FEATURES:      

E-Commerce on The Fly     
Do You Compute?     
Laws of Lanugage     

Laws of Language

Story by Tom Ventsias
Photography by John T. Consoli

How young children build meaning out of a cacophony of words.

When noted linguist Noam Chomsky announced in 1957 his theory of "universal grammar," wherein he stated that there is a complex biological component to language acquisition, he dramatically altered the scientific landscape concerning research in linguistics and cognitive science. Chomsky proclaimed that each person is born with a "genetic blueprint" of language and grammar skills, and that language emerges in children simply by exposure to an outside language source--usually infants hearing their parents talk. These ideas were in direct contrast to "experiential" linguists, who theorized that young children learned to speak by constantly listening to words and sentences repeated to them over time.

Like Chomsky, many of the diverse faculty in linguistics and cognitive science at the University of Maryland believe that language acquisition is largely a biological rather than experiential occurrence. Although no one claims to fully understand why young children can master a rich and complex language system before they master other cognitive skills--like counting or tying knots--establishing the position of language in cognitive development promises to help scientists better understand the human brain and determine how much of human knowledge is "pre-wired" from birth.

According to Stephen Crain, professor and chair of the university's Department of Linguistics, "there is a virtual explosion of language development in children at about 24 months of age." It is precisely at this stage of their cognitive development, Crain says, where children figure out what language is all about by trying out different grammatical structures. This provides a rich source of information for linguists and psycholinguists, he adds.

Currently, researchers in linguistics at Maryland have five or six ongoing experiments involving almost 60 children recruited from the university's Center for Young Children. Their collective efforts represent one of the most comprehensive linguistics research programs in the United States, drawing from a wide range of academic disciplines, including psychology, hearing and speech science, English, biology and computer science, among others.

One major component of Maryland's linguistics research effort is coordinated through the Neuroscience and Cognitive Science program, or NACS. Within NACS, a team of researchers known as the Language Group attempts to bridge the gap that exists between theoretical, computational, psychological and neuroscientific models of language. What this language group strives for, Crain says, is to draw researchers who work on cognition--the process of knowing--together with scientists who work in areas of neurology and other biological disciplines.

The interaction between different disciplines investigating speech and language problems has increased substantially over the last decade, Crain says. There were pockets of collaborative research before, he explains--mainly between psychologists and linguists--but at Maryland these links are more extensive and the collaborations run deeper. "The linguistics program at Maryland is [nationally] recognized because of its dual emphases," Crain says. "We are continually exploring linguistics theory as well as conducting numerous laboratory studies."

For example, he notes the recent addition to the linguistics department of Howard Lasik, considered to be one of the nation's top language theorists and whose doctoral advisor at the Massachusetts Institute of Technology was none other than Noam Chomsky. Crain also cites groundbreaking research at Maryland by David Poeppel, who has a dual appointment in biology and linguistics and is using magnetoencephalography, or MEG technology, to measure magnetic fields generated by neuronal activity in the human brain. Using the MEG technology helps linguists to study how the brain processes speech and language.

"The combination of research tools used at Maryland offers us different sources of insight into the central questions about language--how language is developed, how it is used, how it is represented in the brain and how it is processed," Crain says. "And the answers to all of these questions are important to our understanding of language as a biologically driven phenomena."

Playtime, Puppets and Language Acquisition

Rosalind Thornton, associate professor of linguistics and director of the department's Language Acquisition Lab, says there is still ample debate within the academic community concerning Chomsky's theory of universal grammar. Some researchers still believe that young children essentially imitate other speakers until they are about 3 years of age. "[But] we think that imitation plays only a tiny role in what 2-year-old children are doing with language," Thornton counters. "A lot of the debate on nature-versus-nurture in language acquisition is focused on this 2-year-old age group."

Until very recently, she says, most researchers studying language acquisition in 2-year-old children generally "played" with the toddlers in naturalistic contexts. They would then observe videotapes of these play sessions or read voluminous transcripts of the children's spontaneous speech patterns that were obtained during the unscripted playtime sessions.

"What we are doing is different," Thornton says. "We assess children's knowledge of language by creating specific situations that can tell us whether or not they are able to produce a specific linguistic structure, or how they interpret a particular sentence that we present to them." To do this, Thornton often asks young children to judge the veracity of a sentence that a puppet uses to describe a story that is acted out in front of the children. Thornton says that a child is much more willing to tell a puppet that it is right or wrong than to tell an adult the he or she is right or wrong.

The research objective is to experimentally control both the context and the linguistic input to children, she says. Typically, one experimenter acts out the stories in front of the child, and the puppet, which is manipulated by a second experimenter, describes something that happened in the story. The child's task is to say whether the puppet was "right" or "wrong." Thornton says that observing how children are able to match sentences to situations is a valuable source of information about their grammars.

"We have also been witness to a lot of things that children don't produce in their spontaneous speech," Thornton says. One example of that is negation. For example, in English, we say, "That doesn't fit" instead of "That not fits." However, other languages differ from English in this regard, and Thornton has discovered that some children learning English speak such a "foreign" language for a brief period, and then switch to English. By testing children's use and comprehension of negative sentences over time, linguists can chart the course of language development to see how closely child language follows the natural seams of grammars, as attested in cross-linguistic research.

Thornton may run these same tests--with very little variations--for months, and sometimes up to a year. "We are interested in seeing how a child's grammar changes over time," she says, adding that by keeping the context of the testing the same, researchers know that it is the child's grammar that changes over time, rather than other aspects of the child's cognitive development.

This is extremely important in evaluating Chomsky's theory of universal grammar, Thornton says, because the theory states that many core grammatical principles are operative from birth. "If we can lower the age threshold, and show that children as young as the age of 2 have immense and detailed knowledge about language--including knowledge of things that can be said [that are grammatically correct] but also knowledge of things that cannot be said--then the findings are difficult to explain on the view that children acquire all of their knowledge about language from experience."

Filtering Sound from Noise

While researchers in the linguistics department are studying how children ages 2 to 5 learn grammatical structures, other faculty members at Maryland are looking at an even younger group.

In her Language Perception and Development Lab, assistant professor of hearing and speech sciences Rochelle Newman is working with infants as young as 4 months of age. "I am interested in the very beginnings of language acquisition," she says. "And in determining how well infants do at picking up sounds of the language, and in learning new words, when they are in situations that aren't [acoustically] perfect."

Newman observes the role that background noise has on an infant's ability to understand simple words. "We know that for adults, background noise is easier to separate when it comes from different [spatial] areas," she says. But for infants, this situation is very different. "If they hear different noises coming from different locations within a room, they are not as able to make the distinctions on exactly where the sound is originating."

Using one research method, Newman will place a 5-month-old infant in a soundproof room to observe the child's ability to identify its own name in the midst of ongoing background conversations. "Infants by about 4-and-a-half months of age will recognize their own name when they are in quiet situations," Newman says, "so we're assessing whether they can still recognize their name while they're in a noisy environment."

The background conversations that make up this "noisy environment" are digital recordings that include a combination of adult voices and children's voices. A flashing light to capture the child's attention and stimulate sensory response precedes the infant hearing either its own name or another name. "We don't know if they realize that their name refers to them specifically," Newman explains, "but we know from other studies that they listen longer to their own name than other names." The research focuses on how long the infant chooses to pay attention to his or her name while simultaneously hearing the background conversations. Newman can measure, within a hundredth of a second, how long the child concentrates on the stimulus of light and sound coming from a certain space in the laboratory setting.

Newman is also working with hearing and speech specialists at Johns Hopkins University in Baltimore. Their collaborative investigation has shown that by the age of 7 months, infants will recognize the link between a person's lips moving and the sounds they are hearing. Just as with adults, Newman says, seeing the movement of the lips helps the infants isolate that sound from background noise.

Another aspect of Newman's research examines whether infants are able to learn language from situations that are not interactive. "Can infants learn anything from television?" she asks. "Once you start looking at 2- and 3- and 4-year-olds, we've found that yes, they can learn things [from television], which is why educational television is important. But we don't really know much about if infants can learn anything from those types of situations."

Ultimately, Newman says that the results of her research can be used to help predict outcomes of children in daycare settings--where background noise from other children talking is much more prevalent than in a home setting. It can even be used to help engineers design better voice-recognition software that can filter out unwanted background noise. "Once we can determine how humans deal with noise," she says, "then we are that much closer to being able to create computers that can do the same thing or can learn the same thing."