hull and Vaid 2007). Thus, although there are differences between bilingual and monolingual individuals in the extent of right-hemisphere involvement in language processing, these appear to be differences in degree. It is not yet clear why there is greater right-hemisphere activation for bilinguals who acquired both 14 languages early, but it is hypothesized that the effect is linked to the cognitive processes involved in processing two languages early in development (Conboy and Mills 2006). “Specialization” of Language Functions • The term functional specializations for language refers to the involvement of particular brain regions for particular language tasks but not other tasks. This does not mean that activation is confined to only one area or that separate brain areas operate in isolation of one another. • Neurons in the left and right hemispheres are involved in language processing, but different neurons are involved in different language functions (e.g., semantics, syntax phonology). • The extent of right hemisphere involvement in language processing varies for bilingual versus monolingual individuals, and early versus late bilinguals; these differences are hypothesized to be linked to the cognitive processes involved in processing two languages early in development. • There are differences across bilingual and monolingual children in functional specializations for language, which should not be interpreted as evidence of a delay induced by bilingualism but rather as a distinct developmental pattern linked to experience with each language. Adult patterns of functional specialization for language are not observed in young children. The functions associated with particular brain areas change during development, and children show a process of moving toward adult patterns of cerebral specialization. For example, unlike adults who sustain brain injury, children who sustain focal brain lesions prenatally or perinatally exhibit significant delays in early language development regardless of whether the lesion was in the left or right hemisphere (Thal et al. 2004). By school age, these children tend to have language test scores in the normal range but continue to show subtle deficits in language (Bates and Roe 2001). These results suggest not only that brain injury disrupts normal language development, but also that children’s brains are less specialized than adults’ brains (a characteristic known as plasticity). The neuroconstructivist framework provides a way of understanding how the infant pattern becomes adult-like. Given that left-hemisphere specialization of language functions has been noted in most adults, there must be an innate bias for left-hemisphere regions to process language. Studies of newborn infants using ERPs, fNIRS, and fMRI have, in fact, indicated that left-hemisphere temporal areas are biased for processing 15 auditory stimuli that have rapidly changing acoustic information, such as speech sounds (e.g., Dehaene-Lambertz, Hertz-Pannier, and Dubois 2006; Peña et al. 2003). However, an innate bias does not necessarily lead to an adult pattern. The early activity in left-hemisphere regions may be shaped by early experiences and result in the left-hemisphere specialization commonly seen in adults. The use of different strategies for processing speech sounds throughout development may be one reason that different patterns of functional specialization are observed at different points in development. For example, over the first year, there are shifts in how speech sounds are processed. Behavioral studies, in which participants are trained to provide an overt response such as turning the head when a sound changes, have indicated that infants can distinguish different speech sounds of both their native language and nonnative languages at six to eight months, but only those from their native language at 10 to 12 months (Werker and Tees 1984). This pattern of developmental change linked to language experience has been replicated in numerous behavioral studies and, more recently, in brain-imaging studies (for reviews, see Cheour, Leppanen, and Kraus 2000; Conboy et al. 2008; Fava, Hull, and Bortfield 2011). In the ERP studies, brain discriminatory responses have been measured using an “oddball paradigm” in which an infrequent deviant speech sound (e.g., “ee”) is inserted into a stream of frequently repeated standard stimuli (e.g., “ah”) (described above). The brain’s detection of the change in stimulus is reflected in increases in the amplitude of the neural activity to the deviant versus standard stimulus at a particular point in the time course of processing the sound. That is, the brain responds with more neural activity when it detects a change from the ongoing stream of repeated sounds (“ah-ah-ah-ah-ah-ah”) to a new sound (“ee”), which is interspersed into the stream of repeated “ah” once every few seconds in an unpredictable manner. If the brain does not detect the difference in sounds (as when the speech sounds being tested are not from a language the person knows), then no amplitude change will be noted at that point in time, because the neurons that respond to a change are not activated. Research using ERPs has shown that as infants’ perception of native language sounds improves, their ability to distinguish speech sounds from a nonnative language (i.e., ones to which they have not been exposed) declines (Kuhl et al. 2008; Rivera-Gaxiola, Silva-Pereyra, and Kuhl 2005). The decline in the ability to distinguish sounds from a language that the infant is not hearing in his or her daily input may be due in part to the use of general cognitive abilities 16 that allows him or her to tune out (ignore) irrelevant sounds while tuning into (attending to) relevant ones (Conboy, Sommerville, and Kuhl 2008; Lalonde and Werker 1995). During this same period of development, infants begin making the sounds of their language, a behavior known as “babbling.” Data from a MEG study in which an “oddball paradigm” was used to measure discrimination showed that activity in response to speech–sound changes in the left superior-temporal (“Wernicke’s”) and left inferior-frontal (“Broca’s”) areas of the brain become coupled between the ages of six and 12 months (Imada et al. 2006). This change possibly reflects the emergence of a functional link between brain areas underlying speech perception (the superior-temporal area) and speech production (the inferior-frontal area). In other words, as infants begin to practice producing speech sounds in their first year, this may change how they perceive speech sounds and lead to the establishment of new neural pathways. The use of different strategies for processing speech sounds may also explain differences in functional specialization and structure that result from various learning situations (e.g., bilingualism versus monolingualism). This last point is considered below. Changes in the functional specialization of brain activity to different types of words are also noted during early childhood, and have been linked to experience. Mills, Plunkett, Prat, and Schafer (2005) conducted a series of studies in which they recorded the brain activity of monolingual infants of different ages who passively listened to words they understood and to unknown words. Infants as young as nine months showed larger ERP amplitudes to known versus unknown words by 200 milliseconds (ms) after the onset of the word; that is, different neurons fired in response to hearing the words that were familiar. Younger toddlers (13-montholds to 17-month-olds) showed the effects bilaterally (in both the left and right hemispheres), whereas older toddlers (20-month-olds) showed effects only at electrodes placed at lefthemisphere temporal and parietal sites (for review, see Mills, Conboy, and Paton, 2005). Moreover, toddlers with larger vocabularies showed more focal ERP responses than their sameage peers who had smaller vocabularies, and when toddlers learned new words in meaningful contexts, a more focal response to those words emerged, reflecting that functional specialization emerged with experience with the words (Mills, Plunkett, Prat, and Schafer 2005). These results show how the functional specialization of the brain emerges during development and is linked to experience with language. As with sound processing, such changes may reflect the use of different strategies for word processing, as well as the increasing specialization of particular 17 forms of processing in neural networks that emerges with increasing experience and skill with language. In the next section we consider evidence from studies of dual language learning infants and children, which have further shown that the brains of DLLs become organized differently from those of monolinguals learners. Most of this research has been conducted with infants and toddlers, who are at the earliest stages of language learning. Experience and the Functional Specialization of the Brain during Dual Language Learning Most research on how the young brain responds to language stimuli has been limited to monolingual developmental situations, but recent work increasingly focuses on DLLs. In addition to examining whether there are differences in bilingual and monolingual brains, several ERP studies of infants and young children who are learning more than one language have shed light on how the organization of brain activity changes with language learning. Two types of populations have been studied—those who were learning both languages naturalistically, in which the amounts and/or timing of input varied across children, and those who received shortterm exposure to a second language through play sessions conducted in a laboratory setting, in which the amounts of input were controlled across children. In both types of studies, differences in input and experience across languages within the same individuals have allowed researchers to tease apart the effects of language experience and maturation on brain development. In one study designed to examine functional specializations for word processing, Conboy and Mills (2006) recorded ERPs to known and unknown English and Spanish words from 19- to 22month-old bilingual toddlers. All of these children were learning English and Spanish simultaneously, but in a variety of ways, and with uneven development across languages. Each child’s dominant language was determined by having parents complete a questionnaire regarding exposure to each language across a range of activities and vocabulary checklists on the MacArthur-Bates Communicative Development Inventories (CDI) in English (Fenson et al. 1993) and Spanish (Jackson-Maldonado et al. 2003). ERPs to known and unknown words were compared for each child’s dominant and nondominant languages. For both languages, ERP amplitudes were significantly larger for the known versus unknown words, as reported for monolingual infants and toddlers. However, the patterns varied for the same children’s dominant and nondominant languages, particularly in the timing and distribution of the effects across the brain. This finding can be explained only by experiential factors, not brain maturation, because 18 maturation was held constant. For example, there was more rapid processing of words in the bilingual children’s dominant versus nondominant language, which may reflect greater word familiarity and ease of lexical access in the dominant language. Unlike monolingual children of the same age, the bilingual toddlers showed effects that were broadly distributed across the brain, rather than limited to left electrode sites. In this sense, the distribution of brain activity of bilingual 19- to 22-month-olds was more similar to that of 13- to 17-month-old monolingual toddlers than it was to that of 20-month-old monolingual toddlers. Given that the bilingual toddlers knew approximately the same numbers of words in each of their languages as the younger monolingual toddlers, the results support the hypothesis that the organization of brain activity for language processing is influenced by toddlers’ experience with particular words. This is not evidence that bilingualism hinders or delays early language learning, although it is consistent with other evidence that bilingual lexical learning leads to initially smaller vocabularies in each separate language than for monolingual learners of those same languages, and that total vocabulary sizes (the sum of what children know in both their languages) in bilingual toddlers are similar to those of monolingual toddlers (Pearson, Fernández, Lewedeg, and Oller 1997); for more information on this topic, see Paper 5, Assessment, and Paper 6, Early Intervention and Special Needs. Thus, the differences noted in brain activity across bilingual and monolingual children should not be interpreted as evidence of a delay induced by bilingualism, but rather, as a distinct developmental pattern of specialization linked to experience with each language. There are many other possible reasons why bilingual children’s processing would be different from that of monolingual peers. One is the need to learn and manage conflicting sets of cues for each language. For example, English has many two-syllable words with a stress pattern in which the initial syllable is of longer duration and higher intensity (loudness) than the second syllable (e.g., “mommy”). Initial consonants in English words are thus perceptually salient, or noticeable, because they tend to be louder and longer than other sounds in the word. Because of this saliency, they provide a fairly reliable cue to the beginnings of words in ongoing speech, which helps listeners recognize individual words. This emphasis on the initial parts of words is not as common in all languages. Research that used both behavioral and ERP methods to test infants’ recognition of English and Welsh words showed that the stress patterns of each language accounted for distinct results across learners (Vihman et al. 2007). Monolingual Welsh-learning 19 infants did not show recognition of consonant-initial words at any point between nine and 12 months of age, but monolingual English-learning infants did so by 10 months, reflecting the stronger cues to word onset provided by initial consonants in English compared with Welsh. Bilingual English–Welsh infants recognized both English and Welsh words by 11 months, a pattern intermediate to those of the monolingual infants. Thus, bilingual infants learning English and another language with a different stress pattern (e.g., French, Spanish, Welsh) may temporarily reduce attention to initial consonants in words from both languages. However, this finding does not reflect a delay induced by bilingualism, because the bilingual infants in this study recognized words at an earlier age than did the monolingual Welsh infants. These findings with infants might extend to preschool children who are learning about word onsets during phonological awareness tasks. If there is a difference in the stress patterns of the words in children’s two languages, this may temporarily change the cognitive strategies used for processing those words. Thus, the particular pair of languages a child is learning may influence how learning in the two languages interacts. When teaching phonological awareness skills, practitioners should always consider the phonological structure (sequences of sounds, stress patterns) of children’s home language and whether it is different from the children’s school language. The processing of words relies on being able to perceive and process speech sounds (phonemes), and the ERP oddball paradigm described above is ideally suited to studying such perceptual skills. The perceptual abilities of bilingual learners may be explained to some extent by relative amounts of experience with each language (Conboy, Jackson-Maldonado, and Kuhl 2009; García-Sierra et al. 2011). For example, García-Sierra and colleagues (2011) presented bilingual infants living in Texas with English and Spanish speech sound contrasts in an ERP oddball paradigm. The infants who heard more English in the home showed a larger discriminatory response for the English than for the Spanish contrast, whereas infants who heard more Spanish at home showed the opposite pattern, and infants with more balanced input across languages showed similar discrimination for each language. Because that study was not designed to directly compare bilingual and monolingual infants, it is not known whether the brain activity of any of the three subgroups resembled more closely that of monolingual infants of the same age. However, when the results from the bilingual infants were compared with the results from a previous study of monolingual infants, it was noted that the younger members of the bilingual 20 group (six-month-olds to nine-month-olds) did not show exactly the same patterns as monolingual infants of that age. These results suggest that relative amounts of experience with each language correlate with how the brain processes language. This finding is important for practitioners, because it shows that amounts of exposure to each language make a difference in how the brain processes each language. In sum, although there are some differences between bilingual and monolingual language learning and processing, there are many similarities. Differences between groups have been revealed using cognitive neuroscience methods that are sensitive to the timing and distribution of brain activity. These differences appear to arise from the unique cognitive demands that learning and processing two languages create, such as selectively attending to each language’s sound patterns and grammatical rules, inhibiting the retrieval of words in one language when using the other language, and being able to translate and process mixed language input (see the next section and Paper 2, Development Across Domains, for further information). It is important that such differences are not viewed as deficits, but rather, as opportunities for expanding the brain’s capacity. At the same time, the slight differences in brain function underscore why monolingual standards are inappropriate for assessing dual language learners (see Paper 5, Assessment). Developmental milestones are achieved at similar ages in dual language learners as in monolingual learners, but might be achieved in slightly different ways. The next section provides examples of some cognitive processes used by bilingual individuals to learn and process language. The research reviewed in this paper has indicated that networks in the brain become established for each language based on experience with that language, and that this process 21 requires time, as well as rich input in each language. Differences Are Not Deficits • There are many similarities between bilingual and monolingual language learning and processing, but there are also some differences. There is evidence that the differences arise from the cognitive demands that learning and processing two languages create for learners, including the need to selectively attend to each language’s sound patterns and grammatical rules, to inhibit the retrieval of words in one language when using the other language, to translate, and to process and produce mixed language input. • It is important that such differences are not viewed as deficits, but rather as opportunities for expanding the brain’s capacity. • Networks in the brain become established for each language based on experience with that language; this process requires time, as well as rich input, in each language. Developmental milestones are achieved at similar ages in DLLs as in monolingual learners, but might be achieved in slightly different ways. Involvement of Different Cognitive Processes in Dual- Versus Single-Language Learning and Use The unique challenges faced by bilingual speakers lead them to process information in ways that are different from those of monolinguals. In addition to sorting out conflicting cues to speech sounds, word structure, and sentence structure, bilingual speakers frequently process language under mixed-language conditions (i.e., hearing words from both languages mixed into the same sentence or conversation). Enhanced functioning on nonlinguistic tasks that require executive functions, such as working memory, inhibitory control, and the ability to control attention to relevant versus irrelevant cues, is seen in bilingual individuals as young as preschool and kindergarten age (e.g., Carlson and Meltzoff 2008) and as old as later adulthood (Bialystok et al. 2004). The attentional skills of infants are also influenced by dual language experience (Kovács and Mehler 2009). Thus the cognitive demands of managing two languages may sharpen abilities in other domains, and these enhanced cognitive abilities may be used to further process and learn language (see Paper 2, Development Across Domains, for a review of the most relevant findings for preschool-age DLLs). Shafer, Yu, and Datta (2011) used ERPs to directly compare the speech sound discrimination skills of bilingual and monolingual infants and young children learning English and Spanish in New York City. Children were tested on two speech 22 sounds (vowels) that are found in English but not in Spanish. The ERPs showed discrimination, but the effects varied by age and language experience (monolingual versus bilingual). In particular, there was evidence that the bilingual infants had higher levels of attention while processing the speech sounds. A follow-up study confirmed this hypothesis (Shafer, Yu, and Garrido-Nag 2012). Enhanced attention during speech processing might be thought of as an adaptive strategy, brought on by bilingualism, and it might be one that allows children growing up with two languages from birth with the ability to keep pace with their monolingual peers in achieving developmental milestones in language. In the research on bilingual toddlers conducted by Conboy and Mills (2006) described above, children showed more enhanced neural activity at right-hemisphere versus left-hemisphere frontal electrode sites while processing words in their dominant language, but this pattern was observed only in toddlers who were tested in a condition in which words were randomly switched between English and Spanish. A group of bilingual toddlers tested in a condition in which words were presented separately for each language did not show the larger effect at right frontal sites, and there was evidence that they processed words more quickly. In other respects, their ERP patterns resembled those of the bilingual toddlers tested in the language-switching condition. Thus, the increased cognitive load imposed by switching between languages may have recruited additional neural tissue and required additional processing time. A follow-up study of the Conboy and Mills (2006) study using a different analysis method showed that brain activity in a frequency band known as theta (between 4 and 8 Hz)—which is linked to effortful processing—differed for children tested in the language-switching and in the single-language conditions (Bosseler, Conboy, and Mills 2012). These results support the hypothesis that different cognitive functions are involved in processing words across such conditions, though further research is needed to understand what those functions are. Kuipers and Thierry (2011) studied two- and three-year-old bilingual and monolingual children growing up in Wales in order to determine whether there were differences in how the two groups processed switches between languages. The children were presented with a picture, and then heard a word that matched or did not match the picture. The bilingual children were faster than their monolingual peers at detecting when the language of word presentation switched from one language to the other, suggesting that they had developed a switching-detection device. The bilingual children showed an ERP effect (larger positivity) to the switch within the first 23 300 ms after the onset of the word onset, a time window associated with phonological analysis, while the monolingual children of the same age did not show an effect until after 300 ms, a time window associated with lexical-semantic analysis. Kuipers and Thierry (2011) interpreted the earlier effect as a particular bilingual language-change detection response that developed only in the bilinguals, and the later effect as a response to word familiarity/meaning that developed in both populations. More research is needed to understand how different cognitive functions are used in bilingual versus monolingual language processing. What the research suggests is that different brain tissue is recruited in bilingual children, as an adaptation to the task of managing two languages. Although practitioners may view the use of additional cognitive strategies as a negative aspect of bilingualism, it may be viewed as a positive adaptation that can enhance other aspects of learning (see Paper 2, Development Across Domains, for more information). Second Language Learning: Is Early Always Better? It is widely believed that language learning is optimal during the early years of life and that a language is best learned within this early period. According to the “critical period hypothesis,” acquisition of a language must begin early in life in order for it to occur normally. There is, indeed, evidence that supports the idea of a sensitive or “critical” period for first-language acquisition (Mayberry and Lock 2003; though see Bruer 2008 for a challenge to the use of the term critical period regarding language acquisition). For example, Mayberry and her colleagues have investigated sign-language acquisition in individuals who were born deaf and deprived of early language input because their deafness was not initially discovered, or because opportunities to interact with a community of sign language users were not available (Boudreault and Mayberry 2006; Mayberry 1993; Mayberry and Lock 2003). This line of research has revealed that delays in exposure to a first language in early childhood are associated with a reduced ability to attain high levels of proficiency in that language. Moreover, Mayberry, Lock, and Kazmi (2002) found that individuals who had exposure to either a signed or a spoken language early in life attained higher levels of proficiency in a second language (signed or spoken) than those who had not had early experience in any language. Thus, early exposure to a first language, in either the spoken or signed modality, is not only crucial for successful firstlanguage acquisition, but also lays down a foundation for successful second-language 24 acquisition. However, in individuals who have had early exposure to a first language, it is not clear whether exposure to the second language also must occur early in life for full proficiency to be attained in that language—that is, whether there is a sensitive or critical period for secondlanguage acquisition (Hakuta 2001). Although research findings suggest that individuals who acquire a second language in later childhood or adulthood seldom acquire it with full proficiency, researchers do not agree that the interpretation of such data is that there is a critical period for second-language acquisition. That is, age-of-acquisition (AoA) effects are open to more than one interpretation (see the next section). Unfortunately, misconceptions about sensitive or critical periods for language acquisition have sometimes been used to support practices and policies on DLLs in the United States (Crawford 1999). Advocates of English-only educational approaches with culturally and linguistically diverse students in the United States have argued that “The critical age hypothesis for second language acquisition has long been recognized by linguists, i.e., that the optimal time to learn a second language is between age three and five or as soon thereafter as possible, and certainly before the onset of puberty” (Porter 1988), and that “Young immigrant children can easily acquire full fluency in a new language, such as English, if they are heavily exposed to that language in the classroom at an early age” (English Language in Public Schools Initiative Statute 1998). These arguments not only assume that the optimal time to learn a second language has been established scientifically, but also suggest that the use of bilingual educational approaches delays the onset of English learning, when, on the contrary, children in transitional bilingual education programs have been shown to benefit from learning academic content in their home language at the same time that they are developing English skills (Krashen and McField 2005). It is important that practitioners know about research findings on AoA effects for second-language learning and understand the possible sources of these effects, so that they make informed decisions about first and second language use in the classroom and give sound advice to parents regarding home-language use (see Paper 3, Program Elements and Teaching Practices, for additional information). Several “long-term attainment studies,” in which adults who started learning their second language at different ages were tested on their proficiency in that language, have shown AoA effects on participants’ long-term attainment of the second language. These studies have shown that the earlier the individuals began learning the second language, the higher the level of 25 proficiency they ultimately attained in that language, as observed in their ability to detect grammatical violations (e.g., Birdsong and Molis 2001; Jia, Aaronson, and Wu 2002; Johnson and Newport 1989), discriminate speech sound differences (e.g., Flege, Yeni-Komshian, and Liu 1999), and produce second-language speech sounds (e.g., Jia et al. 2006). In a few studies, participants who were exposed to the second language before or around six years of age showed native-level performance in all of these areas. That is, there is no evidence from this research that the window of opportunity to become a strong second-language user abruptly closes any time before the age of six years. It has also been shown that there is not any abrupt moment in time when the window of opportunity closes, even beyond the age of six years; instead, there is a gradual decline with increasing age of acquisition throughout childhood and adolescence (e.g., Birdsong and Molis 2001; Hakuta, Bialystok, and Wiley 2003). It is important for practitioners to recognize that research on AoA effects has been correlational, rather than experimental, in nature. In correlational research, other factors that can lead to a change in the variable of interest are not controlled, as is done in experimental research. Thus, it is possible that the higher levels of second-language proficiency achieved by earlier versus later learners may be due to some other factor than AoA. Neuroscientific methods have also been used to examine differences in brain function and structure across individuals who have learned a second language at different ages. In the functional imaging research, even small delays in the onset of second-language acquisition (i.e., one to three years) have been linked to the patterns of brain activity for processing that language (Weber-Fox and Neville 1996). Again, it is important to realize that, in many of these neuroscientific studies, experiential factors, such as frequency of use of a language, have been confounded with the age of first exposure to that language. Several studies have indicated that proficiency in the second language, as well as the age at which the language was acquired, affect patterns of brain activity (e.g., Perani et al. 2003). The influence of language proficiency on the brain was demonstrated most profoundly in a study by Mechelli and others (2004), in which structural magnetic resonance imaging was used to examine gray matter volumes in the cortex. Gray matter density in one area of the cerebral cortex—the inferior parietal area—was greater in bilingual adults compared with monolingual adults, greater for individuals who had learned their second language earlier (prior to the age of five years), and greatest for those individuals most proficient in their second language. Moreover, differences in brain function for the first and 26 second languages have been noted even when the second language was acquired before the age of five years (Conboy and Mills 2006; Kovelman, Baker, and Petitto 2008; Perani et al. 2003; Wartenburger et al. 2003). There is a limited amount of neural imaging research on bilingualism or second-language processing in children older than infants and toddlers, and most of that work has been conducted with older children. In an fMRI study of word processing in seven- to 11-year-old bilingual children with high levels of proficiency in both languages, there were no differences in activation patterns across the two languages, even for the children who acquired their second language after the age of three years (Mondt et al. 2009). In an fMRI study of 13-year-old Japanese twins, the processing of second-language verb forms occurred in the same left frontal cortical areas as for processing first-language verb forms, after only two months of second-language exposure in a classroom setting (Sakai et al. 2004). Moreover, second-language performance was highly similar in twin pairs, suggesting that both shared genetic and shared environmental factors are involved in determining the neural correlates of second-language learning. There was similar neural activity for the second language as for the first language, even at the beginning stages of second-language acquisition, when performance was not native-like, and even when secondlanguage learning began in adolescence. This finding calls into question the notion that later second-language acquisition cannot occur through the recruitment of the same brain areas as first-language acquisition and use. In another fMRI study, Tatsuno and Sakai (2005) compared two groups of adolescents, ages 13 and 19, on a past-tense verb identification task in their first language (Japanese) and their second language (English), which was acquired after the age of 12 years. The groups showed similar activation in a left-hemisphere language region for the first language, but different activation levels for the second language, which were linked to proficiency levels. In the older group, who had achieved proficiency with English past-tense forms, there was greatly reduced activation, suggesting less effortful processing with increased proficiency. Archila-Suerte, Zevin, Ramos, and Hernández (2012) reported fMRI evidence for the use of different brain processes during a speech perception task in six- to 10-year-old monolingual children and bilingual children the same age who had begun to learn their second language between four and nine years of age. The bilingual children recruited areas of the brain involved in executive function (cognitive control) to a greater extent than did the monolingual children, 27 but only when processing their second language. This finding supports the idea that some AoA effects are caused by the use of different cognitive processes. Is There a Critical Period for Language Learning? • It is undisputed that early exposure to a first language is crucial for successful first-language acquisition and lays a foundation in the brain for successful second-language acquisition. However, for individuals who have had early exposure to a first language, it is not clear whether exposure to a second language also must occur early in life for full proficiency to be attained in that language. • Although individuals who acquire a second language in later childhood or adulthood seldom acquire it with full proficiency, scientists do not agree that the interpretation of such data is that there is a critical period for secondlanguage acquisition. Some scientists argue that complete mastery of a second language occurs only when exposure to the language begins in early childhood, before certain brain structures have matured (i.e., that there is a critical period for second-language acquisition). Other scientists propose that, rather than brain maturation, experiential and/or cognitive factors frequently limit the learning of the second language (i.e., a second language can be learned at any age, given the appropriate experiences and cognitive factors). Interpretations of Age-of-Acquisition (AoA) Effects The above-mentioned AoA effects noted in both brain and behavioral research have been explained by different accounts. Some argue that AoA effects provide evidence that there is a critical period for the acquisition of second (and subsequent) languages, and that complete mastery of a language occurs only when exposure to the language begins in early childhood, before certain brain structures have matured (e.g., Johnson and Newport 1989). In contrast, others propose that rather than brain maturation, experiential and/or cognitive factors are the major causes of AoA phenomena (e.g., Archila-Suerte et al. 2012; Bialystok and Hakuta 1998; Elman et al. 1996; Marinova-Todd, Marshall, and Snow 2001). Evidence has been mounting to support the latter claim. One line of evidence that supports an experiential account comes from a five-year longitudinal study in which researchers documented changes in the language proficiency and use of 10 native Chinese-speaking children who immigrated to the United States between five and 16 years of age (Jia and Aaronson 2003). It was found that younger and older children went through different second-language learning processes. The younger children 28 quickly switched to English as their main language due to a sense of peer pressure, a desire to fit in, and because their first language was not yet developed. In contrast, the older children sought out peers of the same home language and culture and continued to use a large amount of the home language. Throughout the five years of the study, the younger children experienced a significantly richer second-language environment than the older children did. These findings support an experiential account in explaining the seemingly greater challenge in second-language learning encountered by older learners. Another body of studies on AoA effects on second-language acquisition has examined firstlanguage, as well as second-language, proficiency. Studies adopting this approach have consistently found that individuals who begin to learn their second language in early childhood become more proficient in their second language than in their first language, whereas older learners show the opposite pattern, maintaining the first language as their dominant language. Such trends have been found for the accuracy of pronunciation of Korean–English bilinguals (Yeni-Komshian, Flege, and Liu 2000), as well as for the speed and accuracy of lexical retrieval of Spanish–English bilinguals (Kohnert, Bates, and Hernández 1999) and of Russian–English bilinguals (McElree, Jia, and Litvak 2000). When considering potential critical-period effects on second-language acquisition, practitioners need to look at the long-term outcomes of those effects and consider children’s experiences with both of their languages, instead of focusing only on whether second-language performance matches that of native speakers. Research has found AoA effects for particular aspects of language, such as speech-sound pronunciation and perception (phonology) and grammar (morphology and syntax), but not meaning-related aspects of language (e.g., vocabulary learning). In fact, several studies have shown that older learners are faster at learning vocabulary than are younger learners (Marinova-Todd, Marshall, and Snow 2001). Practitioners need to consider whether the effects on phonology and grammar noted in later second-language learners are important in the long run, especially in terms of educational outcomes. Jia and Aronson (2003) found that young adults who were older at the time of arrival in the country of the second language made more morphological errors in that language than those who had arrived in the country earlier in childhood, but were equally strong academically. In another study of college students from Spanish–English bilingual backgrounds, Jia, Alvarez, and Pantin 29 (N.d.) found that morphological ability in English had only a weak correlation with students’ grade-point average. The Effects of Language Learning on the Brain Facilitate and Constrain Further Learning For an age-related decline in acquisition ability to be called a critical-period effect, there must exist specific moments in time when the window of opportunity for learning opens and closes (see Bruer 2008). Because there is, instead, evidence of a gradual reduction in secondlanguage learning ability with age, several researchers have argued that AoA effects should not be interpreted as critical-period effects (Bialystok and Hakuta 1994). The neuroconstructivist framework provides an alternative model for thinking about the source of such gradual reductions, which is that learning and development constrain future learning. One way to think about how learning constrains future learning is by considering the phenomenon of perceptual narrowing, which is noted across several domains of young children’s development. Werker and Tees (1984), and other subsequent research (for a review, see Conboy et al. 2008), have shown that infants undergo a process of perceptual narrowing in speech perception during the end of their first year, and this is reflected by how well they process differences between speech sounds from their native language versus those from an unfamiliar nonnative language. Kuhl et al. (2005) have further suggested a functional link between perception of native and nonnative language sounds. As described above, infants vary in their ability to detect native and nonnative sound differences when tested at seven and one-half months of age, an age at which infants typically begin to fail to discriminate nonnative sounds. Using a conditioned head-turning test, in which infants are trained to turn their heads in one direction when they detect a change in a speech sound, Kuhl and her colleagues (2005) found that those infants who had better native-language perception also had worse nonnative sound perception at this age, and those same infants had better native-language vocabulary skills several months later, when compared with infants who had the opposite profile at seven and onehalf months (i.e., worse native language perception and better nonnative language perception). This finding was subsequently replicated using an ERP oddball paradigm (Kuhl et al. 2008). The findings suggest that perceptual narrowing involves a process of neural commitment to the native language, which leads infants to ignore particular features that are not relevant to the native language, and results in making a nonnative language more difficult to acquire with increasing age. A similar concept is that of entrenchment, which is the idea that once a learner’s 30 neural networks become entrenched in responding to linguistic cues (e.g., sound patterns or grammatical rules), they become less able to accommodate conflicting cues (see Elman 1993; Elman et al. 1996; Fava, Hull, and Bortfeld 2011). It is important to consider the time frame in which processes such as perceptual narrowing, neural commitment, and entrenchment constrain learning, and to consider whether such constraints are immutable. It would be expected that children who are exposed to a second language after such narrowing has occurred would have difficulty learning the language or that the effects of narrowing would lead to learning that is less than optimal. This hypothesis was tested in two studies by Kuhl and her colleagues. In the first study, Kuhl, Tsao, and Liu (2003) provided infants (ages nine to 10 months) from monolingual English-speaking homes with Mandarin Chinese exposure in a laboratory setting. The infants interacted with native adult speakers of Mandarin during 12 25-minute sessions, or watched and/or listened to those same speakers on video recordings for the same amount of time. Across all of the conditions, the speakers used naturalistic, infant-directed styles of speech and the same toys and books, to ensure uniformity of exposure to the language across infants. After the sessions, the infants were tested on their discrimination of two Mandarin speech sounds, using a head-turning procedure. The infants who received live, naturalistic exposure to the language discriminated the sounds, and their performance was similar to that of a comparison group of infants being raised in Mandarin-speaking homes in Taiwan. The infants who watched or listened to the videos did not discriminate the Mandarin speech sounds and performed at a level similar to that of a control group who received play sessions conducted in English (and hence did not have exposure to Mandarin). In a follow-up study, Conboy and Kuhl (2011) provided a group of infants, ages nine and one-half months to 10 and one-half months, from monolingual English-speaking homes, with live, naturalistic exposure to Spanish. The infants were tested before, as well as after, the 12 exposure sessions, and the researchers used an ERP oddball paradigm to test discrimination of two Spanish speech sounds. The infants showed changes in the ERPs to both the Spanish speech sounds and to a pair of English speech sounds, from pre-test to post-test. For both languages, the neural correlates of speech-sound discrimination (a difference in the ERPs to the infrequently presented deviant sound and the frequently presented standard sound) showed improvements in discrimination. A discriminatory response that was not present for the Spanish sounds when the infants were nine months old was noted when they were 11 months old. For English, the 31 discriminatory response was present at both ages, but it was enhanced (larger in size and with an earlier onset) at 11 months compared with nine months. There are four important points to consider from the above-mentioned research. The first is that, if perceptual narrowing and neural commitment to the native language are the sources of AoA effects in second-language acquisition, at least with regard to phonology, then this process is not immutable. The tuning out of nonnative speech sounds must be reversible, if infants can develop the same neural discriminatory responses for nonnative speech contrasts as for native speech contrasts after only a small amount of naturalistic exposure to that nonnative language. Other studies in which an ERP oddball paradigm was used with children learning a second language in classroom settings have also shown a similar neural discriminatory response to second-language speech sounds emerging at three to six years of age (Cheour et al. 2002; Rinker et al. 2010; Shestakova et al. 2003). Thus, neural commitment does not completely hinder new phonological learning in children, even during the early school years. In other words, the brain’s increasing specialization for processing the first language during early childhood does not necessarily limit its ability to learn another language. Because the brain is not a limited capacity system, it can accommodate the learning of new information. The fact that synaptic pruning occurs during the same period when the brain becomes neurally committed to processing the native language(s) does not necessarily lead to the conclusion that the ability to learn a nonnative language is present only prior to neural commitment. The second important point from the Kuhl, Tsao, and Liu (2003) and Conboy and Kuhl (2011) studies is that native language processing in early development was not adversely affected by exposure to another language. Infants continued to discriminate their first-language sounds while learning second-language sounds, a finding that is consistent with the research on speech-sound discrimination in bilingual infants reviewed above. Again, these findings suggest that the brain is not limited in its capacity to learn multiple languages. However, there is some evidence that infants who have been raised bilingually from birth might not undergo the process of perceptual narrowing in exactly the same ways as monolingual infants. A functional NIRS study demonstrated that 10- to 12-month-old bilingual infants did not show perceptual narrowing for nonnative sounds (i.e., speech sounds from a third language that were not from either of their two native languages [Petitto et al. 2011]). This finding provides evidence that infants growing up with two first languages do not show neural commitment in the form of perceptual narrowing 32 in the same ways as monolingual infants. The bilingual infants adapt their processing in ways that are beneficial for bilingualism (what Petitto and her colleagues suggest is a “perceptual wedge” that allows bilingual infants’ systems to remain more “open” to multiple linguistic cues). Third, the Kuhl, Tsao, and Liu (2003) study highlighted the importance of social interaction for language learning, and a follow-up study has indicated that the degree to which individual infants socially engage with adults who speak to them in the second language, as measured by the infants’ shared eye gaze with those adults, is associated with the infants’ second language perceptual learning (Conboy et al. N.d.). Thus, these results are consistent with the view that infants and young children succeed in learning the phonology of a second language when they are socially engaged with speakers of that language. The studies with infants may be of interest to preschool teachers, because the studies indicate what neural, social, and cognitive mechanisms might be involved in aspects of learning and processing a second language. In sum, the research suggests that many AoA related brain phenomena may be due to relative language proficiency, frequency of use of each language, and social and cognitive factors, rather than critical-period effects. Recommendations and Implications for Program Practice Given the research showing that measures of brain activity reflect the varying experiences children raised in bilingual environments have with each of their languages, practitioners should realize that DLLs will not exactly resemble monolingual children in each of their languages. Therefore, monolingual assessment standards should not be used with DLLs (for more on the topic of assessment, see Paper 5, Assessment). Although differences in how bilingual brains process language are not signs of deficit, the differences point to several instances where a DLL would appear deficient if assessed using monolingual norms. For example, bilingual children tend to have vocabulary knowledge distributed across both their languages, thus they do not always know the same words in both languages. It is typical for young bilingual children to know many different types of words (such as nouns, verbs, and adjectives) in the more dominant language, while knowing fewer word types in the other language (Conboy and Thal 2006). Given this distributed nature of bilingual language development, bilingual children might not be equally ready to learn the same things in each of their languages at particular moments in time. The research reviewed in this paper has indicated that networks in the brain become established for 33 each language based on experience with that language, and this process requires time, as well as rich input in each language. As brain measures have indicated, bilingual children are often more efficient at processing their stronger language than their other language, and this processing efficiency could affect the learning of more advanced language skills in each respective language. Practitioners should also consider that nothing about neuroscience research indicates that the brain ...