Paul Grice is one of those remarkable twentieth-century philosophers’ philosophers who have greatly influenced styles of philosophical thinking (and writing1 ) without ever, or so it seems, gaining even the slightest form of notoriety outside the academic fraternity. In this capacity, he is in the company of equally ‘unmemorable’, more or less contemporaneous, thinkers like J. L. Austin, A. J. Ayer, G. E. Moore, or Gilbert Ryle, all of them based, at one time or another, in Oxford, and most of whom were actively — though in some cases only indirectly or even adversely — involved in the development of something like a world center of ‘ordinary language’ philosophy.
One of the more serious drawbacks of this disposition, however, is the relative unfamiliarity, in the orthodox reception of Grice, with the global (integrated) picture that is being presented, notably in much of his contribution to the philosophy of language. This has largely resulted in the proposal of a whole series of counterexamples to the original ‘Meaning’ (Grice 1957) hypotheses and in some fiddling around with Grice’s conversational maxims (and sometimes with the Cooperative Principle itself), as well as in very technical discussions concerning the status of different types of meaning (semantic, pragmatic). More recently, and in particular following the first (1986) edition of Sperber & Wilson’s Relevance, linguistic debates have focused upon the precise nature of the interface that is postulated to exist between the domains (or rather, the methodologies) of semantics and pragmatics. This interface, then, defines a number of conditions that are hoped to replace various extra (perhaps superfluous) theoretical concepts that were, in the past, prompted by the ‘explosion’ of meaning (Turner 1999a), i.e., its fragmentation into numerous more or less improvised distinctions of meaning types that proved more often than not to be dependent on fairly contingent contextual parameters. In general, therefore, linguists working in the line of Grice have tended to focus almost exclusively on the theme of ‘pragmatic intrusion’, or the idea that the old Gricean dichotomy between what is ‘said’ and what is ‘implicated’ can be refined by distinguishing between minimal propositions, their ‘expanded’ forms, and genuine inferential (argumentative) work on the part of the hearer, as guided by the conversational hypothesis. As such, Grice is one of the prime instigators of a ‘radically pragmatic’ take on meaning that suggests the viability of maintaining a rigorous theoretical distinction between semantics and pragmatics, while ensuring a paradigmatic continuity in the formal study of syntax, semantics, and pragmatics.
Let us not forget, though, that radical pragmatics is largely based on the ultimately unjustified (cf. Grice 1978: 119) assumption “that it is more generally feasible to strengthen one’s meaning by achieving a superimposed implicature, than to make a relaxed use of an expression”. Not only does this pre-theoretical orientation hinge crucially on the acceptance of (propositional) logic as a kind of universal (and sufficient) semantics, which is somehow acknowledged in much of pragmatic work. But many of these and similar assumptions made by Grice himself (and actually presented as suppositions at work in language users’ own understandings of utterances) are also directly related to the generally presumptive or ‘projective’ nature of communication and interpretation, which cannot be explained without reference to the rational properties that should be ascribed to speech participants. Most existing overviews of Grice’s work, especially those targeted at a linguistic and/or cognitive audience, choose to ignore issues of this rational grounding of Grice’s philosophical project. In what follows, I will try to fill out this gap (but not fill it in), indicating links to Grice’s views on (philosophical) psychology, ethics, and metaphysics. It will be suggested, though hardly argued, that it is these nonlinguistic considerations solely that can provide the necessary and ultimate rationale for Grice’s rational account of meaning. The structure of this exposition will explicitly follow some of the lines set out in Grandy & Warner’s (1986a) excellent introduction to Grice’s thinking. Insofar as possible, I will refer to Grice’s original publications, in order to convey a feeling of the historical progression of his thinking. It should be pointed out, though, that most of his papers directly relevant to the development of linguistic pragmatics can be found in Grice (1989), which also contains an important ‘Retrospective epilogue’.
Grice (1913–1988) was academically formed in Corpus Christi College, Oxford, where he began his philosophical curriculum, including the study of Plato and Aristotle (and more specifically of the latter’s Nicomachean Ethics), with W. F. R. Hardie. Grice himself acknowledges the strong influence that Hardie seems to have had on the development of a sense of rationality informing much of his own philosophical work in later times, including the belief “that philosophical questions are to be settled by reason, that is to say by argument” (Grice 1986: 46; emphasis in original). This early period is already of some interest to the student of Grice’s ideas, because in it we see the germs of Grice’s preoccupations with moral theory, especially the kind that is of Aristotelian descent, as well as his relative dissatisfaction with the way in which logic and logical connections are traditionally conceived of as exhausting the range of tools needed for solid argumentation. His demonstration of the largely argumentative structure of discourse (or conversation), moreover, will figure among Grice’s main contributions to the study of (nonlogical) meaning in natural language.
In Oxford, Grice spent the bulk of almost thirty years teaching. In this capacity, he managed to have a considerable impact on (at least) one of his students, later colleague and friend, Peter Strawson2 , as well as on the occasional sabbatical visitor, such as John Searle. In 1939, he became a Fellow of St John’s College, where he would remain until the presentation of his William James lectures in 1967. During this time, Oxford tried to recover from a rude awakening caused by Ayer’s introduction of logical positivism, as a new style of linguistic philosophy. Probably the most renowned answer to this challenge came from what would later be known as ‘ordinary language’ philosophy, in actuality made up of several heterogeneous groups of philosophers, many of whom were at one point or another concerned with the particularities of linguistic usage and their possible relevance to the study of philosophical problems. Grice himself mainly took part in the discussions that took place on Saturday mornings under the leadership of Austin, more affectionately remembered as ‘The Play Group’. Behind these language games, there was first and foremost a very strong legalist interest, oriented towards the types of meaning distinction that can be found in sentence pairs like I shot your donkey accidentally and I shot your donkey inadvertently. Presumably, none of these discussions included the words semantic or pragmatic.
Whether the inspiration came from Austin, Ryle, or Wittgenstein, ‘ordinary language’ philosophers in Oxford stressed the idea that discourse must be grounded in some collection of metaphysical judgments, or world picture. Austin, who felt sympathetic to Moore’s ‘Defence of common sense’ (Moore 1959), recast this idea in terms of a natural metaphysic that could be discovered by the philosophical investigation of detailed features of ordinary, i.e., nontechnical, discourse. Only then, or so the argument would go, could a sound foundation of philosophical thinking be achieved. Austin’s own main concerns were truth and truth telling, and in his philosophical sketches he certainly displayed less respect for logic than, e.g., Grice. Still, this overall picture left room for much diversity. The exact relationship between linguistic phenomena and specific philosophical theses would remain a heated topic of debate, even among the members of the Play Group. Moreover, they had to defend themselves against the perennial accusation of decadent linguistic snobbery haunting (this newest brand of) ‘English Futilitarianism’. From outside, it may indeed have looked at times as if a decaying academic establishment, immersed in a classical education, was trying to keep control of a discipline that threatened to escape its ancient clutches. It would do so by cultivating an idiom that could only be acquired from inside (the establishment) and that was supposed to suggest the refined linguistic sensitivities of its speakers. Ultimately these scholars, it was thought by some, are not concerned with the nature of reality, the proper subject of philosophy, but with its mere representation or appearance.
In 1967, Grice presented the William James lectures in Harvard (revised in Grice 1989). In that year, he was also appointed Professor of Philosophy in the University of California at Berkeley. Globally, this period marks the beginning of Grice’s increased interest in more technical treatments of linguistic phenomena (as well as the beginning of a more productive period in his life, in terms of producing tractable records). Grice had published an important essay on ‘Meaning’ (Grice 1957), which set the stage for subsequent discussions of (and distinctions between) ‘utterer’s meaning’ and ‘utterance-type meaning’. The intentional structure of meaning assignments was herewith revealed (cf. Schiffer 1972 on such meaning or ‘M-intentions’). But it is only in the William James lectures, which received a great deal of attention in America, that Grice would develop, and partly revise, his analysis of the relevant conditions under which meaning can be held to rely on (the structure and content of) a speaker’s M-intention. This elaboration is already succinctly outlined in Grice (1968). At the same time, it was this very project which had somehow forced Grice to find his philosophical fortune in the United States from then on, where he would benefit from closer and more intensive contacts with experts in logic and linguistics. For Grice, Noam Chomsky and W. V. O. Quine are two of his more notable models in this respect. Both scholars have concerned themselves with finding out what a ‘suitable theory’ could be (in formal syntax and the philosophy of science, respectively), and they are both advocates of a strong methodological apparatus for tackling the more intractable regions of philosophical investigation. The methodology in question, when it comes to studying the grammar of ordinary discourse, should ideally lead the researcher to see that grammar as reflecting properties of an underlying logical form. Grice deplored the fact that, though Quine and Chomsky are so strongly united in their quest for methodological rigor, they never seemed to agree on the theoretical fundamentals for the analysis of natural language. Grice himself would try to integrate their respective positions in an unfinished undertaking that aimed at presenting a syntax-cum-semantics with minimal use of transformations. These investigations in formalistic philosophizing were never published. Ironically, Grice, who appreciated the modest level of sophistication marking his own technical machinery, eventually decided to retreat from an overly formalistic treatment of natural language, partly prompted by Hilary Putnam’s remark that his philosophy had become too formal.
Towards the end of his life, Grice became more and more engaged in areas outside the philosophy of language, including philosophical psychology/biology (dealing with the Aristotelian theme of ‘life’), metaphysics, and Kantian ethics (with Judith Baker). These domains, as Grice remarked, are much less amenable to a formal treatment. And so we return to Grice’s initial qualms with the hegemony of logical thinking in argumentation, when technology starts to squeeze a developing philosophical ‘style’ out of operation. Today, what seemed relevant to Grice’s own preoccupations has become even more so in the light of subsequent evolutions in neo-Gricean pragmatics (and related areas of interdisciplinary research). If a formal (i.c., logical) apparatus “began life as a system of devices to combat woolliness [it] has now become an instrument of scholasticism” (Grice 1986: 61; emphases in original). If Grice was dissatisfied with the unnecessary proliferation of meanings attributed to certain natural language expressions, he would certainly have been with the terminological inflation associated with the increasingly technocratic order of current linguistic pragmatics.
A short autobiographical note was included in Grice (1986). In that same volume (Grandy & Warner 1986b), a list of Grice’s main (book- and article-length) publications and ‘unpublications’ can be found as well. Many of his manuscripts are deposited in the Paul Grice Archives, UC Berkeley.
‘Meaning’ (Grice 1957) constitutes Grice’s first serious attempt to distinguish between what is involved when we say (roughly) that ‘something means something’, and the concerns expressed by the collocation that ‘someone means something’. The first type of meaning,3 as in Smoke means fire or That guy means trouble, is called ‘natural meaning’, in that it points at a ‘conventional’ relation between a sign (or even an arbitrary token) and what it means. This type of meaning is not dependent on anything some instance may have meant by the sign in question. The second, ‘nonnatural meaning’ or ‘meaningnn’, defines instances of intentional communication, as in When a man says yes, he means yes, as opposed to the incidental transfer of information. Meaningnn will ultimately have to be explained in terms of one or another occasional sense of mean, or rather, language, as a human institution, functions as an artificial substitute for natural signs. The best way to go about this explanation, according to Grice, is to reduce the so-called ‘timeless’ meaning of a linguistic expression to instances of intentional communication in which an agent — an utterer — is involved on a particular occasion of the use of that expression. For Grice, the content of a meaning (intention) is identified with the intended effect that an utterance (qua object or event) should trigger.
In his paper, Grice cuts up the concept of utterer’s (or speaker’s) meaning into three parts: A speaker ‘meansnn’ something by x, which can be any meaningful sign (from a mere hand wave to full-blown linguistic phrases or sentences), if and only if she intended x to produce an effect in the hearer by means of the recognition of this intention. The effect at issue may be anything from a belief to an action prompted by the utterance of x, and the recognition involved in interpreting meaningful behavior reveals the reflexive (yet nonparadoxical; cf. Grice 1957: 384) nature of communicative intentions.4 Later, utterance-type, or sentence, meaning is seen as equated, though in a largely underdetermined way, with underlying M-intentions. The result of this move is that utterance-type meaning is effectively handled as standardized or conventionalized5 utterer’s meaning, which fits in with the Gricean program of reducing semantics to (propositional attitude) psychology. This account is meant, by Grice, to supplant so-called ‘causal’ theories of meaning, which can only deal with standard, semiotic meaning and which in any case do not seem to be able to go beyond the identification of tendencies, attributed to utterances, to produce certain (cognitive or behavioral) effects in an audience. In contrast, Grice considers that what users do (or should) meannn by a particular utterance is brought about by the audience’s recognition of an M-intention, and that this recognition is presented as a reason, not a cause, for believing or doing something. Such reasons, as opposed to causes, are intrinsically argumentative and thus constitute the core of Grice’s rational account of communication.
In Grice (1968), a subtle modification is introduced with respect to the original proposal. Instead of the straightforward ‘interventionist’ semantics devised previously, in which a speaker exerts a direct influence upon thoughts and/or actions of her audience, the hearer, in this second version, should only intend to do something, or think that the speaker believes something, as an effect of understanding an utterance. Accordingly, Grice is able to distinguish the direct intended effect from an indirect one, e.g., (for indicatives) that the hearer thinks that p herself (an effect which, even if it is indirect, may actually still constitute the prime interest of an informing utterance). Grice also pays considerable attention to the problem of (grammatical) mood in the determination of utterance-type meaning. The general idea is that, if the interpretation of what the speaker meant by x hinges on the identification of an M-intended response, generic differences in types of response will correlate with generic differences in what is meant (Grice 1969: 165). Mood operators, like the indicative or the imperative, mark (quite literally as distinct auxiliary representations in syntactic form) how a psychological state, taken as a ‘radical’ propositional6 content, is to be construed in relation to the hearer: in terms of a belief (one type of generic response) or an action. Thus, the contextual ‘occasion-meaning’ of an imperative like Pass the salt could be noted as follows: An utterer U meansnn !(the salt is passed) if and only if U produces some utterance x such that she M-intends the hearer (i) to think that U intends (to bring it about) that ‘the salt is passed’, and (ii) to intend herself that ‘the salt is (somehow) passed’ partly because of, or through, her recognition of U’s ‘first’ M-intention. Insofar as the meanings of different sentence types tend to correlate with different types of speech act, moreover, the latter can henceforth be distinguished on the basis of the various propositional attitudes that can be expressed.7 The other side of the speech act equation, that of propositional content, is in turn simply reduced to mental content (see especially Searle 1986), so that we get a general picture of language as a vehicle for the communication of thoughts, which are analytically more basic, and not as the medium of thought itself.
Further refinements are proposed in Grice (1969) and Grice (1982), which witness an increasing complexity in the technical apparatus that is being deployed. In Grice (1969), a clear typology of meaningnn specifications is provided. First, the timeless meaning of a ‘complete’ utterance-type is what would usually be considered the conventional meaning of a sentence. On any concrete use of a sentence, parts or the whole of that (token) sentence may turn out ambiguous, though. To resolve the ambiguity, one needs to refer to the ‘applied’ timeless meaning, which is what a sentence means here, i.e., on a particular occasion of use. The ‘occasion-meaning’ of an utterance-type, furthermore, refers to how speakers may mean something by uttering a sentence beyond the composite conventional meaning of that sentence. Finally, all of these meaning types, as Grice purports to show, will depend upon, but not necessarily coincide with, a fourth conception of meaning, that of the ‘utterer’s occasion-meaning’, or what the speaker intended on a particular occasion by uttering x. This fourth distinction is not only theoretically motivated, in that the whole Gricean project depends upon the reduction of linguistic meaning to intentionality, it is also needed if one wishes to deal with instances of indirect meaning (cf. Section 2.2). For ‘word-meaning’, the same distinctions apply, given that a word or nonsentential phrase is taken as an ‘incomplete’ utterance-type.
The main challenge in the remainder of this enterprise proved to be utterance-type meaning, or rather sentence meaning as structured utterance-type meaning. The notion of an automatic ‘procedure’ that a person may have in her ‘repertoire’ (Grice 1968) turns out of central importance here.8 For any (communicative) procedure to be successful, an agent (speaker or otherwise) must assume that her audience will recognize the intention by means of taking a particular token as instantiating the procedure in question. Just executing the procedure then counts as giving the audience a reason to think or intend to do something (routinely associated with that procedure). For this to work, the agent must also assume that the knowledge of such procedures is shared or mutual (Lewis 1969; Smith 1982). Of course, when dealing with sentential utterances, which are compositional, there can be no individual procedure attached to every sentence of a language, because of the generative nature of natural grammars. The interpretation of sentences, as utterance-types, is therefore dependent upon ‘resultant procedures’, which are generated by a finite stock of ‘basic procedures’. The latter apply to particular utterance-types that are components of sentences (lexicon) and to particular ways of combining utterance-types (syntax). In Grice’s project, there is a real need to pay attention to syntax, as a set of basic procedures, and incorporate it into his own meddling with meaning, which explains his interest in Chomsky’s theory.
In Grice (1968: 225), reference is made to the distinction between what a speaker may ‘say’ and what she may ‘implicate’ (imply, indicate, suggest, etc.). The program defined by this distinction is directed at, among other things, a clarification of the notion of conventional meaning, the basis of semantics. This type of meaning, and specifically lexical meaning, ultimately reduces to utterer’s meaning, too. But what a speaker ‘says’ in an utterance, by using a certain combination of words, does not necessarily exhaust the actual meanings conveyed by it (and intended by the speaker). There may be more, or even completely different, shades of meaning that she wishes to implicate. What is important, then, as a matter of methodology, is to make sure that these extra meanings, as functions of the use of words, are not confused with the words’ conventional meanings. This is where the Gricean antecedent of the semantics/pragmatics dichotomy is born. Levinson (2000: 13) refers to this as “cutting up the semiotic pie”. (For a critique of this hypothesis, see especially Cohen 1977.) Grice was particularly apprehensive about the proliferation of senses affecting certain extravaganzas of ordinary language philosophy. His Modified Occam’s Razor thus reads: “Senses are not to be multiplied beyond necessity” (Grice 1978: 118–119). It specifically concerns lexical senses, like the ones that can be distinguished for the verb believe,9 but also, and perhaps more importantly, real and derivative meanings of some grammatical counterparts of logical connectives, like and (pure conjunction, temporal sequencing,…), or (exclusive vs. inclusive), and if (material vs. strict).
There are quite a few reasons to believe that Grice’s longstanding interest in the philosophy of language is very much tailored to the proper analysis of conditional sentences, and in particular of the relation between the material conditional of propositional logic and the indicative conditional that is found in many natural languages. Any debate on the distinction between meaning and use might thus be seen as parasitic upon this more basic concern with meaning as truth-conditional (what is ‘said’), which might point to a potential incompatibility with earlier views on meaning as intentional structure.10 Grice’s point for all of the logical particles listed above is that strictly nonlogical inferences do not belong to their semantics but develop out of typical, sometimes even necessarily associated, features of their use. This does not mean, however, that the way in which such meanings are to be derived is identical for all of these (and other) expressions. In fact, we may distinguish between different ‘genera’ and ‘species’ of meaning, depending on whether they are seen as part of the linguistic code (conventional) or calculated on the basis of rational principles of conversation.
In Grice (1975) and (1978), this fragmentation of meaning is shown to be a general phenomenon, not just applicable to ‘extraneous’ meanings of logical words but to any kind of non-truth-conditional inference based on what is ‘said’. The analysis refers, first and foremost, to the assumption of conversational cooperation, voiced by the Cooperative Principle: “Make your conversational contribution 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 1975: 45) The principle is constituted by four more specific maxims, echoing the Kantian categories of Quantity (informativeness), Quality (truth), Relation (or Relevance), and Manner. Each of these, or any of their combinations, may guide the discovery of nonlogical inferences in an utterance, on the assumption that the speaker is in fact being cooperative. Under this condition, then, any kind of breaching or flouting of a maxim, or of maxims, will prompt the hearer to set up an argumentation (i.e., a series of linked propositions) so as to safeguard the original assumption of cooperation. Thus, if a professor is asked to write a letter of reference about one of her students and limits her remarks to the observation that the student has regularly attended tutorials, the conversational implicature for this particular context would be that the student in question is not a very good one. This implicature can come about because it is obvious, to any audience familiar with the genre at hand, that not enough information has been provided (Quantity). Conversational implicatures, in general, are calculable because the argumentation that leads one from observing a flouted maxim to construing a (nonlogically) derived supposition is transparent and proceeds along rational lines. In addition, they are cancelable/defeasible (nonmonotonic), nondetachable (i.e., based on content rather than form, except for Manner implicatures), and nonconventional or pragmatic.
The resulting typology for those aspects of meaning that do not fall under what is ‘said’ looks as follows. Conversational implicatures, i.e., those calculated on the assumption of cooperation, come in two varieties, particularized and generalized. The first type depends essentially on contextual information that is needed to arrive at a plausible interpretation of what the speaker may have meant over and above what she ‘said’. The second, generalized conversational implicatures (GCIs), share all the properties of their particularized counterparts but also have some sort of default status, in that they apply regardless of contextual specifications (unless they are explicitly overridden). A GCI for expressions containing a quantifier like some, for instance, specifies that ‘not all’ of the members of a designated set are meant, even though the interpretation of some in terms of the stronger quantifier all is not logically incompatible. This type of implicature is in the process of acquiring a whole new theory of its own (Levinson 2000), which stresses the idea that not all of the Gricean inferential work belongs to a rhetorical level of pragmatics, and that in fact some implicatures labeled as GCIs, including cases of disambiguation, fixing reference, and generality-narrowing, go into the determination of what is ‘said’ and thus happen before semantic interpretation. Finally, implicatures can also be conventional (and still not belong to the realm of what is ‘said’), if they are strictly speaking determined by the conventional meanings of the words to which they attach without being ‘part of’ those conventional meanings. Conventional implicatures may look like GCIs, because they, too, are supposed to hold over various contexts, but they are not in principle calculated on the assumption of rational conversational behavior, and they are probably much harder (if not impossible) to defeat. An example of a conventional implicature is the contrastive sense of but (Grice 1961), which attaches to the use of this expression in (virtually) all contexts but cannot properly be seen as a component of its meaning (which is merely conjunctive). All in all, though, it does seem that many of these distinctions are obscured or at least complicated by the fact that it is ultimately not too clear what is meant by ‘what is said’, suggesting a less than “systematic philosophical theory of language” (Grice 1989: 4) on the part of Grice.
Implicatures demonstrate the rational character of conversational behavior in two ways. First, their calculation rests on transparent chains of propositions, which constitute an argument, the stuff that rationality is made of. Secondly, their very identification relies on an original assumption of rationality in the behavior of speech participants. This is the Cooperative Principle. It is portrayed as rational, which means that we should not merely think of it “as something that all or most do in fact follow but as something that it is reasonable for us to follow, that we should not abandon” (Grice 1975: 48; emphases in original). It is not an empirical fact but a rationalist, idealized, and perhaps transcendental assumption, a foundation for Grice’s theory of communication that betrays its moral embedding. But above all, something like the Cooperative Principle demonstrates that people carry around a host of assumptions with them, expectations or projections that should help them deal with the world and that actually turn them into ‘persons’ and allow them to find ‘happiness’. The conception of a person, thus defined, is not a matter of finding some spatio-temporal continuity related to an ‘organism’s’ bodily existence, but rather of construing a chain of mnemonic states (memory) as constitutive of what will then be interpreted as a personal identity (Grice 1941). It is a notion explicitly borrowed from Locke and meant to approach the problem of consciousness from a nonbehaviorist position, where propositions about the self are indeed possible. One such proposition would be that persons do project.
The concept of rationality that is entertained by Grice, and by others in and out of his wake (like Davidson), is a classically Platonic-Aristotelian one. It may be conceived of as a regulator, directing and controlling pre-rational (biological) impulses. It is first of all a category of everyday psychology, “which can be regarded as underlying our ordinary speech and thought about psychological matters, and as such will have to be a part of folk-science” (Grice 1991: 127).11 In defining rationality, Grice refers to the condition that, for a human being to count as a person, she should have a set of ‘evaluative principles’ that allow her to make rational choices, i.e., choices that may count as rational. The idea behind this is that the behavior of ‘agents’, originally conceived of as pure automatons or robots, can be predicted on the basis of a preconceived design of configurations (physical reactions or psychological states) in response to incoming input, but only partly so. While Grice may appreciate certain behaviorist explanations for their attempt to relate psychological concepts to “appropriate forms of behavior” (Grice 1991: 124), he does not condone the urge of such dispositional accounts to exclude these very psychological concepts from their theories proper. For this would also exclude the reflexivity that is necessary to describe meaningful behavior. The way the robot is designed to work — its “genetic” makeup, so to speak — does not necessarily reflect how it actually works, precisely because the robot has also been programmed to be self-regulating. This is where its evaluative principles come in, which determine, for each individual occasion, whether or not to follow one or the other of the robot’s “biological” inclinations. Some of these evaluative principles may be replaced or modified, and the robot even has the capacity to invoke completely new, subsidiary ones, based on previous experiences with its environment. However, a small number of such principles is immune to revisions, and these must constitute the core of what it means to be a person, functioning within a society of other persons. Evaluative principles that are essential to the constitution of a rational agent are also necessarily generalized, i.e., they should count for all agents (in similar situations). If ‘happiness’, then, is one such principle, the notion of ‘personal happiness’, if it is to serve a regulating function, can only be a derivative of the system constitutive of happiness in general. We might say that this aspect of Grice’s substantiation of rationality represents his Kantian inclinations. In passing, we may also note that rationality, thus conceptualized, is more like a cluster of principles than one monolithic notion. This cluster may include the ‘end’ of being happy, next to a host of other evaluative principles (or values).
What are the implications of this ‘psychological’ approach to meaning? For one, it may demonstrate that the way in which a communicative system like language is used by humans can never be entirely predictable. This is not so much because there are irrational components to language use, which there certainly areas well, but rather because linguistic habits can always be negotiated and exploited by speakers in specific circumstances. In fact, many of the so-called pragmatic meaning phenomena depend on this notion of exploitation, and what they show is not a defect of the linguistic system/model (say, semantics) but rationality at work. Thus, semantics should not be chucked out just because there are pragmatic meanings. On the contrary, a Gricean explanation of ‘pragmatics’ should start from the validity of a semantic theory. More generally, the psychological theory envisaged by Grice contains provisions for a feeling of interest involved in our ascriptions of mental states to other minds, and such interests betray a concern for those others. Accordingly, a truly rational person is not some neutral observer of external goings-on, but a ‘passionate’ participant with a number of strong motivations (also affecting communicative interaction) which that person regards as self-justified. One of these, the assumption that people generally use language to refer truthfully to states of affairs, may actually serve as the basis for truth-conditional semantics. Semantics can therefore be seen as the professionalized account of an ordinary principle which we might call ‘charity’ and which can be read as a linguistic manifestation of the general Humean12 affect of curiosity, which is ‘the love of truth (values)’. Humean Projection, or the propensity of the mind “to spread itself on external objects” (Baker 1991: 4), thus figures prominently in Grice’s epistemological discussions. In the process of interpretation, properties of the mind are projected onto the world, and this economy of passions is regulated by principles like charity (which is a kind of trust in the utterer and, as such, a moral category; see Section 3).13 It is exactly this ‘metaphysical routine’ which also shows that semantics is not autonomous, as it is itself constituted by and through Projection. Here, Grice diverges from more classical conceptions of formal semantics, à la Frege, Russell, or even Husserl, in that the latter would disavow any attempt to reduce the problem of truth (semantics) to matters of truthfulness. Truth values in standard formal semantics are not necessarily ‘values’ in Grice’s sense, the latter marking the act of ‘telling the truth’ rather than ‘truth’ itself. Clearly, Grice sees Humean Projection as a principle of semantic organization, too, reducing meaning to psychology once more.
Grice’s conception of everyday psychological explanation offers terra firma for his analysis of utterer’s meaning (Grice 1982). If we wish to realize certain ends (e.g., survive, or be happy), we ‘know’ that an ability to produce beliefs in others will help us meet those ends. As M-intentions are extraordinarily effective ways of achieving just this, I will do something, like produce an utterance, such that my audience will recognize the underlying M-intention, adopt a corresponding belief, and perhaps treat that belief as a reason for doing something else. The point, now, is that we do have the ability to M-intend, i.e., this is an everyday psychological fact. And we expect this ability to be present in others as part of their ‘pre-rational’ makeup. We are all ‘designed’ to M-intend, not because that ability is programmed into us like an automatic reflex, but because “it is rational for us to be so ‘designed’” (Grandy & Warner 1986a: 23). The structure that allows us to M-intend is pre-rational because it is always there and not subject to revision, and it is rational because it has genitorial justification.
A pre-rational structure is, in a way, a transcendental construct, an object of metaphysics. Grice’s Carus lectures on the conception of value, written and delivered in 1983 (see Grice 1991), address such structures in view of his constructivist defense of value. Humean Projection explains “the growth of conceptualization and representation in progressively more complex creatures” (Baker 1991: 2). This type of metaphysical argumentation is called ‘creature construction’, a topic which Grice also tackles in his unpublished lectures on ‘Language and reality’ (Urbana, 1970–1971) as well as in ‘Method in philosophical psychology (From the banal to the bizarre)’, included in Grice (1991) and based on a 1975 APA address. Grice wants to demonstrate the theoretical validity of (his) psychological concepts by developing sequentially distinct psychological theories for various types of ‘pirot’. Pirots are creatures constructed by genitors, or semi-gods (e.g., a philosophical psychologist) who design14 living things. Such design is implemented in animal stuff (‘flesh’) and provides a number of procedures, basic and evaluative, that are deployed in the interaction of our creatures with their environment. Thus, Grice acknowledges the basic behaviorist position of understanding what creatures are in terms of what they do. In turn, ‘very intelligent rational pirots’, with the capacities for thought and action, can put themselves in the genitorial position and, correspondingly, observe that they are themselves designed to realize certain built-in, and more or less constant, ends, insofar as they are allowed by concrete circumstances.15 As already noted, it is also assumed here that, given the dynamic nature of context, pirots are assigned the capacity to redesign themselves, in order to cope most effectively with the ever-changing character of the environment. That is why the rational creatures that we will ultimately call ‘persons’ are also conceived of as ‘end-setters’, i.e., creatures with the ability to adopt new ends and eliminate old ones on the basis of a limited set of evaluative principles. The latter are, of course, mere properties ascribed to creatures from a strictly genitorial perspective, which betrays the genitors’ own natural (or rational) disposition to take the world to be a certain way. But the rationality of this projection itself lies precisely in the fact that we, as genitors, have good reasons to suppose that these evaluative principles are present in (Gricean) end-setters, like ourselves. By projecting such rational structure upon a creature’s ‘diagram’ for dealing with the world, the accidental property of being rational, observable in humans as an empirical fact, is transformed into an essential one. In other words, we can safely assume that, for humans — as a biological type — to reconstitute themselves as (cultural) persons, they must behave rationally (a process which Grice called ‘Metaphysical Transubstantiation’; cf. Baker 1991: 5–14).
A ‘good’ person is someone who is good ‘as a person’, or good at doing what a person should, i.e., finding reasons, constructing arguments, and many more operations derivative of these. Constructing creatures also means building a world that depends critically on the attribution of ends (finality), in the sense of what such creatures ought to do (this ought may be Humean as well). And while Grice’s genitorial account may present itself as entirely mechanistic, he insists that questions of finality and value (evaluation), which lie beyond the biologically useful, will ultimately prove irreducible. A strange union of cybernetics and vitalism is the outcome of this routine — see Grice’s John Locke lectures (1979), delivered in Oxford, on reasons, ends, and happiness (also given as the Immanuel Kant lectures in 1977 and posthumously published in Grice 2001), as well as some as yet unpublished work, like ‘Probability, desirability, and mood-operators’ (1972) and his collaborative project with Baker on Kant’s ethics.
In Grice’s account, rationality attaches to humans only accidentally (Grice 1986: 102). It is what we may deduce from our observations of humans in interaction with the world, including acts of communication with other humans. At this particular level, the maxims that elaborate Grice’s Cooperative Principle describe empirical facts of conversational behavior. But of course their status is not limited to this type of description. The maxims, and the Cooperative Principle itself, appeal to moral categories as well and can be seen as communicative instances of the evaluative principles needed for humans to count as persons (specifically, in linguistic interaction). No new properties are invented here, and a purely mechanistic account of conversational behavior might arrive at the same set of principles. But Grice’s account is thicker, as a result of his metaphysical routine, and it effectively ‘grounds’ the rational project in the essentially reflexive or self-justifying character, not just of the M-intentions involved in language, but of every manifestation of purposeful conduct, i.e., of behavior that needs to be recognized as intended rather than accidental. In his discussion of value, Grice eventually holds that all evaluative principles, whatever they may be and including their potential derivations, are necessarily true (Kant). This allows him to maintain what is probably one of the bolder positions in contemporary metaphysics, viz., the idea that relative personal values, if they spring from persons who are ‘good at’ being persons (in conditions of freedom), borrow their constructed objectivity from the argument for absolute (irreducible) value: ‘What seems good to the good man is good’ (Aristotle). Such absolute value is critical in the projection involved in creature construction, because absolute value, and nothing less, is what we as rational beings legitimately demand. But even ‘outside’ the domain of ethics, these observations still stand. Specifically, M-intentions, grounding Grice’s theory of meaning, are pre-rational structures, too, and they can only be fully motivated with reference to the demands of morality, i.e., “the necessity of rational agents accepting and acting on certain imperatives (in so far as they act rationally)” (Grandy & Warner 1986a: 38).
Grice’s reduction of meaning to psychology involves a number of argumentative steps that need to be individually validated. In the end, however, there is nothing left to validate but the validation procedure itself. At this point, we may ask why certain ‘ordinary’ ways of thinking, using concepts from both ordinary psychology and ordinary moral reasoning, serve the purpose of justifying such highly abstract theoretical constructs so well. The answer to this is metaphysical, not scientific. Grice’s investigation has shown that there are necessary concepts and categories which rational beings cannot avoid applying to reality (including their perception of other rational beings). He develops a kind of “ontological Marxism” (Grice 1991: 131), based on the simple tenet ‘They work therefore they exist’, for theoretical entities which we can quite liberally go about finding: “The entities in these categories are entia realissima. We discover these categories by discovering what parts of everyday psychology are entrenched.” (Grandy & Warner 1986a: 30–31) Any identification of entia realissima, or real objects, happens on the basis of the evaluative principles we have discovered in the course of creature construction. As a direct result of this, the structure of M-intentions can only become as complex as the demands of rationality (implied in creature construction) would allow it to if these entities are to stay ‘real’, and infinite regress is not really an issue (cf. Section 2.1). Moreover, the metaphysical entities that come out of the construction routines — and propositions may form one class of them — derive their truth from the fact that they can be ‘justifiably accepted’ within a metaphysical argument. Truth itself, as the key notion in semantic theory, is still allowed a meaningful part, but only to the extent that it is seen as emerging from a common human activity. In distinguishing between science (presumably including semantics) and metaphysics, Grice stresses the role of the people involved in creature and theory construction, respectively: a small expert elite vs. humanity in general. For Grice, only the latter group may bestow absolute legitimacy upon the outcome of that construction, which is in line with the overall pragmatic orientation of his intellectual program.
Grice’s recourse to psychology in explaining meaning may at first seem perfectly compatible with the heavy emphasis on mind and mental constructs in the cognitive sciences, where Grice is still one of the prime sources of inspiration for tackling phenomena conveniently called pragmatic (see Bach 1999). It is therefore tempting to see in Grice a forerunner of the radical turn from the philosophy of language to the philosophy of mind that has marked the second half of the twentieth century. Yet we have noticed that quite a few premises generally adopted in the cognitive sciences are not shared by Grice. For one, the appeal to truth values grounding a semantic theory of meaning should be seriously qualified, and a belief in the modular architecture of the brain does not by itself warrant a transposition of that principle onto semantic organization, which is manifestly not autonomous for Grice. Also, the kind of materialistic reductionism/eliminativism typifying much of present-day research into the relation between mind and brain is too mechanistic and in any case unwarranted by the Gricean perspective, because it does not provide for the interests that are crucial to explain any act of interpretation. Brain scholars do not investigate persons, they take organisms or systems as their objects of study. And the processing concerns that go with this biological perspective are not identical to the procedures that Grice relies on in constructing his own argumentation. Many ‘neo-Gricean’ models have been set up in various areas of investigation, e.g., that of figurative language (cf. Searle 1993), to produce empirically verifiable predictions on the ‘behavior’ of the brain (or of parts of it, the alleged modules). Yet processing models are not what is needed to pursue Grice’s linguistic goal of eliminating ambiguity in meaning theory by maintaining a sharp distinction between semantics and pragmatics. Finding out which kinds of information are prerequisite to which kinds of meaning assignment (semantic or pragmatic), as illustrated, e.g., in Grice’s account of implicatures, is not the same as locating, tracking, and comparing physiological reactions to various types of linguistic input.
On a more serious note, Grice wants to make sure that the style of psychology adopted in his inquiries is not to be confused with that of scientific (or cognitive) psychology, with its experimental bias. What is more, it is not to be expected that pirotology, as a brand of everyday psychology, may some day provide a template for the elaboration of its scientific counterpart. On the contrary, Grice explicitly warns against the possible excesses of cognitivism (as the belief that ‘knowledge’ and ‘information’ are the only viable objects in the study of behavior): “We must be ever watchful against the devil of scientism, who would lead us into myopic over-concentration on the nature and importance of knowledge, and of scientific knowledge in particular.” (Grice 1991: 161) A more fabulous illustration of this principle is provided at the very end of that same article, ‘Method in philosophical psychology’. Of course, it is taken entirely out of context here:
The very eminent and very dedicated neurophysiologist speaks to his wife. ‘My (for at least a little while longer) dear,’ he says, ‘I have long thought of myself as an acute and well-informed interpreter of your actions and behaviour. I think I have been able to identify nearly every thought that has made you smile and nearly every desire that has moved you to act. My researches, however, have made such progress that I shall no longer need to understand you in this way. Instead I shall be in a position, with the aid of instruments which I shall attach to you, to assign to each bodily movement which you make in acting a specific antecedent condition in your cortex. No longer shall I need to concern myself with your so-called thoughts and feelings. In the meantime, perhaps you would have dinner with me tonight. I trust that you will not resist if I bring along some apparatus to help me to determine, as quickly as possible, the physiological idiosyncracies which obtain in your system.’ I have a feeling that the lady might refuse the proffered invitation. (ibid.)
Apart from the references already given, a number of additional (standard and new) treatments of, or comments on, Gricean thinking deserve to be cited here. To start with, many important papers relating to Grice’s philosophical project have been collected in Kasher (1998).
On the linguistic side, Bach & Harnish (1979) present a comprehensive and systematic theory of communication within the broader framework of social interaction, adapting Grice’s notion of (reflexive) M-intentions and thereby challenging conceptions of speech act theory as entertained by the likes of Austin, Searle, and Sadock. Grandy (1989) and Neale (1992) go some way to coordinating a number of linguistically important issues in the work of Grice, in a period in which a lot of that material was still inaccessible. Schiffer, an early proponent of Grice (cf. above), has over the years been led to abandon the intentionalist program once and for all (Schiffer 1987). Sperber & Wilson (1995) have used the Gricean theory of implicature to develop their own, heavily cognitively oriented, research project, dubbed Relevance Theory. The turn to a processing interpretation of Gricean procedures is symptomatic for much of the empirical investigation that has been conducted in neo-Griceanism over the past couple of decades. Obviously, this reorientation has also had a decisive impact on psychological models of communication and interpretation, of which Clark (1996) is a good example. Clark’s study basically reinterprets the Gricean and other pragmatic theories of meaning in the light of his perspective on language use as a joint or coordinated activity type. Finally, Horn’s (1989) critical study of negative expressions, which include such diverse categories as refusals, contradictions, lies, and irony, is a classic example of the kind of (philosophically inspired) linguistic pragmatics that has developed out of Grice’s preoccupation with the details of language use. Its recent reissue adds a comprehensive state-of-the-art preface surveying past work on negation.
From a philosophical point of view, a larger picture of the analytic program entertained by Grice is presented in Black (1973), Davidson & Harman (1975), and MacKay (1972). The volume edited by Bar-Hillel (1971) proposes an early appreciation of general philosophical problems conjured up by Grice’s (and others’) essentially logical approach to the acquisition and use of natural languages. Rationality and the Cooperative Principle are the subjects of studies by Attardo (1997a, 1997b) and Sarangi & Slembrouck (1992). Neale (1990) offers a (neo-)Gricean treatment of descriptions, a classic Russellian problem, while Avramides (1989) attempts to show how Grice’s meaning and communication theory fits in with the concern, typical of twentieth-century philosophy, with knowing ‘other minds’. Davis (1998), lastly, offers a fierce yet bizarre criticism of Grice’s implicature model.
ATTARDO, S. (1997a) Competition and cooperation: Beyond Gricean pragmatics. Pragmatics and Cognition 5: 21–50.
—— (1997b) Locutionary and perlocutionary cooperation: The perlocutionary cooperative principle. Journal of Pragmatics 27: 753–779.
AVRAMIDES, A. (1989) Meaning and mind: An examination of a Gricean account of language. MIT Press.
BACH, K. (1987) On communicative intentions: A reply to Recanati. Mind & Language 2: 141–154.
—— (1999) Grice, H. Paul. In R. A. Wilson & F. C. Keil (eds.): 359–360.
BACH, K. & R. M. HARNISH (1979) Linguistic communication and speech acts. MIT Press.
BAKER, J. (1991) Introduction. In P. Grice, 1–21.
BAR-HILLEL, Y. (ed.) (1971) Pragmatics of natural languages. Humanities Press.
BLACK, M. (1973) Meaning and intention: An examination of Grice’s views. New Literary History 4: 257–279.
—— (ed.) (1965) Philosophy in America: Essays. Allen & Unwin.
CLARK, H. H. (1996) Using language. Cambridge University Press.
COHEN, L. J. (1977) Can the conversationalist hypothesis be defended? Philosophical Studies 31: 81–90.
COLE, P. (ed.) (1978) Syntax and semantics, vol. 9: Pragmatics. Academic.
—— (ed.) (1981) Radical pragmatics. Academic.
COLE, P. & J. L. MORGAN (eds.) (1975) Syntax and semantics, vol. 3: Speech acts. Academic.
DAVIDSON, D. (1986) A nice derangement of epitaphs. In R. E. Grandy & R. Warner (eds.), 157–174.
DAVIDSON, D. & G. HARMAN (eds.) (1975) The logic of grammar. Dickenson.
DAVIS, W. A. (1998) Implicature: Intention, convention, and principle in the failure of Gricean theory. Cambridge University Press.
DENNETT, D. C. (ed.) (1987) The philosophical lexicon. Blackwell.
GRANDY, R. E. (1989) On Grice on language. Journal of Philosophy 86: 514–525.
GRANDY, R. E. & R. WARNER (1986a) Paul Grice: A view of his work. In R. E. Grandy & R. Warner (eds.): 1–44.
GRANDY, R. E. & R. WARNER (eds.) (1986b) Philosophical grounds of rationality: Intentions, categories, ends. Clarendon.
GRICE, H. P. (1941) Personal identity. Mind 50: 330–350.
—— (1957) Meaning. Philosophical Review 66: 377–388.
—— (1961) The causal theory of perception. Proceedings of the Aristotelian Society, Supplementary Volume 35: 121–152.
—— (1968) Utterer’s meaning, sentence meaning, and word-meaning. Foundations of Language 4: 225–242.
—— (1969) Utterer’s meaning and intentions. Philosophical Review 78: 147–177.
—— (1975) Logic and conversation. In P. Cole & J. L. Morgan (eds.): 41–58.
—— (1978) Further notes on logic and conversation. In P. Cole (ed.): 113–127.
—— (1981) Presupposition and conversational implicature. In P. Cole (ed.): 183–197.
—— (1982) Meaning revisited. In N. V. Smith (ed.): 223–243.
—— (1986) Reply to Richards. In R. E. Grandy & R. Warner (eds.): 45–106.
—— (1989) Studies in the way of words. Harvard University Press.
—— (1991) The conception of value. Clarendon.
—— (2001) Aspects of reason. [Ed. by R. Warner.] Clarendon.
GRICE, H. P. & P. F. STRAWSON (1956) In defense of a dogma. Philosophical Review 65: 141–158.
HABERMAS, J. (1998) On the pragmatics of communication. Polity Press.
HORN, L. R. (1989) A natural history of negation. University of Chicago Press. [Reissued in 2001, CSLI Publications.]
KASHER, A. (ed.) (1998) Pragmatics: Critical concepts. Routledge.
LEVINSON, S. C. (1983) Pragmatics. Cambridge University Press.
—— (2000) Presumptive meanings: The theory of generalized conversational implicature. MIT Press.
LEWIS, D. K. (1969) Convention: A philosophical study. Harvard University Press.
MACKAY, A. F. (1972) Professor Grice’s theory of meaning. Mind 81: 57–66.
MOORE, G. E. (1959) A defence of common sense. In Philosophical papers: 32–59. Allen & Unwin.
NEALE, S. (1990) Descriptions. MIT Press.
—— (1992) Paul Grice and the philosophy of language. Linguistics and Philosophy 15: 509–559.
ORTONY, A. (ed.) (1993) Metaphor and thought. (2nd ed.) Cambridge University Press.
RECANATI, F. (1986) On defining communicative intentions. Mind & Language 1: 213–242.
SARANGI, S. K. & S. SLEMBROUCK (1992) Non-cooperation in communication: A reassessment of Gricean pragmatics. Journal of pragmatics 17: 117–154.
SCHIFFER, S. R. (1972) Meaning. Clarendon.
—— (1987) Remnants of meaning. MIT Press.
SEARLE, J. R. (1965) What is a speech act? In M. Black (ed.): 221–239.
—— (1986) Meaning, communication, and representation. In R. E. Grandy & R. Warner (eds.): 209–226.
—— (1993) Metaphor. In A. Ortony (ed.): 83–111.
SMITH, N. V. (ed.) (1982) Mutual knowledge. Academic.
SPERBER, D. & D. WILSON (1995) Relevance: Communication and cognition. (2nd ed.) Blackwell.
STRAWSON, P. F. (1959) Individuals: An essay in descriptive metaphysics. Doubleday.
—— (1964) Intention and convention in speech acts. Philosophical Review 73: 439–460.
TURNER, K. (1999a) Introduction: From a certain point of view (seven inch version). In K. Turner (ed.) The semantics/pragmatics interface from different points of view, 1–18. Elsevier.
WILSON, R. A. & F. C. KEIL (eds.) (1999) The MIT encyclopedia of the cognitive sciences. MIT Press.
Related articles: Analytical philosophy, Clinical pragmatics, Conversational implicature, Conversational logic, Humor, Semantics vs. pragmatics, Speech act theory, Truthfulness