Доклад: Language Acquisition in Infants

Running head: Language Acquisition in Infants
                         Language Acquisition in Infants                         
                             Yelizaveta Mirkovskaya                             
                      University of California, Los Angeles                      
                         Language Acquisition in Infants                         
The unknown has always puzzled the human brain. When marvelous scientific
technologies were unavailable to solve the mystery of human intelligence,
scientists made numerous guesses as to which processes might underlie the
smooth and complex functioning of the brain; modern technologies made it
possible to look inside the human scalp and start decoding NatureТs most
complex creation Ц the human mind. Language is an extremely essential
attribute of human beings, which distinguishes them from the rest of the
animals. However, its mystery started to be unwound only recently.
How humans acquire ability to separate and construct words is one of the
central questions of the modern research in the field of language
acquisition. It has been proposed that to construct words, human infants use
sequential probabilities; that is, they are innate mathematicians and use
simple statistics to perceptually categorize particular sound groups,
calculate their frequency in the ambient stream of sounds, and predict which
group of sound should follow the certain other one. For example, Japanese,
Swedish, and English infants (i.e., representatives of different types of
languages) demonstrate increased sensitivity to the phonemes and prototype
vowels specific to their language, and do not respond as highly when hearing
phonemes and vowels pertaining to languages, other than their native one. The
experiment on eight-months-old infants also showed that when discriminating a
change from a particular syllabus to another one in the string of syllabi
(i.e., УdeФ to УtiФ), infants who succeed the most are the ones who learn a
pattern of syllabi in the sequence (i.e., perceive combinations of certain
syllabi as words) by learning that the sequential probability of a certain
syllabi given a certain other syllable is one; that is, the syllable will
undoubtedly occur. As hypothesized, learning to chunk certain syllabi into
words on the basis of statistical frequency reduced perceptual demands of the
lingual task for the infants, and made it easier for them to fulfill the task
of distinguishing one syllable from another. Another experiment on infants,
based on the fact that infants pay more attention to novel stimuli,
demonstrated that once having heard the two-minute strings of invariantly
repeated syllabi (their order was not changed), infants showed novelty
preference when they heard an unexpected order of syllabi when tested by
listening to three-syllabi chops of the two-minute sequence. Since the pure
statistical frequency of occurrence of all of the three syllabi was the same,
it was the sequential probability of occurrence of a certain syllabus that
made the infants heighten their attention to the crossing-word-boundaries
combinations. In addition, even younger infants are able to discern the
prosodic pattern (trochaic, with the stress on the first syllable of the
word, or iambic, with the stress on the second one) of the native language,
based on the sequential computations.
Sequential probabilities are definitely important to acquisition of language
skills in infants, but they are not the only mechanism of mastering a
language. Rigid neural networks, produced in the first year of life, encode
the patterns of native language and interfere with the ones of foreign
languages. This way, it is easier for the infant to keep learning the native
language as more and more neural paths and connections are created in the
brain; growing number of these native connections interferes with creating
different ones for foreign languages. When adults are subjected to hearing
foreign speech, larger areas of their brains are illuminated for longer times
during magnetoencephalography as compared to when they hear native speech,
which indicates that more effort is required on the part of the adult to
process the speech sounds. Similarly, infantsТ language skills at an older
age strongly correlate with their speech discrimination skills later in
development. A longitudinal research on seven- and eleven-months old infants
showed that at the age of seven months, infants still discriminate between
native and non-native consonant combinations, whereas at the age of eleven
months this discrimination is almost absent. This distinction illustrates the
fact that with age, neural networks underlying acquisition of language
strengthen and filter out foreign sound combinations more effectively. If it
were not for neural commitment of neural networks, infants should have
reacted to combinations in each language equally strongly. Neural commitment
is also said to be responsible for the Уsensitive periodФ during language
acquisition, during which, according to the NLNC (Neural Language Neural
Commitment) model, brain networks solidify and become more sensitive to the
patterns of native language. Sensitive period closes, when neural networks
stabilize (i.e., become neurally committed).
As suggested in the article, neural networks stabilize when they stop being
sensitive to speakersТ individual differences in pronunciation of various
speech sounds; that is, when infantsТ representations of sound distributions
become stable and rigid. In this regard, bilingual childrenТs neural networks
should take longer to stabilize since there is simply more of them (at least,
two systems instead of one), and the perceptual load on a bilingual
individualТs brain is much greater. A longitudinal quasi-experiment could be
performed to test this hypothesis. Participants in this experiment -
bilingual and monolingual infants at ages of six, nine, twelve, fifteen,
eighteen, twenty one, and twenty four months Ц will be subjected to hearing
series of computer-generated sounds, demonstrating different pronunciations
of the same vowel. Each vowel will play for about two or three minutes in
order to give infant enough time to familiarize with the task; about five
vowels will be tested with the order of their appearance counterbalanced, as
to control for the order effects. Some distracter sounds will be played
between the presentations of the vowel, so that the infant does not get used
to the simply presence of a monotonous sound from the source.  The dependent
variable in this experiment Ц the sensitivity of an infant to different
pronunciations of the same vowel Ц will be tested by observing infantsТ
instances of behaviors caused by distracting novelty (turning of the head
towards the sound source, interrupting the on-going activity, etc.) while the
sound is playing. The more distracting behaviors observed, the more sensitive
the infant is to the differences in presentation of the same sound. Any
extraneous variables that might interfere with the results will be kept
roughly constant: participant infants will be taken from families of similar
socio-economic status and will be of similar health conditions, sounds played
will be the same throughout the whole experiment, infants will be tested on
the day when they reach the next appropriate age for testing, lab conditions
will be kept constant for all of the infants and throughout the length of the
experiment. The experimenters will make sure that for the bilingual group of
children, they will be exposed to the non-English language equally often as
to the English language throughout the whole experiment; for the monolingual
group of infants, they will not be exposed to any language, other than
English.
The results of this experiment, if consistent with the hypothesis, will show
that more distractive behaviors will be observed in bilingual infants because
of higher perceptual load of two languages on their brain. For the
monolingual group, less distractive behaviors due to novelty stimulation will
be observed. If, for example, for monolingual children, neural networks will
stabilize by the age of one year, then, for bilingual children, the neural
networks will be committed at the age of about fifteen months; that is, later
(these ages are hypothetical).
Language acquisition in human infants is a starting-to-be-resolved mystery to
the mankind; its resolution will help to treat children with linguistic
problems and ameliorate language learning in healthy children.