grice paper final draft.doc

of 122/122
Rationality and Conversation: A Thesis on Grice’s Theory of Conversation Matthew Schoolfield

Post on 10-Feb-2017




0 download

Embed Size (px)


Grice paper

Rationality and Conversation:A Thesis on Grices Theory of Conversation

Matthew Schoolfield

MSc (By Research)

The University of Edinburgh


Table of Contents

1Chapter 1: Gricean Theory

11.1 Introduction

21.2 Grices Theory of Conversational Implicature

91.3 Epistemology and Testability of Gricean Theory

12Chapter 2: Criticisms, Alternatives and Neo-Gricean Theory

122.1 Kasher and Hintikka: Rationality as the Basis for Conversation

182.2 Pre-Eminent Schools of Thought

182.2.1 Neo-Griceans Martinich Levinson Leech

252.2.2 Relevance Theory

292.3 Davis: The Anti-Gricean

33Chapter 3: Conversational Goals

333.1 Explanatory Failures of Gricean Theory

343.2 Social Norms as Presumed Goals

353.3 The Communicative Goal

403.4 The Suasive Goal

403.4.1 Seller-Consumer Relationship

423.4.2 The Bargainer Relationship

443.5 The Epicurean Goal

453.6 Exclusivity, Exhaustiveness, and Conclusion

47Chapter 4: Implicature within Goal-Oriented Conversation

474.1 Quantity: Make Your Contribution as Informative as Is Required

474.1.1 Some, Not All

504.1.2 Tautologies

514.1.3 Other Violations

524.2 Quality:

524.2.1 One Should Make His or Her Contribution One That Will Not Express Something False

544.2.2 Do Not Express That for Which You Lack Adequate Evidence

544.3 Relation: Respond Relevantly

564.4 Manner:

564.4.1 Avoid Unnecessary (or Excessive) Obscurity of Expression

584.4.2 Avoid Ambiguity

594.4.3 Avoid Unnecessary Prolixity

614.4.4 Avoid Unnecessary Disorder

614.5 Conclusion

63Chapter 5: Conclusion

67Works Cited

Chapter 1: Gricean Theory

1.1 Introduction

Paul Grice presents a theory of conversation and implicatures in his essay Logic and Conversation. This work was first presented in his William James Lectures in early 1967 and was discussed at some length before being published in Cole and Morgans Syntax and Semantics, Vol. 3: Speech Acts in 1975 (Chapman 2005, 100). This theory attempts to bridge the gap between what participants in conversation say, and what they mean. Grice presents this theory within the framework of the Cooperative Principle: Make your conversational contributions such as is required at the stage at which it occurs, by the accepted purpose or direction of the talk exchange in which you are engaged (Grice 1989, 26). The theory enables Grice to present an account for the interpretation regarding discourse that would otherwise be considered irrational. This account, however, has drawn much criticism from both those who support this theory of conversation, and those who believe that it is inherently flawed. Neo-Griceans, such as Levinson, have presented updated models of conversational implicature, streamlining the theory in some ways and expanding on it in other ways. Revisionists, such as Sperber and Wilson, have created new theories based on other principles. Detractors to the Gricean account of conversation include Davis, Kasher, Hintikka, and an interesting sociological work by Haviland. Davis provides a cornucopia of counterexamples and suggests a theory of convention in its place. Kasher objects to the idea of the Cooperative Principle always being apparent in conversation; this is due to his rejection of the idea that participants in conversation always have a common goal. Hintikka suggests a game-theoretical approach to conversational implicature, but does not expand on this idea. Finally, others consider the theory to be either insignificant or susceptible to many counterexamples. Primarily, if the Cooperative Principle is not necessary for communication then it has much less force regarding conversation; thus, if the usual counterexamples presented against the theory are trivialized, as is usually the case, the applicability of the theory to conversation in general will be undermined. Supposing that Davis, Kasher, and Hintikka are right in their criticism of Grice, one can present a new version of conversation, taking a more inclusive approach to participants goals in conversation. This would not only expand the framework for conversational implicatures, but provide a means to explain concepts that Grice does not include in his Cooperative principle, such as the way in which one ought to resolve the criticisms of these maxims. It may also allow more room for phenomena such as politeness, which Grice does not account for. Once this new theory of communication, based on both maxims and convention, is fleshed out, it should provide a significantly more inclusive account of implicature, and one with greater explanatory power.

1.2 Grices Theory of Conversational Implicature

It should be helpful to start by discussing Grices theory of meaning. When talking about the meaning of a sentence, Grice notes that the term to mean can be used in some interesting variations. Take the example, Those three rings on the bell (of the bus) mean that the bus is full, this example would be quite different from, say, Those spots mean (meant) measles (Grice 1989, 213-214). This difference involves the fact that while both statements are, technically, cancellable, only the former will still be rational when it is cancelled. Thus, if the bus driver mistakenly rings the bell when the bus is not full, the bell still means the bus is full, though it is not. This is unlike the latter statement, which involves natural meaning. This second statement becomes contradictory if one adds, Those spots meant measles, but he hadnt got measles (Grice 1989, 213). Grice writes that the former statement can be restated as Those three rings on the bell mean the bus is full (Grice 1989, 214). This Grice refers to as nonnatural meaning. Grice uses the term meansNN to refer to this specific nonnatural usage. Grice gives a definition of meansNN when he states, A meansNN something by x is (roughly) equivalent to A intended the utterance of x to produce some effect in an audience by means of the recognition of this intention (Grice 1989, 220). When the bus drive rings the bell, he meansNN the bus is full; but also, if he says Theres not any room back there to a potential bus rider, he still meansNN the bus is full. It is this theory of meaning that will lead Grice to study the implicatures created by the distinction between meaning something and meaningNN something.

Grice, in his essay Logic and Conversation, attempts to provide a framework for the pragmatics of conversation. He does so in this essay by presenting a theory of implicature. His desire is to answer a debate between what he calls formalists and informalists who disagree on the meaning of ~, (, (, (, ((x), (x, (x, and their supposed English counterparts not, and, or, if, all, some (or at least one), the respectively (Grice 1989, 22). Between the formal logic of the formalists and the natural language logic of the informalists, Grice has this approach:

I wish to maintain that the common assumption of the contestants that the divergences do in fact exist is (broadly speaking) a common mistake, and that the mistake arises from inadequate attention to the nature and importance of the conditions governing conversation. I shall, therefore, inquire into the general conditions that apply to conversation as such, irrespective of its subject matter. I begin with a characterization of the notion of implicature. (Grice 1989, 24)

To define his idea of implicature he explains, I wish to introduce, as terms of art, the verb implicate and the related nouns implicature (cf. implying) and implicatum (cf. what is implied) (Grice 1989, 24). With these new terms Grice is able to clearly approach what would otherwise be very awkward subject.

Conventional implicatures, as opposed to the conversational implicatures that Grice wishes to explore, are those which contain an implication explicitly within statements. Grice gives the example, He is an Englishman; he is, therefore, brave; here, the relationship between the consequence, being brave, and the antecedent, being an Englishman, is inherent in the utterance (Grice 1989, 25). The therefore makes this implication explicit; there are other words that are able to do this job as well, for example: but, therefore, moreover, thus, hence, etc. (Bultinck 2005, 15). This explicit relationship will not be there, however, for conversational implicatures, which are non-conventional implicatures.

Next, Grice classifies conversational implicatures as a subclass of non-conventional implicatures. These types of implicatures, due to the vague nature of the implicatum will need some communication tool in order for them to be effective. Thus, there must be a fundamental rule to guide them. Grice points out,

The following may provide a first approximation to a general principle. Our talk exchanges do not normally consist of a succession of disconnected remarks, and would not be rational if they did. They are characteristically, to some degree at least, cooperative efforts; and each participant recognizes in them, to some extent, a common purpose or set of purposes, or at least a mutually accepted direction. (Grice 1989, 26)

Here, Grice makes a key step in providing his theory of conversation with the foundation that there is an implicit shared goal in conversation. There are, however, objections to this move.

It is important, though, to point out that Grice presents this principle as a normative definition of conversation. Griceans contend that participants of talk exchanges will engage in conversation, as a general concept, only when following the Cooperative Principle. Since Grice has presented this principle as a condition for normatively defined conversation, then is no problem with counterexamples emerging. Grice would just contend that the action engaged in by the participants is not conversation. However, this normative definition may be significantly less interesting as a theory of communication when compared to more general theories with more explanatory power. Intuitions regarding conversation may not be fulfilled within the Gricean framework.

Because he imposes this requirement for what counts as conversation, Grice is able to present his rough general principle which participants will be expected (ceteris paribus) to observe, namely: Make your conversational contribution such as is required One might label this the Cooperative Principle (Grice 1989, 26). This Cooperative Principle is the basis for conversational maxims. Grice decides to pay tribute to Kant by creating four categories Quantity, Quality, Relation, and Manner (Grice 1989, 26). The categories, along with there maxims, are as follows:


1. Make your contribution as informative as is required (for the current purposes of the exchange).

2. Do not make your contribution more informative than is required.

Quality: Try to make your contribution one that is true

1. Do not say what you believe to be false.

2. Do not say that for which you lack adequate evidence.


1. Be relevant

Manner: Be perspicuous

1. Avoid obscurity of expression

2. Avoid ambiguity

3. Be brief (avoid unnecessary prolixity)

4. Be orderly (Grice 1989, 26-27)

Under the Category of Quantity Grice presents two maxims Q1 and Q2. These maxims obviously are associated with the amount of information exchanged between participants engaged in conversation. The Q1-maxim is very strong; there seems to be an assumption by Grice that the amount of information necessary for participants to provide will be known through some means, such as context, if it is not made explicit. Clearly, desiderata in most conversational discourse are made explicit. The Q2-maxim here is not particularly the source of much controversy. However, Grice notes that it may be superfluous and, therefore, some may expect it to be worrisome. On one interpretation of the maxim, though Grice neglects to point out, his Q1-maxim may indeed implicitly include the Q2-maxim. This reading presumes that one is to increase the level of information up to the correct amount, the other reading merely sees it as telling participant not to go beyond the desired amount of information. This interpretation, however, does not damage the Quantity Maxims; rather, it may simply be a clarifying distinction, due to some confusion regarding the Q1-maxim.

The second category Grice lays out is that of quality. This, he states, should be understood under the supermaxim, Try to make your contribution one that is true; but, this is separated into two specific maxims:

1. Do not say what you believe to be false.

2. Do not say that for which you lack adequate evidence. (Grice 1989, 27)

These two maxims compose different aspects of this supermaxim, and their specific characteristics differ drastically. The QL1-maxim maintains strict guidelines for participants; whereas, the QL2-maxim will require participants to exercise their judgment as to the adequacy of their responses.

Next there is the Category of Relation. For this, Grice simply gives the single maxim Be Relevant (Grice 1989, 27). This seemingly simple maxim holds that one is expected to make his or her input to the conversation appropriately relevant. This maxim, however, has caused significant amounts of criticism. Grice points out that he even struggles with the problems presented by this extremely vague maxim. Indeed, Searle, Wilson, and Sperber all reject this maxim. Searle remarks that, though it is initially intuitive, it is ultimately problematic (Searle 1992, 14).

Finally there are the maxims of manner. These may be insufficient, as Grice points out, for there may be many other maxims regarding the appropriate way of engaging in cooperative conversational discourse. Also, Grice comments that these may be of lesser importance than the maxims of other categories. He writes, It is obvious that observance of some of these maxims is a matter of less urgency than is the observance of others; a man who has expressed himself with undue prolixity would, in general, be open to milder comment than would a man who has said something he believes to be false (Grice 1989, 27). Manner seem to be more aesthetic than the other categories; however, in terms of importance, it seems that the M2-maxim regarding ambiguity is arguably the most important of the maxims of manner. This is due to its relationship to equivocation and, therefore, a close relationship to the QL1-maxim.

After his discussion of these maxims he suggests that these are not the only maxims employed in conversations. He writes There are, of course, all sorts of other maxims (aesthetic, social, or moral in character), such as Be polite, that are also normally observed by participants in talk exchanges, and these may also generate nonconventional implicatures (Grice 1989, 28). It is notable that he decides to intentionally leave out certain maxims from his categories, though this may be due to the inherent limits on lectures.

Grice, to explain how these implications are to be understood, presents four ways in which maxims may be unfulfilled. These four ways are violation, opting out, being faced with a clash, and flouting. First, there may be a violation; Grice writes that a person may quietly and unostentatiously violate a maxim; if so, in some cases he will be liable to mislead (Grice 30). This could happen in many ways. Obviously, there is a violation of the QL1-maxim, which would simply be an occurrence of lying. This, however, is not the only case in which one could mislead. One could violate the M2-maxim, speaking ambiguously with the intention to misinform. There also seems to be the ability to do the same with the Q1-maxim, providing minimal amounts of information with the intention of deception. This is why courts call for people to tell the whole truth, compelling people to include all the appropriate contents.

It seems, however, that some of the maxims are not able to be violated for the purposes of misleading or deception. A violation of the maxim of relevance, for example, would only cause confusion, rather than deception, so long as all the other maxims are maintained. The same could be said for a violation of the M3-maxim, regarding brevity. It does not appear that being excessive with language could mislead, again supposing that all other maxim are being fulfilled.

Secondly, Grice says that one can opt out. He writes that one may opt out from operation both of the maxim and of the Cooperative Principle; he may say, indicate, or allow it to become plain that he is unwilling to cooperate in the way the maxim requires. He may say, for example I cannot say more; my lips are sealed (Grice 1989, 30). This principle, as the example suggests, can be directly applied to Q1-maxim of quantity. He may also, however, opt out of the Q2-maxim by attempting to filibuster during the conversation. In fact, this opting out may be applied to all of the maxims as an overt sign of unwillingness to cooperate in the conversation. However, the fact that Grice allows for individual participants to opt out at whatever time they like may be problematic to his conception of the Cooperative Principle and his reasons for its status in conversation, because opting out is actually semi-cooperative.

The last two reasons for failing to follow maxims will prove most useful to Grice and his description of conversational implicature. The third reason one may fail to follow maxims is that they may clash. Grice writes, He may be faced by a clash: He may be unable, for example, to fulfill the first maxim of Quantity (Be as informative as is required) without violating the second maxim of Quality (Have adequate evidence for what you say) (Grice 1989, 30). Grice gives the example of one person (A) asking another (B) where someone else (C) lives: A: Where does C live? / B: Somewhere in the South of France (Grice 1989, 32). In this example, A wants to know which city C lives in; however, B may not know this fact, so he might leave the Q1-maxim unsatisfied in order to fulfill the QL2-maxim. Because of this clash, however, Grice states that B implicates that he does not know in which town C lives (Grice 1989, 33). Many examples of clashes between various maxims are possible, each presenting different implications. This is not the primary style in which people implicate things; rather, most implication will come from flouting maxims.

One issue that may concern those critical to Grices account of conversation is the method by which determines which maxim will prevail when involved in a clash. It seems that whenever there is a direct clash between maxims, one much choose the most appropriate maxim to follow; however, Grice merely glosses over this as though it will be obvious. This may be a mistake for Grice. It seems that one must have a methodology of maxim superiority if one is to understand how implications are caused by a clash. Suppose, given the previous example, that B were to flout the QL2-maxim, rather than the Q2-maxim, and just guesses a city. Suppose he or she responds with Marseille?, implying that he or she does not know the city, but wishes to provide the appropriate amount of desired information. Grice gives no reason why this should not be the case. He simply asserts that some maxims are more important than others. Given the number of maxims that may occur under the Category of Manner, there may be various clashes, forcing some participants to leave maxims unfulfilled, yet there is no guide as to which maxims are more or less important in general. Presenting a strategic theory to conversation may provide answers to dilemmas of this kind.

Flouting, as opposed to resolving a clash, is clearly where Grice is gets most of his traction regarding conversational implicature. Grice explains, He may flout a maxim; that is, he may blatantly fail to fulfill it This situation is one that characteristically gives rise to a conversational implicature; and when a conversational implicature is generated in this way, I shall say that a maxim is being exploited (Grice 1989, 30). Here, the participant is supposed to make it clear that he is not opting out, there is no reason to suspect that he or she is unable to fulfill maxims due to a clash, and it is supposed to be clear that there is no intent to mislead; thus, the other participant, the listener, must be somewhat perplexed if the statement is taken literally, which will alert him or her to the implication.

Grice presents many examples of flouting. He explains, A is writing a testimonial about a pupil who is a candidate for a philosophy job, and his letter reads as follows: Dear Sir, Mr. Xs command of English is excellent, and his attendance at tutorials has been regular. Yours, etc. (Grice 1989, 33). This, according to Grice, is a case in which A must be flouting the Q1-maxim. He argues that, since A took the time to write the letter, he must not be opting out. Also, there is no reason to suppose A is being deceptive, this seems obvious. Finally, because there is no apparent reason for there to be a clash of maxims, and A ought to be imparting much more information regarding the skills the student possesses, then He must, therefore, be wishing to impart information that he is reluctant to write down. This supposition is tenable only if he thinks Mr. X is no good at philosophy. This, then, is what he is implicating (Grice 1989, 33). This structure of flouting of maxims will be the model Grice uses to explain the disconnects the often occur between what is said and what may be meant by a speaker.

Finally, Grice lays out his requirements for working out conversational implicatures. He posits that in order to work out implicatures one must rely on the following five pieces of information:

(1) the conventional meaning of the words used, together with the identity of any references that may be involved; (2) the Cooperative Principle and its maxims; (3) the context, linguistic or otherwise, of the utterance; (4) other items of background knowledge; and (5) the fact (or supposed fact) that all relevant items falling under the previous headings are available to both participants and both participants know or assume this to be the case. (Grice 1989, 31)

All of this information allows a participant to work out implicatures that he or she may be presented with. Most of these points are fairly uncontroversial; however, the fourth requirement may prove to be contentious, due to its broad nature. Regardless of this possible controversy, one should be able to work out these conversational implicatures if this information is present.

Grice does create a decent theory of communication to serve his purposes; however, as his critics and contemporaries will suggest, this theory is far from complete. With redundant maxims and potentially significant maxims left out, there is a lot to be improved upon. Still, as far as creating the framework for a general theory is concerned, he has done well by providing an intuitive format for understanding the distinction between what is said and what is meant. After a review of his followers work many of these issues will be resolved. Primarily, there is the general rejection of the relevance maxim, but also serious reformulations of redundant maxims. In addition to this, the critics of the Cooperative Principle will provide theories of implicature based on general principles of reason, rather than this controversial principle.

1.3 Epistemology and Testability of Gricean Theory

One particularly interesting work to look at before moving to criticisms of Gricean Theory is Sadocks On Testing for Conversational Implicature. Here, Sadock first lays out the claims that Grice states distinguishing conversational implicature from conventional implicature, then reviews whether these characteristics are valid in identifying types of implicature. Grice presents six different properties that Sadock discusses:

(a) Conversational implicata are capable of being worked out on the basis, inter alia, of the Cooperative Principle. That is, they are CALCULABLE.

(b) Conversational implicata are CANCELLABLE.

(c) Conversational implicata are NONDETACHABLE.

(d) Conversational implicata are not part of the meaning of the uttered forms. They are NONCONVENTIONAL.

(e) Conversational implicata are not carried by what is said, but by the saying of it.

(f) Conversational implicata may be INDETERMAINATE. (Sadock 1978, 284)

Sadock primarily concerns himself with the first three of these as practical tests for determining implicature, because he cannot see how the last three could provide any real test for conversational implicature.

Sadock almost immediately rejects the first quality (a), because the maxims are so vague that almost anything can be worked out on the basis of almost any meaning,; he ultimately determines, not surprisingly, that Calculability is not a sufficient condition for conversational implicature (Sadock 1978, 285-286). Calculability is certainly a necessary condition for implicature; however, it is not a sufficient one. Therefore, using calculability in a test for conversational implicature is not effective. Sadock then turns to detachability. However, because Grice notes that conversational implicatures are in fact based on the WAY what is said is saidon how it is put, Sadock believes that nondetachability is not a sufficient test for conversational implicature nondetachability is not strict enough to distinguish between entailment and conversational implicature (Sadock 1978, 288). Neither of these two tests work, however, cancellability, as one will see, is better suited as a test for implicature.

Sadock then addresses cancellability, the best of the tests, at length (Sadock 1978, 292). Cancellability allows an implication to be canceled without making a statement contradictory. For example, Gertrude not only just failed to swim the English Channel, in fact she swam it is at least somewhat contradictory, but Its cold in here, but I dont want you to close the door is perfectly acceptable (Sadock 1978, 292-293). One problem that Sadock makes clear, however, is that The test does not distinguish cases of ambiguity from cases of univocality plus possible conversational implicature. One of the senses of a grammatically ambiguous sentence may always be contradicted (Sadock 1978, 293). This point is that if there is an implicature in an ambiguous sentence, one of the interpretations of the statement is always cancelable. Sadock explains, The sad fact is that in the very cases where argument is likely to arise as to whether something conveyed by an utterance is conversationally implicated, the competing claim would be that the utterance is ambiguous (Sadock 1978, 294). Thus, cancellability ultimately fails as a test for conversational implicature.

Sadock does propose one more test, however. This is a test of reinforceability of a conversational implicature. This test looks for redundancy, for example, Its odd that dogs eat cheese, and they do is redundant; however, Some grades were good, but not all is perfectly informative. Still, Sadock points out, only assertions are valid test expressions for the reinforcement test while any expression that unequivocally indicates that a speaker holds a certain belief can function properly in testing for cancellability (Sadock 1978, 295). So, sadly, There is no sufficient tests for conversational implicature and no group of tests that together are sufficient Cancellability and reinforceability fail to be sufficient because, in the very important case of grammatical ambiguity, any one sense is obviously cancellable or reinforceable (Sadock 1978, 295-296). Unfortunately for students of implicature, the ability of testing for conversational implicature is not foolproof; however, Sadock provides for a general schema, using cancellability and reinforceability, for identifying conversational implicata.

Chapter 2: Criticisms, Alternatives and Neo-Gricean Theory

2.1 Kasher and Hintikka: Rationality as the Basis for Conversation

Asa Kasher presents one of the more obvious criticisms to Grices work in his essay, Conversational Maxims and Rationality. Kasher places his emphasis on the dubiousness of the Cooperative Principle; primarily, he examines whether or not the Cooperative Principle is actually inherent to conversation. Kasher begins by elaborating Grices position, primarily noting the Cooperative Principle. He then points to his own objectives, which are to show (a) that the cooperation principle has a problematic presumption, (b) that the connection between the cooperation principle and the principles to be derived from it is a problematic one, and (c) that the latter four principles follow from more general and more basic principles (Kasher 1976, 201). He attempts to show these three points by viewing conversation through a goal-achievement lens. Kasher directly addresses the second of these statements first. He makes a parallel between the Cooperative Principle and a goal-achievement principle, which he lays in the following way:

Is the cooperation principle one on which the four principles of quantity, quality, relation, and manner are well grounded? The general relationship between the cooperation principle and the other principles is parallel to the relationship between the general instruction of (5) and the more specific instructions of (6):

(5) At every stage on a way towards achieving an end of yours, act as required for the achievement of the aim.

(6) (a)Do not use the means you have for achieving your ends more or less than is required for their achievement, ceteris paribus;

(b) Try to achieve your ends by the standard use of the means you have for their achievement, ceteris paribus;

(c) At every stage on the way to achievement of your ends, consider the means being used by other persons to achieve their ends, as you come to determine the manner of your progress at that stage, ceteris paribus; and prefer using your means in a manner which is likely to help the progress of others on their way to the achievement of their ends, over any other use of these means, ceteris paribus;

(d) Give preference to means which lead you to your ends over means which lead you to situations wherein achievement of the ends themselves is just a possible result. (Kasher 1976, 202-203)

Here, Kasher notes (b)-(d) are parallel to (5) in the same way that quality, relation, and manner relate to the Cooperative Principle, respectively. The point that Kasher is making with this long set of parallels is that the Gricean maxims are not derivable from the Cooperative Principle. He begins his argument by showing that (a)-(d) clearly hinder (5). Condition (a) prevents one from using all the means one has to achieve goals, (b) prevents one from using nonstandard means, (c) requires one to consider others goals, etc. The only time when one must follow (a)-(d) is when two people cannot achieve their goals alone and must depend on each others resources. Here, one can see the parallel; as Kasher writes, Can it be that the element of cooperation may make it possible to derive the accompanying principles from the cooperation principle? It is quite clear - with one exception that such is not the case (Kasher 1976, 204). Thus, the general Cooperative Principle Make your conversational contribution such as is required, at the state at which it occurs, by the accepted purpose or direction of the talk exchange in which you are engaged, does not in necessarily lead one to follow the Gricean maxims (Grice 1989, 26). People must only follow them in cases where they cannot achieve their conversational goals without the assistance of others. Kasher also notes, about the exception regarding mutual necessity, this exception does not show that cooperation of aims and means is an essential element in the connection between the cooperation principle and the accompanying principles (Kasher 1976, 204). Thus, Kasher shows, as he intended, that the connection between the cooperation principle and the principles to be derived from it is a problematic one, or at least is not a necessary one (Kasher 1976, 201).

Kasher then lays out the rationality principle that he will rely on. He writes, (R) Given a desired end, one is to choose that action which most effectively, and at least cost, attains that end, ceteris paribus (Kasher 1976, 205). He then shows how each of the sub-principles in (6) is achieved by (R). This principle (R) has in it a clear call for efficiency. Thus, it implies the previous principle (6a) and with it the first Gricean maxim when considered in regards to conversational goals. Secondly, (6d) also follows from this (R) principle, because one would want to raise the probability of achieving ones goals as much as possible, all things being equal. Kasher also notes that the first half of (6c) is also implied by (R). The second half of (6c) and (6b), however, require more explanation. The reason why (6b) is fulfilled by (R) depends strongly on the ceteris paribus cause. Given that all things are equal (Kasher uses the example of striking a nail with a hammer as opposed to striking a nail with a coconut), one ought to opt for the standard hammering implement, due to an inherent risk that is involved when using a tool for a purpose other than its standard function. This risk may be negligible in many cases; however, no matter how small the risk, it is rational to play it safe. In this way, (6b) is fulfilled.

There are some problems, however, in the explanation of the second half of (6c), that one should prefer using your means in a manner which is likely to help the progress of others on their way to the achievement of their ends, over any other use of these means, ceteris paribus. The first notable problem that Kasher accepts is that of Hobbesian theory. In certain circumstances in a state of nature, a person might not want to use means to assist others in achieving their goals, due to a constant mistrust. Because this state of nature is such a nasty place, it is logical for one to assume the goals achieved by others may threaten a person, and may ultimately end in harming him or her. However, as Kasher argues, helping others achieve a goal creates the possibility of attaining the benefit of two things: future help from the person one assists and possible benefits from the goal he or she is trying to achieve (Kasher 1976, 208-209). Therefore, if the person one could help has the goal of, say, building a large weapon that could help him or her injure others, perhaps one should not assist that person; however, if that goal is something that one doesnt think can be used to harm others, all things being equal, the possibility of future benefit should be enough to make it rational to use means which will also benefit others. Thus, the (R) principle satisfies all of (6c).

Kasher still believes that the Gricean model is effective; however, there must be an argument for forming implicatures. Kasher writes, Since we are not accepting the cooperation principle neither as the basis for the accompanying maxims nor itself we should try improving the structure of the characteristic argument for forming implicatures (Kasher 1976, 210). In order to do this, Kasher suggests the rationalization principle. He writes, (RP) There is no reason to assume that the speaker is not a rational agent; his ends and his beliefs regarding his state, in the context of utterance supply the justifications of his behavior (Kasher 1976, 210). This principle, as a premise to Kashers argument, seems essential. On occasion people attempt to converse with non-rational agents: the mentally ill, pets, and even plants. However, supposing one is communicating with another rational agent, one must presume that when that agent begins violating (6a-d), there is a purpose. As Kasher states, If we replace, in the implicature-forming argument structure, the cooperation principle (CP) with the rationalization principle (RP), we shall not lose any power of explanation but rather gain additional power (Kasher 1976, 211). The additional power includes certain replies of silence, for example if I refrain from answering Mortons question, at the threshold of my home, Have you stopped playing the trumpet?, and I stare at him without opening my mouth, my silence has a complicated disjunctive implicature (Kasher 1976, 213).

Here, Kasher has the power to present the implicatures that Grice provides; however, he does so under the rationality principle (R), rather than the Cooperative Principle. This argument against Grice is quite effective, since it encompasses the Gricean implicature model while having greater explanatory power. Kasher accomplishes his three starting goals, showing that sometimes there is no full cooperation, because it is contrary to my interests, to a certain extent Grices cooperation principle does not permit such an explanation without radical changes in its content, its justification and the manner of its operation. We have presented such radical changes in this article (Kasher 1973, 214-215). It is this view of communication that must be exploited to have a full comprehension of implicature in general. However, this view is not widely accepted by philosophers interested in conversational implicature. It is in expounding on these ideas where the bulk of this work will be derived.

Another paper that must be explored is Logic of Conversation as a Logic of Dialogue, by Jaakko Hintikka. In this paper Hintikka complements Grice on a body of work; however, he singles out Logic and Conversation to criticize. Hintikka believes that the Gricean maxims are not, and cannot be, the rock bottom of a satisfactory analysis of the logic of conversation (Hintikka 1986, 273). One of the reasons Hintikka thinks this, is his belief that, when the time comes to conceptualize the results of discourse-theoretical observations Grice often seems to retreat back to formulations that pertain to utterances taken one by one rather than to the interplay of different utterances in discourse (Hintikka 1986, 259). Hintikka is interested in a different, more flexible framework in which the dynamics of discourse are spelt out more explicitly (Hintikka 1986, 259). First-order predicate logic is clearly not the logic of dialogue. This point, as Hintikka wants to explore, leads to a fundamental difference between propositions and the utterances of dialogue. The new strategy for understanding conversation Hintikka wants to employ is explained as follows:

Grice says that one of his avowed aims is to see talking as a special case or variety of purposive, indeed rational, behavior [Grice 1989, 28]. If so, the bag of conceptual tools one can profitably use in studying conversational logic should be a special case, or variety, of the conceptual tools one uses in studying the rationality of human behaviour in general. One such tool is game theory. (Hintikka 1986, 262)

Thus, Hintikka argues that the framework for studying dialogue needs to be shifted from formal logic to game theory. Game theory is geared toward better understanding which appropriate strategies one ought to use in given situations, or games.

Hintikka then sketches a simple schema in which conversations can be viewed game theoretically. He writes, Two speakers make moves alternately. There are four different kinds of moves: (a) Assertoric moves. (b) Interrogative moves. (c) Deductive moves. (d) Definitory moves (Hintikka 1986, 262). Hintikka then explains how each of these steps work. First, a player must make an assertoric move, in which he or she puts forward a new proposition (a new thesis) (Hintikka 1986, 262). An interrogative move is a questioning move, the answer (if one can be given) to which is then added to the list of the answerers theses (Hintikka 1986, 262). The deductive moves are pretty straightforward, it is comprised of a logical conclusion from the totality of his/her opponents theses, and previous conclusions obtained by the same means (Hintikka 1986, 263). Finally, definitory moves are when one introduces a new non-logical symbol by and appropriate explicit definition (Hintikka 1986, 263). These four moves are used to prove all the players theses, but according to Hintikka the goals can be varied.

Hintikka believes that the Gricean maxims can be incorporated into his model. In referring to maxim Q2, Hintikka notes, while Grice remarks that one ought not violate Q2 for fear of confusing the hearer, for Hintikka there is nevertheless operative, in ordinary discourse, a different pressure against extra information. Everything a player of my dialogical games says can be used against him (or her) by the opponent (Hintikka 1986, 270). Here, the player will want his discourse to be as weak as possible; thus, requiring him to prove less by the rules of the game. This is a fundamentally different reason to act in accordance with the Q2-maxim; however the end result is the same.

The same result will be found regarding the maxims of quality, QL1, and QL2. Because one only gets a payoff by proving the maximum amount of statements in the dialogue, one will only want to propose things that he or she may be able to show to be true. Surprisingly, QL1 is also satisfied by this game. Hintikka writes, if my opponent gives true answers to my question, if the opponent is fairly well-informed, and if the effects of my own answers can be discounted, then it is ceteris paribus in my own best interest to put forward true theses (Hintikka 1986, 272). Manner, unlike quantity, quality, and relevance, is not of interest to Hintikka. He states that it is different in kind from the first three (Hintikka 1986, 274). This should become apparent, as arguments to this affect will be made later. Relevance, however, must be addressed and is actually reworded to state that it is a move within the rules to increase ones pay-off. This Hintikka must explain; he states, For instead of the relevance of the several utterances in a dialogue I could collectively speak of the coherence of the dialogue (Hintikka 1986, 273). Hintikka refers to a Sherlock Holmes story in which Holmes solves a mystery about a prize race horse by asking a shepherd an apparently irrelevant question about the recent status of his sheep (Hintikka 1986, 275). However, this question, as is often the case with the solutions to intricate puzzles, was the crucial link between a series of facts that ultimately achieved the goal of solving this mystery.

From all this, one can see that Hintikka has crafted a formal game that models the Gricean maxims. This game, like Kashers work, does not require a Cooperative Principle; and, in fact, becomes a competition between the players of the game. Still, there are clearly some problematic results of this account. For example, intuitions of conversation stray far from this schema. Conversations are certainly not games in which one must prove, or at least hope to prove, all the propositions that one puts forward. Still, the idea of conversation as a goal oriented game, with pay-offs and costs, is certainly an idea which has not been explored, and may have some benefits. The primary significance of these two works, however, is their alternative approach to the theory of conversation, which is based on rationality theory.

2.2 Pre-Eminent Schools of Thought

The dominant theory of conversational implicature has been the Gricean model, though it should be noted that this subject is quite young. There does not seem to be much consensus in the philosophical community regarding the way in which meaning is communicated by these implicatures. However, in the linguistics literature about conversation, Grices seems to be the default theory explaining the nature and existence of implicature. Grices theory has been debated, and there have been three primary schools of thought regarding the subject. Firstly, there are the Neo-Griceans. This group accepts the general theory that Grice laid out and expands on it, making minimal changes to the substance of the theory. They tend not to reject major theses of Grice, such as the Cooperative Principle. Levinson is a prime example of a Neo-Gricean; Leech is also, but to a lesser extent. The second school of though are revisionists. This group usually scraps a large section of Gricean theory, for a simpler model based on fewer principles. Sperber and Wilson are the quintessential example of this with their own Relevance Theory. This theory uses one overarching principle, their Relevance Principle, to encapsulate all the ideas within Gricean theory. Finally, there are those who reject the Gricean theory on the whole and suggest that implicature is ultimately rooted in social conventions. Wayne Davis exemplifies this group providing an exhaustive attack on Gricean theory. As one will see, Davis rejects the ideas presented by Grice and resorts to social conventions as an explanation for the existence of conversational implicature. The consequences of these different schools have only recently been discussed in literature.

2.2.1 Neo-Griceans Martinich

One Neo-Gricean is A. P. Martinich. His reformulation of Grices maxims is probably the least different of all the theories. In fact, Martinich has only two major criticisms to the theory as presented by Grice. The first change Martinich proposes is to the maxims of quality. Martinich argues that they are faulty because they are too rigid:

Both are defective because too [sic] narrow. Cast as they are in terms of truth, falsity and evidence, they apply only to those speech acts that attempt to say how the world is, that is, to statements, assertions and the like. They suffer from the typical philosophical disease of fixating on serious factual statements when people often do other things with language. (Martinich 1980, 219)

He suggests, instead, a maxim that is broad enough to cover the entire spectrum of speech acts (Martinich 1980, 219). He settles on his authenticity supermaxim B:

B.Be authentic. That is, do not knowingly participate in a speech act for which the conditions for its successful and non-defective performance are not satisfied. (Martinich 1980, 220)

This rewording of the Quality Maxim certainly accounts for speech acts, and there is something to this speech act adjustment; however, it initially strikes one as odd that this speech-act-sensitive language is needed. It seems that most speech acts, at least the ones that Martinich lists, such as promising, forgiving, and apologizing all entail non-natural meanings. Grice clearly had this in mind when he proposed his maxims, so this dramatic change may be unnecessary to the theory. For those who are particular about speak act theory it may be notable. Still, it is probably excessive to edit Grices principles in this way. Regardless, Martinichs authenticity supermaxim is only one of the two alterations that he makes to Grices theory.

The second maxim that Martinich modifies is the maxim of relation. He divides this maxim into two submaxims, unlike his modification to the QL-maxim. He calls the first of these two submaxims C1, which is, Make your contribution one that moves the discussion towards its goal (Martinich 1980, 220). This modification does address two of the major criticisms levied against Grice, those of Kasher and Hintikka. By addressing the goal-oriented nature of conversation, Martinich points to one of the major ways to improve the Gricean model. However, Martinich misses one of the main intuitions that Kasher brings up, that in order to address the goal-oriented nature of conversation, one cannot simply address aspects of Grices theory, because of its presumption of the Cooperative Principle as its foundation. The only way in which this submaxim could stand without conflicting with the Cooperative Principle is if all conversational goals were cooperative; however, this is simply not the case. Thus, unfortunately, this submaxim cannot even be addressed without an entire overhaul of the Gricean system.

Martinich refers to the second of the two submaxims of relation as C2. This maxim is as follows: Express yourself in terms that will allow your hearer to tie your contribution into the conversational context (Martinich 1980, 221). This brings light onto another alternative to Gricean theory, which is that of Relevance Theory. Without the addition of C1, however, this submaxim does not seem significantly clear enough to warrant this rewording. Context is important; however, without a general theory of context, as Martinich only provides a vague quote by Strawson to found his theory, the difference between contributions being relevant and being able to be tied into a conversational context cannot be differentiated (Martinich 1980, 221-222). Thus, Martinichs reformulations ultimately fail, but they reiterate the criticisms presented in Hintikka and Kasher. Levinson

Continuing with the modifications of Neo-Gricean theory, one must mention the theory presented in Levinsons Presumptive Meanings: The Theory of Generalized Conversational Implicature. It concerns itself with general conversational implicatures (GCIs) rather than particularized conversational implicatures (PCIs). Thus, Levinsons theory of pragmatics is intentionally incomplete, as he points out when he writes, a theory of GCIs has to be supplemented with a theory of PCIs that will have at least as much, and possibly considerably more, importance to a general theory of communication (Levinson 2000, 22). One of the problems Levinson sees in the Gricean theory is that there is no major distinction between the generalized implicatures and the particular implicatures. Levinson writes:

In the immediate context of a discussion of the distinction, Grice provides only one, none too clear, example viz. the inference from the indefinite article to the assumption that the speaker is not in a position to be specific. Thus the assertion of (6) might normally carry the GCI inference:

I saw a woman in my office.

GCI: I saw someone other than my wife/girlfriend/mother/etc.

because the speaker has failed to be specific in a way in which he might have been expected to be specific, with the consequence that it is likely to be assumed that he is not in a position to be specific [Grice 1989, 37-38]. (Levinson 2000, 17)

However, Levinson claims that Grice was particularly interested in these generalized implicatures; thus, the GCI theory Levinson creates would be of great interest to philosophers with similar interests to Grices.

This GCI theory is comprised of three heuristics, upon which all of the Gricean general implicature should rest. This theory should guide the interpretation of ambiguous sentences. The three heuristics are the Q-heuristic (regarding quantity), the I-heuristic (regarding informativeness), and the M-heuristic (regarding manner). Levinson uses they symbol +> to mean implicates, as in p +> q (uttering p implicates q) (Levinson 2000, xi). This is differentiated from the symbol ++> which, for Levinson means, communicates (the sum of what is said and what is implicated) (Levinson 2000, xi).

The Q-heuristic is supposed to model Grices Q1-maxim, What isnt said, isnt (Levinson 2000, 35). This is important for his theory of scalar implicature between words like all and some, and others like not all and none. It also can be used in regards to the simpler aspects of the Q1-maxim such as Three boys came in +> not four (Levinson 2000, 36). However, the scalar implicatures presented by Levinson are particularly interesting, primarily the relation of the Q-heuristic to the traditional square of opposition. Levinson writes:

Aristotle held that in the case of the modals the I/O relation was logical but in the case of the quantifiers it was a nonlogical suggestion. Hamilton and Jespersen held the relation is logical for all the squares, De Morgan and J. S. Mill that it is nonlogical for all the squares, and so on.

The theory of GCIs helps to explain the confusion. The I (some) corner of the square carries a generalized scalar implicature to the effect that the O (not all) corner also holds. It is the generalized nature of the inference that explains the confusion even among these eminent scholars thinking deep and hard about the problem. Some strongly suggests not all, and Some in fact all or Not all, indeed none, indicating that the suggestion cannot be a logical relationship. (Levinson 2000, 68)

The significance of the Q-heuristic here is obvious. It allows the square of opposition to be understood not for just the typical (all, some( (none(not some), not all( pairs, but can be implemented even with the logical connectives (and, or( (neither/nor, not both( (Levinson 2000, 64-67). Still, the real affect that this Q-heuristic produces is the conditions by which the hearer knows that statements tend to be more efficient than they technically must be. People will often, as this heuristic allows, disregard specifier phrases such as in exactly n things, and Possibly, but not definitely X in casual speech. Thus, the Q-heuristic allows for limits on what one must say, so as not to exasperate speakers or confuse hearers.

The I-heuristic that Levinson presents is somewhat different. He defines this heuristic as, What is expressed simply is stereotypically exemplified; here Levinson mimics Grices Q2-maxim Do not make your contribution more informative than is required (Levinson 2000, 37). This heuristic covers things often left unsaid, for simplicitys sake. Levinson gives the example, If you mow the lawn, Ill give you $5. +> Iff you mow the lawn, will I [sic] give you $5 (Levinson 2000, 37). The I-heuristic also has particular applicability in explaining why conjunction statements usually imply temporal order and causal connection, and why conditional statements almost always imply a causal relation. This is why, for example when one says, He got in the car, turned on the engine, and drove away, typically that person implies that the actions were done in that order, and that there was a causal connection between them. This heuristic, however, will need to work with the M-heuristic to have any serious weight; because, in order for something to be said in the stereotypical way, there must be a way to say something non-stereotypically, which is what the M-heuristic marks for hearers.

The M-heuristic is related to Grices maxim of Manner. Levinson explains, Whats said in an abnormal way isnt normal (Levinson 2000, 38). To explain this principle Levinson continues, The underlying idea here is that there is an implicit opposition or parasitic relationship between our second and third heuristics: what is said simply picks up the stereotypical interpretation; if in contrast a marked expression is used, it is suggested that the stereotypical interpretation should be avoided (Levinson 2000, 38). This heuristic is used in order to interpret oddly worded sentence, which contain superfluous wordage, e.g. excessive phrasing or double negation. The example Levinson gives are of the following form, Bill stopped the car +> (by I) in the stereotypical manner with the foot pedal, which invokes the I-heuristic due to its simplicity; whereas, Bill caused the car to stop +> (by M) indirectly, not in the normal way, e.g., by the use of the emergency brake, which relies on the M-heuristic because of the indirect way in with the statement is made (Levinson 2000, 39). Davis criticisms of this theory of implicature, to be discussed in depth later, will effectively undermine these implicatures; thus, the relationship between these heuristics may be in jeopardy.

One of the main pieces of work Levinson incorporates, that Grice clearly fails to explicitly provide for, is his guidelines for the resolution of implicature clashes. Where Grice addresses the fact that clashes may occur, he does not explicate which of his principles will have precedence over others. This allows for the problematic situation where one must decide which maxim is most important in any given situation, and thus, which maxim one ought to follow rather than another. Levinson sees this problem and addresses it when he writes, As sentences become complex, these inferences may arise from different clauses, and traffic rules will need to be established. In short, we have a projection problem (Levinson 2000, 157). Levinsons resolution schema is as follows:

a:Genuine Q-implicature take precedence over I-implicatures;

b:In all other cases, the I-principle induces stereotypical interpretations, unless:

c:A marked expression has been used where an unmarked one could have been employed instead, in which case the M-implicature defeats the relevant I-implicature, by inducing the inference to the complement of the I-implicature that would have arisen from the unmarked expression. (Levinson 2000, 157)

More simply put, Q-implicatures > M-implicatures > I-implicatures; yet, even Levinson admits that this is much too simple of a model, because This still leaves much detail unresolvedfor example, the resolution of potential implicatures of different subtypes arising under the same principle, or of inferences coming from different clauses (Levinson 2000, 157-158). However, this general schema seems sufficient for Levinsons ends, as his theory is about implicature in general, and if pressed he should be able to provide a more in depth version to suit the needs of different problems.

This theory of generalize conversational implicature, though much more cohesively formed, has still been criticized by Davis, as noted, among others. One of the main issues is that, like Grices theory, Levinson must also incorporate the Cooperative Principle, which many have argued is faulty. Davis argues via counterexamples that this type of system will ultimately fail. Leech

Geoffrey Leech is another example of a Neo-Gricean. He presents a modified version of Gricean theory as a means to better explain pragmatics in his Principles of Pragmatics. The theory presented by Leech, however, is unique from the other Neo-Griceans. Primarily, he separates himself by explicitly stating his acceptance of goal-oriented framework for pragmatics. The second difference between Leech and other many of the Neo-Griceans is his attempt to unite the Gricean Cooperative Principle with other principles he presents, namely his Politeness Principle and Irony Principle.

In relation to his acceptance of a goal-oriented framework for pragmatics, there are many example to support this from his work. When referring to speech situations, he states, I shall often find it useful to talk of a goal or function of an utterance, in preference to talking about its intended meaning, or [the speakers] intention in uttering it (Leech 1983, 13). This position is reiterated when Leech writes, The principles of pragmatics are fundamentally non-conventional, ie [sic] motivated in terms of conversational goals (Leech 1983, 24). The reason Leech thinks that pragmatics are non-conventional, is that Leech believes that a theory of motivation, with the additional principles he will add, will provide a general theory of pragmatics.

The extra principles related to the Cooperative Principle that Leech adds are the Politeness Principle, and the Irony Principle. He formulates the Politeness Principle in a negative way, stating, Minimize (other things being equal) the expression of impolite beliefs, to which there is a corresponding positive version (Maximize (other things being equal) the expression of polite beliefs) (Leech 1983, 81). This Politeness Principle has maxims of tact, generosity, approbation, modesty, agreement, and sympathy (Leech 1983, 132). The Politeness Principle is coupled with an Irony Principle that is parasitic on the other two (Leech 1983, 142). Leech explains in the following way:

The [Cooperative Principle] and the [Politeness Principle] can be seen to be functional by direct reference to their role in promoting effective interpersonal communication; but the [Irony Principle]s function can only be explained in terms of other principles. The [Irony Principle] is a second-order principle which enables a speaker to be impolite while seeming polite; it does so by superficially breaking the [Cooperative Principle], but ultimately upholding it. (Leech 1983, 142).

Here, the complex relationship of these three principles is laid out. The Politeness Principle needed by Leech to explain the instances in which apparently cooperative communication seems to violate the Cooperative Principle, and the Irony Principle to explain how speakers can violate the Politeness Principle while being cooperative. It should be noted that Grice mentions both phenomenon; he suggested a politeness maxim in Logic and Conversation, and gives an explanation for irony in Further Notes.

The primary concern of this theory, however, is its insistence that, while pragmatics is goal-oriented, conversation as a whole is not goal-oriented, and conversational implicature is caused by a violation of maxims. These principles of politeness and irony seem quite adequate; however, his theory of conversation in general will not provide the explanatory power for conversations that are not cooperatively based.

2.2.2 Relevance Theory

The next major, new theory of conversational implicature is Relevance Theory. This theory is presented briefly in Wilson and Sperbers essay On Defining Relevance and much more clearly in Sperber and Wilsons Relevance: Communication and Cognition. This theory, as was briefly touched upon before, is significantly different from Grices and Levinsons theories, because it relies on one overarching principle. This principle is that of relevancy.

Sperber and Wilson begin by establishing their definition for ostensive-inferential communication. This essentially means that one communicates that he or she is intending to communicate something. One basic example of this is the phrase Excuse me, we must inform you that p. This phrase expresses that one intends to express something, and conveys this intention. They define this ostensive communication as such:

Ostensive-inferential communication: the communicator produces a stimulus which makes it mutually manifest to communicator and audience that the communicator intends, by means of this stimulus, to make manifest or more manifest to the audience a set of assumptions I. (Sperber and Wilson 1986, 63)

This definition will ultimately be essential to their conception of relevance, because the principle of relevance that Sperber and Wilson create is essentially about these ostensive communications, and not communication in general.

In order to establish a theory of relevance they must first lay out the conditions for which relevance can exist. They write, We are not trying to define the ordinary English word relevance we believe that there is an important psychological property a property of mental processes which the ordinary notion or relevance roughly approximates What we are trying to do is to describe this property: that is, to define relevance as a useful theoretical concept (S&W 1986, 119). The beginnings of this definition they lay out as, An assumption is relevant in a context if and only if it has some contextual effect in that context (S&W 1986, 122). This definition is clearly lacking; however, Sperber and Wilson provide it to show that context is a condition of relevance. Something can only be relevant if there is a context in which it is relevant. Another way in which this definition is lacking is that it does not show that relevance is a matter of degree (S&W 1986, 123).

The authors want to show that relevance is similar to productivity or yield, which involve some form of cost-benefit analysis (S&W 1986, 123). This cost-benefit analysis ultimately will take the form of processing cost versus contextual effects. Now, contextual effect plays the prominent rolls in determining relevancy. The amount a specific context is changed by given assumptions primarily affects relevance; however, other things being equal, an assumption requiring a smaller processing effort is more relevant (S&W 1986, 125). This yields a comparative definition which is as follows:


Extent condition 1: an assumption is relevant in a context to the extent that its contextual effects in this context are large.

Extent condition 2: an assumption is relevant in a context to the extent that the effort required to process it in this context is small. (S&W 1986, 125)

These extent conditions provide the degree basis by which Sperber and Wilson want to judge relevance.

Sperber and Wilson also define Relevance to an individual by way of a simple extension of their general principle of relevance. They state, An assumption is relevant to an individual at a given time if and only if it is relevant in one or more of the contexts accessible to that individual at that time (S&W 1986, 144). This is not a significant modification; rather, it is simply a formal move to refer to one of the common uses of the term relevance. Sperber and Wilson provide an extent condition based definition for relevance to an individual. They also define Relevance of a phenomenon in a similar way, the only main difference is that phenomena are relevant only if they are relevant to an individual.

Finally, Sperber and Wilson are ready to present their primary claim about relevance. First, there is their presumption. They write, an act of ostensive communication automatically communicates a presumption of relevance; however, they want to say something stronger in their principle. Thus, they provide a definition for a presumption for optimal relevance. This definition is as follows:

Presumption of optimal relevance

(a) The set of assumptions I which the communicator intends to make manifest to the addressee is relevant enough to make it worth the addressees while to process the ostensive stimulus.

(b) The ostensive stimulus is the most relevant one the communicator could have used to communicate I. (S&W 1986, 158)

Here, if the addressee holds this presumption, then the ostensive stimulus will appear to be significant to the addressee, regardless of whether or not he or she knows why it relevant. Next they present their Principle of relevance:

Principle of relevance

Every act of ostensive communication communicates a presumption of its own optimal relevance. (S&W 1986, 158)

This principle deals with ostensive communications; not, as the authors explicitly note, with all forms of communication. Ostensive communicators do not always provide the optimal information to their addressees, however, they necessarily intend the addressee to believe that they do (S&W 1986, 158). In addition to this question, Sperber and Wilson address the conception of filibuster. This is a situation in which do not care whether or not their information is relevant or not; it is only a delaying tactic in arguments or formal proceedings. This, however, is a rare situation in which addressees can assume that the apparent communicator is not really addressing them, and perhaps not communicating at all (S&W 1986, 159). This is not normally the case in communication, and is a notable exception to the rule.

Sperber and Wilson also explain the difference between their work and Grices approach. They note that there are many differences between the two theories, writing, One is that the principle of relevance is much more explicit than Grices co-operative principle and maxims (S&W 1986, 161). This is a quality that any revisionist theory will want to have. They continue, Another is that Grice assumes that communication involves a greater degree of co-operation than we do (S&W 1986, 161). This sidesteps a common criticism of the Gricean analysis.

One of the criticisms Sperber and Wilson bring against the Gricean theory is one of their major differences as well. That is, that Grices theory depends on people knowing, following, and expecting other to follow mutually-constitutive behaviors in order for them to communicate with each other. Sperber and Wilson expound on this when the state, Grices principle and maxims are norms which communicators and audience must know in order to communicate adequately the audience uses its knowledge of the norms in interpreting communicative behavior (S&W 1986, 162). The main difference here and the authors point out is that the communicating is second nature, not the result of cultural upbringing, as is shown by the following:

The principle of relevance, by contrast, is a generalization about ostensive-inferential communication. Communicators and audience need no more know the principle of relevance to communicate than they need to know about the principles of genetics to reproduce. Communicators do not follow the principle of relevance; and they could not violate it even if they wanted to. (S&W 1986, 162).

This difference is quite dramatic and shows that what Sperber and Wilson are dealing with is fundamentally different than what Grice accomplishes with his theory.

The most important difference, according to the authors, between theirs and Grices theories is the difference in the explanation of communication (S&W 1986, 162). They illustrate this difference as follows:

Grices account of conversation starts from a distinction between what is explicitly said and what is implicated. No explanation of explicit communication is given Implicatures are explained as assumptions that the audience must make to preserve the idea that the speaker has obeyed the maxims, or at least the co-operative principle. The principle of relevance is intended to explain ostensive communication as whole, both explicit and implicit. (S&W 1986, 162-163)

This approach is different from Grices in that Sperber and Wilson want to explain much more than Grice does. Grices theory of communication does not differentiate between different kinds of communication; Sperber and Wilsons theory depends on essential differences in the way people communicate in order to explain implicature through relevance. Bultinck (2005) reiterates this point when he writes, Relevance Theory is actually much more than a reformulation of some of Grices insights, it is meant as an outline of the system used by human beings in spontaneous inference, and in normal utterance comprehension in particular (Sperber and Wilson 1995: 94) and has specific psychological ambitions, which ultimately lead to the presentation of a cognitive architecture (Bultinck 2005, 27). Here one can see that Sperber and Wilson have different aims with their work. However, these goals incorporate the ideas Grice wanted to express in his theory of communication.

2.3 Davis: The Anti-Gricean

Wayne Davis, in his book Implicature: Intention, Convention, and Principle in the Failure of Gricean Theory, presents the most pervasive argument against Gricean theory to date. This work is critical of Grices theory as well as the theory of generalized conversational implicature and Relevance Theory. He begins by presenting Grices theory. One principle he adds, though he argues it is already presented in the theory, is what he calls Grices Razor. This principle is as follows: Grices Razor: Other things being equal, it is preferable to postulate conversational implicatures rather than senses, conventional implicatures or semantic presuppositions because conversational implicatures can be derived from independently motivated psychosocial principles (Davis 1998, 19). This principle will be essential to most if not all of Davis claims, so he cites a massive number of sources to support it, which Grandy (1989, 516) supports in addition to this list.

Davis first major argument against Grice is in the matter of quantity. He writes, The existence of quantity implicatures is undeniable. What is false is the claim that quantity implicatures are derived from or explained by the Maxim of Quantity (Davis 1998, 35). One of the main reasons for this maxim is for weaker statements to block stronger statements: The idea is that if the speaker were in a position to make the stronger statement, he should have (Davis 1998, 34). However, Davis produces a myriad of counterexamples to this. In presenting these counterexample Davis uses a different symbol from Levinson; thus, for the propositions Davis puts forward => to be read as implicates and > as does not implicate. Here Davis compares:

Did anyone die?

Some did => -(All died).

Yes > -(All died). (Davis 1998, 35)

However, some of Davis numerous counterexamples of weaker statements not implying stronger statements are as follows:

Some died > -(Only some [a few, a minority] died).

Some died > -(Some were killed [murdered, assassinated, executed,])

Some died > -(35.72% died). (Davis 1998, 35-36)

These arguments, and arguments denying that implicature, as some have considered, can be in force in certain circumstance and not in others arbitrarily, affect Grices and Levinsons theories (Davis 1998, 37).

Davis next attacks tautology implicatures. The common example for this type of implicature is War is war, and is associated with a violation of the Maxim of Quantity, as it provides no information. He begins by noting that neither Grice nor Levinson provide an account for how the Maxim of Quantity generates these implicatures (Davis 1998, 42). Then he presents some counterexamples of non-implicating tautologies to undermine the principle in general; for example, If it rains then it will rain or snow The red car is either red and fast or red but not fast, etc. (Davis 1998, 45). Additionally, he points out that when one considers statements similar to tautologies, War is an armed conflict between groups, they will typically appear to him or her as a definition, rather than consider the speaker to be implicating something (Davis 1998, 46). He writes, tautology implicatures are far and away the exception rather than the rule The fact that most tautologies lack implicatures undermines the claim that observed tautology implicatures can be derived from general psychosocial principles (Davis 1998, 45). Concluding with, the moral is clear. Generalized tautology implicatures are not explained by the Gricean maxims. Convention seems to be the only answer (Davis 1998, 46). His attack of the Gricean account of these implicatures is quite substantial. He follows in the same manner with a criticism of conjunction implicatures.

Davis continues in this fashion undermining much of the work Griceans have attempted to settle, including indeterminate implicatures, relevance implicatures, what he calls close-but implicatures, etc. (Davis 1998, 70-75). He attacks Levinsons theory when he addresses quantity implicatures more in depth, stating that scalar implicatures can easily be extended: , and that this extension undermines implicatures caused by the Maxim of Quantity (Davis 1998, 84). He then moves on to the Relevance Theory of Sperber and Wilson. He writes that beside its connection to the Maxim of Relation, the Principle of Relevance cannot follow Grices model in the following way:

The Principle of Relevance does not imply any of Grices other principles Cooperative Principle: nothing guarantees that the contribution with the greatest number of contextual implications per cost must be the contribution required by the accepted purpose of conversation. Maxim of Quality: nothing requires that the conveyed proposition or any of its contextual implications be true or justified. Maxim of Quantity more, or less, informative propositions might be proportionately less costly to process. Furthermore the Principle of Relevance fails to imply the Maxim of Manner to the extent that brevity involves sacrificing content and not just eliminating unnecessary verbiage. (Davis 1998, 100-101)

Thus, according to Davis, many of the typical implicatures Grices theory wants to present, simply cannot be shown via Relevance Theory. In Sperber and Wilsons defense though, the premise of their paper is solely to explain relevance. Conversational implicature through a relevance principle is merely an added benefit. Still, if the theory is merely about processing cost, there are many types of implicature that are based on excessively verbose language.

Finally, Davis presents his theory of implicature conventions. Primarily, he argues that conversational implicature, unlike Grices claims, is not necessarily non-conventional. He writes, We will see that being conventional does not entail being part of the meaning of implicature-bearing sentences, and that the distinction Grice marked with the terms conversational and conventional is valid even though convention is involved in both (Davis 1998, 133). His theory presents implicature as a taught practice that is perpetuated because it is useful and convenient for the practice to be perpetuated. The four reasons besides precedence for implicature to be perpetuated are that one can form a mental association with an idea, that they are habitual, that they are traditional, and that they are reinforced by social pressure (Davis 1998, 134-135). From these practices, implicatures arrive, often arbitrarily, and persist due to perpetuation. However, one thing that Davis does not directly address is a point that Sadock makes, the principles that ORIGINALLY allowed these expressions to have metaphorical senses are still vital and therefore these conventionalized implicatures are also causes where the Cooperative Principle could be invoked, but where it should not be (Sadock 1978, 287). If an implicature becomes prominent enough to become a convention, its origins must still be accounted for. Davis, not considering this fact, presents his theories for the conventional basis for quantity implicatures, tautology implicatures, disjunction implicatures, modal implicatures, relevance implicatures, close-but implicatures, manner implicatures, and interrogative/imperative implicatures. This theory is quite dramatic in the way it presents conversational implicature. There is much for Davis to explain regarding the pervasiveness of some of the common implicature. Indeed, simply because Davis can provide a genealogy for certain implicatures, does not mean there is evidence that implicatures did, in fact, arise via these methods; and, the real strength of his argument relies on what some call Obviously overly harsh, counterexamples (Bultinck 2005, 29).

Chapter 3: Conversational Goals

One of the main issues regarding implicature that may be overlooked is the purpose of conversation itself. Rather than explicate conversation as a general system of communication, one may consider viewing conversation as a goal-oriented preoccupation. This will produce different conversational kinds. Holdcroft (1979) provides a basic framework for discourse kinds. These share a similar groundwork for kinds of conversation. Through the combined lenses of Holdcrofts kinds and Kashers emphasis on goals, many of the common counterexamples that Gricean theory has come up against will be alleviated. In addition to this, the conversational theory will have greater explanatory power than Grices theory. One example of this can be found in sociological works where social conventions of cultures differ in such a way that violations of Grices maxims are an essential part of conversation. These aberrations in conversational structure can be explained via this alternative theory of conversation.

One example of this is in regards to the Zinacantan language. John Haviland writes, It is notable, parenthetically, that the Zinacantecos also frequently appear to violate Grices maxim of Quality Zinacantecos do tell premeditated lies, routinely; it is often a matter of preventing a leak in the carefully patrolled fences of privacy and domestic confidentiality (Haviland 1988, 97). Haviland points out that Grices maxims do not prevent people from telling lies to one another; in fact, his theory specifically allows for this. However, Haviland goes on to say that there is, in Zinacantn, almost a tradition, and certainly and interpretive technique, for extracting a grain of truth from the great boulders of deception that are routinely thrown about (Haviland 1988, 98). Here, the cultural norms of a language require people to tell partial, or whole, untruths with the presumption that the hearers will interpret the actual facts from these statements. This is tantamount to a culture of mass exaggeration. This evidence can be explained by a serious reformulation of the Gricean premises; this will cover more conventions that his ethnocentric theory does.

3.1 Explanatory Failures of Gricean Theory

One of the primary problems of Gricean theory is that it presumes the Cooperative Principle. This, as Kasher contends, will not always be the case in conversation; however, cooperation is often beneficial to participants in achieving certain goals. The primary goal that cooperation can achieve is reliable information transfer. When two parties have a mutual interest in transferring information to each other, they ought to follow a principle of cooperation. However, there are many instances in which two parties might want to be deceptive in their discourse. These cases occur any time participants have non-cooperative goals.

An interesting point that emerges to reinforce the idea that Grices model is insufficient is how one would come to use deception in a conversation according to Grice. Grice allows for violations of maxims; however, if one is engaging in deception then he or she is not following the cooperative efforts that each participant recognizes (Grice 1989, 26). These deceptive moves may cause the talk exchange to not qualify as conversation. Grice writes, at each stage, some possible conversational moves would be excluded as conversationally unsuitable (Grice 1989, 26). This would suggest that non-cooperative conversations are not actually conversations for Grice. For Grice to contend that one can violate maxims, and thus use deception or other unsuitable moves, within conversation seems self-contradictory. The only consolation for Grice is that he qualifies this by saying that conversations normally follow the Cooperative Principle, and that the Cooperative Principle is a rough principle (Grice 1989, 26). This leaves non-cooperative conversations unexplained by the Gricean model. Thus, an alternative model allowing for non-cooperative goals would have greater explanatory power.

Two non-cooperative goals that can manifest themselves in forms of conversation are persuasion, and entertainment. The ends of each are different; thus, their content will also be quite different. However, these other forms of conversation may reject the Cooperative Principle that communication requires. Participants may have conflicting goals, in which case communication may break down. These conflicts illustrate the uncooperative aspects of conversation, which are much more complicated than Grice allows for.

3.2 Social Norms as Presumed Goals

Conforming to social norms, or conventions, must be considered a presumed goal with regards to conversation. Social norms, like those of the Zinacantecos, will be followed to some extent, even if only out of habit, whenever people interact with each other. Grice notes that these norms will be significant when he writes, There are, of course, all sorts of other maxims (aesthetic, social, or moral in character), such as Be polite, that are normally observed by participants in talk exchanges (Grice 1989, 28). However, these social conventions play a much larger role in conversation than Grice illustrates here. The conversational kinds that emerge will be dramatically affected by these conventions, so much so that many Gricean maxims will no longer hold. For example, the amount of coercion or equivocation that one may use while attempting to persuade someone of something will often be influenced by conventions, rather than any maxims. Thus, for a more fundamental understanding of a maxims necessarily one can investigate the balance of these social conventions with Gricean maxims in different types of conversation.

Participants in conversation will often have to choose between maintaining conventions and achieving goals. Norms may be broken if the goal achieved by the conversation is more important than the goal of satisfying the social norm. For example, suppose a person was being unduly harassed by a friend. He or she may resort to shouting or profanity, in violation of social norms, in order to dissuade that friend from pursuing his or her nuisance. Whereas this would be considered highly uncouth in a normal setting, many situations may prove such that preserving certain conventions will not outweigh the importance of achieving certain goals. Thus, social norms must always be weighed against the goals of a specific conversation in order to determine which is more important.

3.3 The Communicative Goal

The communication of ideas as a goal of conversation is not often considered separate from the concept of conversation itself. It is the basis and presupposition for Gricean Theory. It is often taken for granted as the point of conversation, because diaologue is the primary medium by which overt communication occurs. It is only when communication is contrasted with other uses of conversation that one is able to see that conversation can be a tool for achieving the specific goals through utterances. Here, within the Communicative Goal, conversation is used to transmit information from one participant to another. The efficiency of this transfer will be determined by weighing the importance of the communication against various social factors. Cooperative conversation can be understood in this framework, in contrast to Gricean theory where it cannot be separated from conversation as a concept.

Presuming the Communicative Goal, Kashers and Grices results should come reasonably well. The Communicative Goal attempts to correctly transmit information from one individual to another. Though this may seem obvious, there are many obstacles in conventional conversation to achieving this; issues regarding even the most common social conventions. Because there are very few conversations in which social norms will not inhibit communication, an example can help explain the type of communication represented by this model. One common example is that of a person communicating with his or her doctor. Here, full discloser about the relevant information is beneficial for both parties. Almost all, if not all, of the Gricean maxims will be fulfilled. However, most conversations regarding communication will fall prey to the influence of the social norms held by the participants. Politeness is one of the most common of the infringing norms. Honesty in communication will always suffer to some extent (though often to a negligible degree) when this particular convention is invoked.

The primary similarity between the Gricean model and conversation with the Communicative Goal is that they both require the Cooperative Principle as a prerequisite. The Cooperative Principle is suggested by Grice, but as a principle for what would define conversation in general. This will not hold when looking at suasive conversations. Still, when approaching conversations regarding communication, most of the maxims and the Cooperative Principle will conform. For example, if one does not express true information in a conversation where ones goal is communication, that person cannot be acting in a cooperative way.

The first maxim under the Category of Quantity, Make your contribution as informative as is required, will be necessary (Grice 1989, 26). It is clear that if one intends to share an idea, that person must provide at least the minimum amount of information needed to communicate the idea. This point is fairly obvious; however, it is important to unde