Sorites (Σωρίτης), ISSN 1135-1349
Issue # 20 -- March 2008. Pp. 117-124
Kant and the Expression of Imperatives
Copyright © by Ronald Cordero and Sorites

Kant and the Expression of Imperatives

by Ronald Cordero

1. Introduction

Not far into the Second Section of the Foundations of the Metaphysics of Morals, Kant makes an interesting claim about the expression of imperatives. In the original German it is `Alle Imperativen werden durch ein Sollen ausgedrückt ...'Foot note 1, and in the popular Lewis White Beck translation it is rendered as «All imperatives are expressed by an `ought' ...»Foot note 2. The assertion in the English version is extremely perplexing. Are not most imperatives expressed without an «ought»? In fact, are any imperatives ever expressed with «ought»?

From the point of view of moral theory, the question is important. Kant is going to claim that the moral law, which he believes to be both universal and necessary, is created by the categorical imperative that comes from reason. Can this crucially important imperative be expressed by an «ought» judgment? And if it can, how is it that such a judgment can give rise to apodictic law?

Moreover, such a claim about imperatives and «oughts» has significant consequences for the analysis of ethical utterances. To say that imperatives are expressed by «oughts» is to say that at least some «ought» judgments are expressions of imperatives. If this is in fact what Kant has in mind, then he would seem to be saying that using «ought» can be a way of issuing a command; and this has a distinctly noncognitivistic ring to it. On this interpretation, Kant would appear to be in sympathy with A. J. Ayer's famous position on «ought»:

The sentence «It is your duty to tell the truth» may be regarded both as the expression of a certain sort of ethical feeling about truthfulness and as the expression of the command «Tell the truth.» The sentence «You ought to tell the truth» also involves the command «Tell the truth,» but here the tone of the command is less emphatic.Foot note 3

On the same interpretation, Kant would also seem to be assenting, at least in part, to Hare's insistence that «ought» in moral contexts is used with a prescriptive rather than a descriptive force.Foot note 4 Are we then to suppose that Kant is proposing a noncognitivist or prescriptivist analysis of at least certain «ought» judgments? Does he believe that an utterance like «One ought not to lie,» may be nothing more than the issuance of an imperative against lying -- and thus not capable of being either true or false?

What is needed is a careful analysis first of just what is involved in the expression of imperatives and the creation of laws -- and then of Kant's text. This is the only way to dispel the appearance of misstatement on his part, settle the question of his possible endorsement of a noncognitive interpretation of «ought» judgments, and clarify his claim that morality consists of law created by an imperative.

2. Expressing Imperatives and Making Rules

There are at least two important senses in which one can speak of an imperative being expressed. In one of these, to express an imperative is simply to issue it, to give it as a command; in the other, to express an imperative is to state or report it in indirect discourse. Expressing a certain kind of imperative in the first sense can create a rule. Expressing the same kind of imperative in the second sense can state or report a rule.

When an imperative is expressed in the first sense -- when it is issued -- the utterance expressing it is noncognitive, having only what J. L. Austin would call the illocutionary force of commanding or directing.Foot note 5 Suppose we wish to direct (to order or command) Smith to be here by five. We could, of course, issue the imperative in any of thousands of languages. In English, we could issue it by saying to Smith any of the following ...

The first version simply uses the imperative mood. The second employs a future tense in a way that stipulates rather than predicts. And the third involves a construction that can plausibly be regarded as equivalent to «You are hereby directed to be here by five.» Each of these three utterances can be meant and taken as an issuance of the imperative, and if they are so taken, none of them can be regarded as cognitive. None could be either true or false. In using the imperative mood one is not making a statement. Nor is one doing so by stipulating required future behavior as in (2) or by explicitly directing it as in (3). If «You will be here by five,» were understood as a prediction, it would constitute a statement, not a directive. And if «You are to be here by five,» were taken to mean «You have been ordered to be here by five,» it would constitute a statement about a directive given in the past and not an issuance of the directive in the present. But taken respectively as a stipulation and a directive, the latter two utterances are just as noncognitive as the first.

It is important to understand at this point that imperatives expressed in the sense of being issued may or may not be standing imperatives. They may or may not, that is, concern more than a single action on a single occasion. Consider these examples:

1) Always be home by eleven.

2) Students will not run in the halls.

3) Travelers are to carry identification at all times.

Parents who issue the imperative in (1) to their child are giving an order that will remain in effect until it is rescinded. The directive given by some school authority in (2) is a standing order about conduct by any student at any time in any hall in the school. And the order issued in (3) is meant to apply to an unlimited number of travelers on an unlimited number of occasions. All three examples represent the issuance of imperatives that are, as it were, universally quantified. And this, presumably, is precisely the kind of imperative which Kant takes the categorical imperative issued by reason to be. Reason says to us, as it were, «Always act in such a way that ....» The command reason issues is not just about a single occasion. It is a standing imperative, applying to all times and all places.

The second important sense in which one can speak of imperatives being expressed involves the giving of cognitive reports. Here to express an imperative is not to issue it, but rather to report it. Suppose the media want to report the directive we issued when we said to Smith, «Be here by five.» They could do so in a variety of ways:

1) They said, «Smith, be here by five.»

2) They said for Smith to be here by five.

3) They said Smith was to be here by five.

4) They said Smith should be here by five.

5) They directed Smith to be here by five.

Here the media would not be issuing or reissuing our imperative: they would simply be reporting what we did, and their statement could be accurate or inaccurate, true or false. The first version reports the order in what is called «direct discourse» (using a direct quote). The next three examples use «indirect discourse» to report the same order without a direct quote. Any of the five utterances would count as a statement, a cognitive account of what transpired that could be true or false. And any of the five could also be said to express our imperative by reporting it.

While examples (1) through (5) involve statements about a past event, it should be noted that present-tense statements can report commands just as readily. Suppose we are in the process of parking in a foreign country when a police officer approaches and says something in a language we do not understand. We may ask someone to translate and the translator may say, «The officer says you are to park somewhere else,» expressing the officer's directive in the present tense in a way that reports it cognitively but does not, as it were, reissue it.

As in the case of the issuance sense of the expression of imperatives, it is essential to note that imperatives expressed in the sense of being reported may also be standing, rather than «one-time-only,» imperatives:

1) The king says we are to pay taxes annually.

2) The king said, «All subjects will pay taxes annually.»

In asserting either (1) or (2) we would be reporting a standing command originating from the monarch and could certainly be said to be expressing -- though not issuing -- a standing imperative. We could, of course, be mistaken in either case. We might, for example, simply have misunderstood the king's command.

Here it is important to note that reporting a standing command through a statement is tantamount to stating a rule -- since rules are commonly established through the issuance of standing commands. To report a standing command, that is, can be to state a rule. If we direct our daughter to be home tonight by eleven, we are issuing a «one-time-only» command, not laying down a rule. But if we instruct her always to be home by eleven -- if we issue the standing command «Always be home by eleven,» we are laying down a rule. The standing imperative establishes the rule. And a statement reporting our standing command would be a statement of the rule. Our daughter might, for example, express the rule in indirect discourse by saying, «My parents say I always have to be home by eleven.» Similarly, if the authorities direct all travelers to carry identification, their standing imperative gives rise to a rule, which can be stated by reporting the directive: «The authorities say all travelers are always to carry identification.» The cognitive expression of the standing order is a statement of the rule established by that order.

Similarly, on Kant's theory, the categorical imperative delivered by the reason of each rational being establishes the apodictic moral law. And that law can be stated through a cognitive report of the categorical imperative. The noncognitive standing categorical imperative, «Act always in such a way that ...,» establishes the basic moral law, and that law can be stated through a cognitive report of the categorical imperative: «Reason directs us always to act in such a way that ....»

3. What Kant Says

Turning now to Kant's text, we must deal first with his contention that all imperatives are expressed by a «sollen» (an «ought» in Beck's translation). Should we not reject this claim outright as simply mistaken? After all, we know that imperatives, German and English alike, can be -- and indeed characteristically are -- expressed in the imperative mood. There is absolutely no need for a «sollen» or an «ought» to be used in the expression of any imperative. And this is the case whether we are talking about imperatives that are hypothetical, such as «If you want to have a good reputation, keep your promises,» or imperatives that are categorical, like «Keep your promises.» The imperative mood provides a perfectly good way of expressing imperatives, with no need for «sollen» or «ought.» As a matter of fact, in the Foundations Kant himself repeatedly expresses the fundamental moral imperative simply through the imperative mood, without «sollen»:

Act only according to that maxim by which you can at the same time will that it should become a universal law.Foot note 6

Act so that you treat humanity, whether in your own person or in that of another, always as an end and never as a means only.Foot note 7

[T]he categorical imperative can be expressed also as follows: Act on those maxims which can at the same time have themselves as universal laws of nature as their object.Foot note 8

So why would Kant say that all imperatives are expressed by a «sollen»? He clearly cannot mean that they are always so expressed.

The most reasonable assumption to make here may be that what Kant means is simply that all imperatives are capable of being expressed with a «sollen» -- not that they are always expressed in that way. Such an interpretation saves the passage from being obviously false, though it also leaves unanswered the question of what Kant means when he talks about imperatives being expressed in such a manner. Exactly how does he conceive this expressing? Is he thinking of imperatives being expressed in the sense of being issued or of imperatives being expressed in the sense of being reported?

As far as expression through issuance is concerned, there is, in fact, a good basis for Kant's claim with regard to «sollen.» It is a fact that in German a form of the verb «sollen» is commonly used for issuing commands.Foot note 9 And it is important to notice that this is done in a way that contrasts with a common way in which commands are issued in certain other European languages. Consider for example the following different versions of the commandment about adultery found in Exodus 20:

English: Thou shalt not commit adultery.

Spanish: No cometerás adulterio.

French: Tu ne commettras point d'adultère.

German: Du sollst nicht ehebrechen.

There is an obvious difference here between the way the commandment is issued in German and the way it is issued in the other three languages. In the English, Spanish, and French versions of the commandment, a future-tense form is used -- as indeed was the case in the original Hebrew.Foot note 10 In German, on the other hand, «sollen» is employed and there is no future tense: the «sollst» in the German is a present-tense indicative form of the verb.

As already noted, the use of the future tense to issue commands is not uncommon in English. If we want to tell someone not to park on the grass, we can say, «You will not park on the grass.» And it is significant to observe that in such a case we could usually not issue the order unambiguously with «ought.» If we said, «You ought not to park on the grass,» we would run the risk of not being understood to be giving an order at all. We might be taken to be making a recommendation, offering advice, or a giving a suggestion instead.

German, however, does not use the future-tense for issuing commands.Foot note 11 When it does not simply use the imperative mood, it frequently uses «sollen» in the present tense; and Kant, with his well known pietist background, must have been familiar with moral commandments issued in this way. Indeed, in the Preface to the Foundations he cites the commandment against lying as «Du sollst nicht lügen,»Foot note 12 and Beck translates it using the future tense as «Thou shalt not lie,»Foot note 13 not «Thou ought not to lie.» Again, in the Second Section of the Foundations, Kant formulates the commandment against false promises as «[D]u sollst nichts betrüglich versprechen ...»Foot note 14 and Beck translates «Thou shalt not make a false promise ...,»Foot note 15 not «Thou ought not to make a false promise.»

So we might well expect Kant sometimes to think of the fundamental moral imperative being issued in such a way, using «sollen» rather than the imperative. In fact, he often does so, though he uses the first person rather than the second person. This is perfectly understandable, of course, since Kant is thinking of the categorical imperative being issued to each individual by her or his own reason. Reason is telling us what to do. In the First Section of the Foundations, for example, Kant gives a preliminary formulation of the categorical imperative using «sollen» in this way:

... d.i. ich soll niemals anders verfahren als so, daß ich auch wollen könne, meine Maxime solle ein allgemeines Gesetz werden.Foot note 16

That is, I ought never to act in such a way that I could not also will that my maxim should be a universal law.Foot note 17

Here Beck translates «ich soll niemals» as «I ought never.» But by analogy to the preceding examples, the phrase could also be translated as «I shall never act in such a way that ...»

Evidently, if Kant is thinking of expressing commands in the sense of issuing them when he speaks of imperatives being expressed by a «sollen,» it is far from certain that «ought» is the corresponding term to use in English. It would, in fact, be misleading to say that Kant holds that imperatives can be issued by an «ought.» It would be truer to his meaning to say that on Kant's position, imperatives can be formulated in terms of what one is to do or shall (stipulatively, not predictively) do. If, for example, I express (issue) the categorical imperative in German with «sollen,» the corresponding expression in English might best be realized with «shall» for the stipulative future tense or with «are to»: «Rational beings shall (are to) act only in ways in which they can consistently will everyone to act.» Saying, «Rational beings ought to act only in ways which they can consistently will everyone to act,» could mistakenly give the impression that the utterance was intended as counsel or advice, rather than as a command.

And what of the expression of imperatives in the sense of reporting? Could Kant have had this in mind as well? The relevant syntactic fact here is that «sollen» is indeed commonly used in German to express commands in the sense of reporting them. It is, one might say, the verb of choice for reporting commands in indirect discourse. And in the passage originally referenced, Kant could definitely have had indirect discourse in mind. Indeed, the assumption that that is at least part of what he has in mind certainly helps make sense of the passage.

In German «sollen» is used in the subjunctive, not the indicative, mood for reporting commands in indirect discourse. In formal English the verb form used for the same purpose is «should»:Foot note 18 If last night the director said to us «Get those reports in by Friday,» we would report the imperative in indirect discourse in formal English by saying, «The director said we should get those reports in by Friday.» This would be a cognitive report of a command. In less formal English, of course, we have other options when it comes to expressing imperatives in indirect discourse. We might, for example, prefer a construction with the infinitive: «The director said for us to get those reports in by Friday.»

Further, because a standing command establishes a rule, cognitive reports which express standing commands through indirect discourse can count as statements of rules. To report the rule about getting home at a certain time, our children could say, «Mom and Dad say we should always be home by eleven,» or «Mom and Dad say for us always to be home by eleven.» So since Kant conceives of the basic moral law as something created by a standing command given by reason, he might be expected to express (state) it with «sollen» in the subjunctive in indirect discourse. And so he does -- in what is referred to as the second formulation of the categorical imperative in the Foundations:

Denn vernünftige Wesen stehen alle unter dem Gesetz, daß jedes derselben sich selbst und alle anderen niemals bloß als Mittel, sondern jederzeit zugleich als Zweck an sich selbst behandeln solle.Foot note 19

For all rational beings stand under the law that each of them should treat himself and all others never merely as means, but in every case at the same time as an end in himself.Foot note 20

Here the «solle» at the end of the German passage is the subjunctive form of «sollen» in question -- and Beck renders it with a «should» rather than an «ought.» One could also say: «under the law which says for each of them to treat himself and all others ....» On Kant's position, the basic moral law, established by the categorical imperative issued by reason, can be reported in terms of what reason says for us to do.

But what about «ought»? Can it be used to express imperatives in the sense of reporting them? And can statements about what ought to be done express standing imperatives in such a way as to constitute statements of rules or laws?

First consider non-standing imperatives. Can we, for example, report the directive «Get those reports in by Friday,» by saying «The director says we ought to get those reports in by Friday,»? Here again -- as in the case of issuing imperatives -- to do so would be to run a significant risk of being misunderstood. Someone might think that the director only recommended -- and did not order -- that we get the reports in by Friday. Formal English calls for a «should,» not an «ought,» for the expression of commands in indirect discourse; but even «should» may not avoid ambiguity in these cases. If we say, «The director says we should get those reports in by Friday,» someone might think the director merely advised us -- rather than told us -- to get them in by Friday. This may explain why everyday English tends to prefer «The director says for us to get them in,» to «The director says we should get them in.»

As much can be said for the expression of standing imperatives. If the authorities say that travelers are always to carry identification, it could be misleading to say the authorities say travellers ought always to carry identification. The authorities are not recommending something: they are requiring it. And if what the law says is «All citizens shall pay taxes annually,» it would be misleading to describe the law as saying all citizens ought to pay taxes. What the law says is that they have to do so.

For Kant, the categorical imperative issued by reason does much more than merely issue a recommendation or give advice. Because of this, it would be highly misleading to report it in a way that could be understood to present it as merely recommending or advising. But to assert that the categorical imperative says we ought to act only in certain ways would be to do precisely this -- and so would be to misrepresent the imperative completely. According to Kant, the categorical imperative establishes a universal and necessary moral law. And such a law cannot be adequately or unambiguously stated by talking about what ought to be done.

4. Conclusion

When Kant speaks of the expression of imperatives, whether he has in mind the issuance of imperatives by means of a noncognitive utterance or the reporting of imperatives by means of a cognitive utterance, he is not saying that imperatives can be expressed by «ought» judgments. He does not think that reason phrases its categorical imperative in terms of what ought to be done. Nor is he saying that the basic law of morality can be stated in terms of what ought to be done. The suspicion raised earlier about the possibility of Kantian noncognitivism regarding «ought» judgments in the Second Section of the Foundations can thus be seen to be unwarranted.

Ronald Cordero
University of Wisconsin at Oshkosh
cordero [at]

[Foot Note 1]

Immanuel Kant, Grundlegung zur Metaphysik der Sitten, ed. Karl Vorländer, (Hamburg: Felix Meiner, 1952) 33. (Royal Prussian Academy edition page 413).

[Foot Note 2]

Immanuel Kant, Foundations of the Metaphysics of Morals, trans. Lewis White Beck (New York: Macmillan, 1990) 29.

[Foot Note 3]

Alfred Jules Ayer, Language, Truth and Logic (New York: Dover, 1952) 108.

[Foot Note 4]

R. M. Hare, Moral Thinking (Oxford: Clarendon, 1981). See, for example, 215.

[Foot Note 5]

How to Do Things With Words (New York: Oxford University Press, 1962) 99-103.

[Foot Note 6]

Beck, 38. (RPA 421).

[Foot Note 7]

Beck, 46. (RPA 429).

[Foot Note 8]

Beck, 54. (RPA 437).

[Foot Note 9]

See, for example, George O. Curme, A Grammar of the German Language (New York: Frederick Ungar, 1964) 321.

[Foot Note 10]

For the point about Hebrew, I am indebted to my colleague Marshall Missner.

[Foot Note 11]

See Bill Dodd, Christine Eckhard-Black, John Klapper, and Ruth Whittle, Modern German Grammar: A Practical Guide (New York: Routledge, 1996) 339-41.

[Foot Note 12]

Vorländer edition, 5.

[Foot Note 13]

Op. cit., 5. (RPA 389).

[Foot Note 14]

Vorländer edition, 40.

[Foot Note 15]

Beck, 36. (RPA 419).

[Foot Note 16]

Vorländer edition, 20.

[Foot Note 17]

Beck, 18 (RPA 402).

[Foot Note 18]

T. Herbert Etzler and Harvey I. Dunkle, A German Review Grammar (New York: Odyssey, 1965) 34.

[Foot Note 19]

Vorländer edition, 57.

[Foot Note 20]

Beck, 50. (RPA 433).

maintained by:
Lorenzo Peña
Editor of SORITES