Linguistic Presupposition Research Paper

Thomas regrets that Adam ate the apple.                                                                                           (1)

The information that Adam ate the apple is called a presupposition of (1). This presupposition is said to be linguistically induced by the use of the factive verb ‘regret.’

This way of introducing the notion of presupposition calls for a distinction. And the basic distinction that should be made is the distinction between presupposition as information taken for granted and presupposition as induced, invoked, or triggered by a linguistic expression. The first notion is the notion of presupposition as background information, that is information which is already given or taken for granted in a conversation. Stalnaker and Karttunen call it the common ground. The second notion is the notion of presupposition as information that is conventionally associated with linguistic expressions or syntactic constructions. It is found in the literature under a variety of names, potential or pre-supposition (Gazdar 1979), conventional implicature (Karttunen and Peters 1979), or elementary presuppositions (Van der Sandt 1988). Though the two notions of presupposition are fundamentally different there is a straightforward connection. If a linguistic element induces a presupposition, the sentence containing the inducing element will typically be appropriate in a context which already contains the presuppositional information. Thus (1) will normally be judged appropriate in a context in which the information that Adam ate the apple is somehow established or taken for granted. If not, one should rather convey this information in an assertoric way as in (2).

Adam ate the apple and Thomas regrets that.                                                                                    (2)

1. Linguistic Encoding Of Presuppositions

Linguistic presuppositions can thus be seen as pieces of information triggered or induced by certain linguistic structures. Such triggers can be identified by using specific tests. These tests rely on the fact that presuppositional information tends to escape from environments which block ordinary entailments. The traditional test is constancy under negation. Other tests are embedding under epistemic modals, embedding in the antecedent of a conditional construction and questioning. Of these the negation test is the least reliable since negative sentences can in many cases be interpreted in a metalinguistic way so as to defeat the presuppositional inference.

Applying these tests to the verb stop identifies it as a presupposition inducer.

Hubert stopped smoking (at t).                                                                                               (3)

Hubert used to smoke (during some period preceding t).                                                                  (4a)

Hubert did not smoke (after t).                                                                                                           (4b)

Hubert didn’t stop smoking.                                                                                                               (5a)

It is possible that Hubert stopped smoking.                                                                                        (5b)

If Hubert stopped smoking, he must be saving a lot of money.                                                         (5c)

Did Hubert stop smoking?                                                                                                                  (5d)

The presuppositional content of (3) is given in (4a), the non-presuppositional content in (4b). Only the presuppositional information is preserved as an inference from the embeddings in (5a) through (5d).

The following linguistic structures are widely recognized as presupposition inducers. The presuppositional information that is triggered by the relevant items is given in the b-sentences.

(a) Definite NPs (definite descriptions often including proper names and indexicals; they induce a presupposition to the effect that there is some entity satisfying the description)

The King of France is bald.                                                                                                                (6a)

There is a King of France                                                                                                                   (6b)

(b) Cleft constructions (including ‘that-’ and pseudo-clefts; they trigger an existential presupposition that there is some entity satisfying the VP)

It was Adam who ate the apple                                                                                                                      (7a)

Someone ate the apple.                                                                                                                       (7b)

(c) Focal stress (induces the same existential presupposition as clefts, though many authors want to account for this inference by an independent theory of focus)

ADAM ate the apple                                                                                                                           (8a)

Someone ate the apple.                                                                                                                       (8b)

(d) ‘Wh’-questions (the presuppositional material is obtained by existential quantification over the variable introduced by the ‘wh’-expression)

Who ate the apple?                                                                                                                             (9a)

Someone ate the apple.                                                                                                                       (9b)

(e) Quantifiers (the strong quantifiers ‘all,’ ‘most,’… ; weak quantifiers like ‘some’ are considered to be ambiguous between a presuppositional and a non-presuppositional reading; the presupposition induced is that the domain of quantification is nonempty

All of John’s children are bald.                                                                                                          (10a)

John has children.                                                                                                                               (10b)

(f) Factive verbs (‘realize,’ ‘discover,’ ‘regret,’… ; they induce a presupposition that the presuppositional complement is true)

Thomas regrets that Adam ate the apple.                                                                                           (11a)

Adam ate the apple.                                                                                                                            (11b)

(g) Temporal clauses (‘before,’ ‘while,’ ‘since,’ ‘after,’… ; as in (f) they induce information of a propositional nature.

After Adam ate the apple, he was expelled from Paradise.                                                                (12a)

Adam ate the apple.                                                                                                                            (12b)

(h) Sortal restrictions on predicates (‘bachelor,’ ‘close,’… ; the presuppositional information consists of the requirement that the arguments of the predicate are of the appropriate sort)

John is a bachelor.                                                                                                                              (13a)

John is male, adult,… .                                                                                                                       (13b)

(i) Implicative verbs (‘manage,’ ‘succeed,’… ; put a presuppositional requirement on the action denoted by the VP)

John managed to open the door.                                                                                                         (14a)

It took some effort to open the door.                                                                                                  (14b)

(j) Verbs of transition (‘start,’ ‘stop,’ ‘continue,’… ; see (3) above)

(a) through (d) are examples of existential presuppositions. They all involve an existential quantification over some individualor set-variable. In (f ) and (g) the presuppositional information is of a propositional nature. (h) through ( j) involve requirements on the applicability of a predicate. This is only a small sample of the linguistic structures that have been identified as presupposition inducing. Further examples include focus particles (‘only’ and ‘even,’ adverbs of repetition (‘again,’ ‘too,’ ‘back’), but also, more controversially, counterfactual conditionals, non-restrictive relative clauses, and discourse particles like ‘because’ and ‘although.’

2. Projection

The presuppositions induced by simple sentences tend to survive under embedding, that is whenever a subsentence of a compound φ induces the presuppositional information χ this information will surface as an inference of φ. Karttunen (1973) was the first to note and catalogue numerous exceptions to this generalization. Here are some examples. In the a-sentences the presupposition that John has children escapes from its embedded position, in the b-sentences the same presupposition does not surface as an intuitive inference.

If baldness is hereditary, John’s children are bald.                                                                            (15a)

If John has children, John’s children are bald.                                                                                   (15b)

Either baldness is not hereditary, or John’s children are bald.                                                          (16a)

Either John has no children or John’s children are bald.                                                                   (16b)

It is possible that baldness is hereditary and that John’s children are bald.                                       (17a)

It is possible that John has children and that John’s children are bald.                                             (17b)

In the terminology prevalent in the 1970s and 1980s it is said that, in the a-sentences, the presupposition that John has children is projected to the main sentence, whereas it is said to be filtered or cancelled in the corresponding b-sentences.

3. Semantic, Pragmatic, And Discourse Presupposition

We may distinguish between three main types of theories that deal with presupposition and projection: semantic, pragmatic, and discourse theories.

3.1 Semantic Theories

The semantic view originates in Frege and was revived by Strawson. It construes presupposition as a binary relation between sentences (or propositions). A presupposition is a requirement that has to be fulfilled for a sentence to have a truth-value. Thus ψ is taken to be a presupposition of just in case ψ has to be true in order that has a truth value. This requires a semantics that allows truth-value gaps. Assuming the standard definition of negation as an operator which maps truth onto falsity and vice versa, semantic presupposition can then be defined as follows:

φ semantically presupposes χ if φ entails χ and not- φ entails χ.

The defining characteristic is constancy under negation. The projection behavior of presuppositions in other embeddings follows from the definitions of the logical connectives and other operators. One problem for such theories is that they cannot handle the presuppositions triggered by adverbs like ‘too’ and ‘even.’ Here presupposition failure arguably does not affect the truth-conditions of the sentence, but the appropriateness of the utterance. It also turns out that no definition of the logical connectives can be given which accounts for all the projection data. Defenders of the semantic account are thus forced to either postulate an ambiguity in the logical connectives or transfer the burden of explanation to pragmatics.

3.2 Pragmatic Accounts

Pragmatic accounts view presupposition not as a semantic relation between sentences. Instead presupposition is viewed as a property of utterances. The basic notion is that of a speaker presupposing a proposition in a context of utterance. This allows for certain variability of presuppositions with the context of utterance. The first formally developed theory of this kind is found in Gazdar (1979). The basic idea here is that a presupposition of a sentence is preserved unless it conflicts with contextual information or with an implicature invoked by its utterance. In the case of (15) through (17) this amounts to the following. A typical utterance of one of the b-sentences will invoke the implicature that it is possible that John has no children. This implicature conflicts with the presuppositional assumption that he does and so the latter is cancelled. Since no such implicature is invoked for the asentences, they will be interpreted as presupposition preserving.

3.3 Dynamic Theories

A different style of modeling is found in the dynamic approaches. The two main variants are the so-called satisfaction theory and the anaphoric or binding theory. The first derives from Stalnaker (1974) and Karttunen (1974) and has been developed in particular by Heim (1983) and Beaver. The anaphoric account is found in van der Sandt (1992) and Geurts (1999). Comparisons and different assessments of their relative merits are given in Beaver (1997) and Geurts (1999).

Both accounts start off from the idea that an utterance changes the context in which it is made. The basic requirement of the satisfaction account is that the presuppositional information should be entailed by the context of utterance. If so, the input context is incremented with the information contained in the inducing sentence. An utterance of (1) thus requires a context which entails that Adam ate the apple. If this requirement is not met, the input context is not a suitable one. In this case a mechanism of accommodation is invoked. This means that the input context will (under appropriate conditions) be revised so as to incorporate the relevant presuppositional information thus restoring interpretability.

The anaphoric or binding account is formulated as an extension of discourse representation theory. The basic tenet is that presuppositions are anaphoric expressions. Presuppositions only differ from pronouns and other attenuated anaphors in that they have internal structure and semantic content. The basic requirement is not that the presuppositional information is entailed by the context of utterance, but that the presuppositional anaphor should find a suitable antecedent. If no such antecedent is available, the semantic material carried by the presuppositional anaphor will be accommodated in the discourse structure so as to provide an antecedent after all.

The context-change semantics of the satisfaction theory generally yields weaker predictions than are found in the anaphoric account. This is due to the different status that presuppositional material has in both theories and to the different ways in which this material is processed. For example, for a conditional of the form φ→χ_<ψ> where ψ is triggered by χ the satisfaction theory will predict φ→ψ as presupposition for the whole sentence. This yields for (15a) not the presupposition that John has children, but the presupposition that he has children provided that baldness is hereditary. Just like the classic semantic account, the satisfaction theory has thus to appeal to some pragmatic strengthening mechanism to derive the stronger presupposition that John has children. On the anaphoric account, presuppositional information is never transformed into a new semantic construct. Consider (15a) again. The presuppositional anaphor associated with the consequent cannot be bound to some preestablished antecedent. So some entity representing the set of John’s children will be accommodated in the main context, thus yielding the stronger prediction that the utterance of (15a) presupposes that John has children.

Bibliography:

1. Beaver D I 1997 Presupposition. In: Benthem J A van, Meulen A ter (eds.) Handbook of Logic and Language. Elsevier, Amsterdam, pp. 939–1008
2. Gazdar G 1979 Pragmatics: Implicature, Presupposition, and Logical Form. Academic Press, New York
3. Geurts B 1999 Presuppositions and Pronouns. Elsevier, Oxford, UK
4. Heim I 1983 On the projection problem for presuppositions. In: Barlow M, Flickinger D P, Wescoat M T (eds.) Proceedings of the West Coast Conference on Formal Linguistics Volume 2. Stanford Linguistics Association, Stanford, CA, pp. 114–25 [Reprinted in Davis S (ed.) Pragmatics. Oxford University Press, Oxford, UK, pp. 397–405]
5. Karttunen L 1973 Presuppositions of compound sentences. Linguistic Inquiry 4: 167–93
6. Karttunen L 1974 Presupposition and linguistic context. Theoretical Linguistics 1: 181–94
7. Karttunen L, Peters S 1979 Conventional implicature. In: Oh C K, Dineen D (eds.) Presupposition. Academic Press, New York pp. 363–85
8. Sandt R V van der 1988 Context and Presupposition. Croom Helm Routledge, London
9. Sandt R A van der 1992 Presupposition projection as anaphora resolution. Journal of Semantics 9: 333–77
10. Stalnaker R 1974 Pragmatic presuppositions. In: Munitz M K, Unger P K (eds.) Semantics and Philosophy. New York University Press, New York, pp. 197–214