L. J. MARTIN REBOLLO
II. REVISTA DE REVISTAS (*)
Grice (1981: 185) briefly characterizes the speaker’s implicatum as the content that “would be what he might expect the hearer to suppose him to think in order to preserve the idea that the [conversational] maxims are, after all, not being violated.” The neo-Gricean explanation (Atlas 1975, Atlas and Levinson 1981) of referential, factive, and cleft presuppositions depended on the hearer supposing the speaker not to be violating Atlas and Levinson’s (1981: 40) Neo-Gricean Maxims of Relativity, which were refinements of Grice’s Maxims of Quantity.
(4) Maxims of Relativity
1. Do not say what you believe to be highly noncontroversial, that is, to be entailed by the presumptions of the common ground.
2. Take what you hear to be lowly noncontroversial, that is, consistent with the presumptions of the common ground.
The first maxim is a speaker-oriented production maxim; the second maxim is a hearer-oriented comprehension maxim. It is important that the production maxim is a prohibition, a “do not” maxim, and that the comprehension maxim is an obligation, a “must do” maxim. It is also important to note the difference between a sentence being entailed by a set of sentences in the common ground and a sentence being merely logically consistent with a set of sentences in the common ground. The consistency requirement was designed to permit the kind of informative statement accommodation for referential and factive presupposi-tions that I described in Atlas (1975, 1977a). If a singular term were introduced by its use in a statement that would be more informative under an interpreta-tion requiring the singular term to be a referring term in that statement, and its having a reference was consistent with the previously established common ground, nothing in my maxim would stand in the way of such an informative interpretation, whether or not the existence of the reference of the singular term had already been established as part of the common ground. Atlas and Levinson’s (1981) Maxims of Relativity were designed to accommodate accommodation.
It should also be noted that our Maxims of Relativity were couched in terms of non-controversiality and of common ground, constrained by a mini-theory of Non-controversiality. Among the axioms of that mini-theory were (Atlas and Levinson 1981: 40):
42 Jay David Atlas
(5) Axioms of Non-controversiality
a. If A(t) is “about” t, i.e. if 〈t〉 is a Topic NP in the statement A(t), then if 〈t〉 is a singular term, the proposition 〈t exists〉 is non-controversial.
b. The obtaining of stereotypical relations among individuals is noncontroversial.
Embedded in our neo-Gricean account of referential and cleft “presuppositions”
were the distinct notions Topic NP, non-controversial proposition, common ground, and most informative interpretation of a statement, the interpreta-tion of which is consistent with the proposiinterpreta-tions of the common ground.
What I showed in Atlas (1975) and Atlas and Levinson (1981) and have reviewed above was the reasoning by which the conversational implicatum There is a king of France could be reached from the default understanding of an assertion of The king of France is not bald (see (5a) above). The construction of reasoning to a default interpretation is required if Grice’s third criterion for the existence of a conversational implicature is to be met, and the reasoning I constructed depended upon an elaboration and a revision of Grice’s Maxims of Quantity (being as informative as is required).
Grice (1981: 189) himself adopts a Russellian analysis of The king of France is bald, a conjunction of three independent clauses (cf. Strawson 1950: 5):
(6) The king of France is bald.
(A) There is at least one king of France.
(B) There is not more than one king of France.
(C) There is nothing which is a king of France and is not bald.
The account of presupposition that Grice (1981: 190) gave of the presupposi-tion of the negative statement The king of France is not bald depends upon a distinction between denied and undenied conjuncts:
it would be reasonable to suppose that the speaker thinks, and expects his hearer to think, that some subconjunction of A and B and C has what I might call common-ground status and, therefore, is not something that is likely to be chal-lenged. One way in which this might happen would be if the speaker were to think or assume that it is common knowledge, and that people would regard it as common knowledge, that there is one and only one [king of France].
Thus the speaker who asserts The king of France is not bald would be under-stood to deny only the third conjunct (C) {Nothing that is a king of France is not bald, Whatever is a king of France is bald}, since the argument just quoted was supposed to “show that, in some way, one particular conjunct is singled out”
(Grice 1981: 190). Of course, if one takes it as common knowledge that there is a unique king of France, and then denies conjunct (C), as Grice proposes, one gets the conjunction (i) There is a unique king of France and there is at least one king of France that is not bald, which is supposed to be an interpretation of
Presupposition 43 (ii) The king of France is not bald. The (weak) denial, viz. the interpretation of It’s not the case that the king of France is bald as 〈¬ ((A and B) and C)〉, conjoined with the common ground 〈(A and B)〉, is supposed to give 〈((A and B) and
¬C)〉 as the interpretation of The king of France is not bald. But the question is, WHY should the common ground intervene in utterance interpretation in this way? Grice believes that its commongroundedness is a sufficient and obvious explanation; I do not. A theory of how and why common ground enters into utterance interpretation is needed. Grice does not offer one; Atlas and Levinson (1981) do.
Sentence (i) entails There is a unique king of France, and an utterance of (ii), on Grice’s (1981: 189) own showing, “without waiting for disambiguation,”
implies “(in some fashion) the unique existence of a king of France.” Grice (1981: 189) has already remarked that “what needs to be shown is a route by which the weaker reading could come to implicate what it does not entail.”
But what Grice (1981: 190) has just shown is how the (weak) denial of The king of France is bald, in conjunction with the common ground, entails There is a unique king of France, which is what he has explicitly claimed a speaker implic-ates by the (weak) denial of The king of France is bald and that the (weak) denial does not entail without the common ground.
Does this mean that Grice reduces implicature to a context-relative entail-ment? If he were to do so, he would start to sound like a Sperber and Wilson (1986a) RELEVANCE theorist. But a common-ground-dependent entailment from the (weak) denial is merely a fact about context and the assertoric content of a weak negation. It is not a theory of an inference to the best interpretation of the negative utterance in the fashion of Atlas and Levinson (1981). Unlike me, Grice thinks the negative sentence already has an interpretation; it is the weak negation. (Levinson was classically Gricean about it in Atlas and Levinson (1981), though he (2000a) now takes a more favorable view of semantical non-specificity; see Atlas and Levinson 1981: 55, n.19). The question for Grice is, why should that interpretation generate the “implication” that there is a unique king of France? Pace Grice (1981: 190), the answer cannot be that the existential proposition is ALREADY ASSUMED in the context. The existence of a context-relative entailment is no explanation of “the route by which the weaker reading could come to implicate what it does not entail,” since context-relative entail-ment does not possess the logical properties of implicature or presupposition.
Even though the context provides premises, the relation is an entailment; it is monotonic, unlike implicature (which is non-monotonic, i.e. defeasible), and it is unlike presupposition (which is preserved under main-verb negation). Unlike an entailment an implicatum can also be cancelled (i.e. negated without a resulting contradiction).
The sources of these difficulties in Grice’s (1981) analysis of The king of France is not bald are clear. The first source is his accepting the conjunction of three independent propositions as an analysis of The king of France is bald (for reasons that I have not discussed here) and his concomitant commitment to the scope ambiguity of The king of France is not bald. The ambiguity assumption
44 Jay David Atlas
leads Grice into an incoherent account in his attempt to derive the referential
“presupposition” from his (weak) external negation reading of the negative sentence. It is here that ignoring the subtle difference between ambiguity and semantically non-specific univocality has devastating consequences for the success of Grice’s (1981) attempt to reduce referential presupposition to implic-ature. And it is here that the Atlas’s (1974, 1975, 1979) and Atlas and Levinson’s (1981) analyses succeed where Grice’s (1981) fails.
The second source is his appeal to common ground in the simple way he appeals to it, and the way that Stalnaker (1974) appeals to it in “Pragmatic Presuppositions.” An explanation using speaker’s presupposition is not equival-ent to an account using conversational implicatures.