Lecture 12

IIT Kanpur July 201
20 followers
Mar 2, 2024
[Music]
hello and welcome to the course
introduction to the psychology of
bilingualism and multilingualism I'm Dr
arwma from the Department of cognitive
Sciences it kpur and we have been
talking about critical period and age of
acquisition effects in the first
language in the previous lecture today
I'm going to to extend that discussion
towards learning a second or a third
language now if you look around if you
look at the you know the the way our in
our uh Society is now constructed in
2024 uh a lot of us a lot of individuals
actually have this opportunity to learn
a second language very very early in
life you know all of us including myself
or a bunch of people have actually had
this ability to learn a second language
which which is typically English for a
lot of us across India uh by going to an
English medium school you know whether
you are in in the north the Hindi belt
as they call it or in in the South uh
typically the second language that
people learn uh is by way of going to
school very early in life so our schools
typically start around 2.5 to 3 years of
age and once the child starts going in
that school uh the med medium of
instruction is English so what basically
does is it allows us to acquire English
very very early in life for a second
language also I mean English may not be
the second language for everybody for
for a lot of people it could be let's
say uh you know it could be Tamil versus
telu Tamil Canada etc etc depending upon
their familial and you know social
situation but the idea is that a lot of
us actually have an opportunity to learn
a second or a third language very early
in life but again this has not been the
case always this has not been the case
forever and for a lot of people it has
been observed that their opportunity to
learn a second language if they've not
gone to let's say an English medium
School uh typically comes slightly later
in life you know I know a lot of people
who migrated to a different country and
learned a second language at a much
later age in life so for example I was
living in Belgium and I had a bunch of
friends who had migrated to uh Belgium
after selling their Farms Etc and they
arrived in Belgium and the second
language they were learning was
interestingly not English but Dutch
because you know Flemish actually
Flemish which is spoken in the parts of
Belgium that I'm talking about now the
idea is uh due to these kinds of things
let's say education uh occupation or
immigration uh a lot of us learn a
second language much later in life
whereas a lot of us learn a second
language much earlier in life is the
quality or the overall proficiency that
one would attain in both of these cases
going to be different is and you've seen
uh you know how things go in
simultaneous versus sequential
bilinguals there are you know a bunch of
uh changes that you already experience
there is a bit of a delay when you're
talking about sequential bilingual for
mastering their both of the you know
mastering the phology of both their
languages and so on but if you move
further from there and if we try to see
that overall how well does an individual
end up picking up a second language or a
third language later in life all right
so uh this is basically what I'm going
to talk about today that is learning a
second language late in life difficult
and if it is difficult how is it
difficult what are the uh you know
characteristics of these difficulty and
are people able to you know come over
them and surmount
them so that is the reason precisely why
uh you know a bunch of researchers have
studied the age of acquisition effects
for second language acquisition and uh
you know they have discovered several
interesting insights for the same uh
I'll begin with a base study of Johnson
and Newport in 1989 who actually were
looking for testing evidence for the
critical period hypothesis for second
language acquisition remember we've talk
we've been talking about the critical
period hypothesis in the previous
lecture when you're talking mainly about
the first language uh learners but there
uh can also be a similar version of this
hypothesis for second language learning
as well and there are two versions of
this hypothesis that have some sort of a
bearing on how will individual
eventually fair when they pick up a
second or a third language going forward
so the two versions of these hypothesis
are the exercise hypothesis and the
maturational state hypothesis the
exercise hypothesis basically says that
early in life humans have a superior
capacity to acquire languages but if
this capacity is not exercised that is
if the first language is not picked up
very early uh and in in in a normal
manner then what would happen is because
of this delay in first language learning
they will also have uh you know
impoverished second language learning
later in life as well this is what is
called the exercise hypothesis the
second uh version of this hypothesis is
referred to as the maturational state
hypothesis where it is proposed that
early in life humans have a superior
capacity for requiring languages but
this capability keeps on declining with
the maturation of the brain so here
even if you've learned the first
language earlier or later does not
really matter the ability to pick up a
language later in life is anyways on the
decline and for older people it will
basically mean that learning a second or
a third language at a you know at a
older age is certainly going to be very
very difficult now uh if you look at
both the versions of this critical
period hypothesis you will find that
they actually make different predictions
for second language learning more
specifically while exercise hypothesis
predicts that if an individual has
acquired the first language early in age
the ability to pick up a second language
sort of stays throughout the life and
they can at any point in time they want
pick up a second or a third or a fourth
language because their you know ability
to learn language has been exercised
while they were learning the first
language the maturation State hypothesis
says something different what does it
say it says
that despite or regardless of the fact
whether an IND ual has picked up a first
language or not very early in age the
abilities is on the decline as I was
saying you know while closing the last
Slide the language learning capacity is
typically on the decline and will
continue to decline with age as a
consequences of the maturational changes
in the brain all right so these are the
two types of uh or the two variations of
the critical period hypothesis that were
considered by Johnson and Newport 1989
when they were describing their when
they were designing their study so they
reasoned that in order in order to prove
the maturational state hypothesis they
would need to provide evidence that the
ability to learn a language diminishes
with age decreases over time and there
is also a discontinuity you could
predict that would happen as a result of
the ending of the maturation period of
the brain so for example once the brain
is matured the there should be a steep
drop or let's say a flattened uh sort of
a plateau kind of a thing in the
language learning ability of a person
once the maturation is complete
then there should be no more increases
in the ability to learn language is what
we are saying so they tested these ideas
when they were chose to when they were
examining L2 speakers uh knowledge of
English morphology and syntax you know
different aspects of morphology and
syntax like word order tense information
plural information determiners uses of
auxiliaries and prepositions ET a test
across 276 items uh which Were You Know
spoken uh and recorded in an audible
manner uh there were 46 participants in
this study uh some were Chinese some
were Native Korean speakers who had
migrated to the United States between
the ages of 3 to 39 so you can see that
there's a huge variation between these
uh between their coming to the United
States which sort of provides us with a
nice spread of different times of age of
acquisition uh or different types of
being exposed to English and we are
testing them for English so basically it
means that okay uh somebody who arrived
to the US at 5 years of age versus 35
years of age will obviously have
qualitative differences in the way they
have learned
language they also included only L2
speakers as the you know Baseline for
for Baseline comparisons and some you
know some of them who had obtained a
University degree and had been immersed
in English for at least 5 years you know
basically just to control that all the
participants has reached their eventual
final stage of L2 proficiency now this
is interesting because you can see that
some of these individuals have uh you
know may have arrived earlier in uh uh
their 3 to39 period some of them would
have arrived later in their 3 to39
period so in some sense there is not a
uniform time uh available to them but
the researchers are actually using this
criteria to uh in some sense limit that
okay in their assumption or in their
study everybody has had enough time to
reach the final stage of their English
proficiency because we're testing for
English all right so uh the experimental
task involved grammatical judgment for
each of the 2
276 spoken English sentences presented
by means of an audio tape and they
basically had to tell whether each of
these sentences were grammatical or
ungrammatical that is supposed to be the
test of a individual's Proficiency in
language remember I was talking about
the fact earlier that in some cases
people uh use uh you know self-rated
tests of uh language proficiency but
here we are using a more objective
method that okay out of these 276 how
many of these sentences you can
correctly judge to be grammatical versus
ungrammatical so the results actually
showed a very strong relationship
between the age of arrival in the US and
their grammatical knowledge you can see
it's a very you know it's a significant
a very high correlation of minus 77
wherein it is clear that people who
arrived earlier to the United States
early in their life basically if they
would have arrived three four five six
years of face they would be performing
much better as opposed to people who
have arrived let's say 34 35 36 and so
on more specifically uh if you zoom into
these results and we sort of discuss it
in more detail you would find that the
best performance uh in this task was
shown by the people who had the earliest
age of arrivals let's say between the
ages of three and seven and these
people's performance were very similar
to the performance of the native English
speaking group who had been exposed to
English for a long time
also uh interestingly the performance of
people who were you know in these older
age groups 8 to 11 11 to uh 15 and 17 to
39 years of age actually performed much
worse than to the native speakers so in
some sense the suggestion is that people
who have arrived after their critical
period has passed have not been able to
attain native like proficiency of L2
native life Proficiency in
English these results if you see on the
face of it uh do support the idea of a
critical period now moving on to the
second part whether and how brain
maturation impacts language learning
basically what Johnson and uport did was
that they divided their participants
into two main groups one from 3 to 15
years of age and the other from 17 to 39
years of age the idea is that between 3
to 15 years of age the brain maturation
is still going on by 17 to 39 years of
AG the brain maturation is completed and
basically now if you do a comparison you
will will also see the effect of brain
maturational or maturational state
hypothesis in these
people what did they find the former
group the earlier arrivals those who uh
you know fall between 3 to 15 years of
age show a much higher negative
correlation of 87 than the latter group
of uh you know people arriving between
17 to 39 years of age of just minus uh6
which is in some sense interesting but
it also tells us that it is very very
important time at which they arve to the
United States was a very very
significant factor in their eventual uh
proficiency of learning English all
right uh also an interesting result here
was that there was much lower variance
in the first group as opposed to the
second group so there were people all of
the people uh let's say out of the 46
let's say 25 people fell in the first
group they were all very coherent they
were all you know lying very near to
each other in their uh grammatical
judgment performance as opposed to uh
the older individuals or the older Group
which had a lot of variation uh and
we'll discuss where these variations
might be coming
from based on these two or three
findings the authors actually deduced
that their results supported the
maturational state hypothesis basically
saying that people uh you know do not
have the best of the capability to learn
a language after the maturational period
of the brain has
finished now let's look at this in in a
bit more detail because there were a
bunch of studies which followed up
Johnson and Newport study and and these
studies actually revealed newer insights
into this process they looked at this uh
both of them bird song and mly 2001 and
Deer
22000 uh both of them sort of created
you know replication of Johnson and
Newport study but actually ended up with
very different insights let's dive in
and look at this in more detail so deer
on examining uh L2 English speakers with
uh you know whose first language was
Hungarian concluded that his data
actually agreed with the findings of the
original study and in that sense was
also in support of the maturation State
hypothesis however in contrast data from
Birdsong and mollis who tested their
native speakers of Spanish pointed out
that discrepancies between their
findings and that of Johnson and
newport's findings and in that sense
they suggested that there could be other
factors other than this bounded period
or bounded critical period that we are
talking about that have consequences on
the overall language learning abilities
of
individuals okay let's zoom in a little
bit DK actually when they were designing
their experiment noted three very
important points about Johnson newport's
study what are those first firstly in
their study uh they kept the minimal uh
length of residence of the participants
in the US up to 10 years as opposed to 5
years in the previous study because they
thought that you know if the minimum
length of residence of these people were
just 5 years it is not enough time for
them to have reached the final stage of
that English proficiency so when they
were tested they might still be going
undergoing the process of learning
English and they were probably tested
prematurely and given more time they
would have reached the final Peak so as
opposed to Johnson in newport's 5 years
the cas had actually kept a period of 10
years minimum residency period of 10
years in the US to give their
participants ample time to have picked
up the eventual uh final stage of
English learning second is uh in Johnson
and Newport study the age of arrival has
been confounded with the parman age at
test taking uh it is interesting because
what could be implied if you look more
closely at the results from Johnson and
Newport one would say that some of these
test takers were actually slightly older
35 36 37 uh when they were being tested
and that could have impacted their
overall performance for example in terms
of slightly diminished detentional
skills and you know uh slightly uh
impoverished other senses uh that could
have could have been the factors that
are actually you know uh uh ending in uh
in their slightly poorer performance on
grammaticality judgment and so on
finally uh you know a differential
ability to remain concentrated
throughout this long test now 276 items
is a real long test and for uh basically
uh some of these people may not have
been able to concentrate across the
length of the test which is a long 276
items test and because they sort of they
were they were lapses in concentration
their performance would have been
decreased so deaser uh basically
addressed these things uh and introduced
some vital changes in their Paradigm
while they were still replicating
Johnson Newport study they increased the
minimal length of residence in the US
from 5 to 10 years they corrected for
chronological age in the data analysis
and they also reduce the test items from
276 to just 200
items additionally they also
administered a language learning
aptitude test now remember I was having
this discussion earlier when we just
began this course in the first week that
there are also factors other than
biology and uh let's say you know normal
functioning brain uh that could actually
contribute to how eventually or how well
or how uh uh proficiently a typical
individual can Master a language all
right so that people have talked about
things like the aptitude for learning a
language as well and this is the kind of
test that DK administered they
administered a language learning
aptitude test to test whether the
participant had any individual
differences amongst them uh with respect
to their language learning abilities so
somebody who's not really motivated
somebody who does not have the aptitude
for learning a language irrespective of
the time that they had arrived in the US
will not be able to master English
however somebody who has just arrived
but has a very good aptitude in picking
up a language may actually outperform
these other person who has arrived at a
much earlier uh time in their lives and
so on so this they thought would be able
to explain the large variability in the
data uh on the second half of the data
where we are talking where we are
talking about and uh evaluating the
people who have arrived slightly later
in the United States let's look at this
further now this basically derives from
uh you know what is referred to as the
fundamental difference hypothesis first
forwarded by blle and Roman in 1988
which basically posits that while
children are able to utilize an innate
language learning mechanism AKA chsky uh
that allows them to have implicit and
unconscious language learning adult
second language Learners uh typically
need to resort to explicit conscious
learning strategies that sort of makes
it easier to them or that sort of allows
them to pick up a language okay so they
typically typically are relying on their
individual problem solving cap
capabilities their abilities to reflect
on the structures of the target language
and so on and so forth which would
eventually decide how well they are
finally able to learn the language
individuals who would be good at these
skills you know good at General problem
solving capabilities good at uh you know
adopting the best strategy to learn and
so on and we have higher you know verbal
analytical skills we'll eventually pick
up language more native like language as
opposed to individuals who are lower on
these
skills let's look at the results from
deaser in a bit more detail uh in
agreement with deers 2000 deers
hypothesis individuals who actually
you know had better verbal analytical
skills amongst the later Learners were
actually able to compensate for the fact
that they arrived late and you know
their exposure to the second language
happened later in life uh these were
those few AR late arrivals who would
actually obtain you know very good
scores on the grammatically gram
grammaticality judgment task uh and
those their scores were within the range
uh of those observed uh from the early
uh arrivals as well indeed uh if you
look at it in more detail among the
participants who arrived at the age of
17 or more the correlation between
grammaticality judgment score and the
verbal analytical score was a
significant 33 again it's not very high
but it is something that explains a
bunch of variance here for the early
arrivals before the age of 16 who
arrived before the age of 16 actually it
was it did not matter whether they
possessed very good verbal analytical
skills or not the uh Co was certainly
non significant of minus School 07 but
more importantly it tells us that higher
verbal analytical skills typically come
into play when we are consciously trying
to learn a language which would happen
if we are having this input much later
in life as opposed to when we are sort
of picking up language uh effortlessly
nonchalant because of let us say a
species specific language specific
language learning mechanism that we are
endowed with that we are born with all
right
another interesting finding in the study
was that when all the participants were
grouped together deaser observed again a
very strong negative correlation between
the age of arrival and the
grammaticality test score minus. 63
which is very similar to minus. 77 that
Johnson and Newport had got
interestingly the correlation here was
not significant if the partner if the
participants were divided into two
groups of those who arrived before 16
and after 16 so -26 and- 0.04 so then
here now you can see that this result
that was being interpreted as being in
support of the cph credal period
hypothesis is not as well replicated so
the absence of this age effect in a
group of early arrivals actually in some
sense if you look at it constitutes a
challenge to the critical period
hypothesis as the maturation uh account
of critical period hypothesis would
predict finding this very strong
negative correlation between the age of
arrival and grammaticality judgment
scores this was not the case in uh decer
study also the hypothesis predicts that
the occurrence of specific
discontinuities uh in the age function
can be directly related to maturational
change of the brain all right so it
predicts the more Z kind of age function
which you can see here the hypothesis if
you see typically predicts uh something
like uh you know what you're are seeing
in uh figure a on the top uh which is
the common view of the critical period
hypothesis where you will have a height
end period of sensitivity early closer
to birth which sort of uh flat reaches
it Peaks flattens out and then starts
declining as the maturation is getting
completed the stretch Z function uh
basically refers to uh the ultimate
performance the age of accusation uh you
know to Ultimate performance of language
learning uh which is also you know as
predicted by the critical period
hypothesis now this if you look at the
critical period hypothesis in this way
uh it basically tells us that you know
at some point you know at some early
point in life uh or maybe just soon
after birth humans actually have this
increased sensitivity to linguistic
input that takes a very small amount to
reach the peak it's a very steep curve
if you see the first part on the top
figure and then it reach reaches it Peak
and after some time it starts declining
gradually and then flattens out till the
after the brain maturation is completed
now as for this version of this
hypothesis if the learner were immersed
in this L2 en environment within that
first period or in the full period of
the highest sensitivity native like
efficiency can be obtained however if
acquisition starts slightly later and
coincides only in the second part of the
peak sensitivity continuing into the
stage where sensitivities started to
already decline remember I'm talking
about the peak part and the decline part
uh you know two and three in on the top
figure uh or it only starts somewhere in
the middle phases native like
performance will be hard to
achieve so to conclude the critical
period hypothesis predicts uh you know a
very specific kind of age function that
includes two discontinuities as
visualized in uh you know this figure uh
2.11 B the bottom figure uh but
interestingly deers data do not meet
these requirements they are not talking
they don't uh account for the or the the
data are not able to account for the uh
L part all right given that the
participants within that older group
differ from one another to the extent in
which they been able to profit from the
HZ period of height and sensity so it's
basically uh the data is not really
being able to uh or let's say the theory
is not being able to explain uh The
observed data from deer's experiment
there are also studies which actually go
a bit further in in some sense providing
counter evidence to uh the critical
period hypothesis as was presented by
Johnson and Newport one of the
interesting studies in this regard was
that of bir song and mes and they
basically replicated the same experiment
by Johnson and Newport using typically
the same Paradigm same materials same
stim 276 stim and so on and so forth uh
only difference being was that here the
individuals first language was Spanish
uh whereas in Johnson and Newport if you
remember uh the first language was
Chinese or
Korean so what did the results of birung
and Mish show although they could um
obtain you know the first discontinuity
in the in the age function the shape of
the age function as they have observed
was SL was significantly different from
the shape of the age function that is uh
in some sense predicted by the uh you
know critical period hypothesis you can
see here uh you know the straight line
uh it basically tells us that the first
part of the hypothesis is there but the
second part is is not
predicted when you consider it in more
when you look at this data more closely
you would see that when uh you know you
consider the participants of bird song
and mes together they have obtained the
exactly the same correlation between uh
you know arrival age and grammatical
skills of minus. 77 however if you you
know delve into a little bit more detail
you discover that once you divide the
participants into before 16 years of age
on arrival and after 16 years of age on
arrival uh the findings actually get
reversed which is very interesting
because there was a non-significant
24-24 as opposed to
-87 and uh uh you know a significant
minus. 69 as opposed to -.6 .16 in uh
Johnson in Newport study so you can see
that here in this case uh the age of
arrival is actually being able to
predict the grammaticality Judgment
scores in the older individuals much
better than in the earlier individuals
which is interesting if you look at
it so just sort of stepping back and
looking at it bird song and ms's data
actually deviated from those predicted
by the hypothesis uh in three main
aspects first the earlier AR arrivals
showed little difference amongst them in
terms of ultimate grammatical
performance although among them they
still still differred greatly in the
number of years uh you know uh their
elative acquisition coincided with the
period of this peak sensitivity that we
are talking about secondly uh an effect
of age of arrival was obvious in the
group of late arrivals not in the early
arrivals even though the L2 immersion of
all these participants in this group had
started after the closure of the
critical period so remember we are
talking about 16 to 39 here the age of
uh arrival is actually you know more
coordinated with the grammatical
judgement scores than with the earlier
ones finally uh it does not seem like
there is an end period or to this age
related decline in the final performance
of the late arrivals implying that the
second disconnected continuity that uh
the model predicts uh is missing in The
observed
data So based on these observations uh
birung and mollis reject the critical
period hypothesis and they say say that
uh you know it is very difficult to
produce evidence in favor of the
critical period hypothesis now there are
three studies I've talked about Johnson
and Newport uh which confirmed this data
but two studies later which were bird
song and molish and deaser actually do
not confirm to the predictions of the
critical period hypothesis another
interesting finding from their study was
that the incidence of near native like
performance differs between the two
studies so the eventual near native like
performance that happens for Johnson and
newport's participants remember these
were people having Chinese and Korean uh
and uh birung and mes participants who
were you know whose first language was
Spanish is actually very different the
Peaks that they reach are very different
the uh birung and mol's participants
actually reach a much higher Peak as
opposed to the peak reached by the
Chinese and Korean participants why
would that happen and the researchers
sort of zoomed in and looked at it and
they basically figured that because of
the cultural similarity between Spanish
and English as opposed to both Chinese
and Korean and English on the other hand
because Spanish is closer to English and
Chinese and Korean together are not uh
this may have caused the difference
between the Peak Performance of these
individuals in these two studies so
basically you can see that the
structural similarity between these
languages or the typological distance as
we sometime call it actually also have
an has an effect on how well one is
going to pick up or learn a second
language
further uh you know one of the studies
such as one done by hakuta and
colleagues actually showed that final L2
proficiency also depended upon factors
like educational level at the time of
arrival you know if you uh if people who
are who arrived at a higher educational
level actually or their educational
level predicted their eventual L2
learning performance much more than
their age of being exposed to L2 in some
sense it says that while critical period
may be an interesting
you know concept to follow uh there are
also other factors like verbal
analytical skills or the overall
educational level that actually predicts
how well people will eventually uh learn
a given language there are also other
sources of contor evidence to the
critical period hypothesis uh there
would be instances of native like
ultimate attainment in late L2 Learners
you know if people are super motivated
if it depends uh if their life depends
on it if their job depends on it people
who take asylum in different countries
people who immigrate to foreign
countries they eventually pick up
language I mean you you look around and
you'll find that they do pick up
language uh very very native like or
very good levels of language at a second
or you know even though they have
arrived uh very late in that sense and
this would certainly go very counter to
what we have been talking about under
the critical period
hypothesis also uh a lot of people who
are early L2 Learners can not reach
native like performance because let's
say their aptitudes vary the motivations
vary there are other factors and so on
and so forth in conclusion if you look
at all of this you would see that while
critical period hypothesis does predict
this kind of dises in age functions
these function these a this continuities
do not really materialize in The
observed data we see that only Johnson
and newport's data sort of supported the
hypotheses but not of the other studies
that we discovered after that also the
cphs false predicts that after many
years of using the L2 all early learners
but no late Learners will achieve n like
proficiency but we have seen instances
that late Learners of L2 also sometimes
can reach native life
proficiency finally uh one should sort
of consider as a parting note that if
there is a period of special sensitivity
to linguistic stimulation during let's
say some bounded period early in life 3
years 7 years 12 years maybe uh uh other
factors like Lang educational levels and
structural similarity between the two
languages should not matter but we have
seen like in hakuta study hakuta and
colleagues' study that it does matter so
in some sense uh it seems if you look at
all of this evidence holistically uh the
critical period hypothesis or this
bounded period of height and sensitivity
uh is not a necessary condition for
individuals to learn a second or a third
language later in life all right and it
is based on these kind of uh you know
assumptions or based on these kind of
observations Singleton actually opines
that the critical period hypothesis
cannot be plausibly regarded as a
scientific hypothesis either in the
strict poian sense because it cannot be
falsified you know as I said earlier you
cannot in controlled conditions keep
somebody devoid of language input for
any length of time and then see study
carefully the effects of such a
deprivation also whatever you know
experiments that are being done and a
lot of data that is come out uh with
respect to the cph does not clearly
confirm or support this hypothesis as
well all right so uh good news for
people who want to uh learn a second or
a third language later in life there is
certainly seems to be chance that you
can do it if you are adequately
motivated and so on uh and a little bit
of a uh drubbing for uh you know the
biological theorist where we talk about
that no the critical period hypothesis
is the bounding Factor uh as to how or
how much an individual can learn a
second or third language on that note
I'll end the lecture uh of today uh I
will start from the next lecture talking
about uh a different aspect of language
uh bilingualism and multilingualism
which is more about speech production
all right thank
you
[Music]
[Music]
[Music]
