YT YouTube2Text

YouTube2Text → Transcripts → Lec 3: Science of Language - I

Lec 3: Science of Language - I — Transcript

by NPTEL IIT Guwahati · 11,178 words · 744 segments · language en · Watch on YouTube

Open in transcriber Summarize with AI Download SRT/VTT

Full transcript

0:31hello friends welcome back to discourse on the psychology of language this is lecture
0:37number three and what we are going to do in this lecture is we will look at how do we
0:44do research in language sciences or in psycholinguistics now when i say how do we do research in psycholinguistics
0:53the methods that i am going to tell you here is not explicitly the one that we use in language
0:58studies now these methods are very well commonly used across a number of other scientific disciplines
1:06so some of the principles some of the facts that i am going to reveal here is common to
1:11most of the behavioural sciences or social sciences before we start our journey today
1:17and start looking at how do we do research in the science of language let us do a quick
1:23recap of what we did in the first two lectures lecture number one and lecture number two
1:28now lecture number one and lecture number two we focused on how language evolved
1:35and even before that we started looking at the nature of language what is it so the best
1:43way to define language is basically a medium of communication between people a medium of
1:49expression of ideas between two entities so we started off the beginning of this lecture
1:56series by defining what is the definition of language we started off by pinpointing
2:03how communication and language are different and we started seeing or we start our lectures
2:10by looking at the most primitive form of language known to human beings which is animal communication
2:16so we started looking at mu animal communication systems we started looking at why animals
2:22communicate and what are the communication systems existed in animals from there we devised
2:28or underline certain rules on which animal communication systems really work we looked
2:35at reasons like food finding mates or auh faupointing out enemies and these are the
2:42reasons why animal use a communication system
2:46we then moved on to the development of human language how human language develop from animal
2:53communication systems so there we looked example of laugh as a language so how laughing or
3:05smiling is a model of a language so we took that as a model and we describe what it is
3:11and how laugh never means that something funny is been spoken but laugh has multiple meanings
3:17and that is what we try to or try to explain through how language human language is different
3:24from animal communication
3:26animal communication pinpoints here and now things are very basic ideas whereas if you
3:31look at the laugh that we have in humans is not actually pointing out the fact that something
3:37funny is being said it can sometimes mean i like you it can sometimes mean that i support
3:42your views or your ideas or so many other things so the simple laugh that we have a
3:47smile that we have can mean so many things and so we took this model system and explain
3:53to you how animal languages oranimal communications differ from human languages
4:01then we looked at various features of the human language system and how do they vary
4:07from animal communication system for example the nature of sentence structure the nature
4:13of ourarbitrary symbols and the nature of production so this is how because animals
4:19can communicate on limited ideas and theris their way of communication is not structured
4:24in new ideas and not governed by certain rules
4:26so we pointed out that and we also saw that how arbitrary symbols which is words in different
4:33languages mean the same concept or mean the same idea that we are expressing then we also
4:41looked at how human languages progress and there we pointed out at something called duality
4:47of patterning we saw how basic units add up to form longer languages so basically how
4:55basic units combine together to give bigger units in languages
4:59we then looked at the standard paramedical sideof the pyramidal structure of language
5:05in terms of the phenomes the morphemesah the words the sentence and the discourse the syntax
5:12semantics and the discourse so we looked at how this is built about we pointed out or
5:18look briefly into the fact of how language evolved the evolution of language how did
5:26it start so we looked at the proto humans the neanderthal man the homo sapiens and how
5:33they developed a proto language
5:35and how this proto-language developed into the fully formed language that we think about
5:40and we also gave enough evidences to point out that this kind of revolution would have
5:45beenpossible the very existence of pidgin which is a language which has limited scope
5:52but then can express ideas among two communities which are not common in language gives a strong
5:58support to the fact that language would have evolved from the proto-language
6:02we looked at the process of recursion which is verybasic to the idea of language how sentences
6:10or longer sentencesform so the same set of rules or the same set ofstructure is repeated
6:17to form longer sentences so we looked at that as well then we looked at how syntaxes are
6:22formed and we looked at certainevidences of both the continuity and discontinuity of theory
6:28of language
6:29continuity theory of language believes that language developed as a gradual process through
6:33several stages from animalsto humans and to higher humans discontinuity theory believes
6:40that out of a single mutation language eka evolved very rapidly and so we saw how these
6:46theories are what they prompt and what kind of evidences they are based on and then we
6:51looked at certain kind ofahreasoning or evidences for the fact that language evolvedah in a
6:59sequential manner through afossilised structures
7:03so basically how they evolved or how theyah aaaa developed in in ain a chain sequenceright
7:12from theahthe basic communication that the proto-humans were using to the modern language
7:20so that is what we did in the first two lectures we looked at the evolution we looked at how
7:25they are differentfrom the present language system is different from animal language systems
7:30and we providedevidences look at some primary theories of language development in that kind
7:33of thing
7:34and having said that the first thing that we need to know in anycourse is to know what
7:43are thetools and what are the techniques which are available to conduct research in that
7:49particular area or in that particular field and that is what we are absolutely now going
7:55to do in this part of the course we are going to see the tools and techniques which are
8:00available to us which help us in doing research in language
8:06so i will start with a brief story and i will show to you how language research is done
8:14and then we look at the procedures and techniques which are available for doing research now
8:21in this story we first start off by describing what is an erp an event related potential
8:30and so these erp’s or event related potential are changes in the brain electrical potential
8:49due to changes in perception of certain stimuli
9:07now we will take the story of how the n400 which is a negative going peak how this was
9:24developed or this was found out by someone called marta kutas in the lab laboratory of
9:33hillyard so both are prominent psycholinguist and n400 wasa waveform which was developed
9:43by marta kutas and before the development of n400 the p300 was the signature of language
9:49comprehension or language understanding
9:52now let us look at a little bit of detail of what the erp is and why it is needed that
9:57kind of a thing so the brain at any point of time is doing a lot of activities if you
10:03want to know how the brain is processing certain stimuli we have to study or we have to look
10:12at the electrical changes which is happening in the brain while a stimuli is being presented
10:18or after the presentation of a particular stimuli and compare it with a state when no
10:24stimuli is presented
10:26and then what we do is with cut of those section of the eeg which is of interest to us so normally
10:35if this is what my eeg is a normal eeg looks like and if i present my stimulus here and
10:43in other case i do not present a stimulus what i do is i take let us say if my area
10:49of interest is one thousand milliseconds and i believe that within a thousand millisecond
10:52some changes will happen what i will do is i will take this eeg or this kind of eeg by
11:00presenting the stimulus to let us say one thousand people
11:04and then again presenting no stimulus to one thousand people i will average them and from
11:10this average i will be ba able to separate the basic waveforms in the brain at the default
11:16stage and cut-off those or find out those changes in electrical potential which are
11:24coming from the presentation of the stimulus how do we do that we compare no stimulus versus
11:28stimulus condition
11:29and so we will be able to know how does brain or what areas of the brain or what regions
11:36of the brain what processes in the brain are active while doing a particularahprocessing
11:42or while doing a particular cognitive task and so what what was happening is the most
11:49in in sentence comprehension now sentence compression is something that wuil we will
11:52see in in the future coming classes so basically how does the brain read a sentence
11:59how does the brain extract meaning from a sentence now what was happening is that it
12:05was well known before marta kutas came in and gave and and found out that there is something
12:11called n400 before that a lot of research was done and those in eeg and it was found
12:16out that there is something called a p300 wave a p300 wave is a positive going wave
12:22three hundred milliseconds so this is where the stimulus is presented three hundred milliseconds
12:27after the presentation of the stimulus what i see is a deviation on the positive direction
12:32of course eeg’s are read in the wrong side so this is my positive and this is my negative
12:36and so this deviation after three hundred milliseconds basically suggests since ssh
12:39this is the average deviation that i get and so this basically explains so this basicallytells
12:45me about one common feature of shthe stimuli of interest the stimuli of interest was how
12:53the brain processes those sentences in which something unexpected happens
13:00and so the thequestion of interest here was processing a simple statement the statement
13:08that that was processed is she put on her high heel as we will note that till here it
13:31is in small letters and then suddenly the shoe comes in capital letters and so this
13:37is what the deviation is and for this kind of deviations or oddball as we call them the
13:43brain reveals a p300 umdeflection this p300 deflection says that when um when the brain
13:52comes to this part it is able to notice the difference between capitals and small
13:58so this is intha in terms of the sentence structure syntactic structures the sentence
14:03structure but what happens when the meaning is wrong in a sentence for example let us
14:10look at this so she spreads the warm bread with socks now as you can see the moment you
14:33read the sentence you come to know that this is wrong they should not be there and this
14:40is called the closureah effect and sowhat it really means is that people are able to
14:46predict what kind of words will come next
14:48we we will look at these in in the coming lectures so people have this ability of predicting
14:53what words come next huhumm (coughing) and so as soon as a sentence like this is given
15:00the brain is able to notice this difference now this difference isin as as you can see
15:05the difference here is in terms of sasentence structure so it is capitals and it is small
15:10but here thaall are in capital the problem is in terms of meaning
15:14and so when this happens marta kutas found out that she is not getting a p300 rather
15:20what she was getting here is that right from the start of the stimulus four hundred sh
15:27millisecond past the stimulus she was getting a if this was my eeg an average eeg across
15:34multiple subjects this is what she was getting a deflection in the negative so it is called
15:40n400 a negative deflection four hundred milliseconds past the presentation of the stimulus or this
15:47kind of words
15:49so basically then if structure-wise differences are there the brain gives a positive p300
15:55deflection but if a meaning-wise ahm meaning relatedchanges therea meaning related problem
16:02occurs in a sentence a n400 is resulted or the deflection is in the four hundred negative
16:08side now what does this really mean or why does is of interest it is of interest because
16:14the brain processes syntactic structures and semantic forms in different ways
16:23this was the first time that we are able to see that brain processes the meaning and structures
16:28of sentences in different ways this is one kind of research that we do in language and
16:36so this is an erp and easy research that we do inlanguage and so from there on these two
16:42basic the p300 p300 which is called the oddball component of the eeg and the n400 which is
16:54the semantic component marking the semantic component are the two primary components which
17:02have been widely used insentenceproduction sentenceah comprehension
17:09now there is something called n100 also but this is an intentionalcomponent so we are
17:14not bringing it here so why all these things which was to show you what kind of research
17:20will be huhumm (coughing) possible in language and what kind of things are possible in language
17:27now let us look at how is basically research done in language and as i said this principle
17:33that i am going to tell you here is not common to just language studies or psycholinguistics
17:40it can be used with any behavioural or social sciences
17:45so the first thing before we do research in in anybehavioural science or in language for
17:50specifically this matter we need to explain or we need to understand what a theory is
17:56before finding up a problem or bif before doing research in language we have to first
18:00look at a theory because a gap in the theory will provide you a problem of interest on
18:05which you want to do a research so what is theory then what is the definition of a theory
18:10a theory is believed to provide a conceptual framework for explaining a set of observations
18:17so it is a conceptual framework which explains a set of observations and that is that is
18:21what we need to have here basically the theory its research evidence or its kind of a proposal
18:29a background knowledge which gives the reason or which provides the reason or evidence for
18:37any kind of observations that we have and so the observation that we saw right now the
18:42n400 and the p300 there are lots of theories which says that there are different regions
18:46of the brain that process it and there are different time frames
18:49now the p400the n400 and p300 are two things the p300 is more of an intentional component
18:57and so meaning is not derived and sothis happens very early on but the n400 huhumm (coughing)
19:06requires the brain to exerts the meaning of the word the concept of the word and for that
19:12theah abstract concept are the dilemma of the word and for that particular reason what
19:16it has to do is this kind of activities happen late in the processing
19:21it happens at the temporal lobe or the medial temporal lobe and so it will happen late and
19:25also there will be a positive deflection and the negative deflection so basically the theory
19:29gives you the reason so the medial temporal lobe is the secondary structure or where the
19:34memories are there or where semantic memory is stored and so for this kind of variations
19:38or this kind ofah verification in terms of meaning is only semantic memory
19:43whereas looking at just the capitals and small is more or less of an attentional feature
19:49and so that is why it happened early on and so this is the kind of theory that we can
19:53give for the kind ofresearch that has beendisplayed or that i have explained to you before starting
20:01this lecture so theory gives the reason or theor the evidence or the baseline on which
20:09research is done now it should make predictions about future
20:13now what the theory does is not only explanat explain a set of observations it also predicts
20:20the future observations so basically if a set of observation if an experiment is done
20:25a theory provides it enough evidence of why this is coming thathe theory does not oh only
20:31do that it also gives you enough reason or it also gives youthe line of approach to do
20:38newer set of experiments topredict what future observations can come or whatmake predictions
20:45about if you vary something in the in the initial experiment what kind of results are
20:49you going to get
20:50and these experiments that could be tested in experiment so this is what a theory does
20:55the two main purpose of theory explaining observations and explaining future observations
21:00and how this can be explained by experiments now after the theory we come to know about
21:06or we come at the level of what a hypothesis is so what is a hypothesis hypothesis generally
21:13is a tentative solution to a problem or a prediction which is based on a theory about
21:18a solution to a problem
21:19so let us say i have a huhumm (coughing) problem the problem is how the brain interprets meaning
21:25and syntact ends and syntactic or rules at at the meaning level and at the surface level
21:31of sentence processing if i have a theory that there are two different ways in which
21:34the brain handle that hypothesis or predictions which basicallythe solutions to the fact that
21:42how the brain is actually doing it or what are thepossible solutions to this problem
21:47so prediction that is derived from a theory is basically known as a hypothesis so basically
21:52what a hypothesis is it is a prediction which is derived from a theory and so these are
21:56tentative solutions tentative solutions in the sense that this may not be the correct
22:00solution but this is a possible solution once a hypothesis is laid once a solution is laid
22:06what the researcher then does is collects data based on the prediction of the theory
22:20and then verifies the hypothesis or test the hypothesis to test the theory
22:30so whether the theory is correct or not how it is done we collect data in support of the
22:36theory or ahs tha if the particular theory suggests something we collect some data whiche
22:41which the theory says or thawhich the theory predicts that this is what tha the particular
22:46experiment is going to say so we collect data in its response test the data and test the
22:50hypothesis if the hypothesis isverified or if the hypothesis isisis is supported by the
22:58data then we say the theory is correct otherwise we bound to reform the theory or we re-formulate
23:04the theory
23:05another interesting thing in terms of thea in terms of research or scientific research
23:11is falsification criteria now what is the falsification criteria now the hypothesis
23:17which is derived from a theory should have the possibility or it should have the possibility
23:23the probability of being rejected hypothesis are as i said these are tentative solutions
23:30so any hypothesis which has been said or which have been proposed theythey ahl should ahl
23:36always have a chance to be rejected only then it is a scientific research
23:40so theory must make predictions that can be disconfirmed by data if we make hypothesis
23:45whichwhich is always confirmed by data then it is not a right way of doing research so
23:52basically any hypothesis that we predict should always have the possibility of being disconfirmed
23:58and this patis basically is called the falsification or the falsifiability criteria of research
24:03so any hypothesis that we make thes eisa this is a viable solution but this is a tentative
24:08solution which basically means that the chances are that we can always reject this particular
24:14hypothesis and then reconfirm the theory or predict more hypothesis out of it and that
24:19is whytha uh when we do research we never have one hypothesis we have multiple hypothesis
24:24to test because there are multiple solutions right so we do not have just one hypothesis
24:28to go about and say that this is correct
24:29can never prove a theory true but can prove it false hypothesis gives the possibility
24:36or a probability that a particular theory is correct but it never says that it is hundred
24:42percent correct we never have the probability one but weall we can have a probability of
24:48zero we can always say that the theory is wrong in terms of the hypothesis but the hypothesis
24:52can never with full evidence or conformity say that the theory is or is truly correct
24:58also theories that cannot be falsified cannot be considered as scientific because then they
25:03become facts if you have theories like the sun will rise in the east if that is what
25:08it is and you want to test this theory now this is going to happen till the earth rotates
25:12and so in this kind of things there is no point in making hypothesis because then what
25:16happens is this is not a scientific theory to look at and so this is a fact that we know
25:20and it become fact and so that is whyour theory should be abe should be able to be falsified
25:28two other things that we do in research or we use in research is the process of induction
25:33and deduction now what is induction specification or an generalisation so if you have a large
25:48set of data and from there if we conclude specific results this is called deduction
25:55but if you have very specific results and from that we forecast or we come up with varied
26:03kind of hypothesis or varied kind ofah results out of it then that is called induction
26:09so induction is from specific to general the process of generalisation is called induction
26:14and the process of specification isthe is the prothe process of reduction so if you
26:21are coming from specific a large data set from that if we pinpoint or if we are able
26:27to expact extract specific hypotheses or specific results this is call deduction but but from
26:34very specific results ifee if we inflated if we are able toif we are trying to predict
26:40a number of solutions out of very specific results generalise it in a manner then it
26:44is called the idea of induction
26:46so ina what is induction specific examples to general statement so if there is a specific
26:51example let us say that any spahah solike i saw a parrot eat a chilly and based on that
26:58if i say that all birds eat chillies then this is basically an induction because what
27:02i am trying to do is for the bird category i am trying to say that most birds eat chilly
27:06and this is what i am doing is i am generalising it so observation to theories
27:09if you are from one particular observation if you build a theory this is called the process
27:13of induction in deduction a general statement to specific examples and so ifi if i say that
27:19most birds are able to fly and from there if we reduce the statement that ostrich can
27:24fly whereas they cannot then what i am doing is i am mistaking and i am using the process
27:30of deduction
27:31so since most birds fly and if that is a characteristic feature of a bird from there you are defining
27:36feature of a bird from there if i am trying to say that ostriches can fly which is not
27:41right or kemu can flyemu can fly which is also a bird then what i am doing is i am doing
27:46a deduction in induction what happens is we form theory to hypothesis so if we are coming
27:51from theory to hypothesis we are doing deduction but if we are taking observations and from
27:56there we are building a theory then we tend to do something called inductions
28:02so scientist they make observations to detect patterns from which they propose a theory
28:08then they generate the thypothesis which then prh which are predictions about future observations
28:14so first they collect som observe something and from that observation they give a theory
28:19and then they generate hypothesis based on the theories and predict future observations
28:24these new observations call experiments test hypothesis and provide evidences for or against
28:30the theory
28:31so based on certain kind of data i give certain kind of theory and then based on the predictions
28:36of a theory i run experiments and then validate the theory this is how a research is exactly
28:40done the process from patterns to theory involves induction while the process from theory to
28:46hypothesis involves deduction as you can see this is my theory this is my observations
28:50so if i start with observation look at patterns will the theoryend to hypothesis and experiments
28:55and then again come to observation is what i am doing is deduction if i start this wayuh
28:59i do induction if i start the other way around from theory if i go to hypothesis then observations
29:05and patterns again back to theory what i am doing is actually deduction are so these are
29:09the two basic forms or two basic types of doing research
29:14so we will take two experiments and i will explain further on how experiments are designed
29:22in language and in any behavioural sciences we look at the tha basic of experimentation
29:29in behavioural sciences and we will for that purpose we look at the investigation of short-term
29:34memory capacity two classic examples from the field of cognitive psychology or memory
29:42and will take these as our model systems and explain to you how research is done are experiments
29:48are planned
29:49so the two experiments that we are taking is first stm capacity short-term memory capacity
29:54as number of items miller nineteen fifty six so miller gave something called the magical
29:58number which is seven plus or minus two these are the number of chunks of item which is
30:03stored in short-term memories so this is the most number of items that you can shooah store
30:09in your short-term memory or working memory that is what the definition was not working
30:13memory but short-term memory
30:14so this is what they were testing how many items can be stored in short-term memory the
30:18other that we are doing the other experiment that we are tay trying tolook at is alan baddeley's
30:23experiment which says that what is the length of time for which items can get stored in
30:28short-term memory and remember this is a very classy experiment and based huhumm (coughing)
30:33on this experiment the whole idea of working memory was developed
30:36so let us look at these model experiments the hypothesis was people can hold seven items
30:42init should be seven plus or minus two but let us take go with seven seven items in stm
30:48this is what the hypothesis was the test was in a digit span task in a digit span task
30:53what really happens is number of digit appears to you one by one and you will have to commit
30:59these digits to memory and later on reveal or retrieve those digits so that is what a
31:03digit span ta task is
31:05in front of you you will have a computer and so number of digits will come so one digit
31:08a two digit a three digit a four-digit sequence and so as the number of digits’ increases
31:13what you have to do is remember them and retrieve them back and so wewill what we want to see
31:17iswe want to see how many digits can you actually remember or can you hold in your memory people
31:23can ahre reliably repeat seven words from this this test
31:26it was founded that seven is the maximum number of digits that anybody can hold in short-term
31:32memory or for a brief period of time now we will not come to the time hypothesis but how
31:35many items which are there and the interpretation of the result was results support the hypothesis
31:40and from this result we were able to support the hypothesis that stm capacity is limited
31:46the idi the question was whether stm can store a number of items infinite items or if it
31:52cannot then what is the actual capacity of short-term memory
31:55and from this experimentation we were able to come up with a interpretation was since
31:59seven is the is the number of digits that most people who took part is the mean number
32:04of responses or mean of digits that people were comfortable in retrieving back after
32:09doing the digit span task so this is what the result was similarly there was another
32:13experimentwith how long can you hold the digits in short-term memory and this is called baddeley's
32:20experiment
32:21and so here the hypothesis was people can hold about two seconds of information in stm
32:25that was what the hypothesis was so how long thefrom based on research previous researchers
32:30it was believed that two second is the time of information storage in short-term memory
32:35the test was people can repeat about seven short words only two or threeonly two or three
32:42long words so if we give very short words then people can remember or repeat seven short
32:47words and but if it is longer wordswith five six seven digiletters in it or eight or ten
32:55letters in it then only two or three long words can be repeated
32:58and that was thetest that people wereasked to do and based on that the interpretation
33:04is results falsify miller’s hypothesis support baddeley's hypothesis that it is not so this
33:11isthis is also an example of falsification what miller found out is that irrespective
33:15of time people were able to hold seven plus or minus two items and here it was found out
33:22that only two or three items is what people were able to hold
33:26and so time was of essence and not the capacity so short-term memory was not a capacity driven
33:32store which means that it will not depend on capacity whether it is a time-dependent
33:37store so more number of retaliation the longer it was so stm was defined in terms of how
33:43long can information be stored it does not define or it was not da dadependent on how
33:50many words can be retrieved
33:52so by taking these experiments we will progress further and explain to you the whole process
33:57of doing research so what are the methods of science or what are the basic technique
34:03that we use in ti science the first technique that we use is something called the naturalistic
34:08observation so what is naturalistic observation these are used for this is the process of
34:13observing and describing a phenomenon
34:15so if there is a phenomenon if there is a particular experiment or if there is a particular
34:19set of observation which is out there and we cannot go in there and do any manipulation
34:27what we do is we stay away and look at what is happening and from that we collect data
34:31let us say that we want to see how a certain tribe speaks or what is theway in sae which
34:38sentences are comprehended sentences are produced or what are the language of a particular tribe
34:43now since we cannot go and meet with these tribes and we cannot go and intervene with
34:47them and we do not know their language so basically the best we can do is we stay away
34:51somewhere and observe while a phenomenon while they are exchanging their language and this
34:57is called naturalistic observations since we go to that particular tribe and froma very
35:01distance to watch-on how they are producing sentences or producing whatever of our nature
35:06of interest is this is called naturalistic observation
35:09the goal here is to describe so the maingoal of naturalistic observation is describing
35:17the particular phenomenon or describing how this happens that is the bestnaturalistic
35:23observation can do because it collects a lot of data and from there it will form data sets
35:27and describe so it is exploratory in nature we are exploring nothing is out there and
35:32so we are still exploring then we have something called correlational methods in correlational
35:37methods what we do is a mathematical technique that seeks patterns in data
35:42now let us say thatwe have no variables to start with we do not know what is happening
35:46so we we want to study how people in a particular tribe how do they communicate with each other
35:54with english language or any other language we have very set rules a very setah paradigms
35:59of language we have very setah evidences and experiments well developed experiments to
36:05study how language is produced comprehended meaning extracted from it information passed
36:11on and all those thingsare there forlanguages which are known to the world
36:15but then if we have a try somewhere in africa or somewhere somewhere far off place which
36:20has not come in close contact with the real world they want to see how language develops
36:25or how people use language there what we do iswe have nothing to start with becauseah
36:31the rules that we have for the english language or for any other language may not apply there
36:37so what we do is we first go in and look at when they are communicating
36:41and from there collect as much data or look at the phenomena the most closely that we
36:47can record that phenomena this is called a split research now when we do that from that
36:52particu the number of data points that we have the type of data that we have from that
36:57we will run a correlation correlation is the basically a technique in which we find commonalities
37:03between various variables of interest now what is a variable variable issomething which
37:08varies
37:09so thethings of interest the points of interest we find commonalities between them and that
37:14is what is called correlation in correlation what we do is since we have so much data as
37:19we have so much data points what we do is we run a correlation generally we use something
37:23called factor analysisin doing it when run a correlation we will find bunches of data
37:29or sets of data points which accumulate together and form categories and then they are clubbed
37:36together this is how the factor analysis also works
37:38so they are clubbed together to form the factors of that particular dataset so in experimental
37:43research we collect a lot of data and then we runah correlation on them and based on
37:48the correlation data points which correlate highly they will accumulate together to form
37:54data sets or to form clear-cutgroups within thatnumber of data points and these groups
38:02are what are called the factors or thesethese are what are thepoints of interest or these
38:07are the variables of interest for us
38:09so it is a mathematical technique that seeks pattern in data also basically then refining
38:14oursentence we find patterns and thoughthe goal here is prediction once we are able to
38:20find patterns these patterns will then predict that this is the kind of language that they
38:26are using or this is the kind of rule that they are using for the language that they
38:31have in this particular community and so that is what the correlation is all about
38:35and once a prediction has been done once we have these predictions or once we have thesevariables
38:42of interest or these patterns we actually take these patterns and then manipulate things
38:49in patterns or within the pattern itself we manipulate the iv that is what we call in
38:52experimentation but here what we do is we manipulate something in in those patterns
38:57so we change those patterns by certain degrees and see how a particular behaviour of interest
39:03changes and so that is what the experimental method is all about
39:07so this is from here welook at observations we build a theory from this theory we produce
39:13a hypothesis and then we test the hypothesis this indicates experimentation method so means
39:18of systematically testing hypotheses in control situations what we do is now we take that
39:24patterns and based on the patterns we bring these tribes into the lab and then manipulate
39:29those variables or those data patterns when we do that a particular kind of response will
39:35get generated
39:36so if if it is accuracy we are looking at if it is reaction time we are looking at or
39:40some other variables of interest so that is what we look at and we test this hypothesis
39:45whatever hypothesis we generate out of correlation and this the goal of this particular thing
39:50is called explanation so theory building naturalistic observation correlation producing hypothesis
39:56experimental measurmethods or testing the hypothesis
40:00based on that we produce models and theories so um what is a model based on the experimentations
40:06we then generate data model so models are cimply simple versions of phenomena under
40:10the study so most theories are expressed as models and so which are attempt to explaining
40:16the underlying mechanism typically in the form of a graph a set of mathematical equations
40:23or computer simulation
40:25so basically models are simplified vahversions of a particular phenomena which is under study
40:30and this happens in terms of either a graph a set of equations or a computer program so
40:36model could be explained in terms of a graph for example y equals to mx plus c is a form
40:41of a model or y equals to rr one x one plus r two x two plus something if it is a polynomial
40:48so it will ber one plus x one plus rx two square x three something something and then
40:55an error
40:56so this is basically a model and this model basically explains the how we manipulate this
41:02x one x two x three to generate y that is what the model explaining and this is called
41:05a linear model similarly we can have lin non-linear models so basically a model could be a graph
41:10a simple graph is a model or we can have a set of equations like this or we could have
41:14a sometime we have a computer programs also for example actr or act* or sshsoar all of
41:22these are computer models of cognitive processes or how cognitions really work
41:27so these kind of models are generated what is model it is a very simple or the very basic
41:32phenomena of aof a particularah is of aparticular thing that we are studying or a particularquestion
41:42which we are investigating now computer models what are computer models they mimic behaviour
41:46under the particular study so computer models arewhat they will do is they will have they
41:51will have external inputs or they will have computer inpubasicallysome kind of an input
41:56and this will generate an output
41:57so it is basically computer models are those kind of models which takes in a number of
42:02inputs and based on that build the whole theory out of it sothese are calledcomputer simulations
42:08and they help overcome unwarranted assumptions flaws in logic so why we do him a computer
42:14models we make computer models because computer models are very effective and very costah
42:18very logical in sense
42:20so when you make instead of testing a model or instead of testing a particular prediction
42:24of a theory in the actual subject because once we do that and if it fails then maybe
42:29the study does not respond so what we do is sshwe simulate the situation in a computer
42:33the same kind of phenomena the same kind of results that were expecting in a computer
42:38and we gather results from it
42:39and from the results we would see or we try to assess what is the problem that we are
42:44facing what is the kind of logic that we are using where is the computer failing and it
42:46will produce results so un till the point of time it produces the results that we want
42:50or it gives some kind of a result we never actually go ahead and test the model on the
42:55actual data points so computer models help us in unwarranted assumptions and when we
42:59are making false assumptions we will get somethingyou know computer models we can verify we can
43:03re-done the model
43:05for example look at act-r the way the procedural memory and the collective memory is described
43:09in act-r indigenes conceptah in terms of the spatial orah the thaspecial code and the prepositional
43:18codethe way it is described in act-rthis basically is a good explanation of how a computer model
43:24really works or how they can solve problems in the real world and ssh also flaws in logic
43:29can be pointed out by computer programs now models and theory
43:34good models lend plausibility to the theory if we have a good model it will always build
43:38up a theory and only a data can support or falsify a theory so once we have a theory
43:42we then go collect data based on what the theory is saying and then we either falsify
43:47the theory or we rebuild the theory or ifit it will never be true in any sense so wewe
43:55overcome try to overcome assumptions or or we try to overcome the limitations of a theory
43:59and build up a good theory for any kind of data
44:02constructs another interesting thing in ina research so what is a construct a construct
44:07provides ahs a scientist with a useful way of thinking about the world and so basically
44:15what a construct gives is it it it it gives a outline or it gives a way in which huhumm
44:22(coughing) scientist thinks about externalor the outside world so constructs the labelgiven
44:27to a set of observation that seem to be related
44:32so concerts are basically formats in which scientists are able toconceptualise or think
44:40about the external world or think about phenomena’s in the external world so memory attention
44:46intelligence personality even language are constructs so these are phenomena which are
44:51there and when these phenomena aremade in a processhuhumm (coughing) understood a certain
44:58process that is what a construct is all about
45:01another interesting thing or ano in in in line with construct or something called operational
45:07definitions so what is operational definition and this is very important at least in social
45:10sciences why it is important is that let us say i study a phenomena now when i study a
45:16particular phenomena let us say i am studying happiness now when i am studying happiness
45:19it may be possible that there are certain other researchers which are studying happiness
45:23and so when the result that i gave i have to say that this is the way i operationalise
45:28my depi definition of happiness let us say when i say happiness i say my happiness is
45:33bound by a b c d characteristics right so happiness is when you will havecertain kind
45:40ofelevated feeling high arousal certain kind ofbody structures or facial yex effect and
45:46so on and so forth certain other researchers would also be doing and they would say that
45:50a b c d may not be the primaryway in which happiness is defined and e f g is the primary
45:57way
45:58and so when i operationalise the definition i say that my way of defining happiness isbased
46:02on a b c d and not de e f g and that that is why my theory of happiness or the my experiments
46:09in happiness results from experiment may not fit in some other experiment then this is
46:17what is operationalising a definition so it defines the construct in terms of how it ishuhumm
46:21(coughing) measured so i say my happiness is measured in terms of a b c d
46:24and so some other people may say that it is measured in e f g and so huhumm (coughing)
46:29that is a difference that can happen for example let us look at intelligence so intelligence
46:34is based on score on test and this is what is operationalising my intelligence is based
46:38on scores of a test some people would believe that intelligence is not based on scores of
46:43test it may based on something else and so on and so forth
46:45so defining intelligence is iq or scores on a particular test is one way of operationalising
46:50the definition of intelligence also short term memory capacity isis measured in terms
46:57of digit span or it can be also measured in terms of time and so these two definitions
47:02huhumm (coughing) or these two ways of measuring myshort-term memory is what is the operationalising
47:09the definition (refer time: 47:11)
47:11now another important point that we need to know in doing research in language or other
47:17social sciences is the concept of something called validity and reliability and so what
47:23are these what is validity the degree to which instrument measures what it is claimed to
47:28measure now let us say that i i develop a test i develop a test of intelligence and
47:34when i give it to people that test does not measure intelligence it could be that i have
47:39questions which are not related to intelligence at all
47:42so i develop a test of academic intelligence which does not have questions based onlet
47:46us say reasoning oror verbal reasoning and that kind of a thing now tha in that case
47:51my test is said to be not valid so if my tests of intelligence academic intelligence does
47:57not have questions or does not have questions related to academy academics or intelligence
48:02in academics then it is said to be an invalid scale and so how do we measure validity
48:09what we do is when i have a test of intelligence i take my new test and run it against a test
48:16which is already there in the market which is supposed or which claim to measureintelligence
48:22or academic intelligence and when the scores are correlated and we findcoysa agreement
48:29between the two coursethe two test tha that are there the one which is pre-existing and
48:34the one that i have developed
48:35i say that my test is having a validity because then what it is doing is it is measuring the
48:43same thing that the other testers bring and the other test is a well-established test
48:47of intelligence academic intelligence and so if the scores are in range or basically
48:51are correlated highly oreke in some way are significantly correlated with each other huhumm
48:59(coughing) i say my test is valid now for example let us look at bathrooms scale now
49:04if bathroom scale is not able to measure it is valid for measuring weight
49:08so if in a ba bathroom scale i will stand in it and it does not give me weight then
49:12what is the point of that scale soit is valid for measuring weight and so in a bathroom
49:17scale when i stand it only measures it claims and not measuring intelligence so when i ssh
49:23stand on a scale and it says a weight of eighty this eighty is the weight is the body weight
49:30now let us say that what it is giving in my iq then it is a wrong assumption because bathroom
49:35scales the weight has no correlation with iq
49:39and what the intellig what the test bathroom scale is doing is measuring your iq and in
49:43this case this test is or this particular thing is not valid another interesting thing
49:49is called the reliability now whatever we are focusing on today should be a measure
49:54of your designing experiments inlearning or any behavioural sciences so whether it isdesigning
50:00the hypothesis or whether it is usingah a theory how do you use a theory how experimentations
50:05are done
50:06and towards the end of thissection i i will tell you about how experiments are donewhat
50:12is an experimental design and those kind of a thing so and then we should whenever you
50:17are developing a test or experiment we should be always be aware of what is the validity
50:21and reliability of the scale or the experiment that we are doing now what is reliability
50:25again the validity is basically what my test is claiming to measure and what is measuring
50:32and reliability is the degree to which instrument gets consistent measurement for the same thing
50:36now let us say i have a bathroom scale and i stand on it it gives my weight as eighty
50:44next time i ssh so this is my first reading if i go to the second reading it says eighty-five
50:49then i go here and then it says ninety then i go here it says seventy then i go here again
50:54it says eighty then eighty-five what is happening is my test is not reliable or my scale is
50:59not reliable or not been calibrated properly what is calibration it is the process of sensitising
51:07a particular instrument and so that is what it is and so this is not a it is a non-sensitive
51:11scale and nobody would like a scale like that
51:13so when an instrument gives consistent measure now the range of error could be plus or minus
51:17two kgs so if if it is in the range ofseventy-eight to eighty-two that my weight is with the same
51:25machine always then i will say it is a reliable scale but suppose if it keeps on throwing
51:29seventy sixty then ninety then whatever it wants every day then it is not reliable so
51:34basically maintaining consistency if i do an experiment and that experiment if i am
51:39repeating it again if it is giving some weird results and it keeps on givingdifferent different
51:44results on different differentreplications of the same experiment then it does not has
51:48a reliability
51:49so reliability is basically acquiring the more or less same results over multiple repetitions
51:55so daily measurements of a bathroom scaleif it is one forty-three two eighty-five thirty-seven
51:59one ninety-six it is not a reliable scale but if it is one fifty-seven one fifty-five
52:03one fifty-six and one fifty-eight is a reliable scale now this happens because of sensitivity
52:08and so sensitivity of scales would vary and so it will give you plus or minus five percent
52:12error that is what the error range is but if it is more than that then it is a non-reliable
52:17scale
52:18now before doing an experiment we have to come up with something called a hyp experimental
52:25design so what is an experimental design and before that we have toknow about what are
52:31experiments so what are experiments experiments aretightly controlled situation designed to
52:37test a hypothesis so what are experiments experiments areway of controlling a number
52:45of variables a number of factors that are extraneous
52:49now what is extraneous factors extraneous factors are those factors which may harm or
52:54which may interfere with the result if we do not control it let us say we are measuring
53:01thespeed of reading of a person now the speed of reading of person may be affected by intelligence
53:11may be affected by a number of books that he has read may be affected byahm digit span
53:19may be affected by so many things
53:20so ifi if you want a trulya true test of reading span of howahquickly somebody reads or how
53:29quicklythe eye reads and the brain is able tomind the hypothesis and how the mind is
53:34able to make meaning out of it we need to neutralise these factors of intelligence digit
53:40spanprevious experiences and all and so we control for those variables so what is experiment
53:45experiment is we control for all variables of interest or variables of interest non-interest
53:51which may interfere with the results of our experiment
53:55and we control those and then manipulate only one variable so wevariablewe manipulate the
54:00variablethat is the in in a reading span or in a reena in ina iner ai in a reading span
54:07test what we do is how quickly you can read a variable so we will give differentlet us
54:14sayah a different presentation times so you will present the same sentence for different
54:18presentation time and we will see how quickly you can read a particular variable
54:21and so this is if this is of interestthat how quickly can you read a particularsentence
54:28and then generate meaning out of it what what we want is or what we can do is we can value
54:33the presentation time and thenmay be with control for intelligence agent those kind
54:39of thing i raised develops the more repeatyou develop the moreah chances that you will be
54:44able to rightly predict a particularsentence and so we control for that
54:48and then what we do is we present thesethe sentences in different different speed to
54:53the same intelligence people orah balanced intelligence people and they were able to
54:59generate a particular hypothesis out of it so basically it is a tightly controlled situation
55:02designed to test a particular hypothesis it is a comparison between two groups that are
55:07treated differently so basically then i will have a group in which so first ailli will
55:15control all variables of no interest or we call them extraneous variable in research
55:22and then what we will do is we have two groups the experimental group will see variations
55:27in speed of presentation of the sentences and the control group will have one standard
55:33speed and then later on we will see how accurately are you able to produce the sentence back
55:41or able to comprehend the gist whatever your question of interest is andmy experimental
55:48group in the group in which you are making the manipulation and control group is the
55:52group in which you arenot making manipulations
55:54so we are controlling between two groups that are treated differently so in this group we
55:57are making the manipulations in this group we are not making the manipulations and sothat
56:02is that is how we do again we look at huhumm (coughing) baddeley huhumm (coughing) and
56:05bransford study and we will see how these things work so basically then experimental
56:10design hypothesis are derived from a theory by the logical process of deduction and experiment
56:16is then designed to test this hypothesis
56:18these are the process of regenerate an experiment now experiments compare the performance of
56:23different groups the experimental group is given a different treatment to test the hypothesis
56:29the control group goes without the treatment in order to provide a baseline of the comparison
56:36so what is an experiment experiment can be viewed as a stimulus response test and we
56:42will we will look at iv’s and dv’s for this so basically this is how an experiment
56:45is
56:46and so what we are trying to do is we are trying to take to basic classic studies and
56:51we willah talk about how theseexperiments were done and what are the variables in what
56:57how these experiments what are the extraneous variables inta interest variables and what
57:01are the factors in doing an experiment so one is we have tad baddeley nineteen seventy
57:06five study this is the hypothesis here was stm capacity limited by lengthnot number of
57:12words
57:13so how many words are there that is not of interest how much length of the word is and
57:17the method is group a repeats short-term words group b repeats long-term words as as i said
57:23basically what we did was if this is what my thing is i cari create two groups my first
57:28group looks as short term words see see if it is the question is whether word length
57:34not number of word is the reason or or or is is what determines stm capacity i create
57:42two groups
57:43better it would be to create three groups in one group i have shortah length words in
57:49the other group i have long length words and in the third i will have intermediate words
57:53so basically then i can have three groups or i can have two groups only blince since
57:56we are focusing on baddeley study so you had one group looking atahshort words may be two
58:03plus or minus one word and then seven plus or minus two word kind of a thing and then
58:08that is what how the experiment was done that is the method
58:10in bransford and johnson nineteen seventy-two study the hypothesis was context aids ambiguous
58:15story comprehension but the question here was whether contextlet us say that i am talking
58:21abouti am randomly talking about something so i am telling you that after this i am telling
58:25you a story so after this youyou take some powder put it into the machine put things
58:30onto it and then it it goes around and around takes in water and something something until
58:34i ssho show you the photo of a washing machine you will not be able to comprehend that i
58:39am talking about the washing machine
58:40and so that is what we are trying to see that whether context whether picture of something
58:45can lead to ambiguous story comprehension so basically can the photo of it or when some
58:53kind of context on which the story is being saidthat context help us in in making meaning
58:59out of ambiguous sentences the method here was group a sees the picture hears the story;
59:04group b hears the story no picture
59:06so the group when they see a picture they immediately find out that this is what the
59:11washing machine is or they see a washing machine and then the hear the story so they are able
59:15to narrate the fact that this whatever description is coming off is coming off about the particularwashing
59:20or the washing process so when group b hears the story the random story and does not see
59:26the washing machine they have no idea that we are talking about the washing machine
59:31now for any experiment to proceed most experiments have two type of groups one is called the
59:42experimental group and the other is called the control group let us say i want to do
59:46an experiment any experiment which is there and when i want to do an experiment the first
59:52thing is that i have to define the variables of interest let us say the variable of interest
59:58is whether speed of sentence presentation huumm (coughing) influences so speed of sentence
1:00:10presentation influences comprehension
1:00:17if that is whatmimyah question is so speed of sentence presentation is the iv because
1:00:24that is what we will mameasure and sentence comprehension which is how nicely you can
1:00:29read a sentence or how nicely you can retrieve a sentence back get the gist of the sentences
1:00:33is what is my adi huhumm (coughing) this is called the huhumm (coughing) this is called
1:00:37my ahuhumm (coughing) dv so iv is the speed of sentence presentation and dv is the sentence
1:00:42comprehension and huhumm (coughing) then what we do is we canwe create two groups
1:00:47experimental group is the group in which i will vary the speed of presentations so i
1:00:51will take a sentence and present it for let us say one minute two minutes three minutes
1:00:57or let us say hundredth of a minione thousand of a millisecond one hundred of a millisecond
1:01:03five hundred of a millisecond one thousand milliseconds two thousand millisecond and
1:01:07so on and so forth in the control group i do not manipulate my iv so i have a fixed
1:01:12iv which means that one thousand milliseconds is thetime in which the sentence is being
1:01:18presented
1:01:19and then i measure the sentence comprehension in terms of how accurately can you reproduce
1:01:26the sentence back and that is what is calledthe dv so what is the experimental condition groups
1:01:30that is given treatment to test the hypothesis that is called the experimental group and
1:01:35control group or control condition is group that is not given the treatments so in this
1:01:38case here is the this provides baseline for comparison because this will tell me whether
1:01:43these speeds have any effect or not
1:01:46again looking at baddeley and others nineteen seventy five study the experimental group
1:01:50was repeating long words test hypothesis and the control group are repeating short words
1:01:55replicate digit span desk the digit span task in which we found out that it was two seconds
1:02:01in which people were able to comprehend that
1:02:03in the bransford johnson study nineteen seventy two the experimental group was sees picture
1:02:08hears stories test hypothesis and control group was they hear the story and no picture
1:02:12as i said the control group would not see the picture but hears the story but my experimental
1:02:16group would hear the story as well as see the picture and so this is what the difference
1:02:22between mycontrol group and experimental group is all about
1:02:26nowlet us also look at since we have taken up the idea of what an iv and dv is now first
1:02:31of all what is an variable so any equation in this form any function in this form is
1:02:41called a variable so this x and y are variable what it is saying is x is athafunction of
1:02:45x is y so when i replace when i put values of when i manipulate values of x i will get
1:02:52values of y and this is what a variable is so variable is anything that keeps on changing
1:02:57and so independent variable is the variable that we manipulate that we change so in ainah
1:03:02the speed of presentation test the speed at which we present the sentences is what is
1:03:07called the iv and dv is the gist the comprehension the retrieval of sentence that you are doing
1:03:13which is called the dv so my independent variable is various types of treatment given to different
1:03:18groups in experiment and my dependent variable is measurement of responses of each participant
1:03:23makes to that particular treatment
1:03:25in my baddeley experiment the iv was short words or long words then dv were number of
1:03:30words correctly recalled so this is called accuracy and these are the manipulation of
1:03:36so word length is what the manipulation is in my bransford and johnson study the picture
1:03:40and no picture so basically it is called picture stimuli since both the groups heard the story
1:03:47the only thing that we are manipulating is the picture versus non-picture and so here
1:03:51picture versus non-picture is my iv and dv is subjective rating of difficulty number
1:03:55of items recalled again it is accuracy how correctly how many words can you correctly
1:04:00recall
1:04:01andlet us lastly look at something calledbetween subject and within subject designs then we
1:04:08will look at hypothesis testing may be in the nextclass so basically then there are
1:04:13two type of designs that we use so sometimes what happens is it is very difficult to control
1:04:18forwithinsubject forsubject factors let us say thatah there are certainah there are certain
1:04:25kind of experiment which require the same subject to be repeated
1:04:29because certain subject factors may be responsible for the different results weah that we are
1:04:35getting and in those cases we use something called a within subject design so if the same
1:04:38subject is being repeated in both in the control group as well as the experimental group this
1:04:44is called a within subject design but if the different subjects come in to the experimental
1:04:49group and the control group this is called a between subject designs
1:04:52so what is a between subject design assigns each participants to only one condition whereas
1:04:56the within subject design assigns each participant to every condition right it it may be possible
1:05:01that there is one factor that we are interested in and that factorahmay bemay become different
1:05:08for different people and so what we do is to preserve the sanity of it we repeat the
1:05:12same subject across different conditions different testing and this is called the between subject
1:05:17design so let us againlook at baddeley’s experiment
1:05:20so in baddeley’s experiment each participant repeats both short words and long wordson
1:05:25separate trials and this is a within subject design and then bransford and johnsonah same
1:05:30participants seepic some participants see picture some do not and hear the same story
1:05:34so they use a within subject design now hypothesis testing is something that we will do in the
1:05:38next class and look atah further
1:05:41so let us quickly then briefly go over what we did in today's class so other thangoing
1:05:46back and telling you about what we did in the first class in reviewing with the first
1:05:51class in the present class what we did was we looked at how experimentor design or how
1:05:56experiments are done what are the variousfactors or what are the variousconstituents of a of
1:06:04a particular scientific research experiment we looked at the conditions under which an
1:06:08experiment is done
1:06:09we looked at the mechanism of doing an experiment we look at the process of induction in induction
1:06:14and howwhat are theories what is hypothesis what is aah experimental design what types
1:06:20of experimental designs are there and we took some model systems we took some model experiments
1:06:25and looked at how these experience fit into the conceptualisation of experimentation that
1:06:29we have beentalking about
1:06:31in the next class onwards we will look atah how hypothesis testing is actually done and
1:06:36we will further look at we will take some examples from language and see how to construct
1:06:41an experiment and look at some experiments which have been done in the language and discuss
1:06:46those experiments in detail so in all we will tell youah i will tell youah within these
1:06:51two lectures how research is done inah inlanguage and specifically in language and generally
1:06:59in any behavioural sciences so until we meet again in the next class it is goodbye from
1:07:04here thank you

About this transcript

This page contains the full transcript of Lec 3: Science of Language - I by NPTEL IIT Guwahati, generated from the public captions YouTube serves with the video. The transcript has 11,178 words across 744 segments, with the original timestamps preserved so you can click any line to jump to that moment in the embedded player.

What you can do with it

Use the transcript to take notes, quote the speaker, build a study guide, generate a summary with ChatGPT or Claude via the YouTube Summary tool, or export it as a timed subtitle file with YouTube to SRT. You can also re-open it in the transcriber to translate the transcript into 100+ languages.

Free YouTube transcript tool

YouTube2Text is a free YouTube transcript generator — no signup, no daily limit. Paste any YouTube link and get the full transcript instantly, with timestamps, click-to-jump, translation to 100+ languages, AI prompts for ChatGPT, Claude, and Gemini, and exports to TXT, SRT, VTT, or Markdown.