page 3 of Perfect Speaker Theory

that whenever an associated material conditional is false, its associated natural language conditional is false also, and that there is a problem about what to say of natural language conditionals whose associated material conditionals are true but in which it is obviously the case that the antecedent states no ground for acceptance of the consequent, for some such reason as, say, sheer irrelevance of one to the other.

The PST offers the following way with the matter. For a PS, the governing question concerns what the point is of choosing to say things one way rather than another. Consider the conjunction and disjunction cases. Let us accept that the most natural construction to place on someone's saying 'he jumped into the swimming pool and put on his trunks' is that the circumstances were such that the man referred to put on his trunks in the water. And let us likewise accept that if someone says 'he's either in Austria or Switzerland' that the speaker does not know in which of the two countries he is to be found. In the first case, if the reverse temporal order were meant, the situation would have to be regarded as non-standard, or the speaker as not fully competent. Either way, the point of the utterance is at risk of being obscured. In the second case, if the natural implication is false, a different explanation of the speaker's point offers: the speaker dissembles, jokes, or something like. But both in the account to be given of the natural thing to say about what the speaker intends to convey, and in the account to be given of the ways things can go wrong or differently, the key is the point of the utterance. A PS by his rules seeks to convey exactly his point, and so if he chooses to say 'he jumped into the swimming pool and put on his trunks' rather than 'he put on his trunks and jumped into the swimming pool,' or 'he put on his trunks while jumping into the swimming pool,' or any other variation of the temporal relation between the jumping-in and putting-on events, then that is what he means to say. So a PS means the standard implications of saying something a certain way to be present in saying it that way.

The problems (or supposed problems, as this intuitive move suggests) of the divergence between the natural language and formal cases are on this view purely an artefact of taking too seriously the syntactic similarity of 'and' to ampersand, 'or' to the 'if–then' to arrow. The meanings of the second in each pair are determined by their truth-tables, that is, by purely material considerations. But those of the first in each pair cannot be explained without reference to the pragmatic features which determine a speaker's choice of just this rather than that way of saying something: which is to say, his point. This is revealed simply enough by answers to the question, 'why would a PS say it that particular way?'

So the observation is that application of conditions (1) and (2) (the full and exact expression conditions) make choice of a way of saying something wholly deliberate, so that every natural construction to be placed on saying it that way is what saying it that way is intended to convey, so if – and we accept they do – natural language conjunctions, disjunctions and conditionals carry the implicatures identified, then if a PS did not mean them to be taken in what he says, he would say it otherwise. (He would, say, qualify appropriately).

The same analysis in terms of point reduces the problem in the second case I consider, again briefly, namely that of what might be called condition-failure, although it turns out, as I shall claim, that this is a misnomer. The debate in Austin, Searle and Grice is as familiar as the foregoing, and needs as little setting up here as it did. The problem concerns what account should be given of what is implied by saying that–to take two familiar examples – someone omitted to do something or that someone tried to do something. Using those familiar examples from Grice, it is natural to take it that the sentence 'A omitted to turn on the light' implies that A might have been expected to do so, and 'A tried to turn on the light' that some difficulty afflicted A in this matter, and indeed, most likely, that he therefore failed to turn on the light. So it is natural to take as a condition of the appropriateness of saying 'A omitted to turn on the light', rather than saying he did not turn on the light, that in some way there was an obligation on him to do so which he did not fulfil, or that he had intended to do so but forgot, or chose not to, or the like. If none of these things are the case, then it is inappropriate to use the verb 'omit'. If therefore someone says that A omitted to turn on the light, an audience is thereby licensed to make the appropriate inference. Mutatis mutandis the same applies to the 'tried' case. Our inclination would be to put the point by saying that it is a condition of the use of 'omitted' and 'tried' in these cases that the implications hold.

The problem as standardly conceived relates to the truth-evaluation of sentences for which such conditions fail. Suppose it is false that there was an expectation that A would turn on the light, and A did not turn on the light. Then is one to say that the sentence 'A omitted to turn on the light' is false or truth-valueless? Well, consider the PS. According to his rules his choice of ways of saying what he does is governed by the requirement to make his point explicit. If it is standard to use 'omit' and 'try' in given cases because the point of doing so is given by these conditions for their appropriateness, then only if they obtain would the PS use them. This obstructs the alleged problem by stressing the directionality of the dependence: to use 'omit' or 'try' is to say that these conditions obtain. Here I shall restrict the notion of 'saying that' to: expressly conveys. ('Saying that' is not coterminous with asserts – for a liar 'says that' but does not assert – and we must be reasonable about what the utterer intends by way of implication of what he says: he does not consciously have to imply every implication of what he says, given that their number might be unmanageably large). The point of an utterance using these locutions is not given by contraposing, although what the contrapositive says is true (that if not-X then not-A omitted): so in determining choice of expression a PS, as the speaker who is expressly mindful of what one is taken to be saying in saying something in this particular way, uses the expressions in question only if he means to say that the conditions obtain. To say that 'A omitted to do X' is to say that (giving this idiom full weight) there was an expectation that he would. If there is no such expectation, there is at best no point in using 'omit', and at worse it deceives the audience. For the PS, accordingly, the problem cannot arise. I take this to mean that there is no problem and that a moral offers: that the property of evaluation is not truth-value but 'appropriateness', understood as parasitic on point.

The third case is also a straightforward one for the theory: it shows that the appearance of a difference between two kinds of uses of definite descriptions is an artefact of an underlying problem-generator which the theory exposes. Recall the familiar examples: 'Her husband is kind to her', where the man is not her husband, and 'the man drinking champagne is happy tonight', where the man is drinking fizzy water. And we are familiar with Donnellan's view. Now consider the PS in these cases. He intends to say something about someone, and in the example cases what he wishes to say is that someone is kind to her or is happy tonight. In order to fulfil this intention he has to pick out the target of his remark for his audience. The possibility that a description which he believes applies to the referent might get the audience's attention to the target without in fact being true of it is one which falls within the range of epistemic defeats to which the PS is required by his rules to be sensitive and, where doing so is germane, to articulate sensitivity. So he means, and therefore might say, 'The man whom I take to be her husband', 'the man whom I take to be drinking champagne'. Now the descriptions refer by attributing something to the referent by means of which the audience can pick him out. In such practice all descriptions are thus attributive–they attribute to the referents they pick out a property they are believed to have on the basis of which they are successfully identifiable. For this latter purpose, it is sufficient that on the occasion of use there is a presumption shared by speaker and audience that the belief could be true. This is different from Searle's account, given in terms of aspects.

....4/