1. Introduction
The recent publication of Donald Davidson’s Language, Truth and History (2005)Footnote 1, a collection of his essays published over the past twenty years, encourages one to compare the thoughts espoused in those essays with those written and collected in his well-known collection in the philosophy of language Inquiries into Truth and Interpretation (1984). There is a clear continuity of interests from the early Davidson (pre-1984) and the works that we find in this collection. Reading the section “Language,” however, which includes, among others, the essays “A Nice Derangement of Epitaphs” (1986), “The Social Aspect of Language” (1994), and “Locating Literary Language” (1993), one cannot help but be struck by the fact that Davidson is not here defending or expounding on the earlier ideas found in Inquiries, ideas which revolve around his attempt to develop a program devoted to articulating the requirements necessary for a theory of meaning for a natural language, but rather is developing an account of language that takes as its central idea the concept of the speaker’s communicative intentions. For many, this will seem to be a marked departure from his early work. The first to notice this departure was probably Karl-Otto Apel. He read a preprint of “A Nice Derangement of Epitaphs” in 1984 and concluded that the spirit and tone of that essay marked a new direction in Davidson’s thought about the workings of language, a direction which “… more or less overthrows the well-known older position of papers like ‘Truth and Meaning’ or ‘Radical Interpretation’ and still of ‘Thought and Talk’” (all published in Inquiries).Footnote 2 In that older position, Davidson endorsed what Apel calls a semanticist approach to language, an approach which started out from and further developed the “… approaches of Carnap, Tarski and Quine.”Footnote 3 Apel says that he [Apel] is using “the term semanticist in characterizing his [Davidson’s] approach because it seemed to [him] to make up a counter-position to extremely pragmaticist approaches to almost the same problem, as, e.g., the intentionalist approach of Paul Grice.”Footnote 4
This distinction, the distinction between the semanticist and the pragmaticist approaches, is essentially the same distinction made by Peter Strawson in his inaugural lecture at Oxford in 1969.Footnote 5 There, Strawson makes a distinction between what he calls the formal semantic theorists and the communication-intention theorists, with each group of theorists having its own set of “gods and heroes.”Footnote 6 According to the communication-intention theorist, “… it is impossible to give an adequate account of the concept of meaning without reference to the possession by speakers of audience-directed intentions of a certain complex kind.”Footnote 7 The theorist of formal semantics, however, makes no reference whatsoever to the intentions of the speaker in accounting for the meanings of words and sentences; the general idea “… is that the syntactic and semantic rules together determine the meanings of all the sentences of a language and they do this by means, precisely, of determining their truth-conditions.”Footnote 8 The struggle between these two types of theorists is in part a debate about the priority between meaning and communication. The communication-intention theorists hold that from pre-linguistic cases of communication we are able to develop the notion of meaning and language; the formal semantic theorists hold that it is because of language and the meanings that are contained in it that we are able to communicate. Like Apel, Strawson holds that Davidson is a formal semantic theorist. But Apel points out that in “A Nice Derangement of Epitaphs,” Davidson seems to be developing an account more in line with Paul Grice and other communication-intention theorists.Footnote 9
This paper is concerned to address the following questions: What happened to Davidson’s philosophy of language between Strawson’s assessment in 1969 and Apel’s assessment in 1984? Is there really a change of mind on Davidson’s part, as Apel and others have suggested?Footnote 10 Has Davidson “switched sides” and endorsed the “pragmaticist” or “communication-intention” approach? Or, to put these questions another way, is Davidson a Gricean?
In what follows I will briefly outline the semanticist and pragmaticist, or formal semantic and communication-intention, approaches taken by Davidson (pre-1984) and Grice. I will then compare Davidson’s post-1984 position with Grice, and highlight what are indeed some glaring similarities in views. My aim here, however, is to show that although there are similarities, Davidson has not abandoned his earlier views on language.
2. The Davidson Programme
Davidson’s early work in semantics, sometimes referred to as the Davidson Programme, is usually understood to operate according to two mutually dependent theories. On the one hand we have his theory about what the correct form a theory of meaning for a natural language ought to take; he famously argues that it will be a modification of a Tarskian truth theory for that language. On the other hand, we have his theory that the development of such a theory of meaning will be empirically constrained by the observable and behavioural data found in the radical case of interpretation.
In his now famous 1967 paper “Truth and Meaning,” Davidson suggests how we can develop a theory of meaning that satisfies what he takes to be two essential criteria of such a project: (1) that such a theory shows how the meaning of a sentence is composed of the meaning of its subsentential parts (words)—this is the criterion of compositionality, and (2) that such a theory shows how we can move from a finite base of rules and vocabulary to an understanding of an unlimited number of sentences—this is the criterion of recursivity. In order to achieve this, Davidson modifies Tarski’s notion of a theory of truth for formalized languages. The modifications are many, but there are two that I will highlight. First, Tarski took the notion of meaning (or translation) as basic (in order to articulate object-language sentences in the meta-language) and went on to define truth; Davidson inverts this insight, taking truth as the basic concept in order to give an account of meaning.Footnote 11 Accordingly, this theory will yield theorems that pair sentences of the object-language with the conditions under which those sentences are true (these theorems are called T-sentences):
The definition works by giving the necessary and sufficient conditions for the truth of every sentence, and to give the truth conditions is a way of giving the meaning of a sentence. To know the semantic concept of truth for a language is to know what it is for a sentence—any sentence—to be true, and this amounts, in one good sense we can give to the phrase, to understanding the language.Footnote 12
Tarski argues that such theories can be applied only to formalized languages, so Davidson’s second modification shows how we can make the theory work for natural languages as well. This requires that we amend the theory in such a way that it can deal with the inherently indexical nature of natural languages, i.e., deal with the fact that many sentences in natural languages alter in truth value depending on where and when they are uttered. To achieve this, Davidson suggests that the truth predicate ought to be relativized to a speaker, place, and time and that these relativizations be manifest in the statements of the truth conditions. For example, (1) below would be the T-sentence derived from a theory of truth for the typical English speaker who utters “That book is stolen”:
(1) “That book is stolen” is true as spoken by p at time t if and only if the book demonstrated by p at t is actually stolen.
The second aspect of Davidson’s early semantic project was his theory of radical interpretation. Radical interpretation, which Davidson borrows from Quine’s radical translation,Footnote 13 is the process by which we construct a theory of meaning for another speaker’s language. Like Quine, Davidson imagines that we are all radical interpreters faced with not only the task of interpreting radically foreign tongues, but also the task of interpreting members of our own linguistic community. On this picture, we operate according to the principle of charity, the idea that as interpreters we select as the best interpretation the one that maximizes the truth of the beliefs which would be attributed to the speaker. If we do achieve understanding of another speaker, that achievement will be articulated by a theory of meaning.
3. The Gricean Programme
Both Apel and Strawson take Davidson’s philosophy of language as the epitome of the formal semantic approach and they both contrast that theory with the theory proposed by Grice. Grice’s philosophy of language could be understood as consisting of two strands. The first strand is meant to give us a definition or theory about the literal meaning (what he calls “nonnatural meaning”) of words and sentences. His thesis is that all claims about the meanings of words or sentences can be reduced or explicated in terms of the intentions of a speaker; in particular, the speaker’s intention to produce a particular response in her audience. Let’s call this his semantic strand. The second strand deals not with word or sentence meaning but rather with speaker meaning or, with what a speaker intended to effect in her audience via the semantic meaning. This is where Grice develops his influential account of the conversational implicature. Let’s call this the pragmatic strand.Footnote 14
Grice’s semantic strand was first put forward in his 1957 essay “Meaning.”Footnote 15 Here he was concerned to show how it was that sentences like “The word X means Y” could be elucidated by “Someone meant Y by X on a particular occasion,” which would be further elucidated by “Someone meant Y on a particular occasion.” What Grice shows in “Meaning” is that a speaker’s “occasion meaning” (what the speaker meant by her words on a particular occasion) can be defined or characterized in terms of the speaker’s intentions (in later papers he attempts to show how it is that timeless meaning, e.g., can be defined in terms of speaker’s occasion meaning).Footnote 16 The importance of that paper was that it articulated just what those intentions consisted in.Footnote 17 Now, since it is “meaning” that he is trying to explicate here, Grice holds that the analysis cannot make an appeal to “intentions to mean such-and-such,” as that would clearly be circular: “A means x” and “A intends to mean x” are synonymous, as nonnatural meaning is implicitly intentional. Instead, Grice holds that the speaker intends to produce a certain effect or response in her audience; responses such as beliefs or actions. But not only does he show that the speaker must have that intention (to produce a particular response), Grice convincingly establishes that there are at least two other nested intentions that must be present in order to distinguish cases of a speaker’s nonnaturally meaning something, from cases where the speaker successfully brings about an effect or response in an audience. Grice proposes the following as a definition:
“U meant something by uttering x” is true iff, for some audience A, U uttered x intending:
(1) A to produce a particular response r
(2) A to think (recognize) that U intends (1)
(3) A to fulfill (1) on the basis of his fulfillment of (2).Footnote 18
To take a straightforward example: I utter the words “It is raining” to an audience and I nonnaturally mean that it is raining. This is so because I intend that by uttering “It is raining” my audience would have the belief that it is raining, and I intend that they would recognize that I want to impart this belief in them, and I intend that they would form this belief by recognizing that I have these intentions.
In later papers,Footnote 19 Grice does modify this definition in various ways to deal with criticisms about the sufficiency and necessity of these three intentions, but he never seems to settle on a permanent modification, and retains the three-pronged analysis presented here as the core of the concept. For our purposes, there are two points that need emphasizing: First, it is important to notice that meaning is defined here in terms of the psychological concept of intention. One cannot mean something by a particular utterance unless one intends those utterances to have some particular effect on one’s audience, and further, an utterance would not mean something if the audience did not recognize that the utterer intended to produce that effect. Meaning, for Grice, is not something that words or sentences inherently have themselves. Rather, meaning is the result of the intentional interplay between a speaker and an audience, as defined above. The second point that needs to be emphasized is the reflexive nature of these intentions, sometimes called the Gricean reflex or Gricean circle. The intention to mean something by one’s utterance is really three different, nested intentions, each one making reference to the previous one.
4. Davidson’s Post-1984 Philosophy of Language
From this brief characterization of these two philosophers’ semantic theories, it is relatively clear that they have two radically different ideas about how meaning ought to be analyzed. As Strawson and Apel make clear, it comes down to a difference between a theory of meaning that is cashed out in terms of truth-conditions, and a theory that defines meaning in terms of the speaker’s intention to communicate (produce a response in her audience). Strawson argues that the Gricean position is the correct one. Even if Davidson’s truth-conditional account of meaning gets some of the relations between truth and meaning right, it still does not get to the heart of the matter, i.e., the philosophically important aspect of language. The heart of the matter, according to Strawson, is that our notion of truth in general derives from cases where someone has stated or asserted that something is true, and further, that “… we cannot … elucidate the notion of stating or asserting except in terms of audience-directed beliefs.”Footnote 20 That is, the intention to communicate a belief is more fundamental than truth.
With Apel and Strawson, I think we can agree at this point (pre-1984) that Davidson is a formal semantic theorist and his interests in the philosophy of language have a close affinity to thinkers such as Frege, Tarski, and the early Wittgenstein. Grice is clearly a communication-intention theorist, presenting a theory that more resembles the work of J. L. Austin and the later Wittgenstein. But, as Apel (and others) points out, Davidson’s 1984 paper “A Nice Derangement of Epitaphs” (as well as the other philosophy of language papers published in Truth, Language and History) seems to throw this neat division into doubt. In “Derangements,” Davidson is concerned to outline how it is that we are able to interpret or understand malapropisms and other “mistakes” like slips of the tongue, garbled grammar, mispronunciations, etc. The main target of that essay is, among others, the conventionalist thesis put forward by David Lewis and Michael Dummett which suggests that the meaning of a word or sentence is established by convention or, in other words, is determined by reference to the conventional language to which it belongs.Footnote 21 Davidson argues that this cannot be correct as malapropisms and slips of the tongue have no conventional meanings, no matter which way one defines convention; yet we still are able to understand what those words mean. Using one of his examples, we can agree that there is no convention that associates Mrs. Malaprop’s utterance of “a nice derangement of epitaphs” with a nice arrangement of epithets, yet we know that that is what she meant. It is obvious, he suggests, that we know what she meant because we grasped what she intended to mean by those words on that occasion. Realizing this, the philosopher of language must put forward a theory that gives due weight to the importance of the speaker’s intentions.
It is precisely in these more recent essays from Language, Truth and History that Davidson makes explicit the role of intentions in his philosophy of language; and in many places he applauds the importance of Paul Grice’s work in showing just how the speaker’s intentions are to be worked out and analyzed. Although the importance of Grice has been acknowledged by Davidson in various places prior to these essays, it is only now that we see how much of the Gricean programme he feels himself committed to. This is most explicit in Davidson’s analysis of what he takes to be the proper notion of meaning on which a theory of meaning should concentrate.
Davidson refers to what he has in mind here as “first meaning” and uses the word “first” because he wants to avoid connotations usually associated with the words “literal” meaning or “standard” meaning. Most notably, “literal” and “standard” meaning are too strong. They imply that the meanings of the terms used in a particular utterance depend on certain conventions, rules, or traditions in the language community, all of which the utterer may be unaware. Some would argue that if the speaker was unaware of these traditions or conventions, and didn’t intend the word to have a certain meaning, nonetheless, the word still has that literal meaning. And furthermore, “speaker” meaning is too weak. It does not allow enough room to make a distinction between what an individual means by the words on a particular occasion and what the words themselves mean. But Davidson takes for granted that “… nothing should obliterate or even blur the distinction between speaker’s meaning and literal meaning.”Footnote 22
Davidson suggests that the best way to get the distinction between speaker’s meaning and sentence/word (literal or first) meaning is through the intentions of the speaker. In any action an individual undertakes, there are many intentions to be found, and it seems that every intention can be placed in a nexus of further intentions. Take the act of opening a door. I intend to open the door; this intention is related to a larger purpose (intention) of letting some air into the room, for the further purpose (intention) that the smoke dissipates, so that I do not ruin the drapes, etc. But intentions go the other way as well; that is, just as they are related to further or more ultimate intentions, there are other, say, penultimate intentions that allow the act of the door’s being opened to take place. E.g., I intend to move my arm, get out of the chair, cross the floor, turn the door knob to the right, etc. Speaking is just another kind of action (a speech act) and the intentions involved can be just as complex and numerous. I intend to move my mouth and tongue in this way, I intend these sounds to be emitted from me, I intend to say something with these sounds, I intend that the audience recognize the meaning of these sounds, I intend that by my audience’s understanding these sounds that certain further effects will be achieved (e.g., I utter an imperative sentence), and I want these effects to be carried out because I have the further intention that such-and-such will happen, and so on. Of course, there has been much good work done on the analysis of various kinds of speech acts. Davidson’s account of first meaning is more or less J. L. Austin’s distinction between locutionary, illocutionary, and perlocutionary speech acts.Footnote 23 Davidson makes his distinction between semantic intentions (first meaning), intentions of force, and ulterior intentions.Footnote 24
The intentions that are crucial in understanding first meaning are the semantic intentions. It is these intentions upon which the others are dependent and we cannot grasp the intentions of force or the ulterior intentions (if these are to be made manifest at all) if we do not first grasp the semantic intentions. Further, in order for the act to count as a speech act, we need to grasp the semantic intentions of the speaker and see how the intentions of force and the ulterior intentions are built up out of these. It is certainly the case that we can know what a speaker intends us ultimately to do, or what she ultimately wants to achieve, without understanding her “words” qua words. Screams and sighs, for example, are verbal or oral performances in the sense that they involve, like linguistic performances or speech acts, sounds (or other public marks) being emitted from an individual which have, in some sense, a meaning (not first meaning). Through these sounds we can determine that the “speaker” is frightened or bored, and we could say that that is therefore what is meant by those sounds: “The scream meant he was frightened”; “The sigh meant she was bored.”Footnote 25 What the speaker means or ultimately intends cannot always, or even usually, be identified with what the words that she is speaking mean. As an outsider, not understanding the words of the speaker but only grasping what was ultimately meant, we are no better off than in the case of the sigh and the scream. What makes an act a speech act is that the words are bestowed with a certain meaning, a meaning conceptually independent of the speaker’s ultimate intentions. When we understand the words as the speaker intended us to understand them and then determine what further intentions the speaker had in uttering those words, then this is a speech act, or a case of linguistic communication.
Davidson’s point here is that words have an autonomy of meaning. By this he means that words have a meaning independently of what non-linguistic use an individual wants to put those words to. We frequently use words and sentences to achieve many of our extra-linguistic and ultimate intentions. I utter the words “You are a son of a stickleback fish” with the extra-linguistic and ultimate intention of offending my hearer; people who are present or overhear me can say: “He meant to offend her.” But they cannot say that that is what the words mean. The meaning of the words is independent of these intentions; they meant: my hearer is a son of a stickleback fish (or, the sentence is true if, and only if, my hearer is a son of a stickleback fish). Davidson says:
The ulterior purpose may or may not be evident, and it may or may not help an interpreter determine the literal meaning. I conclude that it is not an accidental feature of language that the ulterior purpose of an utterance and its literal [first] meaning are independent, in the sense that the latter cannot be derived from the former: it is of the essence of language. I call this feature of language the principle of the autonomy of meaning.Footnote 26
5. Is Davidson a Gricean?
What puzzles Apel and others is that the notion of meaning that Davidson presents here, and in the other essays found in Language, Truth, and History, seems to fall more in line with the communication-intention theorists. Besides the almost explicit appeal to Austin’s division between the locutionary, illocutionary, and perlocutionary, Davidson is also clearly attempting to cash out semantic meaning in terms of the speaker’s intentions to produce a response in her audience, a project not too dissimilar from Grice. Here are a two sample quotations from Davidson in Language, Truth, and History:
Meaning, in the special sense in which we are interested when we talk of what an utterance literally means, gets a life from those situations in which someone intends (or assumes or expects) that his words will be understood in a certain way, and they are. In such cases we can say without hesitation: how he intended to be understood, and was understood, is what he, and his words, literally meant on that occasion. Thus for me the concept of ‘the meaning’ of a word or sentence gives way to the concepts of how a speaker intends his words to be understood, and of how a hearer understands them. Where understanding matches intent we can, if we please, speak of ‘the’ meaning; but it is understanding that gives life to meaning, not the other way around.Footnote 27
and in “Locating Literary Language”:
But just as words have a meaning only in the context of a sentence, a sentence has a meaning only in the context of use, as part, in some sense, of a particular language. There would be no saying what language a sentence belonged to if there were not actual utterances or writings, not perhaps, of that very sentence, but of other sentences appropriately related to it. So in the end, the sole source of linguistic meaning is the intentional production of tokens of sentences. (emphasis mine) If such acts did not have meaning, nothing would. There is no harm in assigning meanings to sentences, but this must always be meaning derived from concrete occasions on which sentences are put to work.Footnote 28
That there are similarities between the proposals put forward by Grice and Davidson is undeniable. These two quotations state exactly the view of the communication-intention theorists as outlined by Strawson: that the meaning of a word or sentence is to be cashed out in terms of the speaker’s intentions and that understanding the meanings of these words and sentences can always be traced back to actual actions of the speaker (e.g., acts of asserting). Furthermore, Davidson explicitly endorses the idea of reflexive intentions. These are the intentions that we find in Grice’s three-pronged analysis of meaning. According to that view, I cannot merely intend to mean something: I also must intend that you grasp that I intend to mean something, and I must intend that you recognize my intention that you grasp that intention, and I intend that you grasp that intention, etc.Footnote 29 This is the position of meaning that Davidson puts forward in contrast with the “Humpty-Dumpty” picture of meaning that Dummett accuses him of endorsing in his [Davidson’s] introduction of the notion of intentions:Footnote 30
There are two natural pictures of meaning. One depicts words as carrying a meaning independently of speakers. … The opposite picture is that which Humpty Dumpty is using. On this conception it is the speaker who attaches the meaning to the word by some inner mental operation; so anyone can mean by ‘glory’ whatever he chooses. Each picture is crude; each is easily ridiculed by a philosopher or a linguist. But each theorist of language tends to offer a more sophisticated version of one picture or the other. Davidson’s is a version of the second picture.Footnote 31
The difference between the reflexive account and the Humpty-Dumpty account is that we cannot have intentions that we know cannot be fulfilled. Both Davidson and Grice argue for this. Although I may want or desire or wish it to be the case, e.g., that I am the King of England, I cannot intend that I be the King of England since I do not believe that this will ever come to pass (I do not have any royal ancestors). Likewise, if I believe that it is not in the capacity of my audience to know what I mean by a certain word or sentence, then I cannot intend that the word has that meaning since meaning is partly characterized by the audience’s recognition of my intention. The Humpty-Dumpty position is absurd. Davidson says: “In speaking and writing we intend to be understood. We cannot intend what we know to be impossible; people can only understand words they are somehow prepared in advance to understand.”Footnote 32 And: “Thus for me the concept of ‘the meaning’ of a word or sentence gives way to the concepts of how a speaker intends his words to be understood, and of how a hearer understands them.”Footnote 33 Grice also holds this view:
It is in general true that one cannot have intentions to achieve results which one sees no chance of achieving; and the success of the intentions of the kind involved in communication requires those to whom communications or near communications are addressed to be capable in the circumstances of having certain thoughts and drawing certain conclusions.Footnote 34
We can also see that the adoption of this position requires that we place more emphasis on the role of the particular individuals who are speaking, as opposed to on the language or on the community of language users. Languages and communities do not have intentions, only individual speakers do, and hence it is only individual speakers who can be said to mean anything (in the nonnatural sense, for Grice). Hence we see both Grice and Davidson abandoning the idea that meanings are necessarily conventional and they switch from speaking of languages (except derivatively) to speaking of idiolects. For both, the idea of an idiolect is primary and public languages are to be defined as generalizations or abstractions out of the successful communication of individuals via their idiolects. So, for example, in Grice’s analysis, where timeless meaning is defined in terms of utterer’s occasion meaning (and this in terms of the utterer’s intentions), a necessary step between them is an appreciation and account of the speaker’s idiolect.Footnote 35 Of course, this is not Davidson’s way of dealing with the idea, and it is here that I think some of the differences between them show up.
One major difference between Grice and Davidson is that for Davidson a sentence or word only has meaning in the context of a compositional theory of meaning. When Davidson refers to first meaning, he does not mean that the meaning can exist before a speaker has an idiolect, although this is certainly what Grice seems to hold. For Davidson, first meanings are in part differentiated by the role they play in such a theory (a Tarski-style truth theory); the meaning of a sentence or a word is just one possible derivation from this theory. It makes no sense, for Davidson, that a word or sentence could have a meaning independent of such a structure. This is not the case for Grice. He holds that utterer’s meaning, or nonnatural meaning in general, conceptually, comes first. Since idiolects, languages, and timeless meaning are defined in terms of utterer’s meaning, it is then possible for an utterer to mean something without there being an idiolect or language. In non-linguistic cases, this may be easy to agree with. The idea that a hand-wave means that the utterer wants to greet you does not seem to require a theory in which it is compositionally couched. But in the linguistic case, the cases we are interested in, this rejection is, at least on Davidsonian grounds, dubious.
One reason why it is dubious is that it does not allow any room for the notion of the autonomy of meaning, at least as understood by Davidson. Without that principle, we cannot make a distinction between cases where we are dealing with linguistic utterances and those cases where we are dealing with non-linguistic performances; that is, without the principle, we cannot make a distinction between, say, a scream or a yelp of surprise, and a sentence, say, “You are a son of a stickleback fish.” Nor are we able to make a distinction between an utterance’s first meaning (or literal meaning) and the force or ulterior intentions of the person’s utterance. For Davidson, as we have already seen, the autonomy of meaning is the idea that words have (first) meaning independent of the uses that those words are put to (the uses given by the intentions of force and the ulterior intentions). A scream does not possess a (first) meaning; it is just used to perform certain non-linguistic tasks. “You are a son of a stickleback fish,” however, does have a (first) meaning independently of its various uses. Even though my ulterior intention is to offend somebody (and for them to recognize that I have offended them, and for them to be offended on recognition of this intention), I also intend that the words have a meaning independently of this intention (and I intend that my audience recognize this intention, etc.).
Another reason why Davidson would reject the claim that we can intend a sentence to have a meaning independently of a theory of meaning is the well-known and almost universally accepted idea that our infinite linguistic competence is accounted for by only finite resources. That is, there is no upper bound to the number of sentences we can understand or intend others to understand, even though we possess only finite resources, i.e., a finite store of names and other words. In order to account for this, any decent theory of meaning needs to articulate the recursive and compositional features of this competence. It cannot be the case that we are able to pair a meaning (or effect) with each of these unlimited number of sentences. There are two reasons for this: (1) the learning we would have to undertake would also have to be never-ending, which is contrary to the facts, and (2) the “pairing” theory does not tell us how to go on in new cases. It in no way prepares us to account for the meaning of a sentence we have never heard before, although it may be the case that we understand each of the constituent parts of the sentence. The Gricean programme is no more than a “pairing” theory, or what Davidson calls in other places a “building-block theory of meaning.”Footnote 36
This probably marks the sharpest point at which we can say that the Gricean and Davidsonian accounts diverge. Grice wants to define meanings in terms of non-linguistic intentions, while Davidson holds that that project is ill-fated. In a 1993 interview, Davidson was asked how close his account was to Grice’s. He responded:
Grice got some relations between intention and meaning right. For example, feedback: the fact that you not only intend somebody to interpret you in a certain way by acting as you do, but you intend them to get what you mean through their detecting that you have that intention. That’s a very subtle thought, and he made it clear that you have to say something like that. I think you have to say that about first meaning and about force. If you intend something to be an assertion, part of that intention must be that people take you to be intending that it should be an assertion. So, I’m a Gricean to that extent.
But I don’t think you can define the notion of linguistic meaning on the basis of intentions; it’s a necessary condition, but it’s not sufficient. And even if you could define meaning on the basis of intention, I wouldn’t be very interested, because intention seems to me at least as hard to explain as meaning. So, for those two reasons I’m not Gricean.Footnote 37
For Davidson, defining meaning in terms of the speaker’s intentions would not be a sufficient analysis since meaning is an abstraction out of cases of successful communication. This relies not only on the nested intentions of the speaker but also on the fact that the hearer actually interprets the words as so intended. To leave out this social aspect, the hearer or the community, in a theory of meaning is to endorse a version of the Humpty-Dumpty theory of meaning.
But it is the second reason that Davidson gives in the above quote, however, that sets his programme in stark contrast with Grice’s: intentions are just as hard to define as meanings. If we were able to carry out this reduction, Davidson tells us that it would be uninteresting. The reason is that meanings, beliefs, desires, and intentions are all inter-related semantic concepts, and the attempt to define or reduce any one of these (e.g., say, meaning) in terms of the other (e.g., intention) would not constitute the achievement of reducing a semantic concept to a non-semantic one, but would have resulted only in shifting one’s focus from one semantic concept to another. Davidson has held this anti-Gricean sentiment since his earliest work:
The difficulty is that specific intentions are just as hard to interpret as utterances. Indeed, our best route to the detailed identification of intentions and beliefs is by way of a theory of language behaviour. It makes no sense to suppose we can first intuit all of a person’s intentions and beliefs and then get to what he means by what he says. Rather we refine our theory in the light of the other.Footnote 38
From Davidson’s perspective, it would seem that Grice has shifted our focus while claiming he has effected a successful reduction. The structure of the three-prong analysis, as it is originally set out, is to define an intention to mean something (what Grice sometimes calls an “M-intention”Footnote 39) in terms of intentions that are not M-intentions, i.e., are non-semantic. The intentions that are mentioned in the analysis of M-intentions, according to Grice, make no reference, explicit or implicit, to semantic entities. Davidson would have at least two problems with this.
First, intention, like belief or desire, is an attitude toward a proposition or utterance. It seems that what Grice wants, in order to avoid circularity, is an attitude (intention) without the proposition. But it is generally recognized that an intention is an “intention that …” or “intention to …” where the ellipsis is filled with a sentence or utterance. If one’s intentions are not intentions with respect to sentences (or utterances or propositions), what sense of “intention” is Grice using? Obviously, some notion of a non-linguistic intention. The idea is not absurd, but it is hard to make use of in a theory of meaning, and equally hard to make use of when we are trying to describe linguistic behaviour. It is generally recognized that our desires, beliefs, intentions, etc. can be as fine-grained as our language and theories about the world will permit. But without this fine discriminating system, how are we to distinguish between two related beliefs (or intentions, or desires, etc.) that have slightly different senses? Any attempt to answer this question, that is, any attempt that does incorporate some sort of system that allows us to discriminate between the various intentions, is going to have to be a system that is compositional and recursive, just like the meanings that we are trying to explain. So, either we bring in semantic intentions to account for the complexity of the propositional attitudes that we do have, in which case the reduction of M-intentions in terms of non-linguistic intentions fails (i.e., is circular), or we are stuck in the uncomfortable position whereby we have reduced M-intentions to non-linguistic intentions that in no way go to explain the M-intentions.Footnote 40
This criticism is closely related to the second, more encompassing criticism, viz., that Grice gives priority to thought over language. That is, Grice regards thought as somehow “there,” independent of and prior to language; presumably, if we did not have a means of communicating our thoughts, those thoughts would still exist. On Grice’s picture, our mental life could be just as complex and rich as it is now, without a linguistic ability to communicate those thoughts. Grice does not explicitly state his position in this way, but it is certainly a natural conclusion that can be inferred from his arguments. In fact, there have been several thinkers who have seen the reduction of semantics and language to non-linguistic intentions as a means of partly explaining the relation between mind and body. The line runs: reduce the linguistic and communicative life of an individual in a linguistic community to the intentions of the speaker, and then to the non-linguistic intentions of the speaker; these intentions need make no reference to the public language or the external aspects of the world (as publicity is defined in terms of the private), and hence possibly could be correlated with certain functions in the brain. This is a straightforward reduction of timeless meaning in a community all the way down to certain states in the brains of particular individuals.Footnote 41 But of course Davidson would have no truck with this. In the debate over the priority of thought over language, Davidson has always contended that you cannot have one without the other.Footnote 42 It makes no sense, he argues, to attribute thoughts to one who is not a language user.
6. Conclusion
I conclude that although it is tempting to think that Davidson had changed his mind on some of his central theses in the philosophy of language in his later career, this temptation should be avoided. Although it is certainly true that in Language, Truth and History he does make more reference than before to the role that intentions play in the determination of meaning in the interpretative context, especially the reflexive intentions outlined in Grice’s work, it is unfair to suggest that he has abandoned the main tenets of the Davidson programme as it was first thought of in “Truth and Meaning.” According to that programme, we cannot ask for the meaning of a word or a sentence outside of asking for a meaning theory for that language, a holistic theory that exhibits the recursive and compositional features of language. It is certainly true that the reflexive intentions are a necessary feature to determining what someone means by what they have said, but we cannot define meaning in terms of those intentions. In sum, that is the major divide between Grice and Davidson, and the major divide between formal semanticists and communication-intention theorists in general.
