thumbnail of New aspects of language; Linguistic Hardware
Hide -
If this transcript has significant errors that should be corrected, let us know, so we can add it to FIX IT+
New aspects of language using a language is so much a part of our everyday lives that we do not realize just how complex an activity it is. Styx is the science responsible for investigating this activity. We take so much for granted. During this series some of the tools and methods used by linguists to study the complexities of language will be demonstrated. The series is prepared and narrated by Dr. Frankel associate professor of English and Linguistics at George Peabody college for teachers today Dr. Frankel is joined by J.W. Sanders associate professor of Audiology and speech pathology at Vanderbilt University and Peabody college with a clinical and research appointment to the bill Wilkerson hearing and speech center. And Henry savage audiovisual engineer at Peabody college in a discussion of linguistic hardware. Last time we talked about the linguists were I can when he has an informant with them. I mean the life person from whom you can elicit
today. We want to discuss some pieces of electronic equipment which had the thing with us with his work of analysis and other aspects of his language work. Such pieces are for instance the tape recorder. This as the most important piece of equipment and almost a steady companion for the Ning West they have areas that mountains which we can discuss in a moment. The other pieces we may mention today are they sound spectrograph and the computers are there of course many other this. We may have the opportunity to say just a few words about some of them which are somewhat more difficult to describe for many of these others being used in labs all over the country where frenetic work is going on at a steady pace. Now the first as I said of the tape recorder what does the tape recorder do to the playing West. Well
first of all you can play it steadily. You can use the material as many times as you want to without wearing out the equipment as you would probably do with an informant. That is identity of the material no matter how many times you play it. So there is a possibility to work on material without being confronted with the need to account for diversity. You may also use the loop which gives you the possibility to play without even the threading without its what as really the loop. How do you get it and how do you use it. Well there are different types of loops. Some of them are used in our cartridge. This is an endless loop cartridge which they use on commercial broadcast ations and I believe here and have to stay in the loop merely consist of Usually about two or three. What piece of tape splice together and then inserted him between the pressure roller and
cap something. I am the heads of the tape machine and the machine itself paces they take through the heads of those around and around and if you don't need a special tape recorder for that or not and as you say I just know you know never repeat the right weapon and in fact the man can prepared himself at home is all right and he laid and he Lang and then he can listen it goes on forever and ever. So this is the great advantage and the more often the material is repeated the better. The thing with gets acquainted with the sound it has and can analyze it and to take it down. Now one other great advantage. I would almost say the greatest unfound is the tip to the Horta has from the electronic point of view is the fact that one can play the material back with now the question why should it end with one to play Matilda back with well it's not just for fun. You hit the things men playing Matilde back with which you can
not hit when playing it already. That I people by the way who do it for fun that the well known thing with who has learned to talk in such a way it back with it that when you rebrand the tape and play it well but it sound exactly like the on that song but that is not the point. In fact some words in English. Son can take back wets leg of the English webspace foreign and you can done quite a lot about Sun production by doing just that. Let's have some examples of such words. Roy your role Roy your role you know how wrong and how wrong.
You will have heard differences of course. We cannot at this point discuss the exact differences which exist between the same played forward and backward. But it is obvious that there are possibilities for analysis which give us further inside into the production of sons. And this is of course most important for language as about who is X X phonetic production we don't have sufficient information which we have to learn from just such operations at the one produced at this point. There is another use of the tape recorder. When you played back where and this is the solution of certain arguments for instance certain sounds in English have characteristics about which linguists may not agree. Let's say the question is as a certain initial sound aspirated to an aspirated as a certain final sound a stop fricative a T or whatnot. Now again let's have just one example to show you a few words
for what n Bagwell and the indication will be that when the words are being played backwards the kind of sound produced is heard better and the argument can be sought of on an objective basis. There is no need for a purely impressionistic subjective argument. The tape recorder will produce the objective sound as it was given by the informant. Hit they hit they hit that that they hear that they hear that again this is just a sample and we can unfortunate not go into details of the analysis itself as time doesn't permit such a thing.
Another very important piece of equipment is a sound spectrograph duct as sound as can you tell us something about as the previous ss of the spectrograph the pseudo graph humo graph and so on. Well of course there were a number of instruments that were used for sound analysis prior to the spectrograph such as the missile a scope for example that gives you a graphic representation of the sound wave particularly the intensity and frequency of relationships. It was not. An extremely useful piece of equipment for speech and analysis however because of the complexity of the speech signal there's too much there for the oscilloscope to show quickly. So what is this this book good for it's a musical instrument. Any tone any signal that is sustained in time by continued steady tone speech of course is so changing and time that by the time you have gotten ready to look at what's there it's gone on to something quite different.
Yes now the spectrograph of course gives us an exact wizard picture of what we hear you speak into a microphone and the sound waves are transferred by a stylus on sensitised paper better and you get a permanent record of what has been said. I believe that the length of that stretch is two point two seconds. Or are there any longer ones now have these things change no are 2.4 seconds off we can argue over 200 second However even the latest model is still limited to two point four seconds and not a cause for analysis of a vowel sound or a short phrase. This is as much as is needed. This usually means you cannot get a complete sentence unless it is a very short something else. How can you take such a speck to Graham and simply read it and say oh well this is that one speaker whose focus Absent such a kind of dialect
of English. There are possibilities in this direction. Actually one of the all of the spectrograph certainly was a great advance and terms of our ability to analyze complex sound its very complexity is at the same time one of its drawbacks. Some people complain that it gives you too much that there is more information there than than is really needed and that indeed you not only with the newer model spectrographs can distinguish a person from one part of the country from a person from another part of the country but some people maintain that the spectrograph is as individual as fingerprints and are even calling spectrographs voiceprints. Is that put to use that say in the courtroom sometimes that just like your friends I believe it has been admitted as courtroom evidence in the dedication of a speaker at this event resting. You say there's too much in it. Can individual aspects be isolated out so that you
have let the ONLY loudness or only length. Yes you can do what is called a second time trace on the spectrograph in which you take just one particular segment and time of speech. For example a speech sound or sometimes a phrase used in an analyze it in terms of the frequency intensity that relationships are shown at that particular moment in time. I say let's say something about the name of that piece of equipment spectrograph across the graph isms into his writing obviously and we know what a spectrum is how can you write a spectrum. Well that's the whole function of the instrument of course it is to give a graphic a visual representation of the spectrum of the sound the fundamental plus all of the overtones that are present on the sound. For example when we do a spectrogram of the sound Yes we can then see on our graphic representation the fundamental the basic frequency of the voice itself and then we can see the pattern of overtone to see what is the bass line.
Mr. Stephens Well our baseline would be say starting from zero and working operating amplitude I imagine is you know 0 just infinity as zero mostly on its own no sound at all I see. Yes yes. Aha yeah. How does the whole thing but how can you transcribe audio signals into visual signals when just them well they slow down and I guess in many ways of course like we're talking now our voice is transmitted through an amp a microphone amplifiers and read out on the you meter. But it's good. Rather write it's heard and seen and it's and the meter is mechanically moving. I say if you put an pan on the end of this meter. Yeah along with it moving on a graph of paper moving at a certain speed it would visually record our sound. How better that's why village soul will write the letters of the words we know it. It won't take a computer to do that which they have already have an
operation on and yes some computers that is read that's right yes. But it it merely measures the amplitude of your signals. How does it feel when you see really just a little fine line here. How can you measure that line of business. Measure the line that is everything which is why the show is on the paper. What is the out of most of the paper itself is that the scale is set up. For example if you were looking at the amplitude already changes in strength of the signal. He has scale is present on the paper so that when the incline is at one point your scale indicates that you have so many decibels of intensity and the incline falls to another thought gee increase the signal has decreased and so that's a pretty calibrated again as I say that's the crux of the matter is that the microphone that is picking up the voice is what we call a transducer in changes acoustic energy into electrical energy and the electrical energy maintains the
same pattern of change as the acoustic energy being put into it. Has anything changed in recent years with this bank of man can do. Well this latest model that I was talking about in terms of how the great individuality of the tracing that is produced is certainly one change. The use of the spectrograph cause has changed as we have become more and more sophisticated perhaps yes using we. In my particular field of speech pathology and audiology regarded the spectrograph when I first came along as a solution to a lot of severe problems we had always had an understanding voice disorders what until in this respect for instance when I looking for. Indeed it has been a help. For example we can look at a horse voice as compared with a normal voice and certainly see great differences. We can even there's been a very recent
study out published in which the investigator has been able to describe four different kinds of hoarseness in terms of differences in the spectrographic pattern. And this of course may have a great deal of value not only in diagnosis determining what has caused this horse and has it also in working out particular therapeutic procedures to help the patient get rid of the chronic hoarseness. A lot of horse play. If you are really looking for the same things the linguist is looking for only that you are interested in the pathological aspect you as a linguist is interested in the normal aspect yes but the facets are real is the same the language would be interested in matters of relaying and intensity and various shapes which are created by the sounds produced. Now what creates these blank lines on paper. Mr Savage Why do we have for instance the three Foreman's I mean on strange questions What is it. How do we
get it how would you produce it we see these shapes we go up and they go down and have white spots on the brakes. How does all that come about. Well I imagine that would basically depend on the individual sound itself creating these and I mean you talk variation right. So what you're really saying is that the resonance chambers in RA was featured right guns produce these various Tyree of Faerie and say All right what is the difference between a narrow band and a broadband spectrum. This has to do with the with the with the frequences that is being passed by the set of filters in the spectrograph the analysis section of the spectrograph itself. If you set it for a narrow band spectrogram then the filters will pass on the way. All I believe are depending on the equipment about a band of about 75 cycles per second wide whereas the wide by and would be ordered 50 200 cycles per second wide.
What as if the characteristic feature which one of the two would show which the other does not show the in terms of the work that we do. We stick primarily to the narrow band because we believe it gives us a more specific pattern. It's possible that we might by using the wide band get away from what I was mentioning before having too much information and it's being too highly specific plan. But we have found that the much more specific detail that the narrow band gives us seems to sort of our needs best. Now that I understand the following that if different people let's say produce different Vols or different nasal sounds and you may perceive them as identical but to speculate on which show that they had differences and that's what interesting. More on the other hand in being they perceive them as different and they come out identical so that the objective measurement is quite different if I was objective understanding. How do we do that with let's say stops which do not show anything there's just a
break because there is no resonance. So if you have the impression that different people produce different stops can speculate am RUF that not to my knowledge because when we do a recording of a stop on the spectrogram for example the medial sound in the word stop or Yet the spectrogram simply shows us at the moment of that stoppage of the breath flow as our lips come together to form the the piece sung if it's in the stuff. Yeah you said it was hot and it was a blank spot using a spectrogram so the guilt is dealt with just by different people would not show differently. That's right. Would there be a difference had seen the transmission from other sounds into the stop or back into the other sounds from the stop. This may well be because of course going into the stop where you are ordinarily going from of our sound has ition and we may well see some individual variation here. But even here the same vowel sound spoken by two speakers will be much more similar than different vowel sounds spoken
by one speaker Yes. No we have been saying all along that the speculum is objective. Does the spanking them show anything which the ear will perceive at all. Yes indeed the sound for example which we think of as simply a puff of air right here is an energetic ration. Yes the rest of us and it is not for the Yanks. Yeah but of course it still is produced with resonance and so the spectrogram shows us that the age of sound will have the same four months as the vowel sound that follows it. I see so it is really an un boys mouth that's right whereas a pure breath different. Yes I say. Does this make them show you where sounds are produced only how they operate used can you see immediately whether they are the little sounds or sounds I think that only insofar as you're able to interpret the frequency intensity relationships on the spectrogram in terms of previous knowledge as to what particular places you eat
may do doesn't show it was not with some kind of immediacy you have to study the exact representation we are speaking about the new developments and you said that more sophisticated space you have in existence now which I admit it is quite evidence. Other more sophisticated tape recorders Mr Simon. Yes just in the recent year of the new the new All tape deck has become very prominent. Could you give us some details on that. This is just a fascinating new tape deck as you know a conventional tape deck usually has three motors that take up a regular capstan drive and then a supply real motor. But they knew I knew all system only has one motor and that you can do right. Everything it can do everything it can face fart at tremendous high speeds and there is no tape where quite a number of things you can put a number of channels on a quarter inch tape probably up to 20 or 30 channels on the right
and the right to instate right there is no particle displacement of the type I see in the new system which is one of the hard things to accomplish in the conventional tape day. And if it doesn't get all of each channel carries a different language right. Right it can go face fart or reverse either way and either way you want to by reversing the motor ladies. No this really leads us to the last question on this kind of equipment. Linguists don't usually worry too much about the budget they leave that to the institutions but nevertheless I think if you want to out the question How long would it take before everything with can have his own sphincter grabbed you with a tape deck and so on at home or in his pocket to the computer for that matter. How does this seem to you as it wouldn't be expensive and can only very few businesses afford it. Just as well it's extremely expensive equipment the latest spectrograph for example costs around twelve thirteen thousand dollars.
Other pieces of equipment are also quite expensive. My own feeling is though that if a laboratory is to be equipped for research for learning for further understanding of speech in a language that the linguist can't very well afford to be without it of course it has the advantage that you don't have to buy a new model every half of his DS but uses. How about the tape recorder files the new all day. Should be should cut the average transfer cost down to at least a high because there are less moving parts there's no headway there you know and with less moving parts would reduce your cost preservation and that means that despite the technological sophistication of the work and thinking which goes into the price of the actual piece right the price of all that small amounts have been dropping in the past years. Now we don't use transistors bottom sales anymore we go to the integrated circuit. I see you have a new dimension. Yes we have a complete circuit a complete high water watt amp a far and size of your thumbnail. The cost of the units average about two dollars and eighty
cents a piece. So less reliability too. We don't have to worry about taking it back to the repair so we may be able to afford these things. Yes abs here now if somebody had time to develop it. It's an and quite a few people and companies are developing this particular deck. Our video pictures sound to other uses. Well the linguist can look forward to much more help than that perhaps and he will eventually become outdated and looking forward to the day when the equipment is doing the N.S.A. a duck to send as could be in the 1 minute or so we have let's see something about the x ray photography in conjunction with other equipment for the Ning with I haven't really thought about this in terms of linguistic analysis it is extremely helpful and speech pathology of course and an audiology could enjoy them look up because it shows movement. Yet it shows our structure and helps the speech pathologist to understand for example what is happening in areas that he
cannot visualize. Yes so this is really marginal for the thing was you would say although of course the speech pathologist and the linguist Nessa said it where you have a close conjunction. Well thank you very much gentlemen this was very helpful and I hope you have some opportunity to talk about these things. And now to the computer. We asked Professor Ellis from glasses fellow director Georgia Institute for Research University of Georgia Athens Georgia to give us his views on computational linguistics. Here is a summary of the points in his reply. Point one the computer's outstanding qualities are speed and precision. It can do in minutes or hours what it will take a linguist years to do. The computer in fact will give the answers before the linguist knows what to do with it. A human being will get tired and bored if he does the same thing over and over. He will then make mistakes. The computer will never make mistakes and it is his program to wrong
us. Therefore the linguist should never be his own programmer. He should look upon the person who writes to program as just another linguistic too an example of what can be done. This is a program prepared at the Lockheed corporation all of Webster's Dictionary it was put on punch cards. This information can now be used for all kinds of linguistic purposes such as for instance for reverse dictionary which can be very useful for phonological and morphological research to a new area of research is called computation and Stylistics as an example. A computer may be able to establish the authorship of an apocryphal manuscript by counting the number of occurrences of favorite words which we know a certain author has. Point three words are important but larger units are more important in the last decade or so much work has been done on syntax and it now
extends to paragraphs chapters and whole books. What we really need for this kind of analytical work is the computer who speaks a natural language instead of a computer which understands only secret computer language. This would mean of course that a computer can also handle the day to day can texture the experience of the native speaker point for research as thought they knew all the problems and how to look for ounces. The computer has told them that they even don't know how to ask questions. One thing is clear. Computers can only do what they are told to do. No more no less. If they are given just words in ordinary grammar rules they can not disambiguate sentences. It takes an enormous amount of knowledge of context to do that. Exactly how much context we simply don't know. None of the five existent computer programs for syntax has been able to find that out. This remains for the future.
Point fun. The most important inside computers have been able to convey is that only 50 percent of the message we get from speakers are in the new linguistic code. The wrist is an everyday experience. This fact is important for the understanding of literary works for instance. This insight also forces the linguist to RI and I him 7 his work with other disciplines such as psychology anthropology and philosophy and the lines which he had neglected for a long time. Point 6. Computer Research has went both ways. The linguist found his hotbed valuable tool for problem solving while simultaneously the computer told the linguist to ask general questions about human existence and human communications. In today's discussion of linguistic hardware Dr Franco with joined by J.W. Sanders associate professor of ideology and speech pathology at a better built University in Peabody college with a clinical and research appointment to the bill
Wilkerson hearing and speech center and Henry savage audiovisual engineer at Peabody college. The topic for next week's program will be interacting with Dick's. New aspects of language it prepared and they're rated by Dr. Frankel associate professor of English and Linguistics at George Peabody college for teachers and it's produced in the studios of WPI then there's Richard wideout speaking. This program was distributed by the national educational radio network.
New aspects of language
Linguistic Hardware
Producing Organization
Nashville Public Radio
Contributing Organization
University of Maryland (College Park, Maryland)
If you have more information about this item than what is given here, or if you have concerns about this record, we want to know! Contact us, indicating the AAPB ID (cpb-aacip/500-dn3zx795).
Series Description
For series info, see Item 3622. This prog.: Linguistic Hardware
Media type
Embed Code
Copy and paste this HTML to include AAPB content on your blog or webpage.
Producing Organization: WPLN
Producing Organization: Nashville Public Radio
AAPB Contributor Holdings
University of Maryland
Identifier: 68-36-3 (National Association of Educational Broadcasters)
Format: 1/4 inch audio tape
Duration: 00:28:46
If you have a copy of this asset and would like us to add it to our catalog, please contact us.
Chicago: “New aspects of language; Linguistic Hardware,” 1968-09-23, University of Maryland, American Archive of Public Broadcasting (GBH and the Library of Congress), Boston, MA and Washington, DC, accessed April 20, 2024,
MLA: “New aspects of language; Linguistic Hardware.” 1968-09-23. University of Maryland, American Archive of Public Broadcasting (GBH and the Library of Congress), Boston, MA and Washington, DC. Web. April 20, 2024. <>.
APA: New aspects of language; Linguistic Hardware. Boston, MA: University of Maryland, American Archive of Public Broadcasting (GBH and the Library of Congress), Boston, MA and Washington, DC. Retrieved from