This is a technical term in philosophy of language and pragmatics for an extra
level of meaning in addition to the proposition expressed by an utterance and its implicatures. Use of the notion of presupposition is intended to explicate the pre-theoretical intuition that certain sentences or utterances take something for granted. Here is a well-known example: The king of France is bald. Presupposition 149 In the terminology of presupposition, this has been argued to presuppose that there is a king of France, and, assuming that the presupposition is satisfied, assert that he is bald. In general, definite descriptions (here The king of France) are said to give rise to a presupposition of the existence of the individual described. In the now-standard terminology, linguistic items or constructions that give rise to presuppositions are called presupposition triggers. The following are often associated with presupposition: factive verbs, such as know, and discover, clefting and pseudo-clefting constructions, and whquestions (see the separate entry on presupposition triggers for more). In these examples, the apparent presuppositions are given in italics: Factive: John knows there is life on Mars. (There is life on Mars) Clefting: It is John who arrived. (Someone arrived.) Wh-question: Who called? (Someone called.) What is taken for granted in making an utterance may not be satisfied. The point of the example, The king of France is bald, is that here it is not: there is no king of France. There are two views about what happens when a presupposition is not satisfied. On one view, the semantic theory of presuppositions, a statement with an unsatisfied presupposition has no truth value, that is, it is neither true nor false. This was the view taken by the philosopher Peter Strawson, who reintroduced the study of presupposition. (Reintroduced because in the late nineteenth century, the philosophers Sigwart and Frege had been concerned with the question, and there was a similar distinction in medieval philosophy between what an expression presupposed and what it denoted.) Strawsons views on presupposition have been influential in semantics and philosophy and they led to a great deal of work on systems of logic with a truth-gap or three truth values. However, in linguistics the prevailing view since the 1970s has been that presuppositions are pragmatic phenomena. On this view, introduced by Robert Stalnaker, there are two possibilities when there is an utterance with a presupposition that is unsatisfied. Either the utterance is infelicitous, or (and this is more commonly the case) the presupposition is accommodated, which is to say that the information that is presupposed by the utterance but which was not previously part of the common ground is added to the common ground. In other words, a speaker can express certain information as though it were taken for granted, and if the strategy works, this information is from then on treated as known. As Grice 150 Presupposition had previously said, if a speaker says, My aunts cousin went to that concert, the hearer would be expected to take it from the speaker that she did indeed have an aunt and that her aunt had a cousin. A striking property of presuppositions is that they appear to be preserved under negation, in contrast to entailments of the proposition expressed and most implicatures. For example (on one reading) The king of France isnt bald also takes for granted the existence of a king of France and John doesnt know that there is life on Mars also takes for granted that there is life on Mars. This property is sometimes called constancy under negation and it has been taken as the standard test for presuppositions. However this test is difficult to apply to some sentences and does not always produce clear intuitions. For example, what, if anything, does Who did not call? presuppose? What is more, not every speaker who utters one of these negated sentences presupposes what they are supposed normally to presuppose. For example, hearing an argument about whether the king of France is bald, it is perfectly felicitous to say: The king of France isnt bald. There is no king of France. In such cases, either there was no presupposition, or it was cancelled. Many theorists think that presuppositions are cancellable, as implicatures are. On this view, cancellation may be explicit, as above, but it can also be a result of incompatibility of the potential presupposition with world knowledge. For example, before is said to trigger presuppositions, as in: John learned to type before finishing his book. (John finished his book) But the following has no such presupposition: John died before finishing his book. The idea is that we know that dead people do not finish books, so the presupposition is cancelled by its incompatibility with world knowledge. Returning to constancy under negation, it is one example of what is called the projection problem or the presupposition projection problem, a topic that has had a great deal of attention. (The term projection problem and the framing of the issue in these terms are due to joint work by the linguists D. Terence Langendoen and Harris Savin.) Embedding a phrase or a sentence which usually carries a presupposition inside a larger sentence sometimes does and sometimes does not result in a sentence with the same presupposition. We have seen Presupposition 151 that embedding under negation can preserve presuppositions. A common way of expressing this is to say that negation is one example of a hole: a linguistic expression or construction that, as it were, lets presupposition through. As well as holes there are argued to be plugs, expressions which block presuppositions entirely, and filters, expressions which let only some presuppositions through. Verbs of reporting and saying are said to be plugs. For example, an utterance of Mary said that the king of France is bald does not require the hearer to take for granted the existence of the king of France. Logical connectives such as conjunction (and) and implication (if . . . then . . .) are said to be filters, expressions that sometimes let presuppositions through but not always. The simplest filtering cases are ones in which the first conjunct of a compound sentence asserts what the second part would normally presuppose. For example: John stole the Mona Lisa and he regrets that he stole the Mona Lisa. Uttering the whole sentence does not ask the hearer to take for granted that John stole the Mona Lisa, since that is asserted in the first conjunct, before the hearer gets to the part that is said normally to presuppose it. Several different elaborate theories which try to account for these data have been formulated. The best known are due to the linguists Lauri Karttunen (working on his own and with Stanley Peters), Gerald Gazdar and Irene Heim. Almost all of this work on the projection problem takes for granted Stalnakers view that presupposition is a pragmatic phenomenon, a property of utterances rather than sentences, strictly speaking, and that presupposition failure leads to accommodation or infelicity, rather than lack of truth value. However these theories all look to semantics and syntax for explanations for these pragmatic facts. The ability to trigger presuppositions is seen as a property of lexical items or constructions, and the projection (or otherwise) of presuppositions is explained in terms of the larger sentences components and structure. An alternative view has been maintained by many pragmatic theorists, including Paul Grice, Deirdre Wilson, Ruth Kempson and Stephen Neale. On this view, the pre-theoretic intuitions that certain utterances ask the hearer to take something for granted are best explained without appealing to a separate presuppositional level of meaning. Theories which make presuppositions depend on lexical items and sentence structure are said to be writing the 152 Presupposition trigger properties of a speech act, assertion, into the lexicon. This is undesirable on general grounds and because it fails to predict that non-assertive utterances of sentences (e.g. as a guess, a request or a question) do not generally have the presuppositions that assertive uses would. For example, a guess that Martians criticize each other for looping the loop does not presuppose that they loop the loop rather it guesses that they do. In the case of definite descriptions, the non-presuppositional analysis goes back to Bertrand Russell. On a Russellian view of definite descriptions, the semantics of The king of France is bald amount to: There is a unique king of France and he is bald. Usually, but not in all circumstances, the main point of asserting that the king of France is bald would be to comment on his lack of hair, and the speaker would indeed be taking for granted that the hearer knows, or would accept, that there is a king of France. But according to the pragmatic view these are simply facts about typical contexts and how the expression is often used and do not justify the setting up of a separate level of meaning. This analysis has a natural explanation for the fact that the existence presupposition of definites sometimes does and sometimes does not survive negation. On this view, this is a matter of scope. Depending on the scope taken by negation, It is not the case that the king of France is bald, may mean that there is such an individual and he is not bald, or that there is no such individual. Presuppositional analyses usually explain these facts with an ambiguity theory of negation, distinguishing between internal and external negation, or between standard and metalinguistic negation.