Implicature: Intention, Convention and Principle
in the Failure of Gricean Theory
Wayne A. Davis
Cambridge Studies in Philosophy
(Cambridge: Cambridge University Press), 2007
£22.99. 216 pp. Paperback. ISBN-13: 978-0521038065
Reviewed by Graham Ranger
At first sight, the main title of Wayne A. Davis' book, Implicature, might lead one to believe that this was a student's textbook account of the Gricean phenomenon. However, the smaller subtitle, Intention, Convention and Principle in the Failure of Gricean Theory, gives a much more exact idea of the tone of the work which is a sustained and multidimensional critique of Grice's renowned theory of implicature. This is made quite explicit in Davis' short sharp introduction: ‘I argue that the Gricean theory has been barren […] The illusion of understanding provided by the Gricean theory has only served to stifle inquiry. […] I raise many fascinating questions about implicature, requiring systematic historical and sociolinguistic research for their solution, which did not and could not arise when the Gricean theory held sway’ . In the following review I will present a chapter by chapter account of Implicature before providing a critical assessment the work as a whole.
Chapter 1, ‘Concept and theory’, presents the concept of implicature and the theory which Davis will take issue with in chapters 2-6. Implicature is defined as ‘indirect speaker meaning: meaning one thing by meaning another’ . Davis takes Grice's classic petrol-station example, which we will return to, by way of illustration:
Ann : Where can I get gasoline?
Bob : There's a station around the corner.
Davis recognizes the existence of implicatures in such examples and the important of the associated concept. He considers, however, that ‘the theoretical importance of implicature is independent of the Gricean theory of implicature’ . Gricean theory is presented, following Grice 1975, in terms of a Cooperative Principle (‘contribute what is required by the accepted purpose of the conversation’ ) and the four conversational Maxims of Quality (‘Make your contribution true; so do not convey what you believe false on unjustified’ ), Quantity (‘Be as informative as required’ ), Relation (‘Be relevant’ ) and Manner (‘Be perspicuous; so avoid obscurity and ambiguity, and strive for brevity and order’ ). Implicatures, pace Grice, are generated by hearers assuming that one or other of the maxims is being respected andgenerating extra meaning, so to speak, in virtue of this.
Hence Bob's answer,above, if it is to be relevant (Maxim of Relation), must have as its implicature ‘You can get gasoline around the corner’. Implicatures may, on the other hand, depend not on respect of the Maxims, but on a deliberate flouting of maxims, as in the case of irony, where the Maxim of Quality is ostentatiously ignored. In addition to these aspects, Gricean theory holds that it must be possible to use the cooperative principle and the maxims to generate implicatures in a way amenable to calculation. These features are referred to by Davis as the ‘generative assumption’  and the ‘calculability assumption’ . The last tenet is described as ‘Grice's razor’ , in reference to Ockham's razor, by which it is reputedly better to postulate conversational implicatures for sentences rather than separate meanings since the former can be derived ‘from independently motivated psychosocial principles’ .
One of Davis' main arguments against Gricean theory, which he puts forth in the first chapter, is that it gets things the wrong way round, in a form of post hoc reasoning. And so, he argues, it is not because speakers obey maxims that they use implicatures, but rather it is because they use implicatures that we can, in an explicative movement [how rather than why], say that they have obeyed certain maxims. These considerations lead Davis to introduce the concept of speaker intention, which he will return to at the end of the book. He claims that a speaker uses a sentence Σ to produce an implicature p not because they are obeying the maxims but because they intend to observe the Cooperative Principle. One last aspect of Gricean theory presented in Chapter 1 is what Davis calls the Sufficiency Implication: the notion that the previously mentioned aspects of the theory are not only necessary but also sufficient to explain the existence of implicatures.
Chapter 2, ‘Differentiation’ looks at false positives, that is, implicatures which an application of the theory predicts but which Davis contests. He gives four examples of the problem. The first relates to quantity implicatures of the type allowing us, from Some died to derive Not all died. In typical Gricean approaches, such implicatures are explained by considering that, since the speaker S is assumed to be as informative as required [by the maxim of quantity], had he known that all had died, he would have said so. Hence some died implies not all died. Davis, in disputing this form of derivation, gives a battery of examples in which, he claims, a weaker statement (like some…) does not imply the denial of an associated stronger statement (like all…). Here are some of his counter-examples, approximately as he presents them:
Some died does not implicate that [Only some died] is not the case
Some died does not implicate that [Some were killed] is not the case
Some died does not implicate that [35.72% died] is not the case
John may win does not implicate that [John will win] is not the case
John is in California does not implicate that [John is in southern California] is not the case
John lost a book does not implicate that [John lost his book] is not the case etc. 
In Davis' discussion of these examples, as elsewhere, I confess I ran into a frequent problem I personally find with texts from the philosophy of language which combine considerable philosophical sophistication with what seems to me a remarkable linguistic naivety. The opposition between weaker and stronger is relevant in scalar terms, when dealing with implicatures of the some / not all type, but Davis here appears to construe it in terms of more or less precise or more or less determinate. On the following pages [37-38], Davis rightly takes issue with Horn's discussion (1972) of the opposition between John could have solved the problem and John was able to solve the problem but rather than locate the differences in interpretation in aspectuo-modal properties, Davis surprisingly declares the two utterances nearly identical. We shall return to this aspect of Implicature in our conclusion.
He next goes on to look at Grice's discussion of tautology implicatures of the War is war variety. These are typically explained by saying that the maxim of quantity is flouted provoking a movement of interpretative enrichment. Davis quotes Wierzbicka's arguments according to which the Gricean derivation of tautology implicatures are ethnocentric illusions, arguing additionally that most tautologies [examples he gives are The red car is red or If it rains then it will rain or snow] actually lack implicatures. He concludes that this type of implicature is entirely conventional.
Conjunction implicatures are those which allow us, from John took off his trousers and went to bed to infer that he did so in that order. These are generally explained by way of the Maxim of Manner: Be orderly, by which one presents things in the order of occurrence. Davis considers again that this is unwarranted. He argues that John set a record and cleared 15 feet inverts the order of occurrence, that the same implicatures might be present without a conjunction [John took off his trousers; he went to bed] and that many conjoined sentences have no order to them whatever [John saw Bill and met either Charles or Andrew]. Again from a linguistic perspective, many of Davis's counter-examples, including John set a record… above, bother me considerably. He does, however, make the very serious objection that Grice's maxims often enter into conflict with each other and that, in the absence of any explicit hierarchy, ‘the theory yields no predictions, and thus as a wholeis untestable’ .
A related, and pertinent remark, concerns the imperativeGrice unfortunately uses for his maxims, and which is far too ambiguous for his purposes, concealing, as it does, a variety of applications. Hence be as informative as required might be interpreted in different ways, as Normative: (i.e.) People ought to…, Behavioral: People are…, Motivational: People intend to…, or Cognitive: People believe they ought to… . A short final section deals with idiomatic expressions such as go to the bathroom to mean urinate or defecate. These pose theoretical problems since they are in all probability conventional implicatures which began as conversational implicatures but have evolved over the course of time.
In Chapter 3, ‘Determinacy and Calculability’, Davis turns his attention to the twin requirements of Gricean theory, stated in Chapter 1, that implicatures should be calculable and indeed that this calculation determines the implicature. Much of the chapter is accordingly devoted to ‘false negatives’, i.e. implicatures which should not exist but do. Davis examines a series of different types of problems raised by determinacy and calculability. I will not consider them all here. The rhetorical figure problem, as he terms it, relates to ironic uses of sentences. Typically, these are explained as floutings of the Maxim of Quality, by considering that the hearer H knows that the speaker S thinks differently. As Davis points out, however, ‘the determinacy requirement will always fail in the case of irony and other figures of speech, because S COULD have been speaking literally, believing what is said, or using a different figure ofspeech’ .
The problem is in fact present for all purported floutings. Indeterminate implicatures are illustrated by sentences such as My love is a red rose  in which the implicature is notoriously hard to pin down. One can formulate the concomitant implicatures as a disjunction (I1 or I2 or In...) or as a conjunction (I1 and I2 and In...), but the first option seems too exclusive a solution and the second too inclusive. ‘Quantity implicatures: the possibility of ignorance’ evokes the problems of utterances like Some died in which S's lack of specification comes from ignorance. In this case the possibility that All died is not excluded and the Maxim of Quantity does not function. Another situation one might envisage is that where the utterance Some died is proffered on a ‘need to know basis’, S considering that the speakerdoes not need to know any more than some died.
This again imperils theprinciples of determinacy and calculability since ‘nothing […] rules out the possibility that he is obeying the Maxim of Quantity by dispensing information on a need to know basis’ . Interestingly it would appear from research by Keenan and Ochs that ‘Malagasy speakers often withhold information in culturally accepted ways we find alien’ . Davis returns briefly to tautology implicatures and conjunction implicatures before raising the larger and crucial problem of conflicting principles. The Maxim of Manner, in particular, often clashes with the others, since ‘it would always be more perspicuous to 'explicate' a proposition rather than implicate it’ . The Gricean principles additionally clash with the non-Gricean but nonetheless real principles of Style and Politeness. Davis considers possible modifications to the theory which might redeem it, in the form of work by Brown and Levinson (1987), Grice himself (1978) or, at more length, Wilson and Sperber's Relevance Theory (1986).
Wilson and Sperber try to avoid clashes, notably by subsuming Grice'sprinciples in an overarching Theory of Relevance, by which ‘[the] relevance of a proposition increases with the number of contextual implications it yields and decreases as the amount of processing needed to obtain them increases’ [Sperber and Wilson 1986: 382]. Davis is no more indulgent with Relevance Theory than with Gricean Theory, however, remarking pertinently that ‘no one in the world knows how to measure either quantity of information or processing cost with nearly enough precision to calculate the requisite ratios’ . He notes: ‘One who is antecedently convinced of the principle can always insist that because a certain proposition is implicated, it must have the maximal relevance ratio. But the possibility of such ex post facto reasoning does not confer any more predictive power on the Principle of Relevance than it would on the Cooperative Principle’ .
Chapter 4, ‘Presumption and mutual knowledge’, considers the part of Gricean Theory which supposes that the cooperative principle is operative for implicatures to be generated. Davis prefers to considers rather that this notion confuses inference and implicature, since ‘implicatures need not be recognized, and their recognition does not depend on any specialized reasoning process’ . To show this, he presents cases where implicatures are indeed present without any presumption that the Cooperative Principle obtains. The simplest case is that where Ann mistrusts Bob, in the petrol-station example given above, Bob's answer There's a station around the corner appears to produce the same implicature p whether Ann presumes that Bob believes that p or not. More generally, ‘[the] cooperative presumption condition may fail because the conversation has no accepted purpose, as when S is changing the subject, insulting his audience, engaging in idle chitchat, talking at cross-purposes, arguing with H, orexplaining why he is unwilling to give H the requested information’ .
Thismight sound like bad faith (and some of the arguments do tend in that direction). Davis' point however, is that if we reduce our expectations as to what constitutes cooperation in a conversation, then we risk making ‘the Cooperative Principle […] too weak to support the maxims’ . After considering a number of confusions between meaning and communication, and between inference and implicature, the author moves on to an argument which he will be defending later on in the work, namely, that ‘to properly interpret the speaker's utterance, we need evidence as to his intentions’  evidence which the context and the Conversational Principles may help us to infer, but which they do not necessarily determine. As he succinctly puts it, ‘conversational principles play an inessential, indirect, and nonunique role in figuring out what speakers are implicating’ .
In the first four chapters, Davis has provided exhaustive criticism of Gricean and related theories of implicature. In the last two chapters he turns to providing an alternative approach. His fundamental thesis in Chapter 5, ‘The existence of implicature conventions’, is that conversational implicatures depend upon conventions, defined as ‘arbitrary social custom or practice’ . Conventions perpetuate themselves, and Davis shows this with his highway code analogy. Americans drive on the right of the road, and this is an established, but arbitrary convention. Because of this first convention, steering wheels are placed on the left. This convention, although initially motivated, comes to perpetuate itself independently, since people are used to driving that way, factories are set up to install left-hand drives etc. In addition to precedence, conventions also depend upon association, tradition, habit and social pressure for their continued existence [134-135]. Davis argues that ‘implicature practices are conventional and arbitrary despite the fact that most scholars believe they are natural, being deducible from general psychosocial principles’ . He admits the existence of ‘regularities in implicating’ but considers these to be arbitrary .
With this in mind, he reexamines a number of already mentioned problems,including quantity implicatures, of the some-implicates-not-all type, which are claimed to work ‘because that particular sentence form has been used that way before, not because weaker statements are generally used to implicate the denial of stronger statements’ . The conjunction implicatures [i.e. p and q implicating p and then q] are also considered to be arbitrary, and Davis goes so far as to claim ‘We could have used 'e and f' to mean '—[e before f]', 'f before e,' or 'e because f' or we could have used it without any such implicatures’ . Such blatantly counter-intuitive affirmations, and there is no shorter of them in this relatively short chapter, obviously require further justification, and that is the purpose of the final, sixth chapter.
Chapter 6, ‘The nature of implicature conventions’, begins with the statement that ‘[i]mplicature practices are arbitrary to some extent rather than completely determinate’ . This seems to represent something of a comedown, after the more extreme affirmations of arbitrariness quoted above from the previous chapter. Davis begins his discussion of the nature of implicature conventions by establishing an opposition between first- and second-order semantic rules. ‘The first-order rules are conventions for using sentences to directly express certain thoughts. The second-order rules are conventions for indirect expression, rules for expressing further thoughts by expressing thoughts assigned to first-order rules’ . Implicature conventions are, Davis continues, ‘second order because they involve a regularity in indirect meaning’ .
Second-order semantic conventions can evolve to becomefirst-order, as when an indirect expression is no longer felt to be indirect [kick the bucket, go to the bathroom etc.]. Such conventions do not represent rules of a particular language or of language in general, but rather rules for the way a given language is used. If implicatures are considered to be conventional, following Davis, then we might well wonder where they originate. Davis argues that implicature conventions of the some-implicates-not-all type might realistically arise by an inferential reasoning processes without any appeal to Gricean maxims. By this reasoning process, ‘the function of implicature practices is to promote […] cooperative, efficient, polite, and stylish communication’ , the norms for which will vary from culture to culture. Implicatures appear adventitiously, Davis claims, and then may become conventional if they serve the purposes of cooperation, efficiency, politeness,and style .
The hierarchy of these different features must depend onspeaker intention. We have seen that Davis vigorously opposes the Gricean notion of implicature calculability but he does accept what he terms the principle of antecedent relation  by which, although a sentence does not possess an implicature in itself, speakers may find it natural for it to do so in a given situation: ‘On my view, S means or implies I by uttering Σ only if S utters Σ with the intention of providing an indication that he believes I, where an indication is a natural sign’ . In a final section the author considers the existence of similar implicature conventions in different languages. This does not, in his view, indicate universality. He speculates that ‘the most universal implicature conventions arose in the original language community and were sustained by their effectiveness even as the first-order semantic conventions slowly evolved into the thousands of languages found in the world today’ . The concluding paragraph of Implicature is worth quoting in full, as it admirably sums up the thrust of Davis' arguments:
Gricean theory fails because speaker implicature is a matter of intention, sentence implicature is a matter of convention, and neither is calculable from or generated by psychosocial principles. Conversational rules instead codify social goals motivating intentions and sustaining conventions .
I found Davis' book a stimulating read. It begins with a very ambitious remit, stated clearly in the Introduction, and carries out a highly critical, exhaustive reexamination of Gricean pragmatics, also taking into account a number of other related models, from the Anglo-American philosophical tradition, including Sperber and Wilson's Relevance Theory, Horn's Scales or Brown and Levinson's work on Politeness. The edition I am reviewing is an unrevised paperback edition of a work which first appeared in 1998, which might suggest that Implicature has since become required reading for the student of pragmatics and the philosophy of languagein general.
The bibliography is complete, within the Anglo-American tradition,and the Index a useful aid. A preface would be welcome and might additionally help contextualise the work in the light of contemporary pragmatics. The arguments against the calculability and determinacy of Gricean implicature are particularly compelling and one has the impression that the acceptance accorded more or less unquestioningly to the Gricean model enables Davis to argue his point all the more vehemently.
The book is not, however, easy of access. This is no doubt due in part to the subject matter. However, I feel that Davis' arguments would have gained in force had the work been organised in a significantly different manner. Over the different chapters, Davis returns to the same problems and examples, considering them from different perspectives. This means that Chapters 2, 3 and 5 all include sections entitled ‘Tautology implicatures’ or ‘Conjunction implicatures’, for example. The effect on the reader is particularly unsettling, and I, for one, would have preferred the author to have developed his argument thoroughly on the basis of one type of implicature before going on to look at other types, by way of confirmation.
Part of the problem is perhaps Davis' obvious desire to leave no stone unturnedin his indictment of the Gricean model, but this exhaustivity often leaves the reader wondering where he is being taken. I suspect also that the dominant idiom of the philosophy of language contributes to the difficulty of Implicature. Some of Davis' arguments appear, as I have mentioned above, to be in somewhat bad faith. He wonders, for instance, in the petrol-station situation, what would happen if Bob misspoke, saying coroner instead of corner, concluding that ‘the maxims would still seem to require that Bob believe there is a station around the corner, where Ann can get gasoline’ . The author would probably counter that the way in which adepts of Gricean pragmatics ignore patent counter-examples is no better!
While I fully agree with much of Davis' criticism I am prevented from following him too far by the surprising linguistic naivety of many of his arguments (again this is possibly part of the idiom of the philosophy of language which prefers to leave grammar with as little to do as possible, probably a legacy from the nineteenth century). For example, he states: ‘There seems to be nothing in the meaning difference between “could have solved” and “was able to solve” […] that makes it more natural for the former to implicate “did not solve” and the latter to implicate “did solve” than vice versa’ . This suggests to me a very blunt idea of ‘meaning’ indeed. Similarly, his remarks on the role of stress or reduplicative utterances of the War is war variety are often very disingenuous, while his suggestion that cross-language implicature conventions might derive back to a common ancestor (‘the original language community’ ) comes dangerously close to linguistic science-fiction, for me.
Despite these qualifications, Implicature remains a stimulating and ambitious book which criticises Gricean pragmatics in a cogent and challenging way.
Brown, P. & Levinson, S.C. Politeness: Some Universals in Language Use. Cambridge: Cambridge University Press, 1987.
Grice, H.P. ‘Logic and Conversation’. In Syntax and Semantics: 3 Speech Acts, P. Cole & J. Morgan eds., New York: Academic Press, 1975, 41-58.
Grice, H.P. ‘Further Notes on Logic and Conversation’. In Syntax and Semantics 9: Pragmatics, P. Cole, ed. New York: Academic Press, 1978, 113-128.
Horn, L.R. ‘On the Semantic Properties of Logical Operators in English’. Ph. D Thesis, UCLA, 1972.
Sperber, D. & Wilson, D. Relevance: Communication and Cognition. Cambridge, Mass: Harvard University Press, 1986.
Wierzbicka, A. ‘Different Cultures, Different Languages, Different Speech Acts.’ Journal of Pragmatics 15 (1985) 507-520.