Lecture 13

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 am Dr
arwma from the Department of cognitive
Sciences it kpur we are in the third
week of the course and from today
onwards I'm going to talk to you about
speech production in bilinguals and
multilinguals what is speech you know
speech is in in some sense it's one of
the most fundamental skills that we uh
you know take for granted but basically
that is the language that we sort of you
know that is how we understand our
language or communication abilities uh
writing came much later reading came
much later but one of the primary uh uh
uh manifestations of human communicative
ability is language which is expressed
through uh spoken words spoken sentences
and hence uh is basically referred to as
speech
the question that we are going to talk
about in the coming two three lectures
is mainly concentrated about how is how
do bilinguals speak or how do say for
example uh they manage uh speaking in
two languages without an without a lot
of interference from either one how do
they switch across languages how do they
code mix a lot of times how do they uh
basically uh able to how are they
basically able to communicate
effectively having known uh or you know
having
known two or more than two languages now
the thing is if you need to understand
how you know speech production Works in
bilingual or multilinguals it makes
sense for us to also uh you know pay
some attention to how is speech
production uh accomplished by
monolingual speakers as well because
that would give us a baseline in some
sense to understand how would speech
production actually take place in case
of by or multilinguals okay so uh as I
already said majority of human
communication occurs via speech it
includes a range of steps from
conceptualizing an idea that a speaker
wishes to convey uh and to basically a
set of behaviors that would change the
physical environment so for example uh
if I uh give you a particular topic to
speak or if I intend you know stand here
and intend to speak about bilingualism
uh the first part of the process is for
me to conceptualize what do I want to
talk about the second would be
formulating how do I want to talk about
it and the final would be articulation
where I will create you know some kind
of behavior which is basically moving my
vocal apparatus uh creating sound that
carries meaning which will be understood
by my listeners so it is this tripartite
process of conceptualization formulation
and articulation that basically explains
the process of speech production and
this would be true for monolinguals as
well as bilinguals and multilinguals as
well let's take an example you know for
example a compliment conversation how
does this really start say for example
you could you could you meet somebody
you like let's say their shirt and you
go on and you tell them oh I like your
shirt this is a very good color and the
person would listen it and would react
appropriately saying oh thanks uh and
then you can you know take the
conversation forward by saying oh where
did you get it oh I got it from uh this
particular Market in this city and so on
and so forth so this is just an example
of a typical conversation we have all
kinds of conversations if for example uh
a lot of times you will see that
conversations typically revolve around
generic formulas for example if you want
to have a weather conversation it's
typically you know a decent conversation
starter uh when you're meeting somebody
for the first time so you know typically
people would be oh today is really hot
or cold or rainy or anything uh outside
and the person would say yes yes I am
also feeling cold or I'm also feeling
rather warm and then you could sort of
say okay I hope the weather becomes
warmer or colder uh as it goes ahead and
so on and so forth so typically if you
see this process of having a
conversation based basically has these
multiple steps where you are talking
about a particular topic you are
conceptualizing how you will put
yourself across and then you formulate
the right words the right sentences the
right form of how you want to convey the
message and going forward you can
basically uh you know uh take the
message forward to other topics and so
on and so forth so there are you know it
could be a lunchtime conversation as
well there could be any number of
conversations that we have in the same
sense but if you look at these
conversations uh closely you could ask
yourself certain questions for example
how do we retrieve these linguistic
representations say for example when we
want to talk about let's say somebody's
shirt so I know what a shirt is I have
the conceptual representation of a shirt
or let us say if I'm going to talk about
weather I know what weather is what I am
talking about I'm talking about the
general climatic condition of that place
in that moment so how do we sort of from
these thoughts how do I pick up
linguistic representations uh that you
know uh which are best suited to convey
this idea and then how do we organize
these representations we don't speak uh
you know randomly we don't speak just a
constellation of Words which are not
organized we typically speak in a
particular order we typically speak when
we speak our words are grammatically
correct say for example I will not say
oh the weather how it is oh the weather
how is it we will basically oh how the
weather is today something like that so
we we also organize these
representations in a particular order in
sense of grammar as well as semantics
for example uh if you want to talk about
a particular idea you will basically uh
uh you know uh organize your uh you know
ideas in a particular sense that it
seems coherent and it seems like a
logical flow of uh you know
conversation so uh finally the question
that one can ask is that how are these
representations once you've organized
them retrieved them how are they
translated into a form that the motor
system can use to generate the physical
actual gestures which are utilized to
articulate them see uh the first two
parts conceptualization and formulation
are still abstractions of the process of
speaking but the final part which is
articulation is actually a physical
activity because you need to move your
tongue you need to move your mouth you
need to uh stop the flow of air in your
vocal apparatus in order to speak in
order to create uh Speech so
these are the three very important
questions that we can ponder about
speech production uh in this lecture I'm
going to talk mainly about monolingual
speech production but I would encourage
you to you know take a pause at every
moment and say oh how would this happen
when a person has two languages and
we'll basically see how we sort of go
ahead with this so as I said uh the
three the tripartite process of speech
production has three parts
conceptualization formulation and
articulation very basically
conceptualization means thinking of
something to say what do I want to talk
about formulation how do I figure how do
I you know Express this idea in the best
possible way and articulation is the
final part of moving the muscles to
produce the sounds that can be perceived
by The
Listener a very interesting Model A very
interesting theory on this account was
presented by William levelt uh in a
model which he named as the Weaver Plus+
model and we are going to sort of go
through uh very briefly with some of the
stages of this model you can see that at
the top here is the conceptual
preparation in terms of lexical Concepts
what is a lexical concept a lexical
concept is basically concept which can
be expressed in the words of our
language in case of bilinguals it could
be uh you know uh it could be a concept
that could be expressed in two languages
or three languages uh you know the
whatever the number of languages we know
uh typically you will come across more
than one lexical concept at any point in
time so you need to do lexical selection
you'll need uh you'll reach a level
after that which is called the LMA level
which which has both semantics and
syntax information uh you go to
morphological encoding where you are
basically wondering about the type of
word that is going to be used you create
a morim then you go to the phological
encoding part where you are sort of uh
you finalize the morim that you have to
uh communicate but you have to uh
assemble sounds for it create the
phological word go to uh how the word
has to be uh you know actually produced
in phonetic encoding and then is the
gestal score which leads us to sort of
you know produce speech Let's go in a
little bit of detail of these models
because at this point in the slide I've
just sort of given you an overview but
we can obviously visit it in a bit more
detail and try and appreciate how this
would change or what will be the special
uh uh tweaks in this model that we'll
need to do if we want to explain
bilingual or multilingual word
production in this so uh as we've seen
spee production can be viewed as a
process involving a sequence of mental
uh uh steps uh each mental process
accomplishes a smaller goal a subgoal
and the output of one mental process
works as the input of the next one now
this is something which is peculiar to
this model it basically for follows what
is called a feed forward kind of a
process so uh let's say we start with
something like okay uh you know
conceptual preparation which is at the
top here and con during conceptual
preparation we are choosing about the
ideas that we want to speak about we're
choosing the idea that we wanting to
speak about oh do I want to have a lunch
time conversation do I want to talk
about weather do I want to talk about
the nice dress that a person is wearing
what is it that I want to really talk
about the output of this process once
you have a sense of what you really want
to talk about is a lexical concept and a
lexical concept is an basically an idea
for which your language has a label so
the idea is say for example if you want
to talk about uh uh you know uh let us
say uh beauty or you want to talk about
a particular person's shirt or weather
uh your language should have words to be
able to express that idea to take a more
concrete example uh for if you want to
talk about a female horse there is a
word in a in English language called May
and you can if you're talk if you're
seeing and talking about a female horse
you can use the word May and it
basically encompasses all the meaning
that you want to convey in the word m
however if one were to talk about a
female elephant there's not a single
word that uh is convert is you know that
encapsulates that idea so you'll be
forced to use both the words female and
elephant and that will be your lexical
concept so this is basically how what we
are doing is we are converting a
preverbal uh uh message into a actual
verbal uh Thing by mapping the thoughts
onto these words which are these lexical
Concepts now one could ask that can all
ideas be very neatly expressed with
individual words probably not because a
lot of times you'll see people telling
you that oh I am so happy that I cannot
express it in words or I cannot describe
in words what I am feeling at this point
in time so basically what is happening
is this process of mapping the uh
thoughts to these lexical Concepts is
referred to as the process called
lexicalization and lexicalization
basically serves as an intermediary
between non-linguistic thought processes
and the linguistic uh items or lexical
uh items that can produce veral
Expressions to convey these thoughts so
this is precisely what we trying to do
now a lot of times it would happen that
the number of flexal Concepts that you
have for a given idea may be more than
one so for example if you want to talk
about let us say a particular emotion
let's say you want to talk about being
happy so you can be talk you can talk
about oh I am very happy I am very
elated I am Overjoyed uh I am uh uh you
know in a blissful State you can
basically come up with any number of
words uh or a large number of words to
sometimes Express a single idea now what
would happen in that kind of a scenario
in that kind of a scenario what would
happen is that you will need to select
you'll need to perform lexical selection
you'll need to select the appropriate
word for that point in time so do you do
you think that Overjoyed is the best
expression to convey what you're feeling
at the moment or just happy is the best
one or elated is one so you select from
any number of words that you you are
made available with you can basically
select okay this is the word that I want
to go with and this at this process the
output that comes out is called the
Lemma now a Lemma is an abstract mental
representation which reflects a slightly
intermediate stage between activating an
idea and activating the exact speech
sound so it is something which is very
similar to a word but it is an
abstraction it is not an exact word
itself it is an abstraction of that word
you can say it's a root idea for a given
word which carries with itself both simp
semantic information what is it that I
want to convey within the word that I am
choosing as well as syntactic
information how do I want that thing to
uh you know play out say for example if
I'm talking about uh let's say a
particular ad activity uh let's say play
uh am I talking about play am I talking
about playfulness am I talking about
playing what is it that I'm talking
about if I'm talking about eat uh am I
talking about he ate a sandwich a day
ago uh he is going to eat the sandwich
now he's eating the sandwich now so you
can see the root form of this word is
the word eight which is the Lemma but it
can carry information both about the
meaning that oh I want to convey
something about the act of eating but is
also carrying the information about the
syntax that is what is the situation
about eating that I want to convey so
this is what a LMA is so therefore we've
said that LMA incorporates information
both about the meaning and the syntax of
a given idea now once you have an
activated set of lemas uh then you move
slightly further you activate the
process or you initiate the process of
activating the sound codes uh that we
sort of Need For Speech to begin so then
what we are doing is then we are sort of
finalizing upon that exact type or that
exact variation of a given word that we
want to go ahead with and we move to
what is called morphological encoding
now before I talk to you about
morphological encoding I'm sure you
would know that morphologic morphemes
are you know the basic units of language
representation basically in any language
uh they are the basic units in which you
know the language can be uh uh cut down
into or reduced to and morphological uh
and morphemes in in that sense are
basically two kinds of unit which carry
meaning first is lexical morphemes which
can stand alone say for example uh any
number of words you can say play eat
sleep uh uh you know run cry etc etc
these are all free morphs because they
can stand alone on themselves but if I
change them say for example if I do uh a
playing then the ing part is called a
grammatical morphine because per say by
itself it does not have a meaning but
once it is attached to the morphine play
it modifies the word plays meaning so it
this one in e r ESS Etc are basically
called grammatical model mumes now in
morphological encoding basically what
one needs to do is one needs to sort of
finalize that okay this is the word that
I'm going to go with this is that
specific form of the word so am I going
with eat or am I going with eating or am
I going with ate so this exact form of
word is basically specified at the level
of morphological
encoding now uh level basically says
that each for each word that we need to
speak we would basically need to specify
uh the morphological form and hence we
would go to the step called
morphological specification uh telling
us how the word is going to behave when
we place it in a larger utterance so for
example as I was saying the
morphological specification for the word
eat includes that it is a root form and
its past tense is eight or its
continuous tense is eating so we sort of
figured out that okay which form of the
word eat am I going to use and then we
can move on further now remember here
here whichever form of the word that you
finalize you will really need to start
assembling sounds of that type so if you
finalized eight then the sounds that you
start assembling will be different if
you finalized eating then the sounds
that you going to assemble will be
slightly
different so having selected a set of
morphemes to produce morphological
encoding starts activating the speech
sounds uh we need to plan the auditory
movements that will eventually create
the speed signal the speech sound one
produces depends on the morphemes that
one has activated and also they have to
be organized in the right sequence now
again uh when we speak we are not
speaking phones in a jumbled order while
we are speaking phones or our words are
obviously concatenation of phones but
they we don't jumble them up they have
to be executed in the same order for
example if we decided that we want to
speak the word eating we want to talk
about the act of eating then what we
will do is we have to not only activate
the sounds for the word eat and for the
uh part Ing we have to also ensure that
it comes out in the right order we
cannot say in eat we have to always
ensure that whenever we speak it comes
out in that same order eat in this is
basically also very very important so
the right sounds need to be activated
and they need to be activated in the
right
order so just summing up what have we
done so far we started with
conceptualization we eventually selected
dmas we went to morphological
specification then we figured out which
morphs we going to go with and then we
had information about okay which sounds
we need to activate and in what order so
that we can go further with the process
of speech production once you have
everything with you once you have the
morphemes slotted into the right
positions once you have activated their
individual speech sounds then what you
are ending up with is that lexical entry
uh called a lexine it is basically uh
how a particular word exists in your
mental lexan which is your mental
dictionary and for this we need to
activate now start activating now the
basic sounds that will create this leim
and will need to be organized in a
particular manner for
production okay so one of the
organizational schemes that is preferred
by human speakers is called salification
we don't speak in individual sounds
rather we speak in syllables let us let
us look at a example of that now
salification has two two parts it
basically requires us activating a
metrical structure say for example if
I'm talking about the word Pat it is a
single syllable or if I'm talking about
the word patter uh it it is two
syllables so you say Pat and then
there's a break and then you say ER so
basically we need to understand the
metrical structure where the stress
would be say for example uh the word
banana has the stress on the second Sy B
na na and the word Panama has the stress
on the first level Panama so basically
what we are doing is we are organizing
things according to their uh according
to their syllable structure and also at
that same point emphasizing which parts
of the syllable will be stressed which
parts will not be stressed and this
information is very very important
because it tells us how exactly we have
to uh produce these things say for
example banana uh will basically involve
stress on the second syllable and we'll
speak the second syllable slightly
louder than we have spoken the first or
the third syllables similarly in Panama
we are going to be slightly louder on
the first syllable as opposed to the
second and the third syllable now this
is basically while it seems you know as
I was saying in the beginning while it
seems that speech production is a very
easy job it basically requires us paying
attention to these very very minute
details uh and uh you know uh managing
and coordinating between these different
levels of representations uh so that
speech production uh happens
seamlessly and it also happens uh you
know rather rapidly uh typically if you
see people don't make a lot of speech
adds we we speak rather fluently we
speak uh not only with the content that
we want to speak but we also have these
intonations these uh uh you know
different emphasis points Etc so that we
can also communicate uh part of our
emotions or part of our the emphasis
nonverbally as well so in that sense you
can see that that all of these
activities are rather sophisticated and
require a lot of attention uh again as I
said uh banana the stress is on the
second syllable you can see the stress
marker is on the second syllable here in
Panama the stress is on the first
syllable and you can say the stress
marker is on the first syllable now uh
people would wonder that oh is
salification an actual process do we
really go through salification and for
that you could look at how we speak you
can just sample any amount of speech
that somebody has spoken and you could
try and see whether that speech segment
is organized according to actual word
forms morphology or basically according
to a stress scheme for example if uh you
are talking about the word escorting uh
escorting typically has two morphemes it
has a free morphine escort it has a
bound morphine uh Ing and typically what
we we should expect is that if somebody
speaking escorting they should basically
do escort and in but if you record
different number of people speaking the
word you will see that when we are
speaking or just observe yourself
speaking the word esorting you will see
that at S I took a pause C I took a
pause and ing so what we are basically
doing is I'm speaking s cor and Ting
which is basically the syllabic
structure in which this word is
constructed so speech production you can
see here is not basically following
morphemic boundaries but rather it is
following syllabic boundaries and this
is an important property of our speech
production system because this is how or
this is how smoothly the vocal apparatus
actually uh moves around to create these
words so uh as should be clear with this
example we do not simply activate
morphemes we activate the PHS that go
with each morim and we produce them in
sequence moreover after these morphemes
are activated we calculate the best way
to organize the sequence of these phes
into syllables and it is the syllables
that actually form the basis of of how
we are going to produce them along with
stress information and so on uh so yeah
while we do not while we need more FES
and words to plan what to speak uh
Speech does not simply involve
activating the speech sounds in
individual words it involves a much more
detailed speech planning and it
basically involves activating uh
morphemes words it figures out the
stress uh uh marking and so on and it in
that sense becomes uh a slightly
sophisticated slightly coordinated uh
activity that eventually uh ends up in
production of speech now once you've
done salification uh what we are ending
up with is called a phological word how
are we going to speak something so for
for example if we are going to speak
escorting we have three parts we have
escorting this is the set of phological
words and this is how we're going to uh
execute this according to the v++ model
of William level uh one can begin to
speak as soon as you have activated all
the syllables of a given phological word
further we can plan each utterance by
activating a number of lemas and
morphemes simultaneously it's a fast
process you can see and basically the
actual speech uh movements are Bas are
planned one phological word at a time so
escort escort and tus or escort anding
and these are basically you know you can
say mini executable programs that that
are executed one after the other in a
you know specific sequence so that
speech production seems very very uh you
know seamless so uh another part that
could me that we could sort of uh you
know talk a little bit about is that uh
how are uh how do we ensure that people
are actually executing this process in
an ordered manner uh and uh this can be
seen and this has been demonstrated
through a bunch of experiments some of
them done by wi and leveled where if you
ask people to monitor how they are
speaking and if you basically ask them
to you know let's say press a button if
a you know particular target phon comes
you will see that they'll press faster
if the target phon comes from the uh
beginning of the word than if it comes
from the middle or the end of the word
okay so let's just summarize how this
weer Plus+ model or how speech
production in monolinguals work or
generally how speech production works we
begin with a set of ideas uh that the V
speaker wishes to express which is the
conceptualization phase in the next
phase we basically tied the overall idea
to lexical Concepts basically Words
which are capable of expressing our
ideas because then because each language
will have its own specific words that it
will use to convey a particular idea uh
but also may require combination of
words for example we have compound words
we have phrases and so on and so forth
now after we have finalized on the set
of flexal Concepts after this have been
activated we'll activate the lemas that
correspond to these specific lexical
Concepts uh activating lemas would
provide us uh about information about
the morphological properties of the
words including information about how
words can be combined after a set of
morphes has been activated and organized
into a sequence the speech sounds uh
that are required can be activated and
again placed in the syllable sequence as
we discussed phological encoding
involves the activation of metrical
structures and syllabification so that
we know how each syllable is to be
produced in what order and with what uh
emphasis the outcome of this is a pH
word consisting of syllable size
frames finally we come closer to
articulation so what we doing here is
during ftic encoding the speech
production system basically consults a
set of stored representations for
specific syllables so it is basically
again as I said mini executable program
so how do I say s how do I say Pat how
do I say B and basically these programs
are concatenated they are sort of uh
brought together and what the system
does is it activates these appropriate
syllable representations and and places
them in the correct order so that they
can be executed very quickly one after
the other as the demands of the speech
production system are this this final
representation is called a gestural
score which basically creates executable
commands of how these syllables are need
are going to be articulated so that is
pretty much a generic understanding or a
generic description of how speed
production really happens you can uh uh
see that in this model we started at the
top with conceptual preparation we had
we tied those ideas to lexical Concepts
we did some selection selected lemas did
some morphological encoding of which
form of the word we are going to use uh
then we spec fixed out which morphemes
we are going to finally select when went
to phological encoding deciding
syllables creating a phological word and
then we did pH phonetic encoding
phonetic gestural scode which is an
executing command and then articulation
happens which basically ends up in a
sound wave which is a physical stimulus
that is received by the ears of the
listener so that comprehension can take
place so that's all that I wanted to
talk about a generic give you a generic
understanding of speech production in
monolinguals and in monolinguals and in
the next lectures we will look at some
of the processing assumptions of this
model and how these processing
assumptions mean need to be sort of uh
tweaked a little bit in order to account
for bilingual or multilingual speech
production as
well thank
[Music]
you
[Music]
[Music]
[Music]
