The Practical Linguist / A sensible theory of language
Marshall R. Childs Special to The Daily Yomiuri
When I began to study language learning in 1990, most of what passed for theory was not very helpful for teaching. No matter how hard you searched, you could never find enough useful theory to float a syllabus.
Language teachers, left high and dry, had to figure out their own approaches or use textbooks. During the past decade, however, there have been many exciting advances in knowledge about how people learn and process language.
There is growing recognition that a new theory, a sensible and useful one, is at hand. I have been waiting for someone to write a book describing the new theory. But it is a synthesis of several disciplines, and experts seem reluctant to generalize beyond their fields of primary expertise. I share the reluctance, but in The Practical Linguist columns, I often use the new theory to explain phenomena of language processing and language learning.
Today, I will try to present a balanced summary of the new theory. In a future column, I will discuss implications of the theory for language educators.
The new theory deserves a name. I call it "attractor theory" for reasons I will explain. Any theory of how we process language must cover (1) how the brain works, (2) how the brain processes language, and (3) what is involved in learning both first and second languages.
(221 words)
===
How the brain works
A sensible theory of language places it within our best understanding of neural processes in the brain. In the 1940s, psychologist Donald Hebb observed that neuronal connections in the brain do not seem to work one at a time (the way we might design a computer program) but in great masses like soccer fans swarming onto the pitch after a game.
Hebb knew that neurons excite each other by means of connections (synapses), and he declared that using a synapse strengthens it. Large masses of neurons working together came to be known as Hebbian cell assemblies, and the rule that using a synapse strengthens it (Hebb's rule) is regarded as the basic process of change and thus of learning.
People used to think that different parts of the brain handled different functions neatly, similar to the way we might divide up the work in a factory. Now, however, the best thinking is that function is related to location only in a general sense. Instead, the brain does networking, in which neurons in many different areas participate in complex processing. What used to be thought of as centers of processing are now described as epicenters of distributed processing.
M. Marcel Mesulam, professor of neurology at the medical school of Northwestern University, described the networks as "partially overlapping large-scale networks organized around reciprocally interconnected cortical epicenters." He said at least five large-scale networks can be identified in the brain, one of them being a language network.
The metaphor of resonance is a handy way of describing mental processing. The implication is that many parts of the brain are activated in concert with whatever thoughts, feelings or environmental stimuli are being processed.
===
How the brain 'does' language It used to be thought that two areas of the brain perform language: Broca's area was the storehouse of speech patterns and Wernicke's area contained a sound-word dictionary.
According to Mesulam, however, these two areas do not act alone but are epicenters in a distributed language network. Wernicke's area, for example, is a portion of the brain that acts as a gateway between sound and meaning. It is necessary but not sufficient for processing words.
Certainly, connections between sound and meaning are tight and simultaneous, not sequential. Hearing and speaking are handled by almost the same parts of the brain and nervous system. I should comment that hearing and speaking are the normal modes of processing language, but the brain is flexible. For deaf or mute people, the brain supports language in other modalities. Since the invention of writing, the brain has made adjustments to process language by reading and writing.
Network processing means that language is not processed in sequential logical steps like a computer's stored program. Patterns in "wetware" are never completely fixed. Instead they have central tendencies that are called, in phonology, categorical perception or "magnets," and in chaos theory, "attractors." Grammatical patterns, too, are attractors. It is true that some parts of any language can be described as following rules, but this does not imply that languages are produced by rules.
Rather, the recurrent patterns are attractors. Rules are not causal, but only after-the-fact descriptions of outcomes. Language processing is mostly subconscious. Sound and meaning are produced and heard mostly automatically, and resonate in our heads only until understanding has occurred. After that, the memory of actual words disappears and only the understanding remains. We rely on these subconscious processes.
We get upset when we can't recall a word (the tip-of-the-tongue phenomenon). And we think it is funny when words come out wrong, as in spoonerisms such as "This is a stickup. Don't nobody muss a moovle!" Speaking a language is a mental state in the sense that our brains resonate to whatever languages we speak fluently in appropriate situations.
Most monolingual people never get out of the mental state of their native language, and can hardly imagine doing so. Bilingual people can switch from one to another linguistic mental state. Perhaps this fundamental switching ability is what gives them an advantage in learning additional languages.
===
What is learning a language? Learning a language is a matter of habituating the body and nervous system to new patterns. For a first language, most attractors are established during a period of rapid neuronal growth. This period begins before birth, when the ear is fully developed.
The growth of neural connections in response to usage never really ends; it just slows down. In adults, it is about one-thousandth of its original rate.
This kind of difference means babies can drink in languages with their mother's milk, but adults--well, they have more difficulty than infants. If people speak two or more languages to an infant, the infant will learn to speak them. A person who is bilingual from birth suffers no damage either to language or to other mental functions.
The brain uses neural capacity that might otherwise have gone unused--not whole new neuronal nets but existing ones resonating to different situations. In the beginning, first-language learning is not self-conscious, just a set of habits one learns in the service of communicating.
School-age children gradually develop the ability to use language self-consciously, adopting different styles for different purposes and calculating the effects of different styles on listeners. Learning a second language after childhood is made difficult by the presence of established first-language attractors.
First-language voice quality, hearing, pronunciation, rhythm, intonation, etc., usurp neural networks before they can accommodate themselves to new words and patterns of the second language. For an adult, fluency in a second language requires thousands of hours of practice to develop strong attractors.
But because adults have the ability to conceptualize a language as a body of new material to be mastered, explicit learning is of some help. Just how much help is not easy to say, but attractor theory permits us to note that extreme positions are wrong.
It is not true, as some have previously thought, that conscious knowledge cannot be converted to subconscious processing skill. Nor is it true, as some others have thought, that you must notice a piece of language to master it.
Fluency in a second language, like fluency in the first, is a matter of mostly subconscious processing. As Nick Ellis, a psychologist of University of Wales Bangor in Britain, is fond of saying: "Our language processing ability emerges from the collaboration of memories of all the utterances we have experienced."
* * * This series of columns is an attempt to reconcile views of language teachers, theorists and bureaucrats. Readers are invited to send e-mail to mrchilds@tokai.or.jp or letters to The Daily Yomiuri. The column will return on May 9. Childs, Ed.D., is a lecturer at Temple University Japan, Tokyo, and Fuji Phoenix College in Gotenba, Shizuoka Prefecture.