Kripke'nin betimlemeciliğe karşı argümanları ne kadar kesin?

Bu metni David Grunberg'in bir dersi için yazmıştım. Ders ayrıntılarına girmiyorum; buraya koyma sebebim, felsefe bölümünü açarken söylediğim şeyin somut bir örneği olması.

Metin İngilizce ve olduğu gibi duruyor; çevirmedim. Bir felsefe metnini çevirmek, terimlerin yerini oynatmak demek ve bu yazının tamamı zaten terimlerin tam olarak nereye oturduğuyla ilgili.

Bu bir ders metnim: bana ait ve Turnitin'den geçti. İntihal yok, yapay zekaya yazdırılmış bölüm yok.

Konusu kısaca şu: Kripke'nin Naming and Necessity'de özel adların betimlemeci kuramına karşı öne sürdüğü üç argüman — kipsel, epistemik ve anlambilimsel — betimlemeciliği gerçekten kesin biçimde çürütüyor mu? Vardığım sonuç iki parçalı: klasik betimlemeciliğe karşı evet, ama betimlemeci sözcük dağarcığıyla kurulmuş her görüşe karşı hayır. Ayakta kalan görüşler, ayakta kalmayı büyük ölçüde Kripke'nin kendi açıklamasını betimlemenin içine yazarak başarıyor.

---

How Conclusive Are Kripke's Arguments Against Descriptivism?

Introduction

In Naming and Necessity, Kripke (1980) attacks the descriptivist theory of proper names with three lines of argument, usually labelled the modal, the epistemic, and the semantic arguments. The standard view in the literature is that these arguments were devastating: after 1970, hardly anyone defends the theory of names found in Frege (1892/1948), Russell (1905), or Searle (1958) in its original form. But "devastating" and "conclusive" are not the same thing. A surprising number of philosophers have continued to defend positions that call themselves descriptivist — wide-scope descriptivism (Dummett, 1973), rigidified descriptivism (Plantinga, 1978), causal descriptivism (Kroon, 1987; Lewis, 1984), and the two-dimensionalist program (Jackson, 1998; Chalmers, 2006). If these positions are genuinely descriptivist and genuinely survive Kripke's arguments, then the arguments are not conclusive against descriptivism as such.

In this paper I want to argue for a two-part verdict. First, Kripke's three arguments, taken together, really do refute classical descriptivism — the view that the meaning of a name is given by descriptions speakers associate with it, and that reference is fixed by descriptive fit. None of the three arguments does this alone; their force is cumulative, and I will try to show how each one closes off an escape route left open by the others. Second, the descriptivist positions that survive do so only by giving up what made descriptivism interesting in the first place. Causal descriptivism, in particular, "survives" the semantic argument by writing Kripke's own causal-historical account into the reference-fixing description. I will suggest that this is better described as a notational variant of Kripke's view than as a rival to it. So my answer to the title question is: conclusive against descriptivism as a substantive theory, not conclusive — and probably not capable of being conclusive — against every view formulated in descriptivist vocabulary.

Before the arguments, the target needs to be stated carefully, because part of the dialectic turns on which thesis is under attack at which point.

What Descriptivism Claims

Classical descriptivism is really two claims bundled together. The first is a claim about meaning: a proper name is semantically equivalent to a definite description, or — in the cluster version of Searle (1958) and Wittgenstein (1953, §79) — to a weighted family of descriptions. On this view "Aristotle" means something like "the pupil of Plato who taught Alexander and wrote the Metaphysics," or, more cautiously, refers to whatever satisfies enough of the descriptions in the cluster. The second is a claim about reference-fixing: the referent of a name, in a speaker's mouth, is whatever object uniquely satisfies (most of) the associated descriptions.

These claims can come apart. One could hold that descriptions fix reference without giving meaning — Kripke himself allows that reference is sometimes fixed by description, as with "Neptune" (1980, p. 79n). This matters because Kripke's three arguments do not all attack the same claim. The modal and epistemic arguments target the meaning claim; the semantic argument targets the reference-fixing claim. A descriptivist response that answers one argument may therefore leave the others untouched, and I think failing to notice this is the main reason the secondary literature sometimes overstates or understates Kripke's success.

Kripke (1980, pp. 64–71) codifies descriptivism in six theses plus a non-circularity condition: the descriptions used to fix the reference of "Socrates" must not themselves contain the notion "the man called 'Socrates'," otherwise the account presupposes what it was supposed to explain. I will come back to this condition at the end, because in my view it anticipates the move that contemporary causal descriptivists make.

The Modal Argument

The modal argument rests on Kripke's distinction between rigid and non-rigid designators. A designator is rigid if it picks out the same object in every possible world in which that object exists (Kripke, 1980, p. 48). Names are rigid; most descriptions are not. Compare:

(1) Aristotle might not have taught Alexander.

(2) The teacher of Alexander might not have taught Alexander.

(1) is straightforwardly true — Aristotle could have died young, or gone into politics.

(2), read with the description inside the scope of the modal, is false. If "Aristotle" just meant "the teacher of Alexander," the two sentences should not come apart like this. More generally, "Aristotle was a philosopher" can differ in truth value from "the teacher of Alexander was a philosopher" at counterfactual worlds, since at those worlds someone else may have taught Alexander. The name follows the man; the description follows the role. The same machinery gives Kripke the necessity of true identities between names ("Hesperus is Phosphorus"), which no classical descriptivist semantics can deliver.

The first descriptivist response comes from Dummett (1973). He suggests that we interpret names as descriptions that take wide scope over modal operators. With this in mind, (1) could be parsed as, “The teacher of Alexander is someone who might not have taught Alexander”, rendering it true and accounting for the data. I do not believe this is a viable solution, for the reason Kripke outlines in his preface (1980, pp. 10–14): rigidity is present even in simple, unembedded sentences. When we evaluate the statement “Aristotle was fond of dogs” across some counterfactual world, we are referring to that exact man, rather than whoever happens to satisfy a description in that specific world. Furthermore, such a simple sentence lacks the scope ambiguity that wide-scope theorists need to exploit. Soames (2002) further argues that wide-scopism yields incorrect predictions when names are subjected to iterated modal and temporal operators. To my knowledge, there are no serious defenders of wide-scopism left, so I will treat the approach as definitively refuted.

The second reply is harder to dismiss: rigidify the description. "The actual teacher of Alexander" denotes Aristotle at every possible world, so it matches the name's modal profile exactly (Plantinga, 1978). Against rigidified descriptivism, the modal argument simply has no traction. The standard objections come from elsewhere: rigidified descriptions behave badly in attitude reports — ascribing to ordinary speakers, including children, mastery of the concept of actuality they plainly do not have (Soames, 2002) — and rigidification does nothing to answer the semantic argument, which I discuss below. What I take from this is that the modal argument is conclusive against the synonymy of names with ordinary descriptions, but on its own it cannot rule out descriptivism. It needs the other two arguments to close the gaps.

The Epistemic Argument

The epistemic argument attacks the meaning claim from a different direction. If "Aristotle" meant "the teacher of Alexander," then "Aristotle taught Alexander, if Aristotle existed" should be knowable a priori — true in virtue of meaning, like "bachelors are unmarried." But it is obviously an empirical claim. A historian could discover tomorrow that Aristotle never taught Alexander, and competent speakers can doubt it without any linguistic confusion (Kripke, 1980, pp. 78–85). Since none of the famous-deeds descriptions associated with a name is knowable a priori of its bearer, none of them gives the name's meaning. The cluster version fares no better: it is not a priori that Aristotle satisfied most of the cluster either.

Kripke sharpens the point with the standard meter case (1980, pp. 54–57). Someone who fixes the reference of "one meter" by stipulating that it is the length of stick S can know a priori that S is one meter long (if S exists) — yet what they know is contingent, since S could have been longer. The lesson is the fixing/meaning distinction: even when a description genuinely fixes reference, it does not become the name's meaning, and the a priori knowledge it generates is an artifact of the stipulation, not of synonymy. This distinction blocks what would otherwise be a natural descriptivist retreat ("fine, descriptions only fix reference, they don't give meaning") — or rather, it forces that retreat into the open, where the semantic argument is waiting.

There is a descriptivist response worth taking seriously here: go metalinguistic. Perhaps "Aristotle" is associated a priori with "the bearer of 'Aristotle' in my community's practice." Kripke (1980, pp. 68–70) objects that this is circular — "the man called 'Aristotle'" presupposes reference — but Bach (1987) and Katz (1994) have argued the circularity can be avoided if the description invokes a name-using practice rather than the semantic reference of the very token being explained. I am inclined to grant that metalinguistic descriptivism is coherent. But notice what it costs. The descriptions doing the work are no longer identifying knowledge about the bearer; reference is determined by a social practice of using the name — which is, in all but name, Kripke's own picture. Kripke remarks that a "trivial" descriptivism of this sort may be unfalsifiable precisely because it has stopped being a competing theory, and I think that is the right diagnosis.

One more complication. Machery, Mallon, Nichols, and Stich (2004) found that East Asian participants were more likely than Western ones to give descriptivist verdicts about Gödel-style cases, which threatens the intuitive evidence Kripke relies on. The result has been challenged — Ludwig (2007) argues the vignettes conflate speaker's reference with semantic reference, and later studies probing actual usage rather than explicit judgments found more uniformly Kripkean patterns. I cannot settle the experimental debate here. What it shows, at minimum, is that the arguments rest on judgments rather than proofs, so "conclusive" should be read as "rationally compelling given the evidence," not "demonstrative." That seems to me the right standard for philosophy anyway.

The Semantic Argument

The semantic argument is, I believe, the deepest of the three, because it targets the weaker and more defensible descriptivist claim — that descriptions fix reference — and because it works through cases that generalize.

The cases come in two kinds. Cases of ignorance: many speakers can associate with "Feynman" nothing better than "a famous physicist," which does not distinguish Feynman from Gell-Mann. Descriptivism predicts that such speakers fail to refer, or refer indeterminately. But they clearly do refer to Feynman (Kripke, 1980, p. 81). Cases of error: the Gödel–Schmidt thought experiment. Suppose all I associate with "Gödel" is "the man who proved the incompleteness of arithmetic," and suppose that in fact an unknown Viennese named Schmidt proved it and Gödel stole the manuscript. Descriptivism entails that my uses of "Gödel" refer to Schmidt. Intuitively they do not: I am referring to Gödel and saying something false about him (Kripke, 1980, pp. 83–84). The Jonah case extends this to total error — even if everything the Bible says about Jonah is legend, the name may still refer to a real person.

In place of descriptivism Kripke offers the causal-historical picture: a name is introduced in a baptism, by ostension or by a reference-fixing description, and then passed along a chain of communication, each user intending to refer to whatever the person they got the name from referred to (1980, pp. 91–97). Reference depends on the history of the practice, not on what is in any individual speaker's head. Kripke insists this is a "picture," not a finished theory, and the qualification turns out to matter.

The strongest descriptivist reply is causal descriptivism (Kroon, 1987; Lewis, 1984): let the reference-fixing description be "the individual at the origin of the causal chain underlying my current use of 'Gödel'." This description is satisfied by Gödel, not Schmidt, so the error cases are handled; the ignorance cases too, since even ignorant speakers stand in the right causal relation. Jackson (1998) generalizes the strategy: whatever the correct metasemantics is, speakers can internalize it as descriptive content, so some descriptivism must be true.

I have two objections to this move, neither original but both, I think, decisive for how we should score the debate. The first is psychological: it is not plausible that ordinary speakers — five-year-olds using "Santa Claus," say — associate sophisticated metasemantic descriptions about causal chains with every name they use (Soames, 2002). Jackson can retreat to implicit knowledge, but the more implicit the knowledge becomes, the less the view looks like a theory of what speakers grasp and the more it looks like a redescription of facts that obtain independently of them. The second objection is dialectical: a descriptivism whose description simply quotes the rival theory has conceded the substantive question. What Kripke wanted to know is what makes it the case that "Gödel" refers to Gödel. The causal descriptivist's answer is: the causal-historical facts — packaged inside a description. The packaging does no explanatory work. This is, I think, exactly what Kripke's non-circularity condition was designed to flag: an account that fixes reference by appeal to the practice of referring is not an independent account of reference at all (Devitt & Sterelny, 1999, make a similar point).

Two-Dimensionalism and Problems for Kripke's Own Picture

The most ambitious neo-descriptivist program is two-dimensional semantics (Chalmers, 2006; Jackson, 1998). Each expression gets two intensions: a secondary intension that is rigid and vindicates Kripke's modal claims, and a primary intension — a function from epistemic scenarios "considered as actual" to extensions — that behaves like a Fregean sense and explains cognitive significance. "Hesperus is Phosphorus" comes out necessary (secondary) but a posteriori (primary), so the framework promises to explain Kripke's necessary a posteriori rather than merely accommodate it. Soames (2005) objects that for most names there is no non-trivial primary intension available to speakers except metalinguistic or causal ones, which reintroduces all the problems above. I find Soames persuasive, but the debate is not settled, and honesty requires saying so: as long as two-dimensionalism is a live research program, Kripke's arguments have not closed the logical space.

It should be mentioned that Kripke’s alternative is not without its faults. Evans (1973) used his "Madagascar" example to show that reference can shift along a causal chain (where the name originally referenced a piece of the African mainland). Because of this, we can say that causal connections cannot explain the whole picture, so users’ descriptive intentions must be included. Devitt and Sterelny (1999) present the qua problem. In such a case, the baptism by ostension is not sufficient to say whether the name picks out the animal itself, the species, or the surface, without some form of descriptive intention on the part of the baptizer. While none of this defends classic descriptivism, it is a sign that the correct approach is a hybrid in which certain descriptive elements restrict causal connections. These points help to illustrate why Kripke was careful to offer a general picture rather than a complete theory. Ultimately, one can present a decisive argument against a rival theory without necessarily having a fully developed replacement ready to take its place.

Conclusion

How should we interpret the title question? By themselves, the three arguments do not provide a sufficient basis for anything conclusive. Rigidification resolves the modal argument. The epistemic argument can be resolved by going metalinguistic. The semantic argument is resolved by causal descriptivism. However, these do not combine to provide anything stable. Each response also sacrifices part of the classical model. Both metalinguistic and causal descriptivisms abandon the idea of identifying knowledge in speakers that ultimately constrains reference, which is essentially the core of the theory. Every descriptive theory that survives the onslaught of Kripke does so largely by becoming Kripkean, while still maintaining a façade of descriptivism.

In light of the arguments presented, I am now able to offer the two-part conclusion that I promised. In regard to Kripke's direct challenge: the three arguments cumulatively expressed are, in the sense that is most appropriate for philosophy, jointly definitive, eliminating any credible version of the theory. In reference to Kripke's challenges in regard to descriptivism as a flexible interpretative framework, the three arguments are not conclusive, and the existing experimental literature suggests that caution is warranted regarding the foundational intuitions. Ultimately, I do not believe that the surviving interpretative frameworks that rely so heavily on mirroring their counterparts and rivals deserve any credit for surviving.