Radical Quotation and Real Repetition more

© Blackwell Publishing Ltd. 2004, 9600 Garsington Road, Oxford OX4 2DQ, UK and 350 Main Street, Malden, MA 02148, USA. Ratio (new series) XVII 2 June 2004 0034–0006 RADICAL QUOTATION AND REAL REPETITION David Roden Abstract In this essay I argue for a constructivist account of the entities composing the object languages of Davidsonian truth theories and a quotational account of the reference from metalinguistic expressions to interpreted utterances. I claim that ‘radical quotation’ requires an ontology of repeatable events with strong similarities to Derrida’s account of iterable events. In part one I summarise Davidson’s account of interpretation and Olav Gjelsivk’s arguments to the effect that the syntactic individuation of linguistic objects is only workable if interpreters make richer assumptions about semantic properties than Davidson can tolerate. In part two I show that the objectivist account of syntactic objects which Gjelsivk’s arguments presuppose is incompatible with one corollary of Davidsonian semantic indeterminacy: namely, the relativity of language to interpretative scheme. In place of this an account of radical interpretation is presented in which a quotational theory of metalinguistic reference furnishes the requisite relativity. In part three I argue that this account requires that particular utterance events must be repeatable to be radically quotable and give reasons why particularity and repeatability are not incompatible. Conceptual affiliations between the work of Jacques Derrida and Donald Davidson are well-attested; belying their antithetical styles and preoccupations. Samuel Wheeler puts the matter well when he writes that their common point of departure is the impossibility of a ‘magic language . . . The language in which we know what we mean, think our thoughts, and form intentions’ (Wheeler 2000, p. 3). If there can be no such ‘self-interpreting’ language it is a mistake to reify meanings or phenomenological contents. Meaning arises, rather, from relations between ‘texts’ and the multifarious ways in which these operate in the world. This concordance over the non-intrinsicality of meaning may naturally subsist alongside deep theoretical differences between the two thinkers. Wheeler suggests these derive from the central- 192 DAVID RODEN ity, or otherwise, of the notion of truth.1 However, another less remarked difference is apparent in Davidson’s endorsement of the classical type/token distinction and Derrida’s principled rejection of it.2 Davidson’s commitment to type/token ontology is evident in claims regarding the hermeneutic adequacy of formal theories in which the abstract constituents of languages are explicitly represented. In this essay I argue that this difference derives from a philosophical oversight on Davidson’s part: an ontology of linguistic types cannot form part of a cogent account of radical interpretation. The position which Davidson should espouse resembles Derrida’s in important respects. It is a constructivist account wherein the idioms under interpretation are literal artifacts of the process of theory construction. I claim that this position is best understood by way of a quotational account of metalinguistic reference. Finally I demonstrate how using ‘radical quotation’ in the metatheory of radical interpretation requires an ontology of repeatable particular events in contradistinction both to Davidson’s ephemeral events and Chisholm’s recurrent ‘eternal events’.3 This requirement confirms the deep connections between Davidson’s and Derrida’s philosophies whilst helping to clarify the metaphysical basis for the latter’s pronouncements concerning the iterability (or differential repeatability) of texts. I Davidson’s project in the philosophy of language is to refine our theories about what it is for written and spoken marks to be meaningful by giving theoretical expression to our capacity for interpreting fellow language users. He argues that knowledge of an empirical theory specifying the truth conditions of arbitrary sentences of a language would suffice for interpreting the utterances of its speakers, given knowledge that the theory in question was interpretative for that language. He also claims, famously, that Tarski’s technique for recursively deriving truth conditions of sentences from the satisfaction conditions of predicates affords a 1 2 3 See Wheeler (2000, pp. 4–8). See Derrida (1988, pp. 50–53). Such an ontology is not a contradiction in terms, as I argue towards the end of III. © Blackwell Publishing Ltd. 2004 RADICAL QUOTATION AND REAL REPETITION 193 compositional structure for such a theory (Davidson 1984, pp. 17–36). If both claims are correct, an account of semantic understanding can get by with a minimal framework consisting of formal satisfaction and reference axioms and an ‘intuitive’ or pretheoretical notion of truth (Davidson 1984, pp. 221–223; 1990, p. 292, p. 300, p. 314). Statements of the truth conditions of sentences in the language under interpretation (the object language) are derivable from axioms as ‘T-sentences’: biconditionals linking the application of a truth predicate and a sentence expressing truth conditions in the language of the theory (the metalanguage). Schematically: (1) s is TrueL if and only if p Where ‘s’ is some expression referring to the object language sentence and p takes the metalinguistic statement of its truth conditions. The subscript ‘L’ simply reminds us that the theory involves a particular language L. Since Davidson understands ‘. . . is True’ pre-theoretically it is a dispensable but useful reminder that the T-sentences are held to apply to all and only the sentences of L (See Davidson 1990, p. 292). If Davidson’s sufficiency claim holds, it must be possible to establish whether a truth theory gives truth conditions for Lsentences without prior semantic knowledge of L. For example, the process of verification cannot require foreknowledge of the referents or ‘senses’ of L’s expressions. Davidson’s project thus hinges on the possibility of interpreting a language ‘from scratch’, or radical interpretation (See Davidson 1984, pp. 125– 137). For him, this involves testing T-sentences by observing whether the sentences on the right hand of each T sentence tend to be true when sentences referred on the left are held true by native speakers: a technique readily applicable where a sentence’s truth depends on transitory states of the environment.4 In so doing the interpreter must apply the principle of charity; assuming that native utterances are largely true and rational (1984, p. 137). Davidson’s empirical interpretation of Tarskian formalism requires that radical interpreters have a stock of primitive terms 4 In order to cope with the indexicality of natural languages, Davidson proposes that we augment Tarski’s one place truth predicate with a three place predicate relativised to persons and times (See Davidson 1984, pp. 17–36). Since this refinement has no essential bearing upon my argument, I will ignore it in what follows. © Blackwell Publishing Ltd. 2004 194 DAVID RODEN referring to constituent expressions of the object language and that these can be assembled into ‘structural descriptions’ reflecting the syntactic composition of its sentences (Davidson 1984, p. 133). For example, an axiom in a truth theory for a language might state: (2) qŸxŸy is satisfied by a sequence of objects if the first member of the sequence is larger then the second member. Where ‘q’ is a primitive name referring to a predicate expression in the language under interpretation. The empirical adequacy of a theory containing 2) will thus depend, partly, on the circumstances under which sentence tokens containing q are held true or false by speakers. For this dependence to hold the structural description ‘qŸxŸy’ cannot refer exclusively to any subset of these tokens, since the theory is intended to capture something general about how sentence composition contributes to truth. ‘q’ is most readily understood as referring to a syntactical type. A type is an abstract object whose nature determines whether or not any mark is an instance or ‘token’ of it.5 It seems to follow that truth theories commit those who employ them to the existence of types; a consequence Davidson views with equanimity: ‘[T]he role of sentences in a theory is merely to make it possible to deal with types of utterances and inscriptions, whether or not particular types are realised’ (Davidson 1990, p. 309: author’s emphasis). However, in ‘Davidson’s Use of Truth in Accounting for Meaning’ Olav Gjelsvik claims that individuating sentences as Davidson proposes entails restrictions upon the kind of situation in which a true truth theory would be semantically informative (Gjelsvik 1994). If languages are individuated by the syntactic types composing their expressions – roughly, by the physical shape and structure of grammatical strings – the semantic properties of their sentences must be non-essential. It is thus possible for a sentence to have different semantic properties in different speech communities. But then a truth theory for one community can be falsified if another uses tokens of these types differently. For example, on Twin Earth, or in some inaccessible valley, a language, Twinglish, might be spoken in which English-shaped 5 Despite the fact that only its tokens have a genuine signifying function (See Goodman 1973, pp. 261–262; Allen 1993, p. 11). © Blackwell Publishing Ltd. 2004 RADICAL QUOTATION AND REAL REPETITION 195 predicates have contrary ‘meanings’. The existence of Twinglish would be enough to falsify the T sentence: (3) ‘Snow is white’ is TrueE if and only if Snow is white Since it is the syntactic string referred to by ‘Snow in white’ which relativises a truth predicate, not its subscripts, there is nothing to distinguish 3) from a statement about a sentence of Twinglish: (4) ‘Snow if white’ is TrueTw if and only if Snow is white If ‘. . . is white’ in Twinglish were a contrary of its English counterpart (meaning is green, say) the ‘only if’ in (3) would make it false. According to Gjelsvik, the only alternative is to specify English sentences semantically. A Tarskian theory achieves this by defining a predicate that holds of all and only the true sentences of a language. But T-sentences flow from such definitions by stipulation and logical necessity. They do not, as with Davidsonian theories, express contingent, empirical claims about semantically uncharacterised sentences.6 To avoid this vitiating course, Gjelsvik argues, a competent radical interpreter must assume that the world’s distribution of semantic properties is not of the Twinglish/English sort (Gjelsvik 1994, p. 34). The problem, here, is that this assumption utilises pre-theoretic concepts of subsentential meaning. Davidson’s minimal sufficiency claim is not sustainable if the metatheory of radical interpretation would have to employ such concepts. One response to Gjelsivk’s argument suggested in the essay ‘Radical Interpretation’ might be to question the centrality of the concept of language. Davidson proposes there that the notion of language or ‘speech community’ be operationalised in terms of effectiveness of truth theories in predicting speech behaviour: ‘speakers belong to the same speech community if the same theories of interpretation work for them’ (Davidson 1984, p. 135). Where there are variant usages of similar sounds or inscriptions, the number of languages involved is determined by how many workable semantic theories are needed to interpret that behaviour. But even supposing that Davidson’s suggestion is ultimately coherent it will not quite do as a rebuttal. Gjelsvik’s arguments do not depend on there being objective conditions of individuation for languages but only for types. 6 On this point Gjelsvik refers to Etchmendy (1988). © Blackwell Publishing Ltd. 2004 196 DAVID RODEN II Davidson’s linguistic deflationism is heuristically useful, though. For it suggests an analogous approach to sentences and subsentential units. Gjelsvik must assume that whether two sets of marks are syntactically identical is an interpretation-independent fact. But if an interpretative theory ‘constructs’ its text by, say, determining what ‘counts’ as an instance of the ‘same’ sentence, the conditions of sentence individuation cannot be objective in the way that his arguments require. To provide a rebuttal of Gjelsvik along these lines we must: 1) demonstrate that the individuation of linguistic objects by way of interpretative schemes makes sense in the light of Davidson’s approach to semantics; 2) give an account of this operation that is both consistent with his principles of interpretation and metaphysically coherent. 1. As it happens, Davidson has an independent motivation for claiming that syntactic properties of texts are determined by the ways in which they are interpreted. Davidson’s semantics, like Quine’s, entails that the grounds for adhering to semantic theories are exhausted by publicly accessible data. This entails semantic indeterminacy since there is nothing to rule out the possibility of incompatible theories equally well confirmed by all available evidence. Thus a theory T1 which glosses Gavagai as ‘Lo, a black rabbit’ and another T2 which glosses it as ‘Lo, a white rabbit’ could both be adequate in the light of all the facts regarding the use of this expression. Ian Hacking has objected that countenancing this involves a reductio of Davidson’s position since this implies that T sentences which assign contrary truth conditions to the same sentence can both be true (Hacking 1975, p. 154). In response, Davidson argues, in ‘The Inscrutability of Reference’, that the indeterminacy is not that of the truth or content of an individual sentence but of which language is being spoken: Our mistake was to suppose there is a unique language to which a given utterance belongs. But we can without paradox take that utterance to belong to one or another language, provided we make allowance for a shift in other parts of our total theory for a person (Davidson 1984, pp. 239–40) Davidson’s response obviates the charge of inconsistency only if each truth theory that is interpretative for the same episodes of verbal © Blackwell Publishing Ltd. 2004 RADICAL QUOTATION AND REAL REPETITION 197 behaviour applies to a distinct and proprietary range of sentences.7 The T sentences of T1 and T2 would then refer to ‘different’ Gavagai’s. No contradiction would ensue since each would deal with disjoint sets of truth bearers. Now if this is what Davidson’s semantics entails, it cannot be compatible with the position that syntactic identity is only a matter of physical shape and structure (or other objective properties), for whether a particular utterance or inscription exhibits these is not plausibly determined by choice of semantic theory. 2. The outlines of an alternative account of the link between an interpretative truth theory and its text are sketched in the work of a number of commentators on Davidson’s work. Bjorn Ramberg argues that Davidson’s interpretationist approach to meaning entails that we cannot view languages as ontologically complete or straightforwardly given (Ramberg 1989, pp. 98–90, p. 100). This position is overdetermined since it is entailed both by the semantic indeterminacy thesis and by Davidson’s rejection of a semantic explanatory role for shared linguistic conventions or rules (See Davidson 1986; 1984, pp. 265–280). In ‘Indeterminacy of French Interpretation: Derrida and Davidson’ Wheeler expresses this ontological thesis in terms of an explicitly constructivist account of the type/token distinction: whereby types are constructed from ‘the potentially infinite repetitions of the token, past and future’ (Wheeler 2000, pp. 22). Two marks or sounds are ‘tokens of the same type’ only relative to some interpretation in which they are ‘posited’ as being repetitions of the same sign. Likewise, any sound or inscription can be a token of different words under distinct interpretative theories. What unites Davidson and Derrida, for Wheeler, is the realisation that this exigency applies to any metalanguage. If, like Davidson, we wish not to reify ‘semantic natures’ there can be no regress to a ‘background language’ whose signs bear semantic properties intrinsically (Davidson 1984, pp. 234–35). Any scheme in which a mark is posited as a repetition of another mark is thus a text, and there will always be divergent ways to interpret it via yet other texts. There can be no necessity governing which marks count as repetitions of the same type under some interpretative 7 Davidson invokes the fact that Tarskian truth definitions are relativised to a language in his response to Hacking (Davidson 1984, p. 239). But if, with Davidson, we posit syntactic types there is no reason why two or more theories should not give different truth definitions for the same set of syntactic types. They would be relativised in giving the extension of ‘True in L’ but for the same L. © Blackwell Publishing Ltd. 2004 198 DAVID RODEN scheme (See Pradhan 1986). It follows that there are no ‘types’ if by this we mean an abstract object instantiated in each repetition of a sign. Although Wheeler denies that signs are repeatable ‘essences’, he concurs with Derrida that all signs must be essentially repeatable across distinct contexts. Repeatability – or as Derrida prefers to say ‘iterability’ – is the basis for the construction of types, though it does not presuppose contingently instantiated abstract objects. Iterability is thus a condition of possibility for signification in general: Every sign, linguistic or nonlinguistic, spoken or written (in the usual sense of this opposition), in a small or large unit, can be cited, put between quotation marks: in so doing it can break with every given context, engendering an infinity of new contexts in a manner that is absolutely illimitable . . . This citationality, this duplication or duplicity, this iterability of the mark is neither an accident nor an anomaly, it is that (normal/abnormal) without which a mark could not even have a function called ‘normal’ (Derrida 1988, p. 12). What a sign means depends upon its prior and future uses: its deviant, or so-called ‘parasitic’, uses are thus part of what it ‘is’ (Wheeler 2000, p. 23). No signifying event can thus be entirely ‘present’ since interpretation necessarily supervenes upon its uses in other contexts. As we have seen, Davidson’s response to Hacking reinforces Wheeler’s contention that his semantics involves an anti-realist stance towards sentence-types and their constituents. Whether such types are constructed, in Wheeler’s sense, merits further investigation. It remains unclear what treating distinct marks as repetitions of the same sign can amount to if it does not involve assignment of expression-types to tokens. Plausibly, it involves relating the production of a sign to something other than a type. Our problem is to characterise the relation and the relata. In what follows I argue that the relation is best understood as a variety of quotation by showing that a variant of quotation can account for the relativisation of language to theory. The constructivist thesis follows from the role of quotation. Quotation, whether direct or indirect, is a kind of reference but unlike names or pure descriptions quotations refer by way of instances or examples of the referent. The expressions ‘ “Metaphor is the dreamwork of language” ’ and ‘The first sen© Blackwell Publishing Ltd. 2004 RADICAL QUOTATION AND REAL REPETITION 199 tence of Davidson’s essay “What Metaphors Mean” ’ each refer to the same things, but the first expression appears to ‘contain’ or exemplify what it refers to while the second does not. While the phenomenology of graphic containment may be an artifact of the use of quotation marks, I see no reason to demur from the claim that exemplification characterises quotation in all forms or media and distinguishes it from other kinds of reference.8 Assuming the correctness of Davidson’s response to Hacking, it must be impossible for expressions in different interpretative theories to refer to the same sign. Otherwise, indeterminacy would allow that equally good theories could make incompatible semantic claims regarding it. How can we explain that there are objects that can only be referred to by signs restricted to a given theory? If metalinguistic reference is to be mediated by arbitrary signs such as names or definite descriptions the restriction cannot be justified. If, however, metalinguistic reference is mediated only by exemplification, the identity of the relata is restricted. Only instances of a thing are examples of it. However, the relation does not commute. Instancing is not sufficient for exemplifying since, as Nelson Goodman points out, exemplification is ‘possession [of a property] plus reference’ (Goodman 1976, p. 53). This entails, of course, that exemplification is not necessary for possession. Though obvious, this is important, since we do not want our examples to refer exclusively to examples. Exemplification only provides the requisite relation if the same sentence-examples cannot belong to different theories. This would be the case if theory-membership individuated the examples. But what is a theory? As we have seen, Davidson attributes no greater degree of determinacy to interpretative theories than to the idioms they interpret. However, although theories are not determinate, it remains legitimate to conceive them in terms of norms or regularities governing the use of their constituent expressions. In formal systems these are represented by explicit rules of inference, substitution, syntax, etc. But the advantages of formalisation are conceptual, not hermeneutic. Mastery of an interpretative theory can be implicit in an interpreter’s practical grasp of the inference relationships of a language and her ability 8 Not all accounts of quotation presuppose the exemplification of the expression quoted. However, theories which ignore this feature are ill-equipped to explain the productivity of quotation conventions in natural language. See Davidson 1984, pp. 79–92. © Blackwell Publishing Ltd. 2004 200 DAVID RODEN to consistently match truth conditions with utterances (Davidson 1990, p. 312). If theories are constituted by norms, the identity of the tokens in an interpreter’s theory is determined by that theory. Whether a particular mark counts as a sentence, what sentence it is taken to be, etc., will depend on which norms it is subject to. Interpreters using distinct theories will draw on different sentences as examples for citing alien utterances. The sentences instanced by utterances according to one theory will differ from those instanced for a different theory. Although two theories might give incompatible truth conditions for the same speech acts there would be never be grounds for equating the sentences assigned to those speech acts. If we understand metalinguistic reference in radical interpretation as mediated by exemplification, the identity of the sentences interpreted will be guaranteed theory-relative in just the way Davidson’s account requires. Moreover, we can circumvent Gjelsvik’s objections because whether a sentence, in Twinglish, say, is referred to in a theory of English depends on whether it is exemplified by the relevant sentence-examples in that theory. But it can only be so exemplified if it matches their truth conditions and inference relationships. There is, then, a sense in which interpretative theories refer to their object languages semantically. However, no compromising reference to the meanings of parts of sentences is required, because the radical interpreter quotes examples of sentences which already have a certain semantic role in her theory and asks whether certain alien utterances are examples of what she quotes. Radical interpreters thus need ‘radical quotation’ to put words of their coining into other’s mouths. Exemplifying reference must be used because syntactic distinctions are abstracted according to largely semantic criteria.9 We can only distinguish between sentences, in the relevant sense, once we know something about their meaning; so in radical interpretation, at least, there can be no rounding up of suspected sentences prior to successful interpretation. The ‘theory’ thus includes a kind of model language in which, as in ordinary language, the semantic roles of expressions are indicated by their shapes. The semantics of this model is stipulative, in a sense. The interpretative theory does not simply describe or define the semantic properties of a language but 9 Daniel Dennett adopts this position in ‘Beyond Belief’ (Dennett 1987, p. 137). © Blackwell Publishing Ltd. 2004 RADICAL QUOTATION AND REAL REPETITION 201 produces them; yielding a semantically individuated object language. But whether someone other than the interpreters speak this constructed idiom remains open to empirical confirmation or disconfirmation. The empirical question to be settled, is precisely as stated in ‘The Inscrutability of Reference’: not ‘What are the truth conditions of these sentences?’ but ‘What language is being spoken here?’ (Davidson 1984, p. 239). Radical interpretation, then, consists in identifying shapes which do not indicate their semantic roles by means of shapes which do. My account of metalinguistic reference provides indirect support for the identity theory of quotation, since this is arguably the only account that works for radical quotation. The tenet of all identity theories is that expressions in quotes refer to themselves and that quote marks are, as Corey Washington argues, dispensable items of punctuation which indicate that the quoted phrase is being used self-referentially (See Washington 1992). A linguistic expression of theoretical norms employing a sign other than the exemplifying expression to refer to that same expression would allow for competing interpretations of the same theoretical terms. Suppose I employ Davidson’s demonstrative theory of quotation in accounting for radical quotation (See Davidson 1984, pp. 79–92).10 Then each substituend within a T sentence would include a demonstrative phrase. For example: (5) Any inscription with the same shape as this (sentence) is TrueL if and only if Snow is white Where ‘This’ refers to some ostended token of a native sentence. In the demonstrative account, quoted material has no semantic role in quoting expressions. The ostended sentence example would have no part to play in the theory itself. Thus its identity would not depend on the theory’s tacit or explicit rules and it would be nonsense to talk of the theory ‘constructing’ its object language. If we allow arbitrary devices of metalinguistic reference to intervene at any point, we vitiate the constructivist account and open Davidson to the objections raised by Hacking. The only adequate linguistic formulation for the axioms and theorems of interpretative truth theories is one in which the terms referring to elements of the object language are those same object language expressions. 10 According to Davidson, quotation marks function as demonstratives allied to definite descriptions. © Blackwell Publishing Ltd. 2004 202 DAVID RODEN The identity theory requires that expressions such as sentences change their grammatical role, since in quotation they are used for singular reference not assertion (See Davidson 1984, p. 85). Metalinguistic examples must be grammatically equivocal in the way that the identity theory demands of all natural language expressions.11 Here the (unrigorous) distinction between ‘deviant’ and ‘normal’ usage is reversed. The normal (i.e. citational) use of examples depends on the possibility of their being used for assertions and other ‘normal’ speech acts; though it is not necessary that they should be so used (See Wheeler 2000, p. 19). Such multifunctionality is precisely what we ought to expect if, as Derrida claims, iterability is a constitutive feature of language. As Shekar Pradhan puts it, iterability implies that no account of the meaning of a sign ‘can connect with all the possible uses of a sign’ (Pradhan 1986: 72). The norms that govern interpretative theories, as with all non-formal languages, are necessarily pliant. This ‘essential iterability’ renders interpretation possible. III Since utterances are most naturally viewed as events it is reasonable to assign their quotations in interpretative theories to the same category. Since these quotations are examples of the sentences ascribed to native utterances both events (utterances and examples) are instances of that sentence. But how are we to view the claim that some tokens in interpretative theories and some interpreted utterances are ‘instances’ of the same sentence? If we eschew types this position cannot be expressed in terms of a coinstantiation relation (holding between two tokens and some abstract entity) but only as a relation holding between the events themselves. For want of a better term, I propose that we call this ‘repetition’. A repetition is similar to what Nelson Goodman calls a ‘replica’. According to Goodman, marks are replicas of one another if their physical differences are not ‘correlated with a 11 It can be objected that this undermines the Tarskian bit of Davidson’s sufficiency claim since formal theories contain no equivocal elements. However, the point of formalism is to show that certain kinds of concepts and information suffice for interpretation. It is not necessary for theories to be completely formalised. © Blackwell Publishing Ltd. 2004 RADICAL QUOTATION AND REAL REPETITION 203 difference in appropriate use’ (Goodman 1973, p. 263). There are no necessary or sufficient grounds of similarity determining whether this is the case, other than contingent sets of signifying practices. Being a replica is not a matter of instantiating the right abstract object but of being subject to rules determining a notational scheme (See Goodman 1976, pp. 127–143). This may not be too far from Derrida’s position, as I read it, but there are differences which makes me hesitate to adopt Goodman’s terminology. If Derrida is right, no theory or system of rules can capture all possible uses of a sign. This incompleteness is shared by interpretative theories themselves and, as Wheeler argues, is implied by Davidson’s account of referential inscrutability. For example, there are, arguably, rules determining whether marks employed within a given theory are co-replicas. All metalinguistic examples of the same sentence are co-replicas in this sense. But given the non-commutativity of exemplification this relation cannot extend to the quoted alien utterances. Their quotability within a given theory is not determined by the norms of the theory but, empirically, by the speech behaviour of the natives. Iteration embraces replication, but extends beyond it. Replication is relative to the rules or traditions of a notational scheme. Iteration can also be non scheme-relative. Derrida has remarked that iterability is ‘a differential function without an ontological basis’ (Derrida 1984, p. 16). By contrast, I believe that if there is any truth to the iterability thesis it must be because the world contains iterables. There are conceivable worlds containing no iterables, either because they are insufficiently differentiated or because their objects are abstract (sets or numbers can be included in alphabets of formal languages, but not uttered or repeated). Since iterability is an ontological fact, not a logical one, it is legitimate to demand a metaphysics of iterability, and of what Derrida calls the ‘text’. My claim that iterability is involved in radical interpretation can be regarded as an instance of his general thesis that iterability is a condition of possibility for linguistic and non-linguistic signification. However, seeing iteration at work in this highly restricted context contributes to the metaphysical goal by providing indications about the kinds of things iterables must be: A) Iterables are events: this is built into Derrida’s account of iterability in ‘Signature Event Context’ (See Derrida 1988) and is consistent with Davidson and Goodman’s views of utterances and inscriptions. © Blackwell Publishing Ltd. 2004 204 DAVID RODEN B) Iterables are particulars: When a radical interpreter ‘repeats’ or iterates a native utterance in her theory it must be the particular utterance that is repeated and hence quoted. If this were not so our local constructivism would entail wholesale linguistic idealism. Different theories would not only construct different languages, they would not be interpretations of the same speech events. This would be hard to reconcile with the claim that interpretative theories are empirical. C) Iterables are repeatable events. Unlike Davidsonian events, they are not ephemera (See Davidson 1980, pp. 189–203). D) Iterables are repeatable particulars. An ontology of potentially occurrent events, such as expounded by Roderick Chisholm, is incompatible with the rejection of types. The recurrence of Chisholmian events is a property of ‘eternal’ events, not of their occurrences (Chisholm 1970). While some might look askance at an ontology of repeatable particulars, it would involve manifest contradiction only if ‘particular’ and ‘repeatable’ are incompatible attributes. Material objects are regarded as paradigmatic particulars but this status is compatible with their occupying different times and places. As far as I can see, the only ground on which such a potentiality could be denied of particular events is that they are restricted to unique spatio-temporal regions (See Quine 1985, p. 167). It is not clear whether this is even a good principle of individuation of material objects since these do not seem tied to the spatio-temporal ‘world lines’ they may occupy. Pace Quine’s ‘object-oriented ontological attitude’, particularity and individuality may come apart in many cases. ‘Diffusions’ (such as clouds) or collectives (such as virus populations) may have identities in the weak sense that they are relatively differentiated or localised, but lack formal criteria of individuation (See Cantwell Smith 1998, p. 132). Repeatable events or iterables may be regarded as falling into a further class: particulars not individuated by place or time but lacking the continuity of objects.12 Viewing them in this way, we can do justice to Wheeler and Derrida’s claim that a particular sign-event is not ‘totally present’; for though particular, it is not completely individuated by any single context. 12 There may be causal continuity conditions associated with repeatable particulars. For example, it seems implausible to suppose that events could be related by Derridean repetition yet be causally isolated. © Blackwell Publishing Ltd. 2004 RADICAL QUOTATION AND REAL REPETITION 205 Conclusion My account makes explicit what is involved in Wheeler’s claim that signs are constructed through interpretation. The concept of radical quotation provides a coherent descriptive framework for such a constructive activity. Treating sign-construction as radical quotation unpacks the metaphor of ‘construction’ in literal terms since the sentences ascribed to events of utterance or inscription are artifacts whose individuation depends upon the theories in which they are embedded. If Davidson is correct in supposing that susceptibility to radical interpretation is a precondition of meaningfulness, the notion of radical quotation should be seen as part of a metaphysical account of why certain kinds of things are ‘texts’: objects susceptible to interpretation. The account of nonscheme relative repetition outlined in the final section is an essential component of such a metaphysics since, on the Davidsonian account at least, meaningful entities are particulars and, as Derrida has argued, significant particulars cannot be such that they can occur only once.13 D.Roden@btopenworld.com References Allen, Barry (1993). ‘Difference Unlimited’, in G. Madison (ed.) Working Through Derrida (Evanston Ill.: Northwestern University Press). Chisholm, Roderick (1970). ‘Events and Propositions’, in Nous, 4, pp. 15–24. Davidson, D. (1980). Essays on Actions and Events (Oxford: Clarendon Press). —— (1984). Inquiries into Truth and Interpretation (Oxford: Clarendon Press). —— (1986). ‘A Nice Derangement of Epitaphs’, in Ernest LePore (ed.) Truth and Interpretation: Perspectives on the Philosophy of Donald Davidson (Oxford: Blackwell). —— (1990). ‘The Structure and Content of Truth’, Journal of Philosophy, 87 (6), pp. 279–328. Derrida, Jacques (1984). ‘My Chances/Mes Chances: A Rendez-vous with Some Epicurean Stereophonies’, I. E. Harvey and Avital Ronell (trans.), in J. H. Smith and W. Kerrigan (eds.) Taking Chances: Derrida, Psychoanalysis and Literature (Baltimore MD: John Hopkins University Press), pp. 1–32. —— (1988). Limited Inc. Samuel Weber and Jeffrey Mehlman (trans.) (Evanston Ill.: Northwestern University Press). Dennett, Daniel (1987). The Intentional Stance (Cambridge Mass.: MIT Press). Etchmendy, John (1988). ‘Tarski on Truth and Logical Consequence’, The Journal of Symbolic Logic, 53, pp. 51–79. 13 I am grateful to Alessandra Tanesini and Adrian Moore for their comments on earlier drafts of this paper. © Blackwell Publishing Ltd. 2004 206 DAVID RODEN Gjelsvik, Olav (1994). ‘Davidson’s Use of Truth in Accounting for Meaning’, in Language, Mind and Epistemology: on Donald Davidson’s Philosophy, Gerhard Preyer, Frank Siebelt and Alexander Ulfig (eds.) (Dordrecht: Kluwer), pp. 21–43. Goodman, Nelson (1973). The Structure of Appearance (Dordrecht: D Reidel). —— (1976). Languages of Art (Indianapolis: Hackett). Hacking, Ian (1975). Why Does Language Matter to Philosophy? (Cambridge: CUP). Pradhan, Shekar (1986). ‘Minimalist Semantics: Davidson and Derrida on Meaning, Use and Convention’, Diacritics, 16 (1), pp. 66–77. Quine, W. V. (1985). ‘Events and Reification’, in Ernest Lepore and Briain McLaughlin (eds.) Actions and Events: Perspectives on the Philosophy of Donald Davidson (Oxford: Basil Blackwell), pp. 162–171. Ramberg, B. (1989). Donald Davidson’s Philosophy of Language: An Introduction (London: Blackwell). Smith, B. C. (1998). On the Origin of Objects (Cambridge Mass.: MIT). Tarski, A. (1956). J. H. Woodger (trans.), ‘The Concept of Truth in Formalised Languages’, in Logic, Semantics, Metamathematics (Oxford: Clarendon Press). Washington, Corey (1992). ‘The Identity Theory of Quotation’, The Journal of Philosophy, 89 (11), 582–605. Wheeler, Samuel (2000). Deconstruction as Analytic Philosophy (Standford: Stanford University Press). © Blackwell Publishing Ltd. 2004