Chapter 5 Chapter 4: Mechanisms and History of Language

Of course, no theory of communication can escape the question of language.In fact, language is, in a sense, another name for communication itself, not to mention that the word can be used to describe the codes by which communication takes place. As we shall see later in this chapter, the use of encoded messages and decoded messages is important not only for humans, but also for other living beings and the machines they use.The communication between birds, between monkeys, between insects, and all that kind of communication uses to some extent signals or symbols that can only be understood by the participants in the code system.

The difference between human communication and other animal communication lies in: (a) the signal code used is delicate and complex; (b) this signal code is highly arbitrary.Many animals can use signals to communicate their emotions with each other, and in the communication of these emotional signals to indicate the appearance of enemies or the opposite sex of the same kind, and such messages are so delicate that they are extremely rich and varied.Most of these messages are temporary and are not stored.Translating these signals into human language, most of them are auxiliary words and interjections; although some signals can be roughly expressed by certain words, they seem to be given the form of nouns or adjectives, but animals do not use them. Any distinction of grammatical form will be made accordingly.In short, one can think that the language of animals is firstly to convey emotions, and secondly to convey the appearance of things. As for the more complicated relationship between things, it cannot be conveyed at all.

Apart from the above-mentioned limitations in the nature of communication of the languages ​​of animals, they are also very generally bound by the species, immutable throughout the history of the species.The roar of a lion is very similar to the roar of other lions.However, some animals, such as parrots, chaffinches, and crows, seem to be able to learn the sounds of their environment, especially the calls of other animals and people; they are also able to improve or expand their vocabulary, although only to a very limited extent. But even the above-mentioned animals are not as convenient as humans to use any sound that can be made as a code to express this or that meaning, nor to convey this code to the surrounding crowd as conveniently as humans, making The code system becomes a recognized language that is understood within the group and has little meaning outside the group.

Birds are able to imitate human speech with great limitations because they all have the following common characteristics: they are social creatures, have a long life span, and have a memory, which is divided by the precise memory of humans. Excellent in every way except as a measure.There can be no doubt that talking birds, when specially taught, can learn to use human and animal sounds, and that, if one pays attention to them, there is at least an element of intelligibility in them. However, apart from man, in the animal kingdom, even the most vocal animals are not as good at giving meaning to new sounds and storing specific information coded sounds within the scope of language memory. , not to mention the ability to form Russell's notation for "advanced logical types" such as relations, classes, and other entities.

But even so, I still want to point out that language is not a unique attribute of living organisms, but something that living organisms and artificial machines can share to a certain extent.I would go a step further: Man's superiority in language represents nothing more than a possibility given to him than to his closest relatives, the great apes. But I will show that this is imposed only as a possibility, and that it must be learned to be useful. Often, we think of communication and language as merely a means of connecting people.But it is quite possible to make man talk to machine, machine to man, and machine to machine.For example, if new settlements were to be built for the sake of the desolate parts of the West of our country and the North of Canada, where many places were constructed for the establishment of power stations, which were at a great distance from any settlements inhabited by the workers, These power stations are too small, although they are small enough to be abandoned by the power system.So the ideal way to run these plants is to leave them virtually unattended for a few months before and after regular visits by managers without having to keep people on board.

To do this, two things are necessary.One is the introduction of automatic machinery which will ensure that it is impossible to make contact with the bus-bars or connecting rings before the generators have attained the correct frequency, voltage and phase; random accidents.If the daily operation of the power station is not interrupted or abnormal, then this management method is sufficient. However, the situation is not that simple.The load on a power generation system is determined by many variables.There are constantly changing industrial needs, there are temporary accidents that shut down a part of the system, and there are even dark clouds in the sky that can turn on the lights in thousands of offices and homes during the day.In this way, automatic plants, like those managed by workers, must be under the constant supervision of a load dispatcher, who must be able to issue commands to the machines, and he sends corresponding coded signals to the plant. This was done by sending the signals either over special circuits designed for the purpose, or over existing telegraph or telephone lines, or by using the wires themselves as the transmission system.On the other hand, for the load dispatcher to be comfortable issuing orders, he must have prior insight into the workings of the power station.In particular, he must know whether the orders he has issued have been carried out or have been hindered by some defect in the equipment.Therefore, the machines in the power station must be able to send a reply to the load dispatcher.Thus, there is an example of a person uttering language and pointing at a machine, and there is also an example of the opposite, that is, a machine speaking to a person.

The reader may be surprised that we include the machine in the sphere of aphorisms, and yet deny the language of ants almost entirely.But it is often very important to us, in making machines, to endow them with human attributes which are not possessed by members of the animal kingdom.If the reader is willing to see this as a metaphorical expansion of our personality, I welcome that view, but I would remind him that for new types of machines, when we no longer give human support, they It won't stop working because of it. In fact, the language we speak to the machine involves more than one step.If only considered from the line engineer's point of view, the signal code transmitted along the line is complete in itself.For this kind of information, all the concepts of cybernetics or information theory can be applied.We can determine its probability in all possible message series, and then take the negative logarithm of the probability of reading according to the theory explained in the first chapter, so that we can estimate the amount of information it carries.However, this is not the message actually transmitted by the line, but the maximum amount of information that the line can transmit when associated with an ideal receiving device.The amount of information that can be transmitted by the actual receiving device depends on the device's ability to transmit or use the received information.

Therefore, we derive a new concept from the method of receiving orders from the power station.In a power station, the opening and closing of switches, the adjustment of phases of generators, the control of sluice flow, and the opening or closing of turbines can all be regarded as a language in itself, with a The behavior probability system given by its own history.In this structure, each possible sequence of commands has its own probability, whereby its own amount of information is conveyed. Of course, the relationship between the line and the receiving device can be handled so perfectly that the amount of information contained in the signal can be seen from the point of view of the transmission capacity of the line.Or from the perspective of machine operation to measure the amount of information that has been executed, it is equal to the amount of information transmitted by the composite system composed of the line and its receiving device.But, generally speaking, from the wire to the machine, there is a transition stage in which information can be lost and never regained.It is true that, except for the final or validating stage, the process of information transmission may contain several successive stages of transmission; and between any two stages, there is an activity of switching information, which may cause the information to escape.Information can be lost and never regained, as far back as the cybernetic form of the second law of thermodynamics.

Up to this point, we have discussed machine-terminated communication systems in this chapter.In a certain sense, all communication systems end in machines, but ordinary language systems end in a special kind of machine called a human.Man, as a terminal machine, has a communication network, which can be divided into three stages.As far as ordinary spoken language is concerned, the first stage of man consists of the ear and that part of the brain mechanism which is in permanent and fixed connection with the inner ear.This instrument, when it is connected with the vibrating instrument of sound in the air or with an equivalent instrument in an electric circuit, represents a machine concerned with phonetics, that is, with sound itself.

The second aspect of language, or semantics, has to do with the meaning of language.For example, when one phonetic is translated into another language, the incomplete equivalence between words and meanings restricts the flow of information in this language to another language, and the difficulties are obvious.According to the statistical frequency of occurrence in the same language, one can select a lexical sequence (two-character or three-character word group) to obtain an obvious similarity to a language (such as English); by The resulting gibberish would also bear a strikingly convincing resemblance to correct English.From the phonetic point of view, this meaningless analogue of meaningful speech is actually equivalent to meaningful language, although it is semantically false; His English, although it has the characteristics of his own country in pronunciation, or he speaks half-literate English, is good in semantics but bad in phonetics.

Furthermore, ordinary people's conversation after tea and wine is good in phonetics, but bad in semantics. It is difficult, though possible, to characterize the phonetic mechanism of a human communication device; thus, it is equally possible, but equally difficult, to determine what is phonetically meaningful information and to measure it. difficult.For example, both the ear and the brain apparently have an effective frequency limiter that blocks the entry of certain high-frequency sounds that can pass through the ear or be transmitted to the ear by telephone.In other words, no matter what information these high-frequency signals can bring to a suitable receiver, they will not bring any meaningful amount of information to the ear.The more difficult problem, however, is to identify and measure semantically meaningful information. The acceptance of semantics requires the help of memory, and the things recorded are kept for a long time.In the important semantic stage, any abstract type should not only establish a fixed connection with the local neuron devices in the human brain, for example, those neuron local devices that must play an important role in the perception of geometric figures, but also They are also associated with abstract detectors made up of parts of the internal pool.Neuron clumps are temporarily assembled for this purpose. They are groups of neurons that can form various larger devices, but the neurons are not permanently enclosed in the device. In addition to these highly organized and fixed apparatuses that undoubtedly exist in the human brain, which we find in those parts associated with particular senses, and elsewhere, there are also Special switches and connections that seem to be temporarily formed for specific purposes, such as learning reflexes and others.To form these special switches, a series of spare neurons must have been assembled for this purpose.Of course, the assembly problem of neurons is related to the synaptic threshold of a series of neurons used for assembly.Since neurons are either in or out of this temporary apparatus, they deserve a special name. As I have already pointed out, I think they are rather close to what neurophysiologists call clumps of neurons. What has been said above is at least one plausible theory of neurobehavior.Semantic receivers do not take in and translate language word for word, but idea by idea, often in a more general way. In a sense, such a receiver is capable of evoking the whole of past experience in which its form has changed, and these long-term preservations are not an insignificant part of the work of the semantic receiver. The third stage of communication is partly the transition of the semantic stage and partly the transition of the earlier stage, the speech stage.This is the transition from personal experience into action, which is observed by others whether he is aware of it or not.We can call this stage the behavioral stage of language.In lower animals, this is the only language phase we can observe after speech input.In fact, everyone has this stage, unless there is such a special person that any given passage for him is specially prescribed.That is to say, one can know the inner thoughts of another only through the actions of the latter.These actions are composed of two parts, one is the direct and obvious action, which we can also observe in lower animals, and the other is the coded and symbolized action system, which is our A spoken or written language known. Theoretically speaking, it is not impossible for us to establish a kind of statistics about semantic language and behavioral language, so that we can measure the amount of information contained in them well.Indeed, from general observation, we can prove that the total amount of information in the speech language reaches the receiver is less than that originally built, or at any rate no more than can be conveyed by the system of transmission to the ear; It can be proved that both semantic language and behavioral language contain less informational speech language.This fact is again a corollary of the second law of thermodynamics, and, at each stage, if we consider the transmitted information as the maximum amount of information that can be transmitted by a properly coded receiving system, then the above fact must it is true. Let me now draw the reader's attention to a question that might have been considered quite unquestionable, namely, the reason why chimpanzees cannot speak.Chimpanzee behavior has long been a mystery to those psychologists who work with these interesting animals. The young orangutan is especially like a child. In terms of intelligence, it is obviously equal to, and may even exceed.But why is it that a chimpanzee raised in a human home and given speech training for a year or two still has no language as a means of expression, except for occasional infantile utterances?Animal psychologists cannot fail to be amazed by this. This situation may be fortunate or unfortunate; most chimpanzees, in fact all chimpanzees ever observed, insist on being a good ape rather than becoming a quasi-human idiot.But even so, I think the average animal psychologist is quite eager for the chimpanzee to embarrass its ape-like ancestors by tainting more human behavior. The failure to do this thus far is not entirely due to the lack of intelligence of chimpanzees, for there are also flawed beasts with brains that would put chimpanzees to shame.It's just not a matter of the brute speaking or demanding to speak. Speech is such a peculiarly human activity that man's next kin and his most ardent imitators cannot master it.The few sounds made by chimpanzees do, indeed, have a great deal of emotional content, but they lack finesse in clarity and repeatable accuracy in organization, both of which are It is a necessary condition for transforming them into a more accurate signal code of a distant call.Furthermore (and this further suggests that these sounds are different from human speech), these sounds made by chimpanzees are usually an unlearned, innate expression, not something learned to express by members of a particular social group. Speech in general is peculiar to men; particular forms of speech are peculiar to members of particular social groups—a fact that deserves the most attention.First, looking at the whole vast sphere of activity of modern man, we can safely say that there is no society which has not been fragmented by the auditory or intellectual incompleteness of the individuals who compose it, because it has no speech style.Second, all styles of speech require a process of learning, and despite the attempts of the nineteenth century to establish a genetic theory of language evolution, we cannot find the slightest general reason for assuming that modern speech All forms of language are derived from a single, primitive form of speech.It was quite clear that if left to live alone a batch of babies would try to talk too.These attempts, however, show only their own desire to produce a sound, and do not thereby suggest that they are imitating any existing form of language.It is almost as clear that if there is a society of children whose members are within the critical age of speech formation but who have not had the opportunity to come into contact with the language of their elders, they will still make certain voices, although they are very Not pleasant to the ear, but undeniably a language. So why can't we force chimpanzees to talk, and why can't we force children not to talk?Why are the general tendencies of speech, and even the general aspects of language expressed and contained, so consistent for large groups of people, while the specific manifestations of these aspects of language are so varied?A partial elucidation of these issues is at least essential to understanding language-based societies.When we say that man, unlike the apes, has a special urge to use a certain language, we are merely stating some basic facts; but the use of a particular language is a matter that should be learned in each particular case. question.We are able to concentrate on the code and the sound of speech, and our concentration on the code can be extended from the aspect of speech to the aspects related to visual stimuli. These are obviously the characteristics of the human brain itself.But no fragment of these codes is imprinted on us as a code of predetermined harmony, as in the courtship dance of many species of birds, or in the way in which ants recognize and expel intruders who intrude on their nest. in mind.The gift of speech cannot be traced back to the common language of Adam's descendants split in the Tower of Babel.It is a purely psychological impulse; words are not innate, the ability to speak is innate. In other words, the barriers that prevent chimpanzees from learning to speak have to do with semantics, not with the phonological phase of language. The chimpanzee has absolutely no mechanism for converting the sounds it hears into something from which to assemble its ideas or into complex behavioral patterns.Regarding the first point in the statement, we cannot verify it, since we have no direct means of observing this phenomenon.On the second point, that is quite a remarkable empirical fact.It may have its own limitations, but it is perfectly clear that man has built such an institution. Throughout this book we have emphasized man's extraordinary capacity for learning as a feature that distinguishes him from other species, a feature that makes his social life an essentially different species from that of bees, ants, and other species. The social life of social insects, though both, has its superficial similarities.Material about children's isolation beyond the normal critical age at which language is usually learned may not be entirely justifiable. The stories of "Wolf Boys" made Kipling write his fantasy "Jungle Books" (Jungle Books). The degree of absurdity in its original form is as unbelievable as it is in The Jungle Book after it has been embellished with fantasy.There is evidence, however, that there is a critical period during which learning to speak is most convenient; after which, if there is no opportunity to come into contact with one's own kind (regardless of race), language learning becomes Restricted, slow and very incomplete. Presumably this is also true of many other abilities that we regard as innate skills.If a child has not walked when he is three or four years old, he may never want to walk for the rest of his life.For normal adults, the walking habit can become more difficult than the driving habit.If a person has been blind since childhood and later regains his sight through cataract removal or corneal transplantation, then when he does the actions that were originally performed in the dark, the restored vision will not only bring confusion for a period of time, but also cause confusion. Certainly useless.This vision could be no more useful than his carefully acquired new talent of dubious value.We may, therefore, hold that speech is the center of the whole social life of man in normal expression, and that if man does not learn to speak in due time, his whole social character ceases to develop. On the whole, man's interest in language seems to be an innate interest in encoding and decoding, which appears to be the most almost exclusive of any human interest.Speech is man's greatest interest and his most outstanding achievement. I was educated as the son of a linguist, about the nature and techniques of language, etc.; questions, which I have been interested in since I was a child.It is impossible that modern communication theory has brought about such a complete revolution in language theory that its fruits would not affect linguistic thought of the past.Since my father was a very heretical linguist whose tendency was to lead linguistics in the very same direction that modern communication theory has had various influences on, I would like to say a little about my personal concerns about the history and An amateur study of the history of language theory. From a very early age, people have considered language to be something mysterious.The riddle of the Sphinx is an original concept of wisdom.Indeed, the word "riddle" itself is derived from the root of "to rede" (to solve, or guess).Among many primitive peoples writing is not much different from magic.In some parts of China, the respect for writing is so great that tattered old newspapers and useless fragments of slips are not willing to be thrown away. Closely related to all these manifestations is "name magic", a phenomenon in which people of a certain culture use several pseudonyms in their lives, with the intention of preventing their real names from being known and used by witches.Among such examples, we are most familiar with the name of the Jewish God Jehovah (Jehovah). The vowels in this word are all taken from another name of God, "Adonai". His name may be profaned by disrespectful words. Starting from name magic, it only takes one step forward to reach a deeper and more scientific interest in language.Like the interest in judging word of mouth and rare manuscripts on the basis of their original texts, so this interest in language goes back to all cultures of antiquity.A holy book must remain what it is.If there are variants, it is up to an expert commentator to decide.Thus, the Bibles of Christians and Jews, the holy books of Persians and Hindus, the classics of Buddhists, and the writings of Confucius all had their early commentators.In order to maintain the true religion, all people's efforts are attributed to the cultivation of words, and the identification of original texts is one of the oldest ways of intellectual training. For much of the last century, the history of language has been reduced to a set of dogmas that often display a startling ignorance of the nature of language.At that time, people's understanding of Darwin's evolutionary model was too rigid and too blind.Since this whole topic is most closely related to our views on the nature of communication, I shall devote some space to it. The early speculation that Hebrew is the language of heaven, and that the confusion of languages ​​began with the construction of the Tower of Babel in Babylon, needs no more interest for us here than as an original sign of scientific thought.However, the subsequent development of linguistic thought has long maintained a similar naive idea.Languages ​​are related to each other, they have all undergone processes of progress, and these changes have finally led to languages ​​​​that are completely different from the past, etc., which could not remain unnoticed for long by the sharp-eyed linguists of the Renaissance.It is impossible to compile a book like Ducange's Glossarium Mediae atque Infimae Latinitatis without making it clear that Roman is not only derived from Latin but also from Latin slang from.There must have been many learned rabbis who were well aware of the similarities between Hebrew, Arabic and Syriac.When the East India Company, at the advice of the notorious W? Wen on the one hand and Sanskrit on the other, are actually cut from the same material.At the beginning of the last century, the work of the brothers Grimm and the Dane Rash not only showed that the Teutonic language should be included in the so-called European language family, but also further clarified the relationship between these languages. relationship, and clarified their relationship to a pre-established original common language. Thus, the theory of evolution of language is the precursor of Darwin's theory of evolution, which has been carefully studied in biology.This theory of evolution is in fact valid, and it quickly begins to show advantages over biological evolution where it cannot be applied.That is to say, one regards language as an independent, quasi-biological entity whose development is determined entirely by its inner forces and inner needs.In fact, language is a phenomenon that arose simultaneously with human communication, and it is influenced by the respective modes of communication of all social forces. Given the existence of hybrid languages ​​such as Franca, Swahini, Yiddish, Chinook, and even English to a considerable extent, attempts have been made to find a single legitimate language for each language. ancestors of the language, and the other languages ​​that took part in the creation of this language are regarded as merely godparents or godmothers of newborn babies.There has been a pedantic distinction between established, legitimate phonetic formations, and such obnoxious, informal things as ad hoc, folk etymology, slang, and the like.In terms of grammar, the first attempt to force all languages, whatever their origins, to don a straitjacket of Latin and Greek was followed by another, almost as vigorous: Each language has its own form of grammatical structure drawn up. Until the recent writings of Otto Jespersen, none of the famous schools of language was necessarily objective enough to give their science a representation of what was actually spoken and what was actually written. What they proposed was rather It is a trite and superficial thing, trying to teach Eskimos to speak Eskimo dialect and Chinese people to write Chinese characters. The consequences of this inappropriate grammatical rhetoric can be seen everywhere.Among these consequences, perhaps, the first is that Latin was killed by its own children, just as the ancient gods were killed by their predecessors. Throughout the Middle Ages Latin, in its changing nature, remained the language of priests and scholars throughout Western Europe, the best of which was fully acceptable to anyone but a pedant, just as Arabic is still the language of many to this day. The same language that is commonly used in Muslim countries.The reason why Latin has such power is that the writers and speakers who use this language are willing to borrow other languages ​​or create new words within the framework of Latin itself, so as to provide a place for discussing the lively philosophical issues of the time. need.The Latin of Saint Thomas is different from that of Cicero, but Cicero cannot discuss Thomas' thought in his own Latin. Some people may think that the rise of European national languages ​​must necessarily mark the end of the role of Latin.But this is not the case.In India, although the new Sanskrit has developed, Sanskrit still shows great vitality until today.I have said that the Muslim world is unified under the tradition of ancient Arabic, although most Muslims do not speak Arabic, and the Arabic spoken today has divided into many very different dialects.It is quite possible that a language that ceases to be the language of ordinary communication has remained the language of scholars for generations or even centuries.Hebrew fell out of use in the time of Christ, but modern Hebrew survives two thousand years after its extinction, indeed, it has become a modern language in daily life again.As for what I am discussing now, it concerns only the limited use of Latin as the language of scholars. By the time of the Renaissance, when the artistic standards of Latin scholars had become higher, there was a growing tendency to eradicate the neologisms of the late classical period.In the hands of famous Italian scholars of the Renaissance, this reformed Latin could be, and often was, a work of art; As much as a secondary training requires, the scientist's main work is after all concerned with the content of language, not with the integrity of form.The result is that those who teach Latin and those who use it gradually divide into two groups, and the distance grows wider and wider, until the point is reached that, except in the most delicate and useless language of Cicero, the teachers do not know each other at all. Teach your students something else.In this state of idleness, they finally lose any role except as specialists in Latin; The role of the Latin specialist.The price we now have to pay for this blunder of arrogance is the lack of a suitable international language that is far superior to man-made languages ​​like Espereanto and better adapted to modern needs. It is a pity that the attitude of the classicists is often incomprehensible to the layman in the intellectual world!I recently had the pleasure of hearing a classicist at a student commencement address bemoan the increased centrifugal force of current learning, which has increasingly distanced natural scientists, social scientists, and writers. Big.He illustrates the situation with an imaginary excursion: he acts as guide and advisor to the revived Aristotle to visit a modern university.He begins with the ridicule of professional jargon in every field of modern knowledge, and enumerates it, thinking that he presents shocking examples to Aristotle.I don't know if I can comment: the whole legacy we have of Aristotle is nothing but the study notes of his pupils, written in the most obscure technical jargon in the history of the world, They were completely incomprehensible to any Greek of the time who had not studied in the Aristotelian Academy.Since this jargon has become canonized by history, it has itself become an object of classical education.This incident has nothing to do with Aristotle, because it happened after Aristotle's death, not during his lifetime.The important thing is that the Greek of Aristotle's time was ready to yield to the professional jargon of a brilliant scholar; the English of his learned and personally respected successors, on the contrary, were not. Concessions are made to the same needs of modern speech. With these caveats in mind, let us return to a modern view that combines the operations of language translation and the related operations of language interpretation by the ear and brain with the performance of artificial communication networks and their coupling processes. agreeable point of view.人们将看到，这个观点实际上是和现代的并曾经一度被看作异端的叶斯柏森及共学派的见解相一致。语法不再象原先那样规范化了。它变成了与事实相一致的东西了。问题不在于我们应该使用什么信码，而在于我们用了什么信码。在我们仔细研究语言的时候，规范化问题的确起着作用，而且非常微妙，这些都是真的。但是，它们是代表通讯问题中的最后成长出来的美丽的花朵，而不是代表通讯问题中的最基本的那些阶段。 这样，我们就给人的通讯的最简单因素奠定了基础：当两个人是面对面的时候，他们是通过语言的直接使用来通讯的。电话、电报以及其他类似的通讯手段的发明，表明了人的通讯能力根本不受个体直接出现与否的限制，因为我们有许多办法把通讯工具带到海角天涯。 在原始人群中，就有效的社会生活而言，社会的大小受语言传送困难的限制。在好几千年里，这个困难足够使国家的最适当人口减编到几百万人左右，一般还要少些。值得注意的是，超越这个限度的大帝国都是靠通讯工具的改善来维持的。波斯帝国的心脏就是皇家大道和沿路设置的传送皇帝诏书的驿站。罗马大帝国之所以能够建立，只是由于罗马筑路技术的进步。这些道路不单是用来调动军团，而且也用来传送御诏。使用飞机和无线电，统治者的话就可以传播到地球的每个角落，以前妨碍建立“世界国家”的许许多多因素现在已经消除了。人们甚至可以作出这样的主张：现代通讯迫使我们去调整不同的无线电广播系统和不同的航空网等国际性的要求，这就使得“世界国家”成为不可避免的东西。 但是，尽管通讯机构变得如此之有效，它们还是要象经常碰到的情况那样，受制于熵增加这一压倒一切的趋势，受制于信息在传送过程中要逸失掉的这一压倒一切的趋势，除非我们引入某些外界的动因去控制它。我已经提到一位具有控制论思想的语言学家所提出的一个有趣的语言学观点——语言是讲者和听者为反对种种混乱的力量而共同采取的对策。以这种描述为基础，B．曼德勃洛特（Benoit Mandelbrot）博士曾在一种最适当的语言中做过若干关于字的长度分布的计算，并且把这些结果和各种现存语言中所算出的分布进行比较。曼德勃洛特的结果表明：在一种最适当的即符合于若干假定的语言中，字的长度非常确定地表现出了一定的分布。这种分布和Esperanto或Volapuk这类人造语中所找到的字的长度的分布是很不相同的。另一方面，它又和大多数的、经过几百年考验的实际语言中的字的长度的分布极为相近。的确，曼德勃洛特的结果并没有给出一个关于字的长度的绝对不变的分布，在他的公式中，还存在着若干必须进行选定的量，或者，如数学家所讲的，还存在着若干参量。但是，适当选用这些参量，则曼德勃洛特理论所导致的结果就和许多实际语言中的字的分布非常密切地吻合，这就说明了它们之中存在着某种自然选择，说明了一种语言形式如果由于自身有用和有生存价值而生存下来的话，那它一定是采取了一种并非不近似于最适当的分布形式的。 语言的磨损可能是由几个原因引起的。语言也许只是力图反抗跟它捣乱的自然趋势，也许只是力图反抗人们有目的地搅乱其含义的企图。正常的通讯谈话，其主要敌手就是自然界自身的熵趋势，它所遭遇到的并非一个主动的、能够意识自己目的的敌人。而在另一方面，辩论式的谈话，例如我们在法庭上看到的法律辩论以及如此等类的东西，它所遭遇到的就是一个可怕得多的敌人，这个敌人的自觉目的就在于限制乃至破坏谈话的意义。因此，一个适用的、把语言看作博奕的理论应能区分语言的这两个变种，其一的主要目的是传送信息，另一的主要目的是把自己的观点强加到顽固不化的反对者的头上。 我不知道是否有任何一位语言学家曾经做过专门的观察并提出理论上的陈述来把这两类语言依我们的目的作出必要的区分，但是，我完全相信，它们在形式上是根本不同的。 我以后在讨论语言和法律的那一章中将进一步讲到辩论式的谈话。 作为一门控制语言意义逸失的学科，控制论应用于语义学方面的愿望已经在若干问题上得到了实现。看起来，在粗糙的信息和我们人类能够有效使用的信息之间，或者，把这句话改变一下，在粗糙的信息和机器能够有效操作的信息之间，作出某种区别是必要的。依我的意见，这里的基本区别和困难是由于如下的一个事实产生的：对行动有重要意义的，与其说是发出的信息量，不如说是进入通讯装置和存贮装置的足以作为行动扳机的信息量。 我已经讲过，用任何方法传递消息或者从外部来干预它们，都会降低它们所含的信息量，除非利用新的感觉或原先处在信息系统之外的记忆馈进新的信息。如前所述，这一陈述就是热力学第二定律的另一说法。现在让我们考虑一下本章前面所讲的那种用来控制小型电力站的信息系统。重要的问题不仅在于我们传送给线路的信息，而且在于这个信息经由终端机械装置去打开或关上水闸、校准发电机以及完成类似的工作时还剩下什么。在某种意义上，这个终端装置可以看作附加于传送线路的过滤器。从控制论观点看来，语义学上具有意义的信息乃是通过线路以及过滤器的信息，并非仅仅通过线路的信息。换言之，当我听到一段音乐时，大部分声音都进入我的感官并达到我的脑子。但是，如果我缺乏感受力和对音乐结构的审美理解所必需的训练的话，那么这种信息就碰到了障碍，反之，如果我是一个训练有素的音乐家，那它就碰到了一个可以对它作出解释的结构或组织，从而使这种模式在有意义的形式中展示出来，由是产生了审美价值和进一步的理解。语义学上具有意义的信息，在机器中一如在人体中那样，乃是能够通过接收系统中的激活机构的信息，尽管存在着人或自然乃至人和自然二者结合起来的捣乱企图。从控制论观点看来，语义学界定了信息意义的范围并使它在通讯系统中免于逸失。