Oral History Transcript — Dr. Kenneth Stevens
[an error occurred while processing this directive]
See the catalog record for this interview and search for other interviews in our collection
Kenneth Stevens; February 13, 2001
ABSTRACT: Born in Toronto, Canada in 1924; University-based high school offered an excellent education in math. Attended University of Toronto for an undergraduate degree in Engineering Physics in 1945 and completed a Masterís thesis on servomechanisms in 1948. Employed as an Instructor for the Canadian Version of the GI Bill, came to MIT in 1948. Worked with L. Beranek in Acoustics lab and completed doctorate in 1952 with a dissertation about the perception of sounds shaped by resonant circuits. A research staff member of MIT from 1952-1954, he accepted a faculty position at MIT in 1955. Promoted to Associate Professor in 1957. Worked with G. Fant while on sabbatical in Sweden in 1962 and brought back x-ray films that formed the basis for early work with J. Perkell. Promoted to professor in 1963. Collaborated with D. Klatt on pioneering speech synthesis project, Klattalk, that formed the basis for DECtalk and many of speech synthesizers in use today. Traveled to England as visiting professor at University College, London 1969-1970. Served as President of the Acoustical Society of America 1976-1977. Awarded Clarence J. LeBel professorship in the department of Electrical Engineering and Computer Science at MIT in 1977. Received Gold medal from the Acoustical Society of America in 1995. Published an acclaimed book, Acoustic Phonetics, in 1998 that uses techniques of circuit analysis and signal processing to elucidate how a discrete linguistic representation is translated into articulatory movements so that their acoustic effects produce speech communication.
Matthies: We are at MIT in Cambridge, Massachusetts, United States of America. The time is around 9:15, and I am about to interview Kenneth Stevens for the Acoustical Society of America on the Technical Committee on Speech Communications. Good morning, Ken. Could you tell me your present address?
Stevens: My present address is in the Research Laboratory of Electronics at the Massachusetts Institute of Technology, Cambridge, Massachusetts, Room 36517.
Matthies: And your phone number?
Matthies: So your present employer is MIT. What is their business?
Stevens: Their business is education and research.
Matthies: What is your current job title?
Stevens: My job title is Clarence J. Lebel Professor of Electrical Engineering.
Matthies: How long have you held that title?
Stevens: I have held that title since about 1965, I think.
Matthies: How long have you been at MIT?
Stevens: I came to MIT as a graduate student in September of 1948 and I went on the faculty in 1955.
Matthies: What do you do here?
Stevens: I am involved with education, teaching. Part of it is teaching in the classroom, part of it is monitoring the research of graduate students, and advising graduate students, and doing my own research in speech communication.
Matthies: Do you remember what year you joined ASA?
Stevens: I think it was around 1950 as a student.
Matthies: So you had already been at MIT for a couple of years?
Stevens: Yes, I was a student member at that time. Right.
Matthies: What area of acoustics were you interested in at that time?
Stevens: I was already interested in speech communication but I also did some work in room acoustics and in noise control.
Matthies: When did you join ASA?
Stevens: Well, when I came to MIT. Shortly after I came to MIT as a graduate student, I began to work in the Acoustics Lab and I was invited to do that by Professor Leo Beranek. Even though, prior to that, I had not any serious interest in acoustics, I developed that interest at the Acoustics Lab and I joined the Acoustical Society.
Matthies: That is quite an invitation then. Leo Beranek. Very nice. What ASA committees were you a member of?
Stevens: Over the years, I was a member of the Speech Communication Technical Committee. I guess right at the very beginning when they started having technical committees, I chaired the Speech Communication Committee. Then I became, at some point, Vice President of the Acoustical Society and that made me Chairman of the Technical Council of the Acoustical Society and a member of the Executive Council of the Acoustical Society. I think subsequent to that, I was a member of the Long Range Planning Committee of the Acoustical Society. And somewhere earlier than that, I was an Associate Editor of the Journal of the Acoustical Society of America.
Matthies: Have you been on some of the committees that put together the Acoustical Society meetings and reviewed the abstracts?
Stevens: I was involved, yes, in one meeting that they had in Cambridge. I was a member of a committee.
Matthies: Was that fun?
Stevens: It was fun. Yes.
Matthies: It is very interesting to have the Acoustical Society meeting in your hometown. That is very different.
Matthies: You look at it through other peopleís eyes.
Stevens: Well, one event that I remember. In fact, at that time I was in charge of arranging the banquet. We decided that we would have a clambake. It was in the summer and so, we had that out someplace on the North Shore. Everybody seemed to enjoy the event. The only thing was that the P. A. system did not work. So when the time at the banquet came for the presentation of awards, that presentation was done without a P. A. system, so it was basically just a little circle of people around one of the tables and no one else was paying any attention. I remember, I think Dave Green won an award at that time. So, I actually failed at my job.
Matthies: You couldnít provide sufficient acoustics.
Stevens: Thatís right.
Matthies: Are there any other particular Acoustical Society meetings that stand out in your mind? They have had some wonderful meetings all across the world.
Stevens: Yes. I think the first one in Hawaii. No, actually I missed the first one in Hawaii, but a couple of the ones in Hawaii I particularly enjoyed.
Matthies: That is a lovely spot.
Stevens: In fact, when I was President of the Acoustical Society, at that time we arranged for the first meeting in Hawaii which would be a joint meeting between the Japanese Acoustical Society and the Acoustical Society of America. I remember meeting up in Tokyo with the President of the Acoustical Society of Japan in planning this particular meeting. But as it turned out, I did not go to the meeting. I think, maybe, my wife had a kid or something like that. I forget. There was some reason I could not go.
Matthies: Do you remember any other humorous or unusual events at ASA meetings?
Stevens: Back in the early days before the days of posters, I remember meetings where people would always present their papers orally. I remember being at those meetings with Arthur House. He was known to get up after a paper and give the presenter of the paper a hard time in pointing out that most of this work was done several years ago by so-and-so. So, people were very wary of Arthur House at those meetings, especially the young people. So, I remember those.
Matthies: When they knew he was in the audience, they got nervous?
Stevens: Yes, when he was in the audience. Right.
Matthies: How about the meeting where you were presented with the gold medal?
Stevens: Yes, that was certainly memorable for me, that one. And then the very last meeting, of course, at Newport Beach where they had this banquet.
Matthies: And the special session?
Stevens: Special session, yes.
Matthies: Now, how were those papers? Were they up to snuff?
Stevens: Yes. They were fine.
Stevens: They covered a variety of topics.
Matthies: ASA members that have especially influenced you. The first person that comes to my mind is Dennis Klatt.
Stevens: Yes. Dennis Klatt came to our lab here shortly after he finished his doctorate with Gordon Peterson out in Michigan. Klatt was a great influence on me and on speech research at MIT. No question. Of course, other people who influenced me early on were Beranek and other people on my committee like Walter Rosenblith and J. C. R. Licklider.
Matthies: And Arthur House.
Stevens: And Arthur House. Thatís right. Yes. Then later on — and we are moving ahead here but — I worked with Amar Bose for a while. We were teaching a course in circuit theory together and we ended up writing a book on introductory network theory. At that time, Amar Bose was becoming interested in acoustics, the loudspeaker aspect. So, we chatted a little about those things which ultimately led, of course, to his formation of his company.
Matthies: So he parlayed that knowledge into a vast empire.
Stevens: He did. But I think more important, as far as influence on me, was actually his attitude towards education. As we worked together on this course and on this book, he just had a very organized mind, and was very concerned about how the material should be taught, and should be developed in a very logical fashion. And so, that was a great influence on me and we wrote this book together and hopefully, it is reflected in that book.
Stevens: That gets off the topic a little.
Matthies: We are supposed to ramble. You have certainly seen the ASA evolve through the past and present. What about ASAís future? Do you think it is viable in the Information Age?
Stevens: Certainly it is. I think one of the strengths of the Acoustical Society is that acoustics covers such a wide range of fields. There are opportunities when you go to those meetings and when you read the Journal to learn about other fields like musical acoustics. If you are in speech, learning about psychoacoustics and noise, things like that.
Matthies: Besides the Acoustical Society, what are some other professional organizations you belong to? Are you in IEEE?
Stevens: I am a Fellow of IEEE. I should not say this, but I never go to their meetings.
Matthies: Are you in ASHA, Ken?
Stevens: I am a member of ASHA in fact, yes.
Matthies: The American Speech and Hearing Association?
Stevens: Right. I am a member of the IPA, International Phonetics Association.
Matthies: Now we are going to delve back into your early years, pre-college. Could you tell me when and where you were born?
Stevens: I was born in Toronto, Canada on March 23, 1924.
Matthies: Before you went to college, did you live any place other than Toronto?
Stevens: No. I lived at home and we lived only in Toronto.
Matthies: Toronto is a beautiful city.
Stevens: Yes. It has changed a lot since then but I felt it was nice then too.
Matthies: What were your parentsí occupations?
Stevens: My father called himself a business manager. He was the business manager of an organization called the Canadian Bible Society. They sold Bibles and things like that. He was not himself particularly religious but he ran that operation and they would also ship these Bibles and other things like that across the eastern part of Canada. He had that job for all his professional life. My mother and father were both born in England, married in England, had one child, my brother, in England, and then immigrated to Canada. I was born a few years after they came to Canada.
Matthies: Your mother was a homemaker?
Stevens: A homemaker.
Matthies: How would you describe yourself during those early years? Were you quiet and studious? Were you a troublemaker? How were you as a child?
Stevens: I think I was probably more in the quiet and studious direction. But I was, from time to time, a troublemaker. I would play with the other kids on the street and things like that.
Matthies: Did you take things apart and put them back together or not put them back together?
Stevens: I did some of that but if I had to write an application for college, I do not think I would say that from a young age I put erector sets together. No. I had dreams and things like that but I was not really big on that. I guess the question will come up later but I always thought that I was going to be a doctor.
Matthies: Really? Thatís what you wanted to be when you grew up?
Stevens: It was what I had in my mind, yes. I had no idea what it meant, really, to be a doctor. My mother was one of seven children and of those seven, two were boys and they were both doctors. One of those in particular was someone whom I looked up to and so, I thought that would be my profession.
Matthies: Did you have some special hobbies or interests?
Stevens: Well, like other kids, I would go skiing.
Stevens: And a little bit of downhill. Mostly locally, around Toronto.
Matthies: You were from Canada. Did you play hockey?
Stevens: Yes, I played hockey but when youíre from Canada, you donít say that is a special interest. Everybody does that.
Matthies: That is like walking and breathing.
Stevens: That is right. In the summers, we would go someplace up a little bit north of Toronto to do swimming and fishing.
Matthies: A place with a lake?
Stevens: A place with a lake, yes.
Matthies: What especially did you enjoy during high school? That is assuming you enjoyed anything in high school.
Stevens: You mean academically?
Matthies: Or extracurricular.
Stevens: Again, I was involved in sports to some extent. A little bit of high jump and things like that.
Matthies: So you did some track?
Stevens: A little bit, yes. I was not a big runner.
Matthies: So you developed your interest in running later in life?
Stevens: Later. That was later, yes. But in high school, I certainly remember the school because it was a school that was attached to the Department of Education of the University. You had to apply to get into the school. You had to have the right academic credentials. There was a small fee and my parents did not have much money but it was quite small. I think it was seventy-five dollars a year. At that time, that may have been a lot. The teachers there were really excellent.
Matthies: You received a good high school education?
Matthies: About how many hours of homework did you have a night? Do you recall that?
Stevens: I do not recall. Lots. Basically, every evening you would be doing homework for a couple of hours, I guess. Incidentally, the sixtieth anniversary is coming up this year. I am thinking of going up to that school.
Matthies: Your reunion?
Stevens: I have never been back. Anyway, one of the teachers there, his name was Johnny Workman, he was a teacher of geometry and he was someone who I especially remember. He had a way of thinking, a way of presenting things that stuck with me.
Matthies: You certainly had a strong math education.
Stevens: Right. Yes. I remember the math teachers.
Matthies: Did you have any other heroes? You had your uncles.
Stevens: My brother was always a hero to me. He was four years older and I followed him around. He taught me a lot of things about skiing and about going fishing.
Matthies: What was your brotherís name?
Matthies: Moving on to your undergraduate years and college, where did you go to college?
Stevens: I went to college in Toronto at the University of Toronto.
Matthies: So you just moved next door to the bigger building.
Stevens: That is right. In fact through college, stayed at home. I was a commuter then. That seemed to be the thing to do at that time in Canada. There were other universities but they have proliferated a lot since then. You would go to Toronto, or you would go to McGill, or Queens, or the University of Western Ontario and that was it.
Matthies: Do you remember what the tuition was?
Stevens: I think I got a scholarship out of the high school and I think it was something like two hundred and fifty dollars a year.
Matthies: A little different from MIT today.
Stevens: Yes. Right.
Matthies: How many orders of magnitude.
Stevens: Well, you figure things double about every ten years. We are talking sixty years, so it is 2 to the whatever.
Matthies: What did you major in, in College.
Stevens: I went to the Engineering School.
Matthies: Right away.
Stevens: In the Engineering School, there was the usual electrical engineering, mining engineering, and mechanical engineering. There was also something called engineering physics, which is now called engineering science which is a mixture of physics and engineering. So we took courses from the physics department as well as the engineering department. Our education consisted of things like, how do you design a machine which has an electric motor and a belt that runs it? How do you design the belt? That sort of practical engineering through to basic physics, at that time, in the physics department. This was during the war, in fact.
Matthies: Which war?
Stevens: The Second World War, which I was unable to go to because I have only just one eye. They did not like people who had one eye. But there was a lot of excitement since the atomic bomb was going to be coming up and the people in the physics world were excited about these kinds of problems in physics.
Matthies: Were you in any clubs or any special school activities as an undergraduate? Was there as physics club?
Stevens: I do not recall there being that, no.
Matthies: You had enough to do studying?
Stevens: Yes. My main activity was there was a bunch of us who played bridge. About once a week we would get together and have a game of bridge.
Matthies: So that was your hobby?
Stevens: Yes. And I also shot pool but I was not in any club. There was, in the student center, a poolroom and I spent some time there.
Matthies: I recall you owning or playing a harpsichord.
Stevens: That was much later.
Matthies: Much later?
Stevens: But at the time I did take piano lessons. I guess that was back in high school. I took piano lessons for a number of years. Another thing I did early in high school, I guess, was sing in a boysí choir. That must have been before. I was not really in high school. It was before my voice broke, anyway.
Matthies: So you were a tenor then?
Stevens: I was a soprano.
Matthies: A soprano?
Stevens: That was not even in high school. I was much younger. So I did that for a few years; two or three years.
Matthies: So you could describe fully what your vocal folds look like?
Stevens: I had no idea what they looked like.
Matthies: Was there anybody in your undergraduate program, either a fellow student or a professor who you particularly remember as having an influence on you?
Stevens: Yes. There certainly were. I was particularly drawn to the mathematics people, the people who taught mathematics. Then later on, we are talking college, perhaps in my senior year or I did a masterís degree there too, there were a couple of people in the department of applied mathematics who were very good. One of them actually had collaborated with Einstein. He was teaching there. And in fact, it was one of those professors who suggested to me that I might want to go to MIT and I had never even thought of it until this suggestion. So in a sense, that was a major influence on me.
Matthies: That set you down a path.
Stevens: Yes, right.
Matthies: Do you remember the public figures of the time, scientists or sports heroes?
Stevens: As kids, either in high school or early college, we certainly remembered the big football players. In fact in high school — as I said it was a school of education — we would occasionally have student teachers. One of our teachers was a big football player. He was on the football team and so we were really impressed by him.
Matthies: I bet you behaved.
Stevens: His name was Barry Gray. What was the question?
Matthies: Sports heroes. But this was during World War II.
Stevens: There was not so much of that. We were very much concerned with the war, so we certainly knew what was going on. My family, being from London originally, was very closely interested in what was happening.
Matthies: And worried about it being destroyed.
Stevens: Absolutely. Yes. They would tune into the BBC every day.
Matthies: To hear what had been bombed?
Stevens: Yes, right.
Matthies: Did you ever participate in a rally or a protest as an undergraduate?
Stevens: No. I think at that time during the war, I do not recall there being any protests. I think everybody in Canada was very united. During the summers, when I was in college, I would work for defense industries. I worked one summer for a company that was developing radar, which was all very new at the time.
Matthies: That was a great summer job for an engineer.
Stevens: Yes, it was really good.
Matthies: Do you remember any special rationing or things that you went without during that time period?
Stevens: We were short of gasoline. I think we did not drive a car that much. I forget even if we had food rationing. I do not think we did. We did all right. My brother was away for the entire War in England. So we always watched for his letters that came.
Matthies: You were glad when he came back then.
Stevens: Oh, yes.
Matthies: Did you watch every game of the Stanley Cup when you were a kid? Is that a requirement in Canada?
Stevens: Yes, yes. Well, in Canada at the time when I was a kid and growing up, there was always a hockey game on Saturday night. We did not have television but it was broadcast on radio by a guy named Foster Hewitt.
Stevens: So he would describe in great detail these hockey games and I would always listen. I think once, maybe twice, I actually got to go to a game.
Matthies: The action in hockey is so fast that Foster Hewitt was amazing.
Stevens: Right. Do you know Foster Hewitt? You have heard of him?
Matthies: Yes. I am from Buffalo, New York which is very close to Canada.
Stevens: Was he still broadcasting when you were young?
Matthies: I am not sure if I ever heard his games or just knew he was legendary.
Matthies: Looking back, would you go to the same undergraduate school? It sounds as if you would have.
Matthies: You received a good education.
Matthies: You were not tempted to study sociology or anthropology?
Stevens: Well, the one thing I regret is that at that time in engineering, they did not, I think, recognize the value of augmenting your education with studies in the humanities.
Matthies: So you did not have liberal arts anywhere?
Stevens: No. At MIT, the students must take an average of one subject. I never did that. In fact, in my application to MIT, they were questioning it a little bit because I had never had any humanities at that time. Before I came to MIT, I had to study up some economics.
Matthies: So you barely squeaked into MIT?
Stevens: I guess you could say that, right. I really am somewhat uneducated in that sense.
Matthies: You were fortunate then, that you had an excellent high school education because if you had had a narrow high school education you would have been in more trouble.
Stevens: Yes. Although I have to admit, in high school, I actually hated Shakespeare.
Matthies: What made you change your mind?
Stevens: Well, one reason I did not like it was that in high school, we would read a few lines and then the teacher would say, ďKen Stevens, can you explain that line?Ē I really could not explain it and so felt a little embarrassed about that. I guess later on, when you are not under that pressure and you do not necessarily have to understand every single line, you can start enjoying it.
Matthies: Did you get a masterís degree or did you go straight for a doctorate when you got to MIT?
Stevens: I did a Masterís degree at Toronto.
Matthies: In Toronto, okay.
Stevens: I finished in 1945, which was when the Second World War finished too. And then I stayed on because they needed young people to be teaching assistants because of all of the people coming back from the war. There was the Canadian version of the GI Bill and so I was involved in that for about three years.
Matthies: What did you teach?
Stevens: It was in various aspects of engineering including engineering drawing and things like that. So my brother was one of the students.
Matthies: That must have been strange.
Matthies: That was your first teaching experience?
Stevens: No, it was not the first actually. While I was doing that, I also did a masterís degree. My first teaching experience was when I was a senior at college where they sometimes recruited some of the seniors to help teach elementary physics to a group of home economics students. You could imagine for the young men in our group that was a plum.
Matthies: That was a good assignment.
Stevens: Yes. Because at that time, of course, we were all men and home economics, which isnít taught quite that way, was all women. That was my first teaching experience.
Matthies: Then you had to teach the veterans of the War.
Stevens: Thatís right. The big guys.
Matthies: Did you enjoy teaching right from the beginning?
Stevens: Yes, I did. I learned at that time that when you teach something, you really learn it and realize how much you did not know about it the first time around.
Matthies: Occasionally, you will get students from whom you learn things too.
Stevens: Yes, absolutely. Or they can ask questions and make you look at something in a way you had not before.
Matthies: Do you remember what your masterís thesis title was?
Stevens: When I did in my masterís, control theory was a big thing. So I did a masterís thesis on what we called servomechanisms at the time. I do not think that it was particularly profound but I got some control system to work. I do not even actually remember who the advisor was.
Matthies: You would have to look that up?
Stevens: Yes. I probably have the thesis around here somewhere.
Matthies: You had a committee, and a topic, and you did the work.
Stevens: For the masterís degree, just one advisor. Actually, and this is rather obvious, as soon as I went to college in engineering, I dropped the idea of being a doctor. Around that time, when I was doing my masterís degree, which is when I also took these subjects in applied math, is when my professor suggested I come to MIT. At that time, I was thinking of control theory, and that was what I might do at MIT. But then I met Beranek and I changed my mind a little bit.
Matthies: Did you take any premed classes in your first year of undergraduate work? Did you just know that you wanted a math and engineering-based curriculum?
Stevens: I guess I just knew I wanted math and engineering. I dropped completely any interest in biomedical things at that time.
Matthies: Were there any other projects that you worked on in getting your masterís degree?
Stevens: I remember also while I was getting my masterís that I worked for a couple of summers at a place called the Ontario Research Foundation. They were developing new metals that would withstand heat and could be used in airplanes. Metals like titanium. So I worked in a group that was studying the properties of these metals, doing a little bit on the theory of it.
Matthies: You had some great summer jobs.
Stevens: That was good too.
Matthies: Working on radar.
Stevens: Although that particular job was mainly doing some lab work. They suggested I try to read up on the theory but it was a little beyond my comprehension at the time.
Matthies: Do you recall any particular faculty or fellow students from your masterís degree work?
Stevens: No. Well, I should have said those people in applied math. That was during my masterís thesis. So those, I did remember them. No, I didnít since I was doing the teaching and so on. I was not involved with other students.
Matthies: I think you had plenty to do.
Matthies: Did you keep up with your bridge playing?
Matthies: So you applied to MIT while you were finishing up your thesis?
Stevens: I guess while I was finishing up my masterís thesis. Thatís right. It was not as an undergraduate.
Matthies: Did you apply anywhere else for doctoral work?
Matthies: Just MIT?
Stevens: Just MIT. I remember when the thick enveloped came.
Matthies: It was a thick envelope.
Matthies: How did your parents feel about you going, not only to the United States, but all the way to the east coast?
Stevens: I think they felt okay about it. But it was interesting, I had lived such a sheltered life because even up through then I had lived at home. And this was, in fact, going to be my first time away from home, and here I was into my twenties.
Matthies: And did you live in a dorm when you got there?
Stevens: Yes. I lived in what is now Ashdown House or graduate dorm or whatever. I shared a three-person suite there.
Matthies: That was good that you didnít have to try to find an apartment in this area.
Stevens: Right. Yes. It worked out okay.
Matthies: What was Cambridge like when you first got here? Was it as busy as it is now?
Stevens: Busy as it is now. I mean, I remember that there were trolleys that would go down Mass Avenue from Harvard Square across the bridge and over to Dudley. Kendall Square, there was nothing there. The place has changed enormously since then.
Matthies: Did you start working with a particular person immediately or were you just generally admitted?
Stevens: What actually happened was I was admitted and Beranek — who was a fairly new professor at the time and just starting the Acoustics Lab — needed a teaching assistant in his course in acoustics. Basically, the same course that Amar Bose is now teaching. Beranek noticed that in my vita, I had once taken a course in acoustics. So he got in touch with me and asked if I would be a teaching assistant for this subject. And I said, okay. Then pretty soon after that, he said, ďWe have a new project in speech here at the lab. Do you want to work with it?Ē And I said, ďSure. Why not?Ē So my transition to acoustics was very much by chance.
Matthies: Was Beranek at the famous Building 20 at the time?
Stevens: Sure. He was in Building 20 right over by where theyíre digging those holes right now.
Matthies: Who else was at Building 20 during those days?
Stevens: Of course, Building 20 was the Radar Lab originally. Then after the end of the War, it became the Research Laboratory of Electronics.
Matthies: So you were here for the beginning of RLE?
Stevens: No. I think RLE began probably in 1945 or 1946. I was here two years later. So there were people like Jerry Weisner and Bob Funnell there.
Matthies: When did Chomsky arrive?
Stevens: Chomsky was more like 1955, I think. Maybe a little before that.
Matthies: How did linguistics get into Building 20?
Stevens: Well, letís see. The story is that the Acoustics Lab had this little project on speech intelligibility, it was really not a very basic project. Then Gunnar Fant came and spent a couple of years at the Acoustics Lab. I think probably it was arranged by Beranek. And so, I had an office next to Fantís. At that time, he became a great influence on me. He was working on things that became ultimately his doctoral thesis.
And then in Building 20, Halle came, probably around 1951 or so, and started working on some phonetics projects in Building 20. And sometime over the next few years, a little circle of people who were interested in speech was formed. Jakobson was one of the people, and Halle, and a guy by the name of Bill Locke, who was head of the modern languages department — And some other people. They had a couple of conferences there; I got to know some of the people. All that stimulated my own interested in speech communication.
Matthies: Do you recall Liberman at that time?
Stevens: It was basically what we called DLC: Delattre, Liberman, and Cooper. By that time, weíre now up to somewhere around 1955. Those were the main players down there. Maybe Leigh Lisker and John Lotts were involved too.
Matthies: Was that a long trip down to Haskins and back?
Stevens: Yes. Well, letís see. Haskins at that time was on 42nd Street, not too far from where the U.N. is. They had a loft, basically, where they were doing their research. No, you would go down by train and it wasnít so bad.
Matthies: Did you interact with anybody from Harvard at the time?
Stevens: Yes. Now that you mention it, that was another influence that I had. Maybe I was still a graduate student. I think I was. So it must have been around 1951, 1952. They had a series of lectures. It was a course up at Harvard at the psychoacoustics lab. A series of individuals who were still there gave lectures. Those individuals included Smitty Stevens, Von Bťkťsy, Ira Hirsh, Licklider, Walter Rosenblith.
Matthies: That was a seminar.
Stevens: That was really great. That was fantastic. Eddie Newman. So that was also a great influence on me. Not so much from the acoustics, the speech production point of view but the auditory and speech perception side. That was a great seminar.
Matthies: They were doing a lot of significant work in early speech intelligibility.
Stevens: Yes. Absolutely.
Matthies: Do you remember the title of your doctoral dissertation?
Stevens: It was something like the Perception of Sounds Shaped by Resonance Circuits. It was sounds that had peaks in the spectrum, basically.
Matthies: Did you have a committee at the time?
Stevens: Yes. It was Beranek and Licklider and Rosenblith, I think.
Matthies: When did Bolt, Beranek, and Newman get formed?
Stevens: It was formed as sort of a loose partnership, I guess, at the time that the U.N. building was going up, which was around 1948. So Bolt and Beranek got together and they advised on the acoustics of the meeting rooms and basically formed this partnership. Then it became more formalized and they brought in Bob Newman and probably around, I donít know, 1950 or so. They first worked just out of the Acoustics Lab in a sense but then they moved up into Harvard Square. When I got my degree in 1952, I wanted to continue some research in speech and so, somehow I was able to get a half-time MIT job as a research scientist and half-time working for BB&N. I did that for a few years.
Matthies: How did you like that?
Stevens: It was fine. I did a lot of late night work and so on. I wasnít married at the time. At BB&N, I got exposed to some other aspects of acoustics. We did work in noise control and did some traveling at that time. Jet engines were coming in and there was a lot of discussion of how to quiet them, and quiet the testing of those engines, and noise around airports, projects like that. Noise on aircraft carriers. I did some of that.
Matthies: When did you write your first grant proposal?
Stevens: It was probably somewhere in there between 1952 and 1955. We were supported by the Air Force. At that time, I donít think it required much of a proposal. So they supported some of our basic work early on. And then, we turned to NIH for the funding. Sometime later, around 1960 to 1965, we started getting NIH support and ultimately dropped the Air Force support.
Matthies: Speaking of the Air Force, there is a section here on military. You have limited vision in one eye, correct? So you were not in the military.
Matthies: The technical business or trade school. Does that ring a bell for you?
Stevens: You mean in terms of myÖ
Matthies: Or correspondence courses?
Stevens: No. When I was actually in college, I taught some electronics at a trade school in evenings. I remember one year doing that.
Matthies: So after college, you were describing that you were a part time research associate at MIT and working for Bolt, Beranek, and Newman. When did you first come on the faculty at MIT?
Stevens: June 1955.
Matthies: As an Assistant Professor?
Stevens: As an Assistant Professor, right.
Matthies: What was the interview process like for that job?
Stevens: It seemed like it just happened. It wasnít as though there was a search committee and you were one of fifty applicants or anything like that. I think they were in need of faculty at the time and I frankly donít remember the details.
Matthies: You donít even remember applying for the job?
Stevens: Well, I guess I remember a conversation with Beranek and he said that he was considering going full time with Bolt, Beranek, and Newman and dropping the faculty appointment. I expressed an interest in the teaching and he said, ďWhy donít we get you on the faculty?Ē
Matthies: That was great.
Stevens: So I dropped the half-time thing. I actually donít remember what the process was.
Matthies: It sounds a little different from the bureaucracy that goes on today.
Stevens: Iíll say.
Matthies: What were your initial projects when you began as an assistant professor?
Stevens: It may sound strange but like what I am doing now. It was the early days of speech synthesis. Partly as a result of Fantís work, one project was an electrical analog with a vocal track. Studying some elements of speech production and doing some speech synthesis with formant synthesizer. Maybe some speech intelligibility work. I have forgotten a lot of it.
Matthies: When did you begin the X-ray project with Joe Perkell.
Stevens: I went on the faculty in 1955 and at that time as a young guy, I was happy to take a sabbatical after seven years. So in 1962, went to Sweden to work with Fant. One of the things that I did while I was there was make this X-ray picture along with Spinnermann. The two of us worked on getting some X-ray pictures. Spinnermann published papers on co-articulation from the work that he did.
They had a facility there in Stockholm that actually had been setup by a guy by the name of Hank Truby, who had started to make some X-ray pictures of this kind. He ultimately went back to the United States. He had been in Sweden for a while. So we sort of got that working again and made these films.
It was probably when I came back a year later; a young man who was beginning dental school needed a part time job, named Joe Perkell. He had learned a little bit of anatomy, of course. So I said, ďHereís this film, if youíre interested. You could do some analyzing and tracing of these thingsĒ. So he did and got kind of interested in it. And of course, ended up publishing that little book. That was a long answer to your question.
Matthies: And his experience with that initial work in speech. He went on to complete his dental degree but then he came here?
Stevens: Yes. He came here.
Matthies: Do you remember when that was?
Stevens: He got his Ph. D. I donít remember exactly when it was. It was probably in the middle 1960s I would think. So he worked here and did his Ph.D. here. I canít remember when he got it. Continued, as you know, on the staff. During the first few years of that, he did continue to do some dental work. I think a day a week he practiced.
Matthies: Yes, up in Dartmouth.
Stevens: Right. Even before that, I think he practiced around here a little bit too. He would commute from Dartmouth.
Matthies: When did Dennis come to MIT?
Stevens: Iíd have to look that up. It was in the 1960s too but I canít remember exactly when. Iím thinking of when his publications were. I canít think of any of his publications earlier than the 1970s, so probably around the late 1960s.
Matthies: Was Klatt drawn to the speech synthesis problem right away? Or did he start out interested in other things?
Stevens: I think he started out interested in other things because his doctoral thesis was on audition, auditory modeling, I believe. So I guess, he saw some of the things we were attempting here, and became interested, and in doing the research that provided enough knowledge to do the synthesis right. Yes. He was a very energetic guy. Driven. He had a great influence on us here.
Matthies: He also seemed to me to be someone who embraced computer techniques and programming.
Matthies: Any aspect of technology.
Stevens: Particularly the computer stuff. That is, he had the basic understanding of the speech process and speech production process. And he knew the mathematics behind it and he could implement it with great efficiency on the computer.
Matthies: When I first got here about twelve years ago, there was a machine called the Klatt Machine. It primarily had a lot of the synthesis programming, Klatt Talk.
Stevens: Itís still on there. You know that?
Stevens: In fact, Laura Dilley, one of our students, is trying to use it.
Matthies: Was Klatt Talk first and then KLSYN, the analysis tool, did that follow?
Stevens: Klatt Talk. Well, at some point, Dennis got involved with Digital.
Matthies: Digital Equipment?
Stevens: The Digital Equipment Corporation. There were a couple of people out there who had an interest in speech synthesis: Ed Brockhurst [?] and Tony Vitale. So Dennis started working with them to develop what ultimately became DEC Talk. But in the meantime, he developed something called Klatt Talk here. The history in my mind is a little blurred because there was also this project of John Allenís on speech synthesis which ended up in a book and Dennis worked on that too. Dennis was a guy who never bothered with the niceties of, you know, whose property is this and so on. What does Digital own? What does MIT own? He just did it.
Matthies: Committed to the science.
Stevens: Very refreshing. And so it all got done. It clearly was a real breakthrough doing that synthesis.
Matthies: Itís work that lives on today.
Matthies: We were talking about other students or faculty who had a great influence on you in your early years. You were an Assistant Professor and then you became an Associate Professor. When?
Stevens: It was probably around 1960. I canít remember.
Matthies: Did that give you tenure?
Stevens: To tell you the truth, I donít remember. But I think tenure must have come a little bit later than that.
Matthies: So you were promoted and then tenured. When did you become a full professor here?
Stevens: I donít know that either. I think I said that I became the Lebel Professor in 1965 but Iím not even sure of that date. So I must have been promoted to full professor before that. I donít know the dates.
Matthies: Did you ever consider going anywhere else than MIT? Or have you always been happy here?
Stevens: During the course of that career here, there were a couple of cases where there were little nibbles which, at least, I looked at. One was at Haskins Labs I think at a time when Frank Cooper might have been considering retiring. Another one was at Central Institute for the Deaf which ended up either I wasnít interested or they werenít. But I certainly was not interested in a job which would have been an administrative job.
Matthies: You wanted to stay in the lab?
Stevens: Right. So that involved maybe one trip and discussion. It didnít amount to anything.
Matthies: You had a sabbatical in Sweden working with Gunnar Fant.
Matthies: Have you had any other sabbaticals?
Stevens: Mainly just one other, another seven years later where I went to London and worked at University College in London. And that sabbatical was not so much doing experiments as doing some writing. That was just the beginning of a thought that maybe at some point I would write a book. But as you can see, that was in 1969 and it took a while. So I did some reading and writing and talking to people. It was in the Phonetics Department.
Matthies: Do you recall who was there at the time?
Stevens: Dennis Fry was there and also the fellow who has the laryngeal graph.
Matthies: We can write it down when it comes to us. I canít think of it either.
Stevens: But they shouldnít transcribe this stuttering, ďCanít remember the name.Ē
Matthies: Theyíre going to give us a typed copy of the transcript that we can add in things and correct. Weíll get a second chance.
Stevens: That name will come back to me.
Matthies: There are other things we have to look up too.
Stevens: There was another sabbatical which I stayed around here which was really not a sabbatical at all.
Matthies: A change of pace. You have taken advantage of a number of traveling opportunities. Youíve lived in Sweden. Youíve lived in London. And then youíve been all over the world at conferences.
Stevens: Yes. And one has to say that that is certainly a positive aspect of this kind of work. The traveling and the meeting of people all around the world. Itís just a wonderful thing.
Matthies: What are some of your favorite places that youíve visited?
Stevens: Well, of course, Sweden is. It was certainly going to France and a conference in Aix-en-Provence. That was nice. Much more recently, I went to New Delhi and then also to Bombay. When you say favorite place, in a sense ďfavoriteĒ because it gave me an exposure to a completely different culture. And of course, Beijing recently was real interesting to go to.
Matthies: MIT has had some wonderful people come here from all over the world. We had someone from South Africa. Tony?
Matthies: Weíll have to look that up too. But he was a lot of fun.
Stevens: He was. Yes.
Matthies: Tony Trail.
Stevens: Trail. He was a lot of fun to work with, and also we learned something about the clicks too.
Matthies: Yes. The acoustic-genetics of a completely different category of speech.
Stevens: Over the years, earlier than that, there was a period when we had this sequence of Japanese people which you heard about. You were at the banquet.
Matthies: Right. And you did have a visiting scientist from India.
Stevens: Yes. Rajesh Farma [?] or something. Weíve had people from China, from Russia.
Matthies: And Ian Maddieson, who has been here.
Stevens: Right. I donít know if he ever stayed very long.
Matthies: He used to come here for the marathons.
Stevens: Thatís right.
Matthies: He would give us a seminar.
Stevens: At the time of the marathon. Thatís right.
Matthies: How about Ladefoged. Does he ever visit us?
Stevens: I think he has on occasion. My memory of Ladefoged was when he first came to this country and took a faculty position at UCLA. On his way there, I believe he stayed with my wife and me.
Matthies: When did Stefanie Shattuck-Hufnagel come to MIT?
Stevens: Golly, sheís been here a good twenty years, I guess. So whenever that was, around 1980 or so. Maybe a little before that. For a while, she was on the faculty at Cornell and she used to commute. Then she decided to settle down here and we were happy to have her work more in our group.
Matthies: So the current group is you, Stefanie Shattuck-Hufnagel, and Joe Perkell, which is quite a productive and congenial core of folks.
Stevens: Yes. Covering a range of interests which is nice. We shouldnít forget people like you.
Matthies: Youíve had some outstanding students recently. Abeer Alwan comes to mind immediately. What are some of the accomplishments of your doctoral students?
Stevens: Over the years, certainly my first doctoral student was Jim Flanagan. So certainly his accomplishments are well documented. You know, with people like that or Abeer, itís easy to say, ďOh, yes, he or she, my student.Ē But when theyíre that good and come to MIT, you really end up not having that much influence.
Matthies: You guide them more than show them the way.
Stevens: Yes. Just provide them with whatever they need and theyíll do it.
Matthies: And get out of the way.
Stevens: And then later on you can say, ďOh, yes. He was my student.Ē Larry Rabiner is one and Shinji Maeda. And of course his wife, Jacqueline was a post doc here. Victor Zue, Stephanie Senneff were my students. And Mark Johnson. So one can be happy to have such good students.
Matthies: Thatís quite an accomplishment. Well, your students have helped you in the lab with their ideas and their work. And one of the major accomplishments of your lab work has been that incredible book, the acoustics-phonetics book. Do you want to tell us a little bit of how that came about? That is just an amazing book.
Stevens: Well I guess it came about because I was teaching this course in speech communication. And it also came about because, well, I think early on, the influence of Arthur House was very strong. He sort of convinced me of the importance of understanding speech production. And then, through Dennis, partly his intellectual influence, but also all this stuff we have on the computers with all the Klatt Tools and so on, it would not have been impossible to do this book without that stuff.
Matthies: So you were able to really relate circuits, and modeling, and analysis, and perceptual mechanisms all together.
Stevens: Right. And on top of that was the influence of the linguistics. And that, you know, this isnít just a machine that generates sound. There is something up in the head that starts the whole thing going.
Matthies: You have to have meaning.
Stevens: Yes. But even below that, at the phonological level. Morris and Jay have been an enormous influence.
Matthies: Jay Keyser?
Matthies: And you had mentioned the book that you had written with Amar Bose. Have you other books that I donít know about?
Stevens: No. Only I edited a book on the vocal folds, the physiology. But no others.
Matthies: What are you current book plans? Is acoustics-phonetics ready for a second edition? Do you have another book in the pipeline? Or is that enough?
Stevens: No. Right now, you know, if there were to be anything of that kind, it would have to be either, well, Iím relating it to the lexical access work that we do. The two main streams, or maybe three, are: How do you extract from the speech signal the information that gets at the words? And then the possible application of this stuff to clinical problems. And then the third stream, really is really sort of still cleaning up the acoustic-phonetics stuff. Thereís still a lot of work there.
Matthies: There are still things that we donít know. Puzzles left to be solved.
Stevens: Yes. Certainly.
Matthies: Now your current lexical access work is kind of a radical departure from the way most people are looking at speech recognition. Youíre not pounding on it with as big a computer and a statistical approach.
Matthies: Give a brief outline of how your approach differs from the mainstream.
Stevens: Well, of course, itís mainly that itís not so much data driven. Itís basically what weíve learned about how words are represented in memory in terms of words, phonemes, and features. And these features are often fairly directly related to the acoustic signal but often in a more abstract way. But if you look at the acoustic signal, hopefully in the right way, you can get at this abstraction. And that the way to look at this abstraction, in the sense that a word, the features of a particular phonemic unit might be buried there and not obvious from the signal. But by making the right kind of acoustic measurement that will get at the articulation, closer to the articulation, will enable you to uncover these abstract units. That kind of thing.
Matthies: And thatís going to be much more likely to translate into clinical problems and disordered speech, than a data-driven approach.
Stevens: Thatís the hope. Yes. So the data-driven approach, basically, looks at things on the surface and tries to relate that to surface acoustic things. So what that means is that any given word might have ten different pronunciations and you might have fifty or sixty different phonetic units or phones, as they call them. Just so that you can relate these surface representations of the acoustics.
Matthies: Thatís exciting. I hope that continues to go well for you. Letís talk a little bit about your family. Youíre currently married to Sharon Manuel. What is your spouseís occupation at this time?
Stevens: She is an Associate Professor in the Communicative Disorders Department at Northeastern University.
Matthies: So, two faculty members in one family.
Stevens: Thatís right.
Matthies: You have a number of children and a couple of grandchildren. Can you tell us a little about them?
Stevens: My first marriage was to Phyllis Fletcher at the time. We were married in 1957. And we had four children; two girls, two boys who are now late thirties, forties. One of them is married, the son, Michael and he lives near Baltimore and has two children. So those are my grandchildren, two boys. Then with Sharon, we have this adopted child from China who is now a little over two years old.
Matthies: The ďmy, myĒ time.
Stevens: Obviously ďNOĒ is her most frequent word.
Matthies: That is a big difference. The role of a dad and parenting in this time compared to back in the 1950s and 1960s.
Stevens: Yes. At the time my kids were growing up, Phyllis, my first wife, was not working. The parenting duties and so on were distributed in a different way to some extent. Not so much the case now.
Matthies: Can you tell us what your children do for a living?
Stevens: My oldest is a lawyer and she is working for the government in the General Services Administration.
Matthies: Is that Becky?
Stevens: Becky. GSA has lots of real estate. Real estate is one of her specialties in law. And the next one is a girl, is Andrea, and she is working. She has her Ph. D. in environmental biology and she is working for the Nature Conservancy in the state of New Jersey.
Matthies: Boy, thatís a tough place to be working.
Stevens: Yes. Although surprisingly, the state of New Jersey has a lot of countryside. You donít see it as youíre driving along I-95 there. So sheís enjoying her job.
Matthies: And Michael.
Stevens: Michael has his Ph.D. in earth and planetary sciences and is working for the Naval Research Laboratory studying atmospheric trends around the earth as collected from satellites. He has two children who are about the same age as my little one. In fact, weíre going down there in a few weeks to have a little reunion.
Matthies: Thatíll be fun if not exhausting.
Stevens: Yes. And John lives in this area and works for the Reebok Company.
Matthies: Now heís a runner, right?
Stevens: He is a runner. Yes. In fact, all of my kids do running.
Matthies: And some of them have done some marathons, right?
Stevens: John has and Michael has and Andrea.
Matthies: Have you ever run the Boston Marathon?
Stevens: I have not. 10K is the most Iíve ever done. So, John seems to be enjoying this. Heís very much into sports and so working for Reebok puts him in touch with, at least, the equipment.
Matthies: I bet he always has a great pair of shoes.
Stevens: Yes. And in fact, our little daughter, Kendra, has used Reeboks.
Matthies: I think theyíre called Weeboks for the little ones.
Stevens: They are.
Matthies: Do you know what Kendraís going to be when she grows up?
Stevens: She hasnít expressed any interest yet.
Matthies: Your special hobbies when youíre not working. Your work is your passion and thatís very clear. But I know that you sometimes like to get away from Cambridge and head up to Maine. What do you do up there?
Stevens: Well, there are a lot of things to do up there. So part of it is in the wintertime, you can ski and in the summertime you can swim. But there are always odd jobs to do like making a path through the woods or doing repairs on the house at night. Or building some furniture that we might need up there. So I like to do that kind of thing.
Matthies: Enjoying the peace and quiet.
Matthies: Very different from Cambridge.
Stevens: Yes. Nowadays at least, during the winter you go up and youíre spending two or three hours shoveling snow so you can actually get to the house.
Matthies: Besides working, do you have another favorite form of entertainment? Playing with the toddler?
Stevens: Playing with the toddler. Before the toddler came along, I would play the piano a little bit. And now I do but itís mainly ďItsy-Bitsy SpiderĒ and stuff like that with accompaniment. Lately, I have sort of cut back on those kind of other outside activities.
Matthies: I suspect you have a whole new appreciation for language development and phonological development.
Stevens: Yes, itís really interesting because Kendra will come up with something and Sharon and I will look at each other and say, ďWhere did she get that?Ē Like she sees something in a movie and she says, ďOh, thatís personís sad.Ē Where did she get that sad concept? Where did she get that?
Matthies: A two-year-old is such a little sponge. Even though she is mostly using ďnoĒ, itís probably growing rapidly.
Matthies: Do you get much of a chance to read or go to the movies?
Stevens: No. Not to the movies. And as I say, we used to go regularly to plays, I had season tickets, but weíve sort of cut back on that for the last year or so. Thereís not much I can say. I occasionally find time to read books but usually theyíre fast reads, something I can read on an airplane. From time to time, in the past, I would read books on history or particular biographies and so on. But I havenít been doing that lately.
Matthies: And what do you think the chances are that the Red Sox will win the World Series?
Stevens: I think itís going to be a few years. Maybe the city needs to be a little more behind them and build them a better stadium and so on. But I have been to see them on a few occasions, not lately, as I say. Maybe sometime weíll take Kendra.
Matthies: That would be fun.
Stevens: Yes, right.
Matthies: How about the Boston Bruins? Do you watch hockey anymore?
Stevens: I donít. As a kid I was very much interested, but I donít find it so interesting anymore.
Matthies: The game has changed a lot since Rocket Richard.
Stevens: Yes. So, no. I wouldnít watch.
Matthies: Bobby Hull is long gone. How about music or art? Are you a classical music fan?
Stevens: Iím certainly a classical music fan. Yes.
Matthies: Do you prefer Baroque and harpsichord?
Stevens: Iím fine up through Beethoven, but to get a little farther into Brahms and so on, itís okay.
Matthies: Too new-fangled.
Matthies: So Charles Ives is out.
Stevens: Actually, thatís right. On the other hand, Kendra likes, say, the music from the movie Tarzan and I think thatís not so bad. Itís pretty good.
Matthies: Some of our best composers are working for Hollywood these days. Do you have anything else you would like to add? Or do you have a special quote or inspiration you would like to leave us with?
Stevens: I think we dwelt a lot on the earlier days and the influences of people like Arthur House and Dennis Klatt and Gunnar Fant and Morris Halle. But I didnít get a chance to point out the influence of people like Joe Perkell and Stefanie Shattuck Hufnagel who bring these different points of view to the field, not to mention Melanie Matthies. So that in these past few years and along with, of course, continuing with Jay Keyser and Morris has continued to keep me stimulated and have fun.
Matthies: So you have no plans to retire. You have no plans to do anything other than what youíve currently been doing?
Stevens: I think at the right point in time I could cut back a little bit on the teaching and the mentoring of students. I think it would be better if they were mentored by younger people, once we get those people in here. But at this moment, I donít have any plans.
Matthies: You currently have several graduate students in process?
Stevens: Yes. I try to keep count but sometimes itís a little hard to know which are my graduate students. But there are probably about five or six students that I would list as my students.
Matthies: And last spring and summer, you had three doctoral students graduate, right?
Stevens: Yes. I donít know how many doctoral students Iíve had but I figure itís about one a year since Iíve been on the faculty, so weíre talking forty. But Iíve never actually tried to count them.
Matthies: Well, if the Acoustical Society wants a list, we can ask Elaine to help us to compile one because she would be the one who could access all that information.
Stevens: Yes. In terms of the travel, going back to that, there was a period — I think it was actually between my two marriages — where I would go to Portugal once a year, usually at IAP for a week or two and work with them on some phonetics work. That was great fun.
Matthies: We were starting to wrap up and we got diverted on to some of the contemporary work that you are doing. I thought we would just finish up with that because thereís certainly some very exciting projects that youíre working on now. The speech synthesis has certainly evolved over the years. One of the key people I remember working on the project was a gentleman from India whom we called Anantha [Padmanaba]. Do you recall what he was doing here?
Stevens: Anantha was certainly a very knowledgeable person about speech because he had already worked with Gunnar Fant in Sweden for a year or two. I donít recall him doing work on speech synthesis. He was interested in properties of the glottal source and in doing speech analysis. And so, he was here for a while.
Matthies: Did he add to the synthesizer?
Stevens: He added not to the synthesizer but to the Klatt Tools. Included in the Klatt Tools that Dennis developed was a synthesizer and a set of analysis tools. He was working on the analysis. He developed some new averaging techniques, which have been very useful in doing the analysis.
Matthies: So he was primarily interested in the voice as a source?
Stevens: Yes, he was. Right.
Matthies: One of your students, Helen Hanson, has done excellent work in that area.
Stevens: Right. Yes. She is really, I would say, done some definitive work on what are the properties of the source as measured from an acoustic point of view for a wide range of male and female speakers.
Matthies: Now she currently works for Sensimetrics?
Matthies: Do you consult occasionally?
Stevens: I consult for them occasionally, right. Thatís a little company that does research often supported by the government under small business grants in various aspects of speech and hearing, including speech synthesis.
Matthies: And then another project at Sensimetrics is HLSyn and you worked on that with Corine Bickley?
Stevens: Originally with Corine Bickley, yes.
Matthies: Can you tell us a little about what that means? What is HLSyn?
Stevens: It means Higher Level Synthesis and it basically takes the kind of synthesizer, starting out with Gunnar Fantís original synthesis idea, and was expanded it by Dennis Klatt. Where Dennisí synthesizer involves a large number of parameters that control some sources and filters to synthesize speech. HL synthesis tries to take all those parameters and recognize that there are constraints between them. Constraints because of the human articulatory constraints and reduce that large number of say, 40-odd parameters, to a number like 10 or 15 parameters, which are more related to articulation than directly to acoustic properties. Some work has been done at Sensimetrics to get that synthesizer to produce speech from rules and now thereís a company which was a spin-off from Digital Equipment Corporation that has become interested in using HLSyn to basically upgrade the DEC Talk Synthesizer.
Matthies: Thatís exciting. Are there other things going on at Sensimetrics we should know about?
Stevens: Helen Hanson at Sensimetrics is working on things related to this HL synthesis but looking at the prosodic aspects. And looking at them from the point of view of what happens is happening to the articulatory system rather than just thinking of the prosodics as some sort of acoustic data, thinking about where the prosodics come from as far as the production and the physiology is concerned.
Matthies: That is a natural evolution of her doctoral work.
Stevens: Yes, it is.
Matthies: There are a couple of papers of yours that had a profound influence on me and I was wondering if you could talk a little bit about that work. One is the quantal theory and the other is the invariance work that you do with Sheila Blumstein.
Stevens: Talk about invariance work as Sheila and I — probably back in the late 1960s, early 1970s — had this view that we thought might be valid that for any one phonetic feature, there was an invariant acoustic correlate. So that would mean, for example, for a place of articulation, letís say an alveolar consonant there would be some particular acoustic property that remained invariant, independent of speaker, independent of context and so on.
And so we worked on that together for a while, and she would come up once a week or occasionally I would go down to Providence where she was teaching at Brown University. And showed that there was some truth to this view although it was mainly for consonants in initial and CV position and so on. I donít know whether Sheila — how strongly she believes it, but I think there was something to that theory. I think what we are coming to recognize is that there are many acoustic cues for a particular phonetic distinction but that perhaps one of these cues or one property is kind of a defining property. Itís the property that makes this particular feature distinct from another feature. And then there are other properties that might enhance or add to that distinction.
So that was a big influence too, working with Sheila who is a linguist. She actually originally came to the Lab soon after she got her degree, I guess. And she had worked with Roman Jakobson for her thesis and Jakobson advised her to maybe come and work a little bit, collaborate with us a little bit. That worked out well.
The other thing, the quantal theory, was where I had noticed, and Fant as well, that the acoustic properties associated with different phonetic units are often not arrayed along a continuous scale. That when you move the articulators around — that it often happens that as you examine the sounds that are made as you continuously move these articulators — that those sounds go through abrupt changes. And that there are regions of the articulation where you get a certain acoustic property and then you suddenly jump to some other region where you get a different acoustic property. And that would suggest that even though the articulatory system that would be used to produce speech seems to be kind of an analog continuously varying system, the sounds that come out of that system have some quantal properties, some discrete properties. And that maybe those discrete properties have something to do with the discrete representation of an utterance that we seem to have in our head and that the linguists postulate on the basis of their data.
Matthies: Thatís a very unifying theory then.
Stevens: Attempting to do that. So the story is still not complete.
Matthies: Does that have implications for your synthesis work or is that still too remote?
Stevens: I would say it does. Yes. One of the things that come out of this quantal theory is a very basic thing and that is when you make a consonant that creates a discontinuity in the signal. That is acoustically what a consonant is all about. When you make a vowel that creates some sort of peak in the signal. And that in doing speech synthesis, one could imagine laying out a set of landmarks as to when these consonant closures and releases occur, and when the vowels occur. So it can influence the timing and the general way one thinks of the synthesis.
Matthies: Your interest in clinical populations is long standing. I remember work youíve done with speech of the deaf. But more recently it seems like dysarthria has captured your interest.
Stevens: Yes. It has. Thatís partly because we had this student, Hwa-ping Chang, who got his degree in 1995 and he was interested in — as a student in mechanical engineering actually a Ph. D. student — developing ways that could help speakers who have dysarthria, help them communicate with others. So maybe there were sounds that they made that were made more consistently than other sounds. Maybe they could be, first of all, taught to or persuaded to use just those sounds. Or maybe there could be some sort of single processing that could help improve the intelligibility. So, I got interested in dysarthria because of that work and one of his legacies was to leave a lot of recordings that he had made of a bunch of speakers. So we started studying those recordings. Not necessarily with the goal that he had but for just the goal of trying to understand more about what is the nature of dysarthric speech.
Matthies: What was going wrong?
Stevens: What was going wrong? We have this general interest in can you, through acoustic analysis, collect or get some better understanding of what a particular speaker is doing that might help the clinician in the diagnosis and then ultimately the remediation of this.
Matthies: And you also have done a lot work in very thorny issues like nasalization. I know that Marilyn Chen has made a contribution in that area with your help.
Stevens: Yes. There has always been this uncertainty as to exactly what is the acoustic correlate of nasalization. Through looking first of all at the speech of the deaf which is often inadvertently nasalized, and through looking at other languages that have nasal vowels, and having the patience to look at lots of data, which Marilyn is very good at, was able to put these things together and come out with better theories of nasalization.
Matthies: Is she currently doing post doc work with you?
Stevens: We call her a post doc. Yes. Sheís working on the lexical access process.
Matthies: Itís wonderful that youíve been able to establish these long-term colleagues.
Stevens: Sheís very good. I donít know how long sheís going to stay.
Matthies: Another line of work thatís a little off the beaten track for you is the infant speech perception work that youíve done with Pat Kuhl. Do you want to talk a little bit about that?
Stevens: Pat was working on examining the influence of early exposure to language on the perception of, in our case, perception of vowel sounds and classification of vowel sounds by infants. Her thought was that if a child is exposed to the vowels of one language, they will respond differently to those vowels than a child of a similar age who is exposed to a different set of vowels or a slightly different set of vowels from another language. And so she had noted that in Swedish, there is a vowel ďEĒ [eee] not unlike the one we have in English and thereís a high rounded vowel, ďUĒ [you]. And as a consequence, the Swedish ďEĒ is a little different from the English ďEĒ. So she was examining the perception of kids who were English and those that were Swedish and was showing that, in fact, they respond differently to these vowels. They discriminate these vowels in a different way in these two languages. And my role was simply to synthesize a series of vowels that would cover both of those languages. But certainly seeing her work and talking with her as has gotten me interested in the study of speech and language development of children. Not that Iím doing anything on it but Iím convinced that that is one of the major areas where we have got to do some work.
Matthies: Well, the cross-linguistic paradigm adds a lot.
Matthies: What are you doing with Jay Keyser these days? I know youíre meeting.
Stevens: Having coffee. We are writing a paper, which I talked a little bit about at Newport Beach. We are trying to understand where the variability in speech comes from, quite apart from the variability in different peoplesí vocal tracts and so on. We've started looking at two things. What we do to enhance a phonetic contrast is one thing. Often that enhancement is dependent on the context in which a particular phonetic unit occurs. So you will enhance the voicing for a final stop consonant in a different way from when you would enhance voicing for a syllable initial stop consonant.
And the other that we often will talk casually and weíll overlap our articulatory gestures and that will lead to some modification of the acoustic representation of a contrast. And trying to understand what the principles of that are and in fact, is it true that you will overlap but only to the point where youíre still maintaining the distinction and you never lose it acoustically. And so, itís those kinds of things. It is all thought stuff. Itís not very much experimental.
Matthies: It sounds like a theoretical approach that gives you a good framework from which to make some predictions for experiments.
Stevens: And sometimes as a result of what we say we come to look at something in the lab. But never in a very consistent way, unfortunately.
Matthies: Is there anything else you would like to add about your current work or past work or MIT?
Stevens: No. Of course MIT has provided an environment for doing all of this. And of course, all the students and all the people who work around here make it more fun, which is what itís all about, after all.
Matthies: When you enjoy your work, your life is greatly enriched.
Matthies: Good. I think that will do it.