1. We shall distinguish between norm and norm-formulation. The norm-formulation is the sign or symbol (the words) used in enunciating (formulating) the norm.

When the norm is a prescription formulating it in language is sometimes called the promulgation of the norm.

Norm-formulations belong to language. ‘Language’ must then be understood in a wide sense. A traffic-light for example normally serves as a norm-formulation. A gesture or a look even when accompanied by no words sometimes expresses a command.

The distinction between norm and norm-formulation is reminiscent of the distinction between proposition and sentence. We do not however suggest that the former distinction be regarded as a special case of the latter. Under a sufficiently comprehensive use of the term any norm-formulation could perhaps be called a ‘sentence’. But whether any norms can be called ‘propositions’ is debatable and that some (types of) norms cannot be so called is obvious (see below Section 8).

It is common to distinguish between the two ‘semantic dimensions’ of sense (connotation meaning) and reference (denotation) (cf. Ch. II Sect. 2). It is plausible to say that the sense of a descriptive sentence (Ch. II Sect. 2) is the proposition which it expresses. Some logicians and philosophers would wish to say that the reference of a descriptive sentence is the truth-value of the proposition which it expresses. It seems to me more plausible to say that the reference is the fact which makes the proposition expressed by the sentence true (cf. Ch. II Sect. 5). In this terminology we would have to say that only sentences which express true propositions have reference. Sentences which express false propositions lack reference. But they do not lack sense.

As far as I can see it would be misleading to conceive throughout of the relation between norms and their expressions in language on the pattern of the above two ‘semantic dimensions’. At least norms which are prescriptions must be called neither the reference nor even the sense (meaning) of the corresponding norm-formulations. The semantics of prescriptive discourse is characteristically different from the semantics of descriptive discourse. It must not be thought that the conceptual tools for dealing with the latter can as a matter of course be applied to a study of the former type of discourse as well.

What then is the relation between norm-formulation and norm if the second is neither the sense nor the reference of the first? We shall not discuss this question in detail. The following observation on the relationship under consideration will suffice:

When the norm is a prescription the promulgation of the norm i.e. the making of its character content and conditions of application (see Ch. V Sects. 2–6) known to the norm-subjects is an essential link in (or part of) the process through which this norm originates or comes into existence (being). The use of words for giving prescriptions is similar to the use of words for giving promises (cf. Ch. VII Sect. 8). Both uses can be called performatory uses of language. The verbal performance moreover is necessary for the establishment of the relation of norm-authority to norm-subject and of promisor to promisee.

For the reason just mentioned prescriptions can be called language-dependent. The existence of prescriptions necessarily presupposes the use of language in norm-formulations. This is not in conflict with the fact that prescriptions which have not been overtly formulated may sometimes become deduced as logical consequences of other prescriptions. What such deduction means will be discussed in Chapter VIII.

2. Are all norms language-dependent? Can there for example exist rules of a game which are never formulated in language and which are not logical consequences of formulated rules? One can learn to play a game without being told (all) its rules—e.g. by watching it. But this does not prove that the rules need not have been at some stage formulated. It is on the contrary reasonable to think that norms which are rules are language-dependent too. But the way in which rules are language-dependent is not exactly the same as the way in which prescriptions are language-dependent. The formulation of rules of a game is not a ‘performatory use of language’ at least not in the same sense in which the giving of orders or promises is so.

Can technical norms i.e. norms concerning the necessary means to given ends exist without being formulated in language? It can of course be ‘objectively’ the case that some agent ought to do a certain thing in order to attain a certain end of his but that neither he nor anybody else is aware of the necessary connexion. The connexion is not for its existence dependent upon a formulation in words. But the anankastic relationship is not the same as the technical norm (see Ch. I Sect. 7). Therefore one cannot from the language-independent character of the former conclude to the language-independent nature of the latter.

Customs we have said (Ch. I Sect. 6) exert a ‘normative pressure’ on the members of a community. Customs it seems are largely adopted through a process of imitation. In this they differ characteristically from norms which are prescriptions (laws regulations orders). Customs are not ‘laid down’ in the way rules (of a game) normally are; nor are they ‘promulgated’ as are laws and other prescriptions. Thus in the origination of customs language plays no prominent or typical role. Of all the things which may reasonably become included under the heading ‘norms’ customs are probably the least language-dependent. It is a question of some interest whether it should be regarded as essential to customs that they can exist only within communities with a language or whether one can speak of customs proper in animal communities too; i.e. the discussion of this question may contribute interestingly to the formation of the concept of a custom. We shall not however discuss it here.

Even if one cannot maintain the language-dependent character of norms without qualification and restriction it is obvious that there is a characteristic difference between norms and values in their respective relationship to language. Perhaps there are also types of valuation which are language-dependent in the sense that they are not logically possible among beings who do not master a language. But it is also obvious that there are reactions deserving to be called valuations on a pre-language level—among animals and infants. Roughly speaking: valuation is conceptually on a level with pleasure and want; norms are conceptually on a higher level. Norms can I think be said to presuppose logically valuations—but valuations can exist independently of norms. And that which substantially marks norms as conceptually higher than values is the dependence of the former upon language.

3. We shall here disregard norm-formulations such as gestures or signposts which are not ‘language’ in a narrower sense of this term. Disregarding them there are two grammatical types of sentence which are of particular importance to the language of norms. The one type is sentences in the imperative mood. The other is sentences which contain what I propose to call deontic auxiliary verbs. The principal deontic verbs are ‘ought’ ‘may’ and ‘must not’. We shall call the first type imperative sentences and the second type deontic sentences.

It is useful to raise separately the following two questions concerning the relation of imperative sentences to norms:

• (a) Are imperative sentences used chiefly or even exclusively as norm-formulations?
• (b) Can all norms be formulated by means of imperative sentences?

‘Imperative’ means in origin the same as ‘commanding’. From this it does not follow however that all uses of the imperative mood are for commanding. There are several typical uses of it which are not for this purpose. One is in prayers. ‘Give us this day our daily bread’ ‘Look upon us in mercy’. To say that these sentences express commands would not only be to depart grossly from ordinary usage; it would also be to ignore important features of logic. (The logic of prayer is different from the logic of command.) Prayers are not norms of the kind which we call prescriptions nor of any of the other kinds which we have distinguished. As we know the meaning of the term ‘norm’ is vague and flexible. There is no good ground however why prayers should be called norms.

Other typical uses of the imperative mood which are not for commanding are in requests (‘Please give me…’) and warnings (‘Don't trust him’). Requests and warnings are not norms of any of the kinds which we have distinguished. They could perhaps be called norm-like categories. They are more like norms than are prayers.

Consider also such forms of expression as ‘Don't be afraid’ ‘Take it easy’ ‘Let us assume that…’. These are common and typical uses of the imperative mood. But only under a strained use of the term ‘norm’ could we call the sentences in question norm-formulations.

The answer to the first of the above two questions is thus in the negative.

The answer to the question whether every norm can be enunciated in the imperative mood is complicated by the fact that the morphological character of the imperative mood in most languages seems to be rather indistinct. Whether a verb is said to be in the imperative mood often depends upon how the context in which it occurs is understood. ‘You take it easy.’ Is ‘take’ in the indicative or in the imperative mood? The question cannot be answered on the basis of considerations of grammatical form alone.

Imperative sentences which are used as norm-formulations are mainly used to enunciate prescriptions. There is some plausibility in thinking that every prescription of the O-character i.e. command and prohibition can be expressed by means of a sentence in the imperative mood—although part of the plausibility springs from our inclination to make the meaning of the sentence a criterion for calling its mood imperative. But permissive prescriptions or prescriptions of the P-character are ordinarily expressed by means of deontic sentences using the verb ‘may’ in combination with the verb for doing the permitted thing. If we take the view that permissions are prohibitions addressed to a ‘third party’ we could argue that they can be formulated obliquely in the terms of imperatives (‘Don't interfere…’ ‘Let him do…’). But even then the fact would remain that permissions when addressed directly to the permission-holder are normally expressed by means of ‘may’-sentences.

There is however a kind of imperative sentence whose normal function seems to be to enunciate permissions. I am thinking of the form ‘Do so-and-so if you want to’ or ‘Do so-and-so if you please’.

Occasionally imperative-sentences of the categorical form ‘Do so-and-so’ also express permissions and not commands or prohibitions. If when walking along the pavement I arrive at a street corner and the traffic light reads ‘Cross now’ the norm (prescription) addressed to me with these words is a permission to cross the street and not a command to do so.1

To say that the permission is incorrectly formulated because it is in the imperative mood would be sheer pedantry. But it seems plausible to regard imperative sentences of the categorical form ‘Do so-and-so’ when used for enunciating permissions as abbreviated or elliptic forms of hypothetical imperative sentences ‘Do so-and-so if you wish’. Thus the traffic light ‘Cross now’ addressed to pedestrians is short for ‘Cross now if you wish’.

Although imperative sentences as norm-formulations are mainly used for enunciating norms which we call prescriptions it would be a mistake to think that they are as norm-formulations used exclusively for that purpose. To say ‘If you want to make the hut habitable then heat it’ is grammatically no less correct than to say ‘If you want to make the hut habitable then you ought to heat it’. Both sentences would ordinarily be understood to mean the same. It would not be right to say that with the first sentence a command is given and with the second sentence a rule concerning means to an end. The function of the imperative mood in ‘If you want to make the hut habitable then heat it’ and in ‘If it starts raining then shut the window’ is different. The first imperative sentence expresses a technical norm the second a hypothetical prescription (command order).

4. There is a prominent tendency in contemporary philosophy including moral philosophy to lay strong emphasis on language. ‘Ethics’ a recent writer says2 ‘is the logical study of the language of morals.’ And moral language he thinks is ‘prescriptive language’3 so therefore ‘the study of imperatives is by far the best introduction to the study of ethics’.4 He is aware of the fact that imperatives ‘are a mixed bunch’5 but nevertheless decides ‘to follow the grammarians and use the single term “command” to cover all these sorts of things that sentences in the imperative mood express’.6 This is done because the author is interested ‘in features that are common to all or nearly all these types of sentence’.7 That there are such features he seems to take for granted and also that his readers are ‘no doubt familiar enough’ with the differences between the various kinds of imperative.8

I doubt the usefulness of the suggestion that philosophical ethics should start from a logical study of language in the imperative mood. I hope that some of my reasons for disagreeing with this view are plain from the above brief observations (in Section 3) on imperative sentences and their meanings. Neither as a morphological nor as a semantic category is the notion of the ‘imperative mood’ clear and homogeneous enough to make even a provisional identification of norms with the meanings of sentences in this mood plausible.

To characterize the language of norms as ‘prescriptive’ would not be unplausible. It would however imply either a much broader use of the term ‘prescriptive’ or a much narrower use of the term ‘norm’ than we are making here. Prescribing and prescriptions in our use of the words certainly play an important role in the moral life of man. But unless we take a theonomous view of morality moral norms (principles) can hardly be regarded as prescriptions in our sense of the word. And regardless of whether we call moral norms ‘prescriptions’ or not it is doubtful whether moral norms can be formulated in the imperative mood. Consider for example the principle that promises ought to be kept. We can and often do urge people to keep their promises by addressing them with ‘Keep your word’ and similar imperative sentences. One can using such sentences command people to keep their word and prohibit people from breaking their word. This is prescriptive use of language. It is use of language for moral purposes and in this sense ‘moral language’. But the moral norm (principle) that promises ought to be kept is hardly the same as the command (or prohibition) which ‘Keep your word’ and similar imperative sentences may be used to enunciate. The proper linguistic medium for formulating moral principles is not language in the imperative mood.

Ethics moreover is concerned with values as well as with norms. To characterize the language of valuations as ‘prescriptive’ seems to me rather misleading.9 And therefore to base the philosophical study of values on a logical study of imperatives would be misleading too.

There is a sector of linguistic forms which may be said to bear to value-judgments a relation which is somewhat analogous to the relation which norms bear to sentences in the imperative mood. These are the part of speech and the syntactical category called interjections. Very roughly speaking: To value is more like exclaiming than like prescribing. To say this is not to deny that evaluative and prescriptive discourse are logically closely related. Nor is it to suggest that the study of interjections is the best or even a good introduction to the study of value.

5. The two questions which we raised in Section 3 concerning the relation of imperative sentences to norms can be raised mutatis mutandis also for deontic sentences:

• (a) Are deontic sentences used chiefly or even exclusively as norm-formulations?

• (b) Can all norms be formulated in terms of deontic sentences?

In answering the questions we have to make allowance both for the unsharp nature of the concept of a deontic sentence and for the unsharp nature of the concept of a norm.

It is reasonable to think that the answer to the second of the above questions is affirmative. One could make it a partial definition of ‘norm’ that every norm is to the effect that something ought to or may or must not be or be done. It would then follow trivially that every norm can become expressed in a deontic sentence.

Quite apart from the question of definition of ‘norm’ however it is obvious that deontic sentences have a much richer semantic capacity as norm-formulations than imperative sentences. This is so for two main reasons. One is the absence of a peculiar form of ‘permissive imperative’ corresponding to the deontic word ‘may’; the other is that the imperative form when used in norm-formulations is used typically for expressing norms which are prescriptions. Deontic sentences it seems have no such alliance with one particular type of norm.

The answer to the first of the above two questions is without doubt negative. Besides the use of deontic sentences as norm-formulations there are two other equally common and typical uses of them.

The one is the use of deontic sentences to state anankastic (see Ch. I Sect. 7) relationships. ‘If the hut is to be habitable it ought to be heated’ does not express a norm but states a fact about necessary connexions in nature. It is on the loose definition we have given to be counted as a deontic sentence.

Although sentences which state necessary connexions often use the word ‘ought’ to express the necessity they can also be formulated using the word ‘must’. For example: ‘If the hut is to be habitable it must be heated.’ It may be thought that the ‘must’-sentence is a more adequate expression of the anankastic relationship than the ‘ought’-sentence. In any case it seems always possible to replace an ‘ought’-sentence which is used to state an anankastic relationship by a ‘must’-sentence. But it would certainly be contrary to common usage if we suggested that ‘ought’-sentences which are used as norm-formulations can always be replaced by ‘must’-sentences. ‘Must’ is typically an anankastic word. ‘Ought’ is anankastic or deontic.

Another typical use of deontic sentences other than their use as norm-formulations is for making what I propose to call normative statements. What is meant by a normative statement will be explained later (see below Section 9).

6. It must not be thought that imperative and deontic sentences are the only grammatical types of sentence which are used as norm-formulations. Indicative sentences other than deontic sentences are also quite commonly used for expressing norms.

When the norm is a prescription and its expression in words is an (ordinary) indicative sentence the future tense is often used. ‘You will be leaving the room’ does not necessarily express a prediction. It may just as well express a command—and be synonymous with the imperative sentence ‘Leave the room’ and the deontic sentence ‘You ought to leave the room’.

In legal codes norm-formulations in the indicative mood either in the present or in the future tense seem particularly common. When for example in the Finnish constitution we read: ‘The President of the Republic assumes office on 1 March next after the election’ this is not meant as a description of what the president habitually does but as a prescription for what he ought to do.—I have noted that in Swedish criminal law the indicative form answering to ‘is punished’ or ‘will be punished’ and the subjunctive form answering to ‘be punished’ are used indiscriminately to express norms to the effect that so and so ought to take place. The Swiss penal code I understand consistently uses the indicative form throughout.10

7. I hope that the observations on the language of norms in the preceding sections will have made it clear that norm-formulations linguistically are a very varied bunch. They cut across several grammatical types of sentence without including or being included in any one type. One must therefore warn against the idea of basing the conceptual study of norms on a logical study of certain linguistic forms of discourse. Deontic logic i.e. the logic of norms is not the logic of imperative sentences or of deontic sentences or of both categories jointly—just as propositional logic is not the logic of indicative sentences.

Whether a given sentence is a norm-formulation or not can never be decided on ‘morphic’ grounds i.e. seen from the sign alone. This would be true even if it were the case that there existed a grammatically (morphologically and syntactically) sharply delineated class of linguistic expressions whose ‘normal’ or ‘proper’ function is to enunciate norms. For even then it would be the use of the expression and not its ‘look’ which determines whether it is a norm-formulation or something else.

When we say that it is the use and not the look of the expression which shows whether it is a norm-formulation we are in fact saying that the notion of a norm is primary to the notion of a norm-formulation. For the use to which we refer is itself defined as use to enunciate a norm. We thus rely upon the notion of a norm for determining whether an expression is used as norm-formulation or not.

8. It is appropriate to say something here about the relation of norms to truth. Are norms true or false? Or shall we think on the contrary that norms lack truth-value that norms ‘fall outside the category of truth’?

The question has been the object of much dispute. It is useful to raise it separately for the various types of norm which there are. Maybe the answer is not the same for all types. Here we shall consider it very briefly for some main types of norm only.

Have rules e.g. of a game a truth-value? Of rules of a game we have said (Ch. I Sect. 4) that they determine a concept. Chess e.g. is ‘by definition’ the game which is played according to such and such rules. That a rule of a game cannot be false seems plain. We may be mistaken in thinking that there is a rule to such and such effect or that according to the rules such and such a move is or is not permitted in a certain game. What is false is then a proposition about the rules. The false proposition is not itself a rule—not even a false one.

Since rules of a game obviously cannot be false does it follow that they must be true? Some would I think call them analytic (or necessary) truths. I would myself not call them truths at all; and I should be inclined to take the same attitude to rules generally. It is not necessary however to argue the point in detail here.

Are technical norms true or false? For example that if I want to be at the station in time for the train I ought to break up the party now? What is certainly true or false depending upon anankastic relationships in nature is the proposition that unless I break up the party now I shall not be at the station in time. What is also true or false depending upon my present condition is the proposition that I want to be at the station in time for the train. The technical norm however is not the same as the anankastic proposition. Nor is it the conjunction of the two propositions about necessary relations and wants respectively. The relation of the technical norm to these two propositions is not clear to me nor is therefore the relation of the technical norm to truth and falsehood.

The status of moral norms (principles and ideals) in relation to truth and falsehood we shall not discuss in this work at all.

That prescriptions lack truth-value we can I think safely accept. Or would anyone wish to maintain that the permission given by the words ‘You may park your car in front of my house’ or the command formulated ‘Open the door’ or the prohibition ‘No through traffic’ are true or false?

Those philosophers who have defended the view that norms generally lack truth-value have sometimes it seems implicitly identified norms with prescriptions. If by ‘prescription’ we understand commands and permissions which are given by some norm-authority to some norm-subject(s) the identification of norms with prescriptions must appear much too narrowing. If again we understand ‘prescription’ in some wider sense it may become doubtful whether the thesis that prescriptions lack truth-value can be upheld.

To accept the view that prescriptions and perhaps other types of norm too lack truth-value does not of course constitute a hindrance to saying truly that norm-formulations of prescriptions and other types of norm have meaning or that they make sense.11 Whether we shall say that the sense or meaning of a norm-formulation is the norm which it enunciates is quite another matter. A full discussion of the question would raise problems of philosophical semantics which we cannot treat within the scope of the present work. Some comments on the topic were made in Section 1.

9. Suppose I say to someone for example in reply to a question: ‘You may park your car in front of my house.’ Is this a norm-formulation? It is easy to see that there are two possibilities to be considered here.

In replying with those words I might actually have been giving permission to the questioner to park his car in front of my house. In this case the sentence was (used as) a norm-formulation. It did not say anything which was true or false.

But the same words might also have been used for giving information to the questioner concerning existing regulations about the parking of cars. In this case the sentence was a descriptive sentence. It was used to make a true or false statement. I shall call this type of statement a normative statement.

The very same words may thus be used to enunciate a norm (give a prescription) and to make a normative statement. This ambiguity moreover seems to be characteristic of deontic sentences generally (cf. above Section 5).

Which use is in question in the individual case may not be instantly clear. Sometimes both uses are involved at the same time. One and the same token of an ‘ought’-sentence for example may be used both to remind the receiver of an order of the fact that he has been given this order and to give new emphasis to (reissue) the order itself. The possibility however that the meanings thus mix does not entail that they could not be logically sharply distinguished.

The systematic ambiguity of deontic sentences was as far as I know first clearly noted and emphasized by the Swedish philosopher Ingemar Hedenius.12 He coined for (an aspect of) the distinction between the two uses the terms ‘genuine’ and ‘spurious’ legal sentence. Genuine legal sentences are used to formulate the legal norms themselves. Spurious ones are used to make existential statements about legal norms (normative statements).

10. A normative statement schematically speaking is a statement to the effect that something or other ought to or may or must not be done (by some agent or agents on some occasion or generally unconditionally or provided certain conditions are satisfied). The term ‘statement’ is here used in that which I propose to call its ‘strict’ sense. A statement in the strict sense is either true or false. (The sentence which is used in making the statement expresses a proposition.)

By the truth-ground of a given normative statement I understand a truthful answer to the question why the thing in question ought to or may or must not be done.

Let the normative statement be for example that I may park my car in front of your house. Why may I do this? The answer could be that there is a regulation according to which I am permitted to do this. The existence of this regulation (norm prescription permission) is the truth-ground of the normative statement.

Also with a view to the norm (permission) that I may park my car in front of your house the question ‘Why?’ may be raised. The proper answer to this question ‘Why’ is not that there is this norm (permission). The answer tells us why this norm (permission) has been given. The answer thus makes reference to the aims and ends (motives) of the authority who granted the permission.

Generally speaking the truth-ground of a normative statement is the existence of a norm. This holds good as far as I can see not only for prescriptions but for the other types of norm as well. Why is it that in chess a pawn which has reached the last line may become exchanged for a queen? Because there is a rule which gives this ‘right’ to the players. Why is it that I ought to break up the party now? The answer could be that I want to be at the station in time for the train and that unless I leave now I shall be late. Here the existence of a technical norm is the truth-ground of the normative statement.

The proposition that such and such a norm exists I shall call a norm-proposition. For example: that there is a regulation permitting me to park my car in front of this house is a norm-proposition. The norm-proposition is true or false depending upon whether the norm in question exists or not.

The existence of a norm is a fact. The truth-grounds of normative statements and of norm-propositions are thus certain facts. In the facts which make such statements and propositions true lies the reality of norms. The problem of the nature of these facts can therefore conveniently be called the ontological problem of norms. Some aspects of this problem will be discussed in the next chapter.

• 1. I am indebted to Professor Tranøy for drawing my attention to this clear-cut case of a ‘permissive imperative’.
• 2. R. M. Hare, The Language of Morals (1952), Preface, p. v.
• 3. Op. cit., p. 1.
• 4. Op. cit., p. 2.
• 5. Op. cit., p. 4.
• 6. Op. cit., p. 4.
• 7. Op. cit., p. 4.
• 8. Op. cit., p. 4.
• 9. Hare (op. cit., p. 3) classifies imperatives and value-judgments under the common heading ‘Prescriptive Language’. This tends to obscure the conceptual (logical) difference between norms and valuations. A ‘classical’ example of how distinctions may become blurred in this region is the following quotation from the writings of a distinguished contemporary philosopher: ‘It is easy to see that it is merely a difference of formulation, whether we state a norm or a value-judgment. A norm or rule has an imperative form… actually a value statement is nothing else than a command in a misleading grammatical form.’ For an early criticism of these confusions see the paper by Torgny T. Segerstedt, ‘Imperative Propositions and Judgments of Value’ in Theoria, 11, 1945.
• 10. Cf. O. Brusiin, Über das juristische Denken (Soc. Sci. Fenn. Comm. Hum. Litt. XVII 5, 1951), p. 51.
• 11. Yet there was a time not long ago when it was seriously maintained in some philosophic circles that norm-formulations actually are ‘meaningless’ because removed from truth and falsehood. This illustrates the power of philosophic dogmas—in this case the so-called verificationist theory of meaning—of perverting the philosopher's use of language.
• 12. In his book Om rätt och moral (‘On Law and Morals’, 1941). See especially op. cit., pp. 65 f.