Presuppositions of Questions and Commands

Our definition of presupposition states that a sentence S presupposes sentence P if “in every situation where S is accepted, P is also accepted”. Since the presupposition(s) of S are defined in terms of S's acceptance conditions, it seems that any with a presupposition must have some acceptance conditions.  That is: it seems that such a sentence, S,  must be acceptable in certain situations, in order for S to have a presupposition. If S were the kind of sentence that, in principle, could never possibly be accepted, then we could never apply our definition of presupposition to S (since there simply are no “situations where S is accepted”), and so S could have no presuppositions (as we define them).

So any sentence having presuppositions must be a sentence at least capable of being accepted. Which kind of sentence is that? When we accept a sentence, we accept it as true; so the kind of sentence which is capable of being accepted will just be the kind of sentence capable of being true. Which sentences can be true? Declarative sentences, of course. So: it looks like the definition of presupposition, as it stands, allows only declarative sentences to have presuppositions.

But we can extend the notion of presupposition to cover even sentences not capable of truth or acceptance – that is, to non-declarative sentences – as long as we can find some natural way of associating the non-declarative sentence with some family of declarative sentences. The non-declarative sentence can thereby get a presupposition “by association” with the declarative sentences (which already have presuppositions).  

Put another way: the non-declarative sentence will “inherit” any presupposition possessed by the family of 'associated' declarative sentences.  We can then think of presuppositions of non-declarative sentences as involving a new kind of presupposition inheritance.


1. Questions
.  Questions are not themselves declarative sentences, but any question is naturally associated with a certain set of declarative sentences – namely, the set of possible direct answers to that question. (We take the idea of a “direct answer” for granted here, though it should already be a familiar concept.) Some sentences have a very small family of direct answers. For example a "yes/no” question like
(Q1) Is today Tuesday?
has only two possible direct answers:
(1) (Yes,) Today is Tuesday
(2) (No,) Today is not Tuesday

But other questions can have many more possible direct answers. Consider, for example, a "why?” question:
(Q2) Why did Rex start the fire?
We can easily think of numerous possible answers to (Q2):
(3) Rex started the fire because he wanted to attract attention
(4) Rex started the fire in order to keep warm
(5) Rex started the fire because he wanted to toast marshmallows
In fact, we may not be able to list every possible direct answer to (Q2); but that doesn't mean we can't still talk about the set of all such answers, whatever they may be. (In the same way, we can't list every single even number, but we can still talk about, and know things about, the set of all even numbers.  They are, for instance, all divisible by two.) In particular, it is clear that any direct answer to (Q2) can be phrased in some form such as
(6) Rex started the fire because...
(7) Rex started the fire in order to...
and so on. And from this simple observation we can draw an equally obvious conclusion: any direct answer to (Q2) will be a sentence that presupposes (8):
(8) Rex started the fire
Notice, for example, that (3) through (5) all presuppose (8). Indeed, any sentence of form (6) or (7) will be a “causal/explanatory sentence,” and so presuppose (8) in this way.

On the basis of this observation, we can extend the notion of presupposition to questions such as (Q2).

Presupposition of a Question: If P is presupposed by every direct answer to question Q, then P is presupposed by Q.

So since (8) is presupposed by every direct answer to (Q2), we can now say that (8) is presupposed by (Q2) itself. And that seems intuitively correct, after all: if I ask why Rex started the fire, I do seem to be presupposing that he did, in fact, start the fire. So we extend presuppositions to questions in a way that seems entirely natural and correct. Of course, we don't think every question is necessarily going to have a presupposition. But then, our definition of a question's presupposition doesn't make any such claim either.  It simply allows some questions to have presuppositions – which is, again, only natural.


2. Commands.  The same strategy used to build presuppositions for questions can be used for commands as well: allow the command to “inherit” the presupposition(s) of some set of associated declarative sentences. But while questions have a natural set of associated declarative sentences – their direct answers – it is not so obvious what the associated sentences will be for commands. While a question naturally calls for a (declarative) answer in response, a command often doesn't call for any sentence at all in response, but rather some appropriate action, which complies with the command.  But we can still build a kind of 'sentence counterpart' to the appropriate action, by declaring (in a declarative sentence) that we will, or will not, comply with the command and perform that action. Such a sentence will be called a compliance sentence. So, for instance, suppose someone utters (C1):
(C1) Drive below the speed limit!
There will be two compliance sentences one can make in response:
(9) I will drive below the speed limit.
(10) I will not drive below the speed limit.
But notice that both compliance sentences contain a definite description – “the speed limit” – and so both compliance sentences presuppose (11):
(11) There is a (unique) speed limit (in this particular situation).
We can therefore define 'presuppositions of commands', parallel to the definition of 'presuppositions of questions':

Presupposition of a Command: If P is presupposed by every compliance sentence for a command C, then P is presupposed by C.

So since (11) is presupposed by every compliance sentence for (C1), (11) is presupposed by (C1) itself. Again, this should seem entirely correct: if I command someone to drive below the speed limit in a certain situation, then I do seem to be presupposing that there is, in fact, a unique speed limit in that situation.


3. Presuppositions and Rejections.  So far we have only considered direct answers to questions, and compliance sentences (which basically play the role of direct answers for commands). But those sentences alone do not provide a complete account of how one can respond to a question or command. That is, so far our account of questions (for example) allows a participant in a conversation only one possible response to a question: give a direct answer to it. But consider question (Q3):
(Q3) Why did Zeke cheat on his Logic exam?
As it stands, our model of conversation permits us only a direct answer response – that is, a response of the form:
(12) Zeke cheated on his Logic exam because....
But of course, we might not want to give any such response; we may instead simply reply:
(13) Zeke did not cheat on his Logic exam
Sentence (13) does not constitute a direct answer to (Q3), but rather a rejection of the entire question. Likewise, if a speaker commands us to drive below the speed limit (C1) in a situation where there is no speed limit – say, on a highway in Germany – then neither compliance sentence (saying we will, or will not, drive below the speed limit) is an appropriate response to the command. Instead, we would reject the command:
(14) There is no (unique) speed limit.
In both cases – questions and commands – such rejections seem like natural responses, the kinds of responses we give all the time. So our model of conversation ought to include some account of what such a question- or command-rejection amounts to. We don't yet have any definition of a rejection; but in light of the previous discussion of question and command presuppositions, it turns out that we already have all the parts we need for such a definition. Consider again the question (Q3), its presupposition (15), and its rejection (14):
(Q3) Why did Zeke cheat on his Logic exam?
(15) Zeke cheated on his Logic exam
(14) Zeke did not cheat on his Logic exam?
It is clear what the rejection (14) of the question amounts to: it is simply the denial (the negation) of the presupposition of the question. Likewise, in the case of the command (C1), its rejection (14) is simply the negation of its p resupposition (11):
(C1) Drive below the speed limit!
(11) There is a (unique) speed limit.
(14) There is no (unique) speed limit.
So using the notion of presupposition (which we've already defined), it's easy to define the rejection of a question or command:

Rejection of a Question or Command: The rejection of a question or command is the negation of a presupposition of that question or command.

Of course, the very fact that we would, on occasion, wish to reject a question or command only highlights once more the communicative work that presuppositions do.  If a non-declarative sentence S has a presupposition, then when S is uttered it is not, of course, itself going to be accepted in the way a declarative sentence would be – S does not go into the conversational background, because it makes no sense to “accept a question as true”. But even so, when that question is uttered, something may make its way into the conversational background all the same: S's presupposition.  The ability to reject a question or command is important for precisely this reason: if we couldn't reject the question – that is, if we couldn't negate its presupposition – then the presupposition would automatically be accepted in to the conversational background whenever the question was uttered.  Suppose, for instance, a parent asks a child:
(Q4) Where did you hide the cookies you stole?
The question, with its definite description, presupposes (16):
(16) There is a (unique bunch of) cookies you stole.
If the child couldn't reject the question, he would be left only with the option of citing some place where the stolen cookies are hidden (or else, of remaining silent and not responding at all). So even though questions and commands do not themselves assert anything (as being true), through their presuppositions they have the power to indirectly get sentences into the conversational background, and so to “frame” the conversation in significant ways. Rejection provides a communicative tool for counteracting this sometimes dangerous power of questions and commands.




beakley > 1900 > pragmatics
 
previous: presuppositions


next: more on questions: “how?" and “why?"