快猫短视频

FOR two weeks after her stroke, a 68-year-old patient, EM, was mute. Then she
started to speak. On the face of it, she seemed to have made a complete
recovery. But when her relations came to see her, they were puzzled because she
answered them in Italian, her second language and one that she had rarely used.
The language she had spoken every day of her life from infancy was the Veronese
dialect of northeastern Italy, very different from standard Italian. While she
could still understand her mother tongue, she was no longer able to utter a
sentence in it. As far as her ability to speak the Veronese dialect was
concerned, the stroke had neatly excised it.

EM is not the first patient to have an entire language or dialect wiped out
after a stroke, and will certainly not be the last although it is far more
common for the native tongue to be spared. But it is cases like hers that gave
the first hints that non-native languages might be filed in separate areas of
the brain from the mother tongue.

New research is now showing that where a language is represented in the brain
depends on when鈥攁nd, to some extent, where and how鈥攊t is learnt.
True bilinguals who learn two languages together from birth show different
patterns of organisation from those who learn a second language as a teenager or
adult. And the foundations for that organisation may be laid down by our
experiences very early in life鈥攂efore we even learn to speak. By looking
at what is happening in the brain at that time, researchers hope to explain why
children are so much better at picking up languages than adults. And, perhaps,
to offer us some tips as to how to compete with them.

A newborn baby is a bundle of potential as far as language is concerned.
Infants can learn any language spoken to them and can distinguish between any
sounds from any language. A six-month-old Japanese baby can clearly hear the
difference between the words 鈥渞ight鈥� and 鈥渓ight鈥�, but when it comes to a
Japanese tourist asking directions in New York or London, the potential for
confusion is endless. The distinction between the 鈥渞鈥� and 鈥渓鈥� sounds is not made
in Japanese, and by adulthood it appears to have been wiped from the Japanese
brain. In fact, says Marie Cheour of the University of Helsinki鈥檚 Cognitive
Brain Research Unit, the distinction is erased before a child even learns to
speak.

Last September, her group produced the first neurophysiological evidence that
memory traces for sounds specific to certain languages, in this case Estonian
and Finnish, develop before the age of one (Nature Neuroscience, vol 1,
p 351, 1998). The Estonian and Finnish languages have very similar vowel
structures, except that Estonian has one vowel, 鈥溍碘��, that does not exist
in Finnish. Its sound falls somewhere between 鈥溍垛�� (which sounds like
鈥渆r鈥�)鈥攁 vowel that is common to both languages鈥攁nd 鈥渙鈥� (which sounds
like 鈥渙r鈥�). By monitoring the activity in the auditory cortex of six-month-old
Finnish or Estonian babies, using a set of scalp electrodes, the researchers
could see a clear signal (the 鈥渕ismatch negativity鈥� or MMN response) when an
鈥溍碘�� or 鈥溍垛�� was sounded as the odd-one-out in the middle of a string
of other identical vowels. This shows that the babies can easily distinguish the
different vowel sounds. But by the age of one, the Finnish infants hardly
noticed the odd 鈥溍碘��, while the 鈥溍垛�� still stood out. Their brains鈥�
capacity to hear the difference between 鈥溍碘�� and other vowels had all but
vanished.

鈥淲hen we learn a language we build permanent memory traces that represent our
native language sounds to our brains,鈥� says Cheour. Without those memory traces,
you cannot become a 鈥渘ative listener鈥� of the language, and hence you are less
likely to be able to speak it. As Cheour puts it, 鈥淚f you don鈥檛 hear the
difference between two sounds, how can you say them correctly?鈥�

Cheour鈥檚 finding shows that there is a very early critical period in language
acquisition, for establishing phonetic architecture鈥攖he first and crucial
stage in learning a language. According to Patricia Kuhl of the University of
Washington in Seattle, the timing of this critical period is dictated by our
experience. She thinks that infants are born with certain innate boundaries for
discriminating between speech sounds, and that these are either reinforced or
pruned through experience. Sounds that are heard regularly in the native
language are somehow captured as memory traces.

The memory traces for similar sounds that are considered by the brain to be
examples of the same phoneme are all drawn together and treated as a single
category, while the boundaries between different sound categories widen. In the
Japanese case, 鈥渓鈥� and 鈥渞鈥� come together in a single cluster, but in the English
language there will be a distinct boundary
(see Diagram). Kuhl calls this
the 鈥減erceptual magnet effect鈥�. Hearing the same sound over and over reinforces
a particular representation. The critical period is defined by that
reinforcement, ending when the connections can be reinforced no more. This
collection of memory traces, which corresponds to all the sounds required by a
particular language, can be thought of as a mental map, that allows us to
process sounds in an ideal way for each language, says Kuhl.

When immersed in a foreign language, however, the map in your brain acts as a
filter. Faced with a voluble New York cab driver, the non-English speaking
Japanese tourist forces the speech through that filter. In the process, certain
crucial distinctions are lost. Not only does the Japanese tourist not hear them,
but when he or she responds in rudimentary English, his speech will also lack
those distinctions. Hence it will sound heavily accented. But, says Kuhl, 鈥渋n
true bilinguals there are two distinct maps, one for each language鈥�.

Having two separate maps is important if two languages are to be spoken and
understood well, but this doesn鈥檛 mean that they have to be separated in space.
In fact, the opposite may even be true. Although brain imaging hasn鈥檛 yet
allowed us to pick out the phonetic maps specifically, two years ago Joy Hirsch
of the Memorial Sloan-Kettering Cancer Center in New York and her colleagues
used functional magnetic resonance imaging (fMRI) to look at the speech region
known as Broca鈥檚 area. They showed that in bilingual people who learnt both
languages early in life, the two are represented in overlapping regions of the
left frontal lobe. 鈥淭hese data suggest that if multiple languages are acquired
early in life, the brain treats them as if they were one language,鈥� says
Hirsch.

Of course, it is possible that the two languages are separated on a finer
scale that we can see using brain imaging methods. But the apparent overlap begs
a question that often worries bilingual parents鈥攚hether they will muddle
their child by speaking two languages together. If it鈥檚 a case of two languages,
one location, why don鈥檛 the two don鈥檛 get confused?

Arturo Hernandez of the University of California at Santa Barbara thinks he
may have the answer. Two years ago, he used fMRI to eavesdrop on the brains of
Spanish-English bilinguals as they named pictures first in one language, then in
the other, or alternating the two. In the first two instances, their brains
showed similar activation, but in the alternating task, the left prefrontal
cortex seemed to be working overtime. Hernandez concluded that it was playing
some role in reducing interference between the languages.

But it鈥檚 only when two languages are learnt together early on that they seem
to occupy overlapping areas. If a second language was acquired later, Hirsch
found that two separate regions about 8 millimetres apart were activated within Broca鈥檚 area
(This Week, 12 July 1997, p 7). And this pattern isn鈥檛 just seen in
speech areas. The variability gets even greater if you take other areas that
deal with semantics and syntax into account.

In 1997, a group led by Stanislas Dehaene of the CNRS, the French national
research institute in Paris, asked late bilinguals鈥攖hose who, like EM,
acquired their second language after the age of seven鈥攖o listen to stories
in one or other of their languages. This task required not only an analysis of
the phonetic structure of the speech, but also an understanding of semantics,
grammar, and so on. They used fMRI to show that the second language was
associated with a variety of brain regions distributed across both hemispheres,
and sometimes with the right hemisphere alone. Most children start speaking
their first language at about the same age, explaining, perhaps, why the native
language is normally associated with the same regions of the left hemisphere.
But the second language may be acquired at any time, or even haphazardly over
many years, which may explain the variability.

鈥淭he main message is that there are many critical periods,鈥� says Helen
Neville of the University of Oregon in Eugene. The basic phoneme contrasts may
have been carved out in the brain by the age of one鈥攁s Cheour鈥檚 research
demonstrates. And ideally, syntax and pronunciation should be in place by the
age of seven鈥攍ong before most English-speaking schoolchildren have even
learnt to say 鈥淏onjour鈥�. But word power is more malleable. You can go on
building a vocabulary with ease well into adulthood.

The earlier you start to learn a foreign language, the better you will speak
it. Nobody disputes that. But timing is not the only factor determining where a
second language is represented in the brain. There are other influences, too.
For one thing, it may depend on what the two languages are. 鈥淚 think it is very
likely that the first language learnt and its degree of similarity to a second
language learnt strongly affects the organisation of the second language,鈥� says
Neville. Spanish and Italian might be organised differently from, say, Spanish
and German. And research presented to a meeting of the British Psychological
Society in April suggests that where a second language is represented in the
brain depends not only on when it was learnt, but where.

Judith Evans of the University of Glamorgan in Wales found that among
teenagers who spoke both English and Welsh, the lateralisation of the second
language, Welsh, depended on which part of Wales they came from. Among those who
had grown up in the west of Wales, where both languages are spoken, Welsh was
represented in the left hemisphere, regardless of when it was learnt. But in the
English-speaking community of South Wales, the age when Welsh was learnt made a
real difference: those who had learnt Welsh before the age of six showed a left
hemisphere bias, while those who had acquired it later showed a bias to the
right.

Just having a language wash over you in childhood, says Evans, ensures that
the brain selects for it鈥攐r, to use Kuhl鈥檚 terminology, creates a distinct
phonetic map for it. You don鈥檛 actually have to learn to speak it. As far as
carving out the phonetic contrasts is concerned, listening is enough, although
you still have to learn grammar and vocabulary. In fact, she adds, if you were
to regularly play a foreign language radio station in the child鈥檚 bedroom, it
could have the effect of recreating that dual language environment. Your child
wouldn鈥檛 instantly start speaking the language, but if she learnt it later on
she might pick it up more quickly and speak it with less of an accent.

But while it鈥檚 true that if you learn your second language after puberty,
your chances of speaking it without an accent are reduced, it is not the case
that linguistically, life ends at 13. Nobody ever said that adults couldn鈥檛
learn a foreign language. If that were the case, Berlitz would be out of
business, remarks Steven Pinker of MIT. 鈥淭he difference,鈥� he says, 鈥渋s that
children don鈥檛 need to be trained with artificial speech, don鈥檛 need structured
lessons and always succeed nonetheless.鈥�

But if we get the structure of the lessons right, there is still a surprising
amount of give in the adult brain, as a paper soon to be published in the
journal Psychophysiology demonstrates. In collaboration with the
Finnish group, Istvan Winkler of the Hungarian Academy of Sciences in Budapest
found that a Finnish phoneme contrast that does not exist in the Hungarian
language elicited almost no MMN in native Hungarians who knew no
Finnish鈥攁s you might expect. But in Hungarians living in Finland who had
learnt the language as adults, it elicited a large response, showing that the
phonetic map can be taught some new tricks in adulthood. Which may mean that
those memory traces are not as permanent as Cheour suggests.

Indeed, a group led by Jay McClelland of the Center for the Neural Basis of
Cognition at the Carnegie Mellon University in Pittsburgh believe they have
found a successful strategy for recreating some of the phonetic distinctions
lost in the first year of life. At a meeting of the Society for Cognitive
Neuroscience in Washington DC this April, McClelland described how in just three
20-minute sessions, 16 native Japanese speakers were trained to distinguish
鈥渞oad鈥� from 鈥渓oad鈥�, and 鈥渞ock鈥� from 鈥渓ock鈥�. They did this by using a computer to
manipulate the first parts of the words to make the contrast between 鈥渞鈥� and 鈥渓鈥�
as distinctive as possible, and the listeners tapped different computer keys
according to which sound they thought they heard. The better they got, the less
distinctive the sounds became, until eventually they were distinguishing sounds
from the equivalent of slurred speech.

It鈥檚 a promising finding, because it suggests that the phonetic maps, and
hence the fate of our language skills, are not fixed in the sense that neurons
are pruned and lost forever. Instead it looks as if connections can be regrown,
and new language maps learnt, given the right learning methods. Most of the
research suggests that the best way to learn a language is to listen to foreign
sounds in your cot, to master the grammar by the time you can tie your
shoelaces, and then to relax and learn the vocab at your leisure. But perhaps
there is, after all, some kind of latent capacity to learn new languages, to
which we have yet to find the key.