Chapter 14 CHAPTER XIV. UNIVERSAL, SPECIFIC, AND NAMES

Since I gave up monistic logic, I have spent a lot of thinking about the problems related to the general and the particular, as well as to proper names.These are old questions, at least since Aristotle in fact.These questions occupied an important place in the speculation of medieval scholastics.The achievements of scholastics on this aspect are still worthy of attention.The difference between psychology and metaphysics in general was one of the points of conflict between Continental philosophers and British empiricists in the seventeenth and eighteenth centuries.I have written some traditional opinions in the form of an allegory and published them in the magazine "Polish" (1946, Part II, pp. 24-25): travel.They found a simple inn and asked for food to eat.The owner Dong promised to give them a large slice of beef.But when the meat was brought up, they thought it didn't taste good.One of these philosophers was a follower of Hume.He is a veteran of long journeys.He called the owner over and said, "This is not beef, it's horse meat."He didn't know that the innkeeper had lived a good life, but because he was devoted to philosophy and neglected his affairs, the situation was not as good as before; so the philosopher was surprised by the innkeeper's answer, and the innkeeper replied: "Sir, I Surprised to hear what you said. You know that what you say is meaningless. According to you, 'beef' and 'horse meat' are just words that have no meaning in the non-linguistic world. So this is just Literal controversy.

If you like 'horse meat'; that's fine, but I think it's better to say 'beef'". As soon as the innkeeper replied, the philosophers immediately began to talk about it.Among them was a philosopher who was a disciple of Roscelyn.He said: "The shopkeeper is right. 'Beef' and 'horse meat' are just the sounds of human breath, and neither can express this hateful piece of meat that cannot be chewed."A Platonist replied: "Absurd, this piece of meat came from an animal, and that animal was a copy of the immutable horse in the sky, not a copy of the immutable ox".An Augustinian said: "'Beef' and 'horse meat' are concepts in God's mind. I think the concept of God's beef must be different from this piece of meat." Only one thing these philosophers agree on is that, regardless of Whoever sells this unpalatable food as beef should sue the officials and say he is cheating.The owner of the shop knew that the magistrates did not understand philosophy, and he was terrified when he heard philosophers say this.He brought another piece of meat, which everyone ate to their satisfaction.

The point of this parable is that the "universal" problem is not just a matter of words.This question arises out of a desire to tell the truth. Speaking of me, I have come to two aspects: one is due to the study of Leibniz; the other is because many basic concepts of mathematics require asymmetric relations, and this asymmetric relation cannot be transformed into related relations. The predicates of the items cannot be transformed into the predicates of the whole composed of each item.After being convinced of the "reality" of the relationship, I can neither believe in the subject-predicate logic nor the opinion of empiricism, thinking that it is only special.

In the development of my philosophy after the abandonment of monism I have always retained (albeit with some changes) certain fundamental tenets which, though I do not know how to demonstrate them, I cannot bring myself to doubt them.The first of these is so obvious that I am ashamed to say it were it not for the contrary opinion being asserted.This first tenet is that "truth" depends on a certain relation to "facts."The second tenet is that the world is made up of many related things.The third tenet is that syntax, that is to say, the construction of sentences, must have something to do with the construction of things, those unavoidable aspects of syntax, (rather than peculiar to this or that language) , it must be so.Finally, there is a principle of which I am not so sure, but which I am willing to hold to, unless there is a very strong reason to forego it.

The principle is that to state the parts of a complex, and the relation of the parts to one another, without mentioning the complex, is to explain the complex. The use of symbols in Principia Mathematica implies the assumptions stated above.The notation used in this book assumes that there are "things" that have properties and relationships to other "things."There are basically two kinds of symbols I used in sentence construction. The first one means that a "thing" is an item in a category, and the second one means that a "thing" is related to another "thing". some kind of relationship.I use lowercase Latin letters for "things", lowercase Greek letters for classes, and uppercase Latin letters for relations.But classes were gradually replaced by attributes, and eventually, except for notational convenience, disappeared altogether.

The metaphysical beliefs involved in my symbolic logic I first attempted to illustrate in the fourth chapter of my Principia Mathematica, entitled "Proper Nouns, Adjectives and Verbs" .Roughly speaking, my thoughts at that time were related to the value specified by the variable.I have replaced variables with lowercase latin letters, and the possible values ​​of these variables are entities with attributes or relationships.A Greek letter refers to an attribute, or a class of things having that attribute. Capital Latin letters refer to relations.At that time I thought that to assign a value to a lowercase Latin letter was to replace the variable with a proper name.For example, if we know that whatever X is, if X is a person, X is mortal, we can substitute "Socrates" for "X".Likewise, we can substitute an attribute for a Greek letter; a relation for an uppercase Latin letter.This substitution of a constant term for a variable term is the process of applied logic.This process is outside the sphere of logic, since the logician, as far as he is concerned, is unaware of the existence of Socrates or of anything else.

My opinions at that time had the candor of the morning.This vigor disappeared after the hard work and heat of the day.At that time I thought that if a word contributes to the meaning of a sentence, that word must have a meaning. On this point I would like to quote from the forty-seventh section of Principia Mathematica: There are distinctions which are commonplace in philosophy, which are almost the same; I mean subject and predicate, substance and adjective, this and What's the difference between. With regard to these distinctions of the same nature I shall now point out the truth which I have seen.This is an important question, because the debate between monism and monadism, between idealism and empiricism, and between those who claim and reject truth as a matter of being, rests entirely or in part on the basis of what we have learned on this question. Take the doctrine for transfer.But we discuss this question here only because it is central to the theory of numbers or of the properties of variables.The relation of this question to philosophy in general, though not unimportant, I leave entirely alone.Whatever can be an object of thought, whatever appears in a true or false proposition, whatever can be counted as one, I collectively call a term.This is the broadest word in the philosophical vocabulary.I also use some words to be synonymous with the word item, that is, unit, individual, entity.The first two of these three are to emphasize that each item is singular, and the third is because each item has its existence, that is to say, in a certain doctrine, it is real.A person, an instant, a number, a species, a relation, a monster, and everything else that can be talked about is, of course, an item.It must always be wrong not to admit that something is an item.

It may be thought that a word so widely used cannot be of much use. But this opinion, which arises from some widely circulated philosophical doctrines, is incorrect.In fact, all terms have the attributes that nouns have.First, each term is a logical subject, i.e. each term is the subject of a proposition which is itself a term.Not only that, each item is immutable and indestructible.An item is what it is, and to conceive of any change in it must destroy its identity and make it another item.Another characteristic of an item is that it is numerically identical with itself and different from other items.The sameness and diversity of numbers are the source of the one and the many.So admitting that there are many terms destroys monism.This seems undeniable: every constituent of every proposition counts as a constituent, and every proposition contains at least two constituents.So term is a useful word because it indicates dissent from various philosophies.Another reason is that, among many statements, we are going to speak of any one or something.

In this passage, there are many things that I later thought were wrong.I changed my opinion because of the narrative doctrine and the typology doctrine.The narrative doctrine has convinced me that a word can contribute to the meaning of a sentence, but in isolation it can mean nothing.I used to think, for example, that the word "this" referred to something strange that a good logician might hope to encounter in Plato's heaven.The narrative doctrine made me give up this hope.The theory of types also made me abandon the naive and simple ideas in "Principles of Mathematics".I used to feel that certain words would lose all meaning if they were replaced by other words.I noticed that gerunds have the same meaning as verbs, but can be used as the subject of a sentence, for example, in the sentence "to kill is not to murder", "kill" is the case.Then I thought that sentences of this kind, if not meaningless, were shortened sentences in which the verb was a real verb, not a noun.For example, the sentence "killing is not murder" can be expanded to "if A kills B, it is not necessarily A who murdered B".If such a translation is impossible, the sentence is meaningless. The sentence "Socrates and killing are two" is an illegitimate sentence according to the typology; so is "Socrates and killing are one".

Another class of difficulties has to do with the strong doctrine against ontology.The particulars which I have represented by lowercase Latin letters seem to be substances in the sense of making sentences, though not necessarily with the indestructibility which is customarily assumed to be possessed by substances.If the statement "x has such-and-such properties" is always meaningful and unanalyzable, it seems that we can therefore say that x is something different from the sum of all its properties, and must be different. In another particular case, y, the difference is purely numerical.So it should be logically possible that all the properties of the two particular items x and y are shared by both.Of course, we cannot know that they are two, because that needs to know that x is different from y (y is not the case), in fact x will become an unknowable matrix, or it can be said to become an invisible like hams hanging from the beams of a farmhouse.With these points, "special phase"

There are difficulties with this concept, and we have to find a way to avoid difficulties. As regards the particular problem, my first attempt to deal with the above-mentioned difficulties was an article read to the Aristotelian Society in 1911 entitled "On the Relation of the Universal and the Particular."The presence of Bergson at that time did a great deal to honor the meeting.He felt very surprised and said that I seemed to think that what needs to be proved is a special existence, not a universal existence.In this article I analyze a hypothesis and think that it cannot be established (since then until now, I think it can be adopted).This assumption means that there is no need for a particular to be the theme to which attributes are attached. According to this assumption, lumpy attributes can replace particularity.The reason why I rejected this hypothesis at that time was because of the complexity problem of numbers and its relationship with time and space.At that time, I believed that mental phenomena were nothing more than the relationship between subject and object, and that the subject was a very subtle particular, which is the characteristic of the subject.First, based on the relativity of time and space, I assert that there must be something special in the world of sensations. Next, I have a claim that is very similar to what has been said above regarding the difference between two people.I said: From the space of perception, we can deduce the multiplicity of numbers.The argument for the multiplicity of numbers is strengthened by a similar argument, that of the mental content of peoples of all nations.This is at least theoretically possible: if two people both believe that two plus two equals four, the words two, plus two, equals, and four have the same meaning in their minds, so, as far as they are concerned, As far as the object of people's belief is concerned, it is really impossible to distinguish the two.Even so, it is evident that there are two realities, one is what one person believes, and the other is what another person believes.A particular belief is a complex in which there is an element which we may call the subject.As far as the example we have given is concerned, different subjects produce different beliefs.But these subjects are much more than just bunches of general properties.Assuming that one of these two persons has the characteristics of benevolence, stupidity, and puns, it would be incorrect to say "benevolence, stupidity, and puns believe that two plus two equals four."Even with the addition of many more general attributes, it would not be true to say so.Not only that, but no matter how many attributes we add, it is still possible that other subjects also have these attributes; therefore, the agents are not different because of the attributes.The point at which two different subjects must differ is their relation to particulars.For example, a subject has a relationship to another that it does not have with itself. But it is not logically impossible that everything pertaining to one subject, which otherwise pertains only to the general, might apply to another subject.So, even if the differences mentioned above occur, it is not because of these differences that the two subjects are different.Therefore, the subject must be regarded as particular, distinct from the collection of general attributes that the subject may have. Later I thought that these arguments could not be established.Regarding the sensory world, after thinking about it, it is obvious that the positions in the experience space are just like the positions in the physical space, and they are not relative.In my momentary field of vision, location is defined by properties.At the center of the field of vision there is a quality which we may call "centrality," and everything else I see in that moment has two qualities in varying degrees: up and down and left and right.But this is not the most important point that makes me abandon the opinion in that paper.The most important point is related to the logical properties of the space-time relationship. I think this relationship can generate continuity.For the sake of brevity, we speak only of time, and even of time in one's experience.We think that if A is before B, A and B must be different.We hold that if A precedes B and B precedes C, then A precedes C.If there is any doubt about these features of temporal relations, it is not easy to see how temporal continuities can be constituted.In 1911, I felt that the temporal continuum and the geometrical space could not be constructed without the material having a space-time location, which I felt could not be found without the recognition of particulars. The question of the composition of the dot-instant has been lingering in my mind since 1911.Soon Whitehead began working on the problem.I have developed a little on this question in my book Our Knowledge of the Outside World.At that time, I seemed to have seen that the nature of the particular (if there is a particular existence) used to constitute space-time should not belong to a point, but should be a point with a certain extent, as required by physics. The properties of the being belong only to bundles of particulars, each of which has a definite extent.But at that time, I did feel that if there were two reds in two places, there were two special reds.The reason we have to think of them as two has something to do with the relativity of position. At that time, I thought that the two pieces of red were only different in position, and because position is not a property (or I think it is not), the position must first have a variety as a condition, and it cannot constitute a variety.It is a different matter to admit that the position of the space of sensations is absolute.The two red blocks on my right can be a compound of two properties of red and right; the two red blocks on my left can be a compound of two properties of red and left.Left and right, as well as up and down, have various degrees of logical properties required by geometry, and the two red blocks seen at the same time are plural because of the combination of left and right and a certain property (such as red). I apply a similar reason to the order of time.Suppose a quality occurs twice in one's experience, such as a clock telling time.What is the reason why you recognize two taps as two taps instead of repetition of one thing?I have come to the conclusion that some cognitions depend on a quality which we may call the "subjective past".The contents of my mind, so far as they relate to the events of experience, may be arranged in a series beginning with sensations, followed by sensations of balance, then very recent memories, followed by sensations somewhat distant from the present memory.This produces a subjective time series, the terms of which, from an objective point of view, are now.When you hear the very similar sound repeated by the clock telling the time, the sound you have already heard has "gradual decay" to varying degrees.It is the complex formed by sound plus "fading" that is the majority, not the actual nature of the sound that is the majority.I developed this doctrine in my book Human Knowledge.I still find this doctrine satisfactory, and the reason why I like it more is that its belief does not require the assumption of unknowable and unknowable entities, and the result of denying it is necessarily to think that Particulars are those unknowable, unknowable entities. But there was another difficulty, which I considered insurmountable in 1911.It cannot be considered logically impossible that the two states of mind are identical.It may be said that this does not occur in one's experience because of the difference in the accompanying memories of the two periods.But according to the proof of logic, this kind of complete similarity can appear between the experience of two people, A and B.If so, my theory mentioned above obliges me to say that the mental state of A and the mental state of B are numerically the same.At first glance, this may seem unreasonable.We feel that it must be possible to see or constitute things of such a nature that, if one thing precedes the other, the two things are numerically different.But I think this view is due to the intrusion of experience into the realm of logic.As far as experience is concerned, we never see this complete reproduction.So far as we can find empirically, the whole content of a man's mind at one time is never quite the same as that of that man at another time, nor with that of any other man at any time. For those who dislike a priori intuitions outside the sphere of logic, my doctrine has an advantage.This strength lies in dealing with some instances of prior comprehensive knowledge. "If A precedes B, and B precedes C, then A precedes C", this sentence is undoubtedly synthetic, and it makes people feel as if it is a priori.According to my doctrine, this statement is still on the one hand synthetic, but not a priori, but generalized from our experience that the complexes that make up the mental content of an instant are never the same. appear again.From an empiricist's point of view, this is indeed an advantage. Now let me talk about a topic that is closely related to the universal and the particular, that is, the problem of proper names.But before I get down to it, I would like to say a few words about the controversial issue of logical languages.In my opinion, a logical language should be a language in which everything we want to say in terms of clear propositions can be said, and in which the structure can always be made explicit.In this language, we have to use words to denote structures, and we have to use words to denote items with this structure.At that time I argued that these terms could be indicated by proper names.I think this formation of language is a great aid to clear thinking, though I have never thought it useful for everyday life. At one time, Wittgenstein agreed with me, and also thought that a logical language would be useful in philosophy.In my introduction to his Tractatus I said that he had this opinion.It is a pity that by that time, not only had he given up that idea, but he seemed to have forgotten that he ever had that idea.So what I said about this opinion seemed to him to be inconsistent with the facts.His disciples have since strenuously denied the notion that a logical language could be useful. On an important point, I am willing to admit that their criticism is fair.At the beginning, like Leibniz, I believed that all complex things are composed of simple things, and when considering analysis, it is important to take simple things as our goal.Now I think that although we know that many things are complex, we cannot? know what is simple.Nay, the names of complexes are mentioned in sentences which may be quite correct, though the complexes are not considered complex.There are many advances in science that have been found to be complicated in what was once thought to be simple. Molecules, for example, are made of atoms, and atoms have a structure that has been understood in recent years. As long as we avoid saying that a thing under consideration is simple, what we say about it need not be contradicted by later discovery that it is complex.The whole question, therefore, of whether there is anything simple to be obtained by analysis is unnecessary. This has something to do with the question of proper names.I thought that, if we were omniscient, we would have a proper name for every simple thing, but we would not have a proper name for a complex thing, because to say the simple elements of which it is composed and the elements The structure of these complex things is clarified.I now discard this view. But I discard this insight, leaving many questions about the role of proper names. Traditionally, there are two kinds of nouns: proper names and public names. "Socrates" is a proper name; "Man" is a common name.But the common name is not necessary. The two sentences "Socrates is a man" and "Socrates belongs to mankind" mean the same thing, so the common name "man" is redundant and can be replaced by the attribute "of man".Attributes and properties are different and must be distinguished.The latter is a broader concept that includes the former.An attribute occurs in a proposition that contains only one name. "Socrates was of man" is an example. A property is what remains from a proposition in which a name occurs when that name is removed or substituted by a variable. For example, you could say, "If Socrates was willing to reconcile, he wouldn't have to drink poisonous wine." This can be regarded as saying that Socrates has a property, but it is not adding an attribute to him. Traditionally, the difference between a proper name and a common name is that a common name can have instances, while a proper name refers to something unique.But the concept of an instance is related to the concept of a class and is not logically fundamental.What logic requires are propositional functions, that is to say, expressions in which one or more variables are given a value, and the result is a proposition.The instance is thus the value of the variable of this correct propositional function.A variable can represent a variable "thing", or a variable attribute, or a variable quality, or a variable relationship.The invariant value that can be added to a variable depends on what the variable belongs to, and there are differences.If the total class of values ​​is added incorrectly, it becomes meaningless.Take the proposition "Socrates belongs to man" as an example. If you replace "Socrates" with the name of any other person or animal, regardless of whether the resulting proposition is true or false, the proposition still makes sense; not only that, If you substitute any other attribute for "of human beings," the resulting proposition still makes sense.If your proposition is a relational proposition, such as "Socrates loved Plato," you can substitute any other relational word for the word "love" without rendering the proposition meaningless, but you cannot substitute In any word that does not indicate a relationship. The above discussion implies a syntactic definition of proper names.We may say that a proper name is a word which does not denote an attribute or relation, and which may appear in a proposition containing no variable. (In ordinary language, the occurrence of a variable can be indicated by the occurrence of "a", "this", "some", "all", etc.).So far as syntax is concerned, I don't think there is much more to say about proper names. But we also have to do epistemological considerations.If a proper name is to fully fulfill its function, it should not have to borrow other words to delineate its definition.It should signify something that we perceive directly.But this aspect of proper names raises difficulties.If someone mentions Socrates, and you've never heard of him before, you can go to the encyclopedia and use what you find to define the name Socrates.Then, seriously, "Socrates" is not a name for you, but a substitute, a substitute for the narrative.Obviously, since other words have to be used to define words, there must be words whose meaning we do not know by definition.A child learns to know the names of the people in his family who are called by those names when they are around.Even if his parents are in an encyclopedia, the child does not learn who they are or what they are called from the book.This is the original use of the proper name, and the description of the proper name as omission is transferred.If you were once born in Athens and you said, "Who is Socrates?" the person you asked might point to and say, "That's Socrates." There is this distant connection, so that the proposition about Socrates is part of history, not part of the fiction, whereas the proposition about Hamlet is part of the fiction. "Hamlet" pretends to be a name, but it isn't.All propositions about Hamlet are fabricated.These propositions are only true if we substitute "Hamlet" for Hamlet.This is an illustration of one of the characteristics of proper names.This feature is that proper names, unlike narratives, are meaningless unless they denote an actual thing.Although France is now a republic, I can make some propositions about the present king of France.Although these propositions are fabricated, they are not without meaning.But if I falsely claim that he is Louis XIV, the proposition that "Louis XIV" is used as a name is not fabricated, but meaningless. I am not suggesting that in ordinary language or in grammar we should refuse to refer to (say) "Socrates" Think of it as a name.But, from an epistemological point of view, our knowledge of him is very different from our anecdotal knowledge of things.All that we know about Socrates is complete only if the narrative about him takes the place of his name, because, for us, our knowledge of the word "Socrates" comes entirely from the narrative. I have always advocated a principle that, if we can understand the meaning of a sentence, the words that make up the sentence must be exclusively words that refer to things we have witnessed or are defined by such words.This principle still seems to me to be perfectly valid.With regard to logically used words, such as, or, not, several, all, etc., it may be necessary to place some restrictions on this principle.If we limit the application of our principles to sentences containing no variables, or sentences consisting of sentences, we need not restrict ourselves.In that case, we may say that if our sentence adds an attribute to a subject, or asserts a relation between two or more terms, then the word used as the subject, or the terms of the relation, are not It is a proper name in a narrow sense. If we take this opinion, we have the problem of having to decide whether ordinary language is a word that contains proper names in the sense just mentioned.The question of the particular and the universal is related to our current question, but the relationship is not simple.We have to ask ourselves: What are the words we can understand without literal definitions?Not only that, except for the words used in logic, the words that we can understand without literal definitions must, in a sense, be words that refer to things that can be pointed out.For example, "red" and "blue" are words denoting a certain experience. We know the meaning of these words because we hear these two words when we see red or blue things.As for words in psychology, it is more difficult, such as "memory", but the principle is the same.If you see a child thinking about something, say to him, "Do you remember?" He will gradually know what you mean by that word.Only after this procedure can words establish a relationship to reality. This narrow designation can only be applied to things experienced, whether sense or thought.Whether the experience is simple or complex is irrelevant.But it's not an unrelated problem that we don't experience that little bit of particularity.The particularity about minutiae discussed earlier in this chapter is considered unnecessary.If the things discussed in psychology and the particles in physics can be understood by people, we must think of them as clusters of properties and relationships obtained through experience, or because of relationships already known from experience, and this The cluster-by-cluster properties are related.According to the above theory, the basic utensils for making proper names in common language must be made of the properties commonly called, rather than substances, such as red and blue, hard and soft, pleasant and unpleasant, etc. are examples of these properties . This requires a new arrangement in syntax.If there is a red thing in the center of our field of vision, we should not say: "This is red", but should say: "Red co-occurs with the center".If the red thing is not in the center of our field of vision, we have to replace the center with the appropriate left, right, up and down. I repeat, I am not suggesting the abandonment of ordinary speech in favor of this eccentric way of speaking.The matter may perhaps be made clearer by what I call a "minimum vocabulary." "minimum vocabulary" The definition is as follows: Assuming that there are several sentences whose meaning we understand, how many of them are the minimum words to be used to define the other words in these sentences?In general, there is more than one answer to this question, but each of these possible answers contains some words that are common to these answers.These words represent the center of experience upon which the relation of these sentences to the nonlinguistic world depends.I do not believe that one of these words has the unique quality that the special has.We might as well give a definition to the matter that constitutes the world, that is: the matter that constitutes the world is what some words refer to, and those words, if used correctly, are the subject of the predicate, or the item of the relation.In this sense, I think that the material that constitutes the world is made of things like "white", not objects with the property of white.This is the main conclusion of the lengthy discussion above.The main point of this conclusion is that it does not admit that the material that makes up the world is the mind and the finer things. If the above doctrine of qualities is admitted to be correct, the question of the status of the universal acquires a somewhat new form.传统上把性质如白、硬、甜等算做普遍，但是如果以上所讲的学说是正确的，这些性质在句子的构造上说是与物体更近一些。正如传统上的看法，这些性质之与物体有所不同，是因为没有空时的连续性，常识以为人和物件是有这种空时的连续性的。 有一些复合体是由共现的性质合成的。若是一个复合体的成分都彼此相共现，但是不与复合体以外的任何东西相共现，我称这样的复合体为一个“共现的完全复合体”。这样的完全的复合体代替了特殊，我们不说：“这是白的”，而说：“白是由我现在的意识内容组成的一个共现的复合体之中的一个成分”。 但是，虽然以上的学说适用于很多的传统上的普遍，却不能废掉对于普遍的需要。 仍然有宾辞所指的那些普遍，如颜色、声音、滋味等。显而易见，所有颜色都有一些共同的东西。我们可以举例来说明，你能经过极细微的颜色的浓淡从任何一种颜色到任何另外一种颜色。声音也是如此。但是没有法子从一种颜色渐次到一个声音。因为这些理由，我认为“红是一种颜色”是一个真正的主辞——宾辞的命题，把颜色这种性质加在红这个“实体”上。 但是比这种主辞——宾辞的命题更重要得多的是表明关系的命题。除非一种语言有方法说“甲是在乙的前面”，“甲是在乙的右边”，“甲比丙更象乙”，这种语言就不能把一切我们对于世界的知识表示出来。“在以前”，“相似”这一类的字或和这些字同义的字是语言不可少的一部分。也许实际上的这几个字不一定是必须的。用各种不自然的方法拿“相似”来代替很多的（如果不是一切的）表示关系的字是可能的。但是“相似”仍然是表示关系的字。假如这个表示关系的字必须保留，取消了别的表示关系的字是没有显著的好处的。表示关系的字是最顽强的字。在某种意义上，这些字的意思是属于普遍的。 在几乎一切关于普遍与特殊的学理中，有一点是向来为人所忽视的。那些不喜欢“普遍”的人一向认为普遍大概不过是一些字而已。这种主张的困难是，一个字本身就是一个普遍。“猫”这个字代表实际上很多猫。在口语上这个字是一套相似的声音，在文字上这个字是一套相似的形相。如果我们象唯名论者那样用力否认普遍，就没有象“猫”这个字这种东西，只有这个字的一些实例。这种讨论把我们引到普遍的更困难的一方面，也就是普遍在形而上学上的地位的问题。 当我们从表明事实的句子来到句子所表明的事实的时候，我们不得不问我们自己，句子的什么特征必是属于所表明的事实。“腓力普是亚历山大的父亲”，“亚历山大是在凯撒以前”，象这一类的句子显然是表明客观世界的一些事实的。 从前一些唯心论者说关系是心之所产；康德以为实在的事物并不存在于空间和时间里，那个时空体系却是我们主观的装备之所造。但是这一种关于关系的见解完全是根据一种错误的逻辑，只有那些看不出这种逻辑的含意的人才会承认那种见解是对的。至于我，象“甲早于乙”这一类表示关系的事实必定是有的。但是因此就能说有一个物件，它的名字是“更早”吗？这一个问题的意义是很不容易了解的，至于能够看出回答这个问题的方法是更加困难了。没有疑问，确是有些具有结构复杂的整体，若是不用表示关系的字我们就无法叙述这种结构。如果我们要发现关系字所表示的实体，能隐约中潜存于那个整体以外，我们能不能发现这实体是很不明确的。我以为明确的是一件关于语言的事，就是，从前已经提到过，关系字只应用来发生联系作用，用这种字来做主语的句子只有能译成另一些句子，关系字在里面表示项与项之间的关系的时候才有意义。换句话说，动词是不能没有的，动名词却不是不能没有的。这并不能回答以上所说普遍在形而上学上的地位的问题，但是这是我所知道的最接近的回答。 这一个题目的全部曾经在我的《对意义与真理的探讨》一书的最后一章里讨论过。 我所要说的话在那里已经说尽了，现在没有什么新义要说，所以我把那本书的最后两段引在下面：有些含有“类似点”这个辞的命题能够代以含有“相似”这个辞的同义命题，有些却不能。不一定要承认有些命题不能这样代替。例如，假定我说：“相似点是存在的”。 如果这里所说的“存在”是和我说：“美国的总统是存在的”那句话里的“存在”的意思是一样的，则我说的话是荒谬的。首先，我的意思可以用这样一句话来表示：“有些事件，为用文字来叙述，需要有象'甲和乙相似'这样形式的句子。”但是这件语言上的事却暗指一件事实，和所叙述的事件有关，也就是指我说“甲和乙相似”的时候所表明的那种事实。当我说：“相似点是存在的”的时候，我所要表明的是关于事物界的这件事实，不是关于语言的一件事实。 “黄”这个字是不能不有的，因为有黄的东西；“相似”这个辞是不能不有的，因为有一对一对相似的东西。两个东西的相似之点其确为一个非关语言的事实，正和一个东西之黄之为一个非关语言的事实是一样的。 在这一章里我们已经得到了一个结果，这个结果在某种意义上说是所有我们的讨论的目的。我心目中所指的结果是如此：完全的形而上学的不可知论是和支持语言上的命题相矛盾的。有些近代的哲学家主张关于语言我们知道的很多，但是对于别的任何东西则一无所知。这种见解忘记了语言也是一种经验上的现象，也忘记了一个主张形而上学的不可知论的人，当用一个字的时候，不得不否认他有所知。至于我，我相信略借造句法的研究，我们可以获得不少关于世界结构的知识。