 Good morning. Good morning sir, lovely. I want you to be both physically and mentally present in the class. You cannot participate in any exercise unless you do it whole heartedly. We have been talking about production of a speech sounds. We have done two modules now. We began with talking about what makes language a unique system of communication, you know design features of language. And then all of last few weeks how do we produce different kinds of speech sounds, how many different kinds of speech sounds there are, how do we describe one sound differently from the other, the terms, the terminology, the phonetic transcription and all that sort of thing. I know it is a lot of new terms for you, but with a little effort you can remember them and you can use them. At the quiz I am going to ask you also about the definitions of these terms later this month, plus transcription, plus fill in the blanks, etcetera, etcetera. A lot of objective information. Please bear with me. Beginning today we will take up the third module of the course that is phonology. Now what is phonology? Phonology, what is the difference between phonetics and phonology? Very crudely, very briefly, in phonetics we have been talking about individual sounds. One sound at a time, how is it produced? What are the characteristics of this sound? Whether it is palatal or bilabial, whether it is voiced or voiceless, whether it is back or front, whether it is a vowel or consonant, a diphthong or monothong, we have been talking about its characteristics. But beginning today we will talk about the groups of these sounds. A sound in a speech sound in context. Phonology is the study of a speech sound in use. Please write. How do you pronounce this word? Phonology. What was the earlier subject called? Phonetics. Phonetics, not phonetics. That was phonetics. This is phonology. What does it do? It studies the use and organization of speech sounds or sounds of speech sounds of natural languages. Now, when you say use, it automatically means organization. When you say organization, it means use. It is redundant to have both the words. What I am trying to tell you is that none of us, none of us ordinarily speaking, there may be one or two sounds occasionally when we say a mango. But even a mango comes before mango. We do not speak single sounds in isolation. Nobody goes and says pa and keeps quiet. Or nobody goes and says pa and keeps quiet. Human situations are very dynamic. You can say when somebody is tired, fatigued, irritated, bored or had a quarrel with his wife or his boss, then he says ha. These are exceptional situations. Natural situations, human situations are very dynamic. That is why we cannot statistically always predict human behavior. That is why human behavior remains beyond formulaic descriptions. But by and large, let us say 80 percent, 85 percent of the times. We use speech sounds in a context. We never say just pa or ha. Do we do that? Does anyone do that? So, whatever I have told you during the last few weeks is nothing but fiction. It is idealization. It is trying to understand a very complex thing in an extremely simple manner, oversimplified manner. It does not happen that way in real life. Sounds are distorted. Sounds are glorified. Sounds are exaggerated. Sounds are whispered. So, all these descriptions of saying this is voiced, this is voiceless, this is bilevial, this is x, this is y is nothing but fiction. You do not attest it in real life. In you know those of us who are from computer science and you know if you tomorrow do some project in a speech processing, you will find that this entire lot is only an idealized description. In real life, if you look at human language, if you watch yourself speaking, you will find that sounds merge into one another. They get distorted. They get exaggerated. They become longer than long. They become shorter than short. You know when you are angry, etcetera, etcetera. A whole lot of things happen and yet that is where you know the beauty of mind comes. And yet we understand one another. We expect to be understood by one another. We speak without worrying whether I am producing voice or I am not. It is not like motor car where you have to worry whether your car is in first gear or in the third. Whether you are going at the right speed or greater or lower, lots of Indians speak. Lots of Chinese speak at very half rapid tempo and yet they are understood. Lots of others speak at a very slow tempo and yet they are understood. Human language, language in use has much greater variation than the idealized description we saw last week. Do you agree? Yes or no please? Yes it is so. It is idealized. What is phonology? Phonetics is study of sounds in isolation but phonology is the study of sounds in context one next to the other. You might say just this is voiceless. When nobody speaks like that, you say house. So voicing spreads to say house, house, him. It vibrates. So he is not voiceless. It is voiced because it is in context but even voiced sound can become voiceless. If it comes in a context of, I am not saying will become voiceless. I am saying can become voiceless. If it comes in the context of voiceless sounds, just as there is social context, there can be phonological context. Phonology studies speech sounds in context and it is called phonological context. It is not just phonological context. When you use language, then there is more than phonological context. There can also be social context. For example, this is phonological context but there can be social context. Who are the phonologists or are you talking to? Are you talking to your mother or to your father? Do you think people use different languages when they talk to mother and to their father? Yes or no? Do you use the same kind of Telugu with your mother and your father? Do you or don't you? Please speak up. I am going to cry if you don't speak. Do you speak the same kind of Telugu with your mother? No. With your father? What's the difference? Plus minus or respectful word. I will say there is respect also for mother but with mother you might not say, you might not say, you might not say, you might not say, you might not say, you might not say, you wait. I don't have time. We can have different kinds of words, different kinds of pronunciation. Social context also influences language. Are you angry? Are you happy? Are you irritated? Are you bored? What kind of good? Are you speaking standard Are you speaking a standard language? Are you speaking slang? Are you speaking to a friend? Are you speaking to an enemy? Are you speaking to someone you are very close to? Are you speaking to someone you hardly know? Are you speaking when running for the railway train? Okay? A variety of social situations. Change language. Change your pronunciation. Change your use of speech sounds. Similarly, a variety of phonological contexts change your speech sounds. What kind of words? What kind of situation? Is it next to a voice sound? It is after a voiceless sound. Does it come before a lot of consonants? Does it come after a lot of vowels? Does it come at the end of the word? Does it come at the beginning of the word? There is a difference between at the beginning of the word and at the end of the word. There is a difference between pot and top. Say pot and top. When you say pot, both your lips part. But when you say top, both your lips stay shut tight. Say it? Pot. You don't say top. Do you say that? We don't. So all this fiction that is bilabial, voiceless, plosive is fiction. It doesn't happen that way in real life. So there is phonological context. All speech sounds can in context undergo these processes. Please note, lesion, they can be deleted. How often we say do not? I do not want it. We sometimes say don't want it. What happens? Oh, you know, phonological processes. There may be a lesion, may become, what happens to? What happens to a? A bout may become bout. Birthday may be deleted. It may be just come from a birthday. Look at this sound. I will use phonetic transcription. Birthday may become birthday. How often do we say can you please come for my birthday party? We say can you join me for birthday party? And yet we understand one another. Because the context provides that context. Or there may be insertion. An extra sound may come in. Insertion, an extra sound may get in. You know, when you go to England and you hear a lot of English people, you don't have to go to England. You can hear them on YouTube. Many of them pronounce this word as, how do they pronounce it? Lord and order. This word comes in. When they have to say India and Pakistan, they quite often end up saying India and Pakistan. India and Pakistan. An extra word gets in, becomes this. Gets in. Extra. Or, you know, in the pronunciation of many of us, film becomes film. Right? Have you heard that? Yes or no? Yes sir. In the pronunciation of many of us in Bihar, Bengal, we don't say station. We say a station. We don't say speech. We say a speech. Or in Punjab, you know, they give sapeech. They go to satiation. They play supports. It happens, you know. An extra sound may come in. An extra sound may be lost. Or sound may become longer. Next to someone in Sanskrit, you know, we have the phenomenon of sandhi. In Telugu, we have the phenomenon of sandhi. Am I right? Yes sir. Okay. When you say maha and aakash, it becomes mahaakash. Okay. It becomes so much longer. Or it may become shorter. When it comes to a very long sound, you know. Otherwise, ordinarily speaking, it is dine. But it becomes dinner. I becomes e. In English, dine, dinner, reside, residence. I becomes e. Lots of, you know, these processes happen. In other words, in other words, in context, in context, features change. It is no longer the same thing as it is in isolation. Because we don't speak in isolation. The context, the contextual influence is very, very important. The point here is, the question here is, can we write a grammar of these things? Can we say only x will come before y? Only y comes next to x. That is the entire goal of phonology. And if you can do that, then you can create machines which can speak like human beings, which can understand human beings. Okay. Let's look at some variations. The same language. But look at, you know, you can close your eyes and you can say they are not the same people. Would you mind closing your eyes? Don't look at them. Just listen and see if it is the same person speaking. Don't look at that person, please. Just keep your eyes closed and I'm going to ask you some questions. Open your eyes. Who is speaking? A man or a woman? A young man or an old man? Old man. How do you know? Because you know the voice. Okay. Indian or non-Indian? Indian. An Indian. Okay. Can you listen again and pay attention to some peculiarity which you may not find in somebody else's voice? Does Gandhi say senses or senses? What does he say? Come on, make a guess. Gandhi won't get angry with you. Right. You know, we have these individual little, little variations. That is why even on phone when your speakers are not before you, you can immediately recognize whether it's your friend or your worst enemy speaking. Okay. How do we know that? Because our mind is tuned to recognizing tiny variations of language. Okay. You know, absolutely little, small variations of language. I'm going to play something else. Okay. Close your eyes. Don't look at the audio video and just listen. Tell me who's speaking. Don't look at the audio. Don't look at the video, please. I'm going to ask you again the same questions. Close. Keep your eyes closed. Who's speaking? A man or a woman? Young or old? Old. Old. Indian or any other country? Indian or any other national? African. African. Do you saw the video, please? I gave the game away. But please open your eyes. Okay. He sounds different from Gandhi. Nelson Mandela says Gandhi influenced him much more than anyone else. Okay. Did you think Gandhi also influenced him in his English? He has his own voice, very characteristic. Can you recognize anything typically African in his voice? Does he say ugly or ugly? Okay. Please listen. Try and see if he sounds any different. What gives you the impression that he is an African? If you didn't look at him, there are, you know, I have not trained you. We will look at these features. We'll look at these individual features. Okay. Let's listen to something else. Okay. And learn to identify variations. Close your eyes once again, please. Close your eyes. Okay. I'm sorry it's in Hindi. How many of you know Hindi? Okay. Not a fair test anyway, but let me see. Who is shouting? A man or a woman? How old? Okay. Listen again. Who is this speaking? A man or a woman? A woman. Old or young? Very old. Very old or old? Okay. An old lady. How do we know? The quiver in her voice, yes. Vocal cords don't remain as flexible as they are in your case, you know. Old people's vocal cords don't vibrate as rapidly as they vibrate in your case, you know. That is how we recognize, okay. Here is an old man. And all of us can mimic. Close your eyes once again. Who is this man or a woman? Close your eyes. Young or old? Indian or European? Listen again. Indian or European? Did you see it or do you know from the voice? Voice. Each of us brings some peculiarity. Our own. I can give you lots of examples. You can talk the same song, the same speech. Two people produce it in two entirely different ways. Okay. Two people produce it in two different ways, you know. And phonology is the study of these differences. And yet this is important. And yet these similarities, okay. Yet we understand in spite of these distortions, in spite of the quivering of the old speaker, in spite of the voice of the very young, in spite of being in a rage or giving a lecture happily, we understand one another. And phonology is a study of these processes. Okay. You know, there is no point giving you many more examples. I'm sure you agree that variations come in natural languages because of variation in contexts. There is phonological context. There is social context. There are other kinds of emotional linguistic and para-linguistic or you know extra linguistic factors that influence your speech. Sometimes you deliberately do this. Sometimes you do it involuntarily without realizing that you are speaking rapidly. Without realizing you are speaking loud or that you know in all kinds of other features. In other words, come back to generalizations. Theoretically, are we together? So far all I have tried to tell you is that phonetics talks about individual speech sounds but individual speech sounds do not happen in natural languages, correct? Number two. Languages, speech sounds occur in contexts. There can be at least two different kinds of contexts. Close your eyes and tell me what kinds of contexts there can be. There can be phonological context. Sounds coming next to other sounds. Sounds coming between other sounds. Sounds coming at the end of other sounds. There can be phonological context. There can be social context. You are speaking to someone who is very close to you, intimate. Someone who is a stranger. Someone with whom you are not on good terms. Someone much older than you. Someone much younger than you. All of us speak to young people, you know, or very young children in a very different way. Can anybody give me an example of how you speak in Telugu to a child who is about one or about two year old? Invite the child to your, you know, or just ask the child what is his name. Can you do it in Telugu? Give me an example. Let Telugu come in the... Do you say me, pair and d? Do you say that? Okay. What do you say? Or simply pair, pair, pair, pair, pair, pair. Don't we do that? Do you go to a young child and say, can I have your full name, please? Do you do that? What do we ask? Name, okay, pair, okay. Give me some example in Tamil or Telugu. Speak up, please, so that the camera catches you. Speak to the camera, please. How would you say that in Telugu? Okay, and if it is a girl child, further, more polite. More polite, you know. There has actually been actually an experiment. Some western researchers, some American researcher one day at a birthday party, dressed a child, a girl child in boy's dress and lots of guests came and said, hi, lion, how tiger, hi, king, how are you? Okay, and then the same child at another party was dressed as a girl and everyone said, hi, sweetie, hi, honey, okay. You know, this is social factor. It's not that one is being good or bad. It is not that one is being respectful or disrespectful. Language is a part of culture and there are culture-related variations just as there are linguistic variations. Now, culture-related variations are not so easily predictable. We cannot say in this party how many guests will tell this girl hi, honey, hi, sweetie, or hi, ugly little thing, get away, okay. We don't know. You know, people might, we have not reached those levels of knowledge where we can confidently predict human behavior, but linguistic variations can be predicted. So, these are the two things that I have tried to tell you so far that none of us speaks or produces a speech sound in isolation. They come in a context, okay. And there are at least two contexts. What are those? Close your eyes and tell me. Phonological context and social context. Social context is not so easily predictable. Not yet. We do not know enough. But phonological context is predictable. We can say whether the sound coming here is likely to be voiced or voiceless, front or back, this or that, X or Y, okay. Why does that happen? Let's move further. Are we together? Do you understand? Theoretically, all of us can produce all kinds of sounds, but we don't. There are some sounds which are peculiar to Telugu. Peculiar to Tamil. Tamil is not Tamil. What is it? You have to roll your tongue. Okay. What is it? Tamil. It is Tamil. Okay. How many non-Tamil speakers can speak that way, please? Can you? Okay. Are there similar sounds in Telugu? Malayalam has... It's not Malayalam. It's Malayalam. Okay. Malayalam is a speaker here. Okay. Right? How do you pronounce the name of your language? Right. It's not all. Okay. Similarly, there are sounds in English. When you say vision, V-I-S-I-O-N. Okay. It's peculiar to you. Actually, it comes from French. It's not even English. Okay. Only Frenchmen and some Englishmen can do it. In Indian language, only Urdu has this. Okay. So, even when theoretically all of us human beings can produce any sound, we don't. Through influence of culture, through the fact that we are born in a particular community, in a particular family, our language excludes some sounds, includes some others. Okay. That is why. Languages have only limited number of sounds. No language has 200 different kinds of sounds. Do they? How many sounds are there in Telugu? Make a guess. Make a rough guess. There is no punishment for going wrong. As I said, English has 44. Okay. How many? That's the letters. Okay. Not speech sounds. When you can go back, you can count. Please. You know, we are learning. In Hindi, there are 39. Okay. In my mother tongue, maybe there are 36 or 37. Okay. You know, even when theoretically it is possible for any language to have any number of speech sounds, languages have relatively few sounds. We are yet to come across a language which has 60 discrete sounds. Mostly they are between 35 and 50. Okay. Making use of. Mostly. There may be, if there is an exception, I don't know. So, these sounds differ in number and differ in kinds. There are many languages which don't have as many vowels as English has, which don't have as many consonants as Sanskrit has or as Malayalam has. Okay. There are differences. Moreover. Moreover. You know. Not only that, there are restrictions. There are fewer. Speech situation. And this is a baffling problem. This is a baffling challenge for engineers who want to do engineering in natural language processes in NLP, in speech processes. No person speaks on two occasions exactly alike. Not even the best of singers. Some variation. A nanosecond longer vibration of vocal cords. Or a nanosecond, you know, or a nanocentimeter. More or less opening of jaws. Or pressure on a different part of tongue. All of these things together change the sound. Is never, never exactly again. Sorry. It is never, never exactly the same variation. Variations do come in. Actually there has been a philosopher, a Greek philosopher, and we have similar opinions in Indian philosophy as well, which says that we do not swim in the same river twice. Okay. By the time you have gone to the river, the river has flown. Okay. It's another set of water. Another drop of water. Another mass of water. Similarly, you know, no matter how hard you try, it is almost unlikely that you can produce all speech sounds the same way as you did about a minute ago, or about a second ago. We don't speak alike on all occasions. And yet we recognize the speech sounds. We understand. We use them. We feel happy and unhappy because of those sounds in a party. A lot of people are speaking. There is noise, and yet two people can talk, understand one another. Computer cannot. There have been attempts, computer recording. It's called, please write, cocktail effect. Okay. Lots of people speaking, but computer cannot recognize. For computer that's a noise. Whereas human beings can cut all the other sounds. And if you want to talk only to your mother, or to wife, or husband, or brother, or father, you just listen to her. Even in a crowded railway station, in a crowded airport, in a train, in a party, anywhere. Okay. This is the capacity of human mind. But the point is, how does this capacity work? This capacity works because your mind remembers, recognizes, and uses. Recognizes, remembers, and uses. Rules concerning variations of speech sounds. What kind of shape is it likely to take in this situation compared with another situation? Do you understand? Are you with me? Two people are with me. Are you with me? Yes. Right. Please keep saying that. Otherwise, I feel very nervous. Okay. So, and no sound is uttered alike in two different contexts. Look at the example. I have on this screen, we call it bilabial. Okay. But look at in, as I give you the example earlier, pot and top. When you say pot, lips part. When you say top, lips do not part. They just stay put. Look at pot in pit and in put. When you say pit, your lips are spread. Also for pot. But when you say put, your lips are rounded. Also for pot. Why is it rounded? Can you give me an answer? Because of you. Because of who? The following sound is who? Why does it spread? Because the following sound is E with flat lips. So, you know to say that this is bilabial, this is labiorental, et cetera, et cetera, is nothing but idealized representation of speech. Or look at E itself, you know. Look at the next sound E. E is not the same again. I told you, can you give me three term level for E? Is it voiced or voiceless? It's a vowel, therefore it is voiced. What are the three term levels? Front or back? E, E as in leave, as in grieve. Is it front? E is front, closed or open? Closed. Rounded or unrounded? Unrounded. I said E is front, closed, unrounded. Now look at E in both, in pit and in give. Just say pit and say give. Where does E come from? Does it come in give? Does it come from the front of the term? Say pit and give. Please say again and again. Pit, give, pit, give. Do you think E in both the instances is produced from the same point? Where does it come in the first? Front. Where does it come in the next? Back. Back, you know. So these things are extremely dynamic and yet, even though, this is the beauty of the mind. This is the power of mind. Even when it has shifted from front to back, we recognize, this is, partly we recognize because we know the word and we think it is this way. But who has sat down and told a child? Child, child, E is not the same thing in give as it is in pit. Does anyone do that? Do fanaticians do that to their children? Nobody does that. It's the power of the mind. You know, what is that power? What is that algorithm? You know, that is what we have to capture. People speak differently on different occasions. Yet, listeners recognize the words. How do they? What kinds of factors are there in play? Look at some more examples, you know. In English, we say plural marker is sss. But it's not the sss in every, everywhere, you know. After cat, it is pronounced as sss so that it is cats. But after dog, it is zzz. How is it pronounced? Dogs. Why does it happen? Can you, can you make a guess why it happens? Why cats but not dogs? Or why dogs but not cats? The effect of the context. Immediately preceding sound is voiced or voiceless. And speech being dynamic does not get enough time to adjust. But it may if the speaker so likes. Many of us in India don't say dogs. We say dogs. We can. But ordinarily speaking in standard English, this is what happens. The question is why? So, once we look at this kind of data, we can say look, look in English. We can predict. We can tell our computer, computer, computer. When you see s coming after voiced sound, then produce it as zzz. Elsewhere produce it as sss. We can, we can make these rules. Similarly, look at another example. Past tense in English written with ed but produced differently. After ask, it is voiceless. It is asked. But after voiced sound, it is voiced. It is not moved but it is moved. It is voiced. Context, phonological context can be predicted to a great extent. You know, in ordinarily you see in every language. You see in Telugu. You have chepa, fish. But when you add plural, does it become chepa-ulu? Chep-ulu. Chep-ulu. You know, goes away. If something else comes in, harmony comes in. Speech and you can predict that chepa with lu would become not chepa-lu. It will become chep-ulu. Similarly, you know, here we can. We have the same article but before vowel it is an. So you say an orange, an egg, an enemy. But a friend, a guava, a mango. Purely phonological context. Or the sound, nasal sound an. Before t, it is an. You can say and, and. But when there is the following sound is ke, then you say thank, sank, rank. Even in Indian languages, you do not say gan-ga. You say gan-ga. Do you say gan-ga? Do we say gan-ga? Do we say telan-gaana? What do we say? Telangana. We go anga. Because the following sound, which is the velar sound, also influences the nasal. So there are, you know, in all languages of the world, there are influences of context and, you know, you can, you can find examples in any language you like. But, you know, look at the three words there. Peak and kept and keep. Okay? When you say peak, lips are spread, lips part. But when you say kept, okay? Lip has already, tongue has already moved to the position of alveolar ridge for tu. And you say kept, you know, your, your lower lip slightly moves towards teeth. And when you say keep, your lips do not part. Okay? So with, so with, there are some variations which bring a change in meaning. But there are some other variations which don't induce such, you know, changes. Look at variation in these words. If you, if, you know, you have pit, p, i, t, upper, e, t. But if you change tu with n, meaning changes. Okay? There are all kinds of variations. Change n with l, it becomes another word. Change l with k, it becomes another word. But if you change it with, you know, spread lips or lips together, the meaning doesn't change. Whether you call it top or top, meaning doesn't change. There are different kinds of variations in this module in phonology in the next few weeks. We will study speech variations and how we can predict them. Thank you. Have a good day.