My present concern is the meaning of value terms. In section I I shall first suggest a satisfactory criterion for a theory of meaning, namely that it should give a coherent explanation for both fact and value expressions. Then I shall argue that, none of the currently available theories of meaning satisfy this criterion. Section II will develop an inferential conception of meaning and apply it to the expression of facts. Section III discusses its appli-cation in value expression. According to this conception, the meaning of a word or a judgement consists in an inference or a syllogism. That is , by an inference, we understand the meaning of a word from the major premise (usually a definition), and under-stand the meaning of a judgment from the major premise to conclusion. Applied to value expression, this consists in the spea-ker’s conception of value implied in the major premise and shown in the conclusion.

Nevertheless, in my opinion, at least three questions play a central role in the theory of meaning, that is (i) what is meaning or what are we saying about a word when we say what does it mean?; (ii) what is the criterion of meaning i.e., what requirements must a language satisfy if it is to be called meaningful?; and (iii) how to clarify the meaning, i.e., how many different kinds of meaning are there in our language or what sort of meaning does this sentence have? The currently available theories of meaning all give a certain kind of answer to these questions in some way or other.

The problem is that although every theory of meaning gives some answers to the questions of meaning to some extent, none of these theories can give a coherent explanation for both fact expression and value expression. In other words, a criterion for a satisfactory theory of meaning should have a coherent application to all kinds of linguistic expression.

1. Let us look first into referential theory, which is the most natural answer to the question of meaning. This holds that the meaning of a word is that to which it refers, or a kind of object (referent) or bearer of this word. The meaning of a sentence is its truth value: "Dartmouth" means a city to which this word refers. The difficulty of this theory is that, as some philosophers realized, it seems that meaning is the object itself. Then if the bearer of a name died, the meaning of this name no longer exists. Also, the naive version of this theory cannot explain the fact that some words have different meanings, but one referent (Venus-morning star and evening star), and the same meaning has different referents (I, you, here, etc.). And, since most value terms do not refer to any object (in a broader sense they do refer to some relations or properties), most value judgements do not represent facts. Therefore, the referential theory meets an obvious difficulty when applied to value expressions. In the example: "The democratic system is good." what is the meaning of the word "good", to what object does it refer? for the whole sentence, "the democratic system is good" has a meaning. But according to referential theory, since this sen-tence does not represent any fact, in what does the meaning of this sentence consist? In another example: "Utopia is a beautiful paradise." Although Utopia is a fictional paradise, the sentence has a moral and aesthetic meaning. However, according to referential theory, this sentence is meaningless, since we cannot find a place called "Utopia", and we cannot prove the truth or falseness of this sentence.

2. Facing a word like "Dartmouth", ideational theory will say that the meaning of Dartmouth is the idea or association when you hear this word. You might have an imagination that Dartmouth is the mouth of River Dart, or a port city in England. Ideational theory avoids some difficulties of referential theory, that is, we can talk meaningfully about something which no longer exists or never has existed, and a word cannot mean the same for all who speak or hear it, since each person has a different idea. But there are more serious difficulties involved in this theory. Firstly, an idea is private so that we cannot prove whether the idea in my mind is the same as the idea in another’s mind. Secondly, we cannot have any idea of some words, such as, `if’, `when’, `course’, or some phrase like `a four-dimensional space.’

Facing the question of the meaning of a value judgement, ideational theory appeals to the idea or imagination. The trouble is that if we have no idea of what Utopia is, we would have no imagination at all. Moreover, what is the idea of good, right or beautiful? Everyone has his or her own idea. In the case of judgements of fact, such as it is a table, it seems possible for two persons to have almost the same idea by looking at some table from the same direction; however, in the case of the value judgement, there seems no way to know what the speaker’s idea of ‘good’, ‘right’ or ‘beautiful’ is, since there is no such thing called ‘good’ we can refer to or compare with.

3. Generally speaking, use theory finds the question of the meaning of Dartmouth difficult to answer, since they hold that only in a linguistic context can a word have meaning. The meaning of a linguistic expression is its use in the language game. To understand a word, is to grasp its different usages in different contexts.

The main advantage of use theory is that it gives some insight that language, like a game, has different rules and uses, so the meaning of a linguistic expression is not a particular object or entity. Wittgenstein even warns "don’t look for meaning, but look for use." By this sentence, Wittgenstein tries to avoid a traditional mistake, namely, if we ask a question in the form "what is meaning" we naturally wish to find a something which is the meaning of this sentence. According to Wittgenstein this would lead to an ignorance of the multiple uses of linguistic expressions.

However, we must notice that the admonition "don’t look for meaning but use" does not imply that there is not anything called "meaning." Wittgenstein tries to tell us that the best way to understand the meaning of a word is to understand its use in a lan-guage, or to grasp its rules in a language game. Strictly speaking, use theory is a theory of how to understand the meaning of a word in a language.

Therefore, use theory does not mean that meaning is use itself. If consistent, it never asks a question like "what is the meaning?" replying "meaning is the use" for that would conflict with Wittgenstein’s own principle by taking the use itself as an entity.

The vagueness of use theory is that by understanding Witt-genstein’s claim that meaning is the use in the language people usually forget the purpose of grasping the use of a word. In fact only when one wants to know the meaning of a word does one seek its use; by grasping this use one can understand the particular meaning of a word in a certain context. The best way then to ex-press use theory is that meaning consists in the use in a language, rather than that meaning is the use in a language.

Use theory in some sense has an advantage inasmuch as it can give an explanation for such value terms as "good’ and "right". For example regarding the meaning of "good" use theory simply holds that to understand its meaning is to grasp its uses in different con-texts. However, facing some more complicated value judgements, use theory fails to give any useful interpretation.

"She is a virgin" is a typical example of a descriptive-eva-luative judgement. If used in a hospital’s examination room, it is a descriptive sentence; if you hear it at a wedding in a remote village in China as an old woman shows a bloodstained sheet to the guests at a wedding, the sentence means "she is a morally good woman" or "she is virtuous". If you come from a Western country you do not understand that and must ask people in this situation what does it mean.

Use theory can tell you to try to grasp the use of "virgin" in different contexts. Still you cannot find the evaluative meaning, because the moral meaning of virgin is not contained in the use of the word "virgin" in some countries. The deep conception of moral value implied in this word can be revealed only by showing a tra-ditional moral value in China through an inference.

All those who are virgins until getting married are morally good.

She is still a virgin on her wedding day.

So she is morally good.

Obviously, the key to understanding this sentence is to know Chinese moral values about sex and marriage, which was concealed in the major premise.

A means nn by X that P

That is, by uttering X, "She is a virgin", the speaker’s intention is that P, "She is morally good." For Grice, the key to understanding this non-natural meaning is the audience’s recognition of the speaker’s intention. So to say that A means nn something by X is to say that A intended the utterance of X to produce some effects in an audience by means of the recognition of this intention.

Grice’s theory, according to Searle, is a useful start on an analysis of meaning, because it shows the close relationship between the notion of meaning and the notion of intention, and secondly because it captures something which is essential to spea-king a language: in speaking a language I attempt to communicate things to my hearer by means of getting him or her to recognize my intention. Searle then criticizes Grice’s theory for failing to dis-tinguish the different kinds of effects, and for failing to account for the extent to which meaning is a matter of rules or convention. In my opinion, the fatal defect in Grice’s theory is that, although he gives us three steps to show that meaning consists in the reco-gnition of a speaker’s intention, he fails to gives us a clear ex-planation of how to let the audience recognize the speaker’s in-tention. Therefore, when people actually ask "what does A mean nn by X that P", what they can be told from this theory is that so long as you recognize the speaker’s intention, you can understand the meaning of a nonnatural sentence.

Once again, my challenge is that, although it is true that within a certain linguistic community people can understand some non-natural meaning of a sentence by showing some linguistic con-vention, how can I understand a linguistic convention if I come from a foreign linguistic community? In the case of "She is a virgin", while other Chinese know what this sentence means by their conventions, the Westerner still has to wait for an answer to what this convention means.

So as I have shown above, all these theories fail to give us a satisfactory explanation of how to let the audience know your in-tention, or your community’s conventions. As I have shown earlier, by an inference one immediately gets a definition of a certain value term, or value criterion, or a convention from the major premise, and an understanding of the speaker’s meaning or intention from the conclusion.

To turn to the emotive theory of ethics, it concerns the question: what is the nature of ethics? Ethical naturalism believes that the nature of an ethical sentence is one class of empirical description. Ethical intuitionism holds that the nature of an ethical sentence is its nonnatural characteristics. The emotive theory insists that the nature of an ethical sentence is neither descriptive nor nonnaturalist, but just an emotive expression. As a theory of meaning, emotive theory tells us the kind of meaning an ethical sentence has, that is, the ethical sentence is meaningful for it expresses the speaker’s emotion or attitude. To say someone is "good" is to appreciate him or her. In this sense the ethical sentence has emotive meaning whereas a fact sentence has descriptive meaning.

Emotive theory was developed under the influence of the view that the meaning of language is to be looked for in the use to which it is put. By showing different uses of an ethical term, emo-tive theory tells us that the basic meaning of ethical terms is to ex-press appreciation, condemnation, and so on.

Take an example, "the table is square": what kind of psy-chological effect will it cause if you understand this sentence? To explain the meaning of a fact expression in terms of psychological disposition, in most occasions is far from accurate. Take as another example giving a command to somebody. According to this theory the meaning of this sentence should be an action it causes, but if the command is not obeyed no action is involved. Therefore, a com-mand which was disobeyed would be meaningless, which ob-viously is absurd.

S means M can be replaced by

S means P and further, since "means" is ambiguous too,

S means P can be replaced by

S is true if and only if P.

Therefore, a condition for a satisfactory theory of meaning in essence is Tarski’s Convention T that test the adequacy of a formal semantical definition of truth. That is, the meaning of a sentence consists in its truth condition; to understand the meaning of a sentence is to know its truth condition. Thus, a theory of truth for a language is at same time a theory of meaning. Now if you know "snow is white" to be true if snow is white, you know at same time that "snow is white" means that snow is white.

As Davidson locates the meaning of a sentence in its truth condition, this theory can explain the meaning of a sentence in a simple and accurate way. But the problem is, how can this theory apply to a value sentence, since, as Davidson himself realized, moral or evaluation sentence do not have a truth value? That is, how can we understand any value sentence which has no truth value if the truth value is the presupposition of the understanding of meaning? In the end of his article, Davidson admits that this is a difficulty his theory faces, and he suggests that a comprehensive theory of meaning should solve this problem.

The analysis in the last section set up criteria for a satis-factory theory of meaning and showed that the currently available theories of meaning fail to meet these in one way or another. Now it is time to develop an inferential conception of meaning.

An inferential conception of meaning means that the meaning of a word or judgment constitutes an inference or syllogism, that is, by showing an inference. We understand the meaning of a word from the major premise (usually a definition), and understand the meaning of a judgment from the major premise to conclusion. Con-sider these two sentences:

This table is square.

Stealing is wrong.

The first is a fact judgment and hence cannot be understood until you know the word "table", "is" and "square". To understand a fact judgment, usually you need:

(1) to understand the word in this sentence by a major premise or a definition, then

(2) you will understand the meaning of the whole sentence by an inference.

For example:

A shape with four straight equal sides forming four right angles is square.

This table has the same feature as the above definition,

so this table is square. (You know what the whole sentence means.)

As all fact judgments either describe some fact or assert/deny some states of affairs, to understand a fact sentence also means that you know its truth condition. Therefore,

(3) to know the truth condition of the sentence means you will have such an inference:

A shape with four straight equal sides forming four right angles is square.

This table has the same features as the definition,

so "this table is square" is true.

As we can see, if you just want to know the meaning of a word in a sentence or a word without context, you just need step (1). That is, to have a major premise or definition in a certain context, or, to have a general definition without a context is to give a full defi-nition. Therefore, the inference, in the case of the word, not the sentence, becomes a simplified inference. That is, if you do not know what is the meaning of "square" in the above sentence, you just need a major premise or definition, then you will understand "this table is square" even without a full inference. Actually, this is what people do in ordinary life. To understand some fact judge-ments like "this table is square", people do not make a full in-ference. Once they know the meaning of "square", they im-mediately understand why this table is asserted to be square and whether this is true or not. This is because people have a lot of knowledge of fact, and this is why people sometimes just need a simplified inference to understand a fact judgement.

If this is the case, the inferential conception of meaning in the case of a word and some fact judgments can be simplified as a major premise or defining conception of meaning. Thus, we come almost to the common sense most people naturally hold.

The reason for saying that the meaning of a word, in the case of a fact judgment and especially of a word, can be reduced to a definition is that the meaning is always something you intend to tell people by a word or a sentence, or which you grasp when you say you understand a word or a sentence. The best way to grasp this something is the definition of the word which tells what this word means, how many different meanings this word has, and how to use this word in different situations.

This conception shares some basic principles with referential theory in the sense that both theories hold that meaning is always something expressed by a word or a sentence. The difference between the two is that referential theory identifies this as an entity or an object itself, whereas inferential theory holds that this is just what a word means, i.e., a definition in the case of a word, a value criterion (also a definition) in the case of a value term, or a principle of value for a value judgment. This approach can avoid falling into difficulty in a fictional word, meaning entity or fixed meaning of a word in referential theory.

To say that the meaning of a word is a definition does not mean that the meaning is independent of context. Actually, a definition always tells you different uses of this word in different contexts. So the meaning of a word without a context is the completed definition, whereas the meaning of a word in a certain context is one from the many meanings this word contains. When one wants to find out the meaning of a word in a sentence, one normally choose an accurate meaning among several. In the case of value terms, the meaning in one’s value judgment is always one’s value criterion; the definition of a value term is yours, not others.

Therefore, the definition theory of meaning of a word borrows some basic principles from use theory, but avoids its in-ability to give a satisfactory explanation to a word and the problem of some complicated descriptive-evaluative sentences.

The above concerned descriptive expression and concluded that for the meaning of a fact-expression one usually appeals to a simplified inference. This is just a major premise or a definition. Now how about value expressions? Here there is a conventional distinction between the descriptive/emotive (or referential/emo-tive, cognitive/noncognitive) function. Carl Wellman even classi-fies five different functions of language: descriptive, emotive, evaluative, directive and critical. For convenience of expression, I discuss non-descriptive expressions under the name "value ex-pression", although some of them are not really evaluative. For pre-sent purposes this simplification will have little affect on the conc-lusions.

The value judgement will include:

moral judgment The democratic system is a good system.

The man is bad.

non-moral This meal is good.

This is a good car.

aesthetic judgment This flower is very beautiful.

critical judgment This picture is too realistic.

This argument is invalid.

obligation judgment Stealing is wrong.

Telling the truth is right.

A professor ought to be paid more, and

descriptive-evaluation

judgment She is a virgin.

She is very tender (cultured, etc.)

As we have shown earlier, there is a main difference between a fact and a value judgement: the main function of the former is to represent a fact, whereas that of the latter is to express evaluation, criticism, attitude, emotion, etc. As a result, for a fact expression, meaning is connected with the truth value, whereas a value ex-pression does not need such a condition. Also, for understanding a fact expression a simplified inference is enough in most occasions, whereas for understanding a value expression a full inference is necessary. Because the meaning of value term varies from one person to another, to understand a value expression depends to a great extent on one’s understanding of a speaker’s value criteria, of certain conventions in a community and so on, which one cannot grasp until a full inference is drawn.

Since the two kinds of expressions have such obvious dif-ferences, a natural question is whether it is necessary or possible for us to have a coherent theory of meaning for both expressions? Currently available theories of meaning are fully aware of these differences, so they pay attention to one kind of expression (for example, referential theory to fact expression and emotive theory to value expression). The point is that if we cannot find a coherent theory of meaning we cannot say that current theories are unsatisfactory. But if we do find a coherent theory of meaning which can apply to both expressions and at same time does not ignore their differences, then there is valid reason to say that the current theories are unsatisfactory.

Now let us turn to the value expression. Like a fact expres-sion, the meaning of a value expression also is implied in the major premise or definition. How is this meaning to be grasped. Again, we can appeal to an inference by showing a syllogism where we understand the meaning from a major premise to conclusion.

Since a judgement can be used either as a minor premise or conclusion depending on the context, to start an inference in the case of a value judgment, the first step is to determine what role the given value judgement plays in a syllogism. For example:

The democratic system is a good system.

(1) as a conclusion:

Any system which can bring happiness to the people is a good system.

The democratic system can bring happiness to the people,

so the democratic system is a good system.

(2) as minor promise:

All good systems can bring happiness to the people.

The democratic system is a good system,

so the democratic system can bring happiness to the people.

Generally speaking, it is easier to understand a value judgment as a conclusion. To understand your meaning or your intention, we need only to know what is meant by "good system", which I know by a major premise.

This becomes more complicated in the case of a value judge-ment as a minor premise, because with the same minor premise we can draw different conclusions from different major premises. Therefore, if someone says a sentence X means P, as Grice’s A mean nn by X that P, to understand the speaker’s real meaning or intention you must know the major premise.

Within a linguistic community, this major premise can some-times easily be identified or recognized. Suppose your friend wants you to drive him home, and you do not intend to do so, you say: "I drank to much." He will understand immediately what you mean because of a shared conventional principle:

No driving after excessive drinking.

I drank too much,

so I cannot drive.

However in some cases, the major premise or the speaker’s special view of value is too difficult to recognize. For example, after a dinner party two men say the same sentence: "This meal is very good." If they use this sentence as a minor premise, they mean something else by this sentence. The first man’s wife thinks her husband is appreciating her, because her husband has a principle or a convention for a good meal:

All good meals must have been cooked by my wife.

This meal is good,

so this meal must have been cooked by my wife.

The second man’s wife knows that her husband wants to say: "I am going to sleep" as he has a habit or conventional inference:

Whenever I have a good meal will go to sleep.

This is a good meal,

so I am going to sleep.

If neither of them let people know their principle, no one would understand what this sentence really meant.

So, to understand a value judgement as a minor premise, you need two steps:

(1) Whenever you hear a sentence like this is a good X, or this X is good, you naturally treat this judgment as a conclusion. What you need is to know his criterion of a good X, say, a good meal. This can be achieved by an inference.

(2) If he says "No, this is not what I mean," that means he meant something else by the sentence. Now you should take the second step and ask: "What do you really mean by that?" He then gives you his principle. In real life, people usually just give you a principle or a major premise and leave the conclusion to you. That is we ignore the role a syllogism plays in understanding the meaning of a value judgement, or more generally, the meaning of the linguistic expression. Apart from the distinction between a value judgment as a minor premise and a value judgement as a con-clusion, we can find another difference between a value judgment with a basic value term (good, bad, right, wrong and ought) and a value judgment without a basic value term. We call the former first level judgment and the latter second level judgment.

Let us look into first level judgment first. There are two kinds of structure in this level.

(1) X is good.

(2) This X is a good X.

By uttering (1), you are setting up your general value criteria. If people do not understand you, that is because they do not know your own view of "good", not because they do not know the com-mon sense meaning of good, since everyone always presupposes they know the basic meaning of good and bad. Therefore we find an interesting fact that whenever people want of know your view of good, say, in the sentence "The democratic system is good," they simply ask "why" rather than what is the meaning of good. By asking "why", they want to know why you attribute the democratic system to "good". Your answer is:

Anything which brings happiness is good.

The democratic system can bring happiness,

so the democratic system is good.

Your major premise is your conception of "good".

In (2), "good" is used as an attribute. By "good X" you give your value criteria of a certain kind of thing, say: "She is a good doctor." If people do not understand your sentence, most likely they do not know your criteria of a "good doctor", rather than "good". Therefore, by showing your criteria of a good doctor, you give a concrete value criterion.

Now let us turn to the second level. The value terms in this level are: satisfactorily, reasonable, admirable, and so on. These terms express evaluation, criticism, comment, and eventually can be reduced to the basic value term (good, bad, right, wrong, ought). Therefore, to understand this level of judgment, you need first to attribute them to the first level value term, and, according to the linguistic convention, to locate their degree of good, bad, right, wrong.

The difficulty in this level of judgement is that, not every value term has an obvious good or bad meaning. In some cases, the value term is so ambiguous that you do not even know whether it is approving or disapproving, let alone the degree of good or bad. This is very common in art and literary criticism. For example, "This picture is realistic." Firstly, you probably do not know what is meant by "realistic". After being told that "realistic" means "very concrete, not abstract", you still do not know the speaker’s attitude towards this picture, good, or bad? So you ask further: "What do you mean by this sentence?" His answer is:

A good picture should be abstract.

This picture is not abstract,

so this picture is not good.

As we can see, to understand a second level value judgment is much more difficult than a first level value judgment. Moreover as shown above, since with the same judgement as a minor premise from different major premises we can draw different conclusions, the possible meaning for this judgment could be something other than good or bad. For instance:

No modern artist likes a realistic picture.

This picture is realistic,

so this picture must be from a classical artist’s hand.

This shows again that to understand a value judgment we need to make not only a distinction between a judgement as a minor premise and a judgment as a conclusion, but also a distinction be-tween a first level judgement and a second level judgment. In the case of the second level judgement, you sometimes need three steps to find:

(1) the meaning of a second level value term (realistic)

(2) the meaning of the sentence (attribute them to good/bad and the degree of good/bad)

(3) the speaker’s meaning or (as a minor premise) non-basic value term meaning

In short, the meaning of a value expression consists in the speaker’s conception of value implied in the major premise (usually a definition), which can be revealed only by an inference or a syllogism. Therefore, the meaning of a linguistic expression consists in the course of inference.

1. J. Heal, "On the Phrase `Theory of Meaning’", Mind (1978).

2. G. Harman, "Three Levels of Meaning", Journal of Philosophy (1968).

3. P. Grice, "Meaning", Philosophical Review (1957).

4. P.F. Strawson, "Intention and Invention in Speech Act", in his Logico-Linguistic Papers.

5. D. Lewis, Convention.

6. C.L. Stevenson, "The Emotive Meaning of Ethical Term", Facts and Value, also see Ethics and Language.

7. D. Davidson, "Truth and Meaning", Syntheses (1967).