Demonstrating and necessity

Philosophical Review

Volumn 111, Issue 4, 2002, Pages 497-537

  Salmon, Nathan ^a  

^a UNIVERSITY OF CALIFORNIA   (United States)

Author keywords

[No Author keywords available]

Indexed keywords

EID: 34248738147     PISSN: 00318108     EISSN: 15581470     Source Type: Journal    
DOI: 10.1215/00318108-111-4-497     Document Type: Review

References (102)

1. In J. Almog, J. Perry, and H. Wettstein, eds., Themes from Kaplan (Oxford: Oxford University Press, 1989), 481-614.
2. Kaplan's three-tiered theory of character, content, and extension is inadequate. The eternal nature of contents-for example, the fact that a given proposition is unwavering in its truth value-argues in favor of separating the possible world of a circumstance from the time, and drawing a four-way distinction between semantic values by inserting a semantic value-what I call the content base-between the levels of character and (proper) content. Content bases of sentences are proposition-like entities except for being non-eternal. Such things are sometimes called states of affairs. I call them proposition matrices. Kaplan's notion of character is replaced by a semantic value, which I call program, that assigns content bases to contexts: Level 4:program Level 3: content base with respect to c Level 2: content with respect to c and t Bottom: extension with respect to c, t, and w The content of 'I am hungry', when uttered by me at t, is the eternal proposition that I am hungry at t, whereas the content base is a proposition matrix-a recurring state of affairs that, although frequent, is not quite eternal.
3. See my "Tense and Singular Propositions," in Themes from Kaplan, 331-92. I will ignore the need for this significant modification of Kaplan's scheme in what follows when there is no danger of any resulting serious confusion of the relevant issues.
4. More accurately, an indexical determines different content bases depending on context. See the preceding note.
5. In "Afterthoughts" Kaplan proposes replacing demonstrations with "directing intentions" (582-90 and passim). Though the distinction remains somewhat unclear, I believe that nothing said here is affected if the proposed replacement is made throughout.
6. "Der Gedanke," Beitrage zur Philosophie des deutschen Idealismus (1918), translated by P. Geach and R. H. Stoothoff as "Thoughts," in Frege's Logical Investigations (New Haven: Yale University Press, 1977). An alternative translation of the quoted passage occurs there, at p. 10.
7. Tyler Burge, "Sinning Against Frege," Philosophical Review 88 (July 1979): 398-432. Burge argues that Frege's three-way distinction is partially nonsemantic, because Frege's notion of sense is epistemic or cognitive rather than semantic. I am unpersuaded, partly for reasons to be set out shortly. Though Fregean propositions ("thoughts") are mentally apprehended objects of propositional attitude, Frege's notion of sense is no less semantic than Kaplan's notion of content or Alonzo Church's notion of sense. Indeed, the former is a good deal more semantic than, for example, Strawson's notion of the statement made in an utterance (cf. my "Two Cnceptions of Semantics," in Z. Szábó, ed., forthcoming.) The Fregean sense of an expression is precisely what, on Frege's theory, the expression (as supplemented by various contextual elements) expresses and what, in turn, determines the same expression's Bedeutung. The last is a properly semantic notion if anything is.
8. Frege's Posthumous Writings, ed. H. Hermes, F. Kambartel, and F. Kaulbach, trans. P. Long and R. White (Chicago: University of Chicago Press, 1979), 213.
9. I am indebted to observations made by Kripke, who suggested this interpretation and cited some of these points against Burge's reading of Frege in a seminar at Princeton around 1980.
10. The time of utterance would have to present itself in a particular way in order to designate itself (perhaps as the current time, the time being, or the specious present, etc.)
11. since on Fregean theory all designation is secured by means of a sense. Times of utterance, qua self-referential "expressions," would thus provide rare exceptions to the Fregean dictum: There is no backward road from designatum to sense.
12. A complication arises from Frege's explicit assertion that 'today'∼yesterday has the same sense as 'yesterday'∼today. The designated day is the same, but sameness of sense of the supplemented words would require that the sense of 'today' applied to yesterday should yield the very same value as the sense of 'yesterday' applied to today-that is, ∧today∧(yesterday) = ∧yesterday∧(today) (where '∧' is a sense-quotation mark). It is difficult (at best) to reconcile this with Frege's tendency to treat the senses of compound expressions as (metaphorically) being composed of the senses of the component expressions. (How can the sense of 'yesterday' be a component of the proposition expressed by a sentence using the word 'today'?) On the other hand, as several commentators have noted (including Burge and Kaplan), Frege's assertion seems directly contrary to his original motivation for postulating sense as distinct from designatum. But see note 14 below.
13. Burge says (399 n.) that his interpretation of Frege as contrasting his notion of sense with the properly semantic notion of linguistic meaning is further supported by the following passage from Frege's "Logik" (probably 1897). But the passage supports, and even strongly suggests, the very different interpretation offered here:
14. Words like 'here' and 'now' achieve their full sense always only through the circumstances in which they are used. If someone says 'It is raining' the time and place of utterance have to be supplied. If a sentence of this kind is written down it often no longer has a complete sense because there is nothing to indicate who uttered it, and where and when. ... [T]he same sentence does not always express the same thought, because the words require supplementation to obtain the complete sense, and this supplementation can vary according to the circumstances. (Frege's Nachgelassene Schriften, ed. H. Hermes, F. Kambartel, and F. Kaulbach (Hamburg: Felix Meiner, 1969), at 146. An alternative English translation occurs in Frege's Posthumous Writings, at 135.)
15. Burge also says (400) that his interpretation is neutral concerning whether it is the indexical expression itself (for example, the word 'that') or the accompanying circumstance (a demonstration) that actually expresses the sense that determines the relevant designatum. One way or the other, the sense associated with the indexical relative to the context varies with the context, whereas the meaning of the indexical itself remains constant in all its relevant uses. The reasoning is mistaken. There is more than one sense "associated" with the indexical relative to a context of use: there is the sense of the indexical itself, and there is the sense of the supplemented indexical. Insofar as it might be a third sense, there is also that of the contextual supplement-which, like the supplemented indexical, functions as a distinct expression from the mere indexical. It is irrelevant that different demonstrations will express different senses (as Frege undoubtedly held). Crucial to Burge's interpretation is the claim that the indexical itself ('that') does not, by Frege's lights, express a sense that remains unchanged with variations in context. But Frege is best seen as holding precisely that the mere indexical's sense remains unchanged despite changes in the accompanying contextual elements (and the senses thereby expressed).
16. Frege maintained that it is not the exponent itself (and not the word 'squared') that designates the relevant function, but the incomplete expression '-2' (likewise,'-squared'). On the interpretation suggested here, Frege saw the mere word 'yesterday' as also being incomplete, its argument place to be filled not with a syntactic entity but with the time of utterance (qua self-designating "expression").
17. Other specifications of the content, even as a function of context, do not fix the character. There is an exactly analogous distinction between a meta-linguistic biconditional in a theory or definition of truth, like ''Snow is white' is true-in-English iff 'is white' applies in English to the English designatum of 'snow'', and those special theorems called T'-sentences that appropriately fix the nonsemantic truth conditions. Kaplan represents an expression's character in his formal apparatus by the function-in-exten- sion from contexts to contents fixed by a content rule, but further remarks (e.g., "Demonstratives," 505) suggest that the character is something more like the function- in-intension expressed by the character-building content rule. For present purposes an expression's character may be identified with the meta-proposition expressed by its character-building content rule, as distinguished from the other content rules. One who does not know this meta-proposition does not understand the expression.
18. David Braun makes a similar observation, but a significantly different positive proposal, in "What is Character?" Journal of Philosophical Logic 24 (1995): 227-40.
19. Cf. "Demonstratives," 529-32, 597.
20. Kaplan does not articulate the issues concerning knowledge by description and acquaintance as I have. He sees the matter in terms of a supplemented demonstrative's potential for having a different content while retaining its character, in that the same demonstration has different demonstrata in different contexts. I think of the matter instead in terms of the descriptive manner in which the character-building content rule presents the content as a function of context. The difference between the two perspectives is subtle but significant. (See also note 28 below.)
21. To illustrate, Kaplan has introduced the name 'Newman-1' (not an indexical) for whoever will be the first child born in the twenty-second century (in "Quantifying In," Words and Objections: Essays on the Work of W. V. Quine, ed. D. Davidson and J. Hintikka (Dordrecht: D. Reidel, 1969), 206-42, at 228-29). Kaplan agrees that 'Newman-1' has no semantic potential for having a different content (unlike the corresponding ' dthat'-term), since the content is the same no matter the context. Still, its character-building content rule presents that content in a special manner (albeit not as a non-constant function of context): With respect to any context c the (English) content of 'Newman-1' is whoever will be the first child born in the 22nd century, if there will be a unique such person, and nothing otherwise.
22. The character's perspective on content underlies the phenomenon that has been called "the essential indexical" in explaining behavior by invoking indexical reports of certain beliefs or other attitudes (for example, the belief that one's pants are on fire). Cf. John Perry, "The Problem of the Essential Indexical," Nous 13 (1979): 3-21
23. reprinted in Propositions and Attitudes, ed. N. Salmon and S. Soames (Oxford: Oxford University Press, 1988), 83-101. If I am correct, however, a contextual perspective on content is quite inessential to what is semantically expressed by 'I do believe that my pants are on fire'.
24. The idea of accounting for cognitive value in terms of meaning rather than content (or the "statement" made) is found in P. F. Strawson, "On Referring," sec. 5.b, where he says:
25. [O]ne becomes puzzled about what is being said in these sentences [sentences like 'Today is Smith's birthday' and 'Yesterday was Smith's birthday']. We seem ... to be referring to the same [thing] twice over and either saying nothing about [it] and thus making no statement, or identifying [it] with [itself] and thus producing a trivial identity.
26. The bogy of triviality can be dismissed. This only arises for those who think of the object referred to by the use of an expression as its meaning, and thus think of the subject and complement of these sentences as meaning the same because they could be used to refer to the same [thing].
27. Is Frege stymied here? Perhaps not. If the problem for him cited in note 9 above can be solved, he might accommodate the alleged difference in informativeness between 'Today is Smith's birthday' and 'Yesterday was Smith's birthday' through his doctrine of indirect sense (ungerade Sinn). In fact, Kaplan's identification of the cognitive value of a sentence with its character, qua a kind of description of the relevant proposition, is highly reminiscent of Frege's notion of indirect sense.
28. Cf. my "A Problem in the Frege-Church Theory of Sense and Denotation," Noûs 27 (1993): 158-66
29. and "The Very Possibility of Language: A Sermon on the Consequences of Missing Church," in Logic, Meaning and Computation: Essays in Memory of Alonzo Church, ed. C. A. Anderson and M. Zeleny (Boston: Kluwer, 2001).
30. Kaplan sometimes use the term 'utterance' for the supplemented expression, reserving the term 'sentence' for the mere sentence. This terminological difference should not eclipse the fact that on Kaplan's view, as on Frege's, it is the supplemented sentence, not the mere sentence, that expresses a proposition when occurring in a context (see note 33 below).
31. Kaplan overstates KT2 by saying that "indexicals, pure and demonstrative alike, are directly referential" ("Demonstratives," 492). This statement gives the misleading impression that the fact that indexical words are directly referential (in Russell's terminology, logically proper names, in Kripke's, Millian) obtains somehow in virtue of their context-sensitivity. Both the statement and the misleading suggestion are refuted by the context-dependence of such non-rigid phrases as 'his wife' and 'my hometown'. Also, indexical sentences typically express contingent truths and falsehoods ('He lives in Princeton, New Jersey'), hence do not rigidly designate their truth value. By contrast, index-ical words are directly referential not by virtue of their context sensitivity, but presumably because their extensions are not secured through a semantic computation (as with definite descriptions and sentences) but given by a default semantic rule for all simple (for example, single-word) singular terms, indexical or otherwise: the Russellian rule that designatum = content.
32. In Syntax and Semantics, vol. 9: Pragmatics, ed. P. Cole (New York: Academic Press, 1978), 221-43.
33. reprinted in Contemporary Perspectives in the Philosophy of Language, ed. P. French, T. Uehling Jr., and H. Wettstein (Minneapolis: University of Minnesota Press, 1979), 383-400.
34. "Demonstratives," 490-91.
35. "Demonstratives," Ibid., 527.
36. A complex demonstrative like 'that man' may be seen as the combination of a mere demonstrative with a sortal term, standing in need of further supplementation by a demonstration that is facilitated by the sortal. Thus, an utterance of 'He is a spy' is a natural-language analogue of: Dthat [the male x: x is suspicious-looking & x is wearing a brown hat] is a spy.
37. "Demonstratives," 521-22.
38. "Demonstratives," Ibid., 527.
39. "Demonstratives," Ibid., 521.
40. "Dthat," 237.
41. Cf. the designation rule 11 of the inductive definition of extension ("truth and denotation") in "Demonstratives," at 545-46.
42. See note 12 above. The result of instantiating the meta-linguistic variable 'a' in (D) to the quotation-name of 'the suspicious-looking guy I saw yesterday wearing a brown hat' is a content rule that fixes the function-in-extension from contexts to contents, but does not express the actual character. By contrast, the "multiplied through" character-building content rule displayed in the text fixes the intended function-in- intension, thereby expressing the relevant character.
43. Cf. "Demonstratives," 520-22.
44. But see note 33 below.
45. K) itself is instead Kaplan's contextual definition of the mere word 'that'. Have I misinterpreted Kaplan? Or is his claim that (Tk) gives the character of a supplemented demonstrative an oversimplification of his view? (It does fix the character, specifying the character by description.)
46. "Demonstratives," 528-29.
47. "Demonstratives," Ibid.
48. Other writers have made this observation about Kaplan's account-for example, Howard Wettstein, "Has Semantics Rested on a Mistake?" Journal of Philosophy 83 (1986): 185-209, at 196 n.
49. David Braun presses a related point in "Demonstratives and Their Linguistic Meanings," Nous 30 (1996): 145-73, at 149-50. Braun assumes that Kaplan holds that a mere demonstrative is devoid of character while nevertheless having a univocal meaning, and objects that this is inconsistent with Kaplan's proposed identification of linguistic meaning with character. I provide an alternative interpretation in the next paragraph of the text.
50. c. Alternatively, the character might be identified with the appropriate function from singular-term characters to directly-referential- singular-term characters (for example, from the character of 'the suspicious-looking guy I saw yesterday wearing a brown hat' to that of the corresponding ' dthat'-term).
51. David Braun in "Demonstratives and Their Linguistic Meanings" (see note 31 above) makes a proposal similar to the second identification of characters mentioned above. The similarity is superficial. Braun's specific proposal has at least two significant defects. First, Braun takes the arguments of the functions he identifies with the meanings of demonstratives to be demonstrations themselves rather than their characters. This would be analogous to taking the meaning of 'the mother of' to be a function from its singular-term arguments (instead of their meanings) or the character of 'not' to be a function from sentences. This defect might be forgivable, if demonstrations are arguably part of a universal language (unlike singular terms). More important, Braun's central idea is to assign an additional kind of "meaning" to mere demonstratives: a fourth semantic value beyond character, content, and extension (of the supplemented demonstrative). By contrast, the proposal in the text (to be rejected presently) assigns a character, content, and extension to a mere demonstrative itself. The character of the mere demonstrative determines that of the supplemented demonstrative from that of a given supplemental demonstration, whereas the content or extension of the mere demonstrative determines that of the supplemented demonstrative from the content (in both cases) of the demonstration.
52. Kaplan explicitly acknowledges some of these points in "Afterthoughts," at 579-82. Discomfort over the cost of mediation seems to have prompted a disorderly retreat from KT1. Kaplan says that, precisely because the singular term is meant to be directly referential, he had intended the designating term to be simply the word ' dthat', rather than the compound expression Tdthat[the ø], and that the supplemental description [the ø] was to be merely a "whispered aside" that was "off the record" (581; Kaplan adopted these latter phrases from suggestions by Kripke and me, respectively). Since the supplemental term is no part of the term ' 'dthat', he says, as originally intended ' dthat' is not a rigidifier of something else but a term unto itself. He writes:
53. The word 'dthat' was intended to be a surrogate for a true demonstrative, and the description which completes it was intended to be a surrogate for the completing demonstration. On this interpretation 'dthat' is a syntactically complete singular term that requires no syntactic completion by an operand. (A "pointing," being extra-linguistic, could hardly be a part of syntax.) The description completes the character of the associated occurrence of 'dthat', but makes no contribution to content. Like a whispered aside or a gesture, the description is thought of as off-the-record (i.e., off the content record). It determines and directs attention to what is being said, but the manner in which it does so is not strictly part of what is asserted. . 'Dthat' is no more an operator than is 'I'. . The referent of ' dthat' is the individual described. . It is directly referential. Although Frege claimed that the context of use was part of "the means of expression" of a thought, he never, to my knowledge, attempted to incorporate "the pointing of fingers, hand movements, glances" into logical syntax. Can an expression such as the description in a 'dthat'-term appear in logical syntax but make no contribution to semantical form? It would be strange if it did. But there is, I suppose, no strict contradiction in such a language form. (581-82).
54. k) and the supplemental demonstration. In whatever sense it is true, as Kaplan says above, that the supplemental term a "completes" the character, it is equally true (if not even more so) that 'dthat' alone is incomplete without a supplemental term and that the complete term has the form [dthat[α] ] .
55. k), which Kaplan explicitly endorses (527). In fact, Kaplan's acknowledgment above of the consistency of the envisioned prospect is tantamount to an acknowledgement that there is no valid argument from the noncompositionality of content of a complete ' dthat'-term to the supplemental term's not being an essential component expression. On the contrary, the envisioned consistent prospect is the very reality Kaplan has produced with his operator. There does seem to be a kind of inconsistency-not in the operator as stipulated, but between the very two paragraphs quoted above. In fact, the very notion of a demonstrative that is on the one hand noncompound and univocal, but on the other variable in character depending on the designata of "whispered asides," is straightforwardly inconsistent.
56. The remarks in "Afterthoughts" (579-82) fail to provide a coherent interpretation of "Demonstratives." I conclude that Kaplan, on reflection, has misjudged his own original intent for 'dthat' above (and his own theory of demonstratives!) and that the theory is the one explicitly proffered in "Demonstratives" (at 521-27 and passim): that the complete term is the supplemented term comprised by the union of the mere demonstrative with a supplemental demonstration.
57. So does the content base. (See note 2 above.)
58. Cf. my Frege's Puzzle (Atascadero, Calif.: Ridgeview, 1986), especially 57-60, 87-92.
59. Performing the very same demonstration of the same object twice over in a single utterance of 'That is that' is in fact very difficult to accomplish. For convenience, I assume throughout that pointing simultaneously with both hands is a way of accomplishing this feat (though this assumption is strictly false).
60. "Demonstratives," 562-63.
61. I am thinking here of a context as the setting or environment in which an utterance occurs, rather than as the proposition, or set of propositions, assumed by all conversational participants. The case of the answering machine demonstrates that a contextual parameter need not be at the location of the context at the time of the context, since the agent of the utterance of 'I am not here now' is typically asserting a truth. Though the agent of the context of such an utterance is, in some sense, absent from the context, he or she is nevertheless playing an active, or "real," role in the context-there is an assertion in absentia by the agent-and I conjecture that it is this fact that warrants including the absent agent as a contextual parameter. By contrast, the demonstratum of a particular demonstration may be entirely passive, utterly inert, a mere demonstratum. (Thanks to Ben Caplan for forcing me to be more explicit about this matter.)
62. The pronouns 'he', 'she', and 'that' may differ in this respect from the special demonstrative 'this', for which the designatum is arguably always present in the context of use (or present by proxy?). If something closely resembling the Bare Bones Theory is applicable to 'this', it is so because of some such special restriction governing its appropriateness. (In effect, the Bare Bones Theory may mistake 'that' for 'this'. Or is it the other way around?) Contrary to Kaplan's claim (echoing Peter Geach) that anaphoric pronouns may be seen invariably as bound variables ("Demonstratives," 572). Perhaps the issue of whether the 'he' in (ii) is a bound variable is to some extent terminological. But the terminology of 'bound' and 'free' is not without constraints. If it is insisted that the 'he' is a bound variable, then what is the variable-binding operator that binds it to its grammatical antecedent? The 'his' in 'No author inscribed his book' is not a designating occurrence; it is genuinely a bound variable. By contrast, the 'he' in ( ii) designates Ortcutt. Nor is the 'he' a "pronoun of laziness" or an abbreviation for the description in (i). The speaker's suspicion is not merely a de dicto thought to the effect that whoever is a uniquely suspicious-looking guy seen the day before wearing the relevant brown hat is a spy. It is de re concerning Ortcutt: that he is a spy. All indications are that the 'he' in (ii), although anaphoric, is syntactically free, with its grammatical antecedent functioning as a kind of verbalized demonstration.
63. T in possible world cw (etc.). It is far from obvious, however, that such a restriction is desirable. Is the sentence 'That object (assuming it exists) is now being demonstrated', for example, to be regarded as true solely by the logic of 'to demonstrate'?
64. 1.') Still, the form of words evidently yields an invalid argument. Compare: 'He is taller than him, hence, he is neither shorter than nor the same height as him'.
65. Gertrude Stein on seeing her childhood town after it had been torn down: "There is no there there.
66. Braun ("Demonstratives and Their Linguistic Meanings") objects to taking demonstrations as aspects of context on the question-begging grounds that doing so obliterates Kaplan's contrast between demonstratives and the so-called pure indexicals. On the contrary, this is precisely one important reason for putting demonstrations into the context, exactly where they belong. Braun also notes that, unlike other aspects of context (for example, time and place), demonstrations are typically produced under the voluntary control of the agent and are not themselves the contents of the demonstratives they accompany. Here again, these are insufficient grounds to banish demonstrations from their proper place. Demonstrations have important features in common with such contextual aspects as time and place: they are all recognizable as features of the circumstances surrounding an utterance that fix the contents of uttered indexicals.
67. It is for similar reasons that substitution of 'Barbarelli' for 'Giorgione' fails in 'Giorgione was so called because of his size'. Substitution alters the context for the demonstrative 'so'.
68. The construction in the text raises particularly perplexing issues. Consider the following variant: (i) Consider whoever is the shortest spy in the world, (ii) he or she is a communist.
69. It seems undeniable that the speaker has asserted of the shortest spy, de re, that he or she is a communist, since the semantic content of (ii") is precisely that very singular proposition. Kaplan concludes (contradicting his earlier arguments in "Quantifying In"-see note 13 above) that a mastery of the semantics of such directly designating devices as demonstratives enables speakers to form beliefs of singular propositions, and even to gain singular-propositional knowledge a priori (for example, about the shortest spy that he or she is a spy, or about the first child to be born in the twenty-second century that he or she will be born in the twenty-second century), in the absence of any "real" connection to the object in question ("Dthat," 241; "Demonstratives," 560 n.; "Afterthoughts," 605). This conclusion leads almost directly to a form of the controversial thesis of latitu-dinarianism with regard to de re belief. But even if de re assertion (assertion of the singular proposition) is in fact accomplished through such means, it by no means follows that de re belief, let alone de re knowledge, follows suit. On the contrary, firm intuitions derived from ordinary language show otherwise. Cf. my "The Good, the Bad, and the Ugly," in Descriptions, ed. A. Bezuidenhout and M. Reimer (Oxford: Oxford University Press, forthcoming).
70. Kaplan observes that there is "a kind of standard form for demonstrations" accompanying a typical utterance of a demonstrative: such demonstrations have a character like that of a definite description of the form, the individual that has appearance A from here now, where the mentioned appearance is "something like a picture with a little arrow pointing to the relevant subject" (525-26). This is plausible. However, by building excess material into the linguistic meaning of the demonstrative, Kaplan inevitably mis-classifies some utterances of synthetic sentences as being utterances of analytic sentences, for example, 'He (if there is such a thing) has appearance A from here now'. Though this sentence is true, a full mastery of its meaning does not by itself give one the knowledge that it is inevitably true, as Kaplan's account implies. Its truth crucially depends on nonlinguistic, empirical information: that the demonstrated male appears a particular way from the speaker's perspective at the time of the utterance. This information is supplied with the demonstration; it is built into the context of the utterance, not into the expression uttered. (Cf. note 41 above.)
71. Cf. my Frege's Puzzle, especially chapters 8-9.
72. "Demonstratives," 562-63, 598.
73. See note 13 above. A name whose designation is fixed by description has a character of a rather special form. In the case of a typical name, the character-building content rule specifies the content for (every context) by name rather than by description.
74. A frequently heard objection to the hypothesis that compound expressions of a given category (for example, definite descriptions) are singular terms is that expressions of the given category can be coherently quantified into (that is, they can contain a variable bound by an external quantifier) while genuine singular terms cannot. The objection evidently originated with Benson Mates, in "Descriptions and Reference," Foundations of Language 10 (1973): 409-18, at 415
75. but has been endorsed or echoed by others (for example, Stephen Neale, in Descriptions (Cambridge: MIT Press, 1990), at 56 n. 28).
76. The objection typically relies on a α-abstraction theorem, to the effect that any sentence ø β containing a genuine singular term β in extensional position, and which is the result of uniformly substituting (β for the free occurrences of a variable a in the open formula øα, is true only if the designatum of (β satisfies øα)). (The assumed abstraction theorem is not generally stated this precisely, if it is stated at all). Mates may rely on an alternative semantic principle: that any sentence øβ of a restricted class C, and containing a genuine singular term (3 in extensional position), is true only if (β designates. The class C might exclude such problematic formulas as [β does not exist] ) The objection has been applied to complex demonstratives-for example, by Ernest Lepore and Kirk Ludwig in "The Semantics and Pragmatics of Complex Demonstratives," Mind 109 (2000): 200-241, at 205-6, 210-22, and passim (where something like the assumed abstraction theorem is explicitly applied): "It is difficult to see how to make sense of quantification into complex demonstratives on the assumption that they are referring terms. . [The abstraction theorem] renders mysterious how the material in the nominal could interact semantically with the rest of the [quantified] sentence" (205-6). "Examples of apparently coherent quantification into the nominals of complex demonstratives supply some of the most important evidence for denying that they are referring terms" (219).
77. Cf. Jeffrey King, Complex Demonstratives (Cambridge: MIT Press, 2001), at 10-11, 20-22.
78. It should be noted in response that complex demonstratives seem especially immune to this objection, since quantification into them is, at best, odd. If the open phrase 'that man she sees at the podium' is used genuinely demonstratively in 'At least one woman here admires that man she sees at the podium' (not as a stylistically altered definite description), the sentence is indeed true if and only if the relevant demonstra-tum satisfies the matrix 'At least one woman here admires x', and the objection collapses. (The example is from Lepore and Ludwig.)
79. Cf. Barry Taylor, "Truth-Theory for Indexical Languages," in Reference, Truth, and Reality, ed. M. Platts (London: Routledge and Kegan Paul, 1980), 182-98, at 195-96;
80. and Neale, "Term Limits," in Philosophical Perspectives, vol. 7: Language and Logic, ed. J. Tomberlin (Atascadero, Calif.: Ridgeview, 1993), 89-123, at 107.
81. 2', etc. The most glaring counterexample is the paradigm of an open designator: the individual variable. The objection is in fact based on an elementary confusion. Designation for an open term (whether compound or a variable) is relative to an assignment of values to its free variables. The variable ' y' is a genuine singular term if anything is. Its designatum (under the assignment of, say, David Kaplan as value) may fail to satisfy the particular open formula '∼(y)(y) is a person ⊃ xis ingenious)' (let this be βα, with α ='x') even though the sentence that results by substituting 'y' for 'x' is true-precisely because the newly introduced occurrence of 'y' is captured by the quantifier, making its value irrelevant. The mistaken abstraction "theorem" can be corrected, and even generalized: An assignment s of values to variables satisfies a formula øα [of the restricted class C] containing a free occurrence of a singular term (β in extensional position, and which is the result of uniformly substituting free occurrences of (β for the free occurrences of a variable α in øα, if and only if the modified value-assignment s' that assigns to a the designatum of (β under s, and is otherwise the same as s, satisfies øα).
82. This corrected version effectively blocks the objection. (There is likewise a corrected, generalized version of Mates's apparent assumption: An assignment s of values to variables satisfies a formula øβ) [of the restricted class C] containing a free occurrence of singular term (β in extensional position only if (β designates under s.) Cf. my "Being of Two Minds: Belief with Doubt," NoÛs29 (1995): 1-20, at 18 n.) 26.
83. By stipulation, ' zat'-terms are genuine singular terms. Their stipulated content rule (Z) allows for the possibility of quantification in. (See the previous note.)
84. Stefano Predelli, in "Complex Demonstratives and Anaphora," Analysis 61 (2001): 53-59, challenges those who deny that complex demonstratives are compositional with regard to content to explain how the anaphoric pronoun 'her in 'That man talking to Mary admires her' (uttered while pointing to one of several men talking to Mary) obtains its content. It is tempting to suppose that any anaphoric pronoun occurrence whose antecedent is a singular term simply inherits as its content the very content contributed by its antecedent to the content of the sentence in which the antecedent occurs. But according to (T), the antecedent term in this case contributes no component to the content of the complex demonstrative in which it occurs.
85. In response I note that the naive rule of content inheritance is falsified in cases in which the antecedent is a singular term that is not directly referential, as perhaps in 'The number of planets is such that, necessarily, it is odd' and 'Ralph believes of the man seen at the beach that he is a spy'. If the naive rule were correct (and if, contrary to Russell, the definite-description antecedents are singular terms), these sentences would be de dicto rather than de re. A more promising rule of anaphora-applicable even to anaphoric pronouns whose antecedents are singular terms that are not directly referential-is that a simple (non-reflexive) anaphoric pronoun occurrence whose antecedent is a singular term, if it is not itself a bound variable, typically takes as its content the object customarily designated by its antecedent. There is no requirement that the antecedent contribute its customary content to the content of the sentence in which the antecedent occurs. Although this rule is also subject to counterexamples, it is applicable to a significantly wider range of cases than the naive rule of content inheritance and it seems likely that some restricted variant is correct. Consider: 'That man talking to the actress honored here tonight admires her'. Although I hold the description 'the actress honored here tonight' does not contribute its customary content to that of the sentence in question, and instead merely contributes toward a restriction on admissible contents for the complex demonstrative, the description itself has a customary designatum (assuming it is a singular term), and it is that customary designatum, though she makes no appearance in the content of the demonstrative itself, that the anaphoric pronoun takes as its content.
86. K) that "obvious adjustments are to be made to take into account any common noun phrase which accompanies or is built-in to the demonstrative" ("Demonstratives," 527). Kaplan is interpreted as incorporating condition (ii) by Emma Borg, "Complex Demonstratives," Philosophical Studies 97 (2000): 229-49, at 242, where a designation rule entailed by my content rule (T) is defended at some length.
87. A similar designation rule, though couched within the Bare Bones Theory, is proffered by David Braun, "Structured Characters and Complex Demonstratives," Philosophical Studies 74 (1994): 193-219, at 209.
88. Whereas the mere complex demonstrative 'that' NP is devoid of character, content, and designatum, the content of the completed expression'that' NP α is defined to be the demonstratum of α (in the context), if there is a unique such demonstratum and NP applies to it (with respect to the context), and to be nothing otherwise.
89. In the sentence 'If there had been an atheist elected to the U.S. Senate, then that Senator's atheism would have been concealed during the political campaign' (on its most natural reading) the phrase 'that Senator' is evidently not correctly formalized using ' zat'. Yet it is a rigid designator. The sentence seems to have a form something like that of 'For every possible individual i, if i had been an atheist who was elected to the U.S. Senate, then i's atheism would have been concealed during the political campaign'. Though not a demonstrative phrase, the variable ' i' is a rigid designator of its value under any value-assignment. Simple individual variables are rigid designators par excellence. (By contrast, see note 38 above.)
90. The same remark applies to analogous bound-variable uses of pronouns ('., then he would have concealed his atheism .'; cf. note 49 above). Michael McKinsey, in "Mental Anaphora," Synthese 66 (1986): 159-75, uses an example like the following to argue that such pronouns are not rigid (161): An atheist was once elected to the U.S. Senate, but his atheism had been concealed during the political campaign.
91. According to McKinsey, the pronoun designates different possible individuals with respect to different possible worlds-to wit, whoever in that world is an atheist elected to the U.S. Senate. The argument is echoed by Scott Soames, in his review of Gareth Evans's Collected Papers, in Journal of Philosophy 86 (1989): 141-56, at 145
92. and endorsed by Stephen Neale, in "Descriptive Pronouns and Donkey Anaphora," Journal of Philosophy 87 (1990): 113-50, at 130, and again in Descriptions, at 186. It assumes, following Evans, that such pronoun occurrences (so-called "donkey" pronouns) are unbound singular terms or descriptions.
93. Pace Evans, McKinsey, et. al., there is every indication that the pronoun here is (as Peter Geach maintains), or at least is naturally taken to be, a bound variable-like the last occurrence of ' i' in 'It was once the case that for some atheist i, i was elected to the U.S. Senate but i's atheism had been concealed during the political campaign'. (In this case the pronoun 'his' might be regarded as bound by the restricted quantifier 'an atheist'. But compare this with the plural pronoun in 'Few current atheists have been elected to the U.S. Senate, and their atheism was concealed during the political campaign'. Though also a bound variable, the 'their' is bound not by the restricted quantifier 'few current atheists' but, as it were, by a related unarticulated restricted universal quantifier. The sentence is true iff: (i) few individuals who satisfy the open sentence 'X are current atheists' also satisfy the open sentence 'X have been elected to the U.S. Senate', and (ii) those individuals that satisfy both 'X are current atheists' and 'X have been elected to the U.S. Senate' also satisfy the further open sentence 'Their atheism was concealed during the political campaign'.) For any simple pronoun occurrence, if it is a bound variable it is also an occurrence of a rigid designator. Consider: 'A girl sprang from the particular gametes s and e, and it is a necessary truth that whoever sprang from s and e did not spring instead from the entirely different particular gametes ś and e vs. 'A girl sprang from the particular gametes s and e, and it is a necessary truth that she did not spring instead from ś and é́ . Consider also substituting 'that girl' for 'she'. (The foregoing remarks have benefited from discussion with Alan Berger, who realized independently that McKinsey's argument is incorrect.)
94. Cf. "The Good, the Bad, and the Ugly.
95. Contrary to Lepore and Ludwig ("Semantics and Pragmatics," 222-26), this is not a matter of demonstrative phrases always, or typically, taking wide scope: 'Consider: That graduate student is not in graduate school today. The proposition is, of course, false. But its falsity is quite accidental. Indeed, it would have obtained if we had not lowered our admission standards'.
96. If the demonstratum is not F, the sentence 'That Fdoes not exist' is a true negative existential. Such things are rare. Cf. my "Nonexistence, " Nous 32 (1988): 277-319.
97. I assume here that the parenthetical antecedent is false if the demonstrative 'that graduate student' lacks a designatum.
98. Cf. Borg, "Complex Demonstratives," 239-41. Any theory that assigns logical attributes to propositions rather than to sentences or their meanings (such as is defended by Kripke) is unable to accommodate the validity of this inference, assuming (T), without S as an additional premise. Such theories miss the important distinctions illustrated by S.
99. Again, contrary to Lepore and Ludwig ("Semantics and Pragmatics," 213, 222-226). In any context in which the demonstratum is a graduate student, the fact or state of affairs described by S could have been otherwise. (Philosophers indoctrinated in the Quinean tradition may have a tendency to misconstrue 'necessary' as a term for analyt- icity-a semantic notion-rather than for the peculiarly metaphysical notion of a fact or state of affairs that could not have been otherwise.)
100. Kaplan mentions similarly analytic though typically contingent sentences of the form dthat[α] = [α]-he specifically mentions 'He is the male at whom I am now pointing' (see note 39 above)-claiming that all such sentences are a priori ("Demonstratives," 518, 538-39). (Braun, 'Structured Characters," 211-12, 215-216, considers an example exactly like S, correctly deeming it logically valid. Braun does not discuss its epistemological status.) Kaplan offers as an explanation of the existence of such contingent yet (allegedly) a priori truths that alethic modal attributes (metaphysical necessity, possibility, contingency, etc.) are attributes of propositions, whereas apriority and aposteriority are attributes of proposition-characters (that is, of characters that, given a context of use, yield a proposition) or of sentences, not propositions. This confuses epistemological matters (apriority) with properly logico-semantic matters (analyticity), and thus misses one of the important philosophical lessons of demonstratives. Though the sentence ' Dthat [the only member of the UCSB Philosophy Department born in Los Angeles] is the only member of the UCSB Philosophy Department born in Los Angeles' is analytic-in-Kaplish-and hence, known to be true solely on the basis of pure Kaplish semantics-there is no learning the contingent fact described thereby (to wit, that I am the only UCSB philosopher born in Los Angeles) except through epistemic appeal to experience.
101. o is contingent a priori. See notes 13 and 42 above. Such sentences should be deemed analytic even though the facts described are neither necessary nor ( pace Kaplan and Kripke) a priori. Although the existence of analytic
102. and especially "Analyticity and Apriority," in J. Tomberlin, Language and Logic, 125-33.