to linguistics menu
A Workshop in Evidence Based Linguistics.
Also presented at the LACUS Conference
at Dartmouth, 2005, as a continution of my presentation
Is Evidence Based Linguistics the Solution?
Is Voodoo Linguistics the Problem?
(read this piece first by clicking here)
Presents the "Six Laws of Linguistics,"
a system of linguistic cartography, a measuring
unit, and a system of differential diagnosis for language
(and much more).
The Chairperson for this session was
Professor David Bennett,
University College, London.
Those entering the hall see
the following slide projected
on a screen.
(SLIDE: List of Workshop Topics)
2. Six Laws of Language and Linguistics in Draft Form.
4. My Adventures with Voodoo and Evidence Based Linguistics.
5. The Differential Diagnosis of Language.
6. Some Further Details about my Gedankenexperiment.
7. Discussion: Can EBL Be Taken Any Further?
8. The Clarification of Context as the Brain's Chief Language Function: Not Grammar but Linguistic Cartography and a Unit for Measuring Language.
10. Anecdotes and Conclusion.
Okay, if you look at this slide, you'll have some idea of where we are right now, and where I hope we'll be going during this session. As you can see, it has two or three discussion periods built in. During our planning discussions we referred to this session as a workshop-slash-tutorial, and I think I'd like to hold to that concept, even though in our program it's now called a workshop. In either case, what this means is that unlike Wednesdaywhen it was mainly me talking and you listeningtoday we're going to try being slightly more informal by alternating, at least part of the time, my talking with your reactions.
But I hope it's just part of the time, because I've figured out that by conservative estimate, I have at least another hour and a half of stuff I want to tell you about, and some of it is fairly interesting too. But I won't be able to do that if the whole session turns into an absolute free-for-all, so I'm hoping our Chairman will humor me and let me get away once again with doing most of the talking. That's because I have only one opportunity to tell you about this other material, while over the coming days & weeks and months I hope all of you will have tons of time to get back to me via email or otherwise about your reactions and any questions that may occur to you. But I promise, I'm going to let you all have a chance to speak quite soon.
We're located in the "Intro" right now, and the next thing I do after this is I read aloud the Six Draft Laws of Language and Linguistics mentioned in the slide, and then I give each of you a copy you can take home with you and think about and hopefully get back to me with your comments. I'll be saying a few words about some of the laws, and then I'll open the floor to questions or criticisms. But I'm also warning you in advance that most of your questions and criticisms will probably be answered or at least covered in all these other sections that come afterwards, which I hope will get me off the hook from having to spend too much time on questions when I'm just getting started. But I certainly can't stop you from asking.
As you can see, there are seven or eight further sections with weird names or making weird claims, stuff like differential diagnosis or clarification of context or fractal linguistics, and I really hope we can go through it all together. As you can also see, there two possible bits of rough terrain, which I've marked in red with the word "LEAP!!!", and I'm hoping I may actually get you to take those leaps with me as well.
They're not leaps of faith, I promise you, nor do they involve any real leap in logic. But in each of those places there is a leap of conceptualization that has to be carried out in order to make it safely across the gaping abyss. I'll do my best to prepare you for both those leaps before we take them, and I don't mean to sound condescending if I assure you that I'm pretty confident most of you will find them relatively easy.
A lot of people, both linguists and non-linguists, academics in related fields, and even educated members of the general public still can't grasp how there can be anything all that conceptually difficult about language. It's that old story in the German proverb, after all we all speak it, don't we, so we must understand how it works.
Now take something like higher mathematics, we can respect that as being truly deep and conceptual, but languagenah, language is a cinch. Just look at the way someone like Einstein or Gödel or Paul Erdös would fill a whole blackboard with brilliant formulas and calculations and then, when he had no more room to write, he would point to the blackboard and say "Ladies and gentlemen, I want to represent all of these figures and their relationships by the letter N." And he would erase a little space at the top of the blackboard and write the letter N in it, and then he would erase everything else and start all over again and fill the blackboard once more with completely new formulas and calculations, all based on the meaning of the letter N.
Now that, as some of you may believe, is truly advanced conceptualization, We have nothing in language that comes close to approaching such a level.
But if you any of you were to say that or believe that, then you would be quite mistaken. What we all of us do with language every single day is every bit as complex and abstract and finely balanced and just as exquisitely derived and developed as anything mathematicians can ever do with their calculations. We have whole value systems held in place by supports so tenuous and speedily calculated, themselves arrived at by a process of trial and errorbut mostly errorthat it's amazing that anything holds together at all, and this house of shaky premises, much more like a house of cards than any truly stable structure, we call by the grandiloquent name of our deeply considered world view.
We are simply not aware we are doing this because most of our deriving and developing has been performed over broad periods of time, though it can occasionally spring forth during even shorter durations. What's more, the daily complexities we gladly and effortlessly execute often have extremely practical, 3-D, real-world consequences, while the delightful intricacies of master mathematicians only occasionally collide with the real world, though once in a while with a very big boom that once again inspires our awe for them.
And all of this is perhaps just one of the reasons why we may have no choice but to take a conceptual leap now and then.
2. The Six Laws
First of all what I want to do is begin by actually giving you something you'll be able to react to as much as you want, namely the draft version of my Six Laws of Language and Linguistics. I'm hoping your reactions will help me over time to shape them into a more final version.
I'm going to start by simply reading these draft laws aloud to you, and then I'll give each of you your own actual printed copy you can take home with you and mull over, mark up with doodles or comments, or simply use as a dartboard if the spirit moves you. But I'd be happiest if you'd read them over at your leisure and let me have your considered comments over time via email or snail mail or maybe voice mail (though I hope not) or, god save me, Morse code, drums through the hills, or even passenger pigeons.
Six Laws of Language and Linguistics In Draft Form
1. All communication takes place in shared contextual space, subject to a more or less complex process of disambiguation, depending on the conditions inherent in the other five Laws. That space can be more or less roughly measured according to a specialized system of cartography.
2. The Law of Variable Context
If two people share sufficient context, almost any words, including sheer nonsenseor no words at allwill suffice for them to communicate with each other. If two people do not share sufficient context, then not all the words in the world may be enough for them to grasp each other's meaning. Where intermediate degrees of partial, fragmented, or otherwise limited or "noise-distorted" context are shared, communication will be proportionately difficult and/or unsuccessful.
3. The Law of Communication
Communication never takes place generically between languages and languages, or between dictionaries and dictionaries. All successful communication takes place under specific circum-stances between a speaker and a listener, or a writer and a reader, or between a non-verbal communicator and his or her audience. When the communicator changes, and/or the nature of the audience and/or the circumstances change, often the content of the message must also change to some extent, if something approaching successful communication is to take place. This law holds true both for communication in a single language and for translating and interpreting, since there is essentially no difference between translating a message into another tongue and paraphrasing it within a single tongue. This law also holds true for automatic or electronic communication where the final recipient of information is a human being, and any act of communication appearing to originate from a computer, or to occur between two or more computers, only takes place because a human being has originally programmed it to occur. All the conditions of the first two laws still apply.
4. The Law of Linguistic Entropy
A form of entropy, related to Shannon's concept of information entropy or Prigogine's theory of dissipative structuresor of chaos as found in meteorology and other complex systemsalso exists for language, and any sentence, concept, or act of communication may fall into such entropy or chaos even after it has been accurately repeated a number of times. Where Shannon's concept applies to letters of the alphabet, this one applies to words, phrases, and/or entire sentences. The number of times the message must be repeated to fall into such entropy or chaos depends on the nature of the message, the number of people attempting to repeat it, and whatever ambient or incidental noise of whatever type may be present either in the system they are using or among those attempting to repeat it.
5. The Law of Recapitulation
Just as Haeckel and Von Baer observed and debated the nature of a form of recapitulation in the development of the embryo, so there also exists a process of recapitulation regarding language. During their development from children into adults, all human beings will necessarily pass through a recapitulation of as many of the forms and structures of their language as they possibly can within the limits of utility and the peaceful development of their society.
6. The brain understands the language it hears or reads through a combined comparison of sound, meaning, context, and expected collocations, seeking out a match with other sounds, meanings, contexts, and collocations it has already encountered. Once it has made this match, which may be more or less precise, it assumes it has understood correctly. Grammar plays a relatively small role in this process, sometimes none at all. Said otherwise, the way almost all communication works is by means of a relatively error-prone, quick and dirty matching operation. We know the brain proceeds in this manner, because it sometimes makes mistakes, permitting us to draw inferences about the way it functions. This process has for its source the humble origins of language through evolution from the chemical signals of early life forms to the scent markings of animals to the sound markings of humans, which we interpret as language, thus providing further proof that Darwin's theory of evolution must be true.
That's it, it's time to hand out copies for everyone to take home with them, so here they are. (distributing copies) Let me make just one more relatively formal statement about these laws before I mention quite briefly some of the problems they may present, leaving time for you to say something fairly soon. It is my hope that over time these laws can help to remedy a number of erroneous preconceptions advanced by theoretical and computational linguists. I believe that linguistics truly needs something like these laws if it's ever going to find its way out of the dead end situation it has sentence-diagrammed itself into. Many sciences come equipped with their own laws and measuring units, and I see no reason why linguistics should be an exception.
As you can see, a number of different sciences are represented in these laws, including biology and physiology, Shannon's information theory, translation studies, plus a touch of Asimov's laws of robotics, and even the odd bit of physics, since these are all areas which over the years I have tried to assimilate as well as I can. Before I open the floor to questions and comments, let me see if I can clear up at least a few points about these laws that might be causing confusion. In the first law I've talked about a "space that can be more or less roughly measured according to a specialized system of cartography." If that sounds positively weird or vague as hell, let me promise you that approximately 27 minutes of what follows will be devoted to clearing this up as perfectly as possible.
As for the Second Law, during acts of communication context comes close to being everything, and either two human beings, or even two machines, share sufficient context to clarify seemingly unclear references, or they do not. If they do not share such context, then they will not be able to complete that particular act of communication. This is simply the way that language works.
If the context is not entirely clear, how on earth could meaning be communicated with anything approaching precision? And why would we expect language to be all that precise in the first place? Why should it be precise? How can we expect absolute precision from languages concocted by human primates from little more than sound markings, little bits of sound and sense and spittle? Especially when no two tribes of these primates share precisely the same sound markings, and even within the same tribe factors of birth, wealth, and health determine how well tribe members can ever master their own system of sound markings?
Concerning the third law, I'd like to offer the following clarification. Since I have worked over several decades as a professional translator, some of you just might be confused by the claim that "when the communicator changes, and/or the nature of the audience and/or the circumstances change, often the content of the message must also change to some extent, if something approaching successful communication is to take place." Can this possibly be my policy when I am at work translating a business letter or a scientific text? No, of course not, the responsibility of the translator has always been and remains to rephrase language B as perfectly as possible so that it approaches as closely as possible to what has been said in language A. The key to any seeming contradiction lies in understanding exactly how close "as closely as possible" can be.
Here we come to a crucial point which I have made elsewhere in other terms. Contrary to our facile belief that there can be such a thing as a "good translation" or a "correct translation," no such thing as "generic translation" that will work in all cases may exist at all. It may simply be a convenient fiction we have employed to shield us from the true complexity of the translation process and/or as a way of reassuring ourselves or our clients that we are in all cases capable of producing a "correct translation."
Let me phrase this a bit differently: there is one other crucial factor involved in a translation besides the two languages involved and the nature of the subject matterit concerns the audience and/or the occasion for such a translation. Wherever this audience or occasion changes even slightly, there may have to be a corresponding shift in the tone of the translation. Where either of these factors changes more than slightly, we enter the territory of rejected translations, possibly even charges of incompetence. But even the most conscientious translator or translation editor may not always be prepared to meet every demand these circumstances are capable of hurling at us.
What we have run into hereor perhaps it has run into us with a big stick in its handis the true extent of the complexity of language. It is hard enough for humans to work under such circumstanceshow can we ever expect machines to handle them? The real explanation here may well be that we all make some outrageously false assumptions about language and are totally unaware we are doing so. Once again, we assume that we are all walking around on a level playing field, where anyone can readily communicate with anyone else across a short and easily bridged distance.
But the truth is that we do not inhabit a level playing field at all where language is concerned. On the contrary, if we were to visualize ourselves and everyone around us as walking about on stilts of completely different heights, textures, and stability, so that even our very own two stilts are not necessarily of the same height or composition, we would have a better notion of how we actually move through linguistic space and communicate with others. You can easily persuade yourself that this is true by the way you react to others the next time you are in a social situation.
We each of us have our own store of linguistic tricks and devices, and we look out almost instinctively for those who have comple-mentary tricks and devices. Whenever we meet such a person, we become flushed with enthusiasm, sometimes even love, and go on talking forever. But we just as quickly abandon those who do not respond to our conversational rhythms. True, we also carry on everyday conversations with persons who do not share our interests or language style, but we usually do not speak at length or in detail or about more than a few topics with them. What I am trying to suggest is that there is a whole universe of language habits we are simply unaware of. And if we are not aware of them, how can we suppose that a computer can ever be able to gain such awareness?
I'll like to defer discussion of the fourth law until later, when I'll be saying something more about it and the entire topic of fractal linguistics. As for the fifth law concerning recapitulation and the precise identity and life's work of those two great biologists Haeckel and Von Baer, I would need another half hour, which I don't have, to do justice to this topic. If you do want to know more, perhaps one of the best places to begin would be Steven Jay Gould's remarkable volume Ontogeny and Phylogeny.
As for the sixth law and my continued assertion that grammar plays very little role in helping the brain to decode language, I'll also be paying detailed attention to this in just a few minutesperhaps more detailed than some of you might preferand to the question of what the brain is likely to be doing during the process of understanding language.
With that as an introduction, provided our Chairperson agrees, I would now be happy to entertain as few questions as he will let me get away with, since I really am eager to present my remaining material, which may answer some of your questions but also raise new ones.
I'll now be happy to entertain any brief questions you may have, though as I think I've already warned you, if I appear to be waffling slightly in my answers, it's because I believe that those answers are likely to lie in the material that lies ahead, where I'll be treating various aspects of these laws in great detail.
4. My Adventures with Voodoo and Evidence Based Linguistics
Even though this section contains some useful material, it was omitted during the actual presentation at Dartmouth, partly because of time considerations, partly because I decided it was a bit too argumentative and wanted to concentrate on more positive material. You can however read it at the end of this document.
5. The Differential Diagnosis of Language
Now I want to take you into some comparatively deep water, one of the parts of my research that may be a bit hard to grasp for some of you, though I think I can show how even this is evidence based. This is the point when I start to talk about the Differential Diagnosis of language, and I'll soon be explaining exactly what I mean by that. And I'll also soon be asking you to do your best to take a rather large and precipitous leap with me, a strenuous mental leap that I'm hoping you'll find is worth the effort involved.
But first I'd like to begin by quoting the well-known German translator Per Dohler, who runs a German translators' group, edits their online newsletter, and has also been active in the ATA. Dohler did some polling among translators as to how they felt about linguistics, and discovered that most translators believe that our study belongs to the "pseudosciences" or "fields that don't yet have their jargon under control." I want to stress that my German friend is in fact a defender of linguistics, and though I have a few reservations about the field, I consider myself to be a friend of linguistics as well.
To put us in starting position to take that leap, I'd like you all to ask yourselves if there are any other scientific studies which one might also hear described as primarily "pseudosciences" or "fields that don't yet have their jargon under control." I want to leave to one side sociology, whose fairly obvious faults may spring from simply being too vast and undefined a subject, and ask if any other such sciences come to mind. The answer I am anticipating and want to discuss is that yes, there are two other such sciences frequently enough described in this way: various branches of psychology and also that one particular branch of psychology that studies and claims to treat the pathologies and abnormal conditions of the mind, namely psychiatry.
I have a purpose in asking these questions and in anticipating your likely answers. Views similar to those found by my German friend are indeed often expressed about various trends and schools of psychology, and I believe most of us are also aware of the endless rivalry and vast argumentation that goes on among advocates of rival psychiatric theories and the problems they frequently find in revising their official diagnostic material.
Now I'm ready to take you for that leap I mentioned before. Let's take a purely systems analysis look at the totality of human knowledge and the place of these three disciplines within it. Could there possibly be some reason why such similar charges are so frequently launched at these three would-be sciences: linguistics, psychology, and psychiatry? Could it possibly beand here comes the leapthat the reason these three studies seem so particularly deficient is that they are all in fact ultimately parts of a single science that has been unnaturally divided into three artificial segments?
I'll give you a few seconds to absorb this rather grandiose assertion while I point out that if my answer is correct, this could explain why linguists so often seem to be milling about in an area with few practical applications and can't come close to agreeing about theory, why psychologists have evolved excellent testing and investigative methods but rarely seem to be testing or investigating any theme of major interest, and finally why psychiatrists are unable to theorize or reach final conclusions about some of the major complaints they study. Could it be at least partially because the linguistic and language-producing symptoms of these complaints are of little interest to them?
I'm quite well aware of my full rashness and/or foolishness in making these assertions, for I have now ranged against me not merely several schools of linguistics but all the fierce advocates of various schools of psychology and psychiatry as well. On the other hand, if I were to turn out to be right, the diagnostic knowledge gained by such a merging could well overwhelm us with unprecedented successes in all three fields. The combined science could be called something like Language Functions in Sickness and Health, though I'm sure that can be improved upon and abbreviated. It would indeed be an evidence-based study, to an extent that none of the separate segments has never attained. But long before any of that happens, I wonder if the basic idea doesn't make at least a certain amount of sense even now.
But it's pretty unlikely that I could be correct, I'm sure you will agree. Though stranger things have happened in medicine and, come to think of it, in all the learned disciplines before. It's only recently that we've finally found out about the whole Asperger's/High Autism Spectrum, which turns out to affect quite a few members of society, including a fair number of our colleagues who deal with computers. But in this whole scenario we're back once again with that problem of the reorganization of the university. Not impossible, after all we no longer teach the trivium and quadrivium as the basis of all learning, but such a change is still unlikely to happen during any of our lifetimes.
If just in case some of you may not have been able to follow me in the leap I just tried to take you on, I'd like to suggest that over time it might come to appear a lot less precipitous. I can already hear you asking me to show you the evidence for making such a claim. But in this one case I believe that more than sufficient evidence can be found merely in our ability to ask these questions. In other words, the mere fact that we may be able to agree, at least to some extent, that all three of these studies have their deficiencies is more than evidence enough. Besides which, all of this is intended only as a prelude to the differential diagnosis of language problems, which we are perfectly free to discuss whether or not the theoretical underpinnings are in place, and which I am now about to describe.
What is differential diagnosis anyway? Essentially it's a continuation of that well known diagnostic pamphlet "The Automobile Trouble-Shooter." There the first question is: does the motor start? Second question: if the motor starts, do the wheels turn? Next question (though there could be many choices): is the gear shift in drive position? And so on. Except that in medicine differential diagnosis applies not just to a car's innards but potentially to everything that can go wrong inside the human body.
It is the subject that first and second year medical students spend much of their time studying, learning by heart almost endless lists of symptoms, saying and hearing them over and over again to each other, until they are able to spot the one decisive symptom that defines a specific condition. Stuffy nose, sneezing, sudden bursts of phlegmprobably the common cold. Those same symptoms plus coughingmight also be common cold but perhaps also bronchitis. Same set of symptoms plus an unexplained fever, better run some tests, could be pneumonia or even worse. And so on through the whole catalogue of human complaints. This is how medical students learn to think like doctors. And once they have mastered differential diagnosis and gone into practice, they can often quickly and easily diagnose from 75% to 85% of their patients simply by listening to them describe their symptoms plus asking a few questions.
There are of course some language pathologies already covered by linguistics: aphasia, dyslexia, dysgraphia, voice disorders, various degrees of deafness, and a few others. The point I am making here is that these may be only the tip of the iceberg, and that many other language disorders can not only exist but may actually be commonplace-we simply do not think of them as disorders.
What I am about to show you is very much a work in progress, classifications that are by no means in their final form, but I believe they will give you some idea of the overall concept. That's why I am not giving these out as a handout, unlike the laws of linguistics, which I believe may be sufficiently advanced to profit from your criticism. That's simply because I would not like for this section about diagnosis to be shown out of context as an example of my finished work. Here are just a few sections selected from two out of so far five tables I have drawn up showing some of what I call the "diseases of language," so far as many as twenty of them, with perhaps more to come. For now I'm going to show you just four examples, though I'll try to say at least something about the others as well.
The Diseases of Language--Differential Diagnosis
Paralysisis of both
False starts in
As you can see, there are five columns, and beneath their headings the various candidates for inclusion as "diseases of language" are listed and described. The five column headings across the top read respectively:
"Name of Disease or Syndrome"
"Language/Knowledge/Reality (LKR) Relationship"
"Age of Onset," and finally
"Makes Condition Better"
All but the second of these are standard headings in such differential diagnosis tables, to be found in most advanced medical books. That second heading, described as "the relationship between language, knowledge, and reality" (or "LKR"), is explained in great detail in a chapter of my book in progress entitled Where Does Language Stop and Knowledge Begin? It is not at all a difficult concept and concerns the integration of these three factors inside the mindlanguage, knowledge, and realityof each individual who may suffer from various imbalances in their functioning that influence each condition being described.
As I mentioned, I have so far tentatively described some twenty conditions that may spring from various imbalances between these three planes. The four conditions I would like to call your attention to here are Graphiphobia, Rhetoriphobia, Dysepia, and Dysglossia. Of these four, the names of the last two at least follow the model of two other recognized pathologies in our field, Dyslexia and Dysgraphia, commonly treated by language therapists.
Let me describe each of these four conditions. For Dysepia, the inability to handle simple questions, the symptoms are "embarrassment when asked to give directions or avoiding the issue, problems in providing simple answers, even when one knows them perfectly well." As for Dysglossia, or the fear of appearing foolish while speaking a foreign language, its symptoms are "false starts in foreign languages utterances, laughter, embarrassment, reverting to native tongue." For Graphiphobia, or fear of writing, its sufferers complain about the "certainty that one will be unable to write about a subject, often a self-confirming fear." Finally we have the extremely common condition Rhetoriphobia, or fear of speaking in public, where we find among other symptoms "inability to begin speaking, freezing in mid-speech, avoiding invitations to speak in public."
Concerning the age of onset, all four can occur at any time of life, though Rhetoriphobia is more likely among younger people, and Dysglossia becomes more difficult to treat during later years. Each of these conditions requires a fairly individualized treatment program, including classes, sessions with others who share the problem, tapes or language labs, and for Dysepia special exercises in giving directions.
But despite these treatment possibilities, countless millions of Americans continue to suffer from one or more of these conditions and go through their entire lives without any significant improvement. If this is not a definition of pathology, then I don't know what else to call it. And if this does not provide clear proof that language is most definitely not an innate gift bestowed on all human beings, then it is possible that those who believe in such nonsense simply don't want to hear any evidence to the contrary.
The other diseases of language I have so far identified and attempted to describe range across a wide spectrum of behaviors, including some conditions familiar to all of us, such as "Tip of the Tongue phenomenon" or what I call Cyclorrhea, or talking in circles, or Logicorrhea, a form of conduct not totally unknown among linguists, an absolute fixation on logic, as defined by the speaker. But such breakdowns between Language, Knowledge, and Reality can also encompass far more severe and socially prevalent imbalances such as literal-mindedness, or related disorders springing from an obsession with rules, or perhaps even so widespread a condition as social or personal hypocrisy. I am still working to perfect these categories and believe that I will soon be able to describe them in greater detail.
I hope I have begun to give you a rough idea of the material that will be contained in the differential diagnosis sections of my work in progress, once I have finally found a publisher for it. There are just one or two other aspects of medical linguistics that I'd like to mention briefly. In one piece I made widely available on the web and on USENET newsgroups five years ago, I remarked that one of the chief shortcomings of the MIT-LSA school was their total inability to conceive of the sheer physicality of language.
One would-be critic of my position so completely misunderstood this statement that he assumed that mentioning physicality could refer to one thing and one thing only: gestures as language, which was already well covered by linguists. But of course I did not mean that at all, I meant that even though our most eminent pundits imagine language to exist almost entirely on a sublimely mental and intellectual level, the point I was making was that almost everything about language can just as readily be described as largely physical to one extent or another.
Those who do not understand this will have a great deal of difficulty if they ever try as adults to achieve near-native capability in a foreign language. This is because the mechanisms for saying the sounds and words and sentences of a foreign language correctly are essentially the same as the mechanisms for learning to catch a ball. The only difference is that the muscles involved are far more minute and the degree of swift coordination required is far more sophisticated. Much the same skills are at play in appearing "witty" even in a single languageone may think of a witty remark but still not be able to utter it at precisely the right moment. In this way physical agility also plays a role in perceived mental alertness.
If you will permit me just a touch of levity at this point, I would like to suggest that perhaps the most frightening of these diseases of language is in fact one suffered by many working linguists, namely Morpheme Addiction. But as serious as this can become, we would do well to remember that it can lead to an even more alarming condition: Premature Sememe Emission.
My purpose in including this section today has been to give you a rough idea of where my thoughts are headed in developing a physiology of language. I think I have now succeeded in doing that, and I'd now like to move on to the next topic.
6. Some Further Details about my Gedanken-experiment
(Though this section is presented in full below, to save time during the workshop it was summarized in a few brief paragraphs. What follows here is a shortened version of Gross, 2003, available in its full form on this website by clicking here.)
Now I'd like to say something more about that Gedanken-experiment I mentioned on Wednesday, which fully backs up my contention that there is nothing wrong with translation that is not also wrong with language itself. In other words, when the Harvard philosopher W.V.O Quine discussed what he called the "indeterminacy of translation," what he was really referring to was the "indeterminacy of language." In this description I am borrowing from the version published in one of that seemingly endless series of volumes by John Benjamins. (Note 1)
Our experiment begins with the assumption that a group of highly trained journalists has gathered in a room and is seated around a table. These journalists, all roughly of the same age, are seasoned and respected members of their profession. Moreover, they have all studied at the same school of journalism and received their training under the very same professors. It might therefore be assumed that they not only spring from a common background but that they also share overall outlooks and attitudes towards writing and editing to an uncommon extent.
These journalists are now presented with a paragraph written in their native language, along with blank sheets of paper, and are prompted to compose a paraphrase of this passage. Since paraphrasing is a common journalistic exercise and bears a close resemblance to work they routinely perform each day, namely editing, they all immediately set to work, and each one creates his or her own paraphrase of the very same text.
It would be theoretically possible to assume, given the nearly identical background of the participants, that their paraphrases would turn out to be remarkably similar, differing only in a few slight touches. But I do not believe for an instant that this would prove to be the case, nor do I anticipate that any reader sophisticated about the nature of language will draw such a conclusion. If, in fact, we now collect these paraphrasesfive or ten or however many there may beand read each of them aloud to our circle of journalists, I believe we will be amazed by how many different approaches have suddenly sprung from the same original passage. And if this is our result among this remarkably homogeneous group of journalists, it will surely take place to an even greater extent among writers or journalists coming from more diverse backgrounds. And if we now choose to substitute translation students for journalists and ask them to translate rather than to paraphrase a brief passage, I do not believe that any reader will doubt that something remarkably similar will now take place, on this occasion involving two languages rather than a single one.
I also fearlessly predict that the various individual differences of style and wording we discover, whether among the journalists or the translators, will fall into two general categories. The first of these, by far the greatest number, will consist of slight liberties each of the journalists has taken with the original text in creating his or her paraphrase. In fact, as the various versions are read aloud, our participants may even begin to disagree with each other whether or not a specific word or phrase is an adequate equivalent for the word or phrase in the original.
It is likely that most of their discussion will be devoted to such minor disagreement, most often friendly and collegial in tone. The task of creating an ideal textas neutral in tone as possible while still perfectly representing the originalis one that journalists face each day. It is for this reason that reporters at major publications may, where necessary, rewrite, reedit, and/or "re-tweak" each other's work, always in the hope of approaching ultimate perfection, much as the lonely translatoror the translation editormust do in creating a final draft.
But there is also almost certain to be a second category of divergence present in the journalists' work, one which will provoke somewhat more heated discussion. It may well be discovered that in at least a few instances one or another of these professional writers has committed an outright error of paraphrase, has in fact actually overlooked the meaning of a word or phrase in the original text and replaced it with what can only be described as an incorrect solution.
At this point the purpose of the experiment will have certainly become clear to readers. What we have just discovered while using a single language is so remarkably close to what can happen while translating between a pair of languages as to be for all practical purposes indistinguishable. We all know perfectly well that a group of professional translators sitting around the same table and presented with a paragraph to translate into a second language would surely go through a remarkably similar process. And once their translations were collected and read aloud, these writers would certainly also embark on a similar series of discussions and disagreements.
In other words, the idiosyncrasies of trained writers are indistinguishable from those of trained translators, and vice versa. Whereas society at large tends to accept such variations by journalists, that same society tends to focus on them if translators have been the perpetrators, even to suppose that something in the process has gone seriously astray. Indeed, this plethora of individual variations has not escaped the attention of machine translation specialists, who have in some cases chosen to view them as evidence that human translation is unreliable and must one day be replaced by the trustworthy logic of computer programming.
At this point, a closer look at this experiment can be helpful in providing a few practical examples of what might take place. And since machine translation has been mentioned, it may also be useful to glance at the problems programmers might face in an attempt to improve on the work of human translators. The following sentence, taken at random from a historical text, will play the role of our paragraph to be paraphrased:
In April, 1900, the position of Peking had under these rapidly developed circumstances become so dangerous for foreigners that it was deemed advisable to dispatch a relieving force from the port city of Tientsin.
Now let's see what a paraphrase of this sentence might look like if we change as many of the original words as we possibly can:
By the early spring of 1900, because of these swiftly unfolding occurrences, the situation in the Chinese capital had grown so perilous for outsiders that it was considered prudent to send assisting troops from the coastal town of Tientsin.
We can see immediately that the paraphrase has become longer than the original passage (a common outcome in many translations from English as well), but other differences are also obvious. While the paraphrase conveys the essential meaning of the original, a number of questionable changes have been made, though perhaps only one could be characterized as an outright error. This comes at the very beginning, for by changing "In April, 1900" to "By the early spring of 1900," a certain degree of precision has clearly been lost. Other possible solutions might be "During the fourth month of 1900" or "As April of 1900 began" or "March was barely over when...," but all of these are overly elaborate and/or add something not present in the original. The original text does not claim that this situation arose "early in April" or "during" April, merely "in April."
On the other hand, the wording of the source text does not provide us with sufficient information about the exact time to proclaim any of these variants as being devastatingly wrong. Perhaps a useful rule of thumb in a manual for paraphrasing would be never to change the names of months or days, though there might be instances, as with all rules of thumb, where this too might work less than perfectly.
A similar problem crops up at the very end of the sentence when "the port of Tientsin" is rendered as "the coastal town of Tientsin." Unlike the month of April, there is no way we can avoid using the name of the town . Even though it would be technically accurate to call it "Tientsin, the port city of Peking," this involves inserting a geography lesson into the paraphrase where none occurs in the original. In any case, a "coastal town" is not necessarily a "port city," nor would calling it "the seaside town" or "the embarkation point" help us very much. "Harbor town" might be even less correct, as a harbor town is not necessarily a portthe word "harbor" describes mere topography (ships may occasionally enter a harbor, but no docking system may be present)while the word "port" describes a function and often the presence of complex machinery.
Readers are free to examine the many differences between these two sentences at their leisure and are equally free to decide if they can arrive at any truly preferable solutions. These are likely to be few and far between, even though perfectly valid objections can be directed towards every single element of this paraphrase. "To send assisting troops" is not the same thing as "to dispatch a relieving force," nor can "it was considered prudent" be accepted as a perfect equivalent for "it was deemed advisable."
This intricate conversion has taken place entirely in English, a language which we like to believe enjoys an uncommonly rich vocabulary. Yet no end of legitimate questions can be raised about the overall process of paraphrasing, much as they have been raised about the process of translation. In fact, the very level of disagreement likely to arise among readers of this brief passage serves to prove the point being made-that writing and editing in a single language is no more precise or secure than translating between two languages.
Though some of you may disagree, there is probably no major error as such in this paraphrase, as much as it leaves to be desired. Such an error might have occurred if one of the journalists had substituted "Europeans" or "Westerners" for "foreigners," since during this historical episode (the Boxer Rebellion) Americans and Japanese were also among the combatants. Another error might have arisen if the phrase "it was deemed advisable" had been rendered as "it was judged urgent," since this would introduce a true difference in meaning. It may well be that our languages-all languages-provide us with far less "wiggle room" than we are accustomed to believe. While we are capable of saying almost the same thing in many different ways, not all the synonyms in Roget's Thesaurus will unfailingly enable us to convey exactly the same information in all instancesnot to mention expressive or emotional meaning. Our synonyms may often not be as fully synonymous as we tend to believe.
Although we have been working within a single language, the similarities with translating between two languages are obvious. Clearly no two "natural" languages have ever been constructedwhether in their vocabulary or their syntaxso as to be fully synonymous with each other, which of course explains many of the problems involved in translation, even before cultural factors enter the picture. None of this of course exonerates the translator from attempting to choose the best possible translation for every word in a text, any more than it excuses journalists from seeking out the very best choice among competing synonyms.
7. Discussion: Can EBL Be Taken Any Further?
(At Dartmouth this discussion took place towards the end of the workshop.)
At this point I still haven't made it to what I regard as my most important section, where I will spell out the nature of the basic structural principle underlying language, which I foresee as taking over the role grammar has played during too much of the past, along with the form of cartographic linguistics that can supplement this new structural principle. And I really do want to get to that.
But I also believe it's time to ask for your participation again, and I want to start things off by asking you first of all about your reactions to my session on Wednesday but then as a continuation also whether you believe the idea of Evidence-Based Linguistics may have sufficient merit that it should be taken forward in some practical way. In other words, do you believe that LACUS, or any other entity, should initiate some sort of effort to take the concept of EBL further and by so doing work to challenge and/or diminish the current prevalence of what I've been calling Voodoo Linguistics. I've already been told by a few of you that it's quite unlikely you would in fact want to take any steps in that direction, since I've also been told that the prevailing mood of the group is that there is really nothing practical we can do about the general climate that surrounds us in our field. In addition, I'll also be offering one or two of my own ideas about what could possibly be done to gently and cautiously alter that climate, but I don't want to get too bogged down in this subject matter during our discussion because I have, well, I've already explained why.
So with the hope that we can keep our discussion down to perhaps 20 minutes, I'd now like to open the floor to any questions or comments about the Wednesday session, or simply how you feel about it in retrospect.
(OPEN DISCUSSION: there was general agreement that the notion of Evidence Based Linguistics should be taken forward in some way.)
And now, provided no one objects, I'm wondering if we can move on to the second theme: is there any practical action or any practical goals any of you feel we should be following to diminish the hold of Voodoo Linguistics both on others in our profession and on the general public, to the extent that the public is aware of these views.
(FURTHER DISCUSSION: It was further agreed in general terms that such practical actions should be taken.)
Okay, let me share with you precisely what my own thoughts are about this subject. One of the aspects that bothers me most is the constant press notice and public reinforcement the doctrines of Voodoo Linguistics keep receiving in the media. Every few months it seems that their spokesmen are busy trumpeting another major step forward in their theories, yet further proof of how triumphant and irresistible their ideas must be. Whether it's the discovery of yet another sign language in yet another country or some reasonably well-known oddity of British speech patterns, there they are, the victorious and omniscient experts insisting that any and all such phenomena provide yet further proof of the resplendent innateness and universality of language in every corner of the world.
I served as head of two Committees for the American Translators Association, first their Public Relations Committee and later the Special Projects Committee, so I believe I may have some insight and experience as to how these continuous press triumphs are being generated. It is no accident that these findings, regardless of their significance, are so well publicized. Just as it is no accident that the leaders of this school are so well versed in George Orwell's theories on the nature of propaganda. Orwell meant these essays as a way of detecting and guarding against assaults on public opinion from whatever source, but when applied in reverse they can also serve as the directions for writing quite effective disinformation.
I believe I can assure you that what I'd like to suggest is something fairly conservative and unlikely to cause any kind of deep scholarly confrontation, though it could serve in the long run to reverseat least to some extentcurrent public and scholarly misunderstandings (or perhaps ignorance) concerning those who work in non-mainstream linguistics.
What I'd like to suggest is that someone from this group or another entity get in touch either by phone or email with the principal journalists most likely to write about linguistics and related fields for the New York Times, the Washington Post, Time, Newsweek, and perhaps a few other publications. Perhaps an invitation to a drink or even lunch might be in order. But once one found oneself within personal hearing range of any of these journalists, no hard sell of any sort would truly be needed. One might simply explain to them that there seems to be a great bias in the press favoring one sole school of linguistics. It could simply be suggested that the views of this school are far from being universally shared or even accepted. And a list of the names of dissenting linguists on various topics might even be accepted by these reporters as a welcome gift, since reporting various points of view on any subject is one of most journalists' daily concerns. The names on that list would be likely to include some of the members of this organization. The journalists would then be free to consult them-to consult you-for a dissenting view or not as they saw fit.
Not an earthshakingly radical stance at all, simply assisting journalists to carry out their normal duties. Perhaps you already have someone among you who would be the ideal choice for such a task, or perhaps even more than one such person.
The choiceand how you make that choiceis entirely yours. Assuming you choose to make any choice at all. It just might be a good idea to start something like this fairly soonin less than a year and a half the 50th Anniversary of the publication that launched the MIT-LSA movement will be upon us, and I confidently predict that we will once again be inundated by all the fulsome exaggerations surrounding this school we have grown so accustomed to hearing. I'd now like to open the floor to discussion concerning whether this might be a desirable step for this organization to embark on. And/or if any of you can suggest some other step that could lead in a similar direction.
Mr. Chairman, provided you agree, I would be happy if you would be willing to adjourn the discussion, and I will then proceed with my discussion of the successor to grammar in our field.
8. The Clarification of Context as the Brain's Chief
Language Function: Not Grammar but Linguistic
Cartography and a Unit for Measuring Language
I want to present the cartographical material that follows with a bit less certainty than what has gone before, though I still believe I can present a fair amount of evidence in its favor and that it can prove useful in the future-and perhaps even at present-in helping to advance human understanding and bridging the separations that sometimes arise among both individuals and cultures. It is nothing less than a large-scale attempt to define what language is and how it works. And I will also soon be explaining exactly what the successor to grammar in my opinion is likely to be. I would therefore be grateful if you would all try to put aside previous accounts of our entire field of linguistics, whether based on grammar, or on morphemes, lexemes, and sememes, or on Saussure's Le Signifié and Le Signifiant, and do your best to listen with something like untarnished hearing faculties. I am aware of some imperfections in this account, but I believe that with your aid and comments I may soon be able to correct them.
Some of you may be wondering why I have suggested in my abstract that there can be some connection between linguistics and cartography. But it is precisely this aspect of language which overlaps rather neatly with our conference's other main theme, networks. Almost any map can be converted into a table or data base, just as almost any data base can be subjected to a variety of mapping procedures, and this is of course also true of networks. For me language has always been much more than a mere entity, even more than a process. I have visualized language as an actual place, a territory that contains at least relatively specific locations evolving over time, whose seemingly chaotic and conflicting borders can to some extent be known and mapped. A moving target, to be sure, but one that can nonetheless be predicted, aimed at, and based on one's aim and predictions actually be found.
Can I present any evidence why you ought to believe what I'm about to tell you about cartography? Yes, I think I can. So let me start by confessing that by some unknowable trick of fate, I enjoyed the relatively rare privilege of being brought up among an international family of cartographers, in other words I came to learn the field from the ground up. What follows is in many ways the core of my presentation, so I'd like to proceed a bit slowly and begin by explaining a bit about my experiences as one member of a team of mapmakers.
My father had introduced me to maps almost as soon as I could read, when I was given jigsaw puzzle maps of the US and the world to play with. A few years later he was busy teaching me practical values by assigning me writing and indexing jobs for his various guidebooks. He became a bit alarmed when I moved beyond our planet and fixated on astronomy, during the 'Forties viewed as a totally useless study, and even more concerned when I entered college as a lit major. I had no sooner completed my B.A. degree during the early 'Fifties when he decided that the best way to make me face his version of reality was to exile me to Nassau County to map the many new suburban streets suddenly springing up just outside our city.
I spent an entire summer there and walked up and down perhaps as many as a thousand of those streets. My task was to write down the house number of every street at its intersection with every other street. I was also ordered to make drawings of any unusual curves or turnings in those streets and to visit town and village planning offices to obtain public domain charts for housing projects not yet built. On infrequent visits to our offices in the city, I would also confer with our draftsman, whose job it was to engrave these new streets onto a steel plate, and explain to him the precise shape and position of the new streets I had found. I hope some of you are already beginning to sense the similarities between this process and determining the structure and content of a language.
The main goal of this work was to compile what is called a gazetteer guide to that region, a handy index that could be used by salesmen, policemen, and other visitors to find the easiest route to any specific address. Here is just one sample of such a gazetteer guide-I promise you this will become quite relevant to linguistics in a few minutes-in this case one compiled under my father's directions by my half-sister Phyllis as part of the A to Z Atlas of London:
Weston Dri. Stan. 1E 25
Weston Grn. Dag. 1E 55
Weston Gro. Brom. 1A 122
Weston Pl. N8 4E 31
Weston Pl. SE12C 78
Weston Rise. WC12F 63
Weston Rd. N222F 31
Weston Rd. W42D 73
Weston Rd. Brom.1F 121
Weston Rd. Dag.1E 55
Weston Rd. Enf.2B 10
Weston St. SE12C 78 (3)
In this table abbreviations such as Brom. and Dag. stand for districts in London, such as Bromley and Dagenham. Here's a piece from another gazetteer guide, this one from a rather outdated world atlas: (SLIDE)
Loachapoka, Ala.G5 113
Loami, Ill.D4 124
Loange (river), AfricaF6 63
Loango, Fr. Eq. AfricaJ12 64
Lobatsi, Bech. Prot.D2 68
Löbau, GermanyF3 37
Lobaye (river), Fr.Eq. AfricaK10 64
Lobdell, La.J1 129
Lobdell, Miss.B3 135
Lobeco, S.C.F6 150
Lobelia, W.V.F4 158
Lobelville, Tenn.F3 152 (4)
As you can see from the names of nations no longer in existence, such guides require frequent revisions.
Here is a third such gazetteer guide, which as I will soon be explaining most closely reflects the structure of language:
This one shows at what number on Prairie Street the various streets intersecting it will be found. But it does not show in any way how far any one street necessarily is from the one preceding or following it. To discover this you will need to turn to the map itself, and to understand the reason for any seeming anomalies you might need to visit that neighborhood for yourself. If you do, you may well discover that Cherry Street is much closer to Brewster than Brewster is to Pine, or that alternately Apple Street is much further from Cherry than Cherry was from Brewster. This can happen for a number of reasons, including abandoned or burnt out houses, intervening schools or hospitals that take up more space than one family homes, changes in zoning plans, projected structures that were never built, or hills and declivities in the terrain. A major public highway which appears on the map will not appear in the gazetteer guide because it does not intersect the street but passes over it.
In an analogous way, the precise space between possible meanings of words, whether adjectives, nouns, or verbs, may not be marked by the availability of words to fill each shade of meaning. Words can overlap or even conflict with each other, sometimes we have too many to represent a meaning, sometimes too few, and no language can boast a vocabulary that precisely reflects all possible meanings, simply because it is the language itself that determines which meanings are most essential. And no two languages, as you know, do this in precisely the same way.
Here too I hope you will begin to sense a certain family resemblance between networks of place names and of lexical items. We were of course allowed some leeway in our mapping process, simply because no map can ever be perfectly accurate, any more than can any translation between languages or any paraphrase within a single language or for that matter any dictionary. But what we were not allowed is far more important than what we were. We were absolutely not permitted to make up our own street names or shapes or locations, and it would never have occurred to any of us to limit the number of the types of streets that could be mapped or develop ornate or sublime metatheories about what the purpose of a map might be, much less actually transform such theories into mentalistic, formalistic chimeras. Moreover, every step we took in creating our maps came close to observing scientific method: research and/or observing the evidence, conferring with others about this evidence, and integrating it into existing knowledge.
Which brings me to the never-ending use of those famous sacred diagrams invoked by several generations of linguists. Here's one of them (SLIDE):
I suppose they have a kind of totemistic, fetishistic value, but when my wife first laid eyes on one, she couldn't stop laughing out loud. "Why, they're nothing but Calder mobiles!" she exclaimed. And here, with a bit of help from Ilene, is what she saw in them, a genuine example of the Syntacto-Mobile: (SLIDE)
These diagrams, since I believe I'll soon be showing you something a bit more revealing, have very little value in elucidating the structure of language. I wouldn't be surprised if they end up in history next to quaint phrenological maps of the human brain or charming illustrations of the language of flowers or theories coordinating each color we see with a musical note. This is because they are essentially based on nothing more than grammar, and grammar is simply not the basic underlying structure of language. As I am about to show you, and of course I will be presenting at least preliminary evidence for this statement. I strongly suggest that the following slide will provide you with a far more useful guide into the basic principles of how both language and a significant part of human cognition are likely to work. And once again I can show you some evidence to back it up.
Here, in this page from my work in progress, you see illustrated what I believe the mind does when it hears or sees in print the word "lock," which as you know can have at least four possible meanings. (SLIDE)
Immediately on hearing or seeing this word, our minds look and/or listen to determine what other words we can find being used around it, either in the same sentence or a nearby one.
It's something like the concept of a "proximity search" as used in natural language processing. If the mind hears the words key, locksmith, open, close, broken, locked, unlocked, force, jimmy, door, gate, for example, then it immediately concludes that it is dealing with that kind of lock. It's important to notice that nouns, verbs, and adjectives are all jumbled up together in this assemblage, with no real concern for grammatical roles. The purpose of such a list is to determine that we are dealing with the kind of lock that goes in a door.
While many axons or neuromolecular subroutines may be at work to support this conclusion, this is in fact the nature of the act of cognition that takes place, and I believe that Sydney Lamb has correctly described the abstract process involved in such disambiguation in his own work. In this context, grammar is merely an inefficient add-on with no real consistency between one language and another. The real work being done is in fact described by the first, second, and sixth draft laws of language and linguistics.
If on the other hand the mind hears or sees in print such words as raise, lower, sluice gate, Panama, Suez, water level, shipping, tonnage, or cargo ship, the mind then concludes that we are probably dealing with a lock in a canal.
But what if the mind hears or sees some of the following words: camel clutch, Boston crab, scissors, headlock, hammerlock, leglock, sleeper hold, strangling, smothering, pain, pin, submit, unconscious? Such a mind is likely to conclude that the topic involved is wrestling.
Now let's take a slightly more recherché example and assume that the surrounding words are cut, rape, Belinda, Pope, or tress. Here it turns out that no strange Papal crime is being committed, rather we are dealing with a lock of hair, most probably as part of a discussion of Alexander Pope's famous satiric poem The Rape of the Lock.
Now some of you may object that the last example could only be found in highly informed and literate circles. But the truth of the matter is that the other three examples are equally the product of highly developed and specialized cultural knowledge, only a bit less obviously so. As I said on Wednesday, a great many wordsfar from representing eternal or universal veritiesare in fact culture- and time-dependent artifacts. First of all, locks in canals are a relatively recent development, as are canals themselves, despite the efforts by the Egyptians in 1300 B.C.E. and the Grand Canal built by the Chinese Emperor Yang Guang in 604 C.E. (Note 2)
But there certainly was a time when canals did not yet exist, there are areas of the world where canals are still unknown, and there are doubtless people in our own culture who haven't a clue what a canal lock is. What's more, though various cultures have employed locks on doors, there have been other cultures that have lacked doors to put them in. So this meaning has also not necessarily existed in all cultures at all times, any more than the object itself. We like to assume the word lock is simple, but like many words in all our languages it contains a great deal of what I call "embedded process," encompassing diachronic, social, and cultural development over time and many languages.
I suspect we're all aware that just as light may bend while travelling through space, so human cognition can actually bend and change as it passes through time. And cognition often becomes partially transformed as it passes into other languages. No language other than English uses this same homonym "lock" for all four meanings, which means that the French or German brain will need to sort through quite different choices.
So let's get one thing quite straight. The basic linguistic process is not grammar at all, which can at most be viewed as a supplemental system frequently ignored by those who use language. What's more, I believe I can show you that much the same process is at work even where no homonyms are involved at all. It is in fact this network of systematized disambiguation that runs language.
Disambiguation is a long and clumsy word, so I prefer to call it the clarification of context. If I were speaking French, I suspect I would probably call it le débrouillement de contexte, even though two dictionaries tell me that débrouillement is now rarely used. Even so, the French simply love the verb débrouiller, which means roughly to "defogify," and also the noun or adjective débrouillard, which simply means clever, someone who knows how to defogify, to cut through the fog and the thicket of language and get to the real meaning. And even crosses over into the practical domain, where the French word can mean just knowing how to get things done.
The English also pay tribute to this process when they talk of "muddling through," which means not so much defogifying as getting along despite the fog. And that is what both we and our brains spend most of our time doing, defogifying language and trying to extract the clearest meaning possible. Given the time limits in everyday conversation, it is often a necessarily quick and dirty process, as I have mentioned in the Sixth Law. And often enough we also make mistakes. But it is through these mistakes, if we analyze them, that we can actually peer in at the workings of the brain and grasp what they must be doing, even if that activity on the neurocellular level evades us.
If this were an LSA conference, I'm sure some of my audience might want to fall back into comfortably cushioned thought patterns inside their minds at this point and claim this system is merely warmed over semantics. But it probes much further and far deeper into the nature of language. All semantics has come to mean over time is that we sometimes disagree about the meaning of words. But what I'm talking about here is a tool that can delve far more incisively into both the structure and the development of language, physiologically, synchronically and even diachronically.
I realize of course that some of you here have been wedded for so long to the notion that grammar comprises the true structure of language and have indeed been working for years and even decades to find a way of proving this thesis. I therefore want to offer my sincerest apologies for advancing what may seem to you an altogether unlikely alternative. But as dismaying as you may find such assertions to be, I must confess that I was equally dismayed when I first began my readings in the history of linguistics and discovered that simply because detailed systems of grammar had been devised by the Indians and Greeks and Romans to describe their languages, more modern linguists have accepted quite uncritically such explanations as the only ones worth examining.
Now I'd like to go straight to the crux of my topic and explain what I mean by the unit for measuring language I mentioned in my abstract. Now the moment I mention measuring unit, I can well imagine that at least some of you will be having one or both of two different reactions. How on earth can there be a measuring unit for language, you might well be asking, since both its spoken and written forms are so remarkably irregular? I'm also busy wondering if a few of you may not be harboring some extremely misleading preconceptions about what a measuring unit has to look and sound like. Quite likely, you expect a measuring unit to be precise and unvarying, so precise and unvarying that it can readily be used to measure any quantity at all of whatever it is that is being measured, be it a foot, a quart, a kilo, an astronomical unit, or a light year. After all, what else can a measuring unit be but precise and unvarying?
Actually, as can so often happen, reality is about to intervene and decisively challenge this assumption. One of the most effective measuring units I've encountered, and also one of the best known throughout the world, is in fact the Chinese Inch, or tsun, which canat least theoreticallybe slightly different for every single person on the planet. It is the precise length of the second inside joint on your (or anyone's) middle finger when you join that finger and your thumb in a circle (demonstrating). And since most human bodies are built proportionately, this distance is almost unfailingly precise in making measurement for and on each human being. It is used to measure the position of acupuncture or massage points (or attack points in the martial arts).
I'm assuming it's permissible to introduce a term, a concept, and a tool from Chinese medicine into our scientific discussions. This would not have been the case ten or fifteen years ago, and twenty-five years ago I had an interesting encounter over this issue with a remarkable defender of Western science named James Randi, better known as the stage magician the Amazing Randi. On that occasion it was I who challenged Randi, rather then vice versa, when he made the assertion in a national magazine that all of Chinese medicine was superstitious nonsense. In the end it was Randi who conceded he was mistaken and sent me a gracious apology.
But to get back to the tsun, it is of course quite different for small children than for adults, In other words, it is a truly human measuring unit since it grows as we grow. I believe it is therefore a unit like the tsun that is most effective for measuring language, since language grows as we grow. When we are children, we know comparatively few words well enough to distinguish between them, and we recognize relatively few phrases or collocations or concepts. But as we grow older we gradually become acquainted with a growing number of words, phrases, collocations, and concepts, and what will count as one measuring unit can easily expand in size as we grow more able to organize and mobilize our understanding.
And parenthetically, come to think of it, this seemingly exotic notion is not really so exotic after allwe Westerners do something remarkably similar with kids' shoe sizes or with children's clothing sizes in general. Most important, both the Chinese inch and the standard for language I am about to propose share one outstanding quality: they are commensurate to human beings. Unlike the metric system and even unlike some older non-metric units, they are not an attempt to impose a formalistic, abstract, arbitrary standard onto people, rather they spring out of our very nature as human beings.
So this is essentially the second leap that I am asking you to take with me, to accept that a somewhat variable measuring unit can nonetheless be used to measure somewhat variable areas. It is neither a fuzzy unit nor an imprecise one, any more that kids' shoe sizes are fuzzy or imprecise, it is merely a commensurate unit. If anyone here still finds this leap a bit difficult to take, with all respect I'd like you to consider the following. It's hard enough for most people, myself definitely included, to handle any new combination of words, at least the first time we encounter it. But the leap I'm asking you to take isn't just a new combination of words, I believe it just might qualify as a new conceptualization of how all new combinations of words fit together, perhaps even a part of a new linguistic theory. So it could just be that some of the forces mentioned by Kuhn in his scenario for paradigm shifts might be at work here.
It is not a perfect or precise measuring unit, but then neither are our languages either perfect or precise. Nor are a number of the units used by physicians to measure various physical processes inside out bodies, and frequently enough we find new measuring devices introduced that impose slightly different scales of measurement. But most sciences have developed such measuring units, and it seems to me that it could be quite useful if we linguists could come up with at least some rough form for such a unit as well.
First let me give you an example of such a unit in action. Let's suppose that I've just phoned you up and asked you to meet me in New York at the corner of Fifth Avenue and Thirty-Third Street. "Yes," you reply, "I'll be there, at the corner of Fifth Avenue And Thirty-Fourth Street." I immediately correct you and say "No, not Thirty-Fourth Street, Thirty-Third Street."
The amount of mental energy it requires for me to make this distinction and for youor any other personto register that you have grasped it, that precise degree of language comprehension is equal to one of my basic units for measuring language. Essentially, its purpose is to define and measure the smallest and most irreducible element of meaning.
Its size and shape will vary according to the context, not to mention the age of the language user and the level of that user's vocabulary. Let me remind you again that I am not referring to lexemes or sememes or morphemes, which are indecipherable as such units for most people and not much more useful than etymology, but the words and phrases we use in an everyday manner in everyday conversation.
Other such examples would be:
Not chocolate but vanilla.
San Francisco, not Los Angeles.
I said buy, not sell.
The Baltic nations, not the Balkan nations.
If you prefer a definition more in keeping with the terminology of physics, let's call it:
"the cognitive energy needed to move one basic burst of communication from one mind to another (or to other minds equally attuned), assuming close-to-ideal contextual conditions."
This can vary considerably according to the contextual situation, as I will soon make quite clear. Now that I've explained what this unit is, I would like to provide it with a name: (SLIDE)
WITT PLANE WITT PLANES
W.I.T.T. = Wordlike Increment of Terminological Territory
I call this unit the Witt, and I have clearly done so because its name is reminiscent both of Watts and bits, crucial measuring units in both electricity and computer science. It obviously also bears a close similarity to the English words "wit" and "wits" and also recalls the word "whit," meaning quite appropriately an "iota," "a very small bit." Finally, just as the name "Watt" commemorates a major innovator in the history of science, "Witt" can also be seen as a posthumous tribute to an important pioneer in the attempt to link language and math, namely Ludwig Wittgenstein, who I would like to believe would almost certainly have approved of what follows. If anyone still objects to such a neologism and insists that truly scientific names can only be based on acronyms, then it could also be claimed that "Witt" stands for "Wordlike Increment of Terminological Territory," though this arises largely as an afterthought.
Witts do not function in isolation, but rather cluster together in "planes" with related elements of meaning and word forms, from which they derive their own identity. I've already presented you with a few simpler Witt Planes in my discussion of the word "lock" (previous slide is shown again), for instance as you can see, the words key, locksmith, open, close, broken, etc., and so on for the other meanings of lock, together defining a separate Witt plane. And so on for all the other words in our language, all of which live on one or more of their own Witt Planes, and it is through their predictable existence on those planes that we are able to disambiguate and clarify what we hear and read, permitting us to achieve various levels of cognitive competence. As we know simply from observation, gaining such cognitive competence is a gradual process, one that usually but not always favors the older and better educated among us over those younger and less well educated. We learn by trial and error, we also learn quite differently in different languages according to their systems of Witt Planes, and no magical innate gift for language opens the way to anyone to achieve such competence.
But what can possibly be my evidence for any of this? Actually, I can present four excellent examples of evidence. First of all, I am pleased to tell you that I have communicated with another scholar, Price Caldwell, Professor of International Commun-ications at Meisei University in Japan, and he too has developed a theory confirming the existence of Witts as the major building block of language. Some of you may be pleased to learn that what I call Witts he refers to as molecular sememes and even more pleased that both in his treatment of this subject and his background as a linguist he possesses all the conventional academic credentials that I lack. And yet we are both in substantial agreement that in our separate vocabularies we are talking about essentially the same realities and that whatever differences may exist between our two systems may not end up being that significant.
What is important is the relatively relaxed view we both take of a complex subject where the views of so many others in this field are so stressed and strained. Professor Caldwell and I also agree that current problems in linguistics go back further than the MIT-LSA faction to embrace some key concepts advanced by both Jakobson and Saussure. I might even extend this a bit further to include the Junggrammatiker, whose quest for universal laws of phonology may have spilled over into today's obsession for so-called universals.
And getting back to cartography, even more persuasive as a form of evidence has been the persistence over the past 150 years of a major reference work whose contents are remarkably similar to the excerpts from atlases I presented earlier. It consists of literally thousands of word lists remarkably similar to the following:
make ready 669vb.
church service 988n. (2)
This book, which essentially presents a gazetteer guide to the entire English language, is of course the Thesaurus of English Words and Phrases, classified and arranged so as to facilitate the Expression of Ideas and assist in Literary Composition, compiled by Peter Mark Roget and first published in 1852. In many ways this work stands as the Mother of all Networkscontrasted with those developed over recent decades, it is the only one that has withstood the exacting challenges hurled at it by professional writers, editors, and translators over the last 153 years. What we have just stumbled across here is in fact all the evidence we could possibly wish for, and every bit of this evidence confirms the existence of the many separate but interlocking Witt Planes.
Of course not all of Roget's entries constitute valid Witt Planes, but taking this work together with the Oxford Dictionary of English Collocations and some similar reference works, it could become possible to create a valid mapping procedure-or perhaps several such mapping procedures-for all of our languages.
A third remarkably large body of evidence readily arises from the many errors committed by junior and senior high school students in their attempts at English composition, for instance:
Amenable-anything that is mean.
Assiduity-state of being an acid.
Three types of blood vessels are arteries, vanes, and caterpillars.
If we needed any further proof that a relatively error-prone, quick and dirty matching operation underlies human understanding, it is surely to be found in such endlessly recurrent errors as these. This is how we all learned and to some extent how we all still learn, and I have caught both myself and others routinely making such errors in English, Chinese, and other languages.
This process of making errors is open to measurement under the Witt system and provides further proof that the first, second, and sixth laws of language and linguistics are both valid and active. Those who dismiss these errors as mere educational deficiencies, correctable by stricter methods during "the old days," might wish to reflect that these very examples, similar or even identical to those occurring in our schools today, were in fact produced during "the old days" and have been selected from Mark Twain's 1886 essay English As She Is Spoke.
A fourth class of evidence can be readily provided by playwrights, dramaturgs, and others who write dialog for film, TV, or radio. Since I have played three of these roles, I can attest that the entire art and craft of writing dialog is a continual exercise in avoiding the perils of the Sixth Law. Lines of dialog must be exceptionally clear and unambiguous, since those watching or listening have only one chance to understand themwhen they are first spoken. Unlike printed languageor even printed dialogthere is no way to make a second try, of being able to hear or see them again. In other words, language using professionals in these fields necessarily possess or develop the skills to "defogify" language and steer their dialog safely through the rocks and crags of the Sixth Law.
While I can't refer to either of the following two examples as evidence, I wonder if they may not to some extent serve as precursors to the theory I have been attempting to articulate. It could just be that the English philosopher of language J.L. Austin was himself working in the direction of a similar insight. According to wikipedia.org, Austin:
"proposed some curious philosophical tools. For instance, he uses a sort of word game for developing an understanding of a key concept. This involved taking up a dictionary and finding a selection of terms relating to the key concept, then looking up each of the words in the explanation of their meaning. Iterate this process until the list of words begins to repeat, closing in a `family circle' of words relating to the key concept."
Someone else who I believe may have been skirting close to this approach is the AI researcher Roger Schank in his theory of "scripts," which he thought might facilitate the development of NLP and machine translation. He believed it possible to construct scripts for many life situations containing the key words most frequently encountered in each script. Unfortunately, he could not conceive of how incredibly great the number of scripts might turn out to be, nor could he devise a reliable method to determine where one script ended and another began.
It is these Witt Planes which constitutefar more than any grammar or syntax ever canthe true internal structure of all our languages. And these Witt Planes are structured differently for every single language, following the network of images and analogies, the individual weave of usages and euphonies each language has used to build up its knowledge system. Such a construction plan necessarily rules out the possibility of any so-called "universal grammar" and provides yet further reasons why research into machine translation may have been doomed from the start.
The most basic rules of any grammar for at least some languages can be learned in a few minutes, though their full application can take far longer to realize. But the crucial mastery of these Witt Planeseven in a single languagecan take us all our lives to achieve.
I hope none of you will fail to see the family similarity between these gazetteer guides (shifting back and forth between three slides) and Roget's plan of our meanings. Just as our city atlas provides the location of various street and the world atlas of many nations, so Roget's Thesaurus tells us what nearby "streets" lurk at the intersection of "prime" with "numerical," "prime" with "important," and so on.
Now it this were an LSA conference, I believe I would have provoked a certain amount of heckling by now. After all, Peter Mark Roget was a mere literary amateur of language, I might be told, little more than a language maven, totally lacking in scientific background and the rudiments of computational linguistics.
But this of course would not be true at all. Not only was Roget a medical doctor, so that his work with its six major classes and his 990 categories can stand by itself as something like a comprehensive differential diagnosis of English, but he also qualifies as a major computer pioneer. A great deal about his life sounds positively geeklike: Roget invented a slide rule, wrote a regular column on chess problems, and even created the first inexpensive chessboard. Perhaps most intriguingly, he worked long and hard during the 1840s on inventing a calculating machine. In this endeavor he was active at the same time as Charles Babbage, the better-known English computer pioneer, and even co-authored a report on English science with him.
Let's summarize all of this slightly by throwing in a few more numberssince I know some of you, despite everything I've been sayingstill prefer numbers to words, all based on this New York subway example (though all could readily enough be transposed into appropriate examples for other cities, nations, and cultures).
2 Witts = not knowing where Fifth Avenue or the nearest subway stop is located
5 Witts = not knowing which subway train will bring you to the general area
10 Witts = not knowing how to find that train, or a connecting one, in your own neighborhood
100 Witts = not having been in Manhattan for a few years and forgetting how the Avenues run
300 Witts = having never been in Manhattan before (and at this point a new Witt Plane has probably been entered)
500 Witts = having never been in an American city before
1,000 Witts or one Kilowitt = having never been in any city before (and in both this case and the preceding one, yet other Witt Planes have obviously also been entered)
While these numbers are approximate ones, they are also quite useful both in explaining how Witts and Witt Planes work but also in getting you to your appointment with your friend.
In any conversation we may happen to have, we are as often as not divided from the other party by a distance of one Witt or more. In fact, it would be no exaggeration to suggest that most conversations between a pair of speakers essentially revolve around shortening that distance and/or clearing up that difference of one of more Witts.
Let me show you one or two other tricks you can play with Witts, and then I'll move on. I've mentioned that this measuring device can move from the merely synchronic into the diachronic and can also expose how completely some of our most established words are in fact cultural artifacts, and I'd like to illustrate this with a fairly dramatic example.
Here it is. Let's assume that you are holding a "two by four" piece of lumber, which you wish to saw into two equal sections.
As everyone attuned to our culture and our use of lumber knows, a "two by four" is invariably eight feet long, though both the "two" inches of its breadth and the "four" inches of its width have been atrophied by greed and/or "economic forces" into something less than their full dimensions. By maintaining that you wish to cut this board into two equal pieces, what you most probably mean is that you want to end up with two four-foot-long segments.
You will need to choose the proper saw for this task, which for most beginning carpenters might seem to be a simple choice between "cross-cut" and "rip." But you are actually about to take a trip through a fair portion of recent human evolution. Here, assigning each of the underlying Witt Planes involved in this choice an arbitrary descending value based on our decimal system, is what you are actually reenacting when you perform this task:
10,000On this Witt Plane, the usefulness of having boards of standardized size is recognized (through history, only a few societies have agreed on this).
5,000On this Witt Plane, schemes for using these boards to build houses and other structures have become commonplace.
2,500On this Witt Plane, the means for providing such boards to a broad public has been evolved: lumberjacks, lumber mills, lumber stores, delivery services, etc.
1,000On this Witt Plane, ordinary citizens like you and me have gained the skill to recognize the direction taken by wood grains and to understand why this direction is crucial in making a clean cut with a saw.
800On this Witt Plane, ordinary citizens like ourselves have also gained the ability to distinguish a "cross-cut" saw from a "rip" saw and the knowledge to choose the right one for a specific purpose. It is also assumed that we possess at least one of each kind of saw.
600On this Witt Plane, it is assumed that we have gained the skills for making precise measurements and possess the tools for doing so, namely a straightedge and a pencil.
400On this Witt Plane, it is assumed that we have gained the knowledge of how to hold these tools and have perfected that knowledge over time (this also is a learned process not necessarily shared by all peoples in all societies).
If you're wondering how far this can go, the answer is far enough.
200On this Witt Plane, it is assumed that we have also gained the knowledge of how not to use a saw, straight-edge, and pencil: carelessly or foolishly, or perhaps as weapons or toys.
100On this Witt Plane, it is assumed that we possess a carpenter's vise or other tool for holding the board steady as we saw it.
80On this Witt Plane, you and I are permitted to make the correct measurement on our board and draw our line upon it.
50On this Witt Plane, we need to check that the preceding step has been carried out correctly.
30On this Witt Plane, we actually choose between the cross-cut and the rip saws, choose the former, and double-check the shape of the blade to make sure we have chosen correctly.
20On this Witt Plane, we finally place the saw on the line we have drawn, make sure that our body is lined up in an adequate position for sawing, and saw the first stroke into the wood.
10On this Witt Plane, we check that we are continuing to saw in the right direction and then go on sawing. We have now left the level of Witt Planes and are descending through the ranks of individual Witts, whether of language or knowledge or action.
8, 6, 5...1On each of these Witts we continue to saw, we do our best to keep ourselves well-balanced, but we also remain alert to the possibility of knotholes or other irregularities in the wood that may affect the angle of our sawing.
Provided you and I have faithfully fulfilled every single one of these preconditions and do not splinter the wood on our last stroke, we should now be in possession of two four-foot segments of the original two by four. And we have also been successful in recapitulating at least several centuries of our cultural history.
Once we understand how Witts and Witt Planes function, we have the basis for a powerful system capable of measuring not only language but communication and even knowledge itself. In a very real sense, all of language and human knowledge is essentially a structure of Witt Planes, each composed of numerous individual Witts. And all branches of human knowledge-at least those that are not totally physical or sensual in nature-can also be visualized as finer or more extended Witt Planes, each with its own language adjunct to record and represent it.
It is obvious that no one can ever claim mastery of all Witt Planes, any more than anyone can ever master the totality of knowledge, even within a single language, as it merges imperceptibly into such sheer immensity.
Now I realize that this may sound far too vast to some of you. But what I'm trying to do here is present at least a way of beginning to visualize it. With the addition of one closely related concept, we can in fact begin to place at least some sort of numerical handle on this knowledge. And by so doing we can also start to measure and explain many of the problems that arise in dealing with both language and knowledge.
Based on our likely order of mastering them, the various Witt Planes which compose human knowledge can be assigned to one or another Witt Level. As with Witt Planes, not all human beings will learn or experience all Witt Levels during their life times, though they are likely to encounter others who have experienced many of them.
These Witt Levels can be assigned rough numerical values by means of an analytical structure called the Witt Scale, comprised of ten values ranging from one to ten. It is modeled after the Richter Scale for measuring earthquakes, with each number being exponentially more powerful than the one beneath it. It should be added that differentials between these Witt Levels are not usually as destructive as earthquakes, though they can under certain social conditions prove quite disruptive in their own right.
The Witt Scale is also fairly easy to understand and in some ways reflects our current Western educational system, though an analogous but different scale could be superimposed on other cultures than our own, and even-at least theoretically-on societies of the future or on other planets.
It follows here in its entirety as most of the conclusion of this section. I will let listeners decide for themselves to what extent it clarifies the close relationships between these concepts and our everyday lives.
The Witt Scale
WITT LEVEL 0 (Zero).
Begins shortly after birth and encompasses all language usage and vocabulary learned in infancy, childhood, and early adolescence. Its values are expressed by decimal numbers lower than 1: for instance, various stages of "baby talk" might rank as .18, .21, or .24, the speech and knowledge level of a nine-year old as .5, while the language of a young adolescent could be assigned .7. As later adolescence is reached and language begins to merge with adult vocabulary, numbers over .85 may be assigned. In American culture, roughly equivalent to preschool, elementary, and junior high school vocabulary and learning.
WITT LEVEL 1.
The vocabulary for immediate survival and basic enjoyment of life, including entertainment & sports, has been largely mastered. Stages within this level may be assigned such values as 1.23, 1.47, or 1.63, depending on progress in mastering knowledge and vocabulary in specific areas. Values of 1.85 or higher can be assigned for the beginnings of knowledge belonging to Level 2. Roughly equivalent to late adolescence and the last three years of high school.
WITT LEVEL 2.
The specialized vocabulary and knowledge for advanced understanding and enjoyment of life-including arts, basic sciences, and social attainments-have been studied and perhaps mastered. Values such as 2.25, 2.56, or 2.83 may be assigned, depending on skills in various areas. Learning basic computer skills may also play a role. Values of 2.85 can be assigned for the beginnings of vocabulary and knowledge belonging to higher levels. Roughly equivalent to an American college education.
WITT LEVEL 3.
Basic technical knowledge and vocabulary for attaining professional and technical goals, often with a greater emphasis on computer skills. Not necessarily higher than WITT LEVEL 2 but assigned a different value to distinguish it from the latter. May be undertaken as a separate course of studies or combined with college studies. Typical values assigned might be 3.35, 3.52, or 3.22, depending on attainments. Values higher than 3.85 can be assigned for the beginnings of vocabulary and knowledge belonging to higher levels. Roughly equivalent to technical training after high school.
WITT LEVEL 4.
Advanced technical and professional vocabulary and the knowledge it confers, including the ability to teach these subjects to those at lower levels. Will most likely be gained as post-college "on-the-job knowledge" or as part of graduate training projects. Values such as 4.21, 4.46, or 4.64 may be assigned, depending on attainments. For most who proceed to this point, this is the highest level of vocabulary or knowledge they will attain. Values of 4.85 or higher will rarely be assigned unless subjects later entry to WITT LEVELS 5, 6, or 6A is planned. Off the map for many academic fields.
WITT LEVEL 5.
Highly specialized vocabulary, knowledge, and technical training, mainly limited to medicine, pharmacology, and law.
Values such as 5.20, 5.45, or 5.63 may be assigned, depending on attainments. Many specialized studies for assisting those at lower levels. Values higher than 5.85 would be assigned only to those going on to WITT LEVELS 6 or 6A. There can be a great many parallel Level 5's, corresponding to various specialties.
WITT LEVEL 6.
Advanced specialized knowledge and language in medicine, pharmacology, or law, especially for the purposes of teaching such fields to those at WITT LEVEL 5. Values such as 6.15, 6.37, or 6.58 may be assigned, depending on attainments. Values higher than 6.85 might be assigned only to those beginning to learn the vocabulary of another WITT LEVEL 5 study or proceeding to WITT LEVEL 6A.
WITT LEVEL 6A.
Advanced computer theory and applications. Mainly included because this field is still so new and has links with so many other fields. Values assigned as in WITT LEVEL 6. Values higher than 6.85 assigned only to those learning vocabularies of other WITT LEVELS with a view to creating applications for them. Prognosis for this field unknown, as it is largely off the map of previous knowledge.
WITT LEVEL 7.
Study of vocabulary and technically correct knowledge of extinct languages and their cultures. Possible only in limited terms and cannot in any ultimate sense be accomplished, as the widespread disagreement among scholars in these fields tends to prove. Values difficult to assign, but would probably be in the 7.2 to 7.3 range.
WITT LEVEL 8.
Knowledge and vocabulary of languages of the future. Cannot be accomplished, but should be included as an extension of the premise advanced by the school of language study known as "Glossematics," as suggested by the Danish linguist Louis Hjelmslev (along with the following two levels as well). No values can be assigned.
WITT LEVEL 9.
Knowledge and vocabulary of cultures from "other worlds." Cannot be accomplished, but must also be included for the same reason explained in WITT LEVEL 8. Probably more than one WITT LEVEL should be reserved for such languages-assuming humans could ever learn any of them, there is no reason to suppose that they would all share any one language structure, any more than languages on earth do. No values can be assigned.
WITT LEVEL 10.
Language and vocabulary related to the nature and functioning of the universe. Level of "Virtual Deity." Cannot be accomplished for obvious reasons, but should be included for the reason explained under WITT LEVEL 8. No values can be assigned.
Now I recognize that this may all sound a bit grandiose to some of you, especially that last part. I'm not usually a great believer in "deities," but I thought it just might be a good idea not to exclude it. And I hope it will lend at least some credibility to what I may be saying about deities later. Believe me, if I've hurt anyone's feelings by any of this, I certainly apologize. All I was trying to do was satisfy the reasonable request that might arise for some sort of numerical representation of these Witt Planes and Levels.
By this point some of you must be asking, "But what do Witts do anyway?" What possible application can they have? Let me say only this for the present. The Witt system can not only help to measure language and even knowledge, but it can also help to explain why both our language and our knowledge do not always work. Those who attempt to communicate with someone either one Witt Level higher or lower than themselves are going to encounter problems. If that person is located two Witt Levels higher or lower, communication-the passing on of language or knowledge-may be impossible. This factor alone has profound consequences for all relations between individuals, groups, nations, and cultures. Over the ripeness of time, it is possible that Witts can provide a unit capable of measuring the breadth, depth, and scope of people's minds, replacing the dubious achievement of psychology known as the IQ test.
I hope no one will find this too painful, but I'm afraid I really do believe that the scholar should be "humanity thinking." In other words, since languages are spoken by almost all people regardless of background, nationality or religion, it seems to me that it is our duty as language professionals to provide them with all possible assistance in using their languages as constructively as possible and in mastering the techniques to avoid needless conflicts caused partially or sometimes even totally by confusions arising from language. A whole previous generation of linguists, including Whorf, Pei, the younger Hayakawa, and even Sapir, saw nothing wrong with writing as simply as possible about their subject so as to make it more broadly accessible to the general public.
It is only the most recent members of our profession who have preferred to take refuge in what with all possible charity can best be described as pretentious obscurantism and to look down on their elders for making an attempt to go beyond it. I believe the growing vacuum in American civilization and ideas may be one of the results of this attitude, for at its best the information which linguists could provide about many human and social problems could be truly central and authoritative in influencing any number of discussions. In fact, I believe this stance by the last three generations of American linguists represents a form of treason fully comparable to La Trahison des Clercs, Julien Benda's famous book and phrase of the 'Twenties and sometimes translated as "The Treason of the Intellectual Classes."
9. Fractal Linguistics
When I first began thinking about this topic, it was because I was trying to answer a question I was wondering about myself: precisely how large can a language be? In some ways this question is similar to one the mathematician Benoit Mandelbrot asked back in 1967: how long is the coastline of Great Britain? The answer he came up with was that it could be almost infinitely long, depending on how you measured it, because it was so irregular. If you simply laid down a straightedge along the main lines of the coast, you would get one measurement. But if you actually placed a yard stick or even a one-foot or six-inch or one-inch ruler along all its irregularities, it became remarkably longer.
He called this new way of looking at the coast line a fractal dimension. In some ways, as I will be trying to show you, language possesses a fractal dimension as well.
Let me start by observing that according to some authorities there are five million different species of flora and fauna on our planet. The name of any one of these species, even though it is in Latin, can qualify as a noun in any English sentence. So that's five million possible words right there.
But what if through some fluke we discovered a whole new life-supporting planet nearby and found another five million completely different species awaiting us there? Does anyone doubt that we would, subject to reasonable delays, end up bestowing another five million names on all those as well?
Now some of you may object that we don't truly need those other words. But at least we would possess them, we could plug them into the language whenever we just happened to need them. But that's only the beginning. There are in fact a great many such language plug-ins available to us.
For instance, mathematicians like to boast that they are the only ones who deal with true infinities. But there is a serious problem with this claim. Any number you might care to name, say four-comma-three-six-eight-comma-seven-five-two (4,368,752), which I'm looking at in its numerical form before me, can also be said in words and qualifies as a word in itself, quite separate from the famous first ten numerals that form it:
Four million three hundred and sixty-eight thousand seven hundred and fifty two.
So language has already entered the upper brackets for the number of words it can contain. And all the more exotic numbers so beloved by mathematicians-decimal, negative, irrational, real, complex, or imaginary-can also be phrased as words. And oddly enough, when mathematicians come to teach their field to others, they also make use of relatively ordinary English words to do so, though on a somewhat more exalted plane than ours.
Now some of you may argue that these are merely specialist terms and not true words at all. The problem there is that almost any level of language we use can be described as "specialist" compared to another less specialized level above it. For instance, the way an experienced carpenter describes wood, as compared to how you might describe it, assuming you were both talking about putting in a new door. Plug-ins are a crucial element in language, and as we grew from children into adolescence and adulthood, we came in each case to substitute new and more precise sets of plug-ins for older ones. I'll come to that in a moment.
All of this relates quite directly to how large a language can become, because there's yet another way that any language may prove to be vastly greater than the British coastline is long. That's simply because any one particle of that rough physical shore can interact at best only with itself and a few immediately adjacent particles.
Language is quite different. Absolutely any piece of a language-a sound, word, phrase, sentence, paragraph-can interact, at least theoretically, with any other such piece of the language located anywhere in the rest of the system. All you have to do is place them together in the same sentence.
This leads us to a level of vastness which I believe even mathematicians have not begun to contemplate and may even also help to explain why the methods of computational linguistics have thus far proved so ineffective at machine translation and other language applications. The so-called "syntactic analysis" frequently underlying these methods is simply too primitive to describemuch less transcribethe intricate functions of language. And many of both our computational and our theoretical linguists have simply not been able to comprehend this remarkably simple truth.
There is also a very simple reason why theyalong with members of the general publicare unable to comprehend it, and I believe it furnishes one further proof of the fractal nature of language. Most of us inhabit a relatively small area of language (or of Witt Space, if you will), an area that is both unusually regular and confined, and we rarely leave it. In some ways it is similar to inhabiting one of the remarkably well-ordered shore lines to be found along the sides of the Mandelbrot Set.
But almost any departure from our regular linguistic routines is nonetheless capable of hurling us into the sheer chaos to be found in the Set's outlying areas or even into the black hole of the basin lying at the Set's center. We find ourselves veering in either direction whenever we stray away from the ordered language patterns that surround us. And this is indeed the reality of language we encounter when we are forced to deal either with a foreign language or an unfamiliar subform or dialect of our own language. This explanation might also help to explain why computational linguists have spent the last five decades skirting around the edges of language without ever managing to penetrate into its center.
But that still leaves our original question unanswered, precisely how large can a language be? Then let me try to answer it right now. We really have to forget about numbers and also abandon some of our preconceptions about language. The true answer to this question is not 100,000 words, and also not one million words or even one hundred million words, but the far more pragmatic and powerful statement:
A language can be as large as necessary.
We need to realize that a language is not a set of words or a set of rules or even a set of structures for describing these words and rules. Language is far more than this, it is even more than the cultural attitudes it embodies.
I'm still working on this section of my book, and I think that's about as much as I feel ready to share with you at the present time.
So I'd now like to move forward to my anecdotes and conclusion, after which I'll do my best to answer any questions you may have about anything I've presented here either today or on Wednesday.
10. Anecdotes and Conclusion
And as the title for this section suggests, I'd like to end this presentation with a few relatively light-hearted but still not totally light-hearted observations and then summon up my last remnants of seriousness for two or three final sentences. Anyway, you've made it through the six laws and the two leaps, so perhaps a bit of comparative merriment is in order.
I believe that over the last five decades linguistics has had a great deal in common with science fiction, though not intentionally. So it seems to me only fitting that we should honor this resemblance by examining three or four of the most common science fiction plots and see how an evidence based approach to linguistics could reduce them once and for all to the utter nonsense they have always been.
Let's start with science fiction stock plot 12A, involving the first visit of real live aliens to our planet. We earthlings have been incredibly lucky with this story so far, since virtually all the aliens who've ever made it our way have almost invariably been able to speak not just any earth language but perfect modern American English, though sometimes with a variety of accents. Unless of course they were Jules Verne aliens, in which case they quite naturally spoke perfect Académie Française French. And so on in various languages, after all how could any space invaders in an English film possibly fail to speak perfect RP (standing for what in Britain is called "Received Pronunciation")? Sometimes it gets mentioned that the invaders had spent months circling our planet just to master our lingo, but most sci-fi stories don't even make that concession to the need for foreign language education. Or if it's a Spielberg movie, there's always R2D2 to rely on.
But I'm willing to bet, especially since I believe this is one bet that I'll never be called on, that if the real thing ever happens, the odds are far better than even that we won't be able to converse with our first visiting aliens at all. And that all the king's linguists and all the king's scientists won't be able to help in the slightest. That's how total the language gap between us and our delightful guests is likely to be. We simply won't be up to talking much.
Not that it would necessarily help if we could. Even if we both could speak a few broken segments of each other's tongues, that still wouldn't mean we'd come remotely near understanding each other. Look at what happened to Cortez down in Mexico a mere 500 years ago. Even with all the help he got from his mistress and interpreter Doña Marina AKA La Malinche, it still didn't prevent for a single instant the destruction of Tenochtitlán and the annihilation of the original Mexican inhabitants.
That was sci-fi plot 12A, so now let's try plot 12B: time travel. And even here we'll find there's a linguistic problem that almost everyone else has overlooked. You've already heard all the reasons time travel is impossible, the sheer physical problems of moving the vehicle, figuring out how to aim for one specific date, all the paradoxes and monstrosities such visits could create, and the slight problem of what happens to human protoplasm along the way. But let's assume all these minor inconveniences have now been solved. Even so, the linguistic problems involved are sure to stop things dead in their tracks.
Let's insert just one little requirement into our time travel scenario, and it's a very reasonable requirement if we're going to have any chance of surviving, much less making it back. Let us require that our time travelers must have achieved such perfect fluency in the target language of the target area of the target era that they will make no mistakes that could possibly reveal they don't come from that period. And there's a penalty involved if they are detected as outsidersin the most merciful version they will be thrown into a dungeon and simply left to rot. This would be not an unreasonable penalty at all during most periods of history. I may feel more sensitive to such a punishment than others, since I have made it at least close to passing as native in a few different languages and cultures.
There is simply no way that our professors of, say, medieval French literature could ever master the practical language know-how to pass as natives if they were to actually visit their favorite period, however perfectly they may have deciphered its poetic minutiae. In fact, I'm willing to take this claim several steps further and insist that there is no way time travelers from today could ever visit a US city of fifty years ago and go undetectedand therefore unpunishedeven though we have vast treasure troves of film, TV, and popular literature from the period stored in our libraries they could study beforehand and we even have live access to many citizens who were actually alive during the era.
Language change is something like the mills of the godsit takes a long time, but it never stops its grinding motion, and the cumulative effect even after a few decades can be quite devastating. I've occasionally been called upon to help actors immerse themselves in the language and the outlook of another nation or another age, so I think I have some notion of the level of expertise such a traveler would need to gain simply to pass unnoticed during the 'Fifties. The necessary training for full-time, real-time time travel would be considerably more demanding than what's needed just to pass muster on stage for an hour or two. In fact, it would be so demanding that I'm going to insist that it can't be done. Just one more reason, out of many, why time travel is quite totally impossible.
Time for plot 12C: awakening a century or two later after having been frozen cryogenically. You know that plot, Woody Allen's Sleepers. We know all the reasons why that can't be done either. Freeze a raspberry today, thaw it out tomorrow, and you'll quickly conclude that neither its taste nor texture are anywhere near authentic. Just try doing that with a human being, well, that's the end of that story, isn't it? But let's assume all those problems have now been solved. It's not a whole lot different from time travel, but could anyone be sure their minds could gear up that readily to the language forms being used even a century from now, much less two centuries. All the different names for things and people, to say nothing of changes in technology, social attitudes, or even sexual behavior.
So onward to plot 12D: living forever, not too different from the cryogenics scenario, except your brain doesn't turn into a soggy raspberry. You just keep hearing new language forms and names for things and absorbing new social attitudes and finally learning whole new languages, over and over again. I don't know about you, but I'm 73 and I'm already starting to feel that I can't fit too much more in there. When I went down to Mexico last year to present a paper in two languages, I did my best to bring my Spanish up to where it had been five decades earlier, when I worked as a bilingual radio announcer in Madrid for Radio Nacional de España. I came fairly close to succeeding, but at the cost of losing a few memory cells for my other languages, including English. Things do tend to get crowded in there. So perhaps I could handle living forever if I accepted the idea of having to lose old brain connections in order to gain new ones. But what's the point of living for even three hundred years if I can't hold on to all my languages and experiences and enjoy the pleasure of comparing them with the new ones?
So that's it for science fiction-let's see, do I know any more linguistics-related anecdotes?
Oh well, yes, though this one's not all that good, I suppose it's a bit too serious, though I find it quite funny in its way. Take a look at this:
Now, it is words and their associations which are untranslatable, not ideas...there is no idea...which cannot be adequately produced as idea in English words.
I'll tell you who wrote it later, it's not all that important, because it could have been written by any number of people, both linguists and non-linguists. It reflects some very deep feelings people have about their language, it could be any language and not just English. And it's definitely the sort of opinion one might expect from members of the LSA-MIT school of linguistics. It lies right at the very center of that feeling so many people have that all languages are really just one language, and what's more, any one language can effortlessly handle anything any other language can express. And it all sounds so perfectly logical, and at first glance even objective, scientific.
And yet it isn't the least bit logical or scientific or even the slightest bit true. In fact, it contains a blistering error of logic that ought to cry out to almost anyone reading it. But for the most part opinions like this one have attracted only emulation and almost no criticism at all,
Let me show you precisely what that error in logic is. How could any of us ever possibly know for certain that "there is no idea...which cannot be adequately produced as idea in English words?" How could we ever hope to know that? After all, if there were even a single idea that could "not be adequately produced as idea in English words," what possible way could we ever have of knowing about its mere existence?
As you can see, this observation is nothing more or less than a statement of pure linguistic jingoism, almost but not quite on a level with that famous Texas governor who in vetoing foreign languages for the state's schools simply observed "If English was good enough for Jesus, then English is good enough for me." It is an idea that professional linguists should abhor and not hold up as any kind of touchstone of knowledge or cultural insight. As those of us who have pored over obscure Buddhist doctrines or seemingly simplistic Daoist nature studies or even the works of ancient Greek gnostic authors have become well aware, to suggest that any or all of such material can be "adequately produced as idea in English words" truly begs the question of what it means to "adequately produce"-or reproduce-any text in any language.
Oh well, I guess that wasn't too funny. But wait a second, I do know a funny one, hold on, I've almost got it...yes, now I remember, this really is a great one, and it's all about MT research, though some of you may not enjoy it as much as I do. It takes the form of a three part riddle, it goes like this:
What do we call someone who spends three whole hours trying to fit a square peg into a round hole?
Answer: an utter fool.
What do we call someone who spends an entire week trying to fit a square peg into a round hole?
Answer: an absolute idiot.
Last question: what do we call someone who spends his entire life trying to fit square pegs into round holes?
Answer: a computational linguist.
Well, I guess I owe some of you here an apology for that riddle. If it's any consolation, over the decades I myself have followed the efforts of computational linguists with a whole lot more than merely superficial interest. As have quite a few professional translators, and like yourselves we have been continually bedevilled by the question, could it just be possible after all, could MT truly turn into a workable system for handling languages? And knowing translation from the inside, some of us have even turned our hands to setting up systems of rules and governance based on seemingly immutable transformations. And then we've watched those systems and rules turn into so much runny gelatin as we tried to apply them. So yes, we do have some idea what you've been through and are even able, at least to some extent, to empathize with your own efforts and ambitions.
There's a lot more I'd like to tell you about, I was hoping I'd have a chance to say a bit more about how easily seemingly immutable words can turn into constantly shifting artifacts. But I think I've given you more than enough to think about both today and Wednesday and left you with a few concepts you'll be able to either accept or reject, and in either case I'd be interested in learning what your reasons are.
To conclude, I believe that over the last decades the so-called science of linguistics has on the whole made the wrong assumptions, asked the wrong questions, and studied the wrong subjects to ever permit its practitioners any truly deep insights into the subject they purport to study. In my opionion, those subjects would need to include, at a minimum, physiology and medicine, translation and translation studies, cartography, and an enforced, prolonged, and practical immersion in other cultures.
Improvement, when and if it comes (and I would truly like to believe that it will come) can take place only over time and perhaps by means of merging the goals and methods of linguistics first with those of translation studies and human physiology, and then at a latter date with psychology and whatever the study now called psychiatry eventually comes to be known as.
I certainly won't live to see this come about, but some of you may be privileged to witness and play a role in the beginnings of such a process. I want to thank you again for having invited me to this conference, and if there is any further way I can help you in the future, you need only to ask me. Thanks again.
And now I think it's time for me to listen to your questions. If I may be permitted to steal a phrase from Paul Erdös, my brain is now open.
Here is the section omitted in the preceding text:
4. My Adventures with Voodoo and Evidence Based Linguistics
I'd now like to share with you some further reflections about the differences between Evidence Based Linguistics and Voodoo Linguistics, which I couldn't include on Wednesday due to time considerations. If some of these reflections sound a bit scary, I believe it's because they are scary and can give us an idea of how far we have truly strayed from scientific method simply by our complicity in not objecting to the spread of Voodoo research and Voodoo findings. I want to start with two scenarios that may sound as if they are purely science fiction, and yet they have come close to happening in recent years and could yet-if nothing changes to the contrary-become the total reality we must live in.
We've just been through the whole Terri Schiavo sob story, an affront to science if ever there was one. But what would have happened if this scandal had gone just a few steps further, or if a closely related incident in the near future lurches out of control in a similar way? Let's imagine that the next time around, the popular outcry across the nation that someone like Terri Schiavo is fully alive, despite all evidence to the contrary, were actually to lead to new politicians being elected and new laws being passed.
Under these new laws all decisions related to life and death would henceforth be decided not by doctors and nurses but by government appointed Special People's Medical Examiners, civic leaders and clergymen prominent among them, untrained laymen who would render binding decisions based on their biases as to whether someone is alive and must be kept alive or whether they may be permitted to die. And absolutely anyone wearing a smile on their face, even if that smile is frozen and unchangeable, would from now on be declared as totally alive, disregarding all other factors.
I can almost hear the rhetoric involved: Doctors don't really know anything about life and death, and in any case biology and medicine are not really that complicated, which means that We the People, who can see through all these tricks and poses, are the only ones who need to decide, the only ones capable of deciding.
Let me take another seemingly sci-fi but not totally impossible scenario, especially after last week's problems with the Discovery mission: the US Space program, which still has its supporters in many quarters. Leave to one side what the spokesman for American physics Robert L. Park, among others, has to say in his book Voodoo Science, that humans can never survive the solar or cosmic radiation unavoidable in space flight or while staying alive on other planets right here in our solar system, nor can they survive prolonged periods away from earth without fatal bone loss. Let's imagine that a great public upswing in favor of continuing the space program despite all possible obstacles were to develop among us.
Here too I can hear the rhetoric involved: Scientists and political leaders don't really understand astronomy or the deep meaning of space, and in any case space flight can't really that be all that complicated, there's sure to be a way to get through, a wormhole or something, and besides engineers can do absolutely anything. Which means that We the People, who can see through all these tricks, are the ones who have to decide, we're the only ones qualified to decide, even if we have to appoint Special People's Space Consultants to get the job done, after all we all know the real truth, that it's our destiny as human beings to conquer and inhabit the whole universe and ride around in space ships everywhere while we happily shoot laser cannons at each other.
Nor in our current political environment is it at all hard to imagine similar scenarios related to abortion, drug abuse, foreign immigration, and finally once an for all the choice and establish-ment of one single compulsory God all Americans must unfailingly and unquestioningly accept and worship or suffer the consequences. None of these ideas are truly that far away from the current mindset of many leaders in Washington, not to mention at least half of our citizens across the nation. And every single one of these projects would of course be undertaken amidst glorious rhetoric that now finally, at long last, We the People have broken through all the elite nonsense and pretensions of those in power.
But I can already hear your question: how precisely can any of this be related to linguistics. Actually, it could not possibly be more closely related to our field, because what I have described comes perilously close to precisely what happened to American linguistics in 1957.
A whole new crew of self-proclaimed experts suddenly leapt into power and came close to taking over our field. And their rhetoric went very much as follows: "Hey, we don't need any more of those linguist snobs telling us about language. We know all about language that we need to know, after all, we speak it, don't we? Language must be easy, I mean, everybody speaks it, don't they? What more does anyone need to know than that?
"So we've got to quit all this nonsense about language being complex and stop studying all those lingos only a few savages speak anyway. What we most need is a way to make everybody be able to learn any language easily. And why not have computers learn it too. Hey, maybe especially computers, I bet we can even make a computer that can translate between all languages, that way we can get rid of all those useless translators and really get to the core of what language is all about. Hell, language comes natural to everyone, it's absolutely universal, you know what, I bet it's even innate, and no matter where we live or what language we think we speak, when you get down to it we're all already speaking the same language anyway."
I'm leaving of course to one side the slight matter that most of these latter-day linguists happened to be scientists, or at least considered themselves scientists: they included physicists, statisticians, mathematicians, computer scientists, and also a smattering of logicians and philosophers. So they were not your usual plebeian masses out on the barricades, but they certainly acted as though they were. And they all proceeded as though by bringing less knowledge of language to their newly assigned field of study, they were actually bringing more knowledge. Or anyway were sure to do so in the long run.
Am I truly exaggerating by presenting this account? I honestly don't think I am, and as evidence that this was truly the prevalent attitude, let me read you the words of the master of MIT himself, as cited in a 2003 communist-oriented critique by the well-known linguist Chris Knight:
'In fact, the amount that you have to know in a field is not at all correlated with the success of the field. Maybe it's even inversely related because the more success there is, in a sense, the less you have to know. You just have to understand; you have to understand more, but maybe know less'. (Note 3)
And as for the idea of teaching a computer to translate between all languages automatically, despite later denials by this school do we really have to search any further for a smoking gun than the following statement, dating from 1957:
'We can think of the initial state of the faculty of language as a fixed network connected to a switch box; the network is constituted of the principles of language, while the switches are the options to be determined by experience. When the switches are set one way, we have Swahili; when they are set another way, we have Japanese. Each possible human language is identified as a particular setting of the switches - a setting of parameters, in technical terminology. If the research program succeeds, we should be able literally to deduce Swahili from one choice of settings, Japanese from another, and so on through the languages that humans can acquire...' (Note 4)
And there it is, in those two citations alone, the pandora's box, the mindquake, the supergalactic explosion that together with vast funding from the Department of Defense directly launched the formerly prosaic world of linguistics into an era of almost uncontrolled growth coupled with uncontrollable pretensions.
The first of those quotes says you don't really have to know very much about language to join the party, the second excerpt says that even though you don't know very much, as soon as you find the right gimmicks, with almost no effort at all you're sure to be able to make every language turn automatically into every other language.
Which incidentally was just what the US Governmentand especially its Department of Defensewanted to hear and wanted to have happen and why they were funding all of this to begin with. Anyway, who could say no to such fascinating problems when there was not only money to pay for it but eminent scientists ready to swear that it was all perfectly feasible.
But, ladies and gentlemen, it's now fifty years later, and it simply hasn't happened. And it certainly hasn't been for lack of trying. And hey, I'm not jumping anyone, I was caught up in this too, I'm merely trying to be...I think the correct word in our field is "descriptive."
Face it, these brief excerpts have been the adrenaline rush, the amphetamine fix, the testosterone spurt that have been running linguistics throughout these five decades, nor has this movement been limited to MT alone. The period has also coincided with what may have been the peak of belief in MT's parent study, Strong AI. And it's also been the time when computer power and computer hype went through the roof of its own surge of development. Which has meant, when it first became clear that perhaps the MT and the MIT people were encountering a few more problems than they had anticipated, that we have all of us been casting around for possible solutions that could keep our supply of amphetamine going long enough so we would never have to think about the prosaic process of coming down and beginning to recognize that we had spent an entire fifty years being at least partly under the influence of a remarkably large overdose.
When McCorduck & Feigenbaum wrote their book about the so-called Japanese fifth generation, we all entertained the notion that we had to work hard on the task of beating the Japanese at AI, or else they would solve the problem of language before us. When Danny Hillis came up with the Connection Machine, and of course when Rummelhart and McClelland wrote their Parallel Distributed Processing, we hailed them both as the probable solutions to the problems we kept encountering. And when James Gleick published his book Chaos, we hailed that also as providing strategies that were sure to open up vast new insights into more subjects than we could imagine, including the humanities as well as the sciences.
As a matter of fact, I got suckered into that one myself and published a rather confused paper about what I called Fractal Linguistics. (Note 5) I expected to hear no more about it, but it had scarcely appeared when I received an urgent letter (we all wrote letters, not email messages back then) from a researcher at Microsoft, telling me she was working in the same direction and not so much requesting as demanding that I turn over to her all of my findings for the greater glory of machine translation and natural language processing. I actually didn't have any further findings worth mentioning, though I don't think the researcher believed me, but this theme at least sparked my first attempt to write something book length about language.
During that period I did my level best to uphold the principles of evidence based linguistics, as I then understood them, but it was a real uphill battle. One day in 1988 I received a long distance call all the way from Amsterdam. It was from the editor of a new magazine called Language Technology. He had read a critique I had written of an appeal by someone in this group, urging all translators everywhere to donate all their glossaries of technical terms in all fields to his central data bank. In my critique I had been rather negative about the possibility of translators following his instructions. For some reason the editor in Amsterdam was certain that there simply had to be answers to the problems I had described and what's more, that I very probably already knew them. After all, we all knew that all computer problems have computer solutions, or so we all told each other back then.
We were all supposed to believe that computers could do absolutely anything, and some of us actually did. I had no problem accepting the premise that computers could transform people's lives, in fact I had already experienced this in my own life. But this still wasn't good enough back then-you were expected to believe that computers were the ultimate tool that could completely change the face of the earth, much less what people had hitherto regarded as the realm of ideas. And you were asked to believe this with a burning intensity that made you think of nothing else but computers all day long, a sort of digital fundamentalism. I simply couldn't go that far, though I certainly welcomed them as a remarkably useful tool.
This editor wanted me to write a second article explaining the solutions to the problems I had described. He offered me good money for it and dismissed my reservations as no more than modesty. Well, I wrote him his article, mainly a restatement of what those problems were then and still remain, and tacked on in the very last paragraph something that sounded like a happy ending for him. To my amazement he accepted and published the piece, though in very fine print.
This was part of the madness that natural language computer buffs were going through at that time. There were of course some very good reasons why translators were unlikely to donate their glossaries to holding companies for MT companies, one of course being that almost every issue of computer magazines in those days repeated the boast that soon computers would be taking over all translation work everywhere.
That wasn't even the end of that journalist either, who was absolutely sure that computers had the answers to everything. He changed the name of his magazine from Language Technology to Electric Word, and when this plus his remarkably cluttered pages strewn with graphics and outrageous claims still didn't attract readers, he quit after two years and left Amsterdam for California, determined to try out the same formula in a magazine that took not language as his target but the entire computer world and its spin-offs. His name was Louis Rossetto, and the result was a remarkably glitzy rag called WIRED, though even here he was finally bought out and forced to abandon ship in favor of slightly better journalism by new editors. Yet another example of those adrenaline/amphetamine/testosterone-soaked years.
This period also witnessed the development of Doug Lenat's CYC project and George Miller's Wordnet, each of them hyped to infinity and funded to the skies as likely to bring about the final "Conquest of Language" by computers. Soon, we were told, spin-offs from such approaches as these would not only be translating for all of us and interpreting foreign language phone conversations but also busily composing expert gists and precis of whatever data was fed into them and according to some versions even be capable of taking an assemblage of facts and using them to write a thorough and persuasive article about the subject in any language, far superior to what any mere human writer could possibly turn out.
This last claim was one I seemed to remember hearing elsewhere, and I had. In Swift's satire of experiments by Leibnitz to mechanize human language and knowledge, as carried out by the savants of the Academy of Lagado in Gulliver's Travels. Here's a brief excerpt:
The first Professor...said perhaps I might wonder to see him employed in a Project for improving speculative Knowledge by practical and mechanical Operations. But the World would soon be sensible of its Usefulness; and he flattered himself, that a more noble exalted Thought never sprang in any other Man's Head. Everyone knew how laborious the usual Method is of attaining to Arts and Sciences; whereas by his Contrivance, the most ignorant Person at a reasonable Charge, and with a little bodily Labour, may write Books in Philosophy, Poetry, Politicks, Law, Mathematicks and Theology, without the least Assistance from Genius or Study."
That dates from 1726, almost three hundred years ago.
Throughout the early and mid nineties, MT & NLP & AI research went on apace, some of it fairly comparable to Swift's account, with articles constantly appearing-as they had during the 'seventies and 'eighties as well-to reassure us that the final perfection of machine translation was close at hand and required only a few further, remarkably minor breakthroughs before it would finally be announced. The people behind the Systran translation system became so proud of their efforts that they actually offered professional translators the chance to make use of their program at a cost of merely four cents a word.
They believed in their system so intensely that they must have been genuinely surprised when they discovered there were almost no takers among professional translators, some of whom were in any case working for as little as five or ten cents a word. Even then their system was a joke, no more than a toy, and I predicted in a piece dated 1996 that sooner or later they would be obliged over time to lower the price of their output until they finally discovered that they could not even give it away. And this of course is what ended up happening.
I mention these events because in my opinion they count as just some of the many examples of Voodoo Linguistics we have lived through in recent years. If I seem to be stressing the negative side right now, it's because I believe I've already given you, in the six draft laws of language and linguistics and the other material i've presented, something more positive to think about.
NOTE: This bibliography covers both the Wednesday paper and the Saturday workshop session, which explains why most of the references do not apply to the preceding text.
Bloomfield, Leonard. 1933. Language. New York: (reprinted in great part in 1984, University of Chicago).
Caldwell, Price http://www.hinocatv.ne.jp/~price/
Chomsky, Noam. 1957. Syntactic Structures. The Hague: Mouton.
1988a. Interview. In: The Chomsky Reader. Edited by J. Peck. London: Serpent's Tail, pp. 155.
Craciunescu, Olivia; Gerding-Salas, Constanza; Stringer-O'Keeffe, Susan: Machine Translation and Computer Translation: a New Way of Translating? In Translation Journal, Volume 8, No. 3 July 2004, accessible at:
De Beaugrande, Robert. 2002. Linguistic Theory: Louis Hjelmslev, part of The Discourse of Fundamental Works, accessible at:
Dreifus, Claudia. June 26, 2001. A Conversation with Frans de Waal: Observing the Behavior of Apes From Up Close. In The New York Times.
Gould, Stephen Jay. 1977. Ontogeny and phylogeny. Cambridge, Mass. : Belknap Press of Harvard University Press.
Gross, Alexander. 1993a. MT and Language: Conflicting Technologies?Ariadne's Endless Thread. In Sci-Tech Translation Journal. Poughkeepsie, NY: American Translators Association. Accessible from:
1993b. Selected Elements from a Theory of Fractal Linguistics. In Scientific and Technical Translation, ATA Scholarly Monograph Series, Volume VI. Amsterdam and Philadelphia: John Benjamins Publishing Company.
1994. Translators and Interpreters: The Binding Force Of World Civilization. Proposal for a museum exhibit. Sponsored by the American Translators Association. Viewable in part at: http://language.home.sprynet.com/trandex/smithson.htm
1995. Perfect MT: Logical Certainty or Recurrent Self-Delusion? In: ATA Proceedings, 1995. Downloadable from:
1995/6. Spray It Again, Sam: The Real Story of Language And Translation, A Semi-Humorous Account Part III of Truth About Translation. Washington, DC: American Translators Association. Program downloadable without charge from:
2000a. HermesGod of Translators and Interpreters: The Origins of Language and the Prehistory of Interpreting, invited paper at NYU Translation2000 Conference. New York: NYU Translation Studies Program. Presented later that year at Jornadas Jeronimianas Conference, Mexico City. Downloadable from: http://language.home.sprynet.com/trandex/hermes.htm
2000b. HermesGod of Translators and Interpreters, The Antiquity of Interpreting: Distinguishing Fact from Speculation. Paper commissioned by NYU Translation Studies Department as part of a departmental book project which was later abandoned by the chair. Downloadable from:
2003. Teaching translation as a form of writing. In Beyond the Ivory Tower: Rethinking translation pedagogy, edited by Brian James Baer and Geoffrey S. Koby. Vol. XII, ATA Scholarly Volume Series. Amsterdam/Philadelphia: John Benjamins Publishing Company. Downloadable from:
Hjelmslev, Louis. 1961. Prolegomena to a Theory of Language. Translated by Francis Whitfield. University of Wisconsin: University of Wisconsin Press. (Danish title: Omkring sprogteoriens grundlaeggelse, Copenhagen, 1943)
Knight, Chris. 2003. Noam Chomsky:Politics or Science? In What Next: Marxist Discussion Journal. Downloadable from:
Lamb. Sydney. 1999. Pathways of the brain : the neurocognitive basis of language. Amsterdam; Philadelphia, PA : J. Benjamins.
Language Technology. 198890. Periodical, later renamed Electric Word. Amsterdam: INK Taalservice.
Lao Tze. ca. 600 B.C.E. dao de jing. Hundreds of translations available, including the one cited, see footnote 19 for clarification.
Lavoisier, Antoine Laurent. 1789. Traité Elémentaire de Chimie. Third paragraph. Paris: Chez Cuchet, libraire. Online in French at:
Lichtenberg, Georg Christoph. 1908. The Reflections of Lichtenberg. Selected and translated by Norman Alliston. London: Swan Sonnenschein & Co., Lim.
Mandelbrot, Benoit. 1967. How long is the coast of Britain? Statistical self-similarity and fractional dimension. In Science: 156, 636638.
Montgomery, Scott. 2000. Science in Translation: Movements of Knowledge through Cultures and Time. Chicago: University of Chicago Press, 2000.
Mounin, Georges. 1963. Les problèmes théoriques de la traduction. Préface de Dominique Aury. Paris: Gallimard.
1967. Histoire de la linguistique, des origines au XXe siècle. Paris: Presses universitaires de France.
1990. Teoria e storia della traduzione. Milano: Einaudi.
Pellegrini, Angelo M. Giordano Bruno on Translations. In: English Literary History, 10: 193207.
Pullum, Geoffrey K. 1991. The Stranger in the Bar. In: The Great Eskimo Vocabulary Hoax and other irreverent essays on the study of language. Chicago : University of Chicago Press.
Roget, Peter Mark. 1852. Thesaurus of English Words and Phrases, classified and arranged so as to facilitate the Expression of Ideas and assist in Literary Composition. London: Longman.
Saussure, Fernand de. 1913. Cours de Linguistique Générale, Paris (translated by Wade Baskin as Course in General Linguistics, 1959, New York: Philosophical Library.).
The Edge Foundation. 2005. What Do You Believe To Be True Even Though You Cannot Prove It? The Edge Annual Question. Online at:
Whitfield, Francis. 1969 Glossematics, Chapter 23 of Linguistics, edited by Archibald A. Hill. Washington, DC: Voice of America Forum Lectures, (reissued in the same year by Basic Books).
Whorf, Benjamin Lee. 1956. Language, Thought and Reality, (collected papers). Cambridge, MA: M.I.T. Press.
Yates, Frances A.. 1934. John Florio: The Life of an Italian Shakespeare's England. Cambridge: Cambridge University Press.
1. Gross, 2003. [back to text]
2. Details for both at:
[back to text]
3. Chomsky, 1988a: 17. [back to text]
4. Chomsky, 2000: 8. [back to text]
5. Gross, 1993b. [back to text]