² By an intensional context, I shall here mean any sentential context generated by a sentence of the form ‘X F that p’, where X is a grammatical subject and F is either a psychological verb such as ‘believes’, ‘hopes’, ‘fears’, etc., or a speech act verb such as ‘says’, ‘denies’, ‘explains’, etc. I shall speak of the psychological verbs here as propositional attitude verbs, and the intensional contexts they generate as propositional attitude contexts.

³ I here ignore forms of non-linguistic meaning such as Gricean nonnatural meaning or speaker meaning.

⁴ Meaning (amongst other things) either a certain type of fortified dessert wine or a town or place by navigable water where ships can load or unload.

⁵ See Frege, G. ‘On Sense and Reference’, Translations from the Philosophical Writings of Gottlob Frege, Geach, P. T. and Black, M. (eds. and trans.) (Oxford: Basil Blackwell, 1952)Google Scholar; and Evans, G., The Varieties of Reference (Oxford: Oxford University Press, 1982)Google Scholar, ch. 1. Although Frege applied his model quite generally, I here restrict myself to applying it to singular terms.

⁶ In discussions of context-sensitive terms, it is also important to distinguish what Kaplan, Davidcalls character and content (see e.g. his ‘Thoughts on Demonstratives’, Demonstratives, Yourgrau, P. (ed) (Oxford: Oxford University Press, 1990)Google Scholar. Abstracting from his particular semantical framework, we may view the character of a term as consisting of the general, context-free rules governing its use, while its context-specific content will be what it means in that context, on that occasion of its use. My concern is primarily with the latter—though of course an expression may be ambiguous in character as well as in content. (Thus the pronoun ‘she’, which is sometimes used in accordance with a convention to refer to a previously mentioned female, and sometimes in accordance with a convention to refer to an unspecified human being, is ambiguous in character.)

⁷ This is obviously not meant to be exhaustive. For example, some expressions, notably the logical constants, are arguably best elucidated in terms of their inferential role, which is perhaps most appropriately specified in terms of their introduction and elimination rules, or their incompatibility rules. And expressions may be correspondingly ambiguous by virtue of having different inferential roles; for examples, see Williams, S. G. ‘Ambiguity and Semantic Role’, Identity, Truth and Value: Essays for David Wiggins, Lovibond, S. and Williams, S.G. (eds) (Oxford: Blackwell, 1996)Google Scholar.

⁸ It may be that singular senses are not always wholly conceptual, but involve, at least in part, a form of non-conceptual content. I do not subscribe to this view myself. But I do not see any great obstacle to replacing conceptual content by a more general notion of conceptual or non-conceptual content in most of what follows. For a general defence of the conceptualist position, seeMcDowell, J., Mind and World (Cambridge, Mass.: Harvard University Press, 1994Google Scholar) and Brewer, B., Perception and Reason (Oxford: Oxford University Press, 2000)Google Scholar.

⁹ ‘England won the Test Match’ would be an example of the whole representing the part, and ‘All hands on deck’ of the part representing thew hole.

¹⁰ In the sense roughly of what one plays, not in the sense of a kind of animal hunted for food or sport.

¹¹ If an adequate, non-inductive definition of ‘+’ is possible, then it may or may not express an essentially disjunctive concept.

¹² The Aristotelian echo is deliberate, but I do not commit myself to Aristotle's own account of ‘account’.

¹³ The idea of a family resemblance term comes from Wittgenstein, L., Philosophical Investigations, Anscombe, G. E. M., (ed. and trans.) (Oxford: Blackwell, 1967)Google Scholar, sect. 66. It is often taken to be a condition on something's satisfying such a term that its characterizing disjunction be open-ended; cp. the examples of game and perhaps number above. For reasons of taxonomic simplicity, I do not insist on any such restriction here.

¹⁴ Cp. ‘3 is an odd number and so is the set consisting of the empty set’.

¹³ Or ‘pros hen’ homonyms, as Aristotle called them, homonyms directed towards one thing. The term ‘focal’ we owe to Owen; see Owen, G. E. L., ‘Logic and Science in Some Earlier Works of Aristotle’, Logic, Science, and Dialectic, Nussbaum, M., (ed.) (London: Duckworth, 1986)Google Scholar.

¹⁶ Sometimes I speak of the terms being the focus and the satellites, and sometimes of the concepts recorded by those terms; no confusion, I think, arises.

¹⁷ Cp. also ‘My body is healthy and so is my curiosity’. It is worth remarking that there is scope for extending focal homonymy to include what might be called pros polla homonymy, homonymy directed towards many things. For example, consider the terms ‘headsman’ and ‘headhunter’. The former may be characterized as the head of a whaling vessel or as an executioner whose modus operandi consists in chopping people's heads off; while the latter may be characterized as someone who collects heads or as someone who finds heads (i.e. people) for jobs. In these cases, the prefix ‘head’ is focally related to different, but metaphorically or analogically related uses of ‘head’: head as part of body, head as top of an organization, and head as representative of the whole body. This would be an example of pros tria homonymy, homonymy directed towards three things.

¹⁸ See Eddington, A., The Nature of the Physical World (London: Dent, 1928)Google Scholar and Stebbing, L. S., Philosophy and the Physicists (Harmondsworth: Penguin, 1944)Google Scholar, ch. 3.

¹⁹ Notice that it is not plausible to treat the term ‘solid’ as univocally expressing a relation between objects and degrees of precision. For whatever conception of solidity is recorded by a true sentence of the form ‘X is solid’, it remains intuitively the case that ‘solid’ expresses an intrinsic property of X. So although there may well be such a relation, it will at best be an abstraction based on the ideal use and the particular approximations to the ideal use that are fixed, however vaguely, by the context.

²⁰ See Lewis, D., ‘Scorekeeping in Language Games’, Collected Papers, I, (Oxford: Oxford University Press, 1983).Google Scholar

²¹ Cp. Frege op. cit., 56–7. As Frege himself does in the passage cited, I take for granted that the reference of a (non-empty) name is what it is a name of. For a brief discussion of the kinds of ways of thinking that may be wholly or partially constitutive of the sense of a name, see sect. 16. Note also that I sometimes speak of (identifying) information concerning the referent rather than ways of thinking of it.

²² For Davidson‘s account, see ‘On Saying That’, Inquiries into Truth and Interpretation, (Oxford: Oxford University Press, 1984)Google Scholar. The suggestion that the demonstrative may be taken to refer to the Fregean thought expressed by the second sentence is a variant on a suggestion made by Boër, S. and Lycan, W., Knowing Who (Cambridge, Mass.: MIT Press, 1986).Google Scholar

²³ In such a framework, a clause for ‘believes’ would have to parallel those for standard sentential operators such as ‘not’, and would be something like:

∀^S∀^t∀^a[[t refers to a→ ‘ believes that S’ is true iff a believes that S is true];

but this is obviously false. A truth theory of sorts could of course be constructed, using an axiom like:

∀^Sp∀^t∀^a[[S is true iff p & t refers to a] → ‘t believes that S’ is true iff a believes that p],

where the underlined quantifier is substitutional and the associated substitution class sufficiently large to prevent the axiom from being falsified by linguistic impoverishment. But with the resources of substitutional quantification available, we may just as well use as an axiom something like:

∀Sp∀ta[[S means p & t means a] → ‘t believes that S’ means a believes that p],

in which meaning itself is taken to be the basic semantic notion, and the underlined quantifiers are again substitutional. (For details, see Davies, M. K., Meaning, Quantification, Necessity (London: Routledge, 1981)Google Scholar, esp. chs. 1 and 5.)

²⁴ The present proposal does of course require a general treatment of the ways in which thoughts, and their constituent senses, approximate to one another. And given the variety of ways of specifying the meanings of different types of expression, we may expect a similar variety of ways of specifying sense approximation. In the examples I give, however, it should be clear roughly in what relevant approximations consist. Notice that I construe ‘X believes something approximating to the thought that p’ more or less as follows: there is some thought to the effect that q that approximates to the thought that p, and X believes that q.

²⁵ The assumption is required, since (2) may still be strictly true. Whether it is will depend upon how good the mechanic's conception of a sump really is.

²⁶ See Kripke, S. A., ‘A Puzzle About Belief’, Propositions and Attitudes, Salmon, N. and Soames, S. (eds) (Oxford: Oxford University Press, 1988)Google Scholar.

²⁷ Though not enough for him to realize that ‘Londres’ is the French for ‘London’.

²⁸ Faced with a divergence between belief content and corresponding speech act content, certain philosophers are apt to introduce the distinction between the ‘narrowly psychological’ content of the belief states, and ‘world-involving’ content of the speech acts. The problem with this response, however, is that in the version of the example according to which Pierre is put off by different pronunciations of ‘Buckingham’, etc., even the narrow psychological states should on the face of it be inconsistent. The picture I ultimately want to endorse does not require this distinction: the contents of both the belief states and the speech acts can both perfectly well be ‘world-involving’.

²⁹ Henceforth, I shall omit the qualification ‘vision-based’.

³⁰ See Evans, op. tit., 89–92.

³¹ Grasp of some demonstratives of the form ‘that F’ may also require one to know that their referents are F. I doubt this applies to ‘that sheep’, however, since one could perfectly well grasp an utterance of ‘That sheep is frisky’, even if the utterer were in fact referring to a goat; but let us assume in any case that A has not forgotten that the referent of the first use of ‘that sheep’ is a sheep.

³² Evans, op. cit., ch. 6, esp. sect. 6.4. I here take egocentric location to allow (as in this case) that the specification of the place may be relative to more than one person. Thus each of A and B is interested in the location of the sheep relative to them. For a detailed discussion of further complexities relating to egocentric location, see Brewer, op. cit., ch. 6, who differs from Evans on a number of matters of detail not crucial to the present paper.

³³ This means that ‘A is unable to locate x egocentrically’ is implicitly intensional. (Otherwise, A would still be able to locate the referent of the first demonstrative egocentrically, since the referents of the demonstratives are the same, and he can certainly locate the referent of the second egocentrically.) To bring out the intensionality, we may say, as a first approximation, that for A to be unable to locate the referent of the first demonstrative egocentrically, there must be no place specified egocentrically as s (say), such that A knows that the sheep he was referring to is at s.

³⁴ Bill Brewer speaks of the perceptual experience of an object as ‘display [ing its] spatial location’ (op. cit. 187). With this in mind, we may take the sense of a perceptual demonstrative to consist precisely in a way of thinking of its referent that displays (at least potentially) the location of its referent egocentrically.

³⁵ A second suggestion would be the location of the referent in one's personal history of encounters with things, a proposal that Evans makes in connection with recognitional demonstratives; see Evans op. cit. 299–301. However, whether or not this is a viable proposal for such demonstratives, I doubt that an ability to identify this feature is necessary for understanding names. I may have forgotten every occasion when I encountered London, but still be able to understand the name ‘London’.

³⁶ See Evans op. cit., ch.ll, esp. sect. 11.2. Henceforth, I shall use the expression ‘name-using practice’ as shorthand for ‘proper-name-using practice’.

³⁷ Op. cit., 376.

³⁸ Note that although initially the recognitional capacities exercised by the producers will doubtless enable them to place their encounters with the object in question in their own personal histories, the ability to do so need not be retained. As we observed in fn. 35, people can understand a name without being able to remember any of the occasions when they encountered its referent. This negative aspect of understanding a name is also important for the communality of the sense of a name; see sect 16.

³⁹ This is what is established by Kripke's examples of Feynman, on the one hand, and Gödel and Schmidt, on the other; see Kripke, S. A., Naming and Necessity (Oxford: Blackwell, 1980)Google Scholar, Lecture II. See also Evans, op. cit., ch. 11, where he repeatedly stresses the dangers of confusing understanding a name with being able to use it successfully to refer. Pace Evans, however, it is not clear that a speaker can refer to something only if ‘he manifes[s] which name-using practice he intends to be, and to be taken to be, participating in’ (op. cit., 384). For it is surely possible that someone could succeed in referring to something even if he‘s forgotten which practice he’s participating in, provided that his audience realizes which it is. (Imagine someone who has been told to inform a certain audience that NN is in town. By the time he comes to blurt this out to the relevant audience, he may have forgotten not only who NN is, but also who told him to relay the message or even whether it is a message at all. But the audience can still know perfectly well which practice is being manifested by his utterance.)

⁴⁰ It is therefore the feature that is at least roughly analogous to the egocentric location of the referent when the referring expression is a perceptual demonstrative.

⁴¹ But why are the ways of thinking and the name-using practice separated? If I call something ‘NN’, am I not thinking of the referent of ‘NN’ as being called ‘NN’? Yes; but it is still important to separate them since being called ‘NN’ isn't a recognitional way of thinking of the referent. And with the exception of being called ‘NN’, it is those ways of thinking, at least in the phases of a name-using practice when the input of producers remains central to the practice, which are the partially sense-constituting ways of thinking of the referent. Or so I shall suggest in sect. 16. In the late phase of a name-using practice, however, there seems no reason not to treat being called ‘NN’—or more strictly being actually called ‘NN’— as just one more way of thinking of the referent amongst the rest (see sect. 16 again).

⁴² Op. cit., 393.

⁴³ London was once known as the Great Wen.

⁴⁴ Clearly, someone who is aware of both pronunciations of ‘Newcastle’ will not, on hearing that 'Newcastle is F, reason that since ‘Newcastle is New’castle, New'castle is F. In the language of certain theorists, there is, for that individual, just one ‘dossier of information’ associated with both pronunciations.

⁴⁵ At this point, someone might suggest that we try to solve the sheep puzzle in a similar way, by taking the sense of both occurrences of ‘that sheep’ to comprise the visual presentation of the sheep and the salient ‘perceptual-demonstrative-using practice’ (and thereby dispensing with its egocentric location). However, although there may be other expressions— natural kind terms and technical terms are the obvious ones that come to mind—for which the strategy of introducing expression-using practices may be appropriate, there is an important difference regarding one-off referential devices such as perceptual demonstratives. For in general it will be the fact that the visual presentation of the referent of the demonstrative displays the referent's (egocentric) location to someone that will enable him to know which demonstrative-using practice he is participating in. But there is no such feature for names. There is no single non-linguistic feature coordinate with the ways of thinking of the referent of a name which will enable someone to know which name-using practice a user is participating in. The name-using practice is fundamental.

⁴⁶ Just as the thoughts they have concerning the referents of the names will be approximations to what the producers would think.

⁴⁷ Though with the caveat mentioned in fn. 38 concerning the corresponding recognitional capacities.

⁴⁸ The communality of sense also highlights the importance of the previous footnote. A producer may sometimes know when another producer encountered the referent of a given name; but this will hardly be typical.

⁴⁹ Except when the point of the expression would otherwise be undermined; cp. the first person pronoun.

⁵⁰ Though it would also have the consequence that those who cannot see themselves as others see them would not have a proper grasp of their own names.

⁵¹ To emphasize the separation between the two groups, let us suppose that the mature dog was not recognizable as being same dog as the puppy— perhaps it was only identified by a nametag—and that no one was aware of the dog as it matured so that no appeal can be made to ‘expert’ producers, who associate both wedges of information with the name.

⁵² I treat ‘being called “NN”’ as being cognate with ‘NN’; cp. sect. 3.

⁵³ Whether or not such terms are ambiguous is something I remain unsure about. To say, ‘X believes that a is F and so does Y’, where X is a member of the first group, and Y a member of the second, does not sound odd. But to avoid condemning each group to a necessary ignorance of the senses of the names they use, it might still seem proper to view the names as ambiguous. The above sentence would then have to be understood in more or less the way we eventually interpreted (3). That is to say, when precision matters, it would have to be rewritten as ‘X believes something approximating to the thought that a is F and so does Y’, where the sense of ‘a’ is (say) that of the producers in the group to which the utterer belongs. In cases where the overlap in recognitional information is greater, however, we would probably have to allow that the name is univocal but not fully understood by either group.

⁵⁴ Notice that some consumers may have greater expertise than others— scholars, for example. So deference, this time amongst consumers, may even be appropriate in the late phase of a name-using practice. It is also worth reiterating that in the late phase a specification of the practice can figure as a way of thinking of the referent of the name on a par with other ways of thinking of it; cp. fn. 41.