Semantics, 
Pragmatics,
Donnellan and Kripke⁽¹⁾

Speaker's and semantic meaning

The distinction between speaker's meaning and semantic meaning is not new. It has its roots in J.L. Austin's work and a classic clarification in P.F. Strawson's important article "Intention and Convention in Speech Acts". It has been worked out in almost terrifying detail by Grice and his followers (e.g. Schiffer and Bennett). Kripke himself acknowledges his debt to Grice (p.262), but his own very simple and sketchy account of the distinction as applied to reference does not reflect Grice's subtle work. This would not matter if a cruder distinction were sufficient to do the work Kripke wishes it to perform. But, if we are to use the apparatus of speaker's and semantic meaning to shed light on the relevance of Donnellan's distinction to semantics, subtlety is sorely needed. God, as usual, is in the details.

Kripke wishes to distinguish "between what the speaker's words meant on a given occasion, and what he meant, in saying those words" (p. 263), and the example he gives is of one burglar saying to another "The cops are around the corner", where the words mean that the cops are around the corner, but the speaker means "We can't wait around collecting any more loot: Let's split!". Kripke clearly wants to give us a notion of speaker's meaning which is largely independent of the semantic meaning of the words used. But even if we agree, as Donnellan does, that speaker's referring intentions determine both whether a description is used referentially or attributively, and what the referential referent is, it does not follow that speaker's meaning in this sense is radically independent from semantic meaning⁽²⁾.

As a first step towards getting clearer about the notion of meaning that is in dispute between Donnellan and Kripke, I want to take a brief look at the variety of meanings determined in part by the speaker's particular intentions, that may be distinguished from the semantic meaning of a given sentence uttered on a particular occasion. Grice himself distinguishes "four main forms of meaning-specification":
 
 

next page

  contents
 

(a) Timeless meaning - the meaning a sentence-type has in virtue of the (conventional) semantical rules of the language.

(b) Applied timeless meaning - the timeless meaning the sentence-type expresses on a given occasion (disambiguating, setting the level of vagueness, etc.)

(c) Utterance-type occasion meaning - what the speaker meant the utterance to express

(d) Utterer's occasion meaning - what the speaker intended to say (in a specific sense of "say")

An illustration might be useful here. Sandra says:

(1) I was glad we didn't have the cold fish for dinner today.

The sentence she uttered has an ambiguous Timeless Meaning: it either says

(2) I was quite glad today to have avoided the cold fish for dinner (at some unspecified, but understood, time).

or

(3) I was quite glad [at some unspecified, but understood, time] to have avoided the cold fish for dinner today.⁽³⁾

Now, let us suppose that Sandra intended the first of these conventional meanings which then is the Applied Timeless Meaning of (1).

But actually, what they had for dinner on that occasion was roast beef and, in fact, the semantic meaning of (2) is further ambiguous because of the phrase "to have for dinner" which can mean either to eat for dinner or to invite for dinner. The meaning Sandra intended here is the latter. So the timeless meaning is four-way ambiguous and we have isolated, by consulting Sandra's intentions, which sense is the Timeless Applied Meaning:

(4) I was quite glad today that we did not invite the cold fish for dinner [at some unspecified, but understood, time].

But, unless Sandra is a carp herself, it is unlikely she meant the literal meaning. What she meant, rather, was the metaphorical meaning, roughly:

(5) I was glad today that we did not invite that passionless person for dinner.

That is what Sandra meant the utterance to express on this occasion, it's the Utterance-type Occasion meaning. Of course, Sandra was being sarcastic and, in fact, she was not glad at all, nor did she think Frank, whom her husband had called a "cold fish", was a passionless person. What she meant, in fact, was:
 
 

previous page

next page

  contents
 

(6) I was sorry that the amusing Frank was not invited to the dinner in question.

That is the Utterer's Occasion Meaning.

Now, there is much to get clearer here. For example, the difference between applied timeless meaning and timeless meaning tout court seems to depend on ambiguity. On this picture, a single sentence might have an ambiguous meaning. But one might argue, quite plausibly, that just as "bank" and "bank" are two homophonic word-types, so (1) does not have a timeless meaning unless it is specified which of the four homophonic sentences it actually is. In other words, it might seem that there is no such thing as unapplied timeless utterance-type meaning. It might seem that not much hangs on this, but in fact it is an important issue. For the moment, we might help Grice's point by noting the phenomenon of vagueness. The sentence

(7) I prefer the small salsify.

is vague because "small" is. The range of sizes of salsify in question is, of course, fixed in the context of utterance. But it would be implausible to argue that, absent such specification, the sentence lacks timeless meaning.

Perhaps more importantly, there are two broad theoretical issues that Grice tries to deal with in his two justly famous pieces on meaning, the 1957 article "Meaning" and the Logic and Conversation lectures. The first problem, associated more clearly with the earlier article, is to give a specification of what it is to mean something by (a performance of) an utterance that is independent of the semantic meaning of the sentence used. Grice uses "utterance" in a broad sense to include any act that has (non-natural) meaning. To mean something is to produce a sound, movement, inscription etc. - an utterance for short - with the intention to communicate some information. It is the structure of the intentions to communicate by producing an utterance that is characteristic of meaning. The speaker intends to produce a certain effect in the audience by means, in part, of the audience's recognition of the intention to do so. Much has been written on how to specify the intended effect⁽⁴⁾ and it is not clear that the Gricean programme can deal with the problem.

But for our purposes, the other theoretical issue, associated with the Logic and Conversation lectures is of more interest: can the timeless meaning of a sentence be specified in a way that is independent of any particular use of the sentence.⁽⁵⁾ We do not need to go into the details of Grice's views here. But we should note that this is the part of the programme Kripke is appealing to. The reason the referential/attributive distinction is irrelevant to semantics seems to be that there is a notion of timeless meaning that can be specified without reference to other, pragmatic, meanings.

But the issue is a very complicated and we should note, following Austin, that there are many things we can do with words and that it is by no means obvious where "speaker's meaning" will begin and end once we take them all into account. So Sandra uttered (1), intending the sentence to be

previous page

next page

  contents
 

disambiguated in one way, also intending to use it metaphorically, but meaning by it the very opposite (6) since she was speaking ironically. Now, her husband knows Sandra rather well and saw that she wanted to have the effect of bringing the conversation on the topic to a close. One might say, echoing Kripke's example, that the speaker's meaning in uttering (1) was, in fact:

(8) Let's not talk about this any more.

There is, no doubt, a world of difference between (1) and (6), as Kripke points out. But what about (2)-(5)? Are they all pragmatic meanings as well, or are some or all of them semantic, and how does one draw the boundary? Kripke wants a simple dichotomy between speaker's and semantic meaning. The problem is that the plethora of complications that can arise makes any such dichotomy implausible. So, for example, Sandra is using the expression "cold fish" in its metaphorical sense of "passionless person". Is this a case where semantic meaning and the speaker meaning diverge? Well, that depends on your account of metaphor. One might take the sentence to be semantically ambiguous between the literal and metaphorical meaning, or one might insist that it has but one meaning, the metaphorical one being a speaker's meaning. And so on for the other layers of "meaning", and perhaps many more on top (consider double-ironies, metaphorical uses of double ironies, etc.) But the real trouble for Kripke's interpretation comes at the very beginning of our hierarchy: how does one draw the line between timeless meaning and applied timeless meaning?

What is a semantic ambiguity?

Kripke's aim is to show that Donnellan's distinction has no relevance to semantics. Kripke seems to hold that only if the distinction reflects a semantic ambiguity, has Donnellan made his case against Russell, for only then is it relevant to semantics (p. 262). On the one hand, this seems right. After all, if all we have is different uses of the same phrase, a opposed to different senses, the difference seems to be a pragmatic one -- a difference in speech acts, Kripke would say. Indeed, it is the articulation of this view, widely shared among philosophers of language, that makes Kripke's paper so convincing. And yet, I want to argue, the view is false.

According to Kripke, and I have no dispute with this, semantics is the study of timeless meaning. As traditionally conceived, however, semantics has (at least) two roles, corresponding to two central conceptions of what the meaning of a sentence is.⁽⁶⁾ On the one hand, ever since Frege, we have considered the meaning of a sentence to be its truth-conditions (or assertability-conditions, for the less realist among us) - call it "truth-conditional meaning". On the other, we take meaning to be something that a sentence of a language has, and a competent speaker recognizes it to have, just in virtue of being a sentence of that language - call it "competence meaning". The meaning of a sentence

previous page

next page

  contents
 

in this sense is what enables it to be used in different situations in the performance of different speech-acts. Particular contextual facts serve to disambiguate sentences, fill out what is implicit in them anyway, set the limits of vagueness, but do not contribute to their meaning otherwise. In a perfect language, the contribution of context would be eliminated.

The view that there is a single theoretical construct that will serve both as competence and truth-conditional meaning is denied by Direct Reference Theory. Kaplan's distinction between the character and content of indexicals is the best known example of this denial. Indexicals, such as the first person pronoun, have a competence-meaning or character which can be found in any dictionary, but the contribution they make to the truth-conditions of the sentence they appear in - their content - depends on the context of utterance and is an entirely different sort of thing. A person, in fact.

Kripke's contention is that, unless there is a semantic ambiguity, there cannot be any relevance to semantics. The word "I" is not, on the face of it and on the most plausible accounts, ambiguous in English, certainly not when I utter "I love pressed duck" and you utter "I love pressed duck". And yet, again on the face of it and on the most plausible accounts, what I say has different truth-conditions from what you say. The simple answer is that direct reference theory and, more specifically, the account of indexicality it provides shows that there can be a difference in truth-conditions without a difference in competence meaning.

One way to put this point is that the account of indexicality has shifted the boundary between semantics and pragmatics. On the traditional picture, championed by Russell and, less obviously, Frege,⁽⁷⁾ and brought to full fruition in Quine's Word and Object, a certain purity of semantics is possible. It is acknowledged on all accounts, that most of our utterances lack truth conditions as they stand. I say: "This smoked sturgeon is a little tough today" and everyone knows which smoked sturgeon and which day I mean. In addition, they all figure out that I mean that the fish in question is a bit chewy, as opposed to difficult to negotiate with or too hard to be of use as a pillow. But the sentence specifies none of this. The reason that a semantic interpretation of my utterance is possible is that a different discipline, pragmatics, will give an account of how the context of use fills out the gaps. Pragmatics will, or at least much hand-waving says it will, provide us with an item with genuine truth-conditions: an eternal sentence, a proposition expressible in a canonical language, something that wears its logical form on its sleave. Semantics then has the task of specifying these truth-conditions in a recursive way as well as the inferential relations between my utterance and other sentences of English, also duly filled out. The messy business of pragmatics is irrelevant.

But the account of indexicality offered by Direct Reference theory necessarily denies this picture. The truth-conditional content of my utterance contains a direct reference to a piece of smoked fish, perhaps even the smoked fish itself. The words "this smoked sturgeon", as well as what is intuitively their linguistic meaning, have disappeared entirely from the truth-conditional content. But if the latter is all that semantics is concerned with, then my utterance and Susan's "That stuff, whatever it was,

previous page

next page

  contents
 

was a little tough on that day we were just talking about." are semantically indistinguishable, given the right context and, conversely, my utterance of the very same words on the next day is semantically utterly distinct from the original one. So much, then, for a recursive account of truth-conditions. Hence, what used to be clearly part of pragmatics - the way the context of utterance contributes to the determination of reference - is now necessarily part of semantics. Or, at any rate, the boundary between the two disciplines needs to be redrawn.

We can put the point in the terms articulated in the first part of this paper. On the traditional view, semantics is the study of truth-conditional, timeless meaning. It is true that some, or even most, of the sentences actually uttered by humans lack a full meaning because definite descriptions and other referring terms are incompletely spelled out, predicates vague, terms and structures ambiguous, all manner of detail left implicit. But all of these matters get cleared up once sentence is applied in a context. A different sentence could then be constructed to reflect the truth-conditional meaning, the logical form, of the sentence on that occasion. That sentence has as its timeless meaning, the applied timeless meaning of the original. The semantics of a natural language need only supply the semantics of these "eternal" sentences, since any occasion-meaning expressible by a natural-language sentence is expressible by an eternal sentence and eternal sentences either are or are translatable into natural-language sentences whose timeless meaning remains constant no matter how applied.

The insight we obtain from the current accounts of indexicals is that there are certain sentences of natural language which have a timeless meaning which is not expressible by any eternal sentence. In other words, no semantics can be written for timeless meanings alone. The timeless meaning of a sentence with an indexical has no truth conditions until applied. But the applied meaning cannot be expressed by an eternal sentence whose truth-conditions remain constant no matter how it is applied. The determination of the truth-conditions of such sentences is essentially contextual, they appear only upon application. Hence, what was regarded as a eliminable pragmatic matter - the applied occasion meaning of a sentence - turns out to be essentially semantical if semantics is to yield truth-conditions, as we think it must. The general point can be put this way as well: there are ambiguities, and other differences, that affect the truth-conditional meaning of a sentence which are not semantic, but essentially pragmatic.

Donnellan's referential/attributive distinction, and the referential use in particular, may in fact be the first articulation of this point about semantics and pragmatics. He writes:

Strawson and Russell seem to me to make a common assumption here about the question of how definite descriptions function: that we can ask how a definite description functions in some sentence independently of a particular occasion upon which it is used.(p. 282)

The context of use, primarily the intentions of the speaker, enters into the semantic evaluation of an utterance of a sentence in two different ways, at two different points. The very same sentence, say "Smith's murderer is insane", can be used either with a Russellian (Fregean) logical form of or to

previous page

next page

  contents
 

express what might anachronistically be called a "singular proposition" depending on whether the description is used attributively or referentially. This, one might say, is a pragmatic "disambiguation" of the sentence, much in the way pragmatics fixes the logical form of our utterances on the traditional view.⁽⁸⁾ If the use is attributive, then the referent is whatever satisfies the description uniquely in each circumstance of evaluation. But if it is referential, then the context of utterance, specifically the speaker's referring intention, needs to supply the referent - much in the way the context supplies the referent in the case of indexicals. What object the speaker has in mind is thus relevant to the truth-conditions of the sentence so used, ipso facto to its semantics.

This, in retrospect, is roughly the story Donnellan wants to tell, though Kripke is quite right that he wavers on the issue of semantic relevance and semantic ambiguity. In the original paper Donnellan does indeed fall short of asserting that a referentially used definite description can figure in a true statements even when it fails to pick out the intended object, or even picks out an unintended object of which the predicate is false. Consider the Linsky example Donnellan discusses (p. 302). Suppose that the sentence

(9) Her husband is kind to her.

is used referentially to assert of the lover she is walking with that he is kind, while her husband is in fact mean. What is said ought to be true if the distinction is semantically relevant. But Donnellan seems unsure about this. However, a closer reading reveals that the only reason he doesn't assert that (9) is true is that he lacks the notion of singular proposition and is thus worried about what would count as the statement which is true in such cases. After all, it cannot be that her husband is kind to her, for that is false. What Donnellan is missing, but finding his way towards, in the discussion of the case is precisely the notion of a singular proposition and an account of how we report that someone has asserted or thought one. But it is fairly clear that what he has in mind prefigures the notion of direct reference and its account of reports. He writes, for example, that if the speaker used a description referentially, "we may refer to what the speaker referred to using whatever description or name suits our purpose."⁽⁹⁾

Some Doubts

None of this shows, however, that Donnellan's distinction is relevant to semantics, or even that the referential use really does require us to re-draw the boundary between semantics and pragmatics in the same way as the Kaplan account of indexicals does. There are a couple of remaining problems and they may well undermine such a simple conclusion. It looks prima facie plausible that the referential use of a definite description requires, just like the use of an indexical, that we rely in part on contextual factors to determine the referent and thus the truth conditions of what is said. But

previous page

next page

  contents
 

pragmatics enters into the matter even before we get to this stage, for it determines whether the description is being used referentially or attributively in the first place. There is nothing analogous in the case of indexicals, we might argue, and there is no reason to take the distinction to be a semantic one rather than a pragmatic one. After all, given that "what is said" is multiply ambiguous, as Grice points out, there seems to be room left for saying that, while the meaning of the sentence in the context remains fixed, it is commonly used to express two different sorts of speaker's meanings.

This line of argument, which is reminiscent of the line Kripke wants to take, holds some promise. But it is only a direction of argument, not an argument in itself. There is, after all, some evidence that this "pragmatic" ambiguity has semantic import. What is asserted in the two cases does differ in semantic structure, and we have well worked-out theories of what those structures are - in Russell, Frege and followers in the one case, in Kaplan, Kripke, Donnellan and so on in the other. Furthermore, it seems to me quite hard to draw the line between what is asserted in the referential case and what is asserted using an indexical or proper name: if the latter is not "merely" speaker's meaning, it is hard to see why the former one should be. As for the fact that the disambiguation is itself pragmatic, an argument is needed that it is, therefore, irrelevant to semantics and not on a par with, say, the distinction between using "this" as a pronoun of laziness and using it as an anaphoric pronoun.

A different approach to showing that Donnellan's work lacks semantic import is to concentrate on the fact that the final arbiter of both how a description is being used on a particular occasion, and of what the referent is in referential cases, turn out to be the speaker's intentions. The semantic interpretation of a linguistic utterance ought to be based on more public, preferably rule-governed, features of the context. The strategy is to show that even in directly referential cases, like that of demonstratives, there is a distinction between speaker's and semantic reference. The referential use of definite descriptions could then be argued to fall squarely on the speaker's, as opposed to semantic reference side. A well-known example of David Kaplan's might help illustrate this point. Sitting in his office, Kaplan points at a picture on the wall behind him and says: "Now, there is a great philosopher." His intended referent is Rudolf Carnap, whose picture normally hangs there. But some wag has substituted a picture of Richard Nixon. Who is the referent? A plausible answer is that the semantic referent is Richard Nixon, as determined by the public performance of pointing and the conventional rules associated with it, along with the identity of the person photographed. But there is also the intended referent, Carnap. There is some question, of course, whether Kaplan has succeeded in referring to Carnap in any sense at all her, for all his intentions. Now, the contention is that the referential/attributive distinction, and the referential use in particular, tell us something about referring intentions, but not about linguistic reference itself.

The debate cannot end here, however. One can point out, on the one hand, that there is more to the referential/attributive distinction than just the intentions of the speaker: there are differences in semantic structure. On the other hand, the role of intention in determining reference is a complicated matter and it is hard to argue that it is irrelevant in the case of demonstratives or proper names. To

previous page

next page

  contents
 

flesh out the first point just a little bit, we note that at a minimum, referentially used definite descriptions are rigid, that is function as though they had Kaplan's "Dthat" operator on them, which insures that reference is fixed in the context of utterance and carried to all circumstances of evaluation. Of course, as Kripke points out, rigid definite descriptions might well be attributive nonetheless. Referential ones, however, cannot but be rigid.

But the large difference is in the way context fills out an incomplete or near-miss description. Take the martini drinker case. When the chairman of the teetotaller's union asks the famed question, the context must supply the time and the place and perhaps some other details to flesh out a uniquely satisfied set of attributes. The description is incomplete and the context of utterance must complete it - that is the traditional Russellian view Kripke is defending. But in the case of someone asking the question at a party, all that is needed is that the description point the hearer to the right person, much in the way pointing with your finger might have done, had it not been impolite. The description is no more in need of completion to get a uniquely satisfied set of attributes than a demonstrative would be.

The second point that needs to be made is about intentions. They do play a role in our interpretation of people's assertions and the question "Well, what did she mean?" is ever relevant. It is not true, however, that the only thing we have to go on in determining whether a description was used attributively or referentially are the intentions of the speaker. The linguistic and extralinguistic context of the utterance provides clues and we are generally not at a loss to determine what a speaker has meant because we lack access to the inner reaches of her mind. Donnellan himself proposes the "whoever it might be" test, and suggests various ways a conversation might go in the cases he presents. The situation here is no different from the disambiguation of "I saw the children shooting". Someone might suggest that the ambiguity is merely pragmatic, since one cannot tell from the sentence which reading it has and we have to appeal to the speaker's intentions to know which of the many readings is meant. But in fact, the context can and mostly does disambiguate, and the evidence that an ambiguity exist has largely to do with inferential relations to other propositions. So it is with the referential/attributive distinction. The debate over the distinction's place in semantics and pragmatics is a complex and subtle on and I don't believe myself to have taken it very far at all. The point I want to emphasize, however, is that Kripke's deflationary account of it does not even begin to address the issue and its consequences. The questions of the semantic relevance of the referential / attributive distinction remains open.
 
 

previous page

next page

  contents
 

REFERENCES

Almog, Joseph: Naming Without Necessity, The Journal of Philosophy Vol.LXXXIII, 1986, pp.162-185

Austin, J.L.: How to Do Things with Words, Harvard University Press, Cambridge, Mass., 1962

Bennett, Jonathan: Linguistic Behaviour, Cambridge University Press, Cambridge, 1976

Donnellan, Keith: Reference and Definite Descriptions, Philosophical Review vol. 77,1966, pp.281-304.

Donnellan, Keith: Putting Humpty Dumpty Together Again, Philosophical Review 77,1967, pp.203-215.

Donnellan, Keith: Proper Names and Identifying Descriptions, Synthese, 1970, pp. 335-358.

Donnellan, Keith: Speaker Reference, Description, and Anaphora, Peter Cole (ed.), Syntax and Semantics, Academic Press, New York, 1978, pp. 47-68.

Frege, Gottlob: The Thought, E.D.Klemke, Essays on Frege, University of Illinois Press, Urbana, 1968, pp. 507-536.

Frege, Gottlob: On Sense and Reference, Geach and Black (eds.), Translations from the Philosophical Works of Gottlob Frege, Basil Blackwell, Oxford, 1970, pp. 56-78.

Grice, Paul: Studies in the Way of Words, Harvard University Press, Cambridge, Mass, 1989.

Hahn, Martin: How Not to Draw the De Re/De Dicto Distinction, McNamara (ed.), The Logical Foundations of Cognition, Oxford University Press, Oxford, 1994, pp. pp. 325-353.

Hahn, Martin: Is There a Problem Here? Frege's Puzzle, Direct Reference and Intentional Content, unpublished manuscript.

Hahn, Martin: Donnellan and Kripke on Speaker's Reference, unpublished manuscript.

Kaplan, David: Dthat, Peter Cole (ed.), Syntax and Semantics, Academic Press, New York, 1978, pp. 221-253.

Kaplan, David: On the Logic of Demonstratives, Journal of Philosophical Logic, 1978.

Kripke, Saul: Speaker's Reference and Semantic Reference, in French, Uehling, and Wettstein, Midwest Studies in Philosophy vol. II, University of Minnesota Press, Minneapolis, 1977, pp. 259-76.

Kripke, Saul: Naming and Necessity, Harvard University Press, Cambridge, Mass., 1980.

Perry, John: The Problem of the Essential Indexical, Nous vol. 13, 1979, pp. 3-21.

Récanati, François: Direct Reference, Blackwell, Oxford, 1993, .

Russell, Bertrand: On Denoting, Mind Vol 13, 1905, pp. 479-493.

Russell, Bertrand: Mr. Strawson on Referring, Lackey (ed.), Essays in Analysis, George Braziller, New York, 1973, pp. 120-126.

Salmon, Nathan: Frege's Puzzle, MIT Press, Cambridge, Mass., 1986, .

Searle, John: What is a Speech Act? Max Balck (ed.), Philosophy in America, Cornell University Press, Ithaca, N.Y., 1965, pp. 221-239.

Searle, John: Intentionality, Cambridge University Press, Cambridge, 1983.

Schiffer, Stephen: Meaning, Oxford University Press, Oxford, 1972

previous page

next page

  contents
 

Stalnaker, Robert: Propositions, Mackay and Merrill (eds.), Issues in the Philosophy of Language, Yale University Press, New Haven, 1976, pp. 79-91

Stalnaker, Robert: Assertion, Syntax and Semantics 9, 1978, pp. 315-332

Strawson, P.F.: On Referring, Mind LIX, 1950, pp. 320-44.

Strawson, P.F.: Intention and Convention in Speech-Acts, Philos. Review, vol.4, 1964, pp.436-460

Ziff, Paul: On H.P. Grice's Account of Meaning, Analysis 28, 1967, pp.1-8.

NOTES

1. Saul Kripke's article "Speaker's Reference and Semantic Reference" [1977] is often cited approvingly and included in anthologies both for it's deflationary account of Keith Donnellan's seminal but enigmatic distinction between referential and attributive uses of definite descriptions [1966], and for the distinction promised by the article's title in which the critical account is grounded. I believe that Kripke's interpretation of Donnellan's distinction is mistaken, the arguments given in the paper inadequate, and the distinction between speaker's and semantic reference too crude and thus both uninteresting in itself and irrelevant to the referential/attributive distinction. The present paper will not attempt to support all three of these views, but only the last. The rest of the conclusions are argued for in my [1995].

2. At this point I want to remain quite vague about the notion of interdependence of semantic and speaker's meaning that is relevant. I return to the issue below.

3. We'll ignore, for the moment, the extremely relevant but complicated question whether the utterance-type timeless meaning of sentences with indexicals gives us their truth-conditions. As we shall see, my view is that is does not and that fact is crucial to understanding distinctions like Donnellan's. But we shall return to this.

4. See, for example, Bennett, Schiffer, Grice for positive attempts, Ziff , Searle (1965) for criticisms.

5. Grice, in other words, wants to both defend a meaning-nominalism and to deny one. The thesis he wants to defend is that utterer's occasion meaning is fundamental: the intensionality of language is derivative from the intentionality of thought. If it weren't for the fact that people intend to perform communicative tasks, there would be no linguistic meaning. Hence, it is essential to defend the view that complex Gricean meaning-intentions can be independent of sentence-meaning (it is not necessary for him to show that a single one of our actual intentions is uncoloured by linguistic meaning, however.)

The sort of meaning-nominalism Grice wants to deny, however, is that the theoretical construct of timeless meaning is either incoherent, or useless, or eliminable in favour of occasion-meaning, particular uses, or even idiolects. That is why he is at so much pain to distinguish conversational implicature from timeless inferential relations.

There has been considerable misunderstanding concerning the compatibility of these two views, seen for example in the responses of Ziff, Searle, and others to the meaning article. Curiously, the same incredulity does not seem apply to the holding of the contradictories of each of Grice's views together by, for example, Donald Davidson.

6. For a more detailed discussion of this issue, see my "Is There a Problem Here?".

7. Russell's account of "egocentric particulars" and the importance of the context of utterance is found "Mr. Strawson on Referring". Frege's most complete discussion of the issues is in "The Thought".

8. Even on that view, it is not true that disambiguation is the only role pragmatics plays. The phrase "the bouillabaisse recipe" is not ambiguous between all the possible "eternal" completions, nor are vague terms like "tough". Furthermore, some of the ambiguities context does fix are structural ambiguities of whole sentences or even larger segments of discourse (take the distinction between

previous page

next page

  contents
 

pronouns of laziness and anaphoric pronouns, for example) So, even on the traditional view of the relation between pragmatics and semantics, it is not the case that any semantically relevant distinction between two uses of a phrase must take the phrase to be ambiguous.

9. Kripke objects to Donnellan's discussion of what the statement asserted is and how we might report it on the grounds that "a definite description in indirect discourse is neither referential nor attributive" (p. 261) But, on Donnellan's analysis, to report a referential use of a definite description, the position occupied by the referring phrase in the report is always transparent. This agrees with the general analysis of reports of directly referential speech-acts. It may be wrong, in fact I believe that it is and argue for this in "How not to Draw the De Re / De Dicto distinction". But Kripke nowhere argues against it.

previous page

  contents