Paradigms of Reading: Relevance Theory and Deconstruction

Paradigms of Reading Relevance Theory and Deconstruction

Ian MacKenzie

Paradigms of Reading

pour/maRijc/ and, more pertinently, Kirsten

Paradigms of Reading Relevance Theory and Deconstruction Ian MacKenzie

macmillan

© Ian MacKenzie 2002 All rights reserved. No reproduction, copy or transmission of this publication may be made without written permission. No paragraph of this publication may be reproduced, copied or transmitted save with written permission or in accordance with the provisions of the Copyright, Designs and Patents Act 1988, or under the terms of any licence permitting limited copying issued by the Copyright Licensing Agency, 90 Tottenham Court Road, London W1T 4LP. Any person who does any unauthorised act in relation to this publication may be liable to criminal prosecution and civil claims for damages. The author has asserted his right to be identified as the author of this work in accordance with the Copyright, Designs and Patents Act 1988. First published 2002 by PALGRAVE MACMILLAN Houndmills, Basingstoke, Hampshire RG21 6XS and 175 Fifth Avenue, New York, N.Y. 10010 Companies and representatives throughout the world PALGRAVE MACMILLAN is the global academic imprint of the Palgrave Macmillan division of St. Martin’s Press, LLC and of Palgrave Macmillan Ltd. Macmillan® is a registered trademark in the United States, United Kingdom and other countries. Palgrave is a registered trademark in the European Union and other countries. ISBN 0–333–96833–6 This book is printed on paper suitable for recycling and made from fully managed and sustained forest sources. A catalogue record for this book is available from the British Library. A catalog record for this book is available from the Library of Congress 10 11

9 10

8 7 09 08

6 07

5 06

4 05

3 2 04 03

1 02

Printed and bound in Great Britain by Antony Rowe Ltd, Chippenham and Eastbourne

Contents Acknowledgements

vii

Bibliographical Abbreviations

viii

1 Pragmatic Banality and Honourable Bigotry Synopsis Honourable bigotry

1 6 11

2 Relevance Theory and Spoken Communication Linguistic underdetermination Ostension, inference, intentions and relevance Implicatures, vagueness and poetic effects Metaphor and irony, description and interpretation

16 17 18 22 25

3 ‘Positive Hermeneutics’: Relevance and Communication The hermeneutic tradition Communicative intentions The epidemiology of representations The reader as supplement

29 33 36 41 44

4 ‘Negative Hermeneutics’: Themes, Figures, Codes and Cognition Unconscious identities Reading and time Canonicity Codes and inference Carvers and modellers

47 48 50 53 55 60

5 Words, Concepts and Tropes Nouns and concepts Metaphor, truth, lies, realistic assumptions and surplus value Numbers Dead metaphors and catachreses 6 Rhetoric as an Insurmountable Obstacle ‘What’s the difference?’, ‘Son of a bitch!’ and catastrophic confusions Rhetoric and aesthetics v

62 65 69 74 76 84 86 90

vi

Contents

Tropes and persuasion Irony Resistance to language

95 98 101

7 Words and the World: The Problem of Reference Being and becoming Concepts, metaphors, catachreses and reference Reference and ideology ‘Ich kann nicht sagen was ich nur meine’ Reference and application

107 108 112 115 118 126

8 Mechanical Performatives The purloined ribbon Excuses, fictions and machines Paul de Man’s war

131 133 138 146

9 The Madness of Words and the Enunciating Subject The fallacy of the active sign From intentional subjects to inhuman language Arbitrary signifiers and accountable authors Dialogism and ventriloquism

152 153 157 166 170

10 ‘When Lucy ceas’d to be’ Wordsworth’s use of rock and roll Wordsworth’s murderous spirit Lucy and Freud Lucy as metaphor From beyond the grave Miller’s tale

176 177 180 182 186 188 190

11 Relevance and Rhetoric

196

Notes

199

References

221

Index

232

Acknowledgements This book has greatly benefited from the comments of a number of rigorous readers. Thanks are due to Alan Durant, who made very helpful suggestions (not least the title), Neil Forsyth, always an attentive reader, Richard Watts, who obliged me to clarify some of my linguistic arguments, and Adrian Pilkington, who pointed out various local errors concerning relevance theory. An ominous reader for Palgrave also made useful comments. Most of the chapters about Paul de Man were written some years ago. Back in the early 1990s, I argued at length about every single sentence with my former colleague Martine Hennard-Dutheil. Martine’s problem, alas, is that she is invariably wrong about everything. Except when she is right, of course – but then I always miraculously realise that I do in fact share her opinion, even though I may have seemed to have expressed the exact opposite. After careful consideration, however, I chose not to use Martine’s suggested title Blindness and Blindness. More recently, Kirsten Stirling, who improves life, the universe and everything in general, suggested many minor improvements. Any remaining errors are of course my own, so I suggest you imagine an implicit ‘E. & O. E.’, meaning ‘errors and omissions excepted’, in the bottom right-hand corner of each page, as found, for example, on bank statements. A different version of Chapter 10 has been published as ‘When Lucy ceas’d to be: Optimally and Manifestly Relevant Meanings’, in SPELL (Swiss Papers in Language and Literature) 13: The Limits of Textuality, ed. Lukas Erne and Guillemette Bolens (Tübingen: Gunther Narr, 2000). This book started life as a doctoral thesis, and I am very grateful to Peter Trudgill, who for complicated reasons ended up directing a thesis way outside his normal area of expertise, at a university (Lausanne) at which he was no longer working. I also want to take this opportunity to express my appreciation for the Troisième cycle d’anglais des universités Suisses romandes, a splendid institution which between 1982 and 1992 brought to Switzerland, among others, Stanley Cavell, James Clifford, Denis Donoghue, Catherine Gallagher, Stephen Greenblatt, Fredric Jameson, Hillis Miller, Janice Radway, Richard Rorty, Edward Said, Hayden White and Deirdre Wilson, for a series of inspiring seminars.

vii

Bibliographical Abbreviations In the text, I use the following abbreviations to refer to books by Sperber and Wilson, and de Man: Relevance

Aesthetic

Allegories

Blindness

Resistance

Rhetoric Romanticism

Writings

Dan Sperber and Deirdre Wilson, Relevance: Communication and Cognition, 2nd edn (Oxford: Blackwell, 1995; 1st edn 1986). Paul de Man, Aesthetic Ideology, ed. Andrzej Warminski (Minneapolis, MN: University of Minnesota Press, 1996). Paul de Man, Allegories of Reading: Figural Language in Rousseau, Nietzsche, Rilke and Proust (New Haven, CT: Yale University Press, 1979). Paul de Man, Blindness and Insight: Essays in the Rhetoric of Contemporary Criticism, 2nd edn (London: Methuen, 1983). Paul de Man, The Resistance to Theory, ed. Wlad Godzich (Minneapolis, MN: University of Minnesota Press, 1986). Paul de Man, The Rhetoric of Romanticism (New York: Columbia University Press, 1984). Paul de Man, Romanticism and Contemporary Criticism: The Gauss Seminar and Other Papers, ed. E. S. Burt, Kevin Newmark and Andrzej Warminski (Baltimore, MD: Johns Hopkins University Press, 1993). Paul de Man, Critical Writings (1953–1978), ed. Lindsay Waters (Minneapolis, MN: University of Minnesota Press, 1989).

viii

1 Pragmatic Banality and Honourable Bigotry

Theories of literary meaning all agree on at least one point: that linguistic signs or signifiers never wholly coincide with intended or interpreted meanings. Yet the conclusions they draw from this premise are radically opposed. For pragmatic accounts of language such as relevance theory, the impossibility of delivering determinate messages by way of coded signals merely demonstrates the necessity of inference in linguistic interpretation. For deconstructionist accounts of language, such as Paul de Man’s, this non-coincidence is a source of linguistic instability, and the key to the impossibility of reading and the aberrancy of all understanding. Dan Sperber and Deirdre Wilson’s relevance theory – which has been described, in a splendid hyperbole, as the first radically new theory of communication since Aristotle (Fowler, 1989, p. 16) – recognises that every utterance has a variety of possible interpretations that are combinations of explicit content, context and implied meanings. Yet these interpretations are not all equally accessible to a hearer on any particular occasion. The theory proposes that hearers are equipped with a criterion for evaluating interpretations as they occur to them. The criterion of relevance is based on the assumption that human cognition is relevance-oriented and geared to achieving the greatest possible cognitive effect for the smallest processing effort. Individuals focus their attention on what seems to be the most relevant information available. To communicate is to claim an individual’s attention, and hence to imply that the information communicated is relevant. Thus communicated information creates an expectation of relevance. The criterion of relevance should allow hearers to exclude all but one interpretation: a hearer is entitled to assume that the first interpretation that meets the criterion of relevance is the only one that does so. 1

2

Paradigms of Reading

In the terms of relevance theory, utterances are only interpretive representations of thoughts, so there is invariably a gap between the semantic representation of a sentence uttered and the thought it is designed to communicate. This gap has to be filled by a hearer’s or reader’s inferential recognition of the communicator’s intentions, guided by contextual clues. Just as a large proportion of matter in the universe is invisible to both the eye and radio telescopes, so that its existence can only be inferred, much of the meaning of a spoken utterance or written sentence does not appear in the actual words. Thus the linguistic analysis of an utterance, restricted to ‘the materiality of the letter’, is only one of several steps necessary for interpretation. Unlike this inferential model of human communication and cognition, deconstructionist accounts of language insist on the materiality of language, and disregard the enormous role played by contextual factors in linguistic communication. Paul de Man stated that ‘the impossibility of making the actual expression coincide with what has to be expressed, of making the actual sign coincide with what it signifies’ meant that ‘the interpretation of everyday language is a Sisyphean task, a task without end and without progress’ (Blindness, p. 11). He further argued that since all words, names and concepts are metaphorical at root they are unstable, and deconstruct themselves even as they are asserted. Hence ‘the paradigm for all texts consists of a figure (or a system of figures) and its deconstruction’ (Allegories, p. 205). For de Man, rhetorical tropes and substitutions always render the propositional content of utterances and texts undecidable, and wreak ‘epistemological damage’ (Aesthetic, p. 34) that wholly undermines intended meaning. Rhetoric thus ‘radically suspends logic and opens up vertiginous possibilities of referential aberration’ (Allegories, p. 10), and ‘puts an insurmountable obstacle in the way of any reading or understanding’ (Allegories, p. 131). Hence all forms of knowledge are contingent on the radically uncontrollable and autonomous figural potential of language. De Man went on to describe language as invariably ironic, random, arbitrary, irresponsible, mechanical and inhuman. Rather than merely pointing out that discourse may contain disruptive or ambivalent elements that challenge the normal inferential processes of interpretation, and that there is always the possibility of an ironic disparity between meaning and intent, de Man regards this as an inevitability. Socrates, in Plato’s Phaedrus, compares written texts to a pharmakon which, as Derrida pointed out in Dissemination (1981a, pp. 70ff.), can mean either poison or remedy. Whereas pragmatic linguists are looking for a remedy, de Man cannot see

Pragmatic Banality and Honourable Bigotry 3

beyond the poison. This book will argue that relevance theory is an effective pragmatic remedy to de Man’s doom-laden account of language. My basic claim, following Sperber and Wilson, is that substituting the word inference, or more specifically the word relevance, for de Man’s term rhetoric, very often allows one to explain language use more adequately. For example, the utterance ‘The tank is half empty’, can mean ‘we’d better stop and fill up before New York’, or ‘the fish will die if we don’t add some water soon’, or ‘where have Corporal Smith and Gunner Jones gone?’, or ‘I’m a pessimist’, depending on the context. 1 This range of potential, context-bound, inferable meanings demonstrates how linguistic analysis alone wholly underdetermines interpretation, but not that language is incapable of functioning. I will argue that relevance theory – a cognitive theory of language understanding in general, which includes an account of rhetoric and poetic effect – adequately accounts for everyday spoken language, and for both instrumental and literary modes of writing. Despite the differences between them, it will be seen that pragmatics and de Man’s theory of rhetoric are designed to do the same thing, since in demonstrating how linguistic meaning generally depends on context-related implicatures rather than on semantic and grammatical codes alone, linguistic pragmatics also accounts for the difference between – in terms that de Man borrowed from Walter Benjamin – das Gemeinte (what is meant) and die Art des Meinens (the way in which language means). Furthermore, relevance theory calls into question Heidegger’s notion that ‘language speaks man’. At the very least, it enables us to say that since utterances are only ever interpretive representations of thoughts, and since the linguistic analysis of an utterance very much underdetermines its interpretation, even if language speaks through man, man plays an extremely active role in interpreting what is spoken. What is the point of engaging with de Man’s account of language from a pragmatic perspective? Most students of literature are familiar with the various critical approaches developed between the late 1960s and the early 1980s – structuralist, semiotic, reader-response, psychoanalytic, Marxist, feminist, cultural materialist, and so on. Many of us feel that we can take them or leave them, and appropriate or ignore their insights as we see fit. Yet de Man’s form of rhetorical or ironic reading cannot coexist in a comfortable pluralism with other approaches, because if it is tenable, it irreversibly undermines all alternative approaches and paradigms. De Man’s account of reading, quite simply, cannot be ignored. Anyone wishing to propose a pragmatic account of

4

Paradigms of Reading

reading, or indeed any other account of interpretation, is obliged to engage with de Man’s theory of the impossibility of reading. It is such an engagement that I attempt in this book. De Man died of a brain tumour in 1983, five years after writing that ‘death is a displaced name for a linguistic predicament’ (Rhetoric, p. 81), leaving Derrida to point out the correspondence, in French, between tumeur and tu meurs (Derrida, 1985, p. 16). Yet his name reached a rather wider audience four years later, when the New York Times described him as ‘one of the most brilliant intellectuals of his generation’ and ‘the originator of a controversial theory of language that some say may place him among the great thinkers of his age’ (1 December 1987, pp. B1, B6). Unfortunately, the reason for the newspaper piece was the rediscovery of a large number of articles that the young de Man had written for collaborationist newspapers in occupied Belgium during the Second World War. De Man’s accounts of the aberrancy of all cognition and understanding, and the consequent impossibility of standing by, or feeling guilty about, our actions or statements necessarily gain increased significance from the rediscovery of his wartime journalism. Although the last decades have seen a blossoming of literary critical theories, criticism in general does not seem to have ‘improved’. One surely cannot tell a ‘Whiggish’ or teleological tale in which the interpretation of literature has gradually reached maturity. The problem with most literary theorists is that in a conscious move that cannot be compared to de Man’s account of a necessary enabling blindness,2 they wilfully select one element of the act of reading – that is language, the text, the world, the author or the reader – to the exclusion of some or all of the others, and push it to an extreme. In de Man’s case, it is the tropological nature of language that overwhelms everything else. In this book I am specifically concerned to contest de Man’s notions of the random, mechanical, arbitrarily or aberrantly referential, ironic and inhuman nature of language. I cannot accept that de Man’s apparently wilful neglect of all the pragmatic parameters of language use is the necessary blindness that enables his illuminating insights about rhetoric, since my concern is precisely to contest these ‘insights’. I cannot pretend that my arguments might function, like a Hegelian negation, to highlight the better parts of de Man’s theory, because what I want to do is to propose a pragmatic alternative to this theory. Pragmatics concerns the use and meaning of language in context. The term was first used by Charles W. Morris to name a third branch of semiotics, alongside syntax and semantics, which he defined as the study of ‘the relation of signs to their interpreters’ (1971, p. 43). More

Pragmatic Banality and Honourable Bigotry 5

recent definitions focus on communicators (speakers or writers) as well as their addressees, and the production and recovery of reference (particularly in regard to deictic or indexical words), propositional attitude, illocutionary force and implicatures, including such rhetorical devices as metaphor and irony, by which much more can be meant than is literally expressed. Although it has largely been applied to spoken communication, there are many elements of pragmatics that can, by analogy, be applied to literary texts and to the separate contexts of author and reader – separate because most readers of canonical literature are clearly not the author’s intended addressees. The active human agency taken for granted by pragmatics means that it is incommensurable not just with de Man’s account of rhetoric, but with the whole spectrum of post-Saussurean theories of language. Poststructuralist, deconstructionist and Lacanian theory all treat language and texts as active forces which function without any need for an originating consciousness, and treat both writers and readers as mere assemblages of texts and codes, so that the self is dissolved and its functions ascribed to the impersonal systems which operate through it. Saussure (1974, p. 66) described signification as an automatic imprinting, on passive human brains, of verbal stimuli that give rise to determinate concepts (defined in terms of differences). Derrida (1976, 1988) explodes the idea of determinate signifieds, claiming instead that signs are necessarily and essentially inscribed in a differential chain and therefore contain traces of other signs and concepts which (oddly) both supplement and erase or cancel out what is said by the signifier. He also argues that since signs can always be lifted from any context and grafted into new contexts which the author could not have anticipated, meaning can never be fixed, but is always disseminated or dispersed, in an impersonal or extrapersonal free play of signifiers, apparently regardless of any human volition or intervention. On this view, all readings are necessarily misreadings in that they fail to exhaust all the possibilities of the text. De Man, on the contrary, argues for the irrefutability of correct rhetorical readings which take account of the ‘materiality of the letter’, most notably the inevitable occurrence of tropes and ironies which, being incompatible with both grammar and the referential function of language, invariably undermine intended meaning. Relevance theory, however, sees the signifiers on a page or in an utterance as merely the starting point for an inferential process which, together with contextual information, should lead to the discovery of what a communicator intended to convey. Unlike, for example, the materiality

6

Paradigms of Reading

of architecture, in which matter and form necessarily coincide, and the matter is necessarily sufficient, the materiality of language is only the first step towards communication or cognition.

Synopsis I will first outline relevance theory as applied to spoken communication (Chapter 2), and then assess its explanatory power in relation to competing hermeneutic and literary critical theories (Chapters 3 and 4). I will then set out what might be called a traditional account of rhetoric, before going on to outline – and challenge, from the perspective of relevance theory – Paul de Man’s various accounts of the nature of language, specifically his theories concerning ‘cognitive rhetoric’ or the rhetoric of tropes (in Chapters 5 and 6), the relation between words and the world (Chapter 7), performative rhetoric (Chapter 8), and the inhuman nature of language and the tropological nature of selfhood (Chapter 9). In Chapter 10, I analyse a number of readings of a wellknown short poem by Wordsworth in terms of relevance theory. I will conclude by attempting to show that relevance theory is a more plausible (and useful) account of language than that offered by Paul de Man. Relevance theory argues that communication is successful not when hearers recognise an utterance’s linguistic meaning, but when they infer the speaker’s meaning from it. By extension, for literary communication to take place, a text must be inferentially combined with optimally relevant contextual assumptions – which are those envisaged by the writer. In this aspect at least, relevance theory clearly has many precursors, most notably the ‘romantic’ hermeneutics of Schleiermacher and Dilthey, which assumes that it is possible for a reader to transpose himself into an author’s linguistic context and to duplicate his mental and emotional state. However, as I show in Chapter 3, unlike the German hermeneutic tradition, and twentieth-century critics such as Poulet and Richards who equally posit the possibility of a reader recreating an author’s thoughts or experience, relevance theory only presupposes the possibility of creating a partially mutual cognitive environment. After all, as the ‘radical historicist’ Gadamer suggests, no interpreter can ever simply abandon his own horizon, so that the best that can be hoped for is a fusion of the interpreter’s and author’s horizons or contexts. In this chapter I also consider Dan Sperber’s account of the ‘epidemiology of representations’, which suggests that unlike, say, genes or viruses, mental representations tend to be transformed rather than replicated each time they are transmitted.

Pragmatic Banality and Honourable Bigotry 7

Right at the beginning of the hermeneutic tradition, however, in arguing for the necessity of a psychological reading, a projection into the author’s inner creative process, as well as mere grammatical reading, Schleiermacher also opened the door to applicative reading: interpreting the text according to our own beliefs or desires. Once the need to go beyond the words on the page is postulated, it is difficult to legislate for an end to this process. The attempt to reconstruct an author’s intentions, or to restore an original meaning, might be called a ‘positive hermeneutic’. Yet readers of canonical literature, separated from the author and the original intended audience by time and space and temperament, may well find that their interpretation is clearly not one that could have been shared by the original audience. After all, a working definition of a canonical book might be precisely, as Frank Kermode puts it, ‘that however a particular epoch or a particular community may define a proper mode of attention or a licit area of interest, there will always be something else and something different to say. . . . This is what it means to call a book canonical . . . all grant to the text something like omnisignificance. . . . Interpretations may be regarded not as modern increments but rather as discoveries of original meanings hitherto hidden’ (Kermode, 1985, pp. 62, 75). Kermode describes canonical texts as containing figural qualities that will only be detected at an appropriate moment in the future, so that rhetoric is an inexhaustible resource rather than, as de Man would have it, an insurmountable obstacle. Other readers may be simply uninterested in whether their interpretation somehow reconstructs a posited original meaning, because they do not believe in the authority of the author and prefer to pursue their own thematic readings, either in accordance with the linguistic potential of the text (and an unlimited number of ‘intertexts’), or with their own political, psychological or philosophical beliefs and interests, or because their interpretive strategies and assumptions are determined by the contingencies of their individual unconscious desires and defences. All these approaches, which I consider in Chapter 4, might be thought of as versions of a ‘negative hermeneutic’, which regards the text as merely a heuristic stimulus for the production of new meanings. Relevance theory, however, can account for such readings because – as Sperber and Wilson’s title Relevance: Communication and Cognition suggests – it is also a theory of cognition, designed to explain how individuals make sense of the phenomena around them (including literary texts), independently of anyone else’s intentions, according to a notion of maximal (rather than optimal) relevance. The principle of relevance

8

Paradigms of Reading

was originally defined (1986) in terms of acts of ostensive communication communicating the presumption of optimal relevance, with the theory of cognition, involving the notion of maximal relevance, being given rather less prominence. Yet in the second edition of Relevance (1995), Sperber and Wilson call the claim that ‘Human cognition tends to be geared to the maximisation of relevance’ the First (or Cognitive) Principle of Relevance, and the claim that ostensive communication communicates a presumption of its own optimal relevance the Second (or Communicative) Principle of Relevance.3 Importantly, however, relevance theory’s demonstration of the inferential nature of comprehension shows that (relevance-maximizing) interpretations based on potential meanings enabled by linguistic ‘codes’ are not in fact normative, and need not necessarily destabilise meanings perceived as intended or optimally relevant. Chapter 5 begins with what might be called a traditional account of rhetoric, which is essentially that it is language used to persuade. I draw on both standard sources such as Aristotle and Cicero, as well as contemporary writers such as Donald McCloskey, an economist and economic historian who is one of the leading recent apologists of rhetoric. I then discuss de Man’s theory of tropology, showing how he misreads Rousseau’s Second Discourse, translating the term idée générale as ‘metaphor’ rather than ‘common noun’, and contesting Nietzsche’s claim that since our words and truths are largely dead metaphors they are ‘coins which have lost their image and now can be used only as metal, and no longer as coins’. Since we use words as if they were coins, as their metaphorical origin has been forgotten, they still function, and constitute, as Nietzsche puts it, ‘a sum of human relations’, and so are not merely worn-out pieces of metal. Similarly, de Man’s description of number as a ‘conceptual metaphor devoid of objective validity’ and as ‘derivative and suspect’ is itself devoid of pragmatic consequences. In Chapter 6 I show that de Man’s claim that Nietzsche switched the study of rhetoric from persuasion to tropes and figures, and demonstrated that the paradigmatic structure of language is figural or rhetorical, rather than grammatical, referential, or performative, is untenable. De Man’s conclusions regarding the ‘insurmountable obstacle’ of rhetoric and the ‘catastrophic conclusions’ to which it gives rise cannot be said to derive from Rousseau and Nietzsche. I equally try to show the falsity of de Man’s argument that tropes are wholly cognitive and devoid of any aesthetic function, and call into question his accounts of irony, and ‘technically correct rhetorical readings’, by drawing attention to all the pragmatic aspects of language that his account of rhetoric neglects. 4

Pragmatic Banality and Honourable Bigotry 9

Relevance theory, which argues that semantic representations are abstract mental structures that must be inferentially enriched, takes for granted the non-coincidence of words and the world, and language and phenomenal experience. De Man, on the contrary, suggests that the lack of coincidence between words and things problematises the referential function of language. I contest this argument in Chapter 7, as well as de Man’s claim that because the relationship between words and things is conventional or contractual rather than phenomenal or constitutive, reference is always called into question by tropology. I equally dispute his assertion that grammar can only function by suspending reference, while texts in turn generate referents that subvert grammar, by reference to the hermeneutic principle of application, as developed by Gadamer. I also question de Man’s reading of Hegel’s remarks on deictics (words whose reference changes each time they are used), and his belief that this provides further evidence of the instability of the referential function of language. Wlad Godzich considers the deictic nature of demonstratives in his Foreword to de Man’s The Resistance of Theory, mentioning that ‘Many utterances can be evaluated in terms of their claims to truthfulness only when this deictic dimension is brought into consideration. Neither the logical nor the grammatical structure of a sentence suffices for these purposes; they may only provide assurances of well-formedness but they are helpless to determine whether the sentence is true with respect to a certain state of affairs’ (Resistance, p. xvi). Yes indeed. One might also add that it is not only the truth but the very meaning of the sentence – exactly which ‘state of affairs’ is referred to by the use of he, there, then, and so on – that is determined by the attribution of referents to deictic expressions, and that relevance theory and most other schools of pragmatics take it for granted that the logical and the grammatical structure of a sentence also fail to explain most other aspects of meaning. Yet Godzich describes ‘the indicational capability of deixis’ and ‘the referential capacity of language’ as a longstanding linguistic ‘problem’, that Benveniste’s account of the first and second person pronouns and Jakobson’s account of ‘shifters’ seem to sidestep, and suggests that deictics refer only to the instance of discourse, or to the fact that language has taken place. For most pragmaticians, on the contrary, deictics do much more: they refer to particular context-dependent referents, and as such, are perfectly functional and non-problematic. De Man approaches deictics through Hegel’s brief discussion of them in paragraph 20 of the first volume of the Enzyklopädie der philosophischen

10

Paradigms of Reading

Wissenschaften. I show, however, that de Man inverts Hegel’s argument that language can only express universal truths, so we can only think what we can say, interpreting him instead as arguing that our thoughts have an ineffable content that language is not able to express. In Chapter 8 I consider de Man’s account of the ‘performative rhetoric’ generated by the random and mechanical nature of language, a notion he illustrates with reference to Rousseau. I show how de Man’s description of language as a machine (which has the incidental effect of abolishing the notion of guilt), rests on an extremely dubious reading (and re-writing) of Rousseau, in which he turns his confessions or apologies into excuses or justifications by adding a negative to a crucial sentence, and interprets the adjective machinal as machine-like rather than instinctive, allowing him to construct a general theory in which language is mechanical, arbitrary, gratuitous and irresponsible, unconstrained by human cognition or reference and detached from meaning, intentions and signifieds. Since the rediscovery and re-publication of de Man’s wartime journalism, it has been widely argued that his abolition of the notion of guilt in the essay on the Confessions is related to the opinions he expressed in collaborationist newspapers in occupied Belgium. Consequently I end Chapter 8 with a brief account of de Man’s early writings, and a pragmatic response to Wlad Godzich’s claim that ‘the artifactness of language’ and our awareness of the falsehood of all utterances makes it impossible for us to stand by (or feel guilty about) our actions or statements. In Chapter 9 I contest the general post-Saussurean and Heideggerean notion that language signifies by itself, and that texts are active, while human beings dissolve into various impersonal linguistic systems. I argue that Saussurean linguistics is quite simply inadequate, and needs to be replaced by a model that recognises the active nature of linguistic production and understanding. I suggest that de Man’s particular extrapolation from Saussure depends on an unwarranted and untenable slippage from the uncontested fact that the signifiers in any particular language are arbitrary, to the highly contestable notion that signification itself (the use of signs by specific language users in specific, conventional, pragmatic contexts) is the result of the arbitrary positing power of language itself. The New York Times article referred to above included a lamentably inaccurate one-sentence definition of deconstruction such that it ‘views language as a slippery and inherently false medium that always reflects the biases of its users’. De Man’s account of language is in fact quite the opposite: it is a slippery, rhetorical medium that

Pragmatic Banality and Honourable Bigotry 11

prevents language users reliably reflecting their ideas, biases or anything else. De Man writes of the positing power of language, and invokes ‘the madness of words’ which ‘no degree of knowledge can ever stop’ (Rhetoric, p. 122), but even though poetic language, which uses words for their sound as well as their sense, is evidently dependent on the arbitrary phonetic properties of words in a given language, these phonetic possibilities are still used or rejected by particular language users. I finish this chapter with a further example of the active role of the speaking subject – a consideration of the various ways in which writers can report and paraphrase the ideas of others, and how de Man takes advantage of these linguistic mechanisms to rewrite Rousseau, Hegel and Nietzsche. In Chapter 10 I illustrate relevance theory’s account of cognition and maximal relevance with reference to a dozen published interpretations (including de Man’s) of a very well-known lyric poem, Wordsworth’s ‘A slumber did my spirit seal’. This is followed by a brief conclusion in which I reiterate my reasons for claiming that relevance theory – which, like de Man’s theory of rhetoric, takes the non-coincidence of linguistic signs and intended or interpreted meanings as its starting point – is the more plausible description of the nature of language, since it seems to account for the data without needing to be buttressed by very debatable reinterpretations or rewritings of Rousseau, Hegel and Nietzsche.

Honourable bigotry Is anyone amenable to or convinced by de Man’s account of language likely to be interested in a pragmatic counter-theory such as Sperber and Wilson’s? People can perhaps be divided into two groups: those who divide people into two groups, and those who don’t. Yet regarding theories of language, such a division can certainly be made. As George Craig has put it, succinctly, in a book review: There may be no limit to the number of ways of talking or writing about language, but in practice they fall into one of two groups: what one might call the classical and romantic. What unites the first group is the conviction that language can be understood, and therefore explained; or, putting it more bluntly, that humans are in control. For the second group, language forever eludes our understanding; is indeed that which best reveals the limitations of our understanding. For one, unitary theories vie with each other; for the other, all theories

12

Paradigms of Reading

offer at best an interesting sidelight, a partial grasp. We ourselves, moreover, will tend (though not, on the whole, by deliberate choice) to belong to one group or the other. (Craig, 1995, p. 23) According to this schema, it is clear that my pragmatic opposition to de Man’s description of language as random, mechanical and inhuman places me in the ‘classical’ group. (I also like to think that this is indeed ‘by deliberate choice’.) Consequently, I contest the use of ‘we’ in de Man’s statements concerning ‘the duplicity, the confusion, the untruth that we take for granted in the everyday use of language’ (Blindness, p. 9), or how ‘if we start from a pluralistic perspective, we can never achieve a genuine reading’ (de Man, 1974, p. 50), or the way ‘we write in order to forget our foreknowledge of the total opacity of words and things’ (Allegories, p. 203). It would of course be more accurate to read ‘we’ in these sentences as referring only to those people persuaded of le néant des choses humaines (the nothingness of human matters, a phrase from La Nouvelle Héloïse that de Man quotes in no fewer than seven different essays). 5 Or to Heideggereans who lament ‘the barrenness of our condition’ and ‘the false texture of everyday existence’ (Romanticism, pp. 144, 145), and who dismiss as ‘inauthentic’ the empirical or ontic ‘self that lives in the world of material and pragmatic substances’ (Romanticism, p. 36), and instead seek ontological truth and valorize ‘the authentic self engaged in the interpretation of its own being-there and capable of comprehending the truth of its own fallen destiny’ (Blindness, p. 44). Unfortunately discussion between deconstructionists and pragmatic linguists rarely seems to live up to the conversational standards Habermas (1975) describes as constituting a Sprachethik: don’t lie, pay attention, don’t sneer, cooperate, don’t shout, let other people talk, be open-minded, explain yourself when asked, and so on. Pragmaticians tend to dismiss deconstructionists outright, since they wilfully disregard virtually all linguistic theory apart from Saussure, Jakobson and Wittgenstein. De Man, on the contrary, believed that taking the beliefs and intentions of language users into account merely ‘opens up dreary prospects of pragmatic banality’ (Resistance, p. 19), and his followers tend to dismiss pragmatic linguistics as wholly inadequate to the complexities of textuality. This reaction reminds me of the cardinals who refused to look through Galileo’s telescope because they knew that the sun goes round the earth, as both their eyes and scripture told them so. This analogy is of course unfair, as de Man’s account of language can

Pragmatic Banality and Honourable Bigotry 13

hardly be described as Ptolemaic, yet as Alasdair Fowler (1989, p. 16) has suggested, ‘like Copernicus’s De Revolutionibus, [relevance theory] may take a considerable time to win recognition’. I am also reminded of an anecdote about Wittgenstein related in Tom Stoppard’s play Jumpers: Meeting a friend in a corridor, Wittgenstein said: ‘Tell me, why do people always say it was natural for men to assume that the sun went round the earth rather than that the earth was rotating?’ His friend said, ‘Well, obviously, because it just looks as if the sun is going round the earth.’ To which the philosopher replied, ‘Well what would it have looked like if it had looked as if the earth was rotating?’ (Stoppard, 1972, p. 75) What would language look like if it merely permitted interpretive representations of thoughts, and meanings were generally inferred rather than simply encoded and decoded, and (except in poetry) ‘the materiality of the letter’ was frequently no more than a lure or an irrelevance? Perhaps the gulf between the two positions is simply so wide that it constitutes what Lyotard (1988) describes as a differend – a disagreement in which the two sides have such radically incommensurable presuppositions or perspectives or vocabularies that the very point of difference cannot even be expressed in either ‘language game’ without prejudging the issue. Deconstructionists, after all, choose to disregard the constraints of intentionality, meaningfulness and context, and prefer to look for all the discordant signifying elements that pragmaticists conventionally disregard.6 They generally reject pragmatic arguments as inadequate to what they see as the material, rhetorical, nonphenomenal and nonperceptual nature of language. As Barbara Johnson has put it: A deconstructive reading is an attempt to show how the conspicuously foregrounded statements in a text are systematically related to discordant signifying elements that the text has thrown into its shadows or margins; it is an attempt both to recover what is lost and to analyze what happens when a text is read solely in function of intentionality, meaningfulness, and representativity. Deconstruction thus confers a new kind of readability on those elements in a text that readers have traditionally been trained to disregard, overcome, explain away, or edit out – contradictions, obscurities, ambiguities,

14

Paradigms of Reading

incoherences, discontinuities, ellipses, interruptions, repetitions, and plays of the signifier. In this sense it involves a reversal of values, a revaluation of the signifying function of everything that, in a signified-based theory of meaning, would constitute ‘noise’. (Johnson, 1989, pp. 17–18) My argument is that far from needing to be ‘trained to disregard’ certain ‘plays of the signifier’, we tend to do so automatically, with both texts and utterances, given the inferential nature of interpretation. Once it is accepted that encoding and decoding are merely elements in a more inclusive process of inference, it becomes clear that there are generally good reasons for editing out many of the potential meanings that the signifying code allows – and bad reasons, given that their teasing out can enable interesting textual interpretations. The irony of my position, of course, is that the ideas in de Man’s essays I am concerned to counter are not ‘discordant signifying elements’ in the text’s ‘shadows or margins’, but rather ‘conspicuously foregrounded statements’. Making wholly extravagant claims was a central characteristic of de Man’s style. Since it is his central and largely unambiguous claims that I find impossible to ‘disregard, overcome, explain away, or edit out’, I cannot make him the object of the kind of reading that Thomas Kuhn says he suggests to his students: When reading the works of an important writer, look first for the apparent absurdities in the text and ask yourself how a sensible person could have written them. When you find an answer . . . when these passages make sense, then you may find that more central passages, ones you previously thought you understood, have changed their meaning. (Kuhn, 1977, p. xii) Don Bialostosky has acknowledged the difficulty of displacing people’s identification with and material investment in one hero or position to another. He warns that ridiculing critics’ established and secure positions is unlikely to be successful, and even revoicing their characteristic ideas ‘probably will not change their minds and will only have effect on others less fully invested in those interests and consequently more able to entertain other interests and admire other heroes’ (1992, pp. 33–4). Furthermore, Bialostosky counsels, no one ‘should expect anyone else to abandon a productive investment solely on grounds of its alleged inadequacy to the facts, without supplying hope, promise even, of a

Pragmatic Banality and Honourable Bigotry 15

more productive investment’ (1992, p. 34). He quotes from the end of the Preface to Lyrical Ballads where Wordsworth writes that ‘all men feel an habitual gratitude, and something of an honourable bigotry for the objects which have long continued to please them: we not only wish to be pleased, but to be pleased in that particular way in which we have become accustomed to be pleased’ (Wordsworth, 1992, p. 760). Thus I hardly expect this book to provoke Damascene conversions among deconstructionists. Yet I still consider it necessary to show, from a pragmatic perspective, the inadequacy of de Man’s account of rhetoric to the nature of language, and to counter his insistence that ‘technically correct rhetorical readings may be boring, monotonous, predictable and unpleasant, but they are irrefutable’ (Resistance, p. 19). They are only technically correct according to an inadequate linguistic model that wilfully brackets factors such as intention, inference, implicature, relevance and referential constraint so as to avoid ‘pragmatic banality’.

2 Relevance Theory and Spoken Communication

from birth onwards, humans expect relevance from the sounds of speech (an expectation often disappointed, but hardly ever given up). (Sperber, 1996, p. 114) Linguistic pragmatics takes into account the determining role of the hearer or reader in linguistic communication, and demonstrates how meaning – whether explicit or implicit – always depends on context-related inference as well as semantic and grammatical codes. In particular, Dan Sperber and Deirdre Wilson’s relevance theory argues that utterances are only interpretive representations of thoughts, so that there is necessarily a gap between the semantic representations of sentences and the thoughts actually communicated by utterances, a gap that can be filled by a hearer or reader’s inferential recognition of a communicator’s intentions, guided by contextual clues. The very notion that sign and meaning could ever coincide is based on the dominant but erroneous code model which assumes that meanings are encoded and decoded, and disregards the enormous role of (unconscious, non-demonstrative) inference in linguistic communication. Although they describe the principle of relevance as ‘an exceptionless generalisation about human communicative behaviour’ (Wilson and Sperber, 1988, p. 144), Sperber and Wilson (hereafter S&W) largely concentrate on spoken communication. In this chapter I will introduce the theory, quoting extensively from S&W. In the following chapters, I will consider relevance theory in relation to literary interpretation. 16

Relevance Theory and Spoken Communication

17

Linguistic underdetermination S&W argue that the semantics of natural languages are too weak to encode thoughts, and semantic representations are incomplete logical forms that can only be, at best, fragmentary representations of thoughts. Thus ‘languages do not encode the kind of information that humans are interested in communicating. Linguistically encoded semantic representations are abstract mental structures which must be inferentially enriched before they can be taken to represent anything of interest’ (Relevance, p. 174). Hearers and readers generally begin their interpretation of an utterance or a written sentence by assigning to it a unique propositional form. This involves pragmatic inferences that build on the incomplete conceptual representation decoded from the linguistic form used in the utterance. Hearers typically have to disambiguate the utterance by selecting one of the possible semantic representations permitted by the grammar; complete the utterance if it is in any way elliptic or semantically incomplete; identify the referents of all referring expressions, including the deictic ones concerning time, person and place; enrich any vague terms until they are semantically explicit; and loosen the meaning of any concepts used non-literally. As S&W point out, ‘By definition, the semantic representation of a sentence, as assigned to it by a generative grammar, can take no account of such non-linguistic properties as, for example, the time and place of utterance, the identity of the speaker, the speaker’s intentions, and so on. . . . The semantic representation of a sentence deals with a sort of common core of meaning shared by every utterance of it. However, different utterances of the same sentence may differ in their interpretation; and indeed they usually do’ (Relevance, p. 9). But semantic representations of sentences are mental objects that never surface to consciousness; we merely use them as assumption schemas to identify the propositional form of an utterance. Once a semantic representation has been selected, completed and enriched or loosened in various ways to yield a propositional form, a hearer has to identify an utterance’s propositional attitude. This is because ‘the same sentence, used to express the same thought, may sometimes be used to present this thought as true, sometimes to suggest that it is not, sometimes to wonder whether it is true, sometimes to ask the hearer to make it true, and so on. Utterances are used not only to convey thoughts, but to reveal the speaker’s attitude to, or relation to, the thought expressed: in other words, they express “propositional

18

Paradigms of Reading

attitudes”, perform “speech-acts”, or carry “illocutionary force”’ (Relevance, pp. 10–11). Thus linguistic pragmatics as a whole and relevance theory in particular take it for granted that utterances are generally full of semantic and referential ambiguities, and open to innumerable figurative interpretations. Sentences tend not to correspond very closely to thoughts, and even if an utterance does explicitly convey one thought, it might very well implicitly convey others. S&W see nothing paradoxical in insisting that ‘the utterance of a sentence, which provides only an incomplete and ambiguous representation of a thought, can nevertheless express a complete and unambiguous thought’ (Relevance, p. 34), and that ‘although the linguistic analysis of an utterance very much underdetermines its interpretation, the most striking feature of linguistic communication is that it can achieve a degree of precision and complexity rarely achieved in non-verbal communication’ (p. 174). This is because the gap between semantic representations and the thoughts that utterances are intended to communicate can be filled by perfectly ordinary inferential abilities that code-based theories of language bizarrely disregard. 1

Ostension, inference, intentions and relevance Verbal communication clearly does generally exploit linguistic coding and decoding. Hearers and readers can sometimes recover specific or determinate messages associated to signals by an underlying code. Most people will both decode and inference effortlessly and identically if someone shouts ‘Fire!’ in a crowded theatre – but less so if someone shouts ‘Theatre!’ in a crowded fire.2 Disambiguation, reference assignment, and the recovery of implicatures, propositional attitudes, illocutionary forces and figurative interpretations do not proceed on the basis of a code that automatically integrates properties of the context with semantic properties of the ‘message’, but depend on the interaction of linguistic meaning with contextual factors. The linguistic meaning of an uttered sentence falls well short of encoding what the speaker means: it is merely the input to the inferential processes by which the speaker’s intentions are recognised. Relevance theory takes as a starting point Paul Grice’s (1957) definition of meaning in terms of intention. Hence the coding–decoding process is subservient to a Gricean inferential process, and hearers treat the output of decoding as a piece of evidence about the communicator’s intentions. As S&W put it, ‘Hearers are interested in the meaning of the

Relevance Theory and Spoken Communication

19

sentence uttered only insofar as it provides evidence about what the speaker means. Communication is successful not when hearers recognise the linguistic meaning of the utterance, but when they infer the speaker’s “meaning” from it. . . . According to the inferential model, communication is achieved by the communicator providing evidence of her intentions and the audience inferring her intentions from the evidence’ (Relevance, pp. 23–4). (S&W always use ‘she’ for the speaker and ‘he’ for the hearer.) S&W describe this as ‘ostensive–inferential communication’. Ostension involves providing evidence of one’s intention to convey a thought or some information, that is, making manifest to an audience an intention to make certain information manifest (making it mutually manifest that an ostensive stimulus is being produced). In other words a speaker has both an informative intention – to inform the hearer of something – and a communicative intention – to inform the hearer of her informative intention. (It is of course quite possible for the latter intention to be fulfilled but not the former: communicators do not necessarily succeed in producing optimally relevant stimuli, even though they intend their audience to believe that they do.) Ostension is thus the other side of the coin of inference: ‘Inferential communication and ostension are one and the same process, but seen from two different points of view: that of the communicator who is involved in ostension and that of the audience who is involved in inference’ (Relevance, p. 54). S&W’s central argument is that ostension generally succeeds because it carries a presumption of relevance. To communicate is to claim an audience’s attention, hence to imply that the information communicated is relevant, and people’s attention and thought processes automatically turn toward information that seems relevant, which is to say capable of yielding cognitive effects. This leads to S&W’s two basic principles. The First (or Cognitive) Principle of Relevance is that ‘Human cognition tends to be geared to the maximisation of relevance’ (Relevance, p. 260). The Second (or Communicative) Principle of Relevance is that ‘Every act of ostensive communication communicates a presumption of its own optimal relevance’ (p. 158). 3 S&W take it as given that human communication processes are geared to achieving the greatest possible cognitive effect for the smallest possible processing effort. Hearers and readers automatically focus their attention on what seems to them to be the most relevant information available. For information to be relevant it must have cognitive or contextual effects, of which there are essentially three types – the derivation of new assumptions or representations of the world, and either

20

Paradigms of Reading

the strengthening or contradiction and elimination of existing ones – and ‘other things being equal, the greater the contextual effects the greater the relevance’ (Relevance, p. 119). Furthermore, processing effort is a negative factor, so that ‘other things being equal, the greater the processing effort, the lower the relevance’ (p. 124). In other words, people try to obtain from each new item of information as great a contextual effect as possible for the least possible processing effort. Optimal relevance involves obtaining contextual or cognitive effects without any gratuitous processing effort. S&W specify the nature of addressees’ cognitive environments and the contexts that they construct in the pursuit of maximal relevance. A cognitive environment consists of a set of assumptions that the individual mentally represents and accepts as true (or probably true), as well as all the facts or assumptions he is capable of becoming aware of at that time and place. A context is a psychological construct, a subset of a hearer’s background assumptions about the world, including cultural assumptions and beliefs, memories, expectations about the future, the current physical environment, immediately prior utterances, beliefs about the mental state of the speaker, and so on. People are assumed to have a memory store containing representations that are treated by the mind as true descriptions of the actual world. People’s assumptions are structured sets of smaller constituents that S&W describe as concepts: triple entries stored together in the memory. There is a logical entry, consisting of a set of deductive rules which apply to the logical forms of which that concept is a constituent; an encyclopaedic entry containing information about the extension and/or denotation of the concept, that is about the objects, events or properties which instantiate it; and a lexical entry containing information about the word or phrase of natural language that expresses the concept. Logical entries are of course small, finite, and relatively constant, while encyclopaedic entries are open-ended and change among people and across time. In the search for maximal relevance, the mind picks out the information that has the greatest relevance to the initial context from accessible extensions of the context, from long-term encyclopaedic memory, or the immediate perceptual environment. Newly communicated information has to be combined with selected background assumptions, which then constitute the context of interpretation. This means that every new utterance requires a slightly different context, because the interpretation of the previous utterance has just become part of the context. Thus it is relevance which is treated as given, and context which is treated as variable: hearers assume that utterances are designed to be

Relevance Theory and Spoken Communication

21

optimally relevant (that a communicator wishes to alter their cognitive environment), and try to select a context which will maximize relevance, that is one that enables the best possible balance of effort against effect to be achieved. At every stage in disambiguation, reference assignment and enrichment, hearers automatically choose the solution involving the least effort, and abandon this solution only if it fails to yield an interpretation consistent with the principle of relevance, which is to say ‘an interpretation that the communicator could manifestly have expected to be optimally relevant’ (Relevance, p. 169). This is not, however, ‘an angel theory of language’ 4 which assumes that communication always succeeds: S&W stress that ‘to be consistent with the principle of relevance, an interpretation does not actually have to be optimally relevant to the addressee; it must merely have seemed so to the communicator. Conversely, the first optimally relevant interpretation may happen to be relevant in a way the communicator could not have foreseen; in this case it is not consistent with the principle of relevance’ (p. 169). Equally, an addressee might fail to come up with a single coherent interpretation: grammar and inferential possibilities make successful communication possible, but obviously don’t guarantee it. It is an important part of the theory that both communicator and addressee can only expect one interpretation to be optimally relevant. Although every utterance has a variety of possible interpretations, all of them compatible with the linguistically encoded information, not all of them will be equally accessible to the hearer on any given occasion. The communicator, of course, needs to make correct assumptions about the codes and contextual information (and cognitive abilities) that the audience will have accessible and be likely to use in the comprehension process. Ideally, she will use a stimulus that saves addressees the effort of accessing more than one hypothesis consistent with the principle of optimal relevance; that is the hearers’ first inferential hypothesis (the first interpretation that meets the criterion of relevance) should be the good one, so that there is no need to consider a second one. S&W’s ostensive–inferential model is thus a considerable simplification of Grice’s account of conversational implicatures, and his argument that in normal communication, speaker and hearer abide by a cooperative principle that can be sub-divided into maxims of conversation, namely those of quality, quantity, relation and manner (which is to say that in order to converse in a maximally efficient, rational and cooperative way speakers should speak sincerely, relevantly and clearly, while providing sufficient information) (Grice, 1967/1987). Whereas in Gricean theory both communicators and audience must know the

22

Paradigms of Reading

norms underlying the conversational maxims in order to communicate adequately, and can deliberately flout them in order to generate implicatures, S&W insist that the principle of relevance does not have to be known or learnt: it is automatically triggered by every utterance, so communicators could not violate it even if they wished to.

Implicatures, vagueness and poetic effects S&W assert that, due to the principle of relevance, ‘One of the advantages of verbal communication is that it gives rise to the strongest possible form of communication; it enables the hearer to pin down the speaker’s intentions about the explicit content of her utterance to a single, strongly manifest candidate with no alternative worth considering at all’ (Relevance, p. 60). A communicator whose informative intention involves making a particular assumption strongly manifest will strongly or explicitly communicate that assumption. An ostensively communicated assumption which is inferentially developed from the (incomplete) conceptual representation or logical form resulting from decoding is an explicature. But although the use of language can achieve explicit communication, this ‘is not a typical but a limiting case’ (Relevance, p. 55). Indeed, ‘In many – perhaps most – cases of human communication, what the communicator intends to make manifest is partly precise and partly vague’ (p. 59), and for S&W, it is ‘a major challenge for any account of human communication to give a precise description and explanation of its vaguer effects. Distinguishing meaning from communication, accepting that something can be communicated without being strictly speaking meant by the communicator or the communicator’s behaviour, is a first essential step’ (pp. 57–8). S&W suggest that communicators’ informative intentions are not so much to induce specific thoughts in addressees, as to modify their cognitive environment. If one knows an individual’s cognitive environment – which one does to the extent that it is mutual – one can infer which assumptions he is likely to entertain. Successful communication is ‘a matter of enlarging mutual cognitive environments’ (Relevance, p. 193), and does not require an audience to duplicate exactly a communicator’s thoughts. Unlike the code model, which requires addressers and addressees to have mutual knowledge or assumptions, an inferential model can explain communication and comprehension in terms of mutual manifestness (to be manifest is of course weaker than known or assumed) and mutual cognitive environments.

Relevance Theory and Spoken Communication

23

As well as explicatures, speakers can implicitly communicate weak assumptions or implicatures. An implicature is defined as ‘a contextual assumption or implication which a speaker, intending her utterance to be manifestly relevant, manifestly intended to make manifest to the hearer’ (Relevance, pp. 194–5). Implicatures are not recovered by the linguistic decoding of semantic representations, but ‘by reference to the speaker’s manifest expectations about how her utterance should achieve optimal relevance’ (p. 194). Implicatures are thus derived solely by way of pragmatic inference. Some implicatures are made so strongly manifest that the hearer can scarcely avoid recovering them. For example, if you bang on the door and yell ‘Oi!’ at a teenager who has been locked in the bathroom for forty-five minutes or who has just turned his bedroom stereo on at full blast, the hearer (if indeed he hears) will not so much complete the propositional form of the utterance as recover the obvious (but not coded) implicature that you would like him to get out of the bathroom or turn the music down. In these cases, speakers aiming at optimal relevance leave implicit everything their hearers can be expected to supply with less effort than would be needed to process an explicit prompt. Other implicatures are deliberately made less strongly manifest, and are designed to marginally increase the manifestness of a wide range of weakly manifest assumptions, that is to make small and not entirely foreseeable changes in a hearer’s cognitive environment, perhaps by nudging his thoughts in a certain direction: ‘the hearer can often fulfil part of the speaker’s informative intention by forming any of several roughly similar but not identical assumptions’ (p. 60). The hearer need only pay attention to some of these weaker implicatures for the relevance of the intended interpretatin to become manifest. For example, if someone says ‘John is a soldier’ when we know that he is not literally a member of the armed forces, we might interpret that he is very dutiful, loyal, obedient, a selfsacrificing team-player, and so on. No two hearers need to access exactly the same elements of their encyclopaedic entry for ‘soldier’ to understand the speaker’s general intention.5 The same applies to literary metaphors such as the oft-quoted examples ‘Achilles is a lion’ and ‘Juliet is the sun’. S&W explain stylistic and poetic effects in general in terms of weak implicatures (and the pursuit of relevance): ‘Let us give the name poetic effect to the peculiar effect of an utterance which achieves most of its relevance through a wide array of weak implicatures’ (Relevance, p. 222). Stylistic devices (parallelisms, recurrences, oppositions, metre, rhyme, assonance, alliteration, and so on) and tropes can have rich poetic effects, leading to an unpredictable and diversified expansion of the context by way of a range of weak implicatures. This is especially the

24

Paradigms of Reading

case with devices like poetic metaphors, which can generate so many weak implicatures that after a certain point they can no longer be described as intended, and the hearer or reader must be considered responsible for choosing them: In general, the wider the range of potential implicatures and the greater the hearer’s responsibility for constructing them, the more poetic the effect, the more creative the metaphor. A good creative metaphor is precisely one in which a variety of contextual effects can be retained and understood as weakly implicated by the speaker. In the richest and most successful cases, the hearer or reader can go beyond just exploring the immediate context and entries for concepts involved in it, accessing a wide area of knowledge, adding metaphors of his own as interpretations of possible developments he is not ready to go into, and getting more and more very weak implicatures, with suggestions for still further processing. The result is a quite complex picture, for which the hearer [sic] has to take a large part of the responsibility, but the discovery of which has been triggered by the writer. The surprise or beauty of a successful creative metaphor lies in this condensation, in the fact that a single expression which has itself been loosely used will determine a very wide range of acceptable weak implicatures. (Relevance, pp. 236–7) A famous example of a line that has triggered numerous weak implicatures in the minds of literary critics is Shakespeare’s comparison, in the 73rd sonnet, of old age to that time of year When yellow leaves, or none, or few, do hang Upon those boughs which shake against the cold, Bare ruined choirs, where late the sweet birds sang. At the beginning of Seven Types of Ambiguity, William Empson (1972) uses the line about the choirs as an example of the basic type of ambiguity, which ‘covers almost everything of literary importance’, but his reading equally demonstrates the notion of weak implicatures. Empson says that this line is effective in several ways at once because ruined monastery choirs are places in which to sing, because they involve sitting in a row, because they are made of wood, are carved into knots and so forth, because they used to be surrounded by

Relevance Theory and Spoken Communication

25

a sheltering building crystallized out of the likeness of a forest, and coloured with stained glass and painting like flowers and leaves, because they are now abandoned by all but the grey walls coloured like the skies of winter, because the cold and Narcissistic charm suggested by choir-boys suits well with Shakespeare’s feeling for the object of the Sonnets, and for various sociological and historical reasons (the protestant destruction of monasteries; fear of Puritanism), which it would be hard now to trace out in their proportions; these reasons, and many more relating the simile to its place in the Sonnet, must all combine to give the line its beauty, and there is a sort of ambiguity in not knowing which of them to hold most clearly in mind. (Empson, 1972, p. 21) In relevance-theoretic terminology, this involves taking the basic metaphors – the comparisons of old age to a forest in winter, and a forest to a church – and extending the concept of ‘church’ to include choir boys and monasteries, and extending the encyclopaedic entry of monastery to include their destruction by the protestants, and so on, until the original metaphors have triggered a multitude of weak implicatures which are the result of cognitive activity on the part of the reader. This account of the reader taking responsibility for recovering weak implicatures is an important advance on Grice’s account of ‘speaker meanings’. Grice wrote that ‘A meant something by x’ is (roughly) equivalent to ‘A intended the utterance of x to produce some effect in an audience by means of the recognition of this intention.’ (Grice, [1957] 1989, p. 220) where A is the speaker and x is an utterance or any other form of communicative behaviour, including writing. S&W’s explanation of the functioning of poetic effects allows for much more indeterminacy in intended interpretations, and shows how the audience can construct their own implicatures and metaphors, as communication shades off into cognition.6

Metaphor and irony, description and interpretation S&W point out that according to this logic of weakly implicating a variety of cognitive or contextual effects,

26

Paradigms of Reading

metaphor and a variety of related tropes (e.g. hyperbole, metonymy, synecdoche) are simply creative exploitations of a perfectly general dimension of language use. The search for optimal relevance leads the speaker to adopt, on different occasions, a more or less faithful interpretation of her thoughts. The result in some cases is literalness, in others metaphor. Metaphor thus requires no special interpretive abilities or procedures: it is a natural outcome of some very general abilities and procedures used in verbal communication. (Relevance, p. 237) Furthermore, irony and a variety of related tropes (e.g. meiosis, litotes) fall together with a range of cases which would not normally be regarded as figurative at all. What unites these cases is the fact that the thought of the speaker which is interpreted by the utterance is itself an interpretation. It is an interpretation of a thought of someone other than the speaker (or of the speaker in the past). That is, these utterances are second-degree interpretations of someone else’s thought. . . . If we are right, then the same is true of irony as is true of metaphor: whatever abilities and procedures are needed to understand it are independently needed for the interpretation of quite ordinary non-figurative utterances. (Relevance, pp. 237–8) More specifically, ironic utterances are examples of echoic interpretation: they inform the hearer of the fact that the speaker is echoing a thought attributed to someone else (or merely to a type of person, or to people in general) while dissociating herself from it by way of an attitude that is manifestly sceptical or amused or triumphant or disapproving, and so forth.7 This account of the ordinariness of metaphor and irony makes it clear that S&W reject the notion of ‘literary language’ as a specifically metaphorical or ironic category, so that relevance theory is equally applicable to literary interpretation. S&W’s analysis of irony depends on a further fundamental distinction between descriptive and interpretive uses of language. Since, as outlined above, natural languages do not allow us to encode most of our thoughts, utterances are necessarily interpretive representations of thoughts. Importantly, however, S&W point out that an utterance can represent a thought in two different ways:

Relevance Theory and Spoken Communication

27

Any representation with a propositional form, and in particular any utterance, can be used to represent things in two ways. It can represent some state of affairs in virtue of its propositional form being true of that state of affairs; in this case we will say that the representation is a description, or that it is used descriptively. Or it can represent some other representation which also has a propositional form – a thought, for instance – in virtue of a resemblance between the two propositional forms; in this case we will say that the first representation is an interpretation of the second one, or that it is being used interpretively. (Relevance, pp. 228–9) Thus assertions involve a descriptive relation between the speaker’s thought and a state of affairs in the world, while requests or advice involve a descriptive relation between the speaker’s thought and a desirable state of affairs. Metaphor and irony, on the contrary, involve interpretive relations: in the case of metaphor between the propositional form of an utterance and the thought it represents, in the case of verbal irony (as opposed to a verbalisation of situational irony) between the speaker’s thought and an attributed thought or utterance of someone else. 8 Contrary to a widespread belief, descriptive representations or assertions are in no way paradigmatic. S&W argue that although it is widely recognized that utterances can be based on a variety of types of resemblance – for example onomatopoeia is based on resemblances in sound, verbal mimicry on resemblances of phonetic and phonological form, direct quotation and parody on resemblances in syntactic and lexical form, translation on resemblances in logical form, and paraphrase and summary on resemblances in propositional form – the extent to which utterances may only interpretively represent the speaker’s thoughts or interpret someone else’s thoughts, is generally disregarded:9 Most pragmatists and philosophers of language take for granted that there is a convention, principle or presumption, to the effect that the meaning of the utterance must be a literal expression, i.e. an identical reproduction, of a thought of the speaker’s. We believe that this claim is too strong. People certainly do not express themselves literally all the time, and when they do not, there is no intuition that a norm has been transgressed. There is thus no empirical evidence for a convention of literalness or anything of the sort.10 (Relevance, p. 230)

28

Paradigms of Reading

Literalness – when a thought or utterance shares all the implications of another thought or utterance that it represents – is just a special case of interpretive resemblance, as ‘all interpretations start loosely; some may then be tightened to the point of literalness. By the criterion of consistency with the principle of relevance, an utterance will be understood as loose or metaphorical unless nothing less than a fully literal interpretation will do’ (Wilson and Sperber, 1988, p. 144). The presumption of literalness is yet another consequence of the erroneous code model. S&W’s account of the continuity between both metaphor and irony and other perfectly ordinary forms of language use makes it unnecessary to describe poetic language as parasitic upon the normal, literal use of language, as Austin (1962) did, or as a flouting of the maxim of quality, as in Grice’s account of conversational implicatures. S&W’s account of metaphor and irony as two types of interpretive use has another important consequence that is quite the contrary of Paul de Man’s inflation of the notion of ‘trope’ to cover just about every dimension of language use: Metaphor plays on the relationship between the propositional form of an utterance and the speaker’s thought; irony plays on the relationship between the speaker’s thought and a thought of someone other than the speaker. This suggests that the notion of a trope, which covers metaphor and irony and radically distinguishes them from ‘non-figurative’ utterances, should be abandoned altogether: it groups together phenomena which are not closely related and fails to group together phenomena which are. (Relevance, p. 243) Literary language is not uniquely tropological, as all forms of language use metaphor and irony. Consequently relevance theory is as applicable to literary language as to any other form. I will try to validate this claim in the following chapter, by considering relevance theory in relation to a number of traditional and contemporary hermeneutic theories.

3 ‘Positive Hermeneutics’: Relevance and Communication

Although they concentrate on spoken communication, Sperber and Wilson assert that relevance theory also applies to the writing and reading of literature. They write, ‘we assume, for instance, that the lengthy and highly self-conscious processes of textual interpretation that religious or literary scholars engage in are governed just as much by the principle of relevance as is spontaneous utterance comprehension’ (Relevance, p. 75). It is evident, however, from their rejection of the code model and their treatment of weak implicatures, that they do not (or rather, could not) hold a naive view of the perfect transmission of a univocal authorial message. 1 As they put it at the beginning of Relevance: What a better understanding of myth, literature, ritual, etc., has shown is that these cultural phenomena do not, in general, serve to convey precise and predictable messages. They focus the attention of the audience in certain directions; they help to impose some structure on experience. To that extent, some similarity of representations between the artists or performers and the audience, and hence some degree of communication, is achieved. However this is a long way from the identity of representations which coded communication is designed to guarantee. (Relevance, p. 8) There are, of course, significant differences among the triad myth, literature and ritual. Rituals are by definition participatory, and historically, myths have probably more often been passed down orally than read by ‘outsiders’ to a tradition.2 Thus the primary participants and audiences of rituals and myths more often share a mutually manifest context with 29

30

Paradigms of Reading

the performers than many or most literary readers (and literary-anthropological myth interpreters à la Lévi-Strauss) share a context with authors. The vast majority of readers of canonical literature are far removed in time and place and temperament from the author. Literature clearly lacks the pragmatic context of ordinary speech, as well as the gestures, facial expressions, tones of voice, and so on that help hearers of real-life utterances to draw pragmatic inferences. Yet literature seems to compensate for the absence of a pragmatic context by way of its own special context of traditions, conventions, genres, allusion, formal structures, and so forth. All of these are in many ways analogous to the shared domains of assumption that pragmaticians posit in ordinary discourse, and therefore enable quasi-pragmatic inference. 3 Furthermore, texts can be said to supply their own context to the extent that writer and reader have similar assumptions stored at the encyclopaedic entries attached to concepts. If the understanding of written texts is in any way similar to that of spoken communication (and if S&W are right), the reader will assume that the words on the page provide evidence about a writer’s informative intentions, and are there to lead him towards an inferential recognition of those intentions, which if successful will result in at least a partially mutual cognitive environment. The element of relevance theory that is crucial to the theory of literary interpretation is the thesis that we treat relevance as given and context as variable: the use of established genres, conventions and poetic forms will generally influence an experienced reader in his or her choice of context. Yet there also exist, of course, both texts that do not employ conventional techniques and readers who do not know them or who choose not to be constrained by inferred authorial intentions. At first sight, it might appear that S&W reject the interpretations generated by such readers out of hand. After all, they state that ‘we all know – semioticians included – that communication involves the publication and recognition of intentions’ (Relevance, p. 24), and that ‘the goal of the hearer, and, by extension, of the pragmatic processes, is to recover not just some arbitrary interpretation, but the interpretation intended by the speaker: this is the only interpretation it is worth the hearer’s effort to recover’ (Wilson & Sperber, 1991b, p. 583).4 They also state explicitly that where ‘the first optimally relevant interpretation’ is ‘relevant in a way the communicator could not have foreseen . . . it is not consistent with the principle of relevance’ (Relevance, p. 169). How does this square with the claim that the principle of relevance is an exceptionless generalisation about human communicative behaviour

‘Positive Hermeneutics’

31

that does not have to be known or learnt? The answer is that as the book title Relevance: Communication and Cognition suggests, relevance theory incorporates a theory of cognition as well as a theory of communication. Both are based on notions of relevance, defined in relation to cognitive effect and processing effort. The theory of communication, outlined in the previous chapter, involves the notion of optimal relevance, and is designed to explain how an audience infers a communicator’s intentions, or in other words how they recognize intended relevance. The principle of relevance was originally defined in terms of acts of ostensive communication communicating the presumption of optimal relevance. The theory of cognition, involving the notion of maximal relevance, which is designed to explain how people make sense of the phenomena around them, independently of anyone else’s intentions, was given rather less prominence. However, as mentioned in the previous chapters, in the Second Edition of Relevance (1995), S&W revise this, making the claim that ‘Human cognition tends to be geared to the maximisation of relevance’ – the First (or Cognitive) Principle of Relevance – and the claim that ostensive communication communicates a presumption of its own optimal relevance – the Second (or Communicative) Principle of Relevance. Achieving maximum relevance involves selecting the context that enables one to achieve the best possible balance of effort against effect. More specifically, Each thought process leaves the mind in a state characterised by an initially given context and possible extensions. If we are right in assuming that the train of human thoughts is steered by the search for maximal relevance, then the mind should try to pick out, from whatever sources it has available, including its own internal resources, the information that has the greatest relevance in the initial context: that is, which has the greatest contextual effects and requires the smallest processing effort. Such information is to be sought in accessible extensions of the context, whether they involve encyclopaedic memory, the short-term memory store, or the environment. (Relevance, p. 147) Hence readers whose encyclopaedic memories suggest that most works of literature are essentially symptoms of their authors’ unresolved Oedipus complexes, or merely illustrations of bourgeois writers’ habitual ideological erasure or sublimation of the social, economic and political realities of their time and place, or just more examples of the random,

32

Paradigms of Reading

tropological, mechanical and inhuman nature of language, will find such an interpretation the most accessible extension of the context, and the one that produces the greatest cognitive effects. Such cases of thematic interpretation clearly go beyond communication – the inferential recognition of consciously intended meanings – but S&W make it quite clear that languages, in a reasonably broad sense of the term, can and do exist without being used for communication. . . . The activities which necessarily involve the use of a language (i.e. a grammar-governed representational system) are not communicative but cognitive. Language is an essential tool for the processing and memorising of information. (Relevance, pp. 172–3) In a vivid analogy, they suggest that a grammar-governed representational system needn’t be used for communication any more than a prehensile organ need be an olfactory organ, even though an elephant’s is: the fact that humans use language for communicating is exceptional, like the way elephants use their noses for picking things up! Thus, in the case of literary interpretation, readers can treat the words on the page as information to be processed in any number of different ways, as long as the processing effort produces satisfactory cognitive effects. Some readers and critics, of course, choose to limit their interpretive efforts to the reconstruction of authorial meaning. E. D. Hirsch, for example, famously (or perhaps notoriously) insists that the ‘effort to find out what an author meant’ is ‘the only proper foundation of criticism’ (Hirsch, 1967, p. 57), as without the ‘stable determinacy of meaning . . . there can be no knowledge in interpretation, nor any knowledge in the many humanistic disciplines based upon textual interpretation’ (Hirsch, 1974, p. 1). On the contrary, theorists who see literary reading as going way beyond communication somewhat downplay the role of the author. Roland Barthes, for instance, wrote an essay with the unambiguous title ‘The Death of the Author’, in which he stated that where writing – or the ‘practice of the symbol’ – begins, ‘the author enters into his own death’, as ‘writing is the destruction of every voice, of every point of origin’, a ‘neutral composite oblique space where our subject skips away’ (Barthes, 1977, p. 142); while Michel Foucault, in ‘What is an Author?’, sought to replace the notion of the author by that of the ‘author-function’, arguing that ‘the author is the principle of thrift in

‘Positive Hermeneutics’ 33

the proliferation of meaning. . . . The author does not precede the works, he is a certain functional principle by which, in our culture, one limits, excludes and chooses; in short, by which one impedes the free circulation, the free manipulation, the free composition, decomposition, and recomposition of fiction’ (Foucault, 1979, pp. 157, 158). 5 Rather than attempt to infer manifestly communicated informative intentions, readers can let meaning proliferate, and manipulate, decompose and recompose fiction for as long as the effort brings results. However in this chapter I will briefly outline the history of hermeneutics from Schleiermacher to the present day, and then consider the Communicative Principle of Relevance, and Dan Sperber’s account in his anthropological work of the transmission of representations, in relation to literature. In the following chapter, I will consider the Cognitive Principle of Relevance in relation to various contemporary non-communicative theories of literature.

The hermeneutic tradition ‘Traditional’ or ‘romantic’ hermeneutics, beginning with Friedrich Schleiermacher early in the nineteenth century, suggests that since texts have meanings that are intended by their authors and easily recognized by their original audiences, responsible later readers wishing to overcome their temporal distance from the author need to complement ‘grammatical’ reading by a ‘technical’ or ‘psychological’ reading, a divinatory projection into the author’s mind (Schleiermacher, 1978). Wilhelm Dilthey later wrote about Nacherleben, a transposition into the author’s linguistic context, and an intuitive historical re-experiencing or duplication of his or her mental and emotional state (Dilthey, 1972). For Dilthey, the highest form of interpretation was a state of total empathy with the author, arrived at by an act of creative imagination, working in the reverse direction to that taken by the author. This argument soon became something of a critical orthodoxy. Milman Parry, for example, offered a succinct version in 1922: The literature of every country and of every time is understood as it ought to be only by the author and his contemporaries. Between him and them there exists a common stock of experience which enables the author to mention an object or to express an idea with the certainty that his audience will imagine the same object and will grasp the subtleties of his idea. One aspect of the author’s genius is his taking into account at every point the ideas and the information

34

Paradigms of Reading

of those to whom he is addressing his work. The task, therefore, of one who lives in another age and wants to appreciate that work correctly, consists precisely in rediscovering the varied information and the complexes of ideas which the author assumed to be the natural property of his audience. What I have just said is obviously no more than one of the countless ways of expressing a great truth of scholarship. But if the principle is only too evident, its application is rare in the extreme, is in fact so complex as to be impossible of realization in an entirely satisfactory manner: this goal of scholarship is nothing less than perfection itself. (Parry, 1971, p. 2) Perfectly reconstructing the author’s (or his or her original audience’s) complexes of ideas, or cognitive environment, in a reverse direction, would indeed be no small achievement – as someone once asked, why is Fred Astaire more famous than Ginger Rogers, when she had to do everything he did, only backwards and in high heels? – and Schleiermacher, Dilthey and Parry’s successors generally set interpreters rather more modest targets. Hirsch, for example, argues that the reader need only identify the author’s determining will. He rightly states that ‘meaning is an affair of consciousness not of words. Almost any word sequence can, under the conventions of language, legitimately represent more than one complex of meaning’ (Hirsch, 1967, p. 4), from which he concludes that the valid meaning is that willed or intended by the author. 6 To relate textual meaning to any context beyond itself is not to create meaning but only significance. To reject the author’s will is to reject ‘the only compelling normative principle that could lend validity to an interpretation’ (p. 5). For Hirsch, meaning, as such, cannot change. If a reader responds differently to a text at two different times, it is the response that changes, not the meaning. The same applies if the reader in question is the author him- or herself. Yet how is a reader to transpose himself into an author’s mind or reconstruct her willed or intended meanings? Nietzsche argued that being able completely to suspend one’s own horizon, and therefore any other system of belief, would destroy the ability to believe or create (see, e.g., Nietzsche, 1998, p. 118). He also criticized not only the founder of nineteenth-century romantic hermeneutics but all the German idealist philosophers as Schleiermacher: makers of veils, obfuscators who hide the truth (Nietzsche, 1992, p. 92). Husserl argued that ‘prejudgments’ distort our understanding and prevent objects from revealing their true

‘Positive Hermeneutics’ 35

nature, but Heidegger (1978, pp. 191–2) rejected Husserl’s attempt to build a presuppositionless phenomenology, and insisted that without our very being-in-the-world and our prejudices and presuppositions, we would be vacuous subjects to which objects could reveal nothing: ‘Whenever something is interpreted as something, the interpretation will be founded essentially upon fore-having, fore-sight and fore-conception. An interpretation is never a presuppositionless apprehending of something presented to us.’ Consequently, ‘any interpretation which is to contribute understanding must already have understood what is to be interpreted. This is a fact that has always been remarked’ (Heidegger, 1978, p. 194). This is a version of the ‘hermeneutic circle’. Another example, well-known to Christians, was given by Rudolf Bultmann: in order to understand the Christian kerygma – the message about the advent of God in Jesus Christ (or the bible) – you have to be a believer, but in order to be a believer it is necessary to understand this message. Heidegger argues that ‘if we see this circle as a vicious one and look out for ways of avoiding it . . . the act of understanding has been misunderstood from the ground up’ (Heidegger, 1978, p. 194). There is no question of getting out of the circle; what is important is getting into it in the right way. Thus Hans-Georg Gadamer, following Heidegger and breaking with the ‘romantic’ hermeneutic tradition, insists that it is naive to suppose than an interpreter can simply abandon his own ‘horizon’. On the contrary, our values and beliefs, or prejudgements, determined by history and tradition, are a necessary precondition for all understanding. Consequently, Gadamer describes the interpretation or understanding of a text from the past as always involving a fusion of two horizons or contexts, that of the interpreter and that of the producer. ‘The important thing’, however, ‘is to recognize the distance in time as a positive and productive possibility of understanding. It is not a yawning abyss, but is filled with the continuity of custom and tradition’ (Gadamer, 1982, p. 264).7 For Gadamer, ‘what is necessary is a fundamental rehabilitation of the concept of prejudice and a recognition of the fact that there are legitimate prejudices if we want to do justice to man’s finite, historical mode of being’ (p. 246). ‘The important thing is to be aware of one’s own biases’, so that a work of art ‘may present itself in all its newness, and thus be able to assert its own truth against one’s fore-meanings. . . . It is the tyranny of hidden prejudices that makes us deaf to the language that speaks to us in tradition’ (pp. 238, 239). In other words, we can become aware of our context-bound prejudgements in the attempt to understand works of art from the past. Attempting to

36

Paradigms of Reading

construct a coherent interpretation, the interpreter ‘is exposed to distraction from fore-meanings that are not borne out by the things themselves’, that ‘come to nothing in the working out’, and hence must be discarded (pp. 236, 237). The prejudgements that do not do justice to the true objects (die Sachen selbst) will need to be revised in interpretation until they furnish a unity of meaning. Thus we can learn from a work of art from the past, but what we learn is due to our appropriate prejudgements. 8 On the other hand, since suspension of disbelief is somewhat easier than the suspension of belief, we cannot approach a Medieval or Renaissance or Enlightenment or Romantic text in, say, a pre-Copernican, preDarwinian, pre-Marxian and pre-Freudian state of mind. We almost inevitably attribute anachronistic thematic meanings (or what Hirsch calls ‘significance’) to such texts. Although Hirsch allows that authors generally mean more than they are aware of meaning, he rejects a reading of Hamlet as a Freudian family romance because Shakespeare could not possibly have willed this. But those of us familiar with Harold Bloom’s often repeated claim that it is Shakespeare who invented Freudian psychology cannot possibly hope to transpose ourselves imaginatively from our context into that of Shakespeare and his original audience, thereby re-experiencing Hamlet’s original meaning. The impossibility of simply abandoning our own ‘horizon’ means that try as we might, we will not be able to wholly recreate the context that an author would have regarded as optimally relevant for the audience she had in mind, in which a text from another horizon will become optimally relevant.

Communicative intentions Not unlike the ‘romantic’ hermeneutics of Schleiermacher and Dilthey, relevance theory argues that for communication to be successful, a text or utterance must be inferentially combined with the correct contextual assumptions – which means those envisaged by the writer or speaker. Explicatures are defined as combinations of linguistically encoded and contextually inferred conceptual features; implicatures as contextual assumptions or implications which a writer manifestly intended to make manifest. Thus the fulfilment of a communicator’s informative intention requires that both her explicatures and implicatures are recovered in a particular context. It is wholly naive to assume that it is possible to communicate a particular ‘message’ or informative intention to any audience, no matter what their cognitive environment, or that

‘Positive Hermeneutics’ 37

a writer is ‘writing to be understood’ independently of differences in contextual assumptions. Some ideas, or mental representations, are successfully communicated, as public representations, within given cultures or communities. In Explaining Culture, Sperber describes culture as being made up, first and foremost, of contagious ideas, and develops an epidemiology of representations that explains this contagion. He explains the existence of relatively stable cultural representations (those which are repeatedly communicated and minimally transformed in the process) in terms of our shared interpretive mechanisms and common experience: Public representations are generally attributed similar meanings by their producers and by their users, or else they could never serve the purpose of communication. This similarity of attributed meanings is itself made possible by the fact that people have similar enough linguistic and encyclopaedic knowledge. Similarity across people makes it possible to abstract from individual differences and to describe ‘the language’ or ‘the culture’ of a community, ‘the meaning’ of a public representation, or to talk of, say, ‘the belief’ that witches ride on broomsticks as a single representation, independently of its public expressions or mental instantiations. (Sperber, 1996, p. 81) Sperber is confident that people in general usually interpret the same data in similar ways. As he points out, Anthropologists and psychologists alike tend to assume that humans are rational – not perfectly rational, not rational all the time, but rational enough. What is meant by rationality may vary, or be left vague, but it always implies at least the following idea: human beliefs are produced by cognitive processes which are on the whole epistemologically sound; that is, humans approximately perceive what there is for them to perceive and approximately infer what their perceptions warrant. (Sperber, 1996, p. 85) He explains this by way of an expansion of Chomsky and Fodor’s argument that the human mind or brain is not a general, all-purpose intelligence, but a combination of many differently specialized, pre-programmed devices or ‘modules’ that handle specific cognitive domains and process specific types of information. Fodor (1983) argues that perceptual

38

Paradigms of Reading

processes (and linguistic decoding) are carried out by specialized, domainspecific, genetically specified, ‘hardwired’, autonomous mechanisms, that are part of a fixed neural architecture. 9 Yet despite the posited similarity of all human minds, most public representations are only transmitted within specific cultures. Moreover, they are generally aimed at specific audiences, whom the speaker or writer has in mind during composition. This is why literary authors do not normally expect their words (or mental representations) to present their intended readers with the kind of difficulties faced by ethnologists studying radically different cultures. As Vološinov argued in the 1920s, The word is a two-sided act. It is determined equally by whose word it is and for whom it is meant. As a word, it is precisely the product of the reciprocal relationship between speaker and listener, addresser and addressee. . . . I give myself verbal shape from another’s point of view, ultimately from the point of view of the community to which I belong. A word is a bridge thrown between myself and another. If one end of the bridge depends on me, then the other depends on my addressee. A word is territory shared. 10 (Vološinov, 1973, p. 86) Bakhtin, similarly, explained that When speaking I always take into account the apperceptive background of the addressee’s perception of my speech: the extent to which he is familiar with the situation, whether he has special knowledge of the given cultural area of communication, his views and convictions, his prejudices (from my viewpoint), his sympathies and antipathies – because all this will determine his active responsive understanding of my utterance. These considerations also determine my choice of a genre for my utterance, my choice of compositional devices, and, finally, my choice of language vehicles, that is, the style of my utterance.11 (Bakhtin, 1987, pp. 5–6) More recently, a number of reader-response theorists have described the (hypothesised) reader that authors imagine themselves to be writing for – the literary equivalent of the immediate addressee of a spoken utterance. For example Wolfgang Iser writes of the implied reader, Steven Mailloux of the intended reader, and Umberto Eco of the model reader, foreseen by the author (and in fact created by the text) who

‘Positive Hermeneutics’ 39

shares all the requisite codes. Peter Rabinowitz has argued that where poetry and fiction readily give pleasure, it is often because the actual reader coincides with the hypothetical audience or what he calls the ‘authorial audience’ for whom the author imagines him- or herself to be writing (see Iser, 1978; Mailloux, 1982; Eco, 1977, 1979; Rabinowitz, 1977, 1987; MacKenzie, 1990). While readers are not obliged to follow the roles marked out for them in texts, many readers clearly do attempt to become part of the authorial audience whose hypothetical knowledge and beliefs determined the author’s use of explicatures and implicatures. This account of communication clearly has implications for translators as well as for literary interpreters. As Ernst-August Gutt points out in Translation and Relevance: Cognition and Context, translators have to communicate the set of assumptions (both explicatures and implicatures) the original communicator intended to communicate, in a particular context, to her original audience (although obviously a translator’s beliefs about the author’s assumptions can be mistaken). Gutt gives examples of how today’s reader is almost certain to lack many of the contextual assumptions necessary to understand parts of the Gospels. For example, most present-day readers, unaware of the necessity in the first century to demonstrate that the Messiah was indeed a descendant of King David, are likely to find Matthew I:1–16, listing the genealogy of Jesus (begat . . . begat . . . begat . . . , 39 times), somewhat bizarre, and quite simply inconsistent with the principle of relevance. They will probably be unable to reconstruct an appropriate context without additional help, but Gutt insists that cultural implications are not a matter of linguistics, so that spelling them out is a task of a ‘cultural translation’ involving footnotes and commentary, and not part of the job of the mere translator, for whom the explication of implicatures is both unnecessary and undesirable.12 Gutt (1991, p. 73) labels instances where the contextual assumptions intended by the communicator are employed primary communication situations. Where this condition is not fulfilled he refers to secondary communication situations, and insists that ‘in most cases they will lead to misinterpretations’, because where wrong contextual assumptions are used, ambiguities are likely to be resolved the wrong way, wrong semantic representations chosen, wrong propositional forms derived, metaphorical expressions taken literally, implicatures and explicatures confused, intended implicatures missed, unintended implicatures derived, and so on. In other words, Gutt is implicitly arguing, like Schleiermacher and Dilthey before him, that in order to avoid misinterpretations,

40

Paradigms of Reading

a reader separated by time and space from an author should attempt to recreate the context shared by the author and her intended audience, and thereby establish what the author would have regarded as optimally relevant. Yet unlike nineteenth-century hermeneutics, with its call for the re-experiencing or duplication of an author’s thoughts, relevance theory explains successful communication in terms of enlarging mutual cognitive environments. It thus seems more psychologically plausible than critics who still call for a state of empathy with the author – or, like Georges Poulet, claim to be able to achieve this. Poulet, the ‘Geneva School’ phenomenologist, described reading as an act of self-immolation in which the reader simply gives way to the author’s consciousness. When he reads, Poulet says, I am aware of a rational being, of . . . the consciousness of another [which] is open to me, welcomes me, lets me look deep inside itself, and even allows me, with unheard-of license, to think what it thinks and feel what it feels . . . while reading, I perceive in my mind a number of significations which have made themselves at home there: . . . They are the thoughts of another and this other thinks, feels, acts and suffers within me. . . . When I am absorbed in reading, a second self takes over, a self which thinks and feels for me. . . . The consciousness inherent in the work is active and potent; it occupies the foreground. . . . In opposition, I myself, although conscious of whatever it may be conscious of, play a much more humble role, content to record passively all that is going on in me. (Poulet, 1980, pp. 42–7) However, given the highly active, inferential nature of linguistic comprehension described by relevance theory (in terms of selecting semantic representations, completing them to generate propositional forms, identifying referents and propositional attitude, and, in response to tropes and poetic devices, producing one’s own weak implicatures), and the inevitable gap between the author’s and reader’s horizons, Poulet’s description of reading as the passive perception of ready-made significations – while perhaps revealing an admirable humility – seems wholly naïve and improbable. Furthermore, quite apart from the impossibility of emptying our minds of our own prejudices and prejudgements in order to duplicate an author’s historical and linguistic context, there is no guarantee that our understanding of an author’s words will allow us to re-experience his

‘Positive Hermeneutics’ 41

or her mental and emotional state. I. A. Richards suggested in Principles of Literary Criticism (1926) that authors construct linguistic structures that correspond as closely as possible to their initial experience, and that readers, by a careful and precise study of signifying forms, can work backwards toward the experiences that produced these forms. Paul de Man contested this notion of the continuity between form and experience by way of some examples from William Empson’s Seven Types of Ambiguity (1972). The first is Shakespeare’s famous line from the 73rd sonnet (already discussed in the previous chapter) comparing old age to a forest in winter, or to be precise, to ‘Bare ruined choirs, where late the sweet birds sang’. De Man points out that this ‘metaphor whose perfection is immediately felt’ awakens an indefinite number of experiences: Empson lists a dozen of them and there are many others; it would be impossible to tell which was dominant in the poet’s mind or at which we should stop. . . . instead of setting up an adequation between two experiences . . . it deploys the initial experience into an infinity of associated experiences that spring from it. . . . Far from referring back to an object that would be its cause, the poetic sign sets in motion an imagining activity that refers to no object in particular. The ‘meaning’ of the metaphor does not ‘mean’ in any definite manner. (Blindness, p. 235) This argument is wholly compatible with S&W’s account of poetic effects in terms of the generation of weak implicatures, and communication giving way to cognition. De Man points out that ‘if a simple metaphor suffices to suggest an infinity of initial experiences and, therefore, an infinity of valid readings’ we cannot ‘live up to Richards’ injunction to bring the reader’s experience in line with the typical experience of the author (Blindness, p. 236). On the contrary, ‘The correspondence between the initial experience and the reader’s own remains forever problematic’ (p. 236). Although an author and reader can achieve some similarity of representations, the active reconstruction of an author’s originary experiences is, as de Man insists, just as improbable as the passive absorption of an author’s significations.13

The epidemiology of representations Dan Sperber implicitly offers further objections to Poulet’s account of ‘self-immolation’ in the act of reading, and Richards’ call for the

42

Paradigms of Reading

reconstruction of an author’s experience by a careful and precise study of his or her signifying forms, in his theory of the epidemiology of representations in Explaining Culture. Sperber’s model is specifically opposed to Richard Dawkins’ neoDarwinian model which sees culture as made up of units he calls ‘memes’, which are selected and replicated, like genes, without human psychology playing much of a role (Dawkins, 1976, 1982). Sperber argues, on the contrary, that unlike infectious diseases, which are transmitted by replication of viruses or bacteria, with only rare, accidental mutations, representations tend to be transformed each time they are transmitted. Understanding, as earlier argued in Relevance, is not a reproduction of thoughts, but the production of some degree of resemblance between a communicator’s and an audience’s thoughts. Thus the replication or reproduction of a representation, if it ever occurs, is an exception. Once again, only the erroneous code model would lead one to expect replication and consider alterations in transmission as accidents. In fact, representations are transformed as a result of a constructive cognitive process. There seems to be a certain mutational element in every copying event, because human brains do not copy or synthesize all the information they are presented with, but rather use it as more or less relevant evidence with which to construct representations of their own. Sperber says: Let us generalize: in order to represent the content of a representation, we use another representation with a similar content. We don’t describe the content of a representation; we paraphrase it, translate it, summarize it, expand on it – in a nutshell, we interpret it. An interpretation is a representation of a representation by virtue of a similarity of content. In this sense, a public representation, the content of which resembles that of the mental representation it serves to communicate, is an interpretation of that mental representation. Conversely, the mental representation resulting from the comprehension of a public representation is an interpretation of it. The process of communication can be factored into two processes of interpretation: one from the mental to the public, the other from the public to the mental. (Sperber, 1996, p. 34) Specifically, according to the principle of relevance, representations are transformed in the process of transmission, not in a random fashion, but in the direction of contents that require lesser mental effort and provide greater cognitive effects. This

‘Positive Hermeneutics’ 43

tendency to optimize the effect–effort ratio – and therefore the relevance of the representations transmitted . . . – drives the progressive transformation of representations within a given society towards contents that are relevant in the context of one another. (Sperber, 1996, p. 53) Yet quite apart from representations that can be readily interpreted, and poetic representations that are intended to achieve most of their relevance through a wide array of weak implicatures, there also exist pieces of (poetic) language that are not well-formed enough to allow the reader to select a propositional form. S&W assert that in real-time spoken communication, a hearer’s first inferential sub-task is to assign a unique propositional form to an utterance, by selecting a semantic representation, and assigning a referent to each referring expression. Hence, for example, the German audience in 1961 hearing Kennedy proclaim ‘Ich bin ein Berliner’ had to decide quickly whether the American President was in fact claiming to be a local sort of doughnut. Yet Sperber, in ‘Apparently Irrational Beliefs’ (in Sperber, 1985), concedes that just as many of our utterances are in fact semi-grammatical strings rather than proper sentences, many of our thoughts approximate but do not achieve propositionality, and are thus only semi-propositional. Fortunately, we possess meta-representational abilities that allow us to entertain or believe ideas without fully understanding them, and without giving them a well-formed representation. It follows that we can also utter (or write) incompletely understood ideas, of which we have only a semipropositional representation, thereby only intending to convey a semipropositional representation, the understanding of which will be more a matter of cognition than communication. Thus although logical propositions, as opposed to sentences or utterances, cannot be ambiguous (i.e. true in some interpretations and false in others), but are by definition either true or false, some of our reflective beliefs have several possible interpretations and we can hold them without committing ourselves to any of their interpretations. This ability to hold beliefs of a semi-propositional form gives us, in both childhood and adulthood, a means to process information which exceeds our conceptual capacities, and to deal with situations in which we hold two mutually inconsistent ideas and are reluctant to give up either one of them: we give one of them a semi-propositional form. Similarly, a scientist faced with counter-evidence needn’t reject a theory, but can look for a new interpretation by making some of its terms open to redefinition, during which time the theory is in a semi-propositional

44

Paradigms of Reading

state. An analogy might be made with the evolutionary account of the development of the eye as a very slow process of adaptive change taking over half a million generations. Darwin’s critics asked ‘What use is half an eye?’ Richard Dawkins (1995, p. 90) answers, ‘One per cent better than 49 per cent of an eye, and the difference is significant.’ Representations in semi-propositional form are better than no representations at all. Yet contrary to the Darwinian story, in many cases, the journey might be more important than arriving at the end. As Sperber argues: Semi-propositional representations do not only serve as temporary steps towards or away from full propositional understanding. The range of interpretations and the search through that range, as determined by a semi-propositional representation, may be of greater value than any one of these interpretations in particular. The relativist slogan, the teaching of a Zen master, the philosophy of Kierkegaard, and, generally, poetic texts are cases to the point. Their content is semi-propositional from the start. The speaker’s or author’s intention is not to convey a specific proposition. It is to provide a range of possible interpretations, and to incite hearers or readers to search that range for the interpretations most relevant to them. The ideas that come as by-products of this search may suffice to make it worthwhile, particularly when no final interpretation is ever arrived at.14 (Sperber, 1985, p. 53) Sperber describes religious practices as being typical of cultural representations that are repeatedly communicated and regularly change, as they allow the possibility of endless reinterpretation: ‘A repetitive practice may remain relevant because, without perceptibly modifying its public form, it lends itself to different interpretations according to the agent, the circumstances, and the stage of the life cycle’ (Sperber, 1985, p. 116). The same can be said of many works of literature. Yet this does not preclude readers, at any time, from attempting inferentially to combine a text with the contextual assumptions envisaged by the writer.

The reader as supplement Leaving aside the matter of semi-propositional texts whose authors intend to provide a range of possible interpretations, there remains an objection to the argument that readers interested in communication should attempt to reconstruct the context shared by an author and her

‘Positive Hermeneutics’ 45

original readers. Even if we accept Gadamer’s account of the fusion of horizons, isn’t attempting to limit interpretation to retrieving the author’s informative intention, recognising the continuity of custom and tradition, and becoming the author’s addressee, overly conservative?15 Indeed Robert Scholes has labelled theories that attempt to recapture a so-called original truth predicated on an unchanging authorial intent as ‘reactionary’, although he quickly stresses the necessity of this ‘centripetal’ impulse in combination with the ‘centrifugal’ or ‘exorbitant’ impulse that finds new meanings on the text’s always expanding circumference: ‘I must invent the author, invent his or her intentions, using the evidence I can find to stimulate my creative process’; ‘We must resist the impulse to think of the centripetal as the uncreative, the unimaginative, the oppressive restriction of meaning’ (Scholes, 1989, p. 9; see also Scholes 1985). Without this first step – the postulation of an intentional subjectivity behind a verbal text – we can only ever project our own subjective modes of thought and desire upon a text, and will never find anything to interpret or criticise. But Scholes insists that these backward movements must be followed dialectically, by forward movements in which the reader incorporates the text into his or her own context, whereupon the reader’s self becomes a new centripetal centre.16 However, Tillotama Rajan has pointed out that both these possibilities – postulating the intentional subjectivity behind the text, and incorporating the text into one’s own context – have been implicit since the very beginning of the hermeneutic tradition. As already mentioned, Schleiermacher argued for the necessity of two complementary levels of reading: grammatical reading, understanding a text by analysing its grammar and the words on the page (including rhetorical figures), and psychological reading, a projection into the author’s inner creative process and a reconstruction of his or her intentions. But as Rajan argues in The Supplement of Reading (1990), it is only a short step from Schleiermacher’s psychological reading to Shelley’s account of the necessary gulf between inspiration and composition. In his ‘Defence of Poetry’, Shelley stated that no poet ever entirely captures his vision within his poem because ‘the mind in creation is as a fading coal’, and ‘when composition begins, inspiration is already on the decline, and the most glorious poetry that has ever been communicated to the world is probably a feeble shadow of the original conception of the Poet’ (Shelley, 1965, vol. VII, p. 135). This transforms the poem from an autonomous formal unit to a trace or a re-presentation of an informative intention that precedes it in the creative process, and shifts the locus of meaning away from the written text altogether. The text becomes a pre-text from which the reader must generate meaning by

46

Paradigms of Reading

a process of divination, seeking a fusion with the author on a subliminal or transverbal level. But in displacing meaning from the author’s words to the creative consciousness behind the text, Schleiermacher also opened the door to applicative reading: interpreting the text according to our own beliefs or desires or will-to-power. This transforms the literary work from a product to a site of productivity, a catalyst for a new meaning that is created in the reading process. In other words, Schleiermacher began the erosion of belief in the selfsufficiency or referential stability of the text that culminates in today’s poststructuralist fracturing of the sign in which all the links between signifier, signified, signifying subject and reader are disrupted, and the psychological and interpersonal dimensions of reading disregarded. Once we stop thinking of the reader as recipient, and make him or her a supplement who needs to complete the text, to ‘bridge the gap between conception and execution, and to supply a unity not present in the text’ (Rajan, 1990, p. 2), we open two possible roads. There is, to be sure, the ‘positive hermeneutic’ that either seeks to reproduce what was in the author’s mind by expanding elements present in the feeble shadow of the text, or seeks to make manifest or restore a meaning (a message or kerygma) that was once understood but which, with the passage of time, has become obscure. But we also spawn the ‘negative hermeneutic’ that sees the text as a resource rather than a source: a heuristic stimulus for the production of new meanings involving elements absent from or even in contradiction with the textual surface. Recognising the reader as a creative supplement rather than a recipient necessarily involves renouncing the author’s authority, and accepting that no reader is obliged to follow the roles prescribed within texts or by limited, local, culturally determined critical conventions. Although relevance theory explains communication in terms of an original context, in which particular explicatures and implicatures should ideally be recovered, literary reading need not limit itself to attempted communication. In the following chapter I consider the ‘negative hermeneutic’ in terms of relevance theory.

4 ‘Negative Hermeneutics’: Themes, Figures, Codes and Cognition

The previous two chapters have shown how relevance theory, which describes the inferential processes of recovering propositional attitudes and explicatures, deducing implicatures, trying out figurative interpretations, and creating a context in the attempt to make a text optimally relevant, gives a precise account of how readers can understand an author’s informative intentions, and how successful communication can take place. It also explains our awareness of failures of communication, and how we can be conscious of not sharing what a speaker or writer believed to be mutually manifest assumptions, and thus uncertain about the intended meaning of a text. Untrained readers, for example, can place texts in genres different to those intended by their authors, or fail to recognise literary conventions, references and allusions. Furthermore, skilled readers, in different contexts and cognitive environments, frequently find that the interpretive solution requiring the least processing effort and producing the greatest cognitive effect gives a reading that differs radically from other people’s. Indeed, most interesting interpretive disagreements arise where it can be argued that a wide spectrum of possible readings were intended or weakly implicated by the author. Milton either was or was not ‘of the Devil’s party’ when he wrote Paradise Lost, but neither interpretation can be definitively shown to flout the principle of relevance. There are also several varieties of thematic reading, which might collectively be labelled ‘negative hermeneutics’, that require the postulation of the author’s conscious intention only in order to strip away the surface meaning and reveal unconscious motivations or denounce false consciousness or ‘phallogocentricity’ or whatever. Still other critics are simply uninterested in authors’ communicative intentions, regarding them as only one possible focus of interest in a text. As Richard Rorty 47

48

Paradigms of Reading

puts it, the only genuinely constraining force involved in reading is the physical process that gives us little replicas of the text on our retinas, and ‘from there on it is up to us to make something out of these replicas by telling a story about their relation to other texts, or the intentions of its author, or what makes life worth living, or the events of the century in which the poem was written, or the events of our own century, or the incidents of our own lives, or whatever else seems appropriate in a given situation’ (Rorty, 1991, p. 82). Relevance theory, however, being a dual account of both communication and cognition, equally explains how readers who do not wish to be constrained by considerations of authorial intention use their cognitive abilities to make sense of texts in ways that are manifestly relevant to their own beliefs or interests, either by making thematic connections, or by ignoring the economic principles underlying the theory of optimal relevance and analysing all the possibilities of the words of the text. Interpreters who disregard the economic protocols of reading described by the communicative theory of relevance can often tease intriguing and diverting interpretations out of the potential of codes, without any reference to authors or their intentions. Indeed a critic basing an interpretation on codes can argue that any code ever posited anywhere is operative in any work of art, effectively subverting or adding to its manifest surface content. This is only objectionable if the critic claims that the code-based interpretation should be normative, as, unfortunately, deconstructionist critics have a habit of doing. In this chapter, I will consider these varieties of reading in relation to relevance theory.

Unconscious identities Relevance theory is concerned with conscious intentions, and does not consider the possibility that we are definitively cut off from authors’ communicative intentions because we are bound by interpretive strategies emanating from our own unconscious. Yet the ‘self psychologist’ Norman Holland proposes that each of us has an unconscious ‘primary identity’, a persistent sameness ‘imprinted’ on us by our mothers in our earliest relationship, which determines all our behaviour, including literary interpretation. Formed before speech, this identity can never be put into words. Thus we are all, inexorably and permanently, prisoners of our unconscious, to such an extent that we can never learn anything from reading. Holland (1975, p. 12) argues that there is nothing ‘in’ a literary text, which ‘in an objective sense consists only of a certain

‘Negative Hermeneutics’ 49

configuration of specks of carbon black on dried wood pulp’. Poems and narratives are merely marks on a page, and the reader ‘gives them life out of his own desires’ (1975, p. 12). ‘ “Meaning”, that is, the act of making sense of a text’, which for Holland is a matter of finding a thematic unity, ‘works as a defence against some sort of anxiety. Each reader, therefore, will search out a unifying idea that matches his particular needs for sense and logic’ (p. 14), and use the literary work ‘to create in himself a dynamic psychological process that transforms raw fantasy materials to conscious significance’ (p. 16). During this process, the ‘core’ of the ego regresses to primary process thinking, ‘based on the fusions of infant feeding’ (p. 19) – which is why we ‘devour’ novels, ‘have a taste’ for them, are ‘insatiable’ or ‘voracious’ readers, and so on. Thus Holland believes that readers group details of a work together so as to constitute a central theme without any reference to authorial intention: only a reader whose identity customarily causes him or her to deal with authority by total compliance will respond to the work ‘by seeking out and accepting, totally, uncritically, with a gee-gosh, the authority of its author’ (1980a, p. 125). Yet if ‘all of us, as we read, use the literary work to symbolize and finally to replicate ourselves’ (p. 123), the question arises why we bother to read at all, rather than engage in some other form of behaviour. Holland’s reader can learn nothing by reading, and cannot change or become a better person as a result of transactions with literature, but can only continue to replicate his or her self. Except that, in what seems to be a rather inconsistent argument, Holland also suggests that literary texts can and do resist the wildest identities, or at least that we allow them to resist us, in order to have the pleasure of sharing our interpretations with others. He suggests (1973, p. 117) that ‘different readers can construct different experiences of a poem, with more or less contortion of the words and the plain sense’, and that ‘the writer creates opportunities for projection but he also sets constraints on what the reader can or cannot project into the words-on-the page and how he can or cannot combine them. The reader can violate these stringencies, of course, but if he does so, he loses the possibility of sharing his reading with others’ (Holland, 1975, pp. 219–20). Which is to say that authors, or at least their words on the page, creep back into Holland’s account of readers’ defence mechanisms and the transformation of wish-fulfilling fantasies into themes, and that our primary identity is thus not so grimly deterministic at all. Relevance theory, of course, deals with communicators’ manifest conscious intentions. Its starting point is the recognition of the systematic gap between sentence meaning and speaker’s meaning, but it assumes

50

Paradigms of Reading

that utterances are generally good pieces of evidence of these meanings. There is normally no need to stop to consider that on top of this linguistic underdetermination there is a dynamic unconscious at work interfering with the construction of speakers’ informative intentions, ensuring that whatever they say it means something else. Relevance theory supposes that addressees generally ignore what they perceive to be ‘slips of the tongue’ (at least momentarily, even if they analyse them subsequently). Yet it can still account for unconsciously determined interpretations according to the cognitive theory of manifest relevance. Furthermore, it is far more specific than Norman Holland about the linguistic processes involved in giving meaning to the specks of carbon black on the page, and communicating a ‘plain sense’, and constraining readers’ projections. Holland’s arguments about the active nature of linguistic understanding, which I will outline in Chapter 9, are wholly compatible with relevance theory, but his actual account of interpretation as the replication of the self is ultimately contradictory and unconvincing.

Reading and time Apart from the physical absence of the ‘addresser’, the most obvious difference between hearing an utterance and reading a text is temporal: readers can stop and go back any time they want to. In spoken communication, addressees generally have to keep up with speakers, although of course, as S&W argue, speakers ordinarily try to make things easy for hearers by placing the effort-saving background implications of an utterance among its initial constituents, and putting its effect-carrying foreground implications in its final constituents. This means that with optimally relevant utterances, a hearer will be well on the way to constructing a context, or already there, by the time the last word is . . . 1 Hearers usually construct anticipatory hypotheses, especially if – as is frequently the case – large chunks of an utterance are made up of institutionalized expressions (see MacKenzie, 2000), which is why conversation generally proceeds with only fractional pauses between turns. Unusual word order, by poetic language, on the other hand, frequently is employed. Stanley Fish once based an entire theory of reading literature that he named ‘affective stylistics’ on the anticipatory hypotheses or ‘developing responses of the reader in relation to the words as they succeed one another in time’, because ‘in an utterance of any length, there is a point at which the reader has taken in only the first word, and

‘Negative Hermeneutics’ 51

then the second, and then the third, and so on’ (Fish, 1981, p. 27). Thus word-order and stylistic choices clearly matter, and Fish suggests replacing the question ‘what does this sentence mean?’ with ‘what does this sentence do?’ as the sentence is not ‘an object, a thing-in-itself, but an event, something that happens to, and with the participation of, the reader’ (p. 25). Consequently, we should ‘think of language as an experience rather than as a repository of extractable meanings’ (p. 67). The reader’s projections and assumptions as he reads, his hazarding of interpretations and acts of perceptual closure, even if these turn out to be false starts that are almost instantaneously modified or rejected, are the meaning of the text: ‘everything the reader does, even if he later undoes it, is a part of the “meaning experience” and should not be discarded’ (pp. 3–4). Yet Fish posits a reader who continuously makes premature hypotheses, and interprets the end of a line of verse as completing a thought, before finding that the next line alters the meaning, never seeming to learn from all these reversals and anagnorises. Norman Holland suggests that ‘Fish’s stop-motion method of reading’ could only come from ‘watching too much [American] football on television’ (Holland, 1980b, p. 366); in reality, though readers may occasionally be tricked into making premature hypotheses by deliberately contorted poetic syntax, it seems that they more generally avoid constructing an entire context on the basis of only the initial clause of a sentence. On the other hand, once they have come to the end of a sentence, competent readers of literature can, if they feel so inclined, take the time to explore all possible parsings, disambiguations, potential word meanings, reference assignments, enrichments, resolutions of vagueness, restorations of ellipsed material, allusions, thematic possibilities and illocutionary forces, and take each resulting explicature and extend the immediate context by adding all the encyclopaedic entries of all the concepts appearing in the explicature, and then judge the cognitive effects produced, which may be fascinating. This is clearly not a luxury one can afford in a real-time conversation, although S&W do not rule out such interpretive strategies for utterances: the interpretation of an utterance can proceed, in ever expanding contexts, just as long as the hearer thinks that the rewards are likely to outweigh the processing costs. With certain types of utterance, for example a sacred text or a fortune teller’s prophecy, a hearer might be willing to devote a lifetime’s energy to the interpretation process. In practice, however, expectations of relevance are generally much

52

Paradigms of Reading

lower; there are other demands on the hearer’s processing time; and he generally satisfies himself with establishing relevance in the most immediately accessible context, and leaves it at that. (Wilson and Sperber, 1991a, p. 391). Yet if a reader does take the time to delve into potential meanings in a sentence, the final interpretation will never be the same as the initial one, even if the sentence defeats all subsequent hypotheses or constructions. We do not necessarily make fruitlessly premature hypotheses like Fish’s imagined reader in ‘Affective Stylistics’, but we can and do make interpretive hypotheses after reading a passage of text, which means that we can never read the ‘same’ sentence or poem or novel twice. As Henri Bergson was the first to argue, time is dynamic and therefore irreversible. Because the past survives as memory, and leads to anticipations of the future, consciousness can never go through the same state twice. Every text we read, and every minute deliberately spent in interpretation, changes our stock of knowledge, and causes us to read new texts slightly differently, and to reconsider old ones, in a myriad of unpredictable and heterogeneous ways. The ‘words on the page’ are not actually constant over time, as they will act on a changed person. A static, mechanical, Newtonian model that neglects memory and anticipation is inadequate to explain the act of reading or interpretation. Clearly we do forget (or repress) a great deal over time, but in general, we cannot ignore what we already know, or the hypotheses we have tried out, while reading or rereading a text. S&W obviously accept that extended interpretation of literature is possible, but they consider it to be what might be called overinterpretation, carried out after the reader has come up with an interpretation consistent with the principle of relevance. They write: It is tempting, in interpreting a literary text from an author one respects, to look further and further for hidden implications. Having found an interpretation consistent with the principle of relevance – an interpretation (which may itself be very rich and very vague) which the writer might have thought of as adequate repayment for the reader’s effort – why not go on and look for ever richer implications and reverberations? If we are right, and considerations of relevance lie at the heart of verbal communication, such searches go beyond the domain of communication proper. (Wilson and Sperber, 1992, p. 76)

‘Negative Hermeneutics’ 53

However – although S&W fail to spell this out consistently, especially in the first edition of Relevance – this sort of reading or interpretation is covered by the notion of maximal rather than optimal relevance, which explains how people make sense of phenomena without reference to anyone’s intentions. Readers can expend any amount of interpretive effort, as long as they expect (and find) satisfactory cognitive effects. After all, canonical literature in general and wilfully difficult poetry in particular, seem to require a different type of reading, or what Frank Kermode calls a different ‘form of attention’, from today’s local newspaper.

Canonicity Much literary writing differs from other uses of language in that it is intended to require a large amount of processing effort to come up with an adequate interpretation, or is intended to repay processing effort by allowing a number of conflicting senses. According to relevance theory, writers aiming at optimal relevance do not have to make their writing ‘simple’, but merely to avoid gratuitous or unproductive complexity. Indeed, ‘difficulty’ that requires additional processing or interpretive efforts and yields additional effects is often thought to be a defining characteristic of literature – particularly among recalcitrant secondary school pupils. Literary authors self-consciously deploy the resources of language, both exploiting and resisting antecedent and contemporary uses of words. They might also discourage the fixity of meaning, which is one reason why canonical works are able to generate alternative interpretations over time. Experienced readers of literature, without necessarily going as far as deconstructionists in deliberately bracketing intentionality, meaningfulness and representativity, and doggedly seeking out contradictions, obscurities, ambiguities, incoherences, discontinuities, ellipses and interruptions, do generally approach literary texts with greater expectations than they do when interpreting normal discourse or struggling to understand unlearnt foreign languages. And as S&W suggest in their account of metaphor and poetic effects, after a certain point weak implicatures can no longer be considered as intended, so the reader takes responsibility for constructing them. In other words, communication shades off into cognition. Hence relevance theory is equally able to account for thematic, psychoanalytic, deconstructionist and political interpretations which go way beyond postulated intended meanings. It is a critical commonplace, for example, that most canonical poetry is expressly written in vague or richly figurative language, and is

54

Paradigms of Reading

consequently endlessly reinterpreted: readers approach poetry in a particular way because it seems to be endlessly reinterpretable. As Shelley put it in his ‘Defence of Poetry’, ‘All high poetry is infinite; it is as the first acorn which contained all oaks potentially’ (Shelley, 1965, vol. VII, p. 131). 2 A great poem will never be exhausted, and new generations will uncover new meanings. Bakhtin explained this by reference to his concept of ‘great time’: great literary works live on in the future because their roots extend into the distant past. They ‘are prepared for by centuries, and in the epoch of their creation it is merely a matter of picking the fruit that is ripe after a lengthy and complex process of maturation’, but even then, the author’s original audience will not be able to recognise or reveal the richness of the fruit (Bakhtin, 1987, p. 4). Bakhtin argues that: Semantic phenomena can exist in concealed form, potentially, and be revealed only in semantic cultural contexts of subsequent epochs that are favorable for such disclosure. The semantic treasures Shakespeare embedded in his works were created and collected through the centuries and even millennia: they lay hidden in the language, and not only in the literary language, but also in those strata of the popular language that before Shakespeare’s time had not entered literature, in the diverse genres and forms of speech communication, in the forms of a mighty national culture (primarily carnival forms) that were shaped through millennia, in theater-spectacle genres (mystery plays, farces, and so forth), in plots whose roots go back to prehistoric antiquity, and, finally, in forms of thinking. . . . Shakespeare took advantage of and included in his works immense treasures of potential meaning that could not be fully recognized or revealed in his epoch. The author himself and his contemporaries see, recognize, and evaluate primarily that which is close to their own day. The author is a captive of his epoch, of his own present. Subsequent times liberate him from this captivity, and literary scholarship is called upon to assist in this liberation.3 (Bakhtin, 1987, p. 5) Similarly, Frank Kermode has argued that with canonical literary texts, however a particular epoch or a particular community may define a proper mode of attention or a licit area of interest, there will always be something else and something different to say . . . all we are sure

‘Negative Hermeneutics’ 55

about is that the inadequacies of earlier exposition become astonishingly obvious to later expositors . . . this is what it means to call a book canonical . . . all grant to the text something like omnisignificance, all have canons of interpretation that are permissive rather than restrictive . . . the canonical work, so endlessly discussed, must be assumed to have permanent value and, which is really the same thing, perpetual modernity. 4 (Kermode, 1985, p. 62) Once works become canonical, they will normally be provided with the kinds of reading they require if they are to keep their immediacy to any moment; that is, to maintain their modernity . . . since the canonical work is fixed in time but applicable to all time, it has figural qualities not to be detected, save at an appropriate moment in the future. Interpretations may be regarded not as modern increments but rather as discoveries of original meanings hitherto hidden.5 (Kermode, 1985, p. 75)

Codes and inference According to this account of canonical literature, its infinite richness is due to both authorial and readerly skill with figurative language. The most important aspect of relevance theory in this context, however, is not its recognition of the importance of figurative language, but its demonstration that hearers and readers establish meanings inferentially, and not by decoding. If poetry can generate weak implicatures for hundreds of years, reading cannot be a matter of decoding or decrypting or recognising specific meanings generated by semantic or grammatical codes. As Alasdair Fowler (1989, p. 17) put it in his review of Relevance, ‘once it is conceded that all communication involves inferential interpretation, the nature of literary criticism will be seen in a rather different light. If what constitutes a meaning is optimally relevant context, meaning can hardly be arrived at by freeplay alone (although this may be a useful tactic for generating options).’ Relevance theorists, as much as deconstructionists, take it for granted that language (all language, not only figurative language) usually permits a wealth of interpretations; that canonical texts approach ‘omnisignificance’; and that, as Barbara Johnson (1990b, p. 46) puts it,

56

Paradigms of Reading

‘when one writes, one writes more than (or less than, or other than) one thinks’. But unlike Johnson, they do not take it as axiomatic that ‘the reader’s task is to read what is written rather than simply attempt to intuit what might have been meant’ (p. 46), or that most literary texts ‘require a reading strategy that goes beyond apparent intentions or surface meanings’ (p. 47). A reader interested in communication, in meeting an other (an author), may very well limit his or her interpretive activity to intuiting what might have been meant. Of course readers can always decide that they are uninterested in a posited optimally relevant interpretation, and seek to maximise relevance in any way they see fit, either by relating the text to a larger context according to their own thematic concerns, or by mining the words on the page and trying out in turn all the grammatical, semantic and propositional possibilities of the text, without choosing among them or worrying about thematic coherence. What relevance theory and its account of the inferential nature of interpretation shows, however, is that potential meanings that the ‘code’ allows do not necessarily destabilise meanings perceived as intended or optimally relevant. One might even say that deconstructionist readings which go beyond apparent intentions or surface meanings, and delve into individual word meanings, are ‘plainly and simply parasitical’ on ‘the obvious and univocal reading’ of a text. However Hillis Miller, in a well-known essay called ‘The Critic as Host’ (1991a), has challenged this very criticism, by investigating the etymology of the word parasite, through Greek, back to its Indo-European root, which clearly gives a number of possible meanings. Miller (1991a, p. 143) insists that ‘though a given word in “para” may seem to choose univocally one of these possibilities, the other meanings are always there as a shimmering in the word which makes it refuse to stay still in a sentence’. Supposedly univocal meanings are thus equivocal. Furthermore, the word parasite has no meaning without its counterpart, host, and ‘host is both eater and eaten’, containing ‘the double antithetical relation of host and guest, guest in the bifold sense of friendly presence and alien invader’ (p. 145). Miller shows that host and guest go back to the same etymological root (without bothering to mention that both meanings are still to be found in the modern French hôte). He therefore concludes that ‘ “the obvious or univocal reading” always contains the “deconstructive reading” as a parasite encrypted within itself as part of itself’ (p. 149), stating that To get so far or so much out of a little piece of language, context after context widening out from those few phrases to include as their

‘Negative Hermeneutics’ 57

necessary milieu all the family of Indo-European languages, all the literature and conceptual thought within those languages . . . is an argument for the value of recognizing the equivocal richness of apparently obvious or univocal language, even of the language of criticism. (Miller, 1991a, p. 148) There is no denying that this procedure can be fruitful, and that given such a radically diachronic conception of meaning, any ‘shimmering’ etymologically possible reading is indeed ‘encrypted’ in the text. But this is only a particular reading strategy. The whole family of IndoEuropean languages is a possible rather than a necessary milieu or cognitive environment. Etymology comes from the Greek etumos logos, authentic meaning. But authentic for whom? Just as every attorney knows that a jury will not be able to disregard ineligible information, even if instructed to do so by the judge, a philologist with a knowledge of the etymological and tropological history of a word will not ignore possibilities that existed in language prior to the writing of a poem. These potential meanings exist, even if they were not intended, and even less communicated, by the author, and will be manifestly relevant to any reader who shares Miller’s reading strategies. This is why Bakhtin, for example, champions the ‘unfinalizability’ or ‘unfinalizedness’ of texts: they can always be read by new readers coming from diverse and unforeseeable standpoints, who will enter into ‘dialogic’ interaction with the work and thereby actualize potential meanings and significances. But, contrary to the argument of most deconstructionists, if language is an inferential process none of these contexts, and no meanings thrown up by the possibilities of codes, are necessary.6 The trouble with basing interpretations on encrypted or coded meanings is that there is absolutely no limit to the number of potential codes. Ferdinand de Saussure, for example, thought he had discovered anagrams (or as he preferred to call them, hypograms or paragrams) of the names of Gods and heroes dispersed (as phonemes rather than letters) in the works of Lucretius, Horace, Ovid, Seneca and other Latin poets. He suggested that these hypograms functioned as thematic words that generated the poem. Given the limited number of phonemes or syllables available in any language, they tend to recur regularly in any form of discourse, but although one can probably find any name one chooses to look for spread out in six lines or so of any text, Latin poetry does indeed appear propitious for Saussure’s theory: he usually finds the names he wants within three lines. Saussure assumed that the Latin

58

Paradigms of Reading

reader would have been able to discern these sub-poetic words, but being unable to find any verification for his theory, he left his notebooks unpublished, while still apparently believing it plausible that this elementary and determining rule of versification was regularly passed on from master to pupil without ever once being so much as mentioned by a single theoretician of Latin verse. Jean Starobinski concludes his (1971) book on Saussure’s unpublished manuscripts with the remark that critics are generally unwilling to accept that a poet did not consciously or unconsciously will or intend whatever it is that their analysis reveals. Post Saussure, dispersed anagrams are just as much a potential code as etymologies. A further example of a strategy that might uncover hidden meanings arises from the fact that it was noticed shortly after the release of the Beatles’ Sergeant Pepper album in 1967 that there was a phrase recorded backwards at the end of the second side. This legitimated the practice of seeking hidden messages by turning records backwards (although this has become impossible with compact discs). In the 1980s, a crazed American television evangelist became briefly famous by announcing that the hook-line ‘Another one bites the dust’ on a record by the group Queen sounds like ‘It’s fun to smoke marijuana’ when played backwards (it doesn’t), and that all the kids listening to this song were being programmed to take drugs. He did not reveal how many hundreds or thousands of records he had had to spin backwards before finding a sound that could be given a suitably sinful (and imaginative) phonological and semantic representation, but who knows what he might find if he recorded Latin or English poetry and reversed the tapes. I would suggest, however, that Hillis Miller’s insistence that the entire family of Indo-European languages is a necessary milieu for modern English – and Paul de Man’s claims regarding the materiality and the necessity or the normative nature of tropologically based interpretations, which I discuss in succeeding chapters – are not intrinsically more plausible than the claims of the American evangelist. At the very least, they are open to the same objections that Michael Riffaterre made to the more elaborate analyses of the formalists, in particular Jakobson and Lévi-Strauss’s isolation of phonological, morphological and syntactic patterns in Baudelaire’s poem ‘Les Chats’. Riffaterre argues that not all the structural systems that linguists are able to find in a poem are necessarily poetic structures, relevant to the function or effect of a literary work of art. Structural linguists may well analyse constituents that are imperceptible to all readers except structural linguists (the proof being

‘Negative Hermeneutics’ 59

that prior to the structural analysis, no reader had ever publicly responded to Jakobson and Lévi-Strauss’s patterns) (Riffaterre, 1980; Jakobson and Lévi-Strauss, 1981). Structuralists and poststructuralists, of course, tend to reject Riffaterre’s argument. Jonathan Culler (1985, p. 43), for example, calls it ‘dubious’: ‘The idea that the poetically relevant might be simply identified with what the reader perceives is little more than a pragmatic restriction that permits critics to ignore exhausting formal analyses – analyses which, at the very least, had the virtue of drawing attention to aspects of babble, doodle, and riddle.’7 Culler questions ‘our inclinations to use notions of unity and thematic coherence to exclude possibilities that are manifestly awakened by the language and that pose a problem’ (1985, p. 45). But the possibilities arising from coded structures that Culler describes as ‘manifestly awakened’ are for relevance theorists equally manifestly non-intended. Culler describes poetry as containing elements of ‘babble’; spoken language similarly contains a large amount of what information theorists (and language teachers) call ‘redundancy’. All languages have developed a certain proportion of redundancy, so that extempore utterances invariably contain more elements than are strictly necessary for communication. This both reduces the amount of information in any particular utterance and helps a speaker construct a message. Without elements of redundancy – whether lexical, syntactic or phonetic – we would not be able to absorb information at the speed at which it could be emitted in normal speech. It has been estimated – with the help of logarithms and the theory of entropy – that the English language is no less than 50 per cent redundant (Weaver, 1949). Learners of foreign languages are trained, when listening, to ignore incidental elements and to concentrate on grasping the gist of an utterance, something that native speakers do automatically. Just as a theory from cosmology suggests that the universe must contain vast amounts of invisible particles called ‘dark matter’, so does the normal discourse of native speakers: fillers, hesitations, repetitions, corrections, anacolutha, etc.8 (Conversely, it is noticeable that the discourse of non-native speakers, such as the functional and instrumental ‘international English’ of business people, scientists and travellers, is almost wholly lacking in dark matter – or what Barbara Johnson calls ‘noise’ – as the interlocutors do not share idioms, clichés, a dialect, a cultural background, and so on.) Just as people in conversation do not pay much attention to semantically uninteresting dark matter, literary readers do not need to take account of all potentially encoded material. On the contrary, we can ignore unwanted semantic and grammatical possibilities just as we try to ignore

60

Paradigms of Reading

the annoying squeaking of guitarists’ fingers on the fretboard, the cracks in old master paintings, and the fact that in westerns, unrolling at twenty-four frames a second, wagon wheels usually seem to be turning the wrong way. Wittgenstein (1974, para. 4) pointed out that a viewer could mistakenly take one-inch high figures in a painting to represent one-inch people in the real world, and thereby misunderstand the conventions of painting. One could also try to open Andy Warhol’s Campbell’s soup cans with a can-opener. And one could insist on the importance of every potentially encrypted meaning in a sentence or longer piece of text.

Carvers and modellers In this and the previous chapter I have set up a distinction between interpreters who attempt to reconstruct authors’ communicative intentions, and those who prefer to apply texts to their own lives and maximise relevance by way of linguistic or thematic interpretive strategies that go way beyond the inferential recognition of authorial intentions. A further way of classifying these two groups of readers, which also has a potential psychoanalytic explanation, can be extrapolated from the work of the English art critic Adrian Stokes. In the 1930s, Stokes adopted two terms from sculpting – carving and modelling – to designate, metaphorically, the two most general attitudes an artist can adopt to his medium. Stokes’s distinction may fruitfully be further extended to include the audience of art objects, and in particular interpreters of literary texts and theorists of reading, who may be provisionally divided into those who regard texts as moreor-less self-sufficient objects from which an intended authorial (or historical or generic) meaning can be extracted, and those who consider that the meaning is supplied by the reader. In The Quattro Cento and Stones of Rimini, Stokes described what he considered the supreme attribute of a carver, which he identifies in the sculpture and architecture of fifteenth-century Italy: a love of stone or, more generally, of the artist’s chosen medium. The carver recognises and values the inherent qualities (colour, texture, grain, weight and shape) in the stone or marble he uses, and produces a work that conserves these qualities and respects the vitality and integrity of the material. In Colour and Form, Stokes extended the concept of carving to include painting, a love of colour replacing that of stone. The ‘carver’ – for example, Piero, Giorgione, Breugel, Vermeer or Cézanne – attributes a vitality to the surface of the canvas and endeavours to preserve it, by

‘Negative Hermeneutics’ 61

allowing colour to determine form. On the contrary, the modeller – for example, Michelangelo, Turner or Monet – whether working with clay, plaster or wax and casting in metal, or by extension, working with paint, subjugates his material to his own powers of self-expression. Modelling is usually an attempt to embody a conception, whereas in carving conception is adjusted to the life of the material that the sculptor feels beneath his tool. More generally, the modeller shapes the world into a mould for his own consciousness. 9 Stokes never extended his concepts to literature, but they can be used to distinguish between two kinds of readers and critics. Hermeneuticists and phenomenologists such as Schleiermacher, Dilthey, Hirsch and Poulet are clearly carvers who respect the otherness of the text and the informative intentions it contains, and do not seek to impose their own personality on it. The critics for whom the author’s text and intended explicatures and implicatures necessarily give way to the reader’s constructions can be described as modellers. Thus with or without the underpinning of Kleinian object relations theory, the notions of carver and modeller offer a visual analogy by which to distinguish between readers interested in communication, and readers who seek to maximise the relevance of texts without even attempting the inferential recognition of an author’s intentions.10 Relevance theory can explain both these ways of reading, but its detailed account of ostensive communication, the inferential nature of linguistic interpretation, and the recovery of inferred authorial communicative intentions by way of optimally relevant contextual assumptions, disproves the deconstructionist claim that it is impossible to correctly interpret an utterance or the words on a page, because the linguistic signs do not coincide with their intended meanings. ‘Modelling’ is a choice rather than an inevitability. In the following chapters I will outline Paul de Man’s deconstructionist accounts of ‘cognitive’ and ‘performative’ rhetoric, the arbitrary, mechanical and aberrant character of reference, the inhuman nature of language and the tropological construction of the self, and try to show how relevance theory offers a more accurate account of the same aspects of language and language use.

5 Words, Concepts and Tropes

The core notion of rhetoric has remained largely constant for almost 2000 years, as an amalgam of Aristotle and Cicero’s concept of speech designed to persuade and Quintilian’s educational concept of the science of speaking well and showing moral goodness. (The Platonic dismissal of rhetoric as mere flattery, insincerity and cosmetic ornament is, of course, even older.) In the past two decades, an increasing number of scientists have become aware of the importance of persuading and speaking well. Donald McCloskey (1985, 1990, 1994), for example, has called for a turning away from positivism and towards rhetoric, specifically in economics but implicitly in all the social and natural sciences.1 McCloskey is easily able to show that although a majority of economists pay lip-service to a positivist model dating from the 1940s, they actually use rhetoric, most of the time: ‘In italics: everyone, without exception, “uses rhetoric” in all their verbal or mathematical work, without exception. Rhetoric is merely speech that has designs on the reader [sic]’ (McCloskey, 1994, p. xiv). McCloskey’s account of the scope and tools of rhetoric is much broader than the classifications found in the classical authors, such as the three branches of oratory, the five stages of composition, the various parts of a speech, the three styles, the three modes of persuasion, and the most common tropes and figures. He writes of the ‘rhetorical tetrad’ of fact, logic, metaphor and story, and insists that all four are necessary for human reasoning. We associate facts, derived from induction, by way of stories, which are ways of making contiguous facts or events hang together. We classify similar things by way of deductive logic, or compare them by way of metaphors. We judge metaphors by what Peirce called abduction. (It is clear that McCloskey is using ‘metaphor’ as a synecdoche for rhetorical tropes as a whole.) McCloskey 62

Words, Concepts and Tropes 63

insists that fragments of the tetrad are not enough for full thinking. Scientists and scholars need to be literary as well as factual and logical: they should be able to tell factual stories and to spin logical metaphors. Economists, for example, do this all the time. Although positivist economists, suffering from ‘physics envy’, like to talk about hypotheses, verifiability, observable implications, meaningful statements, science vs. pseudo-science, prediction, hypothetico-deductive systems, axiomatization and formalization, McCloskey insists that 90 per cent of what they do is storytelling (although they think they’re merely constructing time-series), and that there’s also a lot of storytelling in the remaining 10 per cent, which is the construction of metaphors (though economists call them models instead). McCloskey argues that a scientist with half a culture is half a scientist, and that ‘to admit now that metaphor and story matter also in human reasoning does not entail becoming less rational and less reasonable. On the contrary . . . it entails becoming more rational and more reasonable, because it brings more of what persuades serious people under the scrutiny of reason’ (McCloskey, 1994, pp. 63–4). Rhetoric thus includes the supposedly impersonal elements of facts and logic, which are consequently not things a scientist can oppose to the dirty word rhetoric. After all, logic is not timeless and independent of human rhetoric, but needs to be put in the plural: we can appeal to Aristotelian, scholastic, first-order predicate, deontic, modal, relevant, multivalued, informal, intensional, counterfactual, epistemic, paraconsistent, relevant entailment, fuzzy and other logics. Similarly, reliably attested facts are language-dependent. McCloskey writes, rather blithely, ‘That a fact is a fact relative only to a conceptual scheme is no longer controversial, if it ever was. Kant knew it; so should we. Studies of science over the past few decades have shown repeatedly that facts are constructed by words’ (McCloskey, 1994, p. 41). He quotes Niels Bohr, who said ‘it is wrong to think that the task of physics is to find out how nature is. Physics concerns what we can say about nature. . . . We are suspended in language. . . . The word “reality” is also a word, a word which we must learn to use correctly’ (1994, p. 41). Similarly, Heisenberg stated that ‘natural science does not simply describe and explain nature; it is part of the interplay between nature and ourselves: it describes nature as exposed to our method of questioning’ (McCloskey, 1994, p. 41). Good reasoning clearly cares about facts, and is therefore empirical – which is not, of course, to say that it is empiricist.2 Facts may be discourse specific, but they are still facts. Numerous literary theorists have quoted J. L. Austin’s example (1962, p. 142) of an apparently

64

Paradigms of Reading

‘constative’ sentence, ‘France is hexagonal’, that for Austin is ‘true for certain intents and purposes. It is good enough for a top-ranking general, perhaps, but not for a geographer’. Hence ‘it is essential to realize that “true” and “false” . . . do not stand for anything simple at all but only for a general dimension of being a right and proper thing to say as opposed to a wrong thing in these circumstances, to this audience, for these purposes and with these intentions’ (1962, p. 144). Yet Austin always stressed that facts are genuine entities and that words are used to talk about realities; he once stated this pithily as ‘fact is richer than diction’ (1979, p. 195). McCloskey insists that rhetorically aware economists would reassess their devotion to allegedly value-free hypotheses, without abandoning the ordinarily useful fact/value distinction. A non-positivist, rhetorical economics would be more, not less, rigorous, because it would face up to more arguments. McCloskey recommends conversation, guided by the qualities Habermas describes as making up a Sprachethik, in which participants let other people talk, pay attention, cooperate, keep open minds, explain themselves when asked, and don’t lie, shout, sneer or resort to violence or conspiracy in aid of their ideas. He insists that scientists know when a conversation in their field is working well, and don’t require a methodology or epistemological rules to guide them. A conversation will work well when its participants are courteous, accurate in their quotations, honest in their handling of factual data, serious in their scholarship, and so on. These participants will also obviously be constrained by the presence and character and attitudes of their audience, the language spoken in common and the customary style used, the history of earlier and similar talk, the practical purpose to be achieved by communication, and so forth. In short, there need be no distinction between what we call science and what the classical rhetoricians called deliberative rhetoric or oratory. Yet by the time McCloskey began describing science in terms of persuasion, Paul de Man had re-described literary language, by which he meant virtually all language, including the elements of syntax and grammar, as a mechanical and inhuman system of tropes, generally detached from any expressive or persuasive function. In the 1660s, Samuel Butler ridiculed his protagonist Hudibras’s excessive and mannered use of rhetoric: ‘he could not ope / His mouth, but out there flew a trope’ (1905, canto 1, 80–81). In the 1970s de Man tried to show, by various routes, that we all unwittingly share this predicament, as all words, names and concepts are tropological at their inception. This is by no means a recent argument: de Man draws on Rousseau’s Second

Words, Concepts and Tropes 65

Discourse, and on Nietzsche’s essay ‘On Truth and Lying in an Extra-Moral Sense’ and his 1872–3 lecture notes for a course on rhetoric, which in turn were heavily dependent on Gustav Gerber’s recently published Die Sprache als Kunst, which drew on Herder, Humbold, Grimm, Bopp, Goethe, Jean Paul, Hegel, Schopenhauer and others. Furthermore, this argument, with its distinguished pedigree, may indeed be true. Yet de Man goes on from here to argue that rhetoric, which he defines in terms of tropology rather than persuasion, is incompatible with both grammar and logic. For de Man, neither grammar nor intended reference can contain figuration, which invariably undoes or displaces grammatical cognition, rendering the propositional content of utterances undecidable, and wreaking ‘epistemological damage’ (Aesthetic, p. 34) that wholly undermines intended meaning. Rhetoric thus ‘radically suspends logic and opens up vertiginous possibilities of referential aberration’ (Allegories, p. 10), making it impossible to decide between incompatible and mutually self-destructive literal and figural meanings. It ‘puts an insurmountable obstacle in the way of any reading or understanding’ (Allegories, p. 131), so that ‘the impossibility of reading should not be taken too lightly’ (Allegories, p. 245). De Man tended to subvert and to allegorise all his key concepts, including ‘trope’ and ‘rhetoric’, but in this chapter and the next I will attempt to pin down and outline his account of rhetoric, and argue that even if all words, names and concepts were indeed originally tropes, de Man’s conclusions regarding the ‘insurmountable obstacle’ of rhetoric do not follow.

Nouns and concepts In ‘Metaphor (Second Discourse)’, in Allegories of Reading, de Man interprets Rousseau as arguing that all concepts are metaphorical. Yet this involves translating Rousseau’s term idée générale as ‘metaphor’ rather than ‘common noun’. In the Discourse on the Origins and the Foundations of Inequality among Men (or for brevity’s sake, the Second Discourse) (1754), Rousseau gives a short account of the genesis of general ideas or concepts in primitive man. He describes the process of conceptualisation at the simplest level as the invention of common nouns on the basis of the perception of resemblances between objects. He imagines a primitive man (in the ‘state of nature’) first giving different names to every object, so that two oak trees might be called, say, A and B, ‘for the first idea we derive from two things is that they are not the same; it often takes a great deal of time to observe what they have in common’.3 Later, on the basis of resemblances

66

Paradigms of Reading

perceived during accumulated observations, our primitive man was led to substitute an abstraction, a word signifying tree, for his particular denominations, entity A and entity B. The perception of resemblances alone – something of which all animals are presumably capable – is not the same as conceptualisation, the process of forming general ideas, which is verbal. Rousseau writes, ‘When a monkey goes without hesitation from one nut to another, do we think that he has in mind a general idea of this type of fruit and that he compares his archetype to these two individual entities? Certainly not’ (1964, p. 149, quoted in Allegories, p. 145). This is because ‘it is necessary to state propositions and to speak in order to have general ideas’ (1964, p. 150; Allegories, p. 145). Immediately after summarising Rousseau’s narrative about primitive man and oak trees, however, de Man redescribes conceptualisation as a form of rhetorical substitution: The text indeed distinguishes the act of naming (tree A and tree B) which leads to the literal denomination of the proper noun, from the act of conceptualization. And conceptualization, conceived as an exchange or substitution of properties on the basis of resemblance, corresponds exactly to the classical definition of metaphor as it appears in theories of rhetoric from Aristotle to Roman Jakobson. The text would then, in a sense, distinguish between, on the one hand, figurative, connotative, and metaphorical language and, on the other, denominative, referential and literal language, and it would oppose the two modes antithetically to each other. (Allegories, p. 146) In a footnote, de Man quotes from the Poetics (1457b): ‘Metaphor is the transfer [epiphora] to a thing of a name that designates another thing, a transfer from the genus to the species or from the species to the genus according to the principle of analogy.’ In this passage, de Man hedges – the text would, in a sense . . . – but he soon drops the conditional form to assert that Rousseau’s idée générale is a metaphor rather than a common noun. Yet the perception of common properties among n units which leads to the positing of a generic term (a common rather than a proper noun) is not at all the same process as the perception of partial analogies between two compared objects which leads to the transference of the name of the one to the other. Rousseau does not equate the formation of a concept with the invention of a metaphor (see Reichler, 1990). Calling an oak an oak is not the same as saying that ‘all high poetry is infinite; it is as the first acorn

Words, Concepts and Tropes 67

which contained all oaks potentially’, as Shelley does in his ‘Defence of Poetry’ (1965, vol. VII, p. 131). The last conceptualisation of a primitive man – ‘Hmm, this large, gregarious, tawny-coloured, predatory, carnivorous, feline, quadruped mammal with a flowing shaggy blackish mane and a tufted tail that is about to devour me is indeed similar to that one: I shall call them lions!’ – is not at all the same process as calling Achilles a lion, or naming a constellation of stars in the zodiac, or saying that the righteous are as bold as a lion (Proverbs 28:1), or describing the devil as a roaring lion (1 Peter 5:8), or calling the puny puma or cougar an American mountain lion (transferring from the species to the genus, as in Aristotle’s definition). De Man swiftly proceeds to collapse what had appeared to be a distinction in Rousseau between particular denominations or proper nouns (entity A and entity B) and metaphorical conceptualizations. Rousseau writes that ‘the first inventors [of words] were able to give names only to the ideas they already possessed’ (1964, p. 150), and de Man recalls Rousseau’s note claiming that ‘the first idea we derive from two things is that they are not the same.’ He adds that the word idea, despite all pre-Kantian empiricist concreteness, denotes the presence of some degree of conceptuality (or metaphor) from the start, within the very act of naming. . . . But if all entities are the same, namely entities, to the extent that they differ from each other, then the substitution of sameness for difference that characterizes, for Rousseau, all conceptual language is built into the very act of naming, the ‘invention’ of the proper noun. It is impossible to say whether denomination is literal or figural: from the moment there is denomination, the conceptual metaphor of entity as difference is implied, and whenever there is metaphor, the literal denomination of a particular entity is inevitable [because the human mind cannot imagine a tree or a triangle in the abstract, but must visualize a particular one]. (Allegories, p. 148) Having turned proper nouns into figures, de Man extends his logic of dereification, arguing that since figures are necessarily linguistic, all concepts are about language rather than things: conceptualization, as the passage in the Second Discourse on the naming of trees makes clear, is an intralinguistic process, the invention of a figural metalanguage that shapes and articulates the infinitely fragmented and amorphous language of pure denomination. To the

68

Paradigms of Reading

extent that all language is conceptual, it always already speaks about language and not about things. . . . Denomination could never exist by itself although it is a constitutive part of all linguistic events. All language is language about denomination, that is, a conceptual, figural, metaphorical metalanguage. (Allegories, pp. 152–3) For de Man, it follows that ‘if all language is about language, then the paradigmatic linguistic model is that of an entity which confronts itself’ (Allegories, p. 153), as in the ‘exemplary situation’ that Rousseau presents as a narrative parable or allegory in the Essai sur l’origine des langues, about a primitive man meeting other men, experiencing fright, and giving them the metaphorical rather than denominative name giants. Later, after many experiences, or accumulated observations, when the primitive man discovers that the supposed giants are no larger or stronger than he is, he abandons his previous metaphor and coins the word ‘man’. The first conceptualisation – giant – caused by fear, is clearly, as de Man puts it, ‘a wild, spontaneous metaphor which is, to some degree, aberrant’ (Allegories, p. 153). But de Man goes on to describe the second conceptualization – man – ‘the result of a quantitative process of comparison based on measurement, and making deliberate use of the category of number in order to reach a reassuring conclusion’ (Allegories, p. 154), as equally aberrant, because it ‘stems from the use of number as if it were a literal property of things that truly belongs to them, when it is, in fact, just one more conceptual metaphor devoid of objective validity and subject to the distortions that constitute all metaphors. . . . The idea of number is just as derivative and suspect as the idea of man’ (Allegories, p. 154). The implications, for de Man, are far-reaching: the invention of the word man makes it possible for ‘men’ to exist by establishing the equality within inequality, the same within difference of civil society. . . . The concept interprets the metaphor of numerical sameness as if it were a statement of literal fact. Without this literalization, there could be no society. . . . Conceptual language, the foundation of civil society, is also, it appears, a lie superimposed upon an error. We can therefore hardly expect the epistemology of the sciences of man to be straightforward. (Allegories, p. 155)

Words, Concepts and Tropes 69

If, as de Man argues, all nouns were once conceptual metaphors, then numbers would clearly be no different, but it is hard to see exactly what is to be gained by describing number as a ‘conceptual metaphor devoid of objective validity’ and as ‘derivative and suspect’. Conjecture concerning the figurative, connotative and conceptual, or the literal, denominative and referential origin of language in no way changes its function, and we continue to live with (and attempt to improve) political, legal and ethical systems, undeterred by the notion that civil society and the sciences of man depend on ‘a lie superimposed upon an error’.

Metaphor, truth, lies, realistic assumptions and surplus value Rousseau’s argument about concepts and common nouns was echoed by Nietzsche in ‘On Truth and Lying in an Extra-Moral Sense’ (written in 1873 but only published posthumously): The ‘thing-in-itself’ (which would be pure, disinterested truth) is also absolutely incomprehensible to the creator of language and not worth seeking. He designates only the relations of things to men, and to express these relations he uses the boldest metaphors. First, he translates a nerve stimulus into an image! That is the first metaphor. Then the image must be reshaped into a sound! The second metaphor. [ . . . ] When we speak of trees, colors, snow, and flowers, we believe we know something about the things themselves, although what we have are just metaphors of things, which do not correspond at all to the original entities. . . . In any case, the origin of language is not a logical process, and the whole material in and with which the man of truth, the scientist, the philosopher, works and builds, stems, if not from a never-never land, in any case not from the essence of things. Let us think in particular of the formation of concepts. Every word becomes a concept as soon is it is supposed to serve not merely as a reminder of the unique, absolutely individualized original experience, to which it owes its origin, but at the same time to fit countless, more or less similar cases, which, strictly speaking, are never identical, and hence absolutely dissimilar. Every concept originates by the equation of the dissimilar. Just as no leaf is ever exactly the same as any other, certainly the concept ‘leaf’ is formed by arbitrarily dropping those individual differences, by forgetting

70

Paradigms of Reading

the distinguishing factors, and this gives rise to the idea that besides leaves there is in nature such a thing as the ‘leaf’, i.e. an original form according to which all leaves are supposedly woven, sketched, circled off, colored, curled, painted, but by awkward hands so that not a single specimen turns out correctly and reliably as a true copy of the original form. (Nietzsche, 1989, pp. 248–9) This passage, which begins by opposing the notion of Platonic forms, not only echoes Rousseau, but sounds in places very de Manian (which is, of course, to say that de Man’s style is clearly influenced by Nietzsche). There is gratuitous exaggeration (‘strictly speaking . . . never identical’ equals ‘absolutely dissimilar’), and the wholly unmotivated use of ‘arbitrarily’ in ‘certainly the concept “leaf” is formed by arbitrarily dropping those individual differences’. There is absolutely nothing arbitrary about the conceptualisation that results in the word ‘leaf’: forgetting the individual differences does not so much give rise to the idea that there is in nature an original form of a leaf, as give us a word that allows us to talk about the green things that grow on trees in spring and turn brown and fall off in autumn. 4 A possible parallel with the process of conceptualisation that results in common nouns is the philosophy of ‘as if’, or in particular Milton Friedman’s logic of the unrealism of assumptions, outlined in his canonical essay ‘The Methodology of Positive Economics’ (1953). Friedman recapitulates the standard arguments that the ultimate goal of a positive science is the development of a ‘theory’ or ‘hypothesis’ that yields valid predictions about phenomena not yet observed, and that the only relevant test of the validity of a hypothesis is comparison of its predictions with experience. (Factual evidence can never ‘prove’ a hypothesis; it can only fail to disprove it, which is generally what we mean when we say that a hypothesis has been ‘confirmed’ by experience.) Yet Friedman insists that hypotheses only have implications, and not assumptions whose ‘realism’ (by which he means plausibility or truth) can be judged independently of the validity of predictions, or whose conformity to ‘reality’ could be a test of the validity of the hypothesis different from or additional to the test by implications. Indeed insofar as a theory can be said to have ‘assumptions’ at all, the relation between its significance and the ‘realism’ of its ‘assumptions’ is almost the opposite of what many people suppose: truly important and significant hypotheses necessarily have ‘assumptions’ that are wildly inaccurate descriptive representations of reality. In general, the more

Words, Concepts and Tropes 71

significant the theory, the more unrealistic the assumptions (in this sense). A hypothesis that is supposed to explain something important (and permit valid predictions) necessarily abstracts common and crucial elements from a mass of complex and detailed circumstances surrounding the phenomena to be explained. (For example, the general equilibrium model in economics assumes perfect competition and information, full employment, totally flexible prices, etc.) In other words, an important hypothesis must be descriptively false in its assumptions, as it will fail to take account of the many other attendant circumstances that it shows to be irrelevant. Thus the ‘assumptions’ of a theory are never descriptively ‘realistic’, but are rather sufficiently good approximations for the purpose in hand. Friedman gives some examples. The relative density of leaves on the south side of trees (in the northern hemisphere), except on trees on north-facing slopes or trees whose south side is in some way shaded, makes it seem as if leaves positioned themselves in such a way as to maximise the amount of sunlight they would receive, and moved effortlessly around trees to that effect, filling unoccupied sunny locations, even though we know that leaves do not calculate or move, and that trees passively adapt to sunshine. In microeconomics (and this was Friedman’s point), according to marginal theory, businesses behave as if they were seeking rationally to maximise their expected returns, and had full knowledge of the data needed to succeed in this attempt. (More technically, it is as if they knew the relevant cost and demand functions, calculated marginal cost and marginal revenue from all actions open to them, and pushed each line of action to the point at which the relevant marginal cost and marginal revenue were equal.) Even if the businessmen we know personally are not as selfish and money-grubbing as the dismal science supposes, Friedman insists that marginal theory predicts actual behaviour, on the whole, and that businesses which didn’t seek to maximise returns wouldn’t last very long. The parallel with conceptualisations and common nouns is clear: we use them as if they truthfully designated nearly-identical objects, which allows us to converse about them. In ‘On Truth and Lying’, Nietzsche follows his account of concepts originating from the equation of the dissimilar with an account of the metaphorical nature of truth that was destined to become one of the most frequently quoted passages in literary theory: What is truth? a mobile army of metaphors, metonyms, anthropomorphisms, in short, a sum of human relations, which were

72

Paradigms of Reading

poetically and rhetorically heightened, transferred, and adorned, and after long use seem solid, canonical and binding to a nation. Truths are illusions about which it has been forgotten that they are illusions, worn out metaphors without sensory impact, coins which have lost their image and now can be used only as metal, and no longer as coins.5 (Nietzsche, 1989, p. 250) If this is indeed the case, there are clearly considerable implications for epistemology, but not necessarily for the use of language. Even if concepts come into existence, and words are coined, as metaphors, and their metaphorical origin is soon forgotten, they still function, and constitute a sum of human relationships. Words whose metaphorical origins have been forgotten are less like worn-out coins than banknotes which function successfully as a means of exchange, despite the fact that we know that they have no intrinsic value.6 (British paper money carries the promise, by the Governor of the Bank of England, to pay the bearer a certain sum on demand, even though this sum will equally be a piece of paper; American bills too are no longer backed by gold, stating instead ‘In God We Trust’.) But what is important about truths and words and currencies is what they do, not what they are. A further economic analogy can be made with Marx’s account of commodities in the remarkable opening chapter of Capital – a peculiar mixture of classical economics, irony, metaphor and a parody of idealist Hegelian metaphysics, interspersed with bizarre asides and literary references. Although commodities are designed to be used for various purposes, under capitalism they become objects to be exchanged, and their use value is wholly eclipsed. Marx writes that ‘the wealth of those societies in which the capitalist mode of production prevails, presents itself as an “immense collection of commodities,” its unit being a single commodity’ (Marx, 1954, p. 43). Commodities, in turn, present themselves as having natural properties which satisfy human wants, but for Marx this is an illusion because commodities, strictly speaking, are not physical entities or objects of sense perception at all. He warns that although ‘a commodity appears, at first sight, a very trivial thing, and easily understood . . . its analysis shows that it is, in reality, a very queer thing, abounding in metaphysical subtleties and theological niceties’ (1954, p. 76). For example, a few pieces of wood can be sawn, sanded, varnished and made into a table. But a table as a commodity is merely a value, not a use-value but an exchange-value,

Words, Concepts and Tropes 73

‘a congelation of undifferentiated human labour’ (p. 68), that of workers hired by capitalists in a competitive market. For Marx, commodities as exchange-values no longer contain an atom of use-value. Despite their semblance of diversity, they are merely different manifestations and depositories of exchange-value, so that, for example, ‘20 yards of linen = 1 coat or = 10 lbs. tea or = 40 lbs. coffee or = 1 quarter corn or = 2 ounces gold or = 1/2 ton iron or = etc.’ (p. 68). Of course this is not a very practical way of expressing value, so the next step is to select one commodity with which to express the value of equivalent amounts of other commodities, so that we could say, for example, that one coat, and 10 lbs. of tea, and 40 lbs. of coffee, and a quarter of corn, and two ounces of gold, and half a ton of iron, etc., all equal 20 yards of linen. From here it is a short step to ‘the Money-form of value’ in which money replaces linen as the truly equivalent form of value, so that one coat, and 10 lbs. of tea, and 40 lbs. of coffee, and a quarter of corn, and 20 yards of linen, and half a ton of iron, etc., all equal two ounces of gold. The self-evident irony and absurdity of Marx’s exposition of the theory of value are designed to reveal what he sees as the mystical character of gold and silver, and to demonstrate that the capitalist economy, based on the value of commodities, is as absurd as his account of it. The concept of exchange-value is the basis of Marx’s theory of surplus-value, and the nexus of all his major ideas concerning the alienation and exploitation of labour and the class conflict that characterise the capitalist mode of production. (The theory of surplus-value has also been widely rejected, by every mainstream economic school since the marginalists.) Yet notwithstanding the transformation of useful or aesthetic objects into means to another end – the creation of value or profit – and the consequent alienation of labour, both commodities as exchange-values and money, the ultimate exchange value, do have their practical uses. 7 Perhaps pure usable objects, and pure use-value untainted by considerations of exchange, are as illusory as non-figural language, or non-metaphorical truth. Truth may be an illusion or a metaphor, but our metaphorical truths have a pragmatic value, if not an intrinsic one, and enable (or constitute) a sum of human relationships. Our words may not refer directly to the world, but our metaphors, exchanging or substituting properties or values of objects, are usable coins. As Nietzsche writes, man ‘forgets that the original intuitive metaphors are indeed metaphors and takes them for the things themselves’ (1989, p. 252). Thus we use words as if they were coins, without recognizing their

74

Paradigms of Reading

metaphorical or illusory nature even though, diachronically, they can be shown to be merely worn-out pieces of metal. It may be that we obey ‘the obligation to lie according to an established convention, to lie collectively in a style that is mandatory for everyone’, and that man ‘forgets that this is his situation; so he lies in the designated manner unconsciously and according to centuries-old habits – and precisely by this unconsciousness, by this forgetting, he arrives at his sense of truth’ (p. 250). It may be that man ‘has an unconquerable tendency to let himself be deceived’ (p. 255), but if this is the price of using conceptualisations and common nouns, so be it.

Numbers In ‘Metaphor (Second Discourse)’, de Man describes number as a ‘conceptual metaphor devoid of objective validity’ and as ‘derivative and suspect’ (Allegories, p. 154); in ‘Pascal’s Allegory of Persuasion’, he also discusses zero, which ‘is radically not a number, absolutely heterogeneous to the order of number’ (Aesthetic, p. 59). It is evident that numbers are signs, and that, like points, lines, spaces and functions, they do not have an external, referential, pre-mathematical existence, and that the easiest way to demonstrate this is to examine the sign zero, which appeared in the West in the thirteenth century, as a sign about signs, a meta-sign indicating the absence of other mathematical signs. At any place within a Hindu or Arabic numeral, the mathematical sign zero indicates the absence of the numerals 1, 2, 3, 4, 5, 6, 7, 8 and 9 rather than the non-presence of any real ‘things’ that are supposedly independent of or prior to signs which represent them. Thus it can be argued that numbers, far from constituting a system that produces infinitely many signs (numerals) that re-present objects existing in a prior reality, do not in fact exist anterior to the process of assignation performed by the system. The system depends on a meta-sign that both initiates the signifying system and participates in it as a constituent sign. The double, ambiguous role played by zero deconstructs the supposed anteriority to signs of ‘reality’. Quite apart from zero, other signs of the numerical system can become creative and autonomous: we can write numerals that name ‘numbers’ that are unrealisable by any conceivable process of human counting or enumeration. The system becomes the source of reality, and provides the means for describing this ‘reality’ as if it were some domain external and prior to itself. At the end of the sixteenth century (and thus long before Saussure argued that words only have meaning in relation to

Words, Concepts and Tropes 75

other signs within the system), the Dutch mathematician Simon Stevin argued that zero was the proper origin of number, so that all other numbers similarly lack referentiality, or positive content, and are produced by and within arithmetical notation. The same applies to the variables used in algebraic equations, formulas, inequalities, identities, and so on: signs such as x and y denote arbitrary individual numbers which can be substituted for them. The meaning of a variable lies in other, necessarily absent signs; the variable is a metasign representing the virtual, potential but not actual presence of any of the particular signs within its range. Variables thus require an active, constructing, counting subject. This destroys the realist epistemology and ontology of unauthored, subjectless, descriptive referents – numbers, points, lines, invariants, geometrical figures, ratios, constructions, functions, spaces, predicates, relations, sets, and so on – that exist in an extra-linguistic universe, independent of human agency, about which mathematics can make unambiguously true or false statements. Yet although a semiotic account of zero such as this is more coherent than the traditional, realist model, it changes nothing: when zero is conventionally accepted as just a number among numbers its meta-lingual role as the origin of these numbers is forgotten and it becomes a mere naturalized sign.8 De Man’s discussion of zero is part of a reading of Pascal’s ‘Réflexions sur la géométrie en général: de l’esprit géométrique et de l’Art de persuader’, an essay whose title alone disproves Aristotle’s claim that no one would use rhetoric when teaching geometry. De Man gives what is now a more or less orthodox account of zero – that it is not a number, and heterogeneous to the number system – and also discusses one, which ‘as Euclid said . . . is not a number. It is a mere name given to the entity that does not possess the properties of number, a nominal definition of nonnumber’ (Aesthetic, p. 58), but which de Man is easily able to demonstrate is also a number. But he is more interested in Pascal’s discussion of the classical distinction between nominal and real definition, and his assertion that mathematics and geometry recognise only nominal definitions, which give a name ‘only to those things which have been clearly designated in perfectly known terms’ (de Man’s translation, p. 55). Definitions of names in no way influence the substance or properties of the things named, and are thus ‘never open to contradiction’. But Pascal goes on to discuss ‘primitive words’ which de Man says ‘include the basic topoi of geometrical discourse, such as motion, number, and extension’, and which ‘are not subject to definition at all, since their pretended definitions are infinite regresses of accumulated tautologies’ (p. 56). If

76

Paradigms of Reading

primitive words possessed a natural meaning, this would have to be universal, like the science that operates with them, but Pascal writes: It is not the case that all men have the same idea of the essence of the things which I showed to be impossible and useless to define . . . (such as, for example, time). It is not the nature of these things which I declare to be known by all, but simply the relationship between the name and the thing, so that on hearing the expression time, all turn (or direct) the mind toward the same entity. (De Man’s translation, p. 55) De Man elaborates: since these words do not function as signs or names, as in nominal definition, the signs have become tropes, substitutive relationships that have to posit meanings whose existence cannot be verified. The semantic function of primitive terms such as motion and number is structured like a trope. To this one might easily retort that just like the semiotic account of zero outlined above, this changes nothing: if on hearing the expression time, everyone directs the mind toward the same entity, it becomes a mere naturalised sign. Similarly, insisting that numbers are metaphorical signs that do not simply mirror prelinguistic referents changes nothing if we conventionally use them as if they did. To describe civil society as ‘a lie superposed upon an error’ because it depends on the numerical concept of men is to say nothing of any consequence. The metaphorical nature of numbers might problematise the epistemology of all the sciences of man, but it leaves the sciences of man themselves wholly untouched.

Dead metaphors and catachreses In ‘The Epistemology of Metaphor’, de Man seeks to demonstrate the equivalence of concepts and tropes by another route. He begins by analysing Locke’s theory of language and his well-known attack on rhetoric in the Essay Concerning Human Understanding. Locke famously argues that all the art of rhetoric, besides order and clearness, all the artificial and figurative application of words eloquence hath invented, are for nothing else but to insinuate wrong ideas, move the passions, and thereby mislead the judgment, and so indeed are perfect cheat . . . they are certainly, in all discourses that pretend to inform or instruct, wholly to be avoided and, where truth and knowledge are

Words, Concepts and Tropes 77

concerned, cannot but be thought a great fault either of the language or person that makes use of them. (Locke, 1987, Book 3, ch. 10) De Man, of course, goes on to point out that a couple of pages later (in ch. 11), Locke describes language as a ‘conduit’ and warns that an ill use of it may ‘break or stop the pipes whereby [knowledge] is distributed to public use’, thereby committing the sins he has just denounced.9 This much is standard, but de Man goes on – predictably – to describe Locke’s own theory of words and language as ‘in fact a theory of tropes’ (Aesthetic, pp. 36–7). Locke describes words as simply and unproblematically substituting or standing for ideas. While a word can unequivocally signify an idea, ‘the names of simple ideas are not capable of any definitions’ (1987, Book 3, ch. 4). For example, Locke asks, how does one define the idea ‘motion’? To explain it in terms of passage ‘is to translate and not to define’. A number of possible solutions to Locke’s problem have been suggested in the twentieth century. Locke himself stressed the arbitrariness of the signifier; Saussure (1974, p. 120) added the arbitrariness of the signified, and the suggestion that a word, for example motion, simply means what all the other words in the language do not mean. Logico-semanticists have advanced notions about meaning postulates and componential analysis (see Lyons, 1978, volume 1, ch. 9). Peirce (1931–58, volume 1, para. 339; volume 2, para. 228; volume 5, paras 484, 487) described word meaning in terms of the interpretants (further words) they habitually produce; Wittgenstein proposed the concept of ‘language in use’; Vološinov (1973, p. 86) described word meaning as being negotiated between speaker and hearer; while Stanley Fish describes meaning as determined by broader ‘interpretive communities’. De Man, however, examines Locke’s problem and writes: Locke’s own ‘passage’ is bound to continue this perpetual motion that never moves beyond tautology: motion is a passage and passage is a translation; translation, once again, means motion, piles motion upon motion. It is no mere play of words that ‘translate’ is translated in German as übersetzen which itself translates the Greek meta phorein or metaphor. (Aesthetic, p. 38) This much is undeniable. Locke proposes motion as an example of a simple idea, and de Man can easily show that, etymologically, this can

78

Paradigms of Reading

mean translation, which in turn means metaphor. Locke’s second example of a word for a simple idea is ‘light’, which again is fortuitous for de Man, given that the word ‘idea’ (eide), of course, itself means light, and to say that to understand light is to perceive the idea of light is to say that understanding is to see the light of light and is therefore itself light. The sentence: to understand the idea of light would then have to be translated as to light the light of light (das Licht des Lichtes lichten), and if this begins to sound like Heidegger’s translations from the pre-Socratics, it is not by chance. (Aesthetic, p. 38) By establishing pseudo-equivalences by way of regressive etymologies such as these, de Man is led to the conclusion that ‘the discourse of simple ideas is figural discourse or translation’ (Aesthetic, p. 38), so that, by extension, Locke’s theory of ideas is a theory of tropes. Yet he goes on to write that ‘etymons have a tendency to turn into the repetitive stutter of tautology’ (p. 38), so that ‘the utterer of simple ideas’ is ‘something of a stuttering idiot’ (p. 39). This in itself would seem to be a reason to turn to a pragmatic account of word meaning, rather than to seek the passage to enlightenment in definition or etymology. Wittgenstein (1958, section 43), for example, suggested that ‘for a large class of cases – though not for all – in which we employ the word “meaning” it can be defined thus: the meaning of a word is its use in the language’. Similarly, ‘To understand a proposition is to understand a language’ (Wittgenstein, 1974, section 84). Words cannot be defined referentially or ostensively, but only according to their use, in that they are chosen in preference to other words. Etymology may take us from motion to passage to translation, but when we use these words (express these concepts), they are not in need of definition, and we are generally not conscious of using a trope. (For example, all the removal vans in Greece have ‘meta phorein’ written on the side, but even so, if you hire a removal firm when changing apartments in Athens you expect the van to arrive equipped with furniture movers rather than translators.) Locke moves on from simple ideas to substances, and one of his examples is gold. The colour (rather than the substance) gold is to be found in peacocks’ tails. Locke writes, I think all agree to make [gold] stand for a body of a certain yellow shining colour; which being the idea to which children have annexed

Words, Concepts and Tropes 79

that name, the shining yellow part of a peacock’s tail is properly to them gold. (Locke, 1987, Book 3, ch. 9) In other words, according to Locke, children may erroneously think of the gold in peacocks’ tails as the substance rather than the colour. De Man writes that ‘the child who cannot tell the figural from the proper keeps recurring throughout eighteenth-century epistemology as a barely disguised figure of our universal predicament’ (Aesthetic, p. 39). Yet this supposedly universal predicament – the fact that, even though most of us grow out of childhood, we cannot infallibly separate the figural from the proper, which calls into question the very notion of epistemology – has in fact only ever been a problem for the anti-rhetorical philosophers, from Plato onwards, who fear what de Man calls the ‘epistemological damage’ wrought by metaphors, tropes and figural language in general. It also seems to be a problem for de Man. From Locke’s gold example, and others, he makes the generalisation that word-meaning cannot be controlled or authorised: the free usage of ordinary language is carried, like the child, by wild figuration which will make a mockery of the most authoritarian academy. We have no way of defining, of policing, the boundaries that separate the name of one entity from the name of another. (Aesthetic, p. 39) This also follows from de Man’s account of the metaphorical origin of all concepts and proper nouns, which ‘arbitrarily’ suppress differences: if an infinite number of different objects can be subsumed under the same noun, this word can also logically be expanded outward to refer to any number of contiguous, related, or only metaphorically connected entities. De Man continues with a reading of Condillac’s account of conceptual ‘abstractions’ (which we could safely translate as ‘metaphors’) as both unavoidable in discourse and defective or corruptive. In his Essai sur l’origine des connaissances humaines, Condillac argued that ‘true reality’ is not located in things but in the subject. Paraphrasing Condillac, de Man writes that this reality ‘is the result of an operation the mind performs upon entities, an apperception . . . and not a perception’ (Aesthetic, p. 44). Things only become real by being appropriated or seized by the mind, because ‘entities, in themselves, are neither distinct nor defined; no one could say where one entity ends and where another begins. They are mere flux, “modifications” ’ (p. 44). What the mind seizes is

80

Paradigms of Reading

concepts with arbitrary boundaries, and de Man declares that ‘as soon as one is willing to be made aware of their epistemological implications, concepts are tropes and tropes concepts’ (p. 43). Consequently, ‘the use and the abuse of language cannot be separated from each other. “Abuse” of language is, of course, itself the name of a trope: catachresis’ (p. 41). Furthermore, ‘something monstrous lurks in the most innocent of catachreses: when one speaks of the legs of the table or the face of the mountain, catachresis is already turning into prosopopeia, and one begins to perceive a world of potential ghosts and monsters’ (p. 42). In ‘Hypogram and Inscription’, de Man goes further, suggesting that ‘as . . . anyone who has ever wondered about the legs of a table or, like Wordsworth, about the faces or the backs of mountains, knows, prosopopeia is hallucinatory. To make the invisible visible is uncanny’ (Resistance, p. 49). It seems more likely that only Paul de Man begins to perceive an uncanny and hallucinatory world of potential ghosts and monsters. For most people, the metaphorical or catechrestic use of language is generally without grievous consequences. Although there were Victorians who disguised piano legs with specially-made pieces of cloth, because naked legs were deemed indecent, for most people the name ‘piano leg’ is a dead metaphor that goes wholly unnoticed.10 De Man continues ‘The Epistemology of Metaphor’ by citing section 59 of the Critique of Judgement in which Kant distinguishes between schematic and symbolic hypotyposes. According to Kant, so-called mathematical ‘symbols’ are misnamed, because there is no iconic element in their representation of the object (they bear no resemblance to the objects they represent). The mathematical signs for plus and minus and the square root are not iconic symbols in the way that, for example, a triangle is. (The example is de Man’s rather than Kant’s.) There is a relation of similarity or resemblance in both schemata, or objects of the mind (such as, for example, geometric shapes like the triangle), and in (genuine) symbols, or objects of reason, such as Condillac’s abstractions or metaphors, where the similarity is analogical. Kant seems to be suggesting that what he calls schematic hypotyposes are epistemologically reliable, a notion that de Man rejects. Yet he seems to do this by first stating, and then denying, that the word ‘triangle’ is in fact a trope. A triangle is ‘a figure . . . having three angles and three sides’ (Shorter OED), as is indeed suggested by the word itself. If, in geometry, we used the word ‘delta’ instead (or better still, a word that had been specially invented to describe the mouth of a river like the Nile or Ganges and which did not fall back on the triangle-shaped Greek letter), there would be no doubt that it was a trope. Yet even in

Words, Concepts and Tropes 81

this instance, for most people it would probably be an unrecognized dead metaphor. Similarly, when we use the word ‘circle’ we rarely stop to reflect that it is a Latin invention from the word ‘circus’, itself deriving from Greek, or that, as I discovered in the Shorter OED, in Roman times circus buildings were ‘generally oblong or oval’. And when we use the word ‘oval’, we seldom stop to think that ovals tend not to be, as etymology would seem to require, strictly egg-shaped. And so on, and so forth, probably ad infinitum. In what sense, then, is (in de Man’s words), the ‘denominative noun “triangle”, in geometry . . . a trope, a hypotyposis which allows for the representation of an abstraction by a substitutive figure’ (Aesthetic, p. 47)? The word or the symbol can of course be used figuratively – for example, to represent the Trinity – but when used to denominate a figure with three angles and three sides, it is not representing an abstraction. On the last page of ‘The Epistemology of Metaphor’, de Man throws in a couple of further, speculative, complications that follow from his account of metaphor: if one accepts, again merely for the sake of argument, that syntagmatic narratives are part of the same system as paradigmatic tropes (though not necessarily complementary), then the possibility arises that temporal articulations, such as narratives or histories, are a correlative of rhetoric and not the reverse. (Aesthetic, p. 50) We are also led, de Man suggests, to the difficult question whether the entire semantic, semiological, and performative field of language can be said to be covered by tropological models, a question which can only be raised after the proliferating and disruptive power of language has been fully recognized. (Ibid.) Or posited. ‘The possibility arises that . . . ’ is one of de Man’s favourite phrases. Another is ‘it is not a priori certain that . . . ’.11 Here I would use the second to counter the first. The possibility arises that the figural nature of language could disrupt all possibility of determinate meaning if, and only if, one chooses to disregard all extra-tropological aspects such as surmised authorial intention, textual coherence, and the possibility of representation and reference. It is, I contend, not at all a priori certain that this would be a very productive idea. Indeed, the very

82

Paradigms of Reading

radicality of such a move seems to be wholly unproductive, as Rodolph Gasché suggests in the Introduction to The Wild Card of Reading: one cannot but wonder whether the kind of subtle, and often infinitesimal, contradictions, disruptions, and dissymmetries that reading discerns in texts to which de Man ascribes an epistemological intention affects epistemology at all. Does the assumption that the dissymmetries in question effectively subvert all effort at understanding not also reveal an inadequate, extremely exaggerated, expectation of what, indeed, epistemology is to achieve? Hence, if it should be the case that de Man’s radicality can be upheld only on condition that what he calls ‘language’ is, in principle, entirely severed from the phenomenal, including language’s own phenomenality, and is hence a domain entirely of its own, immanent to itself without, however, relating to itself, the disruptions that it performs in the world and in worldly language would be without consequence. (Gasché, 1998, p. 10) Yet whether or not de Man’s account of language (or ‘reading’) affects epistemology – and I am inclined to accept Gasché’s suggestion that it does not – it certainly does not seem to affect the pragmatic use of language. Linguistic pragmatics is concerned with the use of language by people who are, in their vast majority, unaware of, or at the very least unconcerned by, the tropological origin of words and concepts, and for whom everyday language is more, in Nietzsche’s phrase, ‘a sum of human relations’ than Heidegger’s ‘forgotten and therefore used-up poem’. Every conceptualisation and hence every noun, including numbers, may have originated as a metaphor, and every dead metaphor may technically be a catachresis, and it may be impossible to define a simple concept without recourse to other equally indefinable concepts, but even if this were generally recognised, it would not present a problem for relevance theory which has no presumption of literalness. On the contrary, S&W argue that all interpretations start loosely or metaphorically and are only tightened to a fully literal interpretation if necessary. Relevance theory shows how words, or linguistically encoded concepts, can be both enriched (or strengthened or narrowed) and loosened (or broadened or weakened) in interpretation. Examples of enrichment or narrowing include the interpretation of ‘I’ve got nothing to wear for the party’ as meaning nothing suitable, and of ‘the cinema is some distance from the restaurant’ as meaning the distance is too far to walk. Examples

Words, Concepts and Tropes 83

of loosening include Austin’s well-known ‘France is hexagonal’, and ‘this steak is raw’ meaning somewhat undercooked or difficult to eat. Speakers routinely use utterances that are non-literal or loose interpretations of the thoughts they wish to communicate when they judge that a non-literal utterance makes the thought more accessible than a literal one would. Hearers decode the lexically encoded concepts (e.g. nothing, some distance, hexagonal, raw), and select those logical or encyclopaedic properties that appear relevant. 12 Loosening the literal meaning of a noun, adjective or verb is as basic and as ordinary a pragmatic process as reference assignment and disambiguation. To use nouns as if they truthfully designated real-world objects is to use nouns tout court, rather than, as Nietzsche put it, to lie collectively or to let ourselves be deceived. But as soon as we use the ordinary and necessary inferential processes of enrichment or loosening, a concept’s dictionary definition or semantically encoded meaning becomes irrelevant, and its metaphorical origin even more so.

6 Rhetoric as an Insurmountable Obstacle

As already mentioned, in ‘Criticism and Crisis’, a few pages into his first book, de Man wrote of ‘the duplicity, the confusion, the untruth that we take for granted in the everyday use of language’: We know that our entire social language is an intricate system of rhetorical devices designed to escape from the direct expression of desires that are, in the fullest sense of the term, unnameable, not because they are ethically shameful (for this would make the problem a very simple one), but because unmediated expression is a philosophical impossibility. (Blindness, p. 9) De Man’s ‘we’ is a little presumptuous. For my part, writing syllabuses for teaching foreign languages, I have always found it necessary to consider ‘our entire social language’ as an intricate system of nameable communicative functions. Unmediated expression is indeed an impossibility, but it is perfectly possible to regard language as an interpretive medium for our thoughts and communicative intentions. De Man goes on to insist on the impossibility, in everyday language, of making the actual expression coincide with what has to be expressed, of making the actual sign coincide with what it signifies. It is the distinctive privilege of language to be able to hide meaning behind a misleading sign, as when we hide rage or hatred behind a smile. But it is the distinctive curse of all language, as soon as any interpersonal relation is involved, that it is forced to act in this way. The simplest of wishes cannot express itself without hiding behind 84

Rhetoric as Insurmountable Obstacle

85

a screen of language that constitutes a world of intricate intersubjective relationships, all of them potentially inauthentic. In the everyday language of communication, there is no a priori privileged position of sign over meaning or of meaning over sign; the act of interpretation will always again have to establish this relation for the particular case at hand. The interpretation of everyday language is a Sisyphean task, a task without end and without progress, for the other is always free to make what he wants differ from what he says he wants. 1 (Blindness, p. 11) However there seem to be two separate arguments here. The first is that signs do not coincide with meanings, so that there is necessarily a difference between what is meant and the way in which language means. The second is that people can speak dishonestly. To the first argument, one can oppose an inferential account of language: although utterances are indeed only interpretive representations of thoughts, the gap between the semantic representations of sentences and the thoughts communicated by utterances can often be filled by the inferential recognition of a communicator’s intentions, by use of contextual information. This is the basis of relevance theory. The second argument – that ‘the other is always free to make what he wants differ from what he says he wants’ – would be true even if signs and meanings did coincide, and is a fact about people rather than an argument about language. It is people who use language dishonestly, not a reified language that is ‘able to hide meaning behind a misleading sign’. Although, in ‘Criticism and Crisis’, de Man describes social language as a ‘system of rhetorical devices’, this is not yet an argument about rhetoric as such. In the 1960s, de Man was more concerned with the Rousseauvian and Heideggerean themes of le néant des choses humaines, ‘the barrenness of our condition’ and ‘the false texture of everyday existence’ (Romanticism, pp. 144, 145), and the ‘inauthentic’ nature of the empirical or ontic ‘self that lives in the world of material and pragmatic substances’ (Romanticism, p. 36), as opposed to ‘the authentic self engaged in the interpretation of its own being-there and capable of comprehending the truth of its own fallen destiny’ (Blindness, p. 44). In these early essays, de Man excludes literature from his strictures: ‘Literature, unlike everyday language, begins on the far side of this knowledge; it is the only form of language free from the fallacy of umediated expression’ (Blindness, p. 17). In his later work, both literature

86

Paradigms of Reading

and ‘ordinary’ and discursive language are described as being wholly undermined by rhetoric. In ‘Semiology and Rhetoric’, the opening chapter of his second book, Allegories of Reading, de Man argues that neither grammar nor intended reference can contain the figural possibilities of language, so it is impossible to decide between literal and figural meanings. Rhetoric ‘radically suspends logic and opens up vertiginous possibilities of referential aberration’ (Allegories, p. 10). In subsequent essays de Man also argued that tropes are wholly cognitive and devoid of any aesthetic function, and that all tropes are ultimately subsumed in an ever-present and uncontrollable process of irony.

‘What’s the difference?’, ‘Son of a bitch!’ and catastrophic confusions In ‘Semiology and Rhetoric’, de Man gives two examples of ‘rhetorical questions’, one from a television script, the other from a canonical poem, which are designed to illustrate the general linguistic truth that any text or utterance can allow at least two wholly coherent but wholly incompatible readings. The former is from an episode of All in the Family (the American version of the British Till Death Us Do Part), featuring characters called Edith and Archie Bunker. Edith asks her husband whether he wants his bowling shoes laced over or laced under. Irritated, he replies, ‘rhetorically’, ‘What’s the difference?’, whereupon she proceeds to explain precisely what is the difference between these two ways of lacing shoes. De Man describes this as an example of a well-known type of utterance with two incompatible meanings, the ‘so-called rhetorical question’, where there is ‘an apparent symbiosis between a grammatical and a rhetorical structure . . . in which the figure is conveyed directly by means of a syntactical device’ (Allegories, p. 9). This is, in fact, something of a redefinition of the term ‘rhetorical question’, which is usually taken to mean one that doesn’t require a reply, as the speaker implies that he or she believes the answer to be self-evident, rather than one whose syntactical structure allows two interpretations.2 But for de Man what is of interest here is that a ‘perfectly clear syntactical paradigm (the question) engenders a sentence that has at least two meanings’. He insists that It is not so that there are simply two meanings, one literal and the other figural, and that we have to decide which one of these meanings is the right one in this particular situation. The confusion can only be cleared up by the intervention of an extra-textual intention,

Rhetoric as Insurmountable Obstacle

87

such as Archie Bunker putting his wife straight; but the very anger he displays is indicative of more than impatience; it reveals his despair when confronted with a structure of linguistic meaning that he cannot control and that holds the discouraging prospect of an infinity of similar future confusions, all of them potentially catastrophic in their consequences. 3 (Allegories, p. 10) Yet there is no way of knowing from reading de Man’s transcription of Archie Bunker’s utterance whether Edith Bunker ‘misunderstood’ what was ‘obviously’ a rhetorical question – obviously, given the pitch, stress, intonation, facial expression, posture and gestures of an exasperated speaker – or whether she was deliberately seeking to annoy her husband. In the first case, the problem would seem to lie in Edith Bunker rather than language itself (she is too stupid to pick up obvious prosodic and contextual clues); in the second, she is deliberately exploiting the ironic possibilities of language, taking a rhetorical question literally and choosing to misinterpret her husband in order to rebuke him for his gruff response to her solicitous question about how he wanted his bowling shoes laced. The sentence ‘What’s the difference?’ can indeed have at least two meanings, but it would normally only be impossible to choose the meaning of a particular utterance if there were no phonetic or contextual clues. In the case of de Man’s example, it seems probable that Edith is simply misunderstanding: as a character she is essentially kindly but none too bright. But of course this is a fictional example, written by a scriptwriter perfectly in control of the potential ambiguity of this invented utterance, who expects the audience to laugh at this comical misunderstanding. Pace de Man, it seems to be precisely the case that there are two meanings, and we have to decide which one of these meanings is the right one in this particular situation. The confusion can indeed only be cleared up by the intervention of an extra-textual intention, which suggests that intention is an integral part of language use, both productive and receptive. There are many everyday utterances in the form of a question, ‘a perfectly clear syntactical paradigm [which] engenders a sentence that has at least two meanings’. But although requests beginning with Can or Could you . . . (pass the caviar, shut the door, turn the light on) could always, potentially, be interpreted as questions about ability (José, can you see . . . ?), this rarely happens. It is perfectly possible for a sentence to have two literal meanings, or a literal and a figural meaning, without there necessarily being an aporia between them. As Stanley

88

Paradigms of Reading

Cavell (1984, p. 42) has written, de Man’s argument is parallel to the following one: Someone says that the difference between knives and forks is that you cut with a knife and spear with a fork; a second objects that you can also cut with a fork and spear with a knife; whereupon a third concludes that there is an aporia between knives and forks, that there is no stable distinction we can draw between them.4 Cavell questions exactly why we should say that there is something aberrant about a sentence that is grammatically a question but rhetorically (or inferentially) something else, or that the rhetoric deconstructs the grammar, or is in tension with it: ‘You might as well say it is perverse or aberrant of the normally functioning human hand that it can grasp, and make a fist, and play arpeggios, and shade the eyes, and be held up to bless or swear. Does one of these possibilities repress or otherwise oppress the others?’ (1984, pp. 43–4). In other words, there is no reason to assume that there need be an inevitable tension between grammar and rhetoric. As Cavell says, ‘This seems to be the moral of ordinary language philosophy as well, and of the practice of art. Put it this way: Grammar cannot, or ought not, of itself dictate what you mean, what it is up to you to say’ (p. 45). This is also the position of linguistic pragmatics, and in particular relevance theory, which describes utterances as fragmentary interpretive representations of thoughts whose interpretation requires multiple pragmatic inferences. Another critic who does not see grammar dictating meaning is Richard Poirier, who suggests that nearly everybody knows that sign and meaning never coincide in everyday language. He recalls that My platoon in the army consisted of fifteen or so teenagers; most had never finished high school; but when one would say to another ‘you son of a bitch’ he would do so on the confident assumption that every one of his associates knew all about ‘the fallacy of unmediated expression’. . . . Obviously ‘son of a bitch’ is a phrase used all the time to mean anything except what it says, anything from ‘you’re terrific’ to ‘how about that!’ Persons who use the phrase are not identifying themselves with the words, but wholly with the performance, with the tonal pitch that can be given to the words. (Poirier, 1992, pp. 141–2)

Rhetoric as Insurmountable Obstacle

89

Of course, we also know which tone to use in order to say, literally, ‘Your mother is a bitch’, if we feel like picking a fight. In short, de Man’s example is unconvincing, and his reference to an infinite number of catastrophic confusions seems to be, as Brian Vickers puts it, both ‘melodramatic and paranoiac’ (1988, p. 458).5 Of course, this is not to say that linguistic confusions cannot have catastrophic consequences, it is just that de Man’s example hardly proves the point. A better example from Britain would be the tragic case of Derek Bentley, who was executed for murder in 1953 (and posthumously pardoned in 1998). Trapped with an accomplice on a warehouse roof during an aborted burglary, and confronted by a policeman, Bentley allegedly shouted ‘Let him have it’. At his trial, the defence claimed that Bentley (who had a mental age of 11) was telling his juvenile accomplice to surrender and hand over his gun. Instead the boy shot the policeman, for which Bentley was hanged (sparing him ‘an infinity of similar future confusions’) as it was claimed he had given the instruction to shoot. 6 After his fictional example of everyday language, from ‘the subliterature of the mass-media’ (Allegories, p. 9) – presumably selected, if one may speculate as to authorial intentions, because it allows him a splendid pun about Nietzsche and Derrida being arché debunkers7 – de Man moves on to literature proper, stating ‘I would not hesitate to equate the rhetorical, figural potentiality of language with literature itself’ (p. 10). He quotes the concluding quatrain of Yeats’s ‘Among School Children’: O chestnut-tree, great-rooted blossomer, Are you the leaf, the blossom or the bole? O body swayed to music, O brightening glance, How can we know the dancer from the dance? These do indeed appear to be rhetorical questions, requiring no answer: Yeats seems to be saying that just as the tree cannot be separated from its parts, we cannot distinguish between the dancer and the dance itself. (Indeed, an early version of the poem ended by answering the question: ‘It seems the dancer and the dance are one’.)8 As de Man puts it, ‘the line is usually interpreted as stating, with the increased emphasis of a rhetorical device, the potential unity between form and experience, between creator and creation’ (Allegories, p. 11). But de Man argues that it is equally possible to read Yeats’s line literally rather than figuratively, as asking ‘with some urgency’ how can we distinguish between sign and

90

Paradigms of Reading

referent? Hence, ‘two entirely coherent but entirely incompatible readings can be made to hinge on one line, whose grammatical structure is devoid of ambiguity’ (Allegories, p. 11). Yet de Man denies that we can say ‘that the poem simply has two meanings that exist side by side. The two readings have to engage each other in direct confrontation, for the one reading is precisely the error denounced by the other and has to be undone by it. Nor can we in any way make a valid decision as to which of the readings can be given priority over the other; none can exist in the other’s absence. There can be no dance without a dancer, no sign without a referent’ (p. 12). However, the history of Yeats criticism and interpretation does indeed reveal that we cannot say ‘that the poem simply has two meanings that exist side by side’: it has many more than two meanings. For example, both Cavell and John Hollander read know from as infer rather than tell apart. Cavell (1984, p. 46) says that ‘knowing from, in Yeats’s line, goes altogether beyond telling. The line asks how we know the dancer from, meaning by means of, the dance; how it is that the dance can reveal the dancer. . . . But the line equally asks how we can know from, meaning know apart from, the dance, meaning know the dancer away from the dance’. Hollander (1988, p. 37), similarly, writes that know from ‘not only can mean “distinguish between” as the usual and obvious construing of the question suggests; it can also mean “infer from”, “know by means of evidence of” ’. Yeats’s line allows several semantic representations, each giving rise to a different explicature. It cannot be reduced to an aporia between a single literal and a single figural meaning.

Rhetoric and aesthetics De Man acknowledges that ‘Among School Children’ ‘is not explicitly “about” rhetorical questions but about images or metaphors’ (Allegories, p. 12). Conventionally, metaphors are considered to consist of rather more than ‘meanings’: they equally have (and indeed are often motivated by) an aesthetic aspect. However, in various essays on literary theory and theorists posthumously collected in The Resistance to Theory, de Man argued that rhetoric has no aesthetic component. He sets up parallel distinctions between hermeneutics and poetics, and between aesthetics and rhetoric. He raises ‘the fundamental question of the compatibility between the descriptive discourse of poetics and the normative discourse of hermeneutics’ (Resistance, pp. 113–14), and insists that rhetorical figures and tropes belong to the realm of the cognitive rather than that of the aesthetic.

Rhetoric as Insurmountable Obstacle

91

Hermeneutics is concerned with interpretation, and hence the meaning of texts, and so, de Man writes, ‘it postulates a transcendental function of understanding, no matter how complex, deferred, or tenuous it might be, and will, in however mediated a way, have to raise questions about the extralinguistic truth value of literary texts’ (Resistance, pp. 55–6). Poetics, on the contrary, is a meta-linguistic, descriptive or prescriptive, and would-be scientific discipline, concerned with the taxonomy and the interaction of poetic structures. The orthodox starting point in the attempt to discover the meaning of a text is to read it (except for those undergraduates who persist in going to the library to find the chapter by an obliging critic that will save them the effort), but this is not reading in de Man’s sense. In hermeneutics, he writes, ‘reading necessarily intervenes, but like computation in algebraic proof, it is a means toward an end, a means that should finally become transparent and superfluous; the ultimate aim of a hermeneutically successful reading is to do away with reading altogether’ (Resistance, p. 56). De Manian poetics, on the contrary, involves slow, careful, rigorous reading, leading to ‘the formal analysis of linguistic entities as such, independently of signification’ (p. 56). De Man recalls a particular course at Harvard in which he was encouraged to start out from the bafflement that . . . singular turns of tone, phrase, and figure were bound to produce in readers attentive enough to notice them and honest enough not to hide their nonunderstanding behind the screen of received ideas that often passes, in literary instruction, for humanistic knowledge. 9 (Resistance, p. 23) The literary or linguistic reading that de Man seems to be calling for is purely formal, which is to say rhetorical rather than aesthetic. It ‘choose[s] to hear’ what a text ‘actually says’ rather than what it wishes to say; it considers ‘the modalities of production and reception of meaning and value prior to their establishment’ (Resistance, p. 7); and it ‘designates reference prior to designating the referent’ (p. 8). Since the impossibility of reading and of analysing singular turns of tone, phrase and figure independently of signification before making a decision as to their signification hardly needs to be pointed out, 10 it seems necessary, at the very least, to read de Man’s ‘prior to’ in the light of his earlier statement that temporal concepts acquire ‘a particularly rich complexity’ when made to refer to linguistic events (Blindness, p. 144). At the end of ‘Shelley Disfigured’ (in Rhetoric) and

92

Paradigms of Reading

elsewhere, de Man recognizes that reading invariably proceeds by way of understanding, aesthetification, historicisation and monumentalisation, while in ‘Reading and History’, he points out with reference to the metaphor ‘Achilles is a lion’, that ‘one has to have “read” the text in terms of poetics to arrive at a hermeneutic conclusion. One has to have become aware that it is a figure. . . . But one also has to read it hermeneutically to “understand” it as poetics’ (Resistance, p. 56). Without either experience or an intuition of metaphors, the reader would make the hermeneutically disastrous deduction that Achilles was a large, tawny-coloured, predatory, carnivorous, feline quadruped. But without an awareness of the difference between Greek heroes and large, tawny-coloured, predtory, carnivorous, feline quadrupeds, the reader would not be able to recognise the metaphor. Consequently, de Man concedes, ‘hermeneutics and poetics, different and distinct as they are, have a way of becoming entangled’ (p. 56). But so do rhetoric and aesthetics, despite de Man’s claim that ‘attention to the philological or rhetorical devices of language is not the same as aesthetic appreciation, although the latter can be a way of access to the former’ (p. 24). The corollary of close reading is that ‘literature, instead of being taught only as a historical and humanistic subject, should be taught as a rhetoric and a poetics prior to being taught as a hermeneutics and a history’ (Resistance, pp. 25–6). De Man stresses that what most people call close reading but what he calls ‘mere reading’ is ‘deeply subversive to those who think of the teaching of literature as a substitute for the teaching of theology, ethics, psychology, or intellectual history’ (p. 24). Most readers, de Man claims, are wilfully blind to the materiality of language, and the subversive potential of tropes. The close reader, on the other hand, ‘cannot fail to respond to structures of language which it is the more or less hidden aim of literary teaching to keep hidden’ (p. 24). De Man insists that if one reads attentively, without ignoring problematic figural structures and textual aberrations, ‘the attribution of a reliable, or even exemplary, cognitive and, by extension, ethical function to literature . . . becomes much more difficult’ (p. 25). The correct reading of literature casts grave doubt on the Kantian notion that aesthetics can reconcile pure and practical reason, or link a priori judgements with sensuous perceptions: ‘Literary theory raises the unavoidable question whether aesthetic values can be compatible with the linguistic structures that make up the entities from which these values are derived’ (p. 25). Yet de Man’s literary theory, and the close reading (or ‘the return to philology’) that he claims it enables, rest on a moment – that

Rhetoric as Insurmountable Obstacle

93

de Man knows to be impossible – in which the reader studies linguistic structures before they begin to signify. Furthermore, the opposition of aesthetics and rhetoric as mutually exclusive terms is highly tendentious, if not wholly untenable. As already outlined in the previous chapter, the standard definition of rhetoric is that it is part of persuasive communication. Theorists of rhetoric have always argued that persuasion involves, among other things, putting an audience in a suitable frame of mind, by way of pathos. The duties of an orator are traditionally listed as movere, docere and delectare – to arouse an audience’s emotions, to instruct, and to do good or to give pleasure. (Plato, in his underhand attack on rhetoric in the Gorgias, rejects the idea that rhetoric can move and instruct, and has Socrates argue that it can only please, an idea echoed by Kant in the Critique of Judgement.) Classical rhetoric also enumerates five stages of composition, one of which is elocutio, the use of language appropriate to a subject matter, including figures and other stylistic devices. Rhetorical devices have often been subdivided into tropes, which involve a transference of meaning on a conceptual level, and figures which involve the reordering of words into a structure that goes beyond minimal communicative needs. The necessary overlap between rhetoric and poetics (particularly formalism) appears to be almost self-evident. As does that between rhetoric and aesthetics, defined as the perception of (symbolic) beauty. Indeed, as Wilkinson and Willoughby point out in their bilingual edition of Schiller’s Letters on Aesthetic Education, ‘Many have thought that Baumgarten [who first used the word ästhetisch in its modern sense in 1750] was simply giving a new name to an old thing, viz. Rhetoric and Poetics’ (Schiller, 1967, p. 304). De Man’s notion of rhetoric, of course, rather differs from the traditional definition, since he describes persuasion as a derived sense of rhetoric, and argues that all words, names and concepts are metaphorical at root so that language is a system of unstable metaphors that deconstruct themselves even as they are asserted. Far from seeing ‘literariness’ as the outcome of authors consciously using tropes and figures in order to persuade and give pleasure, de Man defines it in terms of the ‘autonomous potential of language’ and the ‘foregrounding of material, phenomenal aspects of the signifier’ (Resistance, p. 10). But foregrounding requires a subject as well as an object, and, as I will argue at length in Chapter 9, if an author or reader chooses to foreground certain elements of language, or if a reader succeeds in recognising effects manifestly intended by an author (while at the same time inevitably giving an interpretation to the words), this is no longer

94

Paradigms of Reading

a matter of the autonomous potential of language. If what is foregrounded is tropes and figures, neither their persuasive aspect nor their aesthetic effect can be ignored. De Man states that the use of words for phonetic rather than semantic reasons gives language ‘considerable freedom from referential restraint, but it makes it epistemologically highly suspect and volatile, since its use can no longer be said to be determined by considerations of truth and falsehood, good and evil, beauty and ugliness, pleasure and pain’ (Resistance, p. 10). This leads him to argue that foregrounding the materiality of the signifier ‘creates a strong illusion of aesthetic seduction at the very moment when the actual aesthetic function has been, at the very least, suspended’ (p. 10). 11 But this argument only holds if one accepts de Man’s list of opposing pairs. If all language is inherently tropological, and the use of tropes makes language ‘epistemologically highly suspect and volatile’, then there does indeed arise a problem concerning truth and falsehood. But only if the claim that ‘beauty is truth, truth beauty’ is accepted as a definition of the aesthetic do the pairs truth and falsehood, and beauty and ugliness, become equivalent. If truth and beauty are not equivalent – and Keats’s aphorism is clearly not all ye need to know about aesthetics – nothing prevents tropes or formalistic techniques having an aesthetic function. Not only paranomasis (the use of words for phonetic reasons), but all the figures involving repetition and parallelism (anadiplosis, anaphora, antanaclasis, antimetabole, homoioptoton, homoioteleuton, isocolon, parison, ploche, polyptoton, and a large number of the other figures listed in any handbook of rhetoric) are often employed (usually by writers or speakers unaware that they are using a particular figure, long since named and classified) for aesthetic purposes – that is to please rather than to make truthful or responsible pronouncements on the nature of the world. Poets, politicians, advertising copywriters and advocates all attempt to use felicitous phrases for aesthetic effect, without too much concern for questions of truth. At the beginning of ‘Anthropomorphism and Trope in the Lyric’, de Man describes Keats’s claim about truth and beauty as merely one instance of the ‘gesture that links epistemology with rhetoric’ which ‘recurs in many philosophical and poetic texts of the nineteenth century’, including Nietzsche’s ‘On Truth and Lie’ (Rhetoric, p. 239). Here he only quotes the beginning of Nietzsche’s description of truth as ‘a moving army of metaphors, metonymies and anthropomorphisms’, although in Allegories of Reading he quotes the complete sentence, which ends, in de Man’s translation, ‘ . . . that are being poetically and

Rhetoric as Insurmountable Obstacle

95

rhetorically sublimated, transposed, and beautified’ (p. 110). Which is to say that Nietzsche links epistemology, rhetoric and aesthetics, while Keats’s aphorism is more concerned with aesthetics than rhetoric. De Man writes that ‘the continuity of aesthetic with rational judgment . . . is the main tenet and the major crux of all critical philosophies and “Romantic” literatures’ (Rhetoric, p. 239), but his dislike of this ‘philosophical phantasm’ (p. 258) leads him to a far too hasty claim that rhetoric contains no aesthetic component.

Tropes and persuasion The examples of ‘rhetorical questions’ in ‘Semiology and Rhetoric’ are supposed to show how any text or utterance can allow at least two wholly coherent but wholly incompatible readings, so that textuality displaces authorial intentionality and necessitates figural criticism. In subsequent essays in Allegories of Reading, de Man takes his account of the consequences of rhetoric much further. In ‘Rhetoric of Tropes (Nietzsche)’, he claims, extremely contentiously, that Nietzsche switched the study of rhetoric from persuasion (or as Plato would have it, duplicity) to tropes and figures, and demonstrated that the paradigmatic structure of language is figural or rhetorical, rather than grammatical, referential or performative. De Man does not disregard persuasion, but attempts to demonstrate ‘a process of reading in which rhetoric is a disruptive intertwining of trope and persuasion or – which is not quite the same thing – of cognitive and performative language’ (Allegories, p. ix). He (re)defines rhetoric as ‘the study of tropes and figures’ per se, describing eloquence or persuasion as a ‘derived’ sense (p. 6). De Man bases these claims on a reading of one short section of the notes Nietzsche made prior to teaching a course on rhetoric in Basel in 1872–3. Yet as Nietzsche’s first German editor pointed out, and his more recent French and English translators echo, he follows a well-beaten track (see Nietzsche, 1922, 1970, 1989). The notes were largely copied from recent German books on rhetoric (by Richard Volkmann, Gustav Gerber and Friedrich Blass), and Nietzsche’s course covered, among other things, the concept of rhetoric; its division into inventio, dispositio, elocutio, memoria and pronuntiatio or actio; the rhetorical nature of all language; canons of clarity, purity and decorum; neologism, barbarisms and solecisms; tropes and figures; the three styles (plain, grand and flowery); and the three branches of judicial, deliberative and epideictic rhetoric. Nietzsche recapitulates the history of rhetoric from the Sicilians

96

Paradigms of Reading

Empedocles, Corax and Tisias (conventionally cited as the founders of the art around 460 BC), through orators such as Protagoras, Pericles, Gorgias, Isocrates and Demosthenes, and summarizes and quotes from all the standard Greek and Roman authors, including Plato, Aristotle, the Stoics, Quintilian, the Ad Herennium and Cicero, probably via his recent German sources. In other words he very clearly outlines the traditional concept of rhetoric as an orator’s construction of an artistic discourse, and the use of eloquence and persuasion to move the feelings. In the third section of his notes, Nietzsche does indeed equate rhetoric and language, as he does in the contemporaneously written ‘On Truth and Lying’. He states that ‘with respect to their meanings, all words are tropes in themselves, and from the beginning’ (Nietzsche, 1989, p. 21), and that ‘the tropes are not just occasionally added to words but constitute their most proper nature. . . . There is just as little distinction between actual words and tropes as there is between straightforward speech and rhetorical figures. What is usually called language is actually all figuration’ (p. 25). Furthermore, it is not difficult to prove that what is called ‘rhetorical’, as a means of conscious art, had been active as a means of unconscious art in language and its development [werden], indeed, that the rhetorical is a further development, guided by the clear light of understanding, of the artistic means which are already found in language. There is obviously no unrhetorical ‘naturalness’ of language to which one could appeal; language itself is the result of purely rhetorical arts. The power to discover and to make operative that which works and impresses, with respect to each thing, a power which Aristotle calls rhetoric, is, at the same time, the essence of language. . . . Language does not desire to instruct, but to convey to others a subjective impulse and its acceptance. (Nietzsche, 1989, p. 21) De Man quotes much of this, in his own fairly free translation, on pp. 105–6 of Allegories, in a paragraph with lots of ellipses that jumps forwards and backwards through Nietzsche’s text. Thus as well as quoting the argument that all words originated as tropes and remain essentially trope-like, so that all discourse is figurative, de Man quotes Nietzsche’s statement that language only desires ‘to convey to others a subjective impulse and its acceptance’ (in de Man’s version, ‘Language is rhetoric, for it only intends to convey a doxa [opinion], not an

Rhetoric as Insurmountable Obstacle

97

episteme [truth]’). De Man seems not to notice that in quoting Nietzsche’s sentence about conveying impulses he is contradicting his argument that ‘Nietzsche moves the study of rhetoric away from techniques of eloquence and persuasion by making these dependent on a previous theory of figures of speech or tropes’ (Allegories, p. 105), or that the course ‘starts out from a pragmatic distinction between rhetoric as a system of tropes and rhetoric as having to do with the skills of persuasion’ (p. 130) (even though it is encouraging to find de Man inventing something as dreary and banal as a pragmatic distinction). It is equally untrue that he ‘contemptuously dismisses the popular meaning of rhetoric as eloquence and concentrates instead on the complex and philosophically challenging epistemology of the tropes’ (p. 130). If all words are tropes, then by definition eloquence and persuasion depend on tropes, but Nietzsche nowhere seeks to use this insight to displace or reject the practice of rhetoric as persuasion. De Man describes this section of Nietzsche’s notes as a ‘straightforward affirmation that the paradigmatic structure of language is rhetorical rather than representational or expressive of a referential, proper meaning’ (Allegories, p. 106), but it does not follow that if language is rhetoric it cannot be used to represent or express or refer. Rhetoric and persuasion and opinion have repeatedly been opposed to epistemological truth, from the Sophists to Stanley Fish. People use rhetorical language to express their opinions, and to represent and refer to the world as they see it, in order to persuade others of their opinions. The ordinariness of tropes such as metaphor and irony is taken for granted by present day pragmatic approaches to language, including relevance theory. Recent work in cognitive linguistics has shown that human cognition is fundamentally structured by various figurative processes, and that metaphor and irony constitute basic schemes by which people conceptualise their experience and the external world. There is experimental evidence that processing utterances used metaphorically or ironically hardly takes any longer than processing the same utterances used literally (Gibbs, 1994, pp. 230–2). 12 Rhetoricians have traditionally distinguished between tropes, which change the meaning of a word (as, for example, in metaphor, metonymy, synecdoche, irony and catachresis) and supposedly merely decorative figures. But the distinction is dubious: the artificial rearrangement of word-order in figures such as hyperbaton, antimetabole and anastrophe, and figures involving repetition, such as ploce, homioptoton and anaphora, also tend to affect meaning. Thus we could, perhaps, collapse the notions of figures and tropes, but de Man’s conception of ‘trope’ or

98

Paradigms of Reading

‘tropological system’ goes even further: not only does it subsume other rhetorical categories, it absorbs most linguistic categories as well. De Man writes of ‘such grammatical tropes as negation and interrogation’ (Aesthetic, p. 49), and of ‘tropes such as grammar and syntax’ (Rhetoric, p. 115). If grammar and syntax themselves are tropes, the figures involving word-order mentioned above, and those that more radically affect grammatical structure (such as anapodoton, asyndeton, polysyndeton, hypallage and hysteron proteron) can no longer be considered as turns or deviations from a norm. The many hundreds of tropes and figures identified and described by the authors of so many classical, medieval and renaissance rhetoric manuals all become, in de Man’s theory, mere instances of not just a ‘master trope’, but a trope to end all tropes: grammar itself.13 The word trope, of course, comes from the Greek word for ‘turn’, but turning all words into tropes has the disadvantage of negating the concept itself, as there remains no literal language to turn away from. It is, of course, easy to argue that grammars are interpretive constructs, rather than objective descriptions of language. As Stanley Fish puts it (1981, p. 167), ‘Verbs, nouns, cleft sentences, transformations, deep and surface structures, semes, rhemes, tagmemes – now you see them, now you don’t, depending on the descriptive apparatus you employ.’ And anyone who has made the effort to keep up with Chomsky, through the Revised Extended Standard Theory to the current extensions of Government and Binding theory, and with counterChomskyan models such as case, daughter dependency, functional, generalized phrase-structure, relational, stratificational, systemic and tagmemic grammars, can attest that there are more ways of describing the grammar of a sentence than Wallace Stevens had ways of looking at a blackbird. But the fact that all attempts to describe the grammatical and syntactic functioning of language may be seen as interpretive constructs does not make grammar a trope.

Irony In the third section of his lecture notes, immediately after defining language as rhetoric, Nietzsche elaborates on synecdoche, metaphor and metonymy. In section seven he discusses many more tropes, including antonomasia, onomatopoeia, catachresis, metalepsis, allegory, irony, periphrasis, hyperbaton, anastrophe and hyperbole.14 De Man, however, although he goes further than Nietzsche in that he describes grammar, as well as all individual words, as a trope, only discusses

Rhetoric as Insurmountable Obstacle

99

a very limited range of tropes, and ultimately describes the whole of language in terms of irony. In fact, as Brian Vickers laments in In Defence of Rhetoric (1988), de Man marks the end-point in the long impoverishment of the notion of rhetoric from classical Athens and Rome to contemporary American literary graduate schools. Already in the Greek period, Vickers argues, concentration on elocutio (as both tropes and figures, and qualities of style), at the expense of the other stages of composition, led to a retreat from a persuasive process involving artist, artefact and audience, and its replacement by what would today be called formalist poetics, with little reference to the speaker–hearer relationship. In the medieval period, rhetoric was partially displaced by grammar. The grammarians analysed figures of speech in poetry as mere forms, divorced from any attempt to persuade or to arouse passions. Elocutio became more ornament than the perfect expression of ideas. Figures were listed as one might today list irregular verbs, without reference to meaning or feeling. Vickers insists, on the contrary, that tropes are modes of the expression of feeling, which have great emotive (rather than didactic or cognitive) power. (There are no fixed one-to-one relations between forms, meanings and feelings, but rather a threeway relationship, which Vickers illustrates by means of a triangular diagram with double-headed arrows, reminiscent of Peirce’s semiotic triad of object, sign and interpretant, Frege’s sign, sense and reference, and Ogden and Richards’s triad of word, thought and referent. Figures, Vickers insists, are polysemous and polyfunctional.) Yet in de Man, the atrophy of rhetoric is complete: it has a secondary (written) role, it is reduced to elocutio alone, and then further to half a dozen or so tropes, detached from any expressive and persuasive function. 15 In various places, de Man further reduced his account of rhetorical language to a discussion of irony, which he defined in a post-lecture question and answer session published with the first version of ‘Rhetoric of Tropes (Nietzsche)’ (de Man, 1974, hereafter Symposium in text); in ‘The Concept of Irony’, a 1977 lecture transcribed in Aesthetic Ideology; and in an interview with Robert Moynihan, recorded in 1980 (Moynihan, 1984). In ‘The Concept of Irony’ – ‘an ironic title, because irony is not a concept’ (Aesthetic, p. 163), and therefore undefinable – de Man repudiates his earlier treatment of irony which described it as a moment in the dialectic of the self, a deliberate selfreflexive duplication of the self, in which one part observes the rest from a certain distance. 16

100

Paradigms of Reading

In the Moynihan interview, de Man says: There’s irony when language starts to say things you didn’t think it was saying, when words acquire meanings way beyond the one you think you are controlling and start saying things that go against your own quest for meaning or admitted intention. So irony is so fundamental, that, for me, it’s no longer a trope. Irony is generally called a trope of tropes, but actually irony is a disruption of a continued field of tropological meaning. (Moynihan, 1984, p. 584) Irony is thus something that language does, rather than a conscious attitude on the part of a language-user: ‘There is an inherent conflict or tension . . . between irony on the one hand and self-consciousness on the other. Irony comes into being precisely when self-consciousness loses its control over itself . . . a break, an interruption, a disruption’ (Moynihan, 1984, p. 580). Indeed, ‘Irony goes beyond scepticism in that it puts the notion of the self in question. Scepticism is an attitude of consciousness while irony is a purely linguistic trope’ (Symposium, p. 45). In de Man’s account, therefore, irony is not something to be taken lightly: How can we take our utterances seriously and give them authority if we cannot justify the possibility of escaping from the ironic mode? It is not the self which wants to remain in this suspended state. Irony is an unbearable way of being because, as Friedrich Schlegel well knew, it is both a condition of total freedom and of total selfdestruction. . . . To expect that we could go beyond it is more easily said than done. (Symposium, p. 47) De Man allows that we can, at times, wilfully use irony: ‘Irony also very clearly has a performative function. Irony consoles and it promises and it excuses. It allows us to perform all kinds of performative linguistic functions’ (Aesthetic, p. 165). But alongside such stable and definite irony, there is another unstable kind. De Man quotes Wayne Booth, who writes that in reading ironies we discover why Kierkegaard finally defined irony as ‘absolute infinite negativity’. As Booth puts it (1974, p. 59 n. 14, quoted in de Man, Aesthetic, p. 166), ‘Irony in itself opens up as soon as its possibility enters our heads, and there is no inherent reason for discontinuing the process at any point short of infinity

Rhetoric as Insurmountable Obstacle 101

. . . pursued to the end, an ironic temper can dissolve everything, in an infinite chain of solvents. It is not irony, but the desire to understand irony that brings such a chain to a stop.’ De Man goes on to endorse Friedrich Schlegel’s suggestion in ‘Über die Univerständlichkeit’ [On the impossibility of understanding, or On incomprehensibility] that (in de Man’s words) ‘what is at stake in irony is the possibility of understanding, the possibility of reading, the readability of texts, the possibility of deciding on a meaning or on a multiple set of meanings or on a controlled polysemy of meanings’ (Aesthetic, p. 167). Then in a splendid rhetorical move he offers the following hyperbole: ‘It would hardly be hyperbolic to say (and I could defend the affirmation) that the whole discipline of Germanistik has developed for the single reason of dodging Friedrich Schlegel’ (p. 168). 17 Thus far from being a controllable aesthetic practice or artistic device, irony is a principle of disruption that makes it impossible for anyone to master their own texts or say one thing unequivocally, or for a critic to give a single definitive reading. It prevents the telling of consistent or coherent stories. As de Man puts it in the interview with Moynihan, texts are always difficult to understand, and even texts that appear to be wilfully ironic can be read ironically. Thus ‘there are no such things as valid texts. . . . Neither is there a valid reading. There is no final authority’, although of course ‘some invalid texts are more validly invalid than others!’ (Moynihan, 1984, p. 585). Irony gives rise to aporias – ‘impasse[s] which cannot be resolved, domesticated, or assimilated by a trope’ (p. 589). Consequently, de Man says, ‘I don’t so much speak of misreading, because misreading supposes a right reading. . . . I speak of unreadability, which means that the text produces not misreadings, but readings that are incompatible’ (p. 591). To use a rhetorical term, as Schlegel does, irony is parabasis – ‘the interruption of a discourse by a shift in the rhetorical register’ (Aesthetic, p. 178), but permanently, at all points.

Resistance to language De Man’s account of the inescapably ironic nature of language – such that words invariably acquire meanings that disrupt their users’ intentions, making it impossible to say one thing unequivocally or to interpret other people or texts consistently or coherently – clearly problematizes most traditional accounts of rhetoric. In ‘The Resistance to Theory’, de Man calls into question ‘the most familiar and general of

102

Paradigms of Reading

all linguistic models, the classical trivium, which considers the sciences of language as consisting of grammar, rhetoric and logic (or dialectics)’ (Resistance, p. 13).18 He insists that contrary to received opinion, the rhetorical function of language disrupts the grammatical and logical functions, and ‘undoes the claims of the trivium (and by extension, of language) to be an epistemologically stable construct’ (p. 17). Tropes and figures of speech straddle the borderline between grammar and rhetoric, thereby creating a latent tension. Their existence ensures that ‘literature is not a transparent medium in which it can be taken for granted that the distinction between the message and the means of communication is clearly established’, and that ‘the grammatical decoding of a text leaves a residue of indetermination that has to be, but cannot be, resolved by grammatical means, however extensively conceived’ (p. 15). De Man suggests that ‘no grammatical decoding, however refined, could claim to reach the determining figural dimensions of a text. There are elements in all texts that are by no means ungrammatical, but whose semantic function is not grammatically definable, neither in themselves nor in context’ (pp. 15–16). Consequently, ‘since grammar as well as figuration is an integral part of reading, it follows that reading will be a negative pocess in which the grammatical cognition is undone, at all times, by its rhetorical displacement’ (p. 17). By way of example, de Man gives the title of Keats’s unfinished epic, The Fall of Hyperion: are we to read the genitive as meaning ‘Hyperion’s Fall’ or as ‘Hyperion Falling?’ (For de Man, ‘the undecidability involves the figural or literal status of the proper name Hyperion as well as of the verb falling, and is thus a matter of figuration and not of grammar’ – Resistance, p. 16.) I will not paraphrase de Man’s two interpretations here, but merely endorse his conclusion that ‘both readings are grammatically correct, but it is impossible to decide from the context (the ensuing narrative) which version is the right one. The narrative context suits neither and both at the same time’ (p. 16). This is a far more compelling example than Archie Bunker’s so-called ‘rhetorical question’ with which to demonstrate de Man’s claims about the absolute impossibility of deciding between conflicting literal and figural referents, the reader’s suspension of judgement, and the text’s lack of any cognitive function. Yet de Man is unable to accept this undecidability, and posits ‘the ineluctable necessity to come to a decision’, while reiterating that ‘no grammatical or logical analysis can help us out’ (p. 16). He therefore calls for ‘a theoretical project of rhetorical analysis that will reveal the inadequacy of grammatical models of non-reading’ (p. 17). However

Rhetoric as Insurmountable Obstacle 103

this rhetorical analysis, which de Man seems to equate with literary theory tout court, turns out to be a form of semiology or Saussurean linguistics: Literary theory can be said to come into being when the approach to literary texts is no longer based on non-linguistic, that is to say historical and aesthetic, considerations or, to put it somewhat less crudely, when the object of discussion is no longer the meaning or the value prior to their establishment – the implication being that this establishment is problematic enough to require an autonomous discipline of critical investigation to consider its possibility and its status. . . . The advent of theory . . . occurs with the introduction of linguistic terminology in the metalanguage about literature. By linguistic terminology is meant a terminology that designates reference prior to designating the referent and takes into account, in the consideration of the world, the referential function of language or, to be somewhat more specific, that considers reference as a function of language and not necessarily as an intuition. . . . Contemporary literary theory comes into its own in such events as the application of Saussurian linguistics to literary texts. (Resistance, pp. 7–8) De Man states that the interest of philologists such as Roman Jakobson or literary critics such as Roland Barthes in semiology reveals the natural attraction of literature to a theory of linguistic signs. By considering language as a system of signs and of signification rather than an established pattern of meanings, one displaces or even suspends the traditional barriers between literary and presumably non-literary uses of language and liberates the corpus from the secular weight of textual canonization. . . . The responsiveness of literary texts to semiotic analysis is visible in that, whereas other approaches were unable to reach beyond observations that could be paraphrased or translated in terms of common knowledge, these analyses revealed patterns that could only be described in terms of their own, specifically linguistic aspects. The linguistics of semiology and of literature apparently have something in common that only their shared perspective can detect and that pertains distinctively to them. The definition of this something, often referred to as literariness, has become the object of literary theory. (Resistance, pp. 8–9)

104

Paradigms of Reading

‘Literariness’, more specifically, is ‘the use of language that foregrounds the rhetorical over the grammatical and the logical function’ (Resistance, p. 14). Although ‘the rhetorical or tropological dimension of language . . . can be revealed in any verbal event when it is read textually’, it is ‘perhaps more explicitly in the foreground in literature (broadly conceived)’ (p. 17). Another way of putting this is that ‘the specificity of literary language resides in the possibility of misreading and misinterpretation’ (Blindness, p. 280). As Rodolphe Gasché puts it (1998, p. 121), reading textually, or as de Man also calls it, ‘mere reading’ is ‘a silent reading in that it silences all intuitive, perceptional, sensual approaches to the written text. As if echoing Saint Augustine’s denunciation of the seductions of the senses, mere reading practices radical asceticism. It is deliberately mute, and deliberately blind.’ Yet this account of reading also raises a number of troubling questions, all of them succinctly posed by Gasché: If a reading in no possible way illuminates the text, what precisely is its relation to the text? Does mere reading’s antiaesthetical stand not preempt it from relating to the text in the first place? And further, why read to begin with? And, granted that reading voids all modes of cognition, how do I know that I am performing a reading and not some other activity? . . . Finally, accepting that the pertinence of mere reading cannot be logically justified, why is it more desirable to read than to resist mere reading? . . . And what is a mere reading doing to and for the text when reading? Does it have any significance? In general, for the text, for the reader? (Gasché, 1998, p. 183) Ultimately such analysis of the autonomous potential of language has no relation whatsoever to the meaning that this language appears to transport. Indeed, a rhetorical reading of a text is not geared toward revealing anything regarding the meaning of that text. It is not about a text, and thus cannot be measured against it. For such a reading, all the distinctive discursive moments and levels blend into one undifferentiated and non-transparent mass. But does this mass, therefore, correspond to that night in which all cows are black? (Gasché, 1998, p. 65) Unlike Gasché, I do not consider these questions to be merely potential objections to de Man’s notion of mere reading, but rhetorical questions requiring no answer. How can the answer to Gasché’s final, Hegelian

Rhetoric as Insurmountable Obstacle 105

question, not be affirmative? ‘Mere reading’ does indeed turn language into ‘one undifferentiated and non-transparent mass’, which surely invalidates de Man’s contention that ‘technically correct rhetorical readings may be boring, monotonous, predictable and unpleasant, but they are irrefutable’ (Resistance, p. 19). They are only ‘technically correct’ according to an inadequate code model which wilfully ignores the inferential elements of discourse. A technically correct ‘mere reading’ of a text is only as irrefutable as the binary digital coding used to convert an analogue recording into a CD, or a scanner’s analysis of a painting, which only sees rows and rows of pixels each consisting of combinations of 256 possible colours, rather than the Mona Lisa or the Girl with a Pearl Earring. De Man dismisses ‘defective models of readingavoidance, referential, semiological, grammatical, performative, logical, or whatever’ (p. 19), but his own model is defective in that it deliberately ignores the possibility that by using inferential processes a reader can overcome the ironic possibilities and the semantic, referential and grammatical ambiguities inherent in language, and thereby understand an author’s intended meaning, or construct a stable meaning of his or her own. Amusingly, de Man writes that ‘none of the objections to theory, presented again and again, always misinformed or based on crude misunderstandings of such terms as mimesis, fiction, reality, ideology, reference and, for that matter, relevance, can be said to be of genuine rhetorical interest’ (p. 12), although of course ‘relevance’ here means political significance rather than Sperber and Wilson’s cognitive theory of language. Yet despite de Man’s misplaced Olympian tone, it is entirely possible to dispute his claims for semiology and his account of rhetoric and the epistemological effects of tropology without being misinformed or crudely and systematically misunderstanding or misrepresenting terms such as rhetoric, mimesis, reality, fiction, ideology, reference or, for that matter, relevance. De Man first asserts that ‘the resistance to theory is a resistance to language about language’ (Resistance, pp. 12–13), and then that it is ‘in fact a resistance to reading’ (p. 15). But one can resist semiotic or deconstructionist approaches to literary language and literature (or what de Man calls theory), from the perspective of linguistic pragmatics, which is nothing but ‘language about language’. Pragmatics takes into account language users’ intentions and the contextual factors which intervene in the determination of reference, propositional attitude, and illocutionary force, and contains a theory of implicatures, including such rhetorical devices as metaphor and irony. Relevance theory, just as much as de Man’s literary theory, takes it for granted that ‘literature is

106

Paradigms of Reading

not a transparent medium’ and that ‘the grammatical decoding of a text leaves a residue of indetermination’. Yet rather than seeing the ‘figural dimensions of a text’ as ‘determining’ (Resistance, p. 15), it concentrates on speakers’ and writers’ intentions. Instead of ‘considering language as a system of signs and of signification’ (pp. 8–9), it considers the crucial role of implicature and inference, alongside the encoding and decoding of signs and explicatures. Relevance theory seeks to show how inference both suppresses and complements rhetorical, grammatical and semantic codes, allowing us to discount seemingly irrelevant rhetorical possibilities, and to construct optimally or manifestly relevant meanings that are irreducible to grammatical, semantic, rhetorical or any other codes. De Man writes that ‘we seem to assume all too readily that, when we refer to something called “language”, we know what it is we are talking about, although there is probably no word to be found in the language that is as overdetermined, self-evasive, disfigured or disfiguring as “language” ’ (Resistance, p. 13). However his equation of language with tropes, or with ironic disruption, does not improve matters. The Russian Formalists, in the 1920s, also had an overall theory of poetic language, but they, or at least Boris Eikhenbaum, a self-appointed spokesman for the group, regarded their accounts of literature as fallible hypotheses rather than apodictic statements. As Eikhenbaum put it: We advance concrete principles and stick to them to the extent that they are justified by the material. If the material requires their further elaboration or alteration we elaborate or alter them. In this respect we are free enough of our own theories, as a science should be if there is a difference between theory and conviction. A science lives not by establishing certitudes but by overcoming errors. . . . In the moment that we ourselves are compelled to admit that we have a universal theory, ready for all the contingencies of past and future and therefore not in need of evolving, we would have to admit that the formal method had ceased to exist, that the spirit of scientific inquiry had departed from it. (Eikhenbaum, Literatura, quoted in Steiner, 1995, pp. 16–17) De Man, on the contrary, appears rather too ready to advance his account of rhetoric as a universal theory.

7 Words and the World: The Problem of Reference

Pragmatic accounts of language such as relevance theory take as their starting point the inadequation of signs and meanings, but seek to show how people can represent or interpret phenomenal experience in language, despite the lack of coincidence between words and the world. In this chapter I will try to show that although, as Paul de Man regularly pointed out, the relationship between words and things is conventional or contractual rather than phenomenal or constitutive, it does not follow that the referential function of language is always unstable, or that we cannot express our thoughts, or that reference is incompatible with grammar. There is, of course, a category of words whose reference changes every time they are used, namely deictics, including personal pronouns, demonstratives such as this and that, and adverbs such as here and there, now and then. For pragmaticians, this fact of language, which has been analysed by Benveniste and Jakobson, among others,presents few problems.1 Yet for de Man, it becomes further evidence of the impossibility of expressing what we mean. De Man analyses Hegel’s brief discussion of deictics in paragraph 20 of the first volume of the Encyclopedia, in which he argues that opinions are particular and idiosyncratic and therefore inexpressible, as language can only express universal truths, or in other words, we can only think what we can say. I will show how de Man turns this on its head, interpreting Hegel as arguing that our thoughts have an ineffable content that language is not able to express. I will also contest de Man’s claims that grammar can only function by suspending reference, while texts in turn generate referents that subvert grammar, by reference to the hermeneutic principle of application. 107

108

Paradigms of Reading

Being and becoming Before considering the arguments about reference and deixis in the essays de Man wrote after his ‘rhetorical turn’ at the end of the 1960s, it will be helpful to briefly summarize some of his claims concerning the disjunction between words and the world, and the consequent impossibility of capturing experience in language, from his ‘phenomenological’ period in the 1950s and early 1960s. Although in much of his better known later work ‘selves’ give way to tropes, and anything human tends to be described as a linguistic complication, de Man had previously been greatly concerned with the theme of subjectivity or inwardness or interiority, or what Hegel called the ‘unhappy consciousness’: the divided or alienated character of being, and the noncoincidence of the self and the world, and of words and things. In a series of essays written between 1954 and 1956, he first attempted to demonstrate the dangers of believing that words, especially the poetic use of words, could correspond to natural reality, when there is in fact an impenetrable gulf between them. De Man was heavily influenced by Alexander Kojève’s (1947) Heideggerean interpretation of Hegel which reinterpreted the Phenomenology of Mind, emphasising the aspects of negation and separation rather than the Absolute Spirit, totality and identity. 2 De Man wrote of the ‘unbearable tension’ of the ‘awareness of a deep separation between man’s inner consciousness and the totality of what is not himself’ (Writings, pp. 14–15), and of ‘the feeling of being trapped, accompanied by a vague premonition that poetry alone could end the oppression, provided one could find access again to true words’ (p. 12). This had been ‘the central subject of all great poetry’ for ‘well over a century’ (p. 13), including that of Hölderlin, Keats, Baudelaire, Rimbaud, Yeats and Rilke. Yet this predicament is not confined to poetry as ‘the same weight paralyzes all human actions and relationships’ (p. 12). De Man describes this ‘experience of alienation or separation’ as ‘universal’ (p. 71), although awareness of this separation was especially prominent in romanticism. Our ‘unhappy consciousness’ desires serenity and is tempted to forget and repress this ontological anxiety. De Man suggests, however, that only an ‘authentic’ attitude toward time will allow us to overcome the division of being, and that this has to be sought in a poetry of (temporal) becoming rather than a poetry of (permanent) being: ‘far from being antihistorical, the poetical act (in the general sense that includes all the arts) is the quintessential historical act: that through which we become conscious of the divided

Words and the World

109

character of our being, and consequently, of the necessity of fulfilling it, of accomplishing it within time, instead of undergoing it in eternity’ (p. 33). De Man’s basic argument is that language is a product of consciousness rather than a reflection of external nature. Poetry derives from the ‘nothingness’ of consciousness, as distinct from the substantiality of things. He repeatedly warns against the seductive power of the ‘romantic’ notion, primarily associated with Coleridge and Schelling, but also common to many post-romantic poets and philosophers and literary critics, that naturalising, organic language, such as that of symbol or metaphor, can actually fuse subject and object, and reconcile the antinomies of consciousness and nature, the mind and the physical world, sensuous intuition and cognitive understanding, perception and apperception, experience and knowledge, thoughts and things, intention and meaning, idea and form, form and content, word and world, word and act, spirit and substance, the phenomenal and the semantic, the ontic and the ontological, the temporal and the eternal, and so forth. De Man insists on the absolute disjunction between these realms, and opposes the ‘aesthetic ideology’ or the belief that poetic genius can reconcile such dichotomies by finding the signifying structures of language that coincide with phenomenal experience, thereby creating an identity between subject and object, mind and nature, or beauty and truth. 3 Unlike Schiller and his successors, proponents of an ‘aesthetic education’ that would resolve the antinomy of form and sense (a project originating, according to de Man, in a misreading of Kant), he insists that empirical, sensuous, phenomenal experience can neither be seized nor represented in language.4 In ‘The Double Aspect of Symbolism’, de Man wrote that the poet ‘is no longer close enough to things to name them as they are. . . . The word, the logos, no longer coincides with the universe but merely reaches out for it in a language that is unable to be what it names – a language that, in other words, is merely a symbol’ (Romanticism, p. 150). The ‘no longer coincides’ seems to presuppose a prelapsarian time in which this separation did not exist, but de Man goes on to argue, pace Eliot, that this ‘dissociation of sensibility’, which ‘may well be at the root of all human consciousness’ (p. 151), has always existed.5 De Man suggests that ‘if the poet finds himself in this state of separation and solitude, his consciousness cut off from the unity of the natural world, his first and natural impulse will be to use poetical language as a means to restore the lost unity. He can then look on the symbol as on a key to reenter the world of unity from which he has been exiled’ (p. 151). The

110

Paradigms of Reading

idea is that in a poetic symbol or metaphor, ‘the infinite distance that separates object from consciousness is crossed at lightning speed, and unity is restored not merely among the diversity of natural objects but also between the spiritual and material world’ (p. 151). Unfortunately, de Man insists, all such attempts fail. Baudelaire and his successors attempted to recapture the fundamental unity of all being by an act of the imagination – the creation of symbols – but they did so in terms of not-being. Inevitably, Baudelairean poets ‘come to express unity in terms of death’ (Romanticism, p. 154), as identification with the natural object results in the death of consciousness rather than a true unification. Mallarmé, on the contrary, strove to avoid direct identification between consciousness and the natural object because consciousness absorbed by the power of being annihilates itself: ‘Annihilation of consciousness is primarily the annihilation of language, and since the poet’s only but irrevocable commitment is to language, he can never accept unity on those terms. He has to be, by his essential choice, on the side of consciousness and against natural being’ (p. 158). In the work of Mallarmé, Consciousness attempts to think through the essential otherness of the object, to transform this otherness into a cognitive knowledge stated in language. Poetry is not an identification with the object but a reflection on the object, in which consciousness moves out toward the object, attempts to penetrate it and then, like a reflected ray of light, returns to the mind, enriched by its knowledge of the outside world. And it is by means of this process of thinking the other that the mind learns to know itself. (Romanticism, p. 158) Yet the Mallarméan symbol fails to overcome the subject–object separation that Hegel outlines in such painstaking detail in the Phenomenology because it is ‘a mediation between the subject on the one side and nature on the other in which both keep their separate identities, but in which a third entity, language, contains within itself their latent opposition’ (Romanticism, pp. 158–9). Language can neither unite consciousness with natural being nor capture the phenomenality of experience. As mentioned in Chapter 3, this also problematizes the widespread account of reading (expounded, for example, by I. A. Richards in Principles of Literary Criticism), according to which an author should construct a linguistic structure that corresponds as closely as possible to the initial experience, and in which correct reading involves recapturing

Words and the World

111

an author’s phenomenal experience from his or her words. De Man denies that ‘language, poetic or otherwise, can say any experience, of whatever kind, even a simple perception’ (Blindness, p. 232), because ‘the experience of the object is not the experience of the consciousness of the object’ (p. 244). Rather than containing or reflecting experience, language constitutes it. In a series of essays on Wordsworth, Hölderlin and Rousseau written in the mid-1960s, de Man suggested that these poets knew the analogy between nature and consciousness to be merely a creation of the human mind, an attempt to hide from our separation from Being and the ‘barrenness’ of our temporal condition. Instead, in what de Man calls authentic romantic thought, perception gives way to the temporal world of the imagination and memory, resulting in an authentic attitude towards time, and a poetry of becoming rather than one of Being. An ‘authentic’ understanding of temporality displaces the necessity of linking the mutable self with the permanence of nature, and finding a defensive strategy by which to escape what Wordsworth called ‘the unimaginable touch of time’. De Man seeks to overturn the conventional argument that the romantic poets invented poetic symbols in which subject and object are supposed to coincide, suggesting that their use of allegory, which necessarily refers to an anterior sign, is of far greater importance. He describes much nineteenth and twentieth century European literature as a regression from the early romantics’ awareness of the temporal relationships existing within allegorical signs.6 At first sight, the very comparison between relevance theory and de Man’s Heideggerian speculations about words and the world may appear to be quite simply irrelevant (not to say impertinent). In Heidegger’s terms, relevance theory is a purely ontic science: a wholly pragmatic (and, one might say, un-becoming) account of how we use language, instrumentally and functionally, in our factitious, everyday existence (or, if you prefer, non-being). It clearly has nothing at all to say about the ontological superiority of the natural object, or the divided and time-bound nature of being and the nearness of death. It does not advocate an openness in the face of Being, or a questioning of its sense, meaning, or purpose, or (most importantly) a submission to language. On the contrary, it takes for granted the non-coincidence of language and phenomenal experience, and demonstrates how language only allows us to construct interpretive representations of our thoughts, and how conventional word meanings can be both enriched and loosened.

112

Paradigms of Reading

Concepts, metaphors, catachreses and reference In Allegories of Reading, de Man repeatedly affirms that the referential function of language cannot be disregarded. In the essay on Rousseau’s ‘Profession du foi’ (in the fourth book of Émile) he writes of ‘the referentiality inherent in the linguistic model’ (Allegories, p. 240). In the essay on Rilke, he states that ‘the notion of a language entirely freed of referential constraints is properly inconceivable. Any utterance can always be read as semantically motivated, and from the moment understanding is involved the positing of a subject or an object is unavoidable’ (p. 49). In the essay on Julie, he stresses that ‘it would be quite foolish to move away from the constraint of referential meaning’ (p. 201), and insists that Rousseau’s ‘radical critique of referential meaning never implied that the referential function of language could in any way be avoided, bracketed, or reduced to being just one contingent linguistic property among others’ (p. 207). In the last of the three essays on Nietzsche, however, de Man points out the paradox or the ‘complication’ that any deconstructive reading ‘states the fallacy of reference in a necessarily referential mode’ (Allegories, p. 125). Yet de Man’s questioning of the referential stability of language is based on the argument that the rhetoricity of all words and concepts entails that concepts are about language rather than things, an argument that I have already outlined, and described as fallacious. Rousseau described conceptualisation as the process of inventing common nouns on the basis of the perception of resemblances between objects. In Chapter 5, I showed how de Man interprets this as a restatement of the classical definition of metaphor, conceived as an exchange or substitution of properties on the basis of resemblance. Thus, for de Man, since the substitution of sameness for difference characterises all conceptual language, it becomes impossible to say whether denomination is literal or figural. Moreover, conceptualisation is an intralinguistic process, so that all concepts are in fact about language. In ‘Self (Pygmalion)’, de Man explains how this account of conceptualisation as metaphor problematises the referential function. He states that: all cognitive discourse . . . has to be about an entity such as ‘man’ in which the noun is a conceptual metaphor that replaces a delusive play between identity and difference. . . . this entity turns out to be the substitution of a definitional for what was only a hypothetical knowledge, an epistemological metaphor substituting certitude for ignorance on the basis of an assumed resemblance. . . . The resulting

Words and the World

113

discourse is complex not just because it has a plurality of perhaps incompatible meanings but because the semantic status of any of these meanings can never be determined. . . . Language can only be about something such as man (i.e., conceptual), but in being about man, it can never know whether it is about anything at all including itself, since it is precisely the aboutness, the referentiality, that is in question. (Allegories, pp. 160–1) This questioning of referentiality, however, is only as valid as the original argument about concepts in fact being metaphors about language. I have already described this as a misreading of the Second Discourse, and also suggested that in any case, the conjectured figurative and connotative origin of language in no way changes its function. In a splendidly succinct paragraph in ‘Allegory (Julie)’ (already partly quoted above in Chapter 5, note 10), de Man indulges in the perilous rhetorical strategy of outlining at length a pragmatic position he is going to question: To understand primarily means to determine the referential mode of a text and we tend to take for granted that this can be done. We assume that a referential discourse can be understood by whoever is competent to handle the lexicological and grammatical code of a language. Neither are we helpless when confronted with figures of speech: as long as we can distinguish between literal and figural meaning, we can translate the figure back to its proper referent. We do not usually assume, for example, that someone suffers from hallucinations merely because he says that a table has four legs; the context of common usage separates the figural meaning of the catachresis (which, in this case, leads to the referent) from its literal denotation (which, in this case, is figural). Even if, as is so often said to be the case for poetic language, the figure is polysemous and engenders several meanings, some of which may even be contradictory to each other, the large subdivision between literal and figural still prevails. Any reading . . . postulates the possibility of distinguishing the literal from the figural. This decision is not arbitrary, since it is based on a variety of textual and contextual factors (grammar, lexicology, tradition, usage, tone, declarative statement, diacritical marks, etc.). But the necessity of making such a decision cannot be avoided or the entire order of discourse would collapse. The situation implies that figural discourse is always understood in contradistinction to

114

Paradigms of Reading

a form of discourse that would not be figural; it postulates, in other words, the possibility of referential meaning as the telos of all language. (Allegories, p. 201) I find little to disagree with in these sentences, and believe that de Man’s concession that common usage generally allows us to cope with catachresis – or indeed blinds us to the fact that much language is, technically, catachretic in the first place – has greater import than he realises. Our blindness to supposed catachreses such as the legs of a table is duplicated in our blindness to the supposedly metaphorical nature of concepts: few of us, when we talk about ‘man’, worry that we might in fact be talking about nothing at all or only about language itself. But, of course, in the following paragraph de Man asserts that ‘the critical thrust of Rousseau’s theory of language, in the Second Discourse and the Essay [on the Origin of Language] undermines this model’ (Allegories, p. 201). As does the Second Preface to Julie, in which Rousseau declines to confirm that he is the author of the ‘collected and published’ letters, leaving the reader to decide whether the correspondence is real, written by the people (or ‘historical referents’) who sign the letters, or fictional (and therefore figural). De Man extrapolates from this particular example: The innumerable writings that dominate our lives are made intelligible by a preordained agreement as to their referential authority; this agreement however is merely contractual, never constitutive. It can be broken at all times and every piece of writing can be questioned as to its rhetorical mode, just as Julie is being questioned in the Preface. Whenever this happens, what originally appeared to be a document or an instrument becomes a text and, as a consequence, its readability is put in question. (Allegories, p. 204) However in the case of Julie ou la Nouvelle Héloïse – itself scarcely a representative example of the innumerable writings that dominate our lives – the contractual agreement seems to have been very widely accepted indeed: de Man is obliged to concede that ‘much has gone astray in the critical reading of Julie’ because no one has even so much as recognized ‘the linguistic problem of referentiality’ (Allegories, pp. 188–9). Unfortunately, this ambient blindness does not lead de Man to doubt, even slightly, the veracity of his ‘insight’ into the metaphorical nature

Words and the World

115

of all concepts, and the impossibility of distinguishing between the literal and the figural.7

Reference and ideology De Man discusses reference again in ‘The Resistance to Theory’, but here the discussion no longer turns on the tropological origin of all concepts, but on the conventional nature of the relationship between words and things. As Saussure pointed out, both elements of the sign – signifier and signified – are essentially arbitrary. There is generally no link between the material properties of a signifier in a given language and the signified itself; and different languages divide up the world in different ways, so that they have different systems of differences among their signifiers. Yet the poetic use of language clearly takes into account the materiality of the signifier, perhaps attempting to find, in de Man’s words, ‘a convergence of the phenomenal aspects of language, as sound, with its signifying function as referent’ (Resistance, p. 9). De Man describes this ‘Cratylism’ as ‘an aesthetically oriented conception’, adding that ‘one could, in fact, without distortion, consider aesthetic theory, including its most systematic formulation in Hegel, as the complete unfolding of the model of which the Cratylian conception of language is a version’ (p. 9). But he immediately proceeds to subsume aesthetics to rhetoric, because the use of words for phonetic rather than semantic reasons (a defining instance of literariness since at least Jakobson’s well-known essay ‘Linguistics and Poetics’, Jakobson, 1981), is a trope, paranomasia or paranomasis, as had been pointed out by both Jakobson himself and Henryk Markiewicz. Markiewicz describes paranomasis as a sound-based form of metaphor in which the phonic resemblance imitates the possibility of a substitution at the level of substance (see Blindness, p. 284), but de Man insists that a particular paranomasis (which, of course, he describes as an ‘effect which language can perfectly well achieve’, rather than an effect which language users can achieve): bears no substantial relationship, by analogy or by ontologically grounded imitation, to anything beyond that particular effect. It is a rhetorical rather than an aesthetic function of language . . . that operates on the level of the signifier and contains no responsible pronouncement on the nature of the world. . . . The phenomenality of the signifier, as sound, is unquestionably involved in the correspondence between the name and the thing named, but the link, the

116

Paradigms of Reading

relationship between word and thing, is not phenomenal but conventional. This gives the language considerable freedom from referential restraint. (Resistance, p. 10) However the poetic use of language – using words for phonetic rather than semantic reasons – is less often undertaken in order to imitate natural objects in the world than to set up phonological, morphological and syntactic patterns or correspondences, in pleasing, but non-Cratylian, lines using repetitions, assonance, alliteration, slightly altered syntax, and so on. Frank Kermode (1985, pp. 53–4), for example, describes Hamlet’s comment on his relation to Claudius – ‘A little more than kin, and less than kind’ – as ‘almost a textbook paranomasia; it depends upon the resemblance of one word to another. “Kin” and “kind” are an imperfect doublet, related phonetically, orthographically, and semantically, yet also distinct in all these ways, and represented as antithetical, thus stressing the horrible disparities implied by so close and profane a union.’ Shakespeare clearly chose the words kin and kind for the material properties of their signifiers, but they still signify (and refer to) ‘kin’ (as in relative or relation) and ‘kind’ (as in good-hearted, compassionate and friendly). The phonic and metrical constraints of poetry often influence the choice of words, but it is usually only in nonsense poetry (‘’Twas brillig, and the slithy toves / Did gyre and gimble in the wabe / All mimsy were the borogoves / And the mome raths outgrabe’) that language is being used non-referentially. It is precisely the conventional nature of the relationship between words and things (and their attributes) and actions that permits the recognition of the deviant use of language in poetry (‘The earth is blue like an orange’, ‘Je est un autre’). De Man writes that: In a genuine semiology as well as in other linguistically oriented theories, the referential function of language is not being denied – far from it; what is in question is its authority as a model for natural or phenomenal cognition. Literature is fiction not because it somehow refuses to acknowledge ‘reality’, but because it is not a priori certain that language functions according to principles which are those, or which are like those, of the phenomenal world. It is therefore not a priori certain that literature is a reliable source of information about anything but its own language.8 (Resistance, p. 11)

Words and the World

117

If, however, one does not share de Man’s obsession with the lack of correspondence between language and phenomenal cognition, it is quite possible to argue that we use language as best we can to express our phenomenal cognitions, using the available conventional signifiers and signifieds, both literally and figuratively, to create interpretive representations of our thoughts and experiences, which we expect our addressees to interpret inferentially. Language is the only thing we have as a model for natural or phenomenal cognition, and the fact that our signs (as opposed to symbols) only conventionally represent what they signify need not be a problem. De Man writes that What we call ideology is precisely the confusion of linguistic with natural reality, of reference with phenomenalism. It follows that, more than any other mode of inquiry, including economics, the linguistics of literariness is a powerful and indispensable tool in the unmasking of ideological aberrations, as well as a determining factor in accounting for their occurrence. Those who reproach literary theory for being oblivious to social and historical (that is to say ideological) reality are merely stating their fear at having their own ideological mystifications exposed by the tool they are trying to discredit. They are, in short, very poor readers of Marx’s German Ideology. (Resistance, p. 11) Yet despite the arbitrary nature of the signifieds of a given language, and the fact that some or many or all of our concepts were originally tropes, we still use a great many words (signifiers related to signifieds) to refer to tangible objects and actions. There are indeed many other words whose content is more abstract, and where confusing reference with phenomenalism leads us into the realm of ideology – the people, the will of the people, the moral majority, conventional morality, public opinion, the enemy within, our way of life, progress, truth, market forces, economic reality, globalization, competitivity, rightsizing, and so on. But the argument that language cannot reliably state anything or refer to anything, and that reading is therefore impossible, is not a particularly useful notion either. Andrzej Warminski paraphrases and endorses de Man’s account of ideology in his Introduction to Aesthetic Ideology, and draws on it in his response to de Man’s wartime journalism. For Warminski, reading involves rejecting even the possibility that referential language could serve ‘as a reliable model for cognition on the basis of which we could take action – for instance, like the act of judging de Man as either guilty

118

Paradigms of Reading

or innocent. Reading suspends: it suspends knowledge and it suspends judgment’ (Warminski, 1989, p. 392). This seems to me a very high price to pay. We may contest the ‘natural reality’ of ‘resistance’, ‘collaboration’ and ‘objectivity’, but must this necessarily involve suspending all judgement on sentences like ‘the necessity of action which presents itself in the form of immediate collaboration imposes itself upon every objective mind’, as de Man wrote in 1941?

‘Ich kann nicht sagen was ich nur meine’ Quite apart from determining the ‘referential mode’ (literal or figural) of any piece of discourse, and deciding whether particular words have been chosen for referential or poetic purposes, it is necessary to determine the intended referent of each and every deictic expression, which, in de Man’s terms, is equally a conventional or contractual operation. Wlad Godzich discusses deictics in his Foreword to The Resistance to Theory, and declares that they are a problematic element of language, but de Man only deals with them at any length in ‘Sign and Symbol in Hegel’s Aesthetics’, in Aesthetic Ideology. Hegel discussed the deictic nature of the first person singular pronoun in the first volume of the Encyclopedia. De Man analyses this passage and interprets Hegel as arguing that ‘I cannot say what I think’. He does this during a discussion of Hegel’s account of the sign and the symbol in the Lectures on Aesthetics – which is unsurprising, given that every single one of de Man’s essay titles indicates (or as we often say, ‘says’) precisely what (he thinks) it is about. De Man writes that The theory of the aesthetic . . . is predicated, in Hegel, on a theory of art as symbolic. The famous definition of the beautiful as ‘the sensory appearance [or manifestation] of the idea [das sinnliche Scheinen der Idee]’ does not only translate the word ‘aesthetics’ . . . but it could best itself be translated by the statement: the beautiful is symbolic. The symbol is the mediation between the mind and the physical world of which art manifestly partakes, be it as stone, as colour, as sound or as language. (Aesthetic, p. 93) However, de Man continues, ‘After having stated casually that the symbol can be considered a sign (das Symbol ist nun zunächst ein Zeichen), he goes on to distinguish between the symbolic and the semiotic function

Words and the World

119

and leaves no doubt as to what side of this dichotomy art is on: “In the case of art, we cannot consider, in the symbol, the arbitrariness between meaning and signification [which characterizes the sign], since art itself consists precisely in the connection, the affinity and the concrete interpenetration of meaning and of form”’ (Aesthetic, p. 93). De Man quotes a further distinction Hegel made between sign and symbol (in paragraph 458 of the third part of the Enzyklopädie der philosophischen Wissenschaften): the sign has long been defined in terms of the arbitrary relation between the sensory component and the intended meaning, and thus ‘differs from the symbol, a perception whose own determination [or meaning] more or less corresponds, essentially and conceptually, to the content it expresses as a symbol’ (quoted by de Man, Aesthetic, p. 96). Glossing Hegel’s distinction, de Man writes, To the extent that the sign is entirely independent with regard to the objective, natural properties of the entity toward which it points and instead posits properties by means of its own powers, the sign illustrates the capacity of the intellect to ‘use’ the perceived world for its own purposes, to efface (tilgen) its properties and to put others in their stead. This activity of the intellect is both a freedom, since it is arbitrary, and a coercion, since it does violence, as it were, to the world. The sign does not actually say what it means to say or, to drop the misleading anthropomorphic metaphor of a speaking sign endowed with a voice, the predication involved in a sign is always citational. (Aesthetic, p. 96) The use of signs in predicative sentences, de Man continues, ‘presupposes an implicit subject (or I) which frames the statement and makes it into a quotation: I say (or I declare, or I proclaim) that . . . ’ (Aesthetic, p. 96). The sign thus ‘touches upon the question of the relationship between subject and predicate in any declarative sentence. From the question of the sign we are taken, then, by the logic of the passage itself, to the question of the subject’ (pp. 96–7). Here, de Man jumps back to paragraph 20 of the Vorbegriff of the first part of the Enzyklopädie (a volume translated into English under the title Logic), in which Hegel distinguishes between the thinking subject and the perceiving subject, in a manner that, for de Man, anticipates his later differentiation between sign and symbol: Just as the sign refuses to be in the service of sensory perceptions but uses them instead for its own purposes, thought, unlike perception,

120

Paradigms of Reading

appropriates the world and literally ‘subjects’ it to its own powers. More specifically, thought subsumes the infinite singularity and individuation of the perceived world under ordering principles that lay claim to generality. The agent of this appropriation is language. ‘Since language,’ says Hegel, ‘is the labor of thought, we cannot say anything in language that is not general’ . . . Thus the sign, random and singular at its first position, turns into symbol just as the I, so singular in its independence from almost anything that is not itself, becomes, in the general thought of logic, the most inclusive, plural, general, and impersonal of subjects. (Aesthetic, p. 97) Hegel does indeed argue that all our concepts and words and thoughts are abstract, general and universal. Continuing from the sentence that de Man quotes, he writes, ‘What I only mean or suppose is mine: it belongs to me – this particular individual. But language expresses nothing but universality; and so I cannot say what I merely mean’ [Was ich nur meine, ist mein, gehört mir als diesem besonderen Individuum an; wenn aber die Sprache nur Allgemeines ausdrückt, so kann ich nicht sagen, was ich nur meine] (Hegel, 1975, p. 31; Enzyklopädie, 1970, p. 74). He had already put forward the same argument in making the transition from sense-certainty to perception, the first and second forms of consciousness, in the first part of the Phenomenology of Mind, and in the Introduction to the Encyclopedia. Nouns are clearly universals, and Hegel argues the same for demonstratives. Although ‘this [individual]’, for example, refers to someone present ‘now’ and ‘here’, these too are universals. Hence even when we intend particular reference, our utterances function as universals, applying indifferently to all members of a class. Proper names would appear to damage this thesis, but Hegel states, in the Preface to the Phenomenology, that they have no conceptual content and are consequently meaningless (i.e. language can express nothing meaningful that is not also universal.) This is, however, a circular argument: Hegel can only maintain the universal character of all meaningful language by asserting proper names to be meaningless; proper names are declared meaningless in the sense that they do not express a universal.9 Furthermore, declaring proper names to be meaningless does not mean that they cannot be used to refer to particulars, unless meaning and reference are to be equated. Yet although de Man appears to present the passage quoted above as a paraphrase of Hegel, it is not. There is nothing in Hegel’s paragraph 20 that might motivate de Man’s shift from the idea that a sign is

Words and the World

121

a man-made (albeit arbitrary) entity that we employ, literally or figuratively, to refer to or act upon the perceived world (‘the sign illustrates the capacity of the intellect to “use” the perceived world for its own purposes’) to the familiar active post-Saussurean sign (which de Man himself describes as ‘misleading’) that somehow uses our sensory perceptions in its own way. Neither does Hegel say that because thought is general, in thought signs turn into symbols. In this section of the Encyclopedia, Hegel is contrasting what we intend (meinen) to do with language and what the nature of language will in fact permit us to do, and playing with the paranomasis between the verb meinen and the possessive pronoun mein. J. B. Baillie’s English translation of the Phenomenology uses mean for meinen and meaning for Meinung (Hegel, 1949, pp. 152–3). William Wallace’s English translation of the first book of the Encyclopedia, quoted above, uses both mean and suppose. This is misleading as the English mean has the senses of both to signify and to intend, whereas meinen can only be synonymous with intend (and can therefore only take people, and not words, concepts, propositions, theories or behaviour, as grammatical subjects). Meinen can also mean to be of the opinion that (or somewhat archaically, to opine), especially in the sense of an unsupported meaning (as opposed to knowledge). Thus a Meinung can only belong to a person (and not a word, concept or proposition), and is best translated as opinion rather than meaning. (A word’s meaning is its Bedeutung.) Hegel is thus arguing that opinions (Meinungen) are particular and idiosyncratic, and therefore inexpressible in language, in contrast to thoughts which are universal and therefore expressible. This is part of his attack on empiricism: he denies the possibility of linguistically expressing what we only sense and feel (the particulars presented to us by sensation and feeling). Hence what language cannot express (the ineffable) is unreal, unimportant and, most importantly, untrue. Language, on the contrary, can only express universal truths. In short, Hegel is arguing that we can only think what we can say. This is a very common argument. A more recent example comes from Wittgenstein (1958, paragraph 329), who claimed that ‘When I think in language, there aren’t “meanings” going through my mind in addition to the verbal expressions: the language is itself the vehicle of thought’, and that it is impossible to have a thought before finding the expression, or have a thought in one’s head but not in one’s words. 10 The relevance-theoretic account of language is the opposite: we can think many thoughts that our language cannot encode, and we can communicate many thoughts that our utterances do not encode. S&W also

122

Paradigms of Reading

distinguish meaning from communication, arguing that ‘something can be communicated without being strictly speaking meant by the communicator’ (Relevance, p. 57). But although ordinary inferential abilities make successful communication possible, they obviously don’t guarantee it, and what we intend (meinen) to do with language can fail. De Man, however, in his reading of paragraph 20 of the Encyclopedia inverts Hegel’s argument that we can only think what we can say, suggesting instead that our thoughts have an ineffable content that language is not able to express. He translates Hegel’s was ich nur meine as what is only my opinion, but then insists that meaning is also connoted, by way of an appeal to public opinion (öffentlichte Meinung): ‘Since language states only what is general, I cannot say what is only my opinion [so kann ich nicht sagen was ich nur meine].’ The German version is indispensable here since the English word ‘opinion’, as in public opinion (öffentlichte Meinung), does not have the connotation of ‘meaning’ that is present, to some degree, in the German verb meinen. In Hegel, the assimilation of ‘meaning’ to ‘me’ (or I) is built into the system, since the generality of thought is also the appropriation, the making mine of the world by the I. It is, therefore, not only legitimate but necessary to hear, in the German word meinen . . . a connotation of meinen as ‘to make mine’, a verbalization of the possessive pronoun mein. . . . ‘Ich kann nicht sagen was ich (nur) meine’ then means ‘I cannot say what I make mine’, or, since to think is to make mine, ‘I cannot say what I think’, and, since to think is fully contained in and defined by the I, since Hegel’s ego cogito defines itself as mere ego, what the sentence actually says is ‘I cannot say I’, – a disturbing proposition in Hegel’s own terms since the very possibility of thought depends on the possibility of saying ‘I’. (Aesthetic, pp. 97–8) De Man continues, defensively, ‘Lest this itinerary by way of the signifier meinen appear too arbitrary to be taken seriously, the sequel to the passage makes explicit what one can already choose to hear in the original sentence’. The sequel is a discussion of the paradox that the most particular deictic or demonstrative designations such as now, here, this – and I – are also the most powerful agents of generalisation. De Man says, ‘Hegel can therefore write the following quite astonishing sentence: “When I say ‘I’, I mean myself as this I at the exclusion of all others; but what I say, I, is precisely anyone [wenn ich sage: ‘Ich’, meine ich mich als diesen alle anderen Ausschliessenden; aber was ich sage, Ich, ist

Words and the World

123

eben jeder]” ’ (Aesthetic, p. 98). For de Man, this demonstrates the contradiction between sagen and meinen, to say and to mean, dire and vouloir dire, and confirms that Ich kann nicht sagen was ich (nur) meine ‘has to be read, next to its ordinary meaning, in the sense of “I cannot say I.”’11 Thus we have an ‘inescapable obstacle’, the fact that ‘the position of the I, which is the condition for thought, implies its eradication . . . as the undoing, the erasure of any relationship, logical or otherwise, that could be conceived between what the I is and what it says it is’ (Aesthetic, pp. 98–9). De Man goes on, in a few dense pages, to argue that the mind clearly needs to shelter itself from this self-erasure, and attempts to do so by way of the aesthetic; that the I is both a sign and a symbol but that the relationship between sign and symbol is one of mutual obliteration; that the Romantic ‘ideology of the symbol’ is a ‘rhetorical model’ and ‘a dialectics of internalization’; that the wholly mechanical faculty of memorisation must be sharply distinguished from both recollection and imagination; that memory effaces resemblance or recollection just as the I effaces itself; and that the faculty that enables thought to exist also makes its preservation impossible. Yet much of his argument – which again takes on added resonance following the rediscovery of the wartime writings – is predicated on a tendentious reading of Hegel, and a misunderstanding of the deictic function of language. Surprisingly, given his pronouncements elsewhere about the irrefutability of technically correct rhetorical readings, de Man describes his interpretation of meinen as something ‘that one can . . . choose to hear’ in Hegel. Raymond Geuss, in a response to de Man, points out that in Hegel the nur in ‘ich kann nicht sagen was ich nur meine’ is not in parentheses, as de Man chooses to write it, but rather the key to understanding the passage (Geuss, 1983, p. 380). Geuss reiterates the canonical reading of this passage: Hegel is merely drawing attention to the fact that it is general concepts which make knowledge and reference possible: We might falsely come to think that our knowledge of the world consists in our standing in some direct ‘nonconceptual’ relation to individual things. ‘Was ich nur meine’ then is the purported object of such a (in fact, impossible) purely nonconceptual reference, and ‘Ich kann nicht sagen was ich nur meine’ means: ‘If one were to try to construct the notion of a particular purely as intended in a direct, unmediated, totally nonconceptual intention, such a particular would not be a possible object of speech (and hence would also be

124

Paradigms of Reading

no possible object of knowledge) for us’, or, more simply, ‘In speech (as in thought) we must use general concepts.’ I can’t say ‘was ich nur meine’ because ‘was ich nur meine’ is defined as the utterly nonconceptual, hence unthinkable, hence unsayable. So ‘ich kann nicht sagen was ich nur meine’ doesn’t at all imply ‘I cannot say what I think’; on the contrary, for Hegel what I think is precisely what I can say. (Geuss, 1983, pp. 380–1) Thus ‘there seem to be no grounds to think that for Hegel the mind tends to “erase itself” ’ (Geuss, 1983, p. 381). Replying in turn to Geuss, de Man accuses him of mishearing the German language in interpreting meinen only as vouloir dire, as intending to refer to some particular thing, rather than as having a connotation of opinion, as in ‘eine Meinung haben’. For de Man, the nur – which he claims to have bracketed only ‘on one occasion, but for entirely different reasons than those attributed to me’ (Aesthetic, p. 189) – is precisely what confirms this normal vernacular use: eine Meinung haben, or meinen, or nur meinen, is inferior to Wissen; and meine Meinung requires the possessive article, whereas truth (die Wahrheit) does not. De Man’s connotative chain of meinen, Meinung, thought, and ‘making mine’ leads him to isolate what he sees as an aporia in Hegel: since he defines thinking as the appropriation or making mine of the world by the I, thought necessarily contains an element of particularity that is incompatible with the supposed universality of truth. A pragmatist such as Rorty, of course, would seek to close off this aporia by way of the anti-foundationalist argument that truth cannot exist independently of the human mind and language, and is a property of sentences rather than the world, which conveniently disposes of the problem of the supposed universality of truth. But Hegel, according to de Man’s reading, had already taken this ‘linguistic turn’ without abandoning the notion of the universality of truth, and run into the problem of deixis, in the sentence quoted above – ‘When I say “I”, I mean myself as this I at the exclusion of all others; but what I say, I, is precisely anyone’ – ‘which names the double function of the word “I” as being, at the same time, the most general and the most partcular of terms’ (Aesthetic, p. 190). De Man writes that by limiting himself to the question of conceptualisation, Geuss cuts himself off from the problem of deixis, but in Geuss’s canonical reading, as from the point of view of much contemporary linguistics, deixis is quite simply not a problem. Geuss suggests that Hegel is merely stating the impossibility of believing that

Words and the World

125

‘I own language’. By demonstrating the dual particular and general nature of the term ‘I’, Hegel seems simply to be anticipating Benveniste’s demonstration of the centrality of deictic elements in language, which is wholly congruent with the dialogical notion that of course language consists of shared, general words and concepts and that ‘we own language’. De Man, on the contrary, makes Hegel say ‘I cannot say I’ and ‘I cannot say what I think’, which makes him a forerunner of the deconstructionist claim that ‘no-one owns language’.12 But this reading of Hegel leads de Man to bracket the ‘only’ in the statement that since language states what is general, I cannot say what is only my opinion, or only what is mine, or only my meaning. But the only makes all the difference in these sentences. De Man seems to be disquieted by the fact that language necessarily transforms the singularity and individuation of the perceived world into generalities, so you cannot say what is only your opinion. But his description of Hegel’s sentence about the deictic nature of the first person singular pronoun as ‘quite astonishing’, seems to reveal a misunderstanding of the nature of deictics. Hegel, according to the canonical reading from which de Man insists that it is essential to ‘swerve’, merely described an elementary fact of language. Far from proving that ‘I cannot say what I think’, the first person pronoun is in fact at the centre of language, and is precisely what enables the subject to utilise the resources of language to say what it thinks. The referent of the first person pronoun is not fixed, and changes each time it is used, depending on the context of utterance: I signifies, and can only be defined by, the person in the particular utterance that includes the word. (Even two successive utterances by the same person containing the word I might not have the same referent, as one of them could well be reported speech or a quotation attributable to someone else.) Each time a speaker utters the word I it inserts him or her in a new moment of time, in new discursive circumstances (as well as evoking, implicitly or explicitly, a you, a hearer or addressee). The entire linguistic systems of space and time are related to the first and second person pronouns. Demonstratives (this, that), relative adverbs (here, there), and all other spatial words serve to localize objects in relation to the speaker. More importantly, time in discourse is always linked to the speech situation. The present is reinvented each time someone speaks, and tense and aspect, mood and modality, and so on, are all related to the present moment – the coincidence of the event spoken about and the discourse itself. The entire verbal system results from the actualisation of the present by the speaker, who at the same time activates the referential function of

126

Paradigms of Reading

language, and places these references in relation to his or her own I/here/now. Other signs by means of which an enunciator can mark his or her presence in an utterance include evaluative, emotive and attitudinal verbs, positional verbs such as come, go, bring and take, explicit performatives (I promise, I swear) and modalising terms (concerning doubt, certainty, possibility, and so on). In other words, ‘the linguistics of enunciation’ (and pragmatics in general) posits a subject as an operational category in linguistic theory, without reintroducing the subjectivism and psychologism rejected by Saussure and subsequent structuralism.

Reference and application Quite apart from the issue of deictics, it is necessary, when considering referentiality, to distinguish between texts such as Rousseau’s Confessions, in which names refer to specific people, and legal and political texts such as Rousseau’s Social Contract, which are designed to apply to a large number of unknown future cases. In hermeneutics, this is known as the problem of application. Given his account of our inevitable prejudgements in his ‘radical historicist’ hermeneutics, Gadamer (1982, p. 275) collapses the distinctions that traditional hermeneutics makes among the aspects of understanding, interpretation and application: just as ‘all understanding is always interpretation’, ‘understanding . . . is always application’. By application, Gadamer means the use of extra-textual presuppositions. In legal hermeneutics, for example, interpretation is never abstract but always linked to a specific case. It is impossible to specify the meaning of a law without referring to cases, to the notion of legal application.13 In ‘Meaning and Significance Reinterpreted’, E. D. Hirsch amended his previous claim that future applications belong to the realm of significance: he now agrees with Gadamer that they can be part of meaning. Hirsch attributes his revision of the distinction between meaning and significance to the realisation that ‘we language users, being limited creatures, intend our verbal meanings to go beyond what we can pay attention to at any moment. We intend our meanings to transcend our momentary limitations of attention and knowledge’ (Hirsch, 1984, p. 202). Thus the originating moment of willed utterance fixes only the principles of further extrapolation, leaving unforeseen possibilities. Using Shakespeare’s 55th sonnet as an example, Hirsch writes, ‘The author’s intention in this poem (and this is characteristic of literature,

Words and the World

127

law, and religion) includes an intention to communicate effectively in the future. Not death nor time shall conquer the powerful speech event’. Shakespeare’s is ‘a historical intention . . . that apparently refuses to fix itself in its originating moment’ (p. 205). With a future-oriented intention we cannot have a rigidly fixed meaning. Hirsch consequently proposes a concept-extension model: ‘the application of literature (and law, and religion) to our own lives depends upon the wide scope of the original textual intention. The unchanging component of that intention is conceptual, and the potential future extension of the conceptual element is immense’ (p. 209). Hirsch thus endorses Emilio Betti’s classification of three types of interpretation: reproductive or re-cognitive, presentational, and normative. Understanding most written language is merely reproductive, unlike translating a text into another language, performing a piece of music, or dramatising a play, which are acts of presentation. Yet interpretation in the fields of jurisdiction and theology is normative: an attempt to derive guidelines for present activity by adapting written texts from the past to changed circumstances. The authors of legal and sacred texts accept that their willed implications go far beyond what they explicitly know: the issue in normative interpretation is not what the author was aware of, but what type of meaning he wills. Similarly, in ‘Promises (Social Contract)’, de Man draws on the orthodox hermeneutic account of the application of legal texts to argue that ‘all laws are future-oriented and prospective; their illocutionary mode is that of a promise’ (Allegories, p. 273). But he develops the logic of the application of general legal texts to particular cases to make an analogy between law and grammar. Because no general legal text can explicitly cover all possible applications, or bind all citizens, de Man argues that ‘the meaning of the contractual text has to remain suspended and undecidable’ (p. 266). Furthermore, From the point of view of the legal text, it is this generality which ruthlessly rejects any particularization, which allows for the possibility of its coming into being. Within the textual model, particularization corresponds to reference, since reference is the application of an undetermined, general potential for meaning to a specific unit. The indifference of the text with regard to its referential meaning is what allows the legal text to proliferate. (Allegories, p. 268) One could helpfully substitute the verb ‘function’ for ‘proliferate’ here, and it is also worth pointing out (as Neil Hertz, 1990, has done), that

128

Paradigms of Reading

there is nothing in Rousseau, nor in de Man’s preceding argument, that requires the characterisation of a legal text as ‘ruthless ‘or ‘indifferent’. Stripped of these examples of de Man’s distinctive anthropomorphisms, this is a perfectly good argument. Yet de Man goes on to combine it with another one: To the extent that a text is grammatical, it is a logical code or a machine. And there can be no agrammatical texts, as the most grammatical of poets, Mallarmé, was the first to acknowledge. Any nongrammatical text will always be read as a deviation from an assumed grammatical norm. But just as no text is conceivable without grammar, no grammar is conceivable without the suspension of referential meaning. Just as no law can ever be written unless one suspends any consideration of applicability to a particular entity including, of course, oneself, grammatical logic can function only if its referential consequences are disregarded. (Allegories, pp. 268–9) This is a non sequitur whereby de Man attempts to establish a parallel between an indisputable claim – that agrammaticality presupposes a grammatical norm – and a wholly disputable one – that grammar presupposes the suspension of reference. A legal or religious proscription such as ‘Thou shalt not kill’ is designed to have general applicability. The sentence contains an explicit pronoun, a modal verb, a negative particle, and an infinitive, and both ‘thou’ and ‘not kill’ refer to everybody. A grammar, on the contrary, contains grammatical categories that can be realised by lexical items, just as mathematical formulas contain symbols that stand in for numbers and mathematical operations, and definitions of diatonic and chromatic scales and keys and chords and intervals in music contain rules rather than specific notes. To write a phrase structure grammar, it is only necessary to rewrite individual nouns, adjectives, verbs and so on with the grammatical names for these parts of speech: e.g. Sentence = Subject + Predicate, or Sentence = Subject + Verb + Object + Complement + Adverbial, or S = NP + VP. To write an abstract universal grammar it is necessary to develop rather more elaborate categories, but even this does not involve a suspension of the referential meaning of the words that fill the slots. Grammars, furthermore, attempt to exclude nongrammatical sentences by specifying selection restrictions designed precisely to prevent aberrant referential consequences. The grammar works independently of the ‘referential consequences’ of individual words, but it does not suspend semantics

Words and the World

129

(and reference) so much as enable it. Similarly, in order to utter the word ‘Nonsense!’, it is necessary to produce a dental-alveolar nasal consonant three times, a dental-alveolar fricative twice, a half-closed front cardinal vowel and an open back secondary vowel, but the fact that phonemes can be thus classified does not mean that the referential meaning of the word that they make up is suspended during their utterance. Thus it is logically false to argue that ‘the logic of grammar generates texts only in the absence of referential meaning, but every text generates a referent that subverts the grammatical principle to which it owed its constitution’, or that there is ‘a fundamental incompatibility between grammar and meaning’ that ‘remains hidden in the everyday use of language’ (Allegories, p. 269). For de Man, ‘the divergence between grammar and referential meaning is what we call the figural dimension of language’, where the materiality of language emerges, and ‘this dimension accounts for the fact that two enunciations that are lexicologically and grammatically identical (the one being, so to speak, the quotation of the other and vice versa) can, regardless of context, have two entirely divergent meanings’ (p. 270). Yet linguistic pragmatics and in particular relevance theory (and also, rather earlier, Bakhtin, whose work was undiscovered in the West for several decades), all demonstrate how ‘enunciations that are lexicologically and grammatically identical . . . can have two entirely divergent meanings’, because of rather than regardless of context. Figural meanings are as underdetermined as literal ones; like non-figural meanings, their interpretation, indeed their very recognition, depends on contextual information. Yet according to de Man’s logic, any text is at the same time a ‘generative, open-ended, non-referential grammatical system’ and ‘a figural system closed off by a transcendental signification that subverts the grammatical code to which [it] owes its existence’, the two systems coinciding with the constative and the performative functions of language. Thus ‘a text is defined by the necessity of considering a statement, at the same time, as performative and constative, and the logical tension between figure and grammar is repeated in the impossibility of distinguishing between two linguistic functions which are not necessarily compatible’ (Allegories, p. 269). Consequently, de Man argues, Rousseau has no control over the rhetorical model of the Social Contract, which promises political change in the direction of greater social justice: the actor or promiser is not so much Rousseau as language itself, inevitably misleading but necessarily conveying the promise of its own truth. The error does not lie with the author; equally, ‘the error is not within the reader; language itself dissociates the cognition from the act.

130

Paradigms of Reading

Die Sprache verspricht (sich); to the extent that it is necessarily misleading, language just as necessarily conveys the promise of its own truth’ (p. 277). The German expression is of course a playful corruption of Heidegger’s well-known formula die Sprache spricht that allows both meanings of versprechen. Hence language both ‘promises (itself)’ and ‘makes a slip of the tongue’. In ‘Excuses (Confessions)’, the concluding chapter of Allegories of Reading, de Man extends this personification of language, and the notion that it performs and makes slips of the tongue, into a more general theory in which language is mechanical, arbitrary, gratuitous and irresponsible, unconstrained by human cognition or reference and detached from meaning, intentions and signifieds. In the two following chapters I will contest this account of language, as well as the more general post-Saussurean notion that language signifies by itself, and that texts are active, while human beings dissolve into various impersonal linguistic systems.

8 Mechanical Performatives

In Chapter 5, I gave examples from de Man’s essays on Nietzsche and on Rousseau’s Second Discourse that showed how de Man broadens the notion of trope to include, almost without exception, all words and concepts, as well as the categories of grammar and syntax. Yet in the closing chapters of Allegories of Reading, and in ‘Phenomenality and Materiality in Kant’ (in Aesthetic Ideology), he also develops a notion of performative rhetoric. Unsurprisingly, this is as unorthodox as his treatment of tropology. In ‘Excuses (Confessions)’, he explains that ‘the linguistic model cannot be reduced to a mere system of tropes’ because ‘performative rhetoric and cognitive rhetoric, the rhetoric of tropes, fail to converge. The chain of substitutions functions next to another, differently structure [sic] system that exists independently of referential determination, in a system that is both entirely arbitrary and entirely repeatable, like a grammar’ (Allegories, p. 300). As de Man implies in his essay on Kant (‘From the pseudocognition of tropes, language has to expand to the activity of performance, something of which language has been known to be capable well before Austin reminded us of it’ – Aesthetic, p. 79), the term performative is generally associated with the work of the ‘ordinary language philosopher’ J. L. Austin, who first applied the concept to utterances which do not only express a proposition but perform an action. The most noticeable of these contain a first-person, present simple, active verb, as in ritual speech acts like ‘I hereby name this ship the Titanic’ or ‘I declare this conference open’, or, in everyday life, ‘I promise’ or ‘I warn you that . . . ’, but the performative verb may equally be deleted, as when umpires utter ‘First service’, ‘Wide’, or ‘Ball’. Austin first opposes performatives to constatives, but he slowly demolishes (or deconstructs) the second category, first warning that it might break down, then suggesting that it 131

132

Paradigms of Reading

does not finally seem to be so different from that of performatives, next adding the qualifier ‘supposed’ constatives, before finally abandoning the concept altogether. After all, once ‘I state that . . . ’ is recognised as a performative, the class of ‘constatives’ collapses and it is seen that all utterances are in fact performatives. Hence Austin’s title, How to Do Things with Words. In Austin’s terminology, all performative utterances, which is to say all utterances, possess illocutionary force, which is the act performed in saying something (to state, promise, warn, and so on), and perlocutionary force, the act performed by saying something, if the utterance produces a specific non-linguistic effect on the interlocutor, such as pleasing, persuading or frightening. When people use language, it does something. De Man, on the contrary, develops a new conception of performance such that language is capable of performing randomly, mechanically, non-referentially and non-cognitively, entirely beyond anybody’s will or control. Like theories of the unconscious, this clearly complicates the notion of responsible speaking human subjects. Yet de Man’s transformation of language into a machine in ‘Excuses (Confessions)’ derives from four highly questionable tropings of Rousseau. He begins by rewriting Rousseau to make him imply that one should never confess or apologise. He then elaborates on Rousseau’s claim that when he once falsely accused someone of a theft he only uttered her name because it was ‘the first thing that offered itself’, to argue that the utterance was thus non-referential, meaningless and entirely innocuous, but falsely interpreted referentially by people who failed to realise that language can posit whatever its grammar allows it to say. He states that Rousseau in the Fourth Rêverie du promeneur solitaire describes such harmless nonreferential utterances as fictions, even though Rousseau explicitly states that he is only referring to fables, tales and novels. He suggests that without a non-signifying, non-referential moment, ‘no such thing as a text is conceivable’, and then asserts that ‘it is always possible to face up to any experience (or excuse any guilt), because the experience always exists simultaneously as fictional discourse and as empirical event and it is never possible to decide which one of the two possibilities is the right one. The indecision makes it possible to excuse the bleakest of crimes because, as a fiction, it escapes from the constraints of guilt and innocence’ (Allegories, p. 293). De Man couples this with a perversely literal reading of Rousseau’s claim that another lie he told was the automatic result (‘l’effet machinal’) of his embarrassment, to conclude that texts operate as machines, driven by ‘the arbitrary power play of the signifier’ which ‘from the point of view of the subject . . . can

Mechanical Performatives

133

only be experienced as a dismemberment, a beheading or a castration’ (p. 296). In this chapter I will take issue with de Man’s reading of Rousseau, before going on briefly to discuss his wartime journalism, because today it is scarcely possible to read de Man’s attempted abolition of the concept of guilt without considering these posthumously rediscovered pieces.

The purloined ribbon Although there are literary theorists who treat all texts, literary and otherwise, as utterances or speech acts or performatives (see, e.g., Pratt, 1977) – despite Austin’s insistence that speech act theory must exclude literary language – there are texts that more obviously (are intended to) perform social or political acts than others. The concluding chapters of Allegories of Reading are about texts that legislate, promise, confess, and excuse, namely Rousseau’s Social Contract and Confessions. In ‘Excuses (Confessions)’, de Man considers Rousseau’s confession of his theft of an old and worthless pink and silver coloured ribbon while he was employed as a servant by the Comtesse de Vercellis in Turin. When accused of the theft, the sixteen-year-old Rousseau in turn accused Marion, a kitchen maid, of having given it to him. This led to both servants being dismissed. Rousseau tells this story at the end of Book II of his Confessions, and refers to it again in the Fourth Rêverie du promeneur solitaire, written ten years later.1 Rousseau states in the Confessions that his intention is not merely to confess (to reveal a truth) but to apologise or excuse himself: ‘je ne remplirois pas le but de ce livre si je n’exposois en même temps mes dispositions intérieures, et que je craignisse de m’excuser en ce qui est conforme à la vérité’. De Man translates this as ‘I would not fulfil the purpose of this book if I did not reveal my inner sentiments as well, and if I did not fear to excuse myself by means of what conforms to the truth’, adding in brackets ‘que je [ne] craignisse de m’excuser en ce qui est conforme à la vérité’. Which is to say that by adding the negative, de Man turns Rousseau’s statement that the confessor should not fear to excuse himself into one that says that it is imperative that he does fear to excuse himself, before correctly quoting Rousseau’s sentence eight pages later (Allegories, pp. 280, 288) – in an essay that explicitly mentions ‘dubious translations, literary falsifications, textual distortions’ (p. 295).2 De Man thus reads the negative in ‘si je n’exposois’ as carrying over to the following clause, although he could have avoided adding

134

Paradigms of Reading

the conspicuous ‘ne’ by translating the sentence as ‘if I did not reveal my sentiments as well, or fear to excuse . . . ’. The English verb to excuse has the senses of both apology and justification: to offer an apology, to make or accept an excuse, to seek (or agree) to justify an action or extenuate a fault, to seek (or grant) exemption, release, forgiveness, pardon, indulgence, exculpation, and so on. Of course for deconstructionist readers all these meanings are always, if not exactly present, at least potential or shimmering every time the word is used. But Rousseau’s claim that his book would not fulfil its purpose if he feared to excuse himself, and indeed the reflexive verb s’excuser in most contexts (and dictionaries) surely involves the meaning of apology rather than (or more than) justification. If Rousseau did indeed believe all apologies to be disguised justifications, de Man’s reading – that the confessor must indeed fear to excuse himself – would be quite right. As de Man writes, The only thing one has to fear from the excuse is that it will indeed exculpate the confessor, thus making the confession (and the confessional text) redundant as it originates. Qui s’accuse s’excuse; this sounds convincing and convenient enough, but, in terms of absolute truth, it ruins the seriousness of any confessional discourse by making it self-destructive. 3 (Allegories, p. 280) Yet however convincing and convenient this is, the French proverbial expression, as de Man obviously knew, is the other way round – qui s’excuse s’accuse. And if Rousseau had intended to argue that his confessions were indeed excuses in the sense of justifications and therefore worthless, he would presumably have added the negative to craindre (fear) himself, rather than leave de Man to do it. He could also have called his book Excuses or Justifications. Rousseau seems genuinely to believe that he does not fear to confess and to apologise, even though his act of confession appears not to have had its desired effect, as he returned to the story of the ribbon in the Fourth Promenade. However, as de Man stresses, many of his apologies do indeed turn out to be excuses after all. (In the case of the ribbon, Rousseau claims that the presence of so many people got the better of his repentance; that if they had given him time to come back to his senses, or if M. de la Roque had taken him on one side, he would undoubtedly have told the whole truth, but they intimidated him when they should have given him courage; and besides, he was still a child.) De Man translates and

Mechanical Performatives

135

quotes the passage in which Marcel Raymond, co-editor of the Pléiade Oeuvres Complètes, notes that Rousseau here moves from confession to justification: ‘By revealing his “inner feelings” which were good . . . it appears that after having stigmatized his misdeed he gradually begins to justify it. The same gliding and swerving motion can be observed more than once in the confessions, especially when Rousseau accounts for the abandonment of his children. He is always led to distinguish the intent from the act’ (Allegories, p. 282). 4 Yet whether Rousseau was concerned to apologise or to offer excuses, de Man points out that we can never know if apologies (or, as he insists on calling them, excuses) are sincere. He elaborates on Rousseau’s distinction between a confession as a revelation of truth, and a confession as an excuse or apology (for example, as an expression of an ‘inner feeling’ of remorse). Referential evidence of a theft (in this case, the ribbon itself) is ‘at least in theory’ literally available, and confessions can or ought to be verified in relation to extra-linguistic events. The evidence for inner feelings, on the contrary, can only be verbal: no ‘possibility of verification exists for the excuse, which is verbal in its utterance, in its effect, and in its authority’ (Allegories, p. 281). Philosophy has traditionally distinguished between phenomenal and physical properties, because with the former there can be no appearance–reality distinction. You can always be mistaken in attributing physical properties but not phenomenal ones – you know when you have a pain, or a terrible feeling of guilt or remorse, for example. Yet convincing others returns you to the physical realm: excuses are performative utterances designed to convince, but we will not necessarily be able to convince anyone of our wish to apologise or to excuse ourselves, and can never know whether other people’s apologies are genuine. Stanley Fish has also illustrated this in two separate essays in which he describes a New Yorker cartoon in which an angry woman is standing next to a seated, television-watching man, saying ‘You look sorry, you act sorry, you say you’re sorry, but you’re not sorry.’ No matter what the man says, for the woman his utterances take on an illocutionary (and perlocutionary) force different from the one he would have wished. The couple share a background, and all the presumed advantages of direct communication, but the woman’s experience of this background, and her resultant acts of interpretation, do not accord with the man’s. Fish concludes that physical and temporal proximity change nothing: all attempts to fix meaning and illocutionary force depend on assumption and belief; no performative is assured of success, and the element of risk is constitutive of all cases.5

136

Paradigms of Reading

Rousseau describes his accusation of Marion forty years earlier as a crime that has caused him unbearable remorse. He claims that the desire to assuage his conscience regarding this shameful episode, which he had never previously revealed to anyone, greatly contributed to his resolve to write his confessions. Yet whereas Marcel Raymond states that ‘la sincérité de l’auteur des Confessions paraît évidente, et on ne peut guère douter de sa véracité quand il dit que l’image de Marion vient troubler ses insomnies’ (Rousseau, 1959, pp. 1272–3), de Man calls Rousseau’s selection of this episode ‘as arbitrary as it is suspicious’ (Allegories, p. 279). He suggests that it is not certain, and indeed that it ‘doesn’t matter much, for our purpose, whether the guilt truly relates to this particular act or if the act is merely made to substitute for another, worse crime or humiliation’ or even ‘a whole series of crimes’ (p. 282). Rousseau’s accusation of Marion was all the more shameful because he had planned to give her the ribbon, which, at what de Man calls ‘this still elementary level of understanding’ (Allegories, p. 284), becomes a substitute, a figure, a metaphor for the girl. Echoing Lacan’s (1966) account of Poe’s stolen letter, de Man writes that ‘Once it is removed from its legitimate owner, the ribbon, being in itself devoid of meaning and function, can circulate symbolically as a pure signifier’ (p. 283). But for de Man, ‘the text does not stay confined within this pattern of desire. . . . Another form of desire than the desire of possession is operative in the latter part of the story, which also bears the main performative burden of the excuse and in which the crime is no longer that of theft’ (pp. 284–5). Although Rousseau claims to have slandered Marion because he feared shame more than death, more than the crime, more than anything in the world, and he wished that he could have sunk to the centre of the earth, de Man argues that ‘the obvious satisfaction in the tone and the eloquence’ of this passage betray Rousseau’s true desire: ‘What Rousseau really wanted is neither the ribbon nor Marion, but the public scene of exposure which he actually gets. The fact that he made no attempt to conceal the evidence confirms this’ (p. 285). Rousseau’s crime was not a theft motivated by greed or love or lust, but a desire for shame and exposure. Rousseau is thus, as Shelley’s Rousseau describes himself at one point in ‘The Triumph of Life’ (lines 394–5) ‘as one between desire and shame / Suspended’ (Shelley, 1965, vol. IV, p. 180). The significantly twice-repeated excuse actually acts as a recapitulation of the exposure. The unfortunate Marion ‘was destroyed, not for the sake of Rousseau’s saving face, nor for the sake of his desire for her, but merely in order to provide him with a stage on which to parade

Mechanical Performatives

137

his disgrace or, what amounts to the same thing, to furnish him with a good ending for Book II of his Confessions’ (Allegories, p. 286). This interpretation of Rousseau’s theft might loosely be called psychoanalytic, but although de Man describes the episode as ‘a truly primal scene of lie and deception’ (Allegories, p. 278), and explicitly mentions Freud – ‘like Freud’s dreams of nakedness, shame is primarily exhibitionistic’ (p. 285) – he offers a tropological explanation in terms of figural rhetoric, metaphor, allegory, symbolic substitution and displacement, without any apparent reference to Lacan’s use of these terms. He focuses on ‘the other sentence in which the verb “excuser” is explicitly being used, again in a somewhat unusual construction; the oddity of “que je craignisse de m’excuser” [here quoted correctly] is repeated in the even more unusual locution: “Je m’excusai sur le premier objet qui s’offrit” (“I excused myself upon the first thing that offered itself”), as one would say “je me vengeai” or “je m’acharnai sur le premier objet qui s’offrit”’ (p. 288). Strictly speaking, in both these sentences Rousseau is not explicitly using the verb excuser at all, but the reflexive verb s’excuser; yet although I have insisted that this verb should generally be translated as to apologise, at this point of his confessions Rousseau is clearly not apologising, but merely relating what happened, and using Marion’s proximity at the moment in question as an excuse. He writes, ‘viciousness was never further from me than at this cruel moment, and when I accused the hapless girl, it is bizarre but true that my friendship for her was the cause. She was present to my thoughts, I excused myself on the first thing that offered itself’ (p. 288) De Man quotes from Marcel Raymond’s commentary, which in turn quotes Ramon Fernandez, who writes that ‘he accuses her as if he leaned on a piece of furniture to avoid falling’. But Raymond describes this as being also ‘an almost dreamlike moment dictated by an unconscious which suddenly feels itself accused and by which he transfers the “misdeed” upon the other, on his nearby partner’ (Rousseau, 1959, p. 1273; Allegories, p. 288n). Thus Rousseau’s explanation appears to be ‘Freudian’: as de Man puts it, ‘Because Rousseau desires Marion, she haunts his mind and her name is pronounced almost unconsciously, as if it were a slip, a segment of the discourse of the other’ (Allegories, p. 288). De Man, however, plays down the psychoanalytic reading in favour of a rhetorical one. He claims that Rousseau, when he writes ‘Elle étoit présente à ma pensée, je m’excusai sur le prèmier objet qui s’offrit’, is in fact arguing that ‘Marion just happened to be the first thing that came to mind; any other name, any other word, any other sound or noise

138

Paradigms of Reading

could have done just as well and Marion’s entry into the discourse is a mere effect of chance’ (Allegories, p. 288). And ‘if her nominal presence is a mere coincidence, then we are entering an entirely different system in which such terms as desire, shame, guilt, exposure and repression no longer have any place’ (p. 289). De Man insists that ‘one should resist all temptation to give any significance whatever to the sound “Marion” . . . The estrangement between subject and utterance is then so radical that it escapes any mode of comprehension’. He also points out, in a congenial flash of wit, that ‘when everything else fails, one can always plead insanity’ (p. 289)! Thus de Man’s earlier incidental mention of Freud would seem to have been merely a preparatory rhetorical move bringing into play a well-known argument for the limits of individual responsibility, before he enacts a remarkable displacement, swerving from the inutility of excusing oneself to the non-existence of guilt.6

Excuses, fictions and machines De Man turns to the Fourth Promenade, and argues that here ‘random lies’ such as that of the Marion episode, with their ‘absence of referential signification’, are renamed fictions, as Rousseau states that ‘to lie without intent and without harm to oneself or to others is not to lie: it is not a lie but a fiction’ (Rousseau, 1959, p. 1029). According to this logic, when he uttered the sound ‘Marion’, ‘Rousseau was making whatever noise happened to come into his head; he was saying nothing at all, least of all someone’s name’ (Allegories, p. 292). De Man offers a pithy definition of the concept of fiction, such that it ‘has nothing to do with representation but is the absence of any link between utterance and a referent, regardless of whether this link be causal, encoded, or governed by any other conceivable relationship that could lend itself to systematization’ (p. 292), adding that ‘the radical irresponsibility of fiction is, in a way, so obvious, that it seems hardly necessary to caution against its misreading’ (p. 293). Yet he also points out that ‘its assertion, within the story of the Confessions, appears paradoxical and far-fetched to the point of absurdity’ (p. 293), since Rousseau’s denunciation of Marion did cause her a great deal of harm, hence Rousseau’s guilt and his numerous autobiographical and confessional writings. Indeed, Rousseau writes in the Fourth Promenade that ‘whatever is contrary to truth and hurts justice in any conceivable way is a lie’, and ‘the absence of any purposefully harmful intent does not suffice to make a lie innocent; one must also be assured that the error one inflicts upon one’s

Mechanical Performatives

139

interlocutor can in no conceivable way harm him or anyone else’ (de Man’s translation, Allegories, p. 292). This would seem to eliminate the possibility of Rousseau’s uttering of the name ‘Marion’ being a harmless fiction rather than a harmful slander, and make Rousseau’s claim ‘je m’excusai sur le premier objet qui s’offrit’, as de Man concedes, ‘an anacoluthon, a foreign element that disrupts the meaning, the readability of the apologetic discourse’ (pp. 289–90). Besides, although one does not learn this from de Man’s essay, Rousseau (1959, p. 1029) makes it quite clear that the lies-which-are-in-fact-fictions to which he is referring are to be found in fables, tales and novels, and not in real-life investigations into thefts. Yet notwithstanding his remarks about absurdity and anacoluthon, de Man proceeds to take his argument to the limit: If the essential non-signification of the statement had been properly interpreted, if Rousseau’s accusers had realized that Marion’s name was ‘le premier objet qui s’offrit,’ they would have understood his lack of guilt as well as Marion’s innocence. . . . Not the fiction itself is to blame for the consequences but its falsely referential reading. As a fiction, the statement is innocuous and the error harmless; it is the misguided reading of the error as theft or slander, the refusal to admit that fiction is fiction, the stubborn resistance to the ‘fact’, obvious by itself, that language can posit whatever its grammar allows it to say, which leads to the transformation of random error into injustice. 7 (Allegories, pp. 292–3) Yet ‘the “fact,” obvious by itself, that language can posit whatever its grammar allows it to say’ is only valid in the abstract, as in Benveniste’s demonstration of the mode of functioning of personal pronouns and other deictics. De Man himself, earlier in Allegories, states that ‘the notion of a language entirely freed of referential constraints is properly inconceivable. Any utterance can always be read as semantically motivated, and from the moment understanding is involved the positing of a subject or an object is unavoidable’ (Allegories, p. 49). Consequently, Rousseau’s utterance ‘Marion’ (as a reply to ‘So who gave you the ribbon?’ or whatever) was a lie rather than an error, and its referential reading was contextually inevitable rather than ‘false’. Although he claims, forty years after the event, that Marion was ‘le prèmier objet qui s’offrit’, Rousseau does not go as far as blaming his interlocutors for failing to realise that he was merely making a non-referential noise.

140

Paradigms of Reading

He clearly considered their attribution of guilt to Marion to be what S&W would call manifestly intended, and an optimally relevant interpretation. De Man’s euphemistic substitution of ‘fiction’ for ‘lie’ – his ‘legerdeman’ – shows him to be very able at doing things with words, but he is obliged to concede that the moment of non-referential fiction can never be isolated, or separated from an erroneous, referential, guilt-imputing reading: ‘It seems to be impossible to isolate the moment in which the fiction stands free of any signification; in the very moment in which it is posited, as well as in the context that it generates, it gets at once misinterpreted into a determination which is, ipso facto, overdetermined’ (Allegories, p. 293). This theory, based on an impossible moment, resembles Zeno’s paradox as adapted by Tom Stoppard in Jumpers, which demonstrates that since an arrow must travel half a given distance, and then half again, and so on to infinity, it can never reach its destination, so that Saint Sebastian must have died of fright. Yet de Man nevertheless bases his theoretical argument on the dubious hypothesis that ‘without this moment, never allowed to exist as such, no such thing as a text is conceivable’. Having demonstrated the necessity of the non-referential moment he is able to make the extraordinary assertion (already quoted above) that ‘it is always possible to face up to any experience (or excuse any guilt), because the experience always exists simultaneously as fictional discourse and as empirical event and it is never possible to decide which one of the two possibilities is the right one. The indecision makes it possible to excuse the bleakest of crimes because, as a fiction, it escapes from the constraints of guilt and innocence’ (p. 293). This is a truly startling pronouncement, which arises from the distinctively de Manian move of leaping from a particular example to a radical general claim, or rather, interpreting Rousseau extravagantly (as he also does with Hegel, Nietzsche, Wordsworth, Shelley and Kleist) in order to come up with examples with which to illustrate his pre-existing theories of language as a dual system of tropological substitution and mechanical performance. The story of the theft of a piece of ribbon leads (via a moment that can never ‘exist as such’) to the invocation of ‘the bleakest of crimes’, and the declaration that it is impossible to distinguish between fact and fiction provides in its turn an unconditional escape-route from constraints of guilt and innocence. De Man then moves on to another episode in the Fourth Promenade in which Rousseau reveals that he lied out of shame in response to a woman who asked whether he had any children. (The author of Émile abandoned all five of his children, who were immediately delivered by

Mechanical Performatives

141

the midwife to the Bureau des Enfants-Trouvés.) Of this lie, Rousseau maintains ‘it is certain that neither my judgment, nor my will dictated my reply, but that it was the automatic result [l’effet machinal] of my embarrassment’. Machinal is for most French speakers a dead metaphor meaning automatic, instinctive, involuntary, reflex, unconscious or without recourse to volition or intelligence, but de Man chooses instead to seize on the literal reading of the excuse being like a machine: By saying that the excuse is not only a fiction but also a machine one adds to the connotation of referential detachment, of gratuitous improvisation, that of the implacable repetition of a preordained pattern. Like Kleist’s marionettes, the machine is both ‘anti-grav,’ the anamorphosis of a form detached from meaning and capable of taking on any structure whatever, yet entirely ruthless in its inability to modify its own structural design for nonstructural reasons. The machine is like the grammar of the text when it is isolated from its rhetoric, the merely formal element without which no text can be generated. There can be no use of language which is not, within a certain perspective thus radically formal, i.e. mechanical, no matter how deeply this aspect may be concealed by aesthetic, formalistic delusions. (Allegories, p. 294) Consequently, ‘the possibility arises of the entirely gratuitous and irresponsible text, not just . . . as an intentional denial of paternity for the sake of self-protection, but as the radical annihilation of the metaphor of selfhood and of the will’ (Allegories, p. 296). Rousseau describes his lie about not having any children as an uncontrollable reflex produced by embarrassment and shame; de Man first makes the reflex into a machine, and then – using his familiar phrase ‘the possibility arises’ – suggests that the machine implies the eradication of the intentional subject. This annihilation is far from painless: if writing is mechanical, it ‘always includes the moment of dispossession in favor of the arbitrary power play of the signifier and from the point of view of the subject, this can only be experienced as a dismemberment, a beheading or a castration’ (Allegories, p. 296). De Man traces his lurid figures of bodily and textual mutilation, and the text as a machine, back to Rousseau, who relates two disfiguring childhood accidents in which he crushed two finger tips in a calender (a clothes press), and was hit on the head by a mallet. (In de Man’s tabloidesque paraphrase, ‘he nearly loses

142

Paradigms of Reading

a hand in the first and comes close to having his brains knocked out in the other’ (p. 297).) De Man argues that: In the general economy of the Rêverie, the machine [the calender] displaces all other significations and becomes the raison d’être of the text. Its power of suggestion reaches far beyond its illustrative purpose, especially if one bears in mind the previous characterization of unmotivated, fictional language as ‘machinal’. . . . Barely concealed by its peripheral function, the text here stages the textual machine of its own constitution and performance, its own textual allegory. The threatening element in these incidents then becomes more apparent. The text as body, with all its implications of substitutive tropes ultimately always retraceable to metaphor, is replaced by the text as machine and, in the process, it suffers the loss of illusion of meaning. The deconstruction of the figural dimension is a process that takes place independently of any desire; as such it is not unconscious but mechanical, systematic in its performance but arbitrary in its principle, like a grammar. (Allegories, p. 298) If the text is a machine, It is no longer certain that language, as excuse, exists because of a prior guilt but just as possible that since language, as a machine, performs anyway, we have to produce guilt (and all its train of psychic consequences) in order to make the excuse meaningful. Excuses generate the very guilt they exonerate, though always in excess or by default. Far from seeing language as an instrument in the service of a psychic energy, the possibility now arises that the entire construction of drives, substitutions, repressions, and representations is the aberrant, metaphorical correlative of the absolute randomness of language, prior to any figuration or meaning.8 (Allegories, p. 299) De Man’s account of the mechanical nature of language and its uncontrollable positing power conveniently leads him to the conclusion that there is a radical disjunction between a cognitive function such as guilt and a performative function such as excusing, and he stresses that ‘the resulting predicament is linguistic rather than ontological or hermeneutic’ (Allegories, p. 300). Another possibility which arises, of course, is that the person who puts forward such an extreme argument is battling

Mechanical Performatives

143

with his own drives, substitutions, repressions, and other mutations of psychic energy, deriving from a prior guilt and all its train of psychic consequences – perhaps, in this instance, de Man’s wartime journalism, to which I will shortly turn. To sum up, de Man begins by turning Rousseau’s confessions or apologies into excuses or justifications by adding a (far from pleonastic) negative to a crucial sentence. He then accepts Rousseau’s claim that once in his life, when accused of a theft, he uttered the first sound that came into his head, which just happened to be the name of the girl he desired. He goes on, extremely tendentiously, to interpret machinal as machine-like rather than instinctive, which leads him to construct a general theory in which language is mechanical and arbitrary. Language becomes a gratuitous and irresponsible power, unconstrained by human cognition or reference and detached from meaning, intentions and signifieds. Language speaks by itself, positing absolutely anything that grammar allows, yet at the same time operating with the implacable repetition of a preordained pattern. This arbitrary power play of the signifier leads to the radical annihilation of (the metaphor of) selfhood and human will, which is necessarily experienced as a dismemberment, a beheading or a castration, and makes it impossible to distinguish between fact and fiction, and consequently possible to excuse the bleakest of crimes. However this account of language, which violently severs the connection between cognition and action, is predicated on a moment in which the utterance is a fiction standing free of any signification, a moment that de Man himself says ‘seems to be impossible to isolate’ because ‘the utterance gets at once misinterpreted’. Furthermore, the reported utterance on which this premise is based is Rousseau’s claim once to have ‘excused himself upon the first thing that offered itself’, so that he was speaking randomly and mechanically, and ‘Marion’ could have meant anything or nothing at all. Yet this specific example in no way implies that all human utterances contain a moment of nonreferential arbitrariness. It seems rather more plausible (as relevance theory suggests) that people construct utterances that interpretively represent their (or someone else’s) thoughts, in which case, for the speaker, no moment during which the utterance can ‘posit whatever its grammar allows it to say’ exists. Words do frequently irrupt in our minds, sometimes simply because we store them in adjacent entries or addresses in the memory, and at other times probably as a result of idiosyncratic obsession-generating cathexes in early family life. We may indeed be lifelong slaves to a few possibly random unconscious

144

Paradigms of Reading

compulsions, and might occasionally be surprised to find ourselves blurting out words or names. But de Man explicitly describes the performative rhetoric which ‘exists independently of referential determination, in a system that is both entirely arbitrary and entirely repeatable’ as ‘not unconscious but mechanical’. For addressees, there is indeed a moment – usually infinitesimally short, though potentially lasting much longer – before they manage to construct an inferential interpretation of an utterance, by creating a context, during which the utterance could potentially posit anything at all. But once interpretation takes place, this moment is of no consequence. One might just as well say that no such thing as an utterance is conceivable without the infinitesimal moment between someone speaking and the sound waves reaching you, even if the speaker is whispering in your ear, but this inevitable time lag is of no more importance than the moment preceding coherent interpretation. Furthermore, these infinitesimal moments are potentially or normally followed by interpretation rather than misinterpretation. Language does not signify on its own – de Man rightly recognises the impossibility of making the actual expression coincide with what has to be expressed, of making die Art des Meinens correspond with das Gemeinte – so inferential interpretation is always necessary. The tiny time lag preceding interpretation in no way entails that an utterance was first a fiction, so that thereafter we cannot tell whether it is a fact or a fiction, but can, consequently, excuse the bleakest of crimes. De Man’s account of performative rhetoric in ‘Excuses (Confessions)’ is thus a sequence of dubious translations, textual distortions, non-sequiturs and illogicalities. It is a fictional discourse that does not seem to come near to describing any empirical events of speaking, writing or understanding. De Man’s reading of Rousseau’s Confessions, like his interpretations of Hegel and Nietzsche, is an example of a thematic interpretation that is only manifestly relevant to someone who already subscribes to a theory of language as a dual system of performative and cognitive rhetoric. De Man also described texts as machines in a posthumously published lecture on irony, and in an essay on Kleist’s story ‘Über das Marionettentheater’. The former draws on Fichte’s account of the absolute self, and on Friedrich Schlegel’s Lyceum Fragments and his novel Lucinde. In a passage that he suspects will have his audience rushing to the library after the lecture, de Man reveals that: There is in the middle of Lucinde a short chapter called ‘Eine Reflexion’ (A reflection), which reads like a philosophical treatise or

Mechanical Performatives

145

argument (using philosophical language which can be identified as that of Fichte), but it doesn’t take a very perverse mind, only a slightly perverse one, to see that what is actually being described is not a philosophical argument at all but is – well, how shall I put it – a reflection on the very physical questions involved in sexual intercourse. Discourse which seems to be purely philosophical can be read in a double code. (Aesthetic, p. 168) De Man concedes that ‘Eine Reflexion’ is a joke, which would suggest that Schlegel knew exactly what he was doing in this case: writing a lampoon of Fichte that can also readily be seen as an extended double entendre about the act of sex, in a wilful piece of irony that seems to be well within the author’s control (see Schlegel, 1971, pp. 118–21.) But de Man insists that this intermingling of two radically incompatible codes ‘represents a threat to all assumptions one has about what a text should be’ (Aesthetic, p. 169). The ironic author disappears and the text again becomes an implacable machine: Words have a way of saying things that are not at all what you want them to say. You are writing a splendid and coherent philosophical argument but, lo and behold, you are describing sexual intercourse. Or you are writing a fine compliment for somebody and without your knowledge, just because words have a way of doing things, it’s sheer insult and obscenity that you are really saying. There is a machine there, a text machine, an implacable determination and a total arbitrariness, unbedingter Willkür, he says [Lyceum fragment 42], which inhabits words on the level of the play of the signifier, which undoes any narrative consistency. (Aesthetic, p. 181) De Man also refers to the mechanical nature of language at the end of his essay on Kleist’s story, where he argues that ‘by the end of the tale, the word Fall has been overdetermined in a manner that stretches it from the theological Fall to the dead pendulum of the puppet’s limbs to the grammatical declension of nouns and pronouns (what we call, in English, the grammatical case)’, so that ‘any composite word that includes Fall (Beifall, Sündenfall, Rückfall or Einfall) acquires a disjunctive plurality of meanings’ (Rhetoric, p. 289). He states that ‘the memorable tropes that have the most success (Beifall) occur as mere random improvisation (Einfall) at the moment when the author has completely relinquished any control over

146

Paradigms of Reading

his meaning and has relapsed (Zurückfall) into the extreme formalization, the mechanical predictability of grammatical declensions (Fälle)’ (Aesthetic, p. 290). De Man goes on to discuss traps, because ‘Fälle, of course, also means in German “trap”, the trap which is the ultimate textual model of this and all texts’ (Rhetoric, p. 290). Yet as Theodore Ziolkowski (1987) pointed out, Fälle is only the plural of der Fall; the word for trap is der Falle, which has a different etymology. Given mistranslations such as this, and de Man’s treatment of Rousseau, Hegel and Nietzsche, one wonders exactly what kind of millennium Hillis Miller (1987, p. 58) has in mind when he states that ‘the millennium would come if all men and women became good readers in de Man’s sense’.

Paul de Man’s war In 1913, a certain Paul Deman won the first Tour of Flanders cycling race. Five years later, towards the end of the First World War, he was captured by the Germans on a secret mission in the occupied Netherlands and sentenced to death. Yet he was saved by the Armistice, and later received French, British and Belgian medals for bravery. He went on to win other major races in the 1920s. Until 1987, no-one writing about Paul de Man the deconstructionist literary critic and theorist, born in Belgium in 1919, would have thought of facetiously mentioning Paul Deman the Belgian cyclist bestowed with medals by the Allies. But then, four years after de Man’s death, Ortwin de Graef uncovered a large number of newspaper articles he had written in 1941–42 in occupied Belgium. Both the context and the content of de Man’s wartime journalism inevitably give rise to speculation as to whether his later statements about the impossibility of distinguishing between fictional discourse and empirical events, and the consequent possibility of excusing any guilt, were to a greater or lesser extent determined precisely by a belated sense of guilt about these writings. If confessing or excusing is merely a form of selfexculpation, or a repetition of the crime (or even the performance of the crime) from which it is supposed to free the person doing the excusing, and as such a performative use of language that cannot succeed, the later de Man did indeed have good reason not to confess his previous commitments. If guilt is merely the product of the misreading of the mechanised excuse making process, de Man cannot be guilty of anything. Space restrictions forbid any lengthy summary or discussion of de Man’s wartime writings, but just a handful of citations from his literary

Mechanical Performatives

147

column in the Brussels newspaper Le Soir and his articles on literature and art in the magazine Het Vlaamsche Land will reveal that he did indeed write things he ought to have felt guilty about. Among other things, de Man quoted, with seeming approval, Bertrand de Jouvenal’s argument that Fascism is ‘an extremely normal and lasting reaction in the face of the circumstance created by world politics’; stated that ‘the necessity of action which presents itself in the form of immediate collaboration imposes itself upon every objective mind’; wrote that ‘the war will have but united more intimately the two quite kindred realities which the Hitlerian soul and the German soul were from the beginning, fusing them into a single unique force. . . . Europe’s future can be anticipated only within the framework of the possibilities and exigencies of German genius’; and wrote that the ‘present revolution’ was ‘a clash between extra-European forces and the representatives of continental civilization’.9 The ‘extra-European forces’ in question were clearly the Jews, and de Man also wrote a more shocking piece, ‘Les Juifs dans la littérature actuelle’ (Le Soir, 4 March 1941). In this article, de Man criticises vulgar anti-Semitism for falsely describing all inter-war culture as degenerate and enjuivé, but immediately displaces the responsibility: this is a myth that the Jews themselves have contributed to spreading, and it is not flattering to contemporary writers to reduce them to simple imitators of an alien Jewish culture. He insists that although the condition of Europe since 1920 had been factitious and chaotic – conditions in which the Jews had played an important role – and there had been great political and economic changes, aesthetic evolution continued to obey its own laws. Contemporary forms of fiction and poetry were the normal and logical continuation of what had gone before. Thus the Jews have in fact had hardly any influence in the evolution of contemporary literature, which is ‘comforting for Western intellectuals. That they have been able to protect themselves from Jewish influence in a field as representative of culture in general as literature, proves their vitality. It would not be possible to have much hope for the future of our civilization if it had allowed itself to be invaded without resistance by a foreign force’. By retaining its character and its originality, ‘despite the Semitic interference in all aspects of European life’, Western civilisation had demonstrated that it was ‘healthy’, (or in other words, ‘uncorrupted’). All of which leads de Man to the conclusion that ‘a solution to the Jewish problem which aimed at creating a Jewish colony isolated from Europe would not have deplorable consequences for the literary life of the West, which would lose, all in all, several medicre personalities, and

148

Paradigms of Reading

continue, as in the past, to develop according to its own evolutionary laws’. Thus de Man defends modernist fiction and poetry, which have their own evolutionary logic and history. He denies what the vulgar antiSemites say about the relation of the Jewish problem to literature, rather than what they say about the Jews in general. He does not object to the repression of the Jews. Even though he describes Jews as mediocre and consequently insignificant in literary terms, he contemplates a solution to the Jewish problem and seems to mention their possible disappearance with equanimity or even acquiescence. At first glance, it seems bizarre that anyone could deny the pro-fascist and collaborationist content of de Man’s wartime writings – especially if one contrasts them with his first academic essays of the 1950s. 10 As Lindsay Waters, one of de Man’s posthumous editors, puts it, ‘The undistorted truth about his activities in 1941 and 1942 is unpleasant. De Man lent his considerable intellect to the service of a bad cause when he was a young man, and later chose not to make his activities then or his attitudes to them afterwards a matter of public record’ (in Hamacher et al., 1989, p. 397). Most other responses are either more virulent, or attempts to find excuses. On the one hand, Richard Klein writes of ‘the abyss of de Man’s moral responsibility’ and his ‘egregious act of complicity with the racist policies of the Occupant’ (in Hamacher et al., 1989, p. 285), while David H. Hirsch offers this desperately sad libel: ‘One suspects that those same scholars who were shocked to learn about de Man’s past would be equally surprised to learn about the existence of German death camps’ (Hirsch, 1991, p. 69). On the other hand, de Man’s pre-war and wartime acquaintance Edouard Colinet writes that he only wrote ‘Les juifs dans la littérature actuelle’ ‘reluctantly, fearing to lose his livelihood’, and that this was ‘a misdeed that was largely made up for by the risks he took in hiding Jewish friends’ (in Hamacher et al., 1989, pp. 431, 432). Several critics point to the unavoidable influence of de Man’s uncle Hendrik, a former socialist government minister and leader of the Belgian Workers Party, who had gradually lost faith in the prospects for revolutionary socialism and then announced his support for collaboration with Nazism, which he judged more likely to bring abut social justice than the capitalist democracy he had always rejected (along with, it would seem, democratic nonrevolutionary socialism). Other readers of de Man’s piece on the Jews claim that his intended meaning might have been far more complex. De Man himself, in a late essay on Bakhtin, refers to ‘double-talk, the necessary obliqueness of any

Mechanical Performatives

149

persecuted speech that cannot, at the risk of survival, openly say what it means to say’ (Resistance, p. 107). He quotes Leo Strauss’s summary, in Persecution and the Art of Writing, of the features of such a use of language: ‘obscurity of the plan, contradictions, pseudonyms, inexact repetitions of earlier statements, strange expressions, etc.’. In Responses, Ian Balfour argues that ‘the contradictory impulses of [‘Les Juifs dans la littérature actuelle’] – the hypothesis of a Europe free of Jewish writers together with an attack on vulgar anti-Semitism and a canonization of Kafka – suggest a scene of writing not unlike those analyzed by Strauss in the light of the hermeneutic proposed for the reading of philosophical texts written under political pressure’ (in Hamacher et al., 1989, p. 8).11 Timothy Bahti argues that since there is evidence that de Man only agreed to write the piece on the Jews ‘under the threat of not being allowed to continue to publish in Le Soir . . . historical circumstances suggest . . . that the article in question is weirdly encoded in and determined by a tactics and perhaps a strategy of wartime publication in occupied Belgium and France’ (in Hamacher et al., 1989, p. 1). This may be true, but the reading strategy involving dissecting de Man’s necessary obliqueness would not have been manifest to the majority of Le Soir’s 275,000 readers. 12 Bahti also suggests that the article on the Jews ‘is specifically an ironic rewriting and rereading’ of an earlier article de Man had written for a student newspaper, and that this ‘would probably have been known by de Man’s friends’. This is an even smaller category than readers wise to the wiles of persecuted writers. In short, as de Graef has latterly put it, most responses to de Man’s wartime writings tend to be ‘overdetermined by the rhetoric of outright denunciation or circuitous exculpation’ (de Graef, 1993, p. 13). Either one asserts de Man’s guilt, laments his failure to come to terms with his past, and sees his argument in ‘Excuses (Confessions)’ as conveniently selfserving, or one tries to find mitigating factors that might lessen the guilt. One exception to this pattern is Wlad Godzich, another of de Man’s posthumous editors, who argues that standing by one’s statements and actions is an impossibility given that ‘language is a conventional system of artificial signs governed by rules proper, and internal, to the system’ (Godzich, 1996, p. 138). Thus: The modern subject . . . must bear in mind that to speak is to submit to the rules of language, to construct, to fictionalize, while at the same time telling the truth by establishing the referentiality of the world of the worldview. This is indeed the modern predicament that so fascinated de Man: the modern subject must both remember and

150

Paradigms of Reading

forget the artifactness of language so as not to be a dupe of his or her own constructions and yet not renounce all pretensions to the cognition of the world and to the communication of this cognition. (Godzich, 1996, pp. 138–9) Godzich distinguishes three different ‘solutions’ to this problem. The first is mere disillusion: the subject discovers that the truth he or she believed in is a verbal construct, or a metaphor, or a falsehood, and recoils at this discovery. This moment ‘ushers in a form of radical nihilism that leaves the subject unable to stand by any utterance that he or she may make except those of a demystifying kind’ (p. 139), such as Nietzsche’s claim that ‘truths are illusions about which it has been forgotten that they are illusions’ (Nietzsche, 1989, p. 250). As Godzich points out, this sort of nihilism is pervasive in de Man’s writings of the late 1960s and early 1970s. The second solution draws upon the decisive experience of the modern subject, who has come to recognize the central role of language in relation to cognitive processes and has begun to come to terms with the fundamentally falsifying nature of language. The subject accepts as true the falseness of language, and makes this truth her own, thereby constructing a new equation in the form of a paradoxical proposition in which she opposes a local and punctual truth to a larger falsehood: I know that what I am saying is false in an absolute sense (that it will not stand up to examination once the constructed nature of language is brought into consideration) but it is nevertheless true in a punctual sense. (Godzich, 1996, pp. 139–40) This is the position of American pragmatism. But Godzich posits the desirability of a ‘third moment’, following the recognition of the constructed nature of language and the impossibility of absolute truth, in which ‘the modern subject shies away from the violence presupposed by the regime of belief, and gives up belief itself’ (p. 141). Accepting the falsehood of all utterances leads to the incapacity, posited in de Man’s later writings, to stand by our actions or statements. Yet he does not establish the necessity of making this move beyond pragmatism into the acceptance of the falsehood of all utterances, but merely reports that this is the direction de Man took in his final work. Godzich mentions the essay on The German Ideology that de Man’s illness and early death

Mechanical Performatives

151

did not allow him to write, and suggests that ‘The best prevention against the inveiglement of ideology is the practice of reading, in which the calculable well-formedness of various logics is constantly being fractured by a pervasive textuality’ (p. 143). The de Manian subject emerges in the act of reading itself, and ‘far from establishing his or her autonomy through the mastery of the text or the imposition of meaning on it, discovers the radical otherness of this text and his or her unredeemable indebtedness to it’ (p. 143). Yet the very existence of a myriad of readings of different texts is equally a prevention against the inveiglement of ideology. The existence of multiple readings and interpretations can indeed be interpreted as evidence of ‘pervasive textuality’, but it can equally be interpreted as a result of human activity. The pragmatist proposal, common to William James, Dewey and Rorty, to replace the foundational concept of truth with that of belief, and to accept the contingency of language and history, does not accept ‘the fundamentally falsifying nature of language’, but simply posits the local, contingent nature of all individual human vocabularies and descriptions. It does not worry about descriptions being ‘false in an absolute sense . . . once the constructed nature of language is brought into consideration’, because the constructed nature of language is all we have. The fact that words do not coincide with the world does not necessarily entail that language is false, or that no utterance can come into being without a non-referential moment, or that all utterances are at one level ‘fictions’. It merely reveals that people are able to do things with language.13 In the following chapter I will defend the essentially human nature of language against the general post-Saussurean notion that language signifies by itself, and that texts are active, while human beings dissolve into various impersonal linguistic systems.

9 The Madness of Words and the Enunciating Subject

I wrote again on my tabletop, scrawled the symbols for the interrelationship between matter and energy as it was understood in my day: E = Mc2 It was a flawed equation, as far as I was concerned. There should have been an ‘A’ in there somewhere for Awareness – without which the ‘E’ and the ‘M’ and the ‘c’, which was a mathematical constant, could not exist. (Kurt Vonnegut, Breakfast of Champions, 1973) In describing language as ‘mechanical’, de Man situates himself in a fairly broad tradition deriving from Saussure, according to which it is language itself, rather than language users, that is active. In this chapter I will argue that Saussurean linguistics needs to be replaced by a model that recognises the active nature of linguistic production and understanding, and suggest that de Man’s particular extrapolation from Saussure depends on an unwarranted slippage from the uncontested fact that the signifiers in any particular language are arbitrary to the highly contestable notion that signification itself (the use of signs by specific language users in specific, conventional, pragmatic contexts) constitutes an arbitrary power play. De Man writes of the positing power of language, and invokes ‘the madness of words’ which ‘no degree of knowledge can ever stop’ (Rhetoric, p. 122), but even though poetic language, which uses words for their sound as well as their sense, is evidently dependent on the arbitrary phonetic properties of words in a given language, these phonetic possibilities are still used or rejected by particular language users. I end this chapter with a further example of 152

The Madness of Words 153

the active role of the speaking subject, by considering the various ways in which writers can report and paraphrase the ideas of others, which is something that de Man did on most of the pages he ever wrote.

The fallacy of the active sign Most contemporary accounts of linguistic understanding, including relevance theory, stress its extremely active nature. Hearers and readers are constantly decoding signs, selecting semantic representations, and carrying out a variety of inferential tasks such as assigning propositional forms, identifying referents, resolving ambivalences and ambiguities, enriching vague, incomplete and elliptical phrases, interpreting tropes, and determining propositional attitudes. Indeed, when you stop to think about it, you sometimes wonder how real-time understanding is possible. De Man once recalled ‘trying to drive down a Swiss street after having just read, in a local newspaper, that for every 100 metres one drives one has at least thirty-six decisions to make’, with the result that he had ‘never been able to drive gracefully since’ (Rhetoric, p. 277).1 Understanding spoken language is a far more complex activity than driving. Post-Saussurean linguistics, on the contrary, rests on an implausible notion of signification, in which hearers and readers are quite simply passive. In his Cours de linguistique générale, Saussure stressed that he wanted to avoid psychology (and sociology and anthropology), and to model language in purely linguistic terms. Consequently, he described a system in which language simply signifies, without any activity required on the part of the addressee. Language is just something that we have, which ‘exists in the form of a sum of impressions deposited in the brain of each member of a community, almost like a dictionary of which identical copies have been distributed to each individual’ (Saussure, 1974, p. 19). Communication is assumed to be automatic: speakers use this shared, identical language, and receivers understand passively. Once sounds reach the ear, activity stops: a verbal stimulus simply ‘imprints’ itself on the brain. Words emit meanings that are identical for all receivers. 2 Signifiers automatically evoke signifieds. Physical sounds (or differences in physical sounds) convert automatically into psychological concepts (or differences in psychological concepts). Even on the phonological level, this picture is evidently false: far from automatically ‘imprinting’ themselves, sounds often require complex inferential and interpretive acts, rather than simple deciphering. It is not simply the case that each phoneme functions in a stable system of

154

Paradigms of Reading

differences with other phonemes: many phonemes have different allophonic realisations depending on what precedes or follows them, and phonemes can further be varied in any number of native and nonnative accents. On the semantic level, Saussure can only account for individual word meanings, and not sentences or discourse. Understanding utterances requires rather more than running together a chain of imprinted meanings, and speaking and writing also involve a multitude of inferential skills. In short, Saussure attributes to language processes that can only take place in hearers and readers. This results in a false model in which language does things and texts are active: words mean, differences signify, and texts control, limit, restrain or ‘imprint’ their readers. The whole range of post-Saussurean literary theory is similarly flawed by its acceptance of this mysterious notion that language itself is active, and signifies without any human interference. As the psychoanalytic critic Norman Holland puts it (1992, p. xii), much contemporary literary theory and criticism is ‘a disease of the intellect’ which ‘rests on a disproven linguistics and a dubious psychology’. One could fill many pages with examples of critical arguments premised on the notion of the active sign, but I will only cite a few notable cases. Various French thinkers sought to ‘abolish’ the human subject in the 1960s, redefining the self as a construct of cultural and linguistic (or ideological) systems over which the individual has no control. Foucault, for example, famously replaced man with language in Les Mots et les choses, describing it as ‘comforting . . . and a source of profound relief to think that man is only a recent invention, a figure not yet two centuries old, a new wrinkle in our knowledge’, and declaring that ‘he will disappear again as soon as that knowledge has discovered a new form’ (Foucault, 1973, p. xxiii) and that ‘man is in the process of perishing as the being of language continues to shine ever brighter upon our horizon’ (p. 386). Derrida (1976, p. 50), meanwhile, insists that ‘from the moment that there is meaning there are nothing but signs. We think only in signs’, and goes on to describe signs as extremely active. For Saussure, signifieds were just negative terms in a system of differences, meaning what all the other signifieds do not mean. Derrida (1981b, p. 26) extends this concept of difference, arguing that signifiers are inscribed in a chain or system within which they refer to other concepts: ‘no element can function as a sign without referring to another element which itself is not simply present’. Thus signs do not just differ from each other, they actively refer to other signs. Furthermore, Derrida argues that signs always contain traces of other signs, which disrupt the sign to the extent of erasing or cancelling out what is said by the signifier, but equally

The Madness of Words 155

function as a supplement to the signifier. Moreover, all signs are iterable, and can be lifted from any context and grafted onto other signs in contexts that the author couldn’t have anticipated. For Derrida (1988, p. 10), the very definition of a ‘signifying form’ is that it can be ‘repeated in the absence not only of its “referent”, which is self-evident, but in the absence of a determinate signified or of the intention of actual signification, as well as of all intention in present communication’. According to this account of the sign, with its traces and supplements, and the ‘structural possibility of being weaned from the referent or from the signified (hence from communication and from its context)’, Derrida substitutes the notion of limitless, indeterminate and infinite play for Saussure’s difference. Yet this free play of signifiers, which endlessly refer to other signifiers (and their related signifieds), appears to be impersonal or extrapersonal, and to occur without any action on the part of the language user. Texts simply give off or scatter or disseminate meanings, as a lamp radiates light. Meanings seemingly impose themselves on ‘the reader’, who doesn’t so much respond or infer or create meanings or apply rules as register the active force of language or textuality. Despite the vividness of these metaphors, the fact remains that language itself is not active, and cannot shine ever brighter on our horizon, or disseminate meanings. Jonathan Culler (1981) has written that: A whole tradition of thought treats man as essentially a thinking being, a conscious subject who endows objects around him with meaning. Indeed, we often think of the meaning of an expression as what the subject or speaker ‘has in mind’. But as meaning is explained in terms of systems of signs – systems which the subject does not control – the subject is deprived of his role as source of meaning. I know a language, certainly, but since I need a linguist to tell me what it is that I know, the status and nature of the ‘I’ which knows is called into question. . . . Although [the human sciences] begin by making man an object of knowledge, these disciplines find, as their work advances, that the self is dissolved as its various functions are ascribed to impersonal systems which operate through it. (Culler, 1981, pp. 32–3) Yet this argument involves a pure non sequitur. I am not entirely sure that I understand the workings of most of my internal organs, but the fact that I would need a physician to tell me what it is that keeps me alive does not call me into question. Furthermore, as Holland points

156

Paradigms of Reading

out, analysing this passage, ‘called into question’ is not the same as dissolved, decentred, decomposed, diffracted or diffused, or all the other destructive metaphors relating to the self frequently used by poststructuralists. As Holland suggests, the entire post-Saussurean tradition is flawed by an unnecessary either/or logic: either we as individuals operate through and control language, or language operates through and controls us; either we use codes or we are an assemblage of codes; either we are individual subjects or we are the result of systems of convention; either the ego makes meaning or the chain of signifiers does; and so on. But why must we use a logic of either–or rather than both–and? Why must it be either man or language? Holland (1992) sums up the negative part of his argument as follows: When you hear someone drop the person out and announce that a text does something, reads itself, supplies a meaning or a context, invites a certain kind of reading, deconstructs itself, or at the root of it all ‘signifies’, you can be sure that you have found one weak point in the argument. At that point the critic or theorist is assuming something that is simply not true of language. (Holland, 1992, p. 219) He also points out that this fallacious argument is by no means new. The approach of the deconstructionists – whom Holland tags the ‘New Cryptics’ – is simply the mirror image of that of the New Critics, who situated objective meanings in the text, and believed that texts defined their own proper responses, requiring only that the otherwise passive reader identify the correct themes and structures and recognise the unity of the poem. Much the same applies to E. D. Hirsch’s text with a fixed meaning: once again the text does all the work, and all the reader has to do is recognise the correct intrinsic genre, which is the one willed by the author. For many of today’s poststructuralist and deconstructionist critics the text, now defined as a ‘signifying practice’ rather than a poetic unity, equally does all the work, creating its own disunity. Language still has meaning(s) in itself, and doesn’t require an active reader. The positive part of Holland’s argument is that we, as interpreters, use the codes and canons of our choice. We can choose to deconstruct or psychoanalyse, to look for themes or inconsistencies, to read for gender, colonialism, social justice, historical background or whatever. We use codes and canons like we use screwdrivers and violins, and they limit us

The Madness of Words 157

like screwdrivers and violins do: they can only do certain things. We try out different hypotheses, which the text either reinforces or defeats. But texts do not force or foresee a response, they merely provide feedback that the interpreter processes. As Holland writes, I use codes. They both enable and limit me. I cannot speak without them. I cannot speak except in their language. In that sense they ‘speak through’ me. Yet in no sense need we assume that they make the I disappear. Just the opposite. The codes demand an I to run them. (Holland, 1992, p. 169) From the perspective of relevance theory, of course, it is regrettable that Holland remains trapped in the code model, and his version of cognitive psychology would benefit greatly from an awareness of inferential processes that do not involve encoding and decoding. But nevertheless, instead of a world of active signs and texts, Holland sees ‘an endless field of human relations, persons relating to persons relating to persons through various cultural codes. It is a field very like the endless sea of textuality, but the players on my field are not texts but people. Imperfectly, flickeringly perceived, but people nevertheless’ (p. 219). Pragmatists and pragmaticians, just as much as psychologists of the self, equally perceive active people and passive texts.

From intentional subjects to inhuman language In his essays from the mid-1950s to the mid-1960s, Paul de Man posited an alienated but intentional human subject or phenomenological self, as opposed to the reified subject (or the self as text or object formed by cultural codes and discourses) posited by structuralism and poststructuralism. Yet in the first of his Gauss lectures (delivered in 1967 but only published in 1993), de Man seized on the ‘antiromantic overtones’ of structuralism. Discussing René Girard’s Mensonge romantique et vérité romanesque, he isolates the ‘central myth of romantic literature’ (Romanticism, p. 9): the notion of a truly autonomous subject possessing a privileged language. Girard, de Man says, describes the illusory autonomy of the self, which leads to a consideration of the subject in and of itself rather than in relation to other subjects and things, ‘a recurrent aberration that plagues the mind of Western man since at least the Renaissance’ (p. 6), but which is particularly strong at the beginning of the nineteenth century. De Man’s series of lectures are concerned to

158

Paradigms of Reading

explode the romantic myth of the self as the independent and generative centre of the literary work. In the essay on Girard, he applauds structuralism’s ‘astute rephrasing of the problem that develops when a consciousness gets involved in interpreting another consciousness, the basic pattern from which there can be no escape in the social sciences if there is to be such a thing’ (Romanticism, pp. 9–10), but also summarises ‘the task of structuralist literary critics’ as being ‘to eliminate the constitutive subject’ (p. 13), so that the reader of literature is no longer a consciousness interpreting another consciousness. In ‘The Rhetoric of Temporality’ (included in the second edition of Blindness and Insight), de Man turns to a more formalist concern with genres and tropes, and suggests that intention is located in the text rather than in the author’s consciousness, yet he still treats rhetorical forms as modes of consciousness, so that texts are conceived of as objective correlatives for states of awareness, and the history of tropes (symbol, allegory and irony) becomes a phenomenology of the sign. Yet in Allegories of Reading, the phenomenological concern with consciousness disappears. After his ‘rhetorical turn’, de Man ceased to write of man’s inner consciousness, arguing instead that ‘selfhood is not a substance but a figure’ (Allegories, p. 170) with an ‘epistemologically unreliable structure’ (p. 187). In one succinct passage in ‘Self (Pygmalion)’, he writes: The rhetorical resources of language, regardless of whether one considers them as mere tropes or extends them to wider patterns of persuasion, are by no means, in themselves, incompatible with selfhood. On the simplest pragmatic level, they obviously offer the self the means by which it can accomplish its own designs, either in full knowledge of its purpose or with the true intent hidden from the subject by bad faith, repression, sublimation, or whatever dynamics of consciousness one wishes to imagine. Rhetoric all too easily appears as the tool of the self, hence its pervading association, in the everyday use of the term, with persuasion, eloquence, the manipulation of the self and others. Hence also the naïvely pejorative sense in which the term is commonly used, in opposition to a literal use of language that would not allow the subject to conceal its desires. (Allegories, p. 173) Rhetoric does indeed ‘all too easily appear as a tool of the self’ in virtually all definitions prior to de Man’s, but he reverses the priority, and

The Madness of Words 159

describes the self as an effect of language. He states that although the rhetorical resources of language are not incompatible with selfhood, there remains ‘the impossibility of replacing the epistemology of figural language by that of the self’ (Allegories, p. 187). In ‘The Rhetoric of Blindness: Jacques Derrida’s Reading of Rousseau’, de Man goes further, defining literary language in terms of rhetoricity, and transferring the cognitive function from the language-using subject, whether writer or reader, to the rhetorical language itself, although of course the cognition is necessarily aberrant, as figural language is inevitably misunderstood. He elaborates on Rousseau’s claim in the Essai sur l’origine des langues that the first ever language was figural rather than literal, illustrated by the allegory (discussed above in Chapter 5) of the primitive man meeting other men, no larger than he was, but experiencing fright and inventing the conceptualization ‘giants’. But Rousseau’s argument is also stated in figural language, and de Man claims that We are entitled to generalize in working our way toward a definition by giving Rousseau exemplary value and calling ‘literary’, in the full sense of the term, any text that prefigures its own misunderstanding as the correlative of its rhetorical nature; that is, of its ‘rhetoricity’. [ . . . ] it follows from the rhetorical nature of literary language that the cognitive function resides in the language and not in the subject. (Blindness, pp. 136–7) Yet, he says, Rousseau’s text also states the necessity of making this statement itself in an indirect, figural way that knows it will be misunderstood by being taken literally. Accounting for the ‘rhetoricity’ of its own mode, the text also postulates the necessity of its own misreading. It knows and asserts that it will be misunderstood. It tells the story, the allegory of its own misunderstanding. (Blindness, p. 136) In other words literary, which is to say rhetorical, texts deconstruct themselves. As de Man put it in an interview in 1983: I assume, as a working hypothesis (as a working hypothesis, because I know better than that), that the text knows in an absolute way what it’s doing. I know this is not the case, but it is a necessary working

160

Paradigms of Reading

hypothesis that Rousseau knows at any time what he is doing and as such there is no need to deconstruct Rousseau. In a complicated way, I would hold to that statement that ‘the text deconstructs itself, is self-deconstructive’ rather than being deconstructed by a philosophical intervention from the outside of the text. (Resistance, p. 118) Yet both here, and in ‘The Rhetoric of Blindness’, there is an oscillation between Rousseau and Rousseau’s text, between Rousseau knowing what he is doing and his text knowing what it’s doing (an oscillation, moreover, that recalls the one in Georges Poulet’s theoretical writings between the self and language as originating source, which de Man, in his essay on Poulet, describes as a theoretical blindness). For those of us suspicious of the notion of the active sign or text, there seems no need to deploy the ‘working hypothesis’ that the personified rhetorical text knows and asserts that it will be misread, and in this sense already deconstructs itself. We would say instead that it is people who postulate, know, assert or tell stories and allegories, often in rhetorical language that lends itself to misinterpretation and renders cognition difficult (or, if you will, impossible or aberrant), but Stephen Greenblatt (1981, p. viii) reports de Man answering a question by saying – ‘perhaps not altogether in jest’ – that ‘if you want to talk about men, you are in the wrong field. We can only talk about letters’. In his last essays, de Man uses purely rhetorical categories, finding metonymies, catachreses, and prosopopeias everywhere, without any reference to an author’s psychology. In ‘Walter Benjamin’s “The Task of the Translator” ’, he finally and radically dissociates man from language. He takes up Benjamin’s distinction in ‘Die Aufgabe des Übersetzers’ between das Gemeinte and die Art des Meinens, and offers various translations, including ‘what is meant’, logos, vouloir dire, and ‘to mean’ for the former, and ‘the way in which language means’, lexis, dire, and ‘to say’ for the latter. He points out that Harry Zohn, the English translator of Benjamin’s essay, translates these two terms as ‘the intended object’, and the ‘mode of intention’, and suggests that this is problematical, as it presupposes that both the meaning and the way in which meaning is produced are intentional acts: But the problem is precisely that, whereas the meaning-function is certainly intentional, it is not a priori certain at all that the mode of meaning, the way in which I mean, is intentional in any way. The way in which I can try to mean is dependent upon linguistic properties

The Madness of Words 161

that are not only [not] made by me, because I depend on the language as it exists for the devices which I will be using, it is as such not made by us as historical beings, it is perhaps not even made by humans at all. To equate language with humanity . . . is in question. If language is not necessarily human – if we obey the law, if we function within language, and purely in terms of language – there can be no intent; there may be an intent of meaning, but there is no intent in the purely formal way in which we will use language independently of the sense or the meaning.3 (Resistance, p. 87) Answering questions after this lecture, de Man elaborated further, stating that although intention is necessarily semantic, it is not necessarily subjective; meaning is intentional, but intention in Husserl is precisely a critique of the notion of the subject as simply the expression of his own wishfulness. He describes the inhuman nature of language as a ‘constant problem’: That there is a nonhuman aspect of language is a perennial awareness from which we cannot escape, because language does things which are so radically out of our control that they cannot be assimilated to the human at all, against which one fights constantly. . . . Things happen in the world which cannot be accounted for in terms of the human conception of language. (Resistance, p. 101) Specifically, The inhuman is: linguistic structures, the play of linguistic tensions, linguistic events that occur, possibilities which are inherent in language – independently of any intent or any drive or any wish or any desire one might have. . . . there is, in a very radical sense, no such thing as the human. (Resistance, p. 96) However the only examples he gives, following Benjamin, of some ‘possibilities which are inherent in language’ are various connotations of the German and French words for bread – Brot and pain – which have developed over time, and which we can neither contest nor control. For example Brot necessarily evokes, for de Man, Hölderlin’s Brot und Wein,

162

Paradigms of Reading

which in French becomes pain et vin, which come included in the price of the meal in a cheap restaurant, which leads to bâtard, another word for the baguette or French loaf, and de Man finds ‘bastard’ to be a long way from Hölderlin, and from Christian connotations of bread and wine. 4 Yet this in no way renders these sets of potential meanings inhuman; it merely shows how words pick up different meanings in different cultures. It is precisely because words come with attached cultural meanings that pragmaticians and reader-oriented critics assume that speakers of utterances and writers of texts wish to communicate an informative intention to a particular context-bound and time-bound audience. Benjamin (1969, p. 69), however, begins his essay on translation by dismissing the notion that art is oriented towards a particular audience, stating that ‘in the appreciation of a work of art or an art form, consideration of the receiver never proves fruitful. . . . No poem is intended for the reader, no picture for the beholder, no symphony for the listener’. Yet it becomes clear that he has in mind a somewhat mystical or messianic account of poetic language as a kind of reine Sprache, a language of pure form, of signifiers devoid of any semantic function. Benjamin asks ‘what does a literary work “say”? What does it communicate? It “tells” very little to those who understand it. Its essential quality is not statement or the imparting of information’ (p. 69). Thus rather than transmitting content, translation ‘ultimately serves the purpose of expressing the central reciprocal relationship between languages’ (p. 72), because there is a ‘distinctive convergence’ between languages which ‘are not strangers to one another, but are, a priori and apart from all historical relationships, interrelated in what they want to express’ (p. 72). This relation lies in language’s intentions: all suprahistorical kinship of languages rests in the intention underlying each language as a whole – an intention, however, which no single language can attain by itself but which is realized only by the totality of their intentions supplementing each other: pure language. While all individual elements of foreign languages – words, sentences, structure – are mutually exclusive, these languages supplement one another in their intentions. (Benjamin, 1969, p. 74) Meaning only ever emerges ‘as pure language from the harmony of all the various modes of intention’ (p. 74). The element in translation that goes beyond the transmittal of subject matter or surface content

The Madness of Words 163

‘points the way to this region: the predestined, hitherto inaccessible realm of reconciliation and fulfilment of languages’ (p. 75). The task of the translator, then ‘consists in finding that intended effect [Intention] upon the language into which he is translating which produces in it the echo of the original’ (p. 76). A good translation ‘intends’ language as a whole, and not just the particular work in the foreign language that is its point of departure. Fidelity to the words of the original is not enough because it ‘can almost never fully reproduce the meaning they have in the original. For sense in its poetic significance is not limited to meaning, but derives from the connotations conveyed by the word chosen to express it’ (p. 78). Brot and pain, for example, ‘ “intend” the same object, but the modes of the intention are not the same’, with the result that ‘the word Brot means something different to a German than the word pain to a Frenchman’ (p. 74). If fidelity to individual words is not the secret of translation, neither is fidelity to syntax. Benjamin states that ‘A literal rendering of the syntax completely demolishes the theory of reproduction of meaning and is a direct threat to comprehensibility . . . it is self-evident how greatly fidelity in reproducing the form impedes the rendering of the sense’ (p. 78). Indeed it is. So, ‘instead of resembling the meaning of the original’ a translation must lovingly and in detail incorporate the original’s mode of signification, thus making both the original and the translation recognizable as fragments of a greater language. . . . For this very reason translation must in large measure refrain from wanting to communicate something, from rendering the sense . . . as regards the meaning, the language of a translation can – in fact, must – let itself go, so that it gives voice to the intentio of the original not as reproduction but as harmony, as a supplement to the language in which it expresses itself, as its own kind of intentio. . . . A real translation is transparent; it does not cover the original, does not block its light, but allows the pure language, as though reinforced by its own medium, to shine upon the original all the more fully. (Benjamin, 1969, pp. 78–9) Yet despite describing a literal rendering of the syntax as a direct threat to comprehensibility, Benjamin suggests that the way to reveal the concealed and fragmentary nucleus of pure language, which ‘inhabits linguistic creations only in symbolized form’ (p. 79), and which ‘no longer means or expresses anything but is, as expressionless and

164

Paradigms of Reading

creative Word, that which is meant in all languages’ (p. 80), is ‘by a literal rendering of the syntax’ (p. 79). By way of example, he cites Hölderlin’s infamous literal translations of Sophocles. Obviously Hölderlin’s word-for-word (and therefore totally unintelligible) translations preserve Sophocles’ Wort at the cost of destroying his Satz or subject-matter, which leads Benjamin to suggest that translation calls into question the compatibility of grammar and meaning (or die Art des Meinens and das Gemeinte). This vision of pure language is clearly wholly incompatible with the pragmatic view that it is speakers rather than language that intend.5 But de Man uses Benjamin to argue that the dual impossibility of translating word-for-word and literally rendering the syntax reveals an instability, and the existence of ‘disruptions which are there in the original, but which the original managed to hide’ (Resistance, p. 98). More generally, there are disjunctions in one’s mother tongue: We think we are at ease in our own language, we feel a coziness, a familiarity, a shelter in the language we call our own, in which we think we are not alienated. What the translation reveals is that this alienation is at its strongest in our relation to our own original language, that the original language within which we are engaged is disarticulated in a way which imposes upon us a particular alienation, a particular suffering. (Resistance, p. 84) Yet it is hard to see how either Benjamin’s essay, or de Man’s reading of it, proves this point. Benjamin’s notion of reine Sprache, or pure language, was later echoed by Heidegger. In ‘Language’ (1971c, p. 192), a lecture given in 1950, Heidegger stated that the conception of speech as the audible expression and communication of human feelings and thoughts does ‘not suffice to circumscribe the nature of language’. Although ‘no one would dare to declare incorrect, let alone reject as useless, the identification of language as audible utterance of inner emotions, as human activity, as a representation by image and by concept’, these ‘correct ideas about language . . . ignore completely the oldest natural cast of language. Thus, despite their antiquity and despite their comprehensibility, they never bring us to language as language’ (p. 193). For Heidegger, ‘human speech, as the speech of mortals, is not self-subsistent. The speech of mortals rests in its relation to the speaking of language’ (p. 208). As he puts it in ‘ . . . Poetically Man Dwells . . . ’, ‘For strictly, it is language that speaks.

The Madness of Words 165

Man speaks first when, and only when, he responds to language by listening to its appeal’ (Heidegger, 1971d, p. 216). Consequently, in ‘The Nature of Language’ (1971b, p. 57), Heidegger recommends that we ‘undergo an experience with language’, and ‘let ourselves be properly concerned by the claim of language by entering into it and submitting to it’, by which he means that we should listen for the etymological, and specifically Greek, meanings of words. Unlike Benjamin, for whom pure language consisted of the combined intentions of all languages, Heidegger privileged the original Indo-European languages, arguing that the earliest words were immediate to the truth or nature of things, and therefore contained the most valid human perceptions, and realised their speakers’ authentic intentions. Thus he postulates a time long past when language was numinous, when light shone through words, a time which can now only be retrieved by examining etymologies, because words have long since become used up and worn out, and lost their incisiveness and their original illumination. Hence only if we uncover Greek etymologies will we approach the logos, and perhaps be able to hear or to stand in the light of Being. Yet relevance theory’s demonstration of how language only allows us to construct interpretive representations of our thoughts, and its illustration of the active, inferential nature of understanding, problematises Heidegger’s claim that ‘language speaks’. Once we recognise that language only interprets our thoughts, we will not look to it to provide the right words for every thought, and cannot endorse Heidegger’s statement that ‘when the issue is to put into language something which has never yet been spoken, then everything depends on whether language gives or withholds the appropriate word’ (1971b, p. 59). Nor can we accept the notion that ‘language speak[s] itself as language . . . when we cannot find the right word for something that concerns us, carries us away, oppresses or encourages us’ and that only when ‘we leave unspoken what we have in mind’ do we ‘undergo moments in which language itself has distantly and fleetingly touched us with its essential being’ (p. 59). On the contrary, language reveals itself to be a splendidly supple instrument when we are able to express something that has never been spoken, despite the absence of pre-existing words with pre-packaged meanings. Relevance theory, in short, implicitly offers an alternative to the notions that syntax and lexis constrain the possibilities of cognition, and that words ‘say’ rather than mean (see Heidegger, 1971a, p. 2). At the very least, it allows us to say that since the linguistic analysis of an utterance very much underdetermines its

166

Paradigms of Reading

interpretation, even if language speaks through man, man plays an extremely active role in interpreting what is spoken. Relevance theory is also diametrically opposed to de Man’s extrapolations from Benjamin’s account of the task of the translator, as it asserts that the mode of meaning, the way in which a person means, is wholly intentional. Far from functioning ‘within language, and purely in terms of language’, hearers and readers necessarily go beyond the words of an utterance or text, and inferentially enrich them, as utterances and written sentences come with context-related assumptions or implicatures. Whereas Benjamin states that ‘consideration of the receiver never proves fruitful’, relevance theory assumes that speakers and writers attempt to communicate particular explicatures and implicatures to a particular audience, in a particular context, and that the audience pays attention to this intention precisely in order to eliminate all the irrelevant possibilities that are inherent in language, independently of a user’s intent or desire. Translators equally have to communicate the set of assumptions the author intended to communicate, and so, just as much as a reader, are involved in the inferential recognition of implicatures. Relevance theory stresses the ordinariness of metaphorical language, and refuses to give any priority to literal or etymological meanings. It does not by any means ignore what de Man calls ‘the materiality of the letter’, but considers this to be merely the input into inferential processes.

Arbitrary signifiers and accountable authors Robert Frost’s negative definition of poetry as ‘what gets lost in translation’ is well-known. Poetry gets lost because the poetic function of language (as Jakobson called it) uses words for their metrical attributes and their sound as well as their sense. Poets obviously have to balance the precision of their thoughts and the search for le mot juste with the constraints of versification. Some poets, and some epochs, revel in highly formal verse forms, others lament that the need to find rhyming words hinders the poet from expressing his thoughts in the most accurate and economical way possible. As Boileau put it, in the seventeenth century, with a flawless rhythm and rhyme: Maudit soit le premier dont la verve insensée Dans les bornes d’un vers renferma sa pensée, Et donnant à ses mots une étroite prison, Voulut avec la rime enchaîner la raison.

The Madness of Words 167

For Boileau, rhyme should, on the contrary, be the slave of reason: Quelque sujet qu’on traite, ou plaisant ou sublime, Que toujours le bon sens s’accorde avec la rime. L’un l’autre vainement il semblent se haïr; La rime est une esclave et ne doit qu’obéir.6 But it can of course be argued that the poet is of necessity subjugated to the demands of versification, so that the poetic use of language is in a sense ‘mechanical’ and out of the hands of the poet, who can (to a very large extent) only use the signifiers the language offers. De Man gives a concise account of this in ‘Shelley Disfigured’, in terms of ‘the bifurcation between the semantic and the non-signifying, material properties of language’: The various devices of articulation, from word to sentence formation (by means of grammar, syntax, accentuation, tone, etc.), which are made to convey meaning, and these same articulations left to themselves, independently of their signifying constraints, do not necessarily determine each other. The latent polarity implied in all classical theories of the sign allows for the relative independence of the signifier, and for its free play in relation to its signifying function. If, for instance, compelling rhyme schemes such as ‘billow’, ‘willow’, and ‘pillow’ or transformations such as ‘thread’ to ‘tread’ or ‘seed’ to ‘deed’ occur at crucial moments in the text, then the question arises whether these particularly meaningful movements of events are not being generated by random and superficial properties of the signifier rather than by the constraints of meaning. (Rhetoric, p. 114) The question does indeed arise, and it is generally agreed that in poetry (and advertising slogans) we often find ‘compelling rhyme schemes’ that are generated by the random or arbitrary phonetic properties of the signifier rather than constraints of meaning. Shelley’s triplet billow, willow and pillow (in ‘The Triumph of Life’, lines 362–6) is clearly less likely to occur in French or German poems, given the material properties of s’élever en tourbillons, saule and oreiller, and wogen, Weide and Kopfkissen. An English poet is more likely than a French one to associate Marion with carrion, while a French poet is more likely to make it rhyme or chime with carillon, but in neither case does this make the poet a mere

168

Paradigms of Reading

marionette jerked around by random and superficial properties of the signifier. Poems and their meaning may well be to a greater or lesser extent determined by phonetic accident, but the rhyming words are still chosen by the poet in the sense that he or she can reject them if not happy with some of the meanings they throw up. We can also search our memories, sometimes laboriously, for compelling rhymes. An example of the poetic function that is frequently cited by translation theorists is Christian Morgenstern’s brief poem ‘The Aesthetic Weasel’, which begins Ein Wiesel sass auf einem Kiesel inmitten Bachgeriesel and which is probably better translated ‘A weasel / perched on an easel / within a patch of teasel’ than the literal ‘A weasel / sat on a pebble / in the midst of a ripple of a brook’, given that the poem continues Das raffinierte Tier Tat’s um des Reimes Willen.

[The cunning animal did it for the sake of the rhyme]

If you don’t like this translation (I had to look up teasel in the dictionary) you can translate much more freely. Morgenstern’s English translator Max Knight suggested various alternatives, including A mink / sipping a drink / in a kitchen sink A hyena / playing a concertina / in an arena A lizard / shaking its gizzard / in a blizzard.7 I have, for fun, tried this out in a Swiss collège class with a not-atypical mix of mother tongues, and they rapidly produced, and disputed the relative merits of, dozens of versions in French, German, Italian, Spanish, Portuguese, Serbian, Albanian and English. This looked to me like a fairly creative exercise, with the children revealing themselves to be just as cunning linguists as the weasel, rather than a case of language operating as a machine. The consequences of the random and arbitrary nature of signifiers are seldom of tremendous importance. 8 Felicitous, phonetically-determined advertising slogans can temporarily increase the sales of one product at

The Madness of Words 169

the expense of another, and thereby contribute to making or breaking a company, but their effect is rarely greater than that. Political slogans too have their uses, and Jakobson (1981) cites the slogan ‘I Like Ike’, but it was not this alone that put or kept Eisenhower in the White House. Poems, meanwhile, usually have less momentous effects than the choice of American Presidents, on an unknown number of individual readers. Yet in ‘Shelley Disfigured’, de Man is concerned with much more than compelling rhyme schemes. He also discusses ‘the arbitrary element in the alignment between meaning and linguistic articulation’ (Rhetoric, p. 114) in terms of figuration. The figural potential of language is, of course, limitless: as de Man marvelled in his essay on Proust (and as is surely evident to anyone who has ever heard children gleefully talking nonsense – or indeed anyone who has ever been a child), ‘There seems to be no limit to what tropes can get away with’ (Allegories, p. 62). In the essay on ‘The Triumph of Life’ he writes, ‘it is the alignment of a signification with any principle of linguistic articulation whatsoever, sensory or not, which constitutes the figure’ (Rhetoric, p. 114), but this seems to have it backwards: we, as readers, more often take a particular ‘linguistic articulation’ and give it a figural signification, or a range of figural significations. Tropes and figural substitutions can indeed be used to say almost anything at all, but considerations of contextual relevance generally radically limit the potential significations of any particular use of figural language. It would only be true that ‘the positing power of language is both entirely arbitrary, in having a strength that cannot be reduced to necessity, and entirely inexorable in that there is no alternative to it’ (p. 116), if there were indeed a moment in which utterances were fictions standing free of any signification, so that – as in de Man’s account of Rousseau’s Confessions – one could blame Marion for anything and everything without telling a lie, indeed without signifying anything at all. De Man himself describes such a moment of non-referential arbitrariness as impossible, but he goes on to elaborate the notion of the mechanical textual machine, which requires him to leap from arbitrary in the sense of ‘due to the particular signifiers of a given language’ to arbitrary in the sense of random and beyond control. In the essay on Shelley, he chooses not to recognise any human limits to the positing power of language, and invokes ‘the madness of words’ which ‘no degree of knowledge can ever stop’ (Rhetoric, p. 122). His reading of the positing power of language and figuration in ‘The Triumph of Life’ even leads him to the conclusion that ‘nothing, whether deed, word, thought, or text, ever happens in relation, positive or negative, to anything that

170

Paradigms of Reading

proceeds, follows or exists elsewhere, but only as a random event whose power, like the power of death, is due to the randomness of its occurrence’ (p. 122). Neil Hertz (1989, pp. 99–100) has described de Man’s use of lurid figures of dismemberment, beheading, castration, maiming and drowning (in ‘Shelley Disfigured’, ‘Autobiography as De-Facement’ and ‘Wordsworth and the Victorians’, as well as in ‘Excuses’) as moments of madness, ‘that “madness of words” that de Man names as such in “Shelley Disfigured” ’, and as an ‘obsession . . . the repeated filling-in of unavoidable structures with images drawn from a limited set of anxieties’. Yet I find myself in agreement with Don Bialostosky (1992) who disagrees with Hertz, arguing instead that the various positings of machines and violent death reveal de Man exercising the verbal power he also analyses more than they show him subject to the verbal pathologies he also diagnoses. I am not convinced that his own writing should be subsumed under his account of a ‘linguistic predicament’ that necessarily and inevitably produces such effects; I hold him responsible for his misreadings and exaggerations and excesses as I would expect to be held accountable for my own. Even when we have been compelled by passion or interest or unconscious motive or language itself to produce such effects, there is nothing to stop us or others from checking and correcting them. If they irrupt into our writing [as opposed to speech], as Wordsworth presents voices as rising up against him in The Prelude, we have the choice, as he did, to decide whether or not to let them stand. 9 (Bialostosky, 1992, pp. 192–3)

Dialogism and ventriloquism A further argument for the importance of the role of the enunciating subject involves the way people report the words or ideas of others. We regularly quote and report and paraphrase or parody other people’s utterances, and despite his insistence that it is language that speaks and posits through us, independently of our intentions, drives, wishes or desires, Paul de Man devotes the majority of his writings to quoting and paraphrasing other writers. Virtually every one of his essays devotes more space to reports and paraphrases of the words and ideas of poets, novelists, critics, theorists or philosophers than to de Man’s own theoretical pronouncements. And despite his reading of Hegel as saying ‘I cannot say what I make mine’, de Man almost invariably ‘makes’ his

The Madness of Words 171

own version of the writers he discusses. I have already shown how de Man occasionally doctors his quotes, but a far more elaborate analysis of de Man’s strategy has been proposed by Mark Currie. In ‘The Voices of Paul de Man’, Currie (1993, p. 183) diagnoses a form of wilful ‘ventriloquy’ at work in de Man. He argues that his ‘analytical prose is deeply dialogic, in the sense that it is a composite of voices, cited, reported, paraphrased, conjectured, which are often difficult to distinguish from the critic’s own commentary and judgement . . . the effect is astounding, since it allows for the formulation of key linguistic tenets upon which the radical reputation of deconstruction rests without de Man taking personal responsibility for that formulation’. 10 Currie finds this procedure crucial: It would be hard to overstate the importance of this evasiveness. In my opinion de Man owes his place in the history of ideas to the intriguing (sometimes irritating) blend he achieves in his prose between his own authoritative propositional voice and that which apparently mediates some metalingual allegory enacted by other texts. Responsibility for the key tenets of deconstruction – the autotelism of language, the inability of language to refer, the figurality of all language, and the self-deconstructing nature of literary texts – is therefore ambiguously divided between de Man and the texts that he reads. . . . De Man makes the texts of Nietzsche, Rousseau, Rilke and Proust speak sometimes in citation, but more often in a mode akin to free indirect speech. The texts seem to speak for themselves, but in fact they are subtly construed, translated and transformed by de Man’s own voice. (Currie, 1993, pp. 183–4) Yet Currie qualifies his description of de Man’s prose as ‘deeply dialogic’ by describing this ventriloquy as ‘a kind of artifice – a single voice presented falsely as a dialogue’ (p. 184). He identifies ‘the mode akin to free indirect speech’ as the even freer (i.e. less faithful to the verbal structure of the original) ‘narrative report of speech act’, a term from Leech and Short’s Style in Fiction. As Currie points out, the identification and analysis of categories of speech and thought presentation are more common in the stylistic criticism of narrative fiction than in criticism of analytical prose, because while novelists are free to present fictional information as they wish, ‘a critic who represents object texts in a mode akin to free indirect speech operates under an obligation to represent with some degree of fidelity the texts that he or she purports

172

Paradigms of Reading

to ventriloquise’ – but of course ‘for de Man the difference between literature and criticism is delusive’ (p. 184). Direct speech in fiction (‘My dear Mr Bennet . . . have you heard that Netherfield Park is let at last?’) creates the illusion of unmediated access to characters. Indirect speech (Mr Bennet replied that he had not) is more obviously mediated and embedded in a narrative sentence. Free indirect speech (He hadn’t heard) reports the content of an utterance rather than the actual words. Citation in criticism is similar to direct speech, although of course a direct quotation is also mediated through selection and recontextualisation, and by being embedded in the critical commentary. Paraphrase and summary are like indirect speech, allowing for a greater transformation of the original text. Currie shows how de Man blends quotation, paraphrase and commentary (as novelists blend direct, indirect and free indirect speech), and argues that by doing so he is able to disguise the extent of his own intervention into his source texts, and, like a ventriloquist, put words into the mouths of the authors he draws on. Currie gives examples from de Man’s discussions of Proust, Nietzsche and Rousseau in Allegories of Reading. I will expand on one of his examples – part of de Man’s treatment of Rousseau’s Second Discourse, discussed above in Chapter 5. Here are two paragraphs from de Man, with the sentences numbered: (1) Animals have no history because they are unable to perform the specifically linguistic act of conceptualization. (2) But how does conceptualization work, according to Rousseau? (3) The text yields information on this point, though not in a simple and straightforward way. (4) It describes conceptualization as substituting one verbal utterance (at the simplest level, a common noun) for another on the basis of a resemblance that hides differences which permitted the existence of entities in the first place. (5) The natural world is a world of pure contiguity: ‘all individual entities appear in isolation to the mind [of primitive man], as they are in the picture of nature. If one oak tree was called A, another was called B . . . ’ (Rousseau, p. 149). (6) Within this contiguity certain resemblances appear. (7) By substituting for A and B the word ‘tree’ on the basis of certain properties that A and B have in common, we invent an abstraction under which the irreductible [sic] differences that separate A from B are subsumed. (8) The perception of these resemblances is not, in itself, a conceptualization: in the case of animals, it leads to acts that satisfy needs but that remain confined to the limits of the particular

The Madness of Words 173

action. (9) ‘When a monkey goes without hesitation from one nut to another, do we think that he has in mind a general idea of this type of fruit and that he compares his archetype of these two individual entities? Certainly not . . . ’ (Rousseau, p. 149). (10) Conceptualization does not proceed on the basis of mere perception: perception and imagination (in the guise of memory) intervene in recognizing the existence of certain similarities – an act of which animals are said to be as capable as men – but the actual process of conceptualization is verbal: ‘It is necessary to state propositions and to speak in order to have general ideas; for as soon as the imagination stops, the mind can only proceed by means of discourse’ (Rousseau, p. 150). [ . . . ] (11) The most incisive evaluations of this and similar passages are those which locate the tension within language itself by stressing that the implied polarity exists within the structure of the linguistic sign, in the distinction established by Rousseau between the denominative and the conceptual function of language. (12) The text indeed distinguishes the act of naming (tree A and tree B) which leads to the literal denomination of the proper noun, from the act of conceptualization. (13) And conceptualization, conceived as an exchange or substitution of properties on the basis of resemblance, corresponds exactly to the classical definition of metaphor as it appears in theories of rhetoric from Aristotle to Roman Jakobson. (14) The text would then, in a sense, distinguish between, on the one hand, figurative, connotative, and metaphorical language and, on the other, denominative, referential and literal language, and it would oppose the two modes antithetically to each other. 11 (Allegories, pp. 144–6) In Currie’s terminology, sentence (1) is a kind of free indirect speech (FIS) (a loose paraphrase of Rousseau). Sentences (2) and (3) are authorial commentary. Sentence (4) is indirect speech (IS), announced by the reporting verb ‘describes’. Sentence (5) is free indirect speech followed by direct speech (or a citation). Currie comments: The FIS and IS in this passage can equally be understood as ‘narrative report of speech act’ (NRSA) . . . which is a mode of speech summary less faithful than FIS and IS to the actual verbal structure of the original; and the ambiguity is amplified by the suggestion in the third sentence that Rousseau’s text autonomously ‘yields’ information without intervention from a critic – a sentence which could also be

174

Paradigms of Reading

considered as a form of NRSA. The ambiguity between intervention and objectivity is also present in the fifth sentence which apparently justifies its FIS with direct citation, surreptitiously translating the text while conveying fidelity to it. (Currie, 1993, pp. 185–6) De Man adds in a footnote that Rousseau in fact wrote, rather cryptically, ‘des premiers Instituteurs’, but ‘the meaning refers to “primitive” men as the “first” inventors who instituted language’ (Allegories, p. 145n). Currie breaks off his analysis here, but the rest of the passage is interesting. Sentence (7) is again FIS or NRSA: Rousseau does not use the expression ‘invent an abstraction’; nor does he describe the differences between trees as ‘irreductible’ (although Nietzsche suggests this of the differences between leaves in a similar passage in ‘On Truth and Lying’). Sentence (8) is a much more faithful report of Rousseau’s text, and sentence (9) is a direct quotation (in de Man’s translation). Sentence (10) is again a combination of FIS and a translated quotation. Sentence (11) is de Man’s voice, commenting on the distinction between the denominative and conceptual functions of language that he attributes to Rousseau. Sentence (12) repeats the claim made in the previous sentence, stressing that it is ‘indeed’ to be found in Rousseau’s text. Sentence (13) is again de Man’s voice, not quoting Rousseau, but equating something he finds in Rousseau with definitions of metaphor across twenty-five centuries. Sentence (14) reads the identity de Man has established between Rousseau’s account of conceptualisation and the definition of metaphor back into Rousseau’s text – ‘The text would then, in a sense, distinguish between . . . ’ – in an almost wholly free NRSA, but for the moment with the hedge ‘would, in a sense’, which is soon forgotten as de Man’s argument continues. Currie states: I do not wish to suggest that this is an exclusively de Manian composite. This is an argument with implications for academic criticism in general, my own included. But de Man does seem to specialise in a kind of sustained ambiguity which this passage illustrates: between different levels of intervention, and therefore different degrees of translation and transformation of the object text. Through the interplay of these voices de Man can disguise the extent of his own intervention and, like the ventriloquist, put words into Rousseau’s mouth. (Currie, 1993, p. 186)

The Madness of Words 175

As I outlined at greater length in Chapter 5, de Man thus quotes and paraphrases selectively in order to claim that Rousseau equates the formation of concepts and common nouns with the invention of a metaphor. He also, as mentioned above, pretends that texts are active and speak for themselves, not unlike fictional characters: for example ‘The text yields the information on this point . . . ’ (Allegories, p. 145). Currie (p. 190) thus shows that in this essay on Rousseau, and in many other essays, ‘De Man controls the response of his readers to controversial propositions . . . by attributing these propositions to his object texts, and by controlling the distance between reader and object text according to his own oscillation between uncritical proximity and critical distance to those texts.’ The very texts in which de Man attempts to demonstrate the figural and self-deconstructing and essentially inhuman nature of language, or the impossibility of saying what we think, by way of appeals to the authority of Rousseau, Hegel and Nietzsche, demonstrate the active and necessarily distorting (or interpretive) presence of the enunciator or speaking or writing subject.

10 When Lucy ceas’d to be

In a late lecture, Paul de Man made the rather extravagant claim that Walter Benjamin’s essay ‘The Task of the Translator’ is ‘very well known, both in the sense that it is very widely circulated, and in the sense that in the profession you are nobody unless you have said something about this text’ (Resistance, p. 73). A rather better candidate for such an institutional claim, it seems to me, would be Wordsworth’s short lyric ‘A slumber did my spirit seal’: A slumber did my spirit seal, I had no human fears: She seemed a thing that could not feel The touch of earthly years. No motion has she now, no force; She neither hears nor sees; Roll’d round in earth’s diurnal course With rocks and stones and trees! (Wordsworth, 1992, p. 164) This brief poem has given rise to a number of interpretations that have frequently been compared, contrasted and trumped in critical articles using the poem as a paradigmatic text with which to demonstrate theories of meaning, intention or interpretation. Among the critics who have used Wordsworth’s lyric in this way are a venerable New Critic, Cleanth Brooks; notable defenders of authors’ intentions, E. D. Hirsch, P. D. Juhl, and the neo-pragmatist double act of Knapp and Michaels; 176

When Lucy ceas’d to be

177

a Grand Old Man of Romantic criticism, M. H. Abrams; better and less known psychoanalytic critics, including Norman Holland and Richard Matlak; and the three Yale ‘boa deconstructors’, de Man, Miller and Hartman. In this chapter, I will consider these various readings from a relevance-theoretic perspective. It is, alas, probably necessary to state at the outset that this chapter will not reveal Wordsworth’s true intentions, or provide a definitive new interpretation of the poem. People who apply relevance theory to literature regularly have to correct misguided complaints that the theory does not allow one to do these things (see, e.g., Pilkington, MacMahon and Clark, 1997). All relevance theory aims to do is to offer explanations of existing readings in cognitive pragmatic terms.

Wordsworth’s use of rock and roll For a long time, the most widely advanced reading was that the poem is about the sudden shattering of a lover’s vain illusion that his beloved is immortal. His spirit sealed in a slumber, the poetic speaker ignored all mortal fears, thereby becoming less than human. At the same time, he made his beloved (in shorthand, ‘Lucy’) more than human, while also, ironically, diminishing and objectifying her by calling her an inanimate ‘thing’ which could withstand the touch of time. These related weak implicatures of ‘thing’ are activated by its juxtaposition with ‘human’ in the preceding line. Nemesis strikes when Lucy dies, but the lyric is an expression of grief recollected in tranquillity, in which present wisdom and an awareness of mortality are contrasted with previous (past tense) innocence. Then F. W. Bateson famously found ‘pantheistic magnificence’ in the poem’s last two lines: ‘Lucy is actually more alive now that she is dead, because she is now a part of the life of Nature, and not just a human “thing” ’ (Bateson, 1950, pp. 33, 80–1). This makes the poem an occasion for happiness rather than despairing resignation: Lucy’s return to the earth is an apotheosis, since eternal oneness with the world constitutes a release and a fulfilment that amply compensate the loss of motion, force, eyes and ears. In this reading, ‘things’ like rocks, stones and trees are spiritually infused, immortal and part of the earth’s natural diurnal cycle. Yet such an interpretation almost certainly also depends on an awareness of sentiments Wordsworth expresses in other poems. Bateson mentions The Prelude (1805) III, 124–6:

178

Paradigms of Reading

To every natural form, rock, fruit or flower, Even the loose stones that cover the highway, I gave a moral life . . . and one might also cite lines 94–103 of ‘Tintern Abbey’: And I have felt A presence that disturbs me with the joy Of elevated thoughts; a sense sublime Of something far more deeply interfused Whose dwelling is the light of setting suns And the round ocean, and the living air, And the blue sky, and in the mind of man, A motion and a spirit, that impels All thinking things, all objects of all thought, And rolls through all things. (Wordsworth, 1981, pp. 118–19) The practice of going to an author’s other works so as to specify the meaning (or the informative intention) of a particular poem is, of course, perfectly acceptable from a relevance-theoretic perspective: if we are familiar with Wordsworth, reading ‘A slumber’ will bring to mind ‘encyclopaedic entries’ based on his use of ‘rock’ and ‘roll’ elsewhere. E. D. Hirsch reactivated and publicised Bateson’s reading a decade later by contrasting it with Cleanth Brooks’s New Critical reading. (Brooks, in ‘Irony as a Principle of Structure’, had concentrated on the lover’s use of the word ‘thing’, to which Lucy’s death gives a bitterly ironic sense.) As outlined in Chapter 3, Hirsch distinguishes between unchanging meaning, as defined by the author’s intentions, and significance, which results from the interaction of textual meaning and a reader’s context, and he claimed that Bateson’s interpretation was more historically concrete than Brooks’s: ‘everything we know of Wordsworth’s typical attitudes during the period in which he composed the poem’ shows that ‘inconsolability and bitter irony do not belong in its horizon’ (Hirsch, 1967, p. 239). In Gricean terminology, Hirsch is claiming that the pantheistic reading is an implicature – a contextual implication that the poet manifestly intended to make manifest to the reader, by way of which the poem will achieve optimal relevance. ‘A slumber’ has since served as an example in other (more simplistic) intentionalist arguments by P. D. Juhl (1980), and by Knapp and Michaels

When Lucy ceas’d to be

179

(1986). Frequent repetition of (Hirsch’s account of) both Brooks’s tale of despair, inertness and the bitter irony of death, and Bateson’s affirmation of pantheism, have made both readings – which Hirsch declared to be incompatible – seem so familiar that some current commentators seem happy to accept them both. One might describe this as an example of Empson’s ‘ambiguity of the seventh type’ (Empson, 1972, ch. 7), in which two opposing or mutually exclusive meanings are held in the mind (of both author and reader) at once, in what has been described as ‘the literary-critical equivalent of quantum mechanics’ or the wave-particle duality (Bate, 1997, p. 315).1 Geoffrey Hartman, for one, suggests that we are not obliged to choose: ‘It does not matter whether you interpret the second stanza (especially its last line) as tending towards affirmation, or resignation, or a grief verging on bitterness. The tonal assignment of one rather than another possible meaning . . . is curiously open or besides the point’ (Hartman, 1987, p. 188). Hartman further suggests that ‘Lucy may be everyone whose spiritual or imaginative potential is unrealized; whose is not?’ (p. 190). Despite the received view that lyric poetry is a purely subjective and monological form of expression, originating in an autonomous, integrated consciousness, which attempts to reproduce the poet’s own mood in the hearer (or ‘overhearer’), this lyric clearly allows a number of readings that can be postulated as intended by the author. But can they all be relevant? As I outlined in Chapter 2, Sperber and Wilson explicitly argue in chapter 3 of Relevance that in relation to utterances, both communicator and addressee can only expect one interpretation to be optimally relevant. A speaker should use a stimulus that saves addressees the effort of accessing more than one hypothesis consistent with the principle of optimal relevance, hence the hearers’ first inferential hypothesis (the first interpretation that meets the criterion of relevance) should be the good one, obviating the need to consider a second one. On the other hand, in chapter 4, they explain stylistic and poetic effects in general in terms of weak implicatures, which give rise to a variety of contextual effects. Successful creative metaphors allow a very wide range of acceptable weak implicatures that are triggered by the writer (most literary critics, of course, would say the text), but for which the reader has to take a large part of the responsibility. A reader like Hartman, who sees no need to choose between the interpretations of ‘A slumber’ as an expression of affirmation and of inconsolability, is implicitly describing these two interpretations as parallel implicatures triggered by the author. A critic like Hirsch, however, who considers the interpretations to be incompatible, is unlikely to accept this account of

180

Paradigms of Reading

implicatures, and would be thus be more prescriptive than S&W in the determination of meaning.

Wordsworth’s murderous spirit A more intractable interpretive problem involves the third and fourth lines of the first stanza. As relevance theory points out, one of the many tasks of a hearer or reader is to attribute referents to deictics. In the case of ‘A slumber’, there is no agreement as to the referent of the ‘she’ of the third line.2 Most critics take it (her) to be Lucy, equally the subject of four related poems, but in 1965, Hugh Sykes Davies pointed out that Lucy is not named in this poem, and denied that there is a Lucy group, simply by looking at the order of the poems in the collections that Wordsworth published. In the 1800 Lyrical Ballads, ‘Three years she grew in sun and shower’ is separate from the triplet of ‘Strange fits of passion I have known’, ‘She dwelt among th’untrodden ways’ and ‘A slumber did my spirit seal’. In the 1815 Poems, ‘Strange fits’, ‘She dwelt’, and ‘I travelled among unknown men’ are printed together among ‘Poems founded on the Affections’, while ‘Three years’ and ‘A slumber’ are among ‘Poems of the Imagination’. The ‘Lucy group’ as such is thus the invention of Victorian anthologists such as Palgrave and Arnold, and consequently, Davies argues, ‘Lucy’ can hardly be assumed to be the human antecedent for ‘she’ in the first stanza of ‘A slumber’. He thus offers a new interpretation which, with the mores of a bygone time, he claims to have kept to himself for ‘some years’ because ‘the general impression it gave was of something over-ingenious, perverse, incapable of proof or disproof, and better forgotten’ (Davies, 1965, p. 135). He suggests that ‘she’ in this autonomous poem refers to the speaker’s spirit, which is in a state of trance. As in the moments that Wordsworth (in The Prelude) called ‘spots of time’, the poet has transcended the boundary between his own being and the rest of the world, and he allows the great restorative principle of nature to assert itself. Davies specifies that in Wordsworth’s verse the word ‘spirit’ is antecedent to a pronoun twenty-five times; six of these pronouns are masculine, ten neuter, and nine feminine. He also enumerates Wordsworth’s use of pronouns with ‘soul’ and ‘mind, and examines his use elsewhere of ‘slumber’ and ‘seal’. Yet this reading of ‘A slumber’, based on Wordsworth’s use of words elsewhere and his ordering of the poems that make up the so-called Lucy group, seems entirely to disregard the second stanza with its present tenses. Davies has been followed by Brian Caraher (1991), who points out in Wordsworth’s ‘Slumber’ and the Problematics of Reading that the feminine

When Lucy ceas’d to be

181

pronoun had been used to designate human and divine spirits in medieval and Renaissance religious writings, and in lyrics by Wyatt, Surrey, Sidney, Jonson, Herbert and Crashaw. Caraher draws attention to the poem’s punctuation, and claims that the colon at the end of line two demonstrates that the antecedent of ‘she’ in line three can only be ‘my spirit’: if we were to assume that it referred to something new to be clarified later, there would be a full stop. (The colon remains at the end of line two in all known versions of the poem, from a letter from Coleridge to Thomas Poole in 1799 to the 1850 Poetical Works, although Wordsworth altered other punctuation in the poem in the 1802 edition of Lyrical Ballads.)3 Thus for Caraher, ‘strictly speaking, the ascription of “Lucy” as the reference or antecedent of “she” in line three is completely unwarranted’ (although in a footnote he distinguishes between ascription and presupposition), as it requires a knowledge of the other Lucy poems and commentaries on them, which is premature and prejudicial: ‘a pronoun in one poem should not be forced to presuppose an antecedent found only in another poem’ (1991, p. 19).4 Caraher suggests that if Wordsworth had wanted to put a full stop at the end of line two he could have done, and that if he’d wanted to work the name Lucy into the metre of line three he’d have done that too, although he also allows that Wordsworth could have intended this referential confusion. Yet Caraher proposes a different interpretation, drawing on a grammatical possibility in the first line of the poem rather than the pronoun in the third. The verb form with ‘did’ allows the meaning ‘my spirit sealed my slumber’ as well as ‘a slumber sealed my spirit’.5 If the speaker’s spirit did seal the slumber, he is neither a belated, slumbrous, insensitive and imperceptive mourner of lost love, nor a passive undergoer of his own vicarious death fantasy, but rather an active, cold-blooded, inhuman agent. According to this reading, which Caraher claims supplements rather than supplants the other two, and articulates what is lacking in both, the slumber that his spirit has sealed is the (imagined) murder of a woman. The narrator is a selfinvolved, morbidly isolated solipsist, with an inhumane desire for power and control over others, who first conjures up an unearthly and insubstantial other, and then murders the object of his fantasy. (We only assume ‘she’ to be a loved one if we look for a Lucy-like referent.) He dreams the death of the other and without a trace of human fears, moves towards imaginative fulfilment of this yearning, turning her into a thing, an unearthly, disembodied, ghostly (or immortal) object at his disposal. Caraher suggests that the speaker of this interpretation of ‘A slumber’ resembles Rivers (in the 1842 version, Oswald) in Wordsworth’s

182

Paradigms of Reading

early tragedy The Borderers, with his imagination, reason (often sophistic), pride, scorn and contempt for society.6 This would make the poem a compact moment in Wordsworth’s cultural critique of the inhumanity and cruelty of modern man. Caraher insists that this Romantic death fantasy is not his own: he would prefer to read the poem as a fantasy of the speaker’s death, and, vicariously, his own, but the particularities of his reading deny him this fantasy. Unlike Brooks’ and Bateson’s interpretations of ‘inconsolability and bitter irony’ and ‘pantheistic magnificence’, it is difficult to explain Caraher’s reading in terms of weak implicatures. Relevance theory states that hearers or readers begin their interpretation by assigning a sentence a unique propositional form, which involves, among other things, identifying referents, and disambiguating the sentence by selecting one of the semantic representations permitted by the grammar. Caraher suggests that Wordsworth could have intended the referential confusion concerning ‘she’ (as either Lucy or the speaker’s spirit). Yet if Caraher’s interpretation involving the speaker’s spirit is to supplement rather than supplant the well-known earlier readings, we are obliged to accept both the alternative readings of ‘did . . . seal’ (that ‘my spirit sealed my slumber’ and that ‘a slumber sealed my spirit’). If this is not to contradict S&W’s argument that only one possible semantic representation can be relevant, and that only one inferential hypothesis should be necessary, we must allow their account of weak implicatures and poetic effects to override their account of semantic disambiguation, and generously interpret ‘poetic effect’ to include such things as ambiguous verb forms that can invert subject and object.

Lucy and Freud Caraher’s reading, unlike those of Bateson, Brooks and Davies, at least has the merit of bringing into consideration the reader’s own psyche, the subject of much current psychoanalytic criticism. In ‘Literary Interpretation and Three Phases of Psychoanalysis’, Norman Holland (1976) uses ‘A slumber’ to illustrate what he calls psychoanalysis of the unconscious, of the ego, and of the self, of which only the third takes into consideration the analytic critic’s own self. Holland offers caricatures of the types of reading typical of literary critics influenced by the first two ‘primitive’ phases. In the hands of a literary critic, the first phase, which contrasted latent and manifest content, typically ‘hurls us from poetry to anatomy, from the wordson-the-page to the depths of the unconscious’ (1976, p. 227). It would

When Lucy ceas’d to be

183

generate a reading of ‘A slumber’ concerning castration, to which ‘she’ at least was invulnerable, although unlike the sealed ‘I’ she could be penetrated. The rocks and stones, meanwhile, would have to symbolise dirt or faeces, while the trees and rocks might also symbolise a phallus – which ‘she’ has also become, but a castrated one, completely vital but completely dead. The second phase of psychoanalysis, in Holland’s account, begins with Freud’s topographical model of superego, ego and id. A literary critic informed by this perspective ‘would try to find ego strategies as they are apparently embodied in the language of a poem’ (1976, p. 228), and which in this case have transformed an unconscious fantasy, about castration or turning into a disembodied phallus, into a conscious theme about pantheism, or cosmic indifference, or death-within-life. Thus the poem embodies the ego strategy or defence of denial: the speaker denies his awareness of his lover as a mortal being in danger, as someone who can feel or move or see or hear, and turns her into a nothing, a mere rock, stone or tree. Holland proposes the same logic of denial to explain the sequences of s’s, n’s and r’s in lines 1, 5–6, and 7–8 (the tonguey s’s are soothing soporifics in the service of denial; the series of n-plus-vowels acts out negations or omissions; the r’s as sounded in the mouth grind and growl in a verbal version of the anger hidden in this brief tale), and claims that in fact, ‘literary analysis wearing the second face of psychoanalysis is very like regular formalist reading’ (p. 229). Yet this approach still implies that mental processes are embodied in the poem rather than the poet’s (or reader’s) mind. Like both regular formalist reading and first-phase psychoanalytic criticism, it is unable to account for differences in people’s reading experiences. The basic polarity of third phase psychoanalysis is no longer conscious and unconscious, or ego and non-ego, but self and non-self. Holland – outrageously – claims virtually all recent psychoanalytic theorists as third phase allies, including Lacan, Winnicott, Erikson, Rycroft, Kohut, Schafer and, in particular, Heinz Leichtenstein, from whom he has developed the notion of the identity theme (and its variations) that is our very essence and which wholly determines all our acts, perceptions and relationships (see Holland, 1985). Thus (as outlined in Chapter 4) when we interpret a poem, although we may strive for objectivity that will command the assent of other informed readers, we are also, ineluctably, being subjective, and true to our identity. There will be a conscious in-mixing of the features of the poem with the reader’s own feelings; ‘There can be as many readings as there are readers to write them. Can be and should be’ (1976, p. 233).

184

Paradigms of Reading

Holland confesses that he dehumanises people in systems and theories, and thus identifies with the speaker whose attempted denial of the abrasions of time and human relationships becomes a kind of monstrous indifference, for which Lucy’s death is a sudden retaliation. Relevance theory explains communication in terms of manifest communicative and informative intentions without so much as a parenthetical mention of the possibility of latent or unconscious intentions. Yet it also seeks to explain how human cognition involves the maximisation of relevance, and although S&W nowhere say as much, interpretations arising from a reader’s wholly determining identity theme would clearly be part of that individual’s habitual way of maximising relevance. Seen in this light, relevance theory, just as much as Holland’s notion of an identity theme, will allow as many readings as there are readers. Not all psychoanalytic critics, however, have followed Holland into the third phase. Richard E. Matlak, in ‘Wordsworth’s Lucy Poems in Psychobiographical Context’, published two years after Holland’s article, attempts ‘to provide a comprehensive psychological and biographical ambience for the Lucy poems that will dispel the mystery of their genesis and account for their intriguing peculiarities’, but without any reference to his own psyche (Matlak, 1978, p. 46). Matlak explains most of the poetry Wordsworth wrote during his five months in Goslar in the winter of 1798–99 (notably the Matthew poems, the first four Lucy lyrics including ‘A slumber’, and the first book of the two-part Prelude of 1799) entirely in terms of the poet’s state of mind. More recently, Matlak has specified how psychobiography considers both dynamic and genetic determinants, because present circumstances as well as early experiences have their part in ‘influencing fantasy and linguistic behaviour through primal desires and fears’ (Matlak, 1986, p. 147). He states that ‘the thematic emphasis’ of the Goslar works is a complex of separation, death and chastisement’ (p. 148), and that the critic’s task is to determine what anxiety stimulated this productive gloom, and what caused the anxiety. In 1798, Coleridge chose to leave behind his pregnant wife and go to study philosophy in Germany, thus bringing to an end what for Wordsworth had been an idyllic year in Alfoxden. Wordsworth elected to follow him, on the pretext of learning enough German to earn a living as a translator of German literature. But while Coleridge went first to Ratzeburg and then to the university town of Göttingen, Wordsworth could only afford to stay in the village of Goslar in the Hartz mountains, because of the financial burden of keeping his sister Dorothy. Matlak reads a letter from Wordsworth to Coleridge and diagnoses anxiety

When Lucy ceas’d to be

185

(although this would be ‘difficult for a layperson to know’) about their separation: ‘The Lucy poems reveal ambivalence toward Dorothy; the Matthew poems and part 1 of the 1799 Prelude manifest the anxiety and distress caused by Coleridge’s waning devotion’ (Matlak, 1986, p. 149). Thus the Lucy poems are – or, to quote scrupulously, ‘can be taught as’ – ‘fantasies of Dorothy’s death, written not to ward off incestuous feelings, as Bateson argued, but to be rid of her inconvenient presence’ (p. 149). ‘Strange fits of passion’ is then the result of a murderous fantasy, which arouses guilt that the poet diffuses through mourning. ‘A slumber’ is neither ‘a muted cry of distress over the frailty of the mortal vessel’ nor ‘a pantheistic affirmation of an almost fortunate return to a spiritually infused universe’, but rather ‘void of affect, indeed, unacceptably void of humane response’ (p. 149). So there we have it. According to Matlak, the true (or optimally relevant) meaning of the poem can only be found by expending a great deal of processing effort and going outside the text to the biographical information and the Collected Letters that disclose the stifling and inhibitive relationship at the time between William and Dorothy, and the boredom, melancholy and pain he suffered in a freezing German winter. This reading owes nothing to the interpreter’s psychobiography, and leaves nothing for his students and other readers to do but agree. There is no question of any reader actively maximising relevance. Matlak’s reader is referred to Freud and to works on psychology-and-literature, but there is no suggestion that literary history, or issues of textuality or tropology, or the reader’s mind or context, are of the slightest importance. The poem is finalised, and offers no scope for dialogue – unless, perhaps, one were to offer alternative psychobiographical theories, for example that it was Annette Vallon, or her daughter (Wordsworth’s illegitimate child), or his future wife Mary Hutchinson, or even Samuel Coleridge himself, rather than Dorothy, that Wordsworth wished dead.7 Neither does Matlak give any suggestion that the biographical evidence might be open to more than one interpretation. Caraher, for example, dismisses Matlak (whose name he misspells throughout), arguing instead that it was rather Coleridge who missed the Wordsworths, and that Wordsworth did want to be with his sister. He uses the same evidence – a handful of letters, largely written by Dorothy and Coleridge, and finds it odd that Dorothy should transcribe the poems that are supposed to be expressions of her brother’s fantasies of being rid of her. He points out that the evidence for Matlak’s psychobiographical context comes solely from sources and authorities other than the poems or anybody’s experience of reading them.

186

Paradigms of Reading

Lucy as metaphor Unlike the psychoanalysts who either seek to identify the real Lucy in the poet’s unconscious, or who are only interested in what ‘A slumber’ means to them as a result of their own unconscious identity, and unlike Caraher who suggests that Wordsworth imagined Lucy only so as to be able to imagine killing her, Frances Ferguson, in Wordsworth: Language as Counter-Spirit, suggests that ‘the Lucy poems figure in little a variety of problems about poetic representation and naming, ultimately calling into question the very possibility of locating an object of representation or a signified’, and ‘issue in the question, “Was Lucy ever there, even in the past being remembered?” ’ (Ferguson, 1977, p. 174). She suggests that Lucy only ever existed as a metaphor. Ferguson quotes Wordsworth’s note to ‘The Thorn’ in which he justifies repetition in terms of ‘the interest which the mind attaches to words, not only as symbols of the passion, but as things, active and efficient, which are themselves part of the passion’ (Wordsworth, 1992, p. 351), and suggests that the passion in the Lucy poems might be less for Lucy than for active and efficient words. Lucy, in the four poems that name her, and ‘A slumber’, is largely silent and unseen. (She is only addressed and given a voice in the final stanza in the 1799 manuscript version of ‘Strange fits of passion’ – ‘Her laughter light / Is ringing in my ears’ – which was deleted before publication) (Wordsworth, 1992, p. 294). She is described as a rose, a violet, a star and a fawn, rather than by human similes and metaphors, and for Ferguson these are ‘figural substitutions for Lucy which stand in for Lucy completely enough to suggest that there may be a fundamental category mistake in seeing her as a human being – she is, perhaps, a flower (or a simile or a metaphor)’ (Ferguson, 1977, p. 175). 8 Despite Hugh Sykes Davies’ denial that there is a ‘Lucy group of poems’, Ferguson points out that ‘critics have from Wordsworth’s day felt the continuities among the Lucy poems so strongly as to suggest that they are commentaries upon one another’ (p. 177). She says that these continuities emerge most clearly in their final (post-1820) ordering, in which ‘they can be seen as mutual revisions of one another. . . . Wordsworth revised without cancelling, so that all five Lucy poems emerge from the palimpsest to depict what we may call a Wordsworthian quest for a poetical object – a quest in which the adequacy of poetic language depends precisely upon the poet’s renunciation of any claims to appropriate the object of his poems’ (p. 178).

When Lucy ceas’d to be

187

The narrator of ‘Strange fits of passion’ is rather like a questing knight of romance (or mock-romance), who fears that his love is dead. ‘She dwelt among th’untrodden ways’ states what the audience of the earlier poem had to assume, that she is indeed dead. Moreover, she had been ‘A Maid whom there were none to praise / And very few to love’, and she had ‘liv’d unknown, and few could know / When Lucy ceas’d to be’, which suggests that no one had ever seen her alive (Wordsworth, 1992, p. 163). In ‘I travelled among unknown men’ the narrator realises that Lucy was to be found at home in England, while he was questing abroad. In ‘Three years she grew in sun and shower’ we learn that Lucy was a child, and thus not a suitable love partner, and that Nature took her at the age of three, leaving the poet a mere spectator. Finally, ‘A slumber’ is an epitaph to a Lucy who now seems never to have been born, which explains ‘She’ and ‘thing’ and why Lucy is not even named. She exists only in the poet’s mind with (not like) rocks and stones and trees. According to this interpretation, ‘the Lucy poems bespeak a poetics which moves steadily toward renunciation of the poet’s claims to knowledge. As the poet appears to learn more about Lucy in each successive poem, he learns increasingly that Lucy is essentially unknowable’ (Ferguson, 1977, p. 193). Furthermore, ‘self-expression becomes a conspicuously vexed enterprise as the poet keeps continually expressing the deficiencies of his previous, deluded notions; and the “reality” of the poetic object comes to seem beside the point in poems which finally assert that such “reality” can never be known, whether it exists or not’ (p. 194). Relevance theory, with its account of how utterances are merely interpretations of thoughts, and of the inferential nature of linguistic activity, attempts to show how readers are able to disregard some grammatical, semantic and tropological possibilities, and to select others, in their quest for a writer or speaker’s inferred intentions. It also describes poetic effects in terms of the generation of a wide range of weak implicatures and an unpredictable and diversified expansion of the context. It does not, of course, exclude all thematic readings which were clearly not intended by the author: just as Hirsch describes such readings as part of the poem’s significance rather than its meaning, Sperber and Wilson merely deny that they can be said to have been communicated. It is, after all, the thematic (and figural) interpretations that canonical works generally allow which make them canonical. I find Ferguson’s thematic reading very congenial, but (from my ‘cognitive environment’ – i.e. my knowledge of Wordsworth and Wordsworth criticism) I take it to be her extrapolation from the poems, doubtless influenced by her

188

Paradigms of Reading

announced friendship with Hartman, de Man and David Ferry, rather than anything explicitly intended by Wordsworth. Yet it is no less valid for that, and unlike other ‘rhetorical’ critics, Ferguson does not insist upon the necessity, inevitability or correctness of her thematic reading, instead filling her chapter with modalising phrases such as ‘suggest that . . . ’, ‘can be seen as . . . ’ and ‘what we may call. . . ’.

From beyond the grave For Paul de Man, the death in ‘A slumber’ is clearly not a cause for pantheistic celebration. Although nature, while subject to decay and mutability, endures, and escapes ‘the unimaginable touch of time’, people do not. Reviewing The Visionary Company in 1962, de Man described Harold Bloom’s ‘vitalist’ representation of mind and the external world in Wordsworth as a ‘serious misreading’. Pace Bloom, nature in Wordsworth ‘always contains an element that can be called divine precisely to the extent that it entirely escapes the possibility of being humanized. No dialogue could ever be possible between man and the profound otherness, the “it-ness” that is always a part of Wordsworth’s nature. . . . Instead of humanizing nature, Wordsworth creates increasingly dehumanized entities, a world in which it is hard to tell men and women from rocks and stones. He may indeed be “a man talking to men,” but he is talking to them about nonhuman beings’ (Writings, pp. 94–5). According to this argument, although Lucy may once have been a living thing, she is now unequivocally a dead one. But unlike Ferguson, who suggests that Lucy never actually existed and thus could not have died, de Man proposes in ‘The Rhetoric of Temporality’ that the death in ‘A slumber’ is that of the poet himself. Although he states that ‘the death alluded to is not the death of the speaker but apparently that of someone else’, he goes on to argue that for the informed reader it is clear that ‘Wordsworth is one of the few poets who can write proleptically about their own death and speak, as it were, from beyond their own graves. The “she” in the poem is in fact large enough to encompass Wordsworth as well’ (Blindness, p. 225). In his Gauss lectures in 1967, de Man had made the same claim for another Wordsworth poem, ‘The Boy of Winander’ – who in an early manuscript was identified with the poet himself. De Man argues that in the second part of this poem, Wordsworth is reflecting on and anticipating his own death. Although the poem appears retrospective – a memory of the boy, who ‘was taken from his mates, and died / In childhood, ere he was full ten years old’9 – it is in fact proleptic autobiography: the

When Lucy ceas’d to be

189

poet is writing an epitaph for himself, and ‘anticipating a future event as if it existed in the past’ (Romanticism, p. 81). Furthermore, de Man states, this temporal perspective of writing from beyond the grave ‘is characteristic for all Wordsworth’s poetry – even if it obliges us to imagine a tombstone large enough to hold the entire Prelude’ (p. 82). Thus in both ‘The Boy of Winander’ and ‘A slumber’, there is a temporal sequence in which an inauthentic past that repressed or forgot human temporality gives way to a demystified present; in both poems, ‘Wordsworth is retrospectively reflecting on a self that is proleptically bringing him into contact with his authentic temporal destiny’ (Romanticism, p. 102). More generally, de Man argues that contrary to ‘the wellestablished custom’ (p. 75), ‘the key to an understanding of Wordsworth lies in the relationship between imagination and time, not in the relationship between imagination and nature’ (p. 92). This account of Wordsworth, and the discussion of ‘A slumber’, is part of de Man’s larger project in the Gauss lectures and in ‘The Rhetoric of Temporality’ to show that the authentic romantic thought of Rousseau, Hölderlin and Wordsworth does not concern the dialectical relationship between subject and object, but rather the ‘conflict between a conception of the self seen in its authentically temporal predicament and a defensive strategy that tries to hide from this negative selfknowledge’ (Blindness, p. 208). On the level of language, this is revealed by the use of allegory, rather than, as is commonly assumed, symbol. De Man’s argument about Wordsworth’s writing ‘from beyond the grave’ begins to become persuasive through repetition, and could be reinforced by reference to poems de Man does not mention.10 Yet the hyperbolic claim that this is ‘characteristic for all Wordsworth’s poetry’ is as reductive as any reading determined by what Norman Holland calls an ‘identity theme’. If de Man could be said to have an identity theme, however, it would be less a concern with reading from beyond the grave than a concern to provide an overall thematic envelope, of whatever kind, for Wordsworth’s poetry. In a remarkable revision of ‘Time and History in Wordsworth’ in the early 1970s, de Man displaced the theme of time by that of rhetoric. The prolepsis – the temporal perspective of speaking from beyond the grave, formerly characteristic of all Wordsworth’s poetry – is re-described as metalepsis, a rhetorical substitution. ‘The Boy of Winander’, in this version of the essay, ‘does not reflect on death but on the rhetorical power of language that can make it seem as if we could anticipate the unimaginable’ (Romanticism, p. 201, n8). As Don Bialostosky (1992, p. 174) puts it, dryly, ‘the object of reflection changes, but Wordsworth again turns out to be reflecting

190

Paradigms of Reading

on what de Man is reflecting on’.11 Anyone unconvinced by de Man’s account of the rhetorical nature of language is likely to be suspicious of the way his ‘rhetorical turn’ begins by using the same material that served him so well in his earlier ‘phenomenological’ period. Once again, what is ‘manifestly relevant’ to the critic clearly depends on his prior thematic concerns.

Miller’s tale While de Man makes the whole of Wordsworth mean one thing, J. Hillis Miller (1991b), in ‘On Edge: The Crossways of Contemporary Criticism’, proposes an iconoclastic reading of ‘A slumber’ that allows it to mean just about anything. Rather than speculate as to the poet’s intentions, or analyse his own response, Miller uses the poem to illustrate deconstructionist claims about tropology and aporias. Before offering both an ‘orthodox’ paraphrase and a deconstructionist reading of ‘A slumber’, he suggests that no fewer than eighteen genuinely exclusive binary oppositions, allowing no dialectical resolution, are present in the poem. Whereas Norman Holland’s caricature of a ‘primitive’ reading predicated on early psychoanalysis restricted its hunt for symbols to a predictable narrow range – faeces and castrated phalluses – one of Miller’s oppositions ranges wide: the poem presents ‘mother as against daughter or sister, or perhaps any female family member as against some woman from outside the family, that is, mother, sister, or daughter as against mistress or wife, in short, incestuous desires against legitimate sexual feelings’ (Miller, 1991b, p. 177). Yet interpreting dialectically and searching for a synthetic key would be futile: Miller argues that it is impossible to posit a logos for ‘A slumber’ that would be ‘the poem’s source and end, its ground and meaning, its “word” or message’, because ‘something is always left over, a plus value beyond the boundaries of each such interpretation’ (p. 177). Much the same conclusion, of course, is taken for granted by a ‘third phase’ psychoanalyst such as Holland, and has been implicit since at least the appearance of Bateson’s pantheistic reading for all but those who wish to legislate in terms of the author’s conscious or unconscious meaning. Miller then proposes a paraphrase which resembles a great many others to be found in Wordsworth criticism, mentioning the speaker’s false assumption that Lucy will not grow old or die; the shift from the past to the present tense, from ignorance to knowledge, from life to death, from Lucy’s and the speaker’s seeming immortality to the perpetual ‘now’ of a universal knowledge of death; eloquently restrained grief and

When Lucy ceas’d to be

191

the calm of mature knowledge; the two senses of ‘thing’ as person and object; and so on. He adds – without mentioning de Man’s ‘Rhetoric of Temporality’ – that Wordsworth had earlier had the strong conviction of his immortality, and that the speaker confronts or knows ‘his own death by confronting the death of another’, as in ‘There was a boy’, the Matthew poems, ‘The Ruined Cottage’, and elsewhere. He suggests, however, that ‘the reader’ will find the poem odder than this too logical and oversimplified account. In a postscript added to reprints of ‘On Edge’, Miller (1991b) chides Meyer Abrams, who took issue with this essay in ‘Construing and Deconstructing’ (1989b), for failing to see that this paraphrase was meant to be ironic and to display its manifest inadequacy, in order to prepare the way for a more complete reading. ‘As someone has said’, Miller writes, ‘there should be a mark of punctuation for irony’ (p. 190). 12 Although Lucy may have achieved immortality by joining herself to the perpetual substance of earth, ‘the speaker by not dying remains excluded from that perpetual vitality. His immortality is the bad one of a permanent empty knowledge of death and a permanent impossibility of dying’ (p. 180). Miller points out that the poet’s ‘I’ is absent from the assertions of the second stanza, and suggests that as a consequence of the girl’s death, the speaker has lost his selfhood. Furthermore, he argues, ‘an obscure sexual drama is enacted in this poem’. The poet’s mother died when he was a child of eight, and ‘in the Lucy poems the possession of Lucy alive and seemingly immortal is a replacement for the lost mother. It gives him again that direct filial bond to nature he had lost with his mother’s death’ (pp. 180–1). (In The Prelude and elsewhere, Wordsworth personified nature as both mother and father, but more often as mother.) Dismissing a major preoccupation of the psychobiographers, Miller blithely writes that ‘it perhaps does not matter greatly whether the reader thinks of Lucy as a daughter or as a mistress or as an embodiment of his feelings for his sister Dorothy. What matters is the way in which her imagined death is a re-enactment of the death of the mother as described in The Prelude’ (p. 181). Lucy was a virgin ‘thing’ who seemed untouchable by earthly years, by time, by death.13 But for Miller, the ‘touch of earthly years is both a form of sexual appropriation which leaves the one who is possessed still virgin if she dies young, and at the same time it is the ultimate dispossession which is death. To be touched by earthly years is a way to be sexually penetrated while still remaining virgin’ (p. 182). The male, adult speaker of the poem is the displaced representative of both Lucy herself and of nature or death. For Miller, the poem is full of blank

192

Paradigms of Reading

contradictions which are not ironic. ‘Lucy is both the virgin child and the missing mother, that mother earth which gave birth to the speaker and has abandoned him. Male and female, however, come together in the earth, and so Lucy and the speaker are “the same”, though the poet is also the perpetually excluded difference from Lucy, an unneeded increment, like an abandoned child’ (p. 182). What this means is that ‘the poet has himself caused Lucy’s death by thinking about it. Thinking recapitulates in reverse mirror image the action of the earthly years in touching, penetrating, possessing, killing, encompassing, turning the other into oneself and therefore being left only with a corpse, an empty sign’ (p. 183). Miller swiftly deflects any possible charge that this reading is grounded in psychobiographical details of the poet’s reaction to the death of his parents, by announcing that ‘it is the other way round. Wordsworth interpreted the death of his mother according to the traditional trope identifying the earth with a maternal presence. By the time we encounter her in his writing she exists as an element in that figure. His life, like his poetry, was the working out of the consequences of this fictitious trope, or rather of the whole figurative system into which it is incorporated’ (p. 183). Moving to a higher level of generality, Miller explains that this poem, ‘in the context of the other Lucy poems and all of Wordsworth’s work, enacts one version of a constantly repeated Occidental drama of the lost sun. Lucy’s name of course means light’ (p. 183). Which is to say that her name – which of course does not actually appear in this poem – derives from the Latin root for ‘light’. 14 Miller suggests that the poem is an allegory of loss in which the speaker is not so much mourning the dead girl as ‘the lost source of light, the father sun as logos, as head power and fount of meaning’ (p. 183). Again more generally, ‘the loss of the radiance of the logos, along with the experience of the consequences of that loss, is the drama of all Wordsworth’s poetry, in particular “A Slumber Did My Spirit Seal” ’ (p. 184). Miller claims that ‘in expressing this, the poem leaves its reader with no possibility of moving through or beyond or standing outside in sovereign control. The reader is caught in an unstillable oscillation unsatisfying to the mind and incapable of being grounded in anything other than the activity of the poem itself’ (p. 182). Except, of course, that the poem does not ‘express’ any of this. Miller’s is a thematic reading that moves way beyond the eight lines and fifty-two words of the poem. Many other readers have found satisfying readings that they grounded on the poet’s biography or various of his other poems.

When Lucy ceas’d to be

193

Miller expressly argues elsewhere (1991a) that all possible etymological meanings are ‘shimmering’ in a word each time it is used, but it is surely preferable to base readings on words that actually appear in the text, as other readers cited above have done. Caraher (1991), for example, hears ‘diurnal’ as die yearn all, and points out that the metrical, alliterative and conceptual stress fall on the middle syllable. Norman Holland, on the contrary, hears ‘in “diurnal” the word “urn” as saying another way the whole earth has been made Lucy’s funerary vessel’ (1976, p. 226), while Geoffrey Hartman, in ‘Timely Utterance’, adds that ‘diurnal’ is ‘followed by “course”, a word that sounds like the archaic poetic pronunciation of “corpse”’ (1987, p. 236 n4). Miller insists that it is false to assume that a literary text has a plain sense, a fundamental level of grammatical meaning that is easily identifiable by any competent reader, to which is added a second, nonessential layer of figurative language, because, on the contrary, ‘figurative language goes all the way down. . . . All language is irreducibly and fundamentally figurative. . . . All good reading is therefore the reading of tropes at the same time as it is the construing of syntactical and grammatical patterns. Any act of reading must practice the two forms of interpretation together’ (Miller, 1991b, p. 188). Furthermore, ‘the inherence of tropes, including the trope of irony, in ordinary as well as in literary language has been known since Plato and the Greek rhetoricians’ (p. 191). Indeed it has: this argument causes no problem for a relevance theorist, and Sperber and Wilson make much the same point. Yet all interpretations of ‘A slumber’ engage with what it means to say that a spirit slumbers or is sealed (or that a spirit sealed a slumber). Miller’s interpretation in ‘On Edge’ has nothing to do with tropes turning aside plain grammatical sense, or ‘the perversion of grammar by rhetoric which deconstruction patiently demonstrates’ (p. 191). It is rather a thematic reading which digs into his ‘cognitive environment’ and draws out Wordsworth’s biography, his other poetry, a thorough acquaintance with de Man’s reading of the poem in ‘The Rhetoric of Temporality,’ a taste for Freudian symbolic inversions, displacements, and condensations, and, as Abrams points out, a desire to find everywhere blank contradictions and aporias. The obscure sexual drama, the turning of Lucy into a sexually penetrated but still virgin prepubertal girl and a replacement for the poet’s mother, and the description of her death as a death fantasy of a surrogate mother, resulting in the loss of light, do not directly derive from specific grammar-subverting tropes. Miller rejects Abrams’ suggestion that he has first construed the poem, in much the same way as many other readers from Cleanth Brooks to

194

Paradigms of Reading

Paul de Man, and then wilfully deconstructed it, by claiming that all language is figural. But the meaning of figures, just as much as the meaning of propositions, is grossly underdetermined. It would not be too difficult to duplicate Miller’s reading of ‘A slumber’ with reference to Lennon and McCartney’s account of tangerine trees, marmalade skies, a girl with kaleidoscope eyes, cellophane flowers, a bridge by a fountain, rocking horse people eating marshmallow pies, newspaper taxis, a train in a station, plasticene porters with looking-glass ties, and Lucy in the sky with diamonds. In short, I find Miller’s account of ‘A slumber’ implausible, and am inclined to agree with Abrams’ (1989a) description of this kind of deconstructive criticism as largely a mode of display oratory or epideictic rhetoric. As I have repeated throughout this book, relevance theory takes it for granted that signs do not coincide with what they signify, and that linguistic communication is not a matter of delivering determinate messages by way of semantically and grammatically coded signals. The linguistic meaning of an uttered sentence falls well short of encoding what the speaker means, and is merely the input to the inferential processes by which an addressee attempts to identify a communicator’s intentions, guided by contextual clues. Utterances are only interpretive representations of thoughts, so the semantic representations of sentences, arrived at by decoding, have to be augmented by implicatures. Hearers or readers will consider implicatures as long as they yield cognitive effects that repay the processing effort. In interpreting ‘A slumber’, Miller clearly expends a great deal of processing effort, offering an array of ‘meanings’ which all appear to be equally relevant. He does not, as S&W say hearers of ordinary speech do, automatically choose the solution involving the least effort, in search of an interpretation that the communicator could manifestly have expected to be optimally relevant. He proposes meanings which strike many of his readers as having little to do with Wordsworth, and which therefore cannot be described as implicatures, as S&W define these as assumptions which a speaker manifestly intended to make manifest. His interpretations are enabled by the fact that the nature of language invariably allows the same linguistic structure to have innumerable incompatible references and meanings. This is not because one can always add a rhetorical overlay to semantic and grammatical ‘codes’, but because language use requires inference as well as decoding. Miller insists that ‘Derrida and I are right . . . about the enigmas introduced into even the most apparently simple passage by its permeation or penetration by figurative language’ (1991b, p. 190), but Sperber and Wilson’s cognitive (and performative, and humanist) account of

When Lucy ceas’d to be

195

language shows, on the contrary, that linguistic meaning generally arises from the combination of intended metaphors or ironies with intended grammatical structures, without the one necessarily perverting the other. Relevance theory does not account for the vagaries of thematic interpretation, but it does offer a thorough account of language which is an antidote to ‘rhetoricist’ claims of the necessary truth of readings that claim merely to reveal unauthored linguistic or tropological ‘meanings’ generated by ‘the materiality of language’. Indeed, if we must invoke the truth of interpretations, those which seek to infer communicative intentions and optimally relevant meanings seem to have a stronger claim. Given Sperber and Wilson’s Gricean demonstration that the very nature of communication is such that it involves a mixture of explicatures and implicatures, one might argue that intended implicatures have a logical priority over thematic extrapolations. The poet, as Wordsworth put it in the idiom of his age, is a man speaking to men (usually figuratively). So too is the critic, who can choose either to make the attempt to infer authorial intentions in search of communicated meanings, or to ignore the author and simply interpret a text – or, as often happens, all texts – according to his own particular interests, regardless of anyone else’s intentions, perhaps making them all manifestly relevant in exactly the same way. Yet a quick comparison of the various readings of Wordsworth’s short ‘Slumber’ poem reveals that the thematic interpretations, unlike those based on what are taken to be conscious implicatures, tend to lose the specificity of the poem being read. They either seem to make the whole of Wordsworth mean one thing (even though, as with de Man, that one thing can change over time), or make a single poem mean absolutely everything. Implicatures may be unresolvable, and ‘optimally relevant’ meanings irretrievable – as the conflicting interpretations of Brooks and Bateson perhaps demonstrate – but the attempt to resolve them, and to cooperate with an author, often seems to produce the most specific, and indeed the most pertinent, readings.

11 Relevance and Rhetoric

Relevance theory describes the inferential nature of communication and cognition, and the way in which the human cognitive system tends towards processing the most relevant inputs available. I have tried to show that this is a more plausible description of language – and, specifically, of the difference between what Benjamin (1969) called das Gesagte or das Gemeinte (what is said or meant) and die Art des Sagens or des Meinens (the way of saying it, or the way in which language means) – than de Man’s account of rhetoric, which renders language uncontrollable, arbitrary, mechanical and irresponsible. Just as much as Paul de Man, linguistic pragmaticians such as Dan Sperber and Deirdre Wilson take as their starting point the fact that linguistic signs never coincide with intended or interpreted meanings. A sentence is merely an interpretive representation of a thought, and both contextual information and inferential work by a hearer or reader are required if a communicator’s intentions are to be recognised, or indeed if any interpretation at all is to be hazarded. The fact that any sentence, if thought of as an authorless string of material signifiers, taken out of context – or rather, transposed into a variety of contexts – can have a wide range of potential meanings, does not entail that language is unstable or that all understanding is necessarily aberrant, because every instance of language use and understanding takes place in a particular context. We focus our attention on what seems to be the most relevant information, and construct a context that seems to maximise relevance, which enables us to disregard the plainly irrelevant linguistic possibilities that, in the abstract, could destabilise any particular spoken utterance or written sentence. This applies equally to situations in which we are endeavouring to recognise a communicator’s informative intentions, and to those in which we only seek to maxiise the relevance of an utterance, sentence or text 196

Relevance and Rhetoric

197

according to our own interests, with no regard for the author’s postulated intentions. As a dual account of communication and cognition, relevance theory is able to account for both of these processes. Relevance theory shows how utterances can represent thoughts both descriptively and interpretively. It explains how hearers and readers have to assign a sentence or utterance a propositional form, and select a semantic representation, and complete any ellipses, and identify all referents, and enrich any vague terms, and make any accidentally ambiguous terms semantically explicit, and recognise tropes, and identify propositional attitude, and recover implicatures and weakly manifest assumptions, and so on. It thereby demonstrates the untenability of all critical approaches to literature grounded on the notion of the passive reception or understanding of an author’s meaning or message. Language cannot ‘posit’ or ‘signify’ on its own, nor operate mechanically. No less than de Man, S&W consider that all language use is or can be loose or rhetorical or non-literal, yet they argue this with specific (albeit invented) examples, rather than with appeals to the authority of Rousseau, Hegel, Schlegel, Nietzsche and others. Consequently, they have no need to offer tortuous reinterpretations and mistranslations of these authors in order to bolster their argument. S&W demonstrate that metaphor and irony are two types of interpretive use which resemble other perfectly ordinary forms of language that are not usually even considered as figurative. Comprehending metaphor and irony is as natural an inferential process as comprehending literal language. Reference assignment is one of a number of inferential processes that language users execute continuously, unconsciously and, generally, effortlessly. Relevance theory is a contemporary cognitive account of language use that has no relation to metaphysical speculations concerning what Hegel called the unhappy consciousness, and the split between subject and object, or the split between words and the world. It shows how words are used to refer to objects or processes, sometimes literally, often loosely. The metaphorical origin of many, most or all words, names and concepts need have no bearing on their subsequent use, if their users (speakers and writers, hearers and readers) are either unaware of or unconcerned by their metaphorical roots, and employ them in accordance with conventional contemporary usage. Nietzsche himself conceded that our ‘worn out metaphors’ nevertheless constitute ‘a sum of human relations’. The fact that most words and expressions potentially have a range of meanings does not mean that these meanings are all equally present at any one time, or that they necessarily render the propositional content of utterances undecidable, or undermine intended

198

Paradigms of Reading

meaning. Moreover, the fact that equivalent words in different languages pick up incompatible connotations, as de Man points out in an elaboration on Benjamin’s remarks on Brot and pain, might just as logically be adduced as evidence of the quintessentially human character of language, rather than, as de Man has it, its nonhuman or inhuman nature. All the pragmatic and inferential processes outlined by S&W take place in the composition and interpretation of both everyday spoken language and literature. Although both de Man and Derrida make claims relating to everyday language, most interesting critical disputes relate to literary meaning. Relevance theory cannot be called upon to adjudicate such disputes. Where inferred ‘informative intentions’ are contested, the critical readers concerned are generally – by definition – not part of the intended ‘authorial audience’, so there is no way of settling their claims concerning ‘manifestly intended’ implicatures and explicatures, and optimally relevant interpretations. Where readers are ‘merely’ proposing interpretations that are manifestly relevant and compelling from their particular theoretical perspective, as de Man does, the issue of communicated authorial intentions does not arise. But relevance theory does seek to account for the way such interpretations are generated. It accounts for the ways in which we express ourselves and interpret what is expressed by others. In doing so, it unquestionably undermines de Man’s claims regarding the irrefutability of readings based on his own highly dubious account of rhetoric, which sees only the words on the page but entirely disregards all the pragmatic aspects of language use. Relevance theory, which sees metaphor and irony as integral to language use, offers an alternative approach to reading and literary interpretation. Rather than postulating the impossibility of expressing what one wishes to say and the aberrancy of all cognition and understanding, as de Man does, it focuses on the potential richness of intended or interpreted meanings, including poetic effects.

Notes Chapter 1 1 I have lifted this example from Rumrich (1996, p. 6), just so as to be able to cite this book, as works of literary criticism informed by relevance theory are still all too rare. 2 In the 1960s, de Man wrote a series of essays (published in Blindness and Insight and Romanticism and Contemporary Criticism) about literary critics who concentrated on the problems and complexities of reading, in whose work he retrospectively perceived a recurrent pattern: a paradoxical discrepancy between the general statements the critics made about the nature of literature and the actual results of their interpretations. Drawing on Gadamer, he attributed this discrepancy to a necessary ‘blindness’, brought about by the critics’ historical situatedness, but he described this blindness as the source of their best insights. 3 This dual account of communication and cognition is thus designed to explain all possible interpretations, from those predicated on inferred authorial communicative intentions to those entirely based on individual readers’ strategies. Relevance theory is thus wide open to the Popperian objection that it is wholly unfalsifiable, and therefore not scientific. Yet as Duhem pointed out in 1906, there are serious drawbacks to the logic of falsification. (See McCloskey, 1985, ch. 1.) More recently, Kuhn, Lakatos, Feyerabend and Quine have all provided compelling arguments against slavishly following Popper’s version of scientific method. From a pragmatic perspective, the utility of a theory lies in what one can do with it, irrespective of its scientific status as defined by Popper. 4 De Man suggested in ‘Language and Literature: A Commentary’ (in Blindness and Insight) that a reading of a misreading can only give rise to a further misreading, but this logic seems to preclude any possibility of demonstrating the incorrectness of a reading or interpretation. 5 Rousseau (1961, p. 693). As Ortwin de Graef (1995) points out, de Man uses this phrase in ‘Criticism and Crisis’, ‘Form and Intent in the American New Criticism’, ‘The Literary Self as Origin: The Work of Georges Poulet’, and ‘The Rhetoric of Blindness: Jacques Derrida’s Reading of Rousseau’ in Blindness and Insight, in ‘Patterns of Temporality in Hölderlin’s “Wie wenn am Feiertage . . . ”’ and ‘Allegory and Irony in Baudelaire’ in Romanticism and Contemporary Criticism, and in ‘Allegory (Julie)’ in Allegories of Reading. 6 Derrida (1988, p. 159), replying to a question by Gerald Graff, claims that ‘grammatology has always been a sort of pragmatics, but the discipline which bears this name today involve [sic] too many presuppositions requiring deconstruction, very much like speech act theory, to be simply homogeneous with that which is announced in De la grammatologie. A pragrammatology (to come) would articulate in a more fruitful manner these two discourses.’ Derrida first mentioned pragrammatology in 1984, but has yet to develop it. 199

200

Notes

Chapter 2 1 The inferential nature of language thus requires almost continuous inferential activity on the part of the addressee, to fill the gap between semantic representations and thoughts. S&W’s account of this inferential activity is far more rigorous than the hermeneutic critic Wolfgang Iser’s account of reading fiction in terms of filling in blanks. Iser (1978) describes works of fiction as immanent structures consisting of multiple perspectives (essentially those of the narrator, the characters, the plot and that marked out for the reader), between which there lies a referential field of blanks that the reader has to fill in to connect the perspectives and construct a coherent and cohesive object. He describes the reader as continually deciding on the relation between the currently foregrounded element and the background, between a particular theme or perspective and the horizon of other perspectives and knowledge acquired from previous reading. S&W’s account of language gives the reader or hearer far more to do: there are gaps to be filled in every single sentence or utterance. 2 I don’t remember where I got this example, but I once tried to use it in a conference talk outlining relevance theory, and it came out as ‘“fire” in a crowded theory’, causing great hilarity among a sceptical audience who all instantly became psychoanalytic critics, convinced of the negligible significance of conscious intentions. 3 As mentioned in the previous chapter, in the first (1986) edition of Relevance, S&W proposed only one principle of relevance, the one they now call the Second (or Communicative) Principle. Yet because, they say, many readers quoted the claim that ‘human cognition tends to be geared to the maximisation of relevance’ as the principle of relevance, they renamed this the First (or Cognitive) Principle in the second (1995) edition. 4 I adapt this term from the economist Uskali Mäki (1995), who disparagingly describes Donald McCloskey’s conception of rhetoric as herrschaftsfrei (dominance-free) conversation, obeying a Habermasian Sprachethik, as an ‘angel theory of truth’. 5 This example comes from Sperber (2000, p. 132). 6 This book is largely about linguistic meaning; for a detailed relevancetheoretic account of poetic effects see Pilkington (2000). 7 The Aristotelian definition of irony of saying one thing and meaning another, or more specifically meaning the opposite, lacks the defining echoic element, and doesn’t explain why anyone would deliberately say something other than what they mean – and as Paul de Man points out (Blindness, p. 209), it is also insufficiently precise, and could equally stand as a definition of allegory. Yet S&W’s concept of derisively echoing another utterance, while certainly accounting for sarcasm, doesn’t explain the ‘cosmic’ or situational irony associated with Schlegel and Kierkegaard, and more recently Paul de Man, such that what is amiss is not other people’s utterances but human reality itself. If you fall and sprain your ankle and declare ‘Oh great, that’s nice’, you are not echoing someone who thinks sprained ankles are nice, but are being angry with the world, or with fate. The object of irony is not the real or imagined originator of an opinion you are echoing, but the incongruity or incompatibility between reality itself and one’s desire not to have a sprained ankle. See Martin (1992) and Gibbs (1994, ch. 8).

Notes 201 However S&W’s account of the ordinariness of various forms of echoic interpretation is to be welcomed. The prevalence of what Bakhtin called ‘double-voiced’ language (including parody, irony and reported speech) is usually underestimated. Bakhtin (1981, p. 339) suggested that ‘we need only keep our ears open to the speech sounding everywhere around us to reach such a conclusion: in the everyday speech of any person living in society, no less than half (on the average) of all the words uttered by him will be someone else’s words (consciously someone else’s) transmitted with varying degrees of precision and impartiality (or more precisely, partiality)’. Bakhtin further suggested that even when we are not echoing specific persons, what we say tends to derive from the community to which we belong: ‘The majority of our information and opinions is usually not communicated in direct form as our own, but with reference to some indefinite and general source: “I heard”, “It’s generally held that . . . ”, “it is thought that . . . ” and so forth’ (p. 338), which leads him to say that in fact ‘all speech is in a sense reported speech’ (p. 293). 8 Although there is necessarily an interpretive relationship between the propositional form of a metaphorical utterance and that of the thought communicated, the thought itself could be descriptive: it is possible to describe states of affairs in the world metaphorically. 9 One could add mimesis – in which an utterance ‘resembles’ something in the world – to this list. As de Man points out, like onomatopoeia, mimesis is just ‘one trope among others, language choosing to imitate a non-verbal entity just as paranomasis “imitates” a sound without any claim to identity (or reflection on difference) between the verbal and the non-verbal elements’ (Resistance, p. 10). 10 Ironically, in recent years, a sure sign that people (in particular radio and television sports commentators) are not speaking literally is when they use the intensifier ‘literally’ (or in French, littéralement), as in the marathon runner who ‘has literally hit the wall’, or the cyclist in difficulty on a climb ‘qui a littéralement explosé’ or ‘est littéralement collé à la route’.

Chapter 3 1 Anyone tempted to dismiss Sperber and Wilson as typically positivist linguists with minimal feeling for literature should immediately read Deirdre Wilson’s novel, Slave of the Passions, and as the Director of the London Institute of Behavioural Science therein recommends, take notes on threeby-five-inch file cards. 2 For example, Milman Parry (1971) demonstrated the oral traditional nature of Homeric poetry in the 1920s. 3 This point is made by Alasdair Fowler (1989). 4 It may appear unfair to quote sentences about speakers and hearers in relation to literary interpretation, but it will be seen that S&W extend their argument to writers and readers. 5 In the light of some of Foucault’s other early pronouncements (about man being ‘only a recent invention, a figure not yet two centuries old, a new wrinkle in our knowledge’ who is ‘in the process of perishing as the being of

202

Notes

language continues to shine ever brighter upon our horizon’ (Foucault, 1973, pp. xxiii and 386), this is a fairly mild statement. Foucault is merely arguing that we imagine or construct an author, either to fix meaning in terms of an informative intention or to justify our own critical practice. 6 Hirsch does at least concede that verification of authorial meaning is impossible, and that one could construe authorial meaning without even knowing it: ‘To verify is to show that a conclusion is true; to validate is to show that a conclusion is probably true on the basis of what is known’ (1974, p. 171). Furthermore, an author’s statement of intended meaning is not necessarily a reliable means of validation because it is ‘very possible to mean what one is not conscious of meaning’ (p. 22). A reader can easily discover meaning that was not conscious to the author because ‘an author almost always means more than he is aware of meaning’ (p. 48). 7 Of course if the gap is filled by tradition, the text has contributed to our horizon and so does not actually have a separate horizon of its own; the ‘fusion of horizons’ is only a metaphor, but positing a separate horizon is a necessary part of the process of understanding. It is worth noting that the notion of horizons is a horizontal rather than a vertical or foundational metaphor – Gadamer’s title is ironic: he opposes the belief that scientific Method will lead to Truth, insisting instead that foreknowledge precedes every act of cognition. Mikhail Bakhtin later independently made a very similar argument about the impossibility of abandoning one’s own context, but his notion of dialogue allows no fusion of horizons: ‘In order to understand, it is immensely important for the person who understands to be located outside the object of his or her creative understanding – in time, in space, in culture. . . . We raise new questions for a foreign culture, ones that it did not raise itself; we seek answers to our own questions in it; and the foreign culture responds to us by revealing to us its new aspects and new semantic depths. Without one’s own questions one cannot creatively understand anything other or foreign (but, of course, the questions must be serious and sincere). Such a dialogic encounter of two cultures does not result in a merging or mixing. Each retains its own unity and open totality, but they are mutually enriched’ (Bakhtin, 1987, pp. 6–7). Other critics deny that we can ever take the slightest step out of our own horizon. Stanley Fish, for example, states that ‘it seems obvious that the efforts of readers are always efforts to discern . . . an author’s intention’, so that ‘discerning an intention is no more or less than understanding’ (Fish, 1981, p. 161), but insists that the text ‘is the structure of the reader’s experience rather than any structures available on the page’ (p. 152). Yet the reader’s experience will be entirely the product of the ‘interpretive strategies’ shared by his or her ‘interpretive community’, so that, in S&W’s terms, every new phenomenon necessarily becomes manifestly relevant in exactly the same way. Unfortunately, Fish never adequately defines interpretive communities, or specifies how ‘a mind that cannot see anything outside its horizons’ (Fish, 1989, pp. 144–5) can change, even though he describes interpretive communities as assimilative and self-transforming (p. 152). 8 I have outlined Hirsch and Gadamer’s positions more fully in MacKenzie (1986).

Notes 203 9 Although Sperber does not say as much in Explaining Culture (1996), this modular process is presumably restricted to understanding repeatedly communicated, stable cultural representations. In ‘Pragmatics and Modularity’ (1991b), Wilson and Sperber insist that unlike grammar, which is a special purpose modular system, pragmatics is not a cognitive system at all. The pragmatic processes used to bridge the gap between the semantic representation of a sentence and the interpretation of an utterance are always context dependent, and therefore open to risk. Like scientific hypothesising, they are also what Fodor would call ‘informationally unencapsulated’, that is they have free access to all potential contextual and encyclopaedic information. At first sight, the modularity of thought thesis seems wholly incompatible with the recognition of human cultural diversity, which would seem to indicate that the mind is infinitely malleable. Yet Sperber suggests that organisms endowed with modular minds can indeed engender truly diverse cultural beliefs, because there exist two classes of beliefs, which achieve rationality in different ways. Intuitive beliefs owe their rationality to essentially innate, hence universal, perceptual and inferential mechanisms, and so do not vary dramatically and are essentially mutually consistent or reconcilable across cultures. Those beliefs that vary across cultures to the extent of seeming irrational from another culture’s point of view are typically reflective beliefs with a content that is partly mysterious to the believers themselves. Such beliefs are rationally held, not in virtue of their content, but in virtue of their source (they come from trusted people). I will return to this notion of reflective beliefs below. 10 The Russian word for ‘word’ – slovo – could equally be translated as ‘utterance’ or ‘discourse’. 11 There is, of course, a good reason for this similarity. Bakhtin’s American biographers, Katerina Clark and Michael Holquist, and the Russian editors of the new collected works of Bakhtin, argue that Bakhtin also wrote the books published under the name of his colleague Vološinov; Vološinov’s American editors, and Gary Saul Morson and Caryl Emerson, authors of various books on Bakhtin, disagree. But even if Vološinov did write the books published under his name, there is no question that he, like the other members of the ‘Bakhtin Circle’, was greatly influenced by Bakhtin. Like Sperber and Wilson half a century later, both Vološinov and Bakhtin stress that even if a speaker or writer exactly infers an addressee’s apperceptive background, there still remains a lot for the addressee to do, and the process of understanding will still be highly active: see Bakhtin (1981, pp. 280–1), and Bakhtin (1987, p. 68). Although Bakhtin called his theory of communication ‘dialogism’, Paul de Man quotes a similar passage in which Bakhtin writes that ‘the speaker breaks through the alien conceptual horizon of the listener, constructs his own utterance on alien territory against his, the listener’s, apperceptive background’ (Bakhtin, 1981, p. 282), and describes this as ‘a gesture of dialectical imperialism’ in which ‘there is no trace of dialogism left’ (Resistance, p. 112). De Man argues bluntly that ‘it is not a foregone conclusion whether Bakhtin’s discourse is itself dialogical or simply contradictory’. But despite the distinctly undialogical terminology that Bakhtin uses here – alien horizons and alien territory – he insists that the point of constructing an utterance

204

12

13

14

15

Notes according to the hearer’s apperceptive background is to provoke an answer. The hearer who understands is more likely to respond. Dialogism, unlike dialectic and rhetoric, claims to be interested in the hearer’s response for its own sake. Even so, this aspect of Bakhtin’s dialogism is very similar to what Kenneth Burke called identification, which he proposed as an alternative to persuasion as the key term underlying the practice of rhetoric. Burke (1969, p. 55) quotes Aristotle (who was quoting Socrates) to the effect that ‘it is not hard to praise Athenians among Athenians’. Burke writes, ‘You persuade a man only insofar as you can talk his language by speech, gesture, tonality, order, image, attitude, idea, identifying your ways with his. Persuasion by flattery is but a special case of persuasion in general’ (p. 55). For Burke, the key to rhetorical persuasion is to give signs of deferring to an audience’s opinions in order to gain their good will. To change their opinion in one respect it will be necessary to yield to their opinions in others. This is very close to Bakhtin’s account of constructing an utterance according to the listener’s ‘alien apperceptive background’. Of course, given the oral origins of the gospels, one could take this one stage further. As Walter Ong has pointed out (discussing Mark rather than Matthew), ‘essentially, what Mark had to do was to interpret. The situation had become such that the old oral kerygma [the preaching or proclamation of the gospel of the Kingdom of God], which began with Jesus and was continued by Jesus’s disciples, could no longer function effectively as it earlier had. Something had to be done to get the implications clear. The matter had to be reorganized. When Mark undertook to put the old oral heritage of stories and preaching about Jesus into writing, this was in effect what he undertook to do: to reorganize and interpret the oral kerygma so as to bring out its current relevancy’ (Ong, 1986, p. 151). De Man’s argument about the disjunction between the author’s and reader’s experience comes from ‘The Dead-End of Formalist Criticism’, an article published in 1956, long before his ‘rhetorical turn’ at the end of the 1960s. De Man in the 1950s and early 1960s was concerned with the ontological (or phenomenological, or Heideggerean) theme of subjectivity. (See Chapter 7, below.) The arguments about language and rhetoric that I contest in this book date from the last decade of de Man’s life. Sperber suggests that this process of using semi-propositional representations to salvage information one cannot represent propositionally – especially information that comes from authoritative sources such as one’s entire cultural group or especially respected elders – goes a long way towards explaining the apparently irrational beliefs reported by ethnologists, anthropologists and historians. There is no reason to assume that they are factual beliefs: indeed the very fact that when assumed to be factual these beliefs appear irrational is reason enough to assume that they are representational beliefs of a semi-propositional content. This dispenses with the need to assume cognitive relativism – the idea that other cultures inhabit different cognisable worlds, so that what appears rational in their world appears irrational in ours. Robert Crosman (1980) has pointed out that E. D. Hirsch’s arguments about validity in interpretation produce a strict hierarchy with final adjudication

Notes 205 coming from a supreme court of professors who themselves decide when they have arrived at the author’s intended meaning. This contrasts with virtually all other areas of liberal education, in which the ability to see several different points of view is valued. 16 Bakhtin (1987, p. 144) describes a similar process: ‘The first task is to understand the work as the author himself understood it, without exceeding the limits of his understanding. This is a very difficult problem and usually requires introducing an immense amount of material. The second task is to take advantage of one’s own position of temporal and cultural outsideness.’ Yet he names this process dialogism rather than a version of dialectic.

Chapter 4 1 Clearly this works better for sentences obeying a ‘finite-state’ grammar than for the more complex sentences that such a left–right choice grammar doesn’t allow. Yet S&W argue – convincingly, I feel – that relevance theory replaces the huge literature on given and new information, topic and comment, theme and rheme, presupposition and focus, presupposition and assertion, and so on (Relevance, pp. 203–17). 2 Of course, as Tillotama Rajan (1990) has pointed out, this is almost exactly the contrary of the claim Shelley makes four pages later, that ‘the most glorious poetry that has ever been communicated to the world is probably a feeble shadow of the original conception of the Poet’. If great poetry contains an infinity of meanings, there seems little reason to search for that first acorn, or to capture the poet’s vision, or to try to reignite the ‘fading coal’ that is the poetic mind in creation. 3 Bakhtin also suggests that all language users (in everyday speech and literary creation alike) imagine a ‘superaddressee’ who will fully understand them. He writes that ‘any utterance always has an addressee . . . whose responsive understanding the author of the speech act seeks and surpasses. . . . But in addition to this addressee (the second party), the author of the utterance, with a greater or lesser awareness, presupposes a higher superaddressee (third), whose absolutely just responsive understanding is presumed, either in some metaphysical distance or in distant historical time. . . . In various ages and with various understandings of the world, this superaddressee and his ideally true responsive understanding assume various ideological expressions (God, absolute truth, the court of dispassionate human conscience, the people, the court of history, science, and so forth)’ (Bakhtin, 1987, p. 126). Bakhtin insists that this personified third party ‘is a constitutive aspect of the whole utterance, who, under deeper analysis, can be revealed in it. This follows from the nature of the word, which always wants to be heard, always seeks responsive understanding, and does not stop at immediate understanding but presses on further and further (indefinitely)’ (pp. 126–7). He also maintains, rather lamely, that this is not a mystical argument. 4 It should be mentioned that in his Introduction to Kermode’s (1985) book, Frank Lentricchia tartly remarks that the conception of the canonical text as ‘omnisignificant’ is a modernist shibboleth, rather than a universal truth, and rewrites Kermode’s permanent modernity as permanent ahistoricity.

206

Notes

5 Here Kermode is talking about the Torah, Halakhic and Haggadic midrash, the New Testament and the Reformation, rather than the literary canon, but his argument extends from one to the other. 6 As essentially code-based interpretive strategies go, Miller’s is not, after all, so extreme. In the 1860s, C. S. Peirce argued that every sign creates in a hearer’s mind an equivalent or more developed sign which he called an interpretant of the first sign (Peirce, 1931, para. 228). In other words, we do not so much recognise signs as re-cognise them, and immediately convert them into other signs, whether synonyms, translations, equivalents in another signifying system, definitions, paraphrases, and so on. The interpretant gives rise to another sign, ad infinitum. There is an infinite regression, or what Umberto Eco (1977) joyfully describes as unlimited semiosis. This allows words a far greater range of meanings than Miller’s etymologies. 7 ‘Babble’ and ‘doodle’ are Culler’s English renditions of Northrop Frye’s more learned sounding melos and opsis (see Frye, 1957, p. 275). 8 This is what makes the exact transcription of speech and the reading of transcripts so tedious, which probably partially explains pragmaticists’ frequent failure to work with authentic material. Gerald Gazdar (1979, p. 11), for example, writes ‘I shall assume . . . that invented strings and certain intuitive judgments about them constitute legitimate data for linguistic research’, while Sperber and Wilson’s Peter and Mary examples rarely have the ring of authentic data. 9 Stokes, who had been in analysis with Melanie Klein since 1930, later correlated these two processes with two fundamental psychic attitudes or ‘positions’ concerning object relations described by Klein: ‘Our relationship to all objects seems to me to be describable in the terms of two extreme forms, the one a very strong identification with the object, whether projective or introjective, whereby a barrier between self and not-self is undone, the other a commerce with a self-sufficient and independent object at arm’s length. In all times except the earliest weeks of life, both of these relationships, in vastly differing amalgams, are in play together, as is shown not only by psychoanalysis but by art, since the work of art is par excellence a selfsufficient object as well as a configuration that we absorb or to which we lend ourselves as manipulators’ (Stokes, 1978, vol. 3, pp. 151–2). Klein argues that during the first few months of life the infant recognises only part-objects: the breast, and parts of its own body. The experience of hunger gives rise to the fantasy of the persecutory ‘bad breast’, an object of hatred and fear, which is split off or ‘projected’, while the ‘good breast’ is idealised and ‘introjected’. Klein describes this as the paranoid-schizoid position. Out of a simple projection grows projective identification, in which parts of the self are projected onto the object in fantasy. Later (from six to twelve months) the infant enters the depressive position (despite their names, these positions are considered normal) in which it is able to experience self-sufficient separate whole objects, capable of both good and bad. The maternal whole object is introjected, the good aspects forming the core of both the integrated ego and the superego, and the bad aspects forming the persecutory aspect of the superego. Yet the child still experiences uncontrollable greedy and sadistic impulses. In fantasy, the loved object is attacked in greed and hatred, and torn to fragments. But this shatters the internal world

Notes 207 as well, leaving depressive feelings of loss and guilt, of pining or mourning. Klein argues that the transition from the first to the second position is never fully achieved beyond all possibility of regression, so that an individual may well oscillate between the two throughout his life, or part-and whole-object relations may be found conjointly (see Klein, 1975a, 1975b, 1975c). The parallels between Klein’s depressive position and Stokes’ conception of carving are evident. The carver esteems his material and his finished work for its otherness; the modeller moulds his material into a form determined by his ego. Stokes initially considered carving to be the intrinsically superior form, but encouraged by Klein he later developed and re-evaluated the idea of modelling. Obviously the explanatory power of these positions is not absolute. Stokes readily grants that psychoanalysis entirely ignores ‘the subjection of the artist to his time, and therefore to the art of his time, inasmuch as art must reflect the typical concatenations of experience, of endeavour, in the milieu in which the artist and his public live; otherwise the artist’s achievement of form seems to be nearly always without urgency or power’ (Stokes, 1978, p. 230). Yet Stokes chooses to grant more importance to the individual subject than to historical or contextual factors. 10 I expand this argument, and relate Klein’s version of object relations theory to other psychoanalytic approaches which might help account for and classify theories of reading – including those of D. W. Winnicott, André Green, Roy Schafer, Heinz Kohut, Ernst Wolf and Anton Ehrenzweig – in ‘Theories of Reading: Carvers and Modellers’ (Mackenzie, 1987b).

Chapter 5 1 The former Donald McCloskey is now Deirdre McCloskey, but I use the masculine pronoun because the books I refer to were written by Donald. 2 Empiricism, conventionally associated with Locke, Berkeley and Hume, although there are also much earlier and more recent versions, rejects a priori reasoning and reflection, and asserts that all claims to knowledge can only be justified by experience. Empiricists believe that experience can be described in isolation, without any reference to the nature of the object that gives rise to it, or to the mind that undergoes or ‘receives’ it. This is clearly antithetical to McCloskey’s accounts of rhetoric, induction, metaphors (models), stories and appeals to authority. 3 A footnote added to the 1782 edition of the Discours sur l’origine et des fondements de l’inégalité (1964, p. 149). All page references are to this edition, but I use de Man’s English translations. 4 Conceptualisations such as ‘tree’ and ‘leaf’ are far from arbitrary. What would be arbitrary is a classification of animals such as the following: ‘(a) belonging to the Emperor, (b) embalmed, (c) tame, (d) sucking pigs, (e) sirens, (f) fabulous, (g) stray dogs, (h) included in the present classification, (i) frenzied, (j) innumerable, (k) drawn with a very fine camelhair brush, (l) et cetera, (m) having just broken the water pitcher, (n) that from a long way off look like flies.’ However this splendid classification, which Foucault says inspired The Order of Things, comes from ‘a certain Chinese encyclopaedia’

208

5

6 7

8

9

10

11

Notes called the Celestial Empire of Benevolent Knowledge, invented by Jorge Luis Borges in his essay ‘The Analytical Language of John Wilkins’ (Borges, 1993, p. 103; Foucault, 1973, p. xv, xx). Cf. Heidegger: ‘Poetry proper is never merely a higher mode (melos) of everyday language. It is rather the reverse: everyday language is a forgotten and therefore used-up poem, from which there hardly resounds a call any longer’ (Heidegger, 1971c, p. 208). Rotman (1987) makes this point in reverse, comparing the use of paper money to the literalisation of metaphors. I discuss the opening of Capital more fully in MacKenzie (1991), a critique of Wolff (1988), which is a remarkable example of what de Man called the aesthetic ideology: Wolff argues that a description of the true reality behind ‘the mystifying appearance of capitalist society can be rendered literarily only by the ironic discourse that we in fact find in Capital’ (p. 31). See Rotman (1987), who traces developments subsequent to the introduction of the mathematical sign zero into Western consciousness, including the vanishing point in painting and, more importantly, the possibility of paper money, not backed by gold or silver, that can only be exchanged for an identical copy of itself. In fact, de Man has Locke argue that language may ‘corrupt the fountains of knowledge which are in the things themselves’, whereas what Locke writes is that ‘an ill use of [language] . . . does not corrupt the fountains of knowledge’ precisely because these ‘are in the things themselves’. Ortwin de Graef (1990, p. 223n) has drawn attention to de Man’s dubious translations and paraphrases, specifically mentioning ‘his manhandling of Locke and Kant in “The Epistemology of Metaphor”’. De Graef warns that ‘as often as not, it is more than advisable to trace de Man’s quotes to their supposed sources . . . to appreciate the sometimes dazzling import of these “minor distortions”’. Corngold (1982), Ziolkowski (1987) and Vickers (1994) have outlined in minute detail other of de Man’s misquotations and mistranslations. I will discuss two far more significant examples of his inventive translations in Chapters 7 and 8, in relation to Hegel and Rousseau. His account of Nietzsche’s notes on rhetoric, discussed in the following chapter, falls into the same category. On the other hand, there is no tradition of covering up the nether regions of pianos and tables in France – which could be because in French they do not have legs but only pieds. De Man is aware of pragmatic objections to his account of metaphor: in ‘Allegory (Julie)’ he concedes that we are not ‘helpless when confronted with figures of speech: as long as we can distinguish between literal and figural meaning, we can translate the figure back to its proper referent. We do not usually assume, for example, that someone suffers from hallucinations merely because he says that a table has four legs; the context of common usage separates the figural meaning of the catachresis (which, in this case, leads to the referent) from its literal denotation (which, in this case, is figural)’ (Allegories, p. 201). Yet this brief pragmatic interlude would itself seem to be something of a rhetorical gesture, as it in no way deflects de Man from his overall reduction of language to a system of metaphors which undermine reference. Eight further examples: ‘it is not a priori certain that [modernity] should be treated differently when discussing lyric poetry than it should, for example,

Notes 209 when discussing narrative prose or the drama’ (Blindness, p. 142); it ‘should not be a priori or naively taken for granted’ that one can leave behind the ‘hermeneutic model’ constituted by ‘the canonical principles of literary history’ (Allegories, p. ix); ‘the historical nature of literary discourse is by no means an a priori established fact’ (Allegories, p. 79); ‘The convergence of the two modes [of confession] is not a priori given, and it is because of the possibility of a discrepancy between them that the possibility of excuse arises’ (Allegories, 281); ‘the possibility arises of the entirely gratuitous and irresponsible text, not just . . . as an intentional denial of paternity for the sake of self-protection, but as the radical annihilation of the metaphor of selfhood and of the will’ (Allegories, p. 296); ‘The possibility thus arises that the postulate of ethical authority is posited for the sake of maintaining the undisputed authority of teachers in their relationship to their pupils’ (Rhetoric, p. 270); ‘it is not a priori certain that language functions according to principles which are those, or which are like those, of the phenomenal world. It is therefore not a priori certain that literature is a reliable source of information about anything but its own language’ (Resistance, p. 11); ‘whereas the meaning-function is certainly intentional, it is not a priori certain at all that the mode of meaning, the way in which I mean, is intentional in any way’ (Resistance, p. 87). 12 These examples are taken from Carston (1996). See also Carston (2002). In S&W’s account, enrichment is considered to contribute to the propositional form of an utterance: for example strengthening ‘nothing’ in ‘nothing to wear’ to ‘nothing suitable’ produces an ad hoc lexical concept, so that the proposition becomes an explicature. Loosening, on the other hand, is said only to produce implicatures, communicated with various degrees of strength: for example ‘this steak is raw’ can mean this steak is insufficiently cooked, this steak is inedible, I am very unhappy with the state of this steak, I want you to replace this steak with a properly cooked one, etc. Carston questions this distinction, and suggests that some loosenings, just as much as strengthenings, can contribute to the proposition expressed by an ad hoc metaphorical concept, constructed from the logical and encyclopaedic information made accessible by the ordinary lexical concept.

Chapter 6 1 The pages (9–15) from ‘Criticism and Crisis’ in Blindness and Insight, from which these quotations are taken also appear, slightly rewritten, in the first Gauss lecture, ‘The Contemporary Criticism of Romanticism’ (Romanticism, pp. 9–13). Cf. ‘The Rhetoric of Temporality’: ‘Sanity can only exist because we are willing to function within the conventions of duplicity and dissimulation, just as social language dissimulates the inherent violence of the actual relationships between human beings. Once this mask is shown to be a mask, the authentic being underneath appears necessarily on the verge of madness’ (Blindness, pp. 215–16). 2 De Man himself uses a series of rhetorical questions 100 pages further into Allegories of Reading, in relation to Nietzsche. For example, he asks, ‘is it not possible to progress from the rhetorical language of literature to a language

210

3

4

5

6

7

8 9

10

11

Notes that, like the language of science or mathematics, would be epistemologically more reliable?’ (p. 110). De Man’s answer is self-evidently negative, but this genuine rhetorical question contains no figure conveyed directly by means of a syntactical device. Vickers (1994, p. 208) has pointed out that both meanings (‘what’s the difference between . . . ’ and ‘what’s the difference to me?’) are in fact literal, but since the second is what we call a rhetorical question, de Man is perhaps justified it calling it figural. Cavell (1984) is actually contesting de Man’s use of Austin’s terms constative and performative here, rather than figural and literal, or ‘the tension between grammar and rhetoric’, but the parallel stands. While I wholly agree with Vickers’ criticism of de Man’s extrapolation from his Archie Bunker example, the Bunker character, like his Cockney forebear Alf Garnett, is a good instance of the difficulties of irony. From the very first episodes of Till Death Us Do Part in the 1960s, there were viewers who took Garnett to be a hero and a role model rather than a caricature of a reactionary bigot. At the Court of Appeal, in 1998, it was claimed – though not accepted by the judges – that Bentley never actually uttered the words in question, and that they were invented by the police in evidence, drawing on another conviction for the shooting of a policeman ten years previously. Cavell (1984, p. 47n) insists, in response to the objection that de Man’s example was motivated by the joke it allows about arché-debunking, that ‘for a critic to choose between accuracy and wit is as fateful as for a poet to choose between reason and rhythm. It gives up the game, or stacks it.’ I agree: given the portentousness of the conclusions de Man draws from this example, levity is no alibi. See Parkinson, (1971, pp. 106–7), cited in Hollander (1988, p. 243 n18). This widely-lauded course was ‘HUM 6’ (or ‘Humanities 6: The Interpretation of Literature’), founded by Reuben Brower. It is worth noting, however, that the intense close reading it encouraged does not inevitably lead to de Manian conclusions. Richard Poirier (1992, ch. 4) relates that he also taught HUM 6 students to ‘read in slow motion’ and immerse themselves in the waywardness of language, and describes recent poststructuralism as merely a late conversion to a kind of linguistic scepticism that had been practised even before Brower by a number of critics in England and the USA, including Empson, Blackmur and Burke. He traces this scepticism or distrust of language back to Emerson, and the American pragmatic lineage of Peirce, James and Dewey, and to poets like Stein, Stevens and Frost. Poirier rejects de Man’s account of the subversive power of literary instruction, describing HUM 6 as just a more subtle and ideologically neutral version of New Criticism. It is, of course, entirely possible to mouth utterances in foreign languages without understanding them. There was a time when learners spent hours in language laboratories doing repetition and substitution drills, frequently without knowing what they were saying, or why. What is known as communicative language teaching is precisely a reaction against the manipulation of literally meaningless sentences. In ‘Reading and History’, de Man also attempts to empty of aesthetic content the historical concretisations described by Hans-Robert Jauss in

Notes 211

12

13

14 15 16

terms of readers’ ‘horizons of expectation’. Although Jauss’s historical concepts ‘seem to dovetail perfectly’ with the structural aesthetics of the Prague circle, and ‘this theoretical alliance achieves a genuine synthesis between hermeneutics and poetics’ (Resistance, p. 63), de Man questions whether the concretisations of literary works described by Jauss, like the concretisations described by the formalists, are part of an aesthetic process. In particular, he rejects Jauss’s description of resistances and linguistic difficulties thrown up by close reading as aesthetic devices. See also Lakoff and Johnson (1980, 1999) and Lakoff and Turner (1989). Unfortunately Lakoff and Turner badly mischaracterise relevance theory (pp. 217–18). In ‘The Resistance to Theory’ and ‘Semiology and Rhetoric’, de Man discusses the ‘uncertain relationship’ or the ‘latent tension’ between grammar and rhetoric, and the problem of whether figures of speech and tropes should be considered as part of grammar as well as of rhetoric. Yet describing grammar itself as a trope would seem to remove this difficulty. As mentioned above, other rhetoricians have classified hyperbaton and anastrophe as figures. See also Vickers (1994, pp. 231–47). In ‘Allegory and Irony in Baudelaire’ (in Romanticism) and the second part of ‘The Rhetoric of Temporality’ (in Blindness), de Man considered the possibility of employing self-conscious irony in the attempt to live in an authentic temporality. Drawing on Baudelaire’s ‘De l’essence du rire’, he describes irony (which Baudelaire also named ‘le comique absolu’) as a deliberate selfreflexive duplication of the self, in which one part observes the rest from a certain distance. Baudelaire describes the capacity for such duplication as rare, belonging specifically to artists and philosophers who deal in language, and who are capable of an act of consciousness that transfers the self out of the empirical world of nature into one constituted in language. In de Man’s reading of Baudelaire, the ironic self is engaged in a quasi-Heideggerean quest for an essential, ontological relationship between understanding and temporality. The ironic self is thus an ontological one, but explicitly linguistic, like a sign. It is constituted by the act of understanding the inauthenticity of the empirical self, but it is unable to apply this knowledge to the empirical world. The ontological self can overlook or observe the worldly self or consciousness, as if from a higher viewpoint, with a privileged perspective of distance, but only as a disinterested spectator. Consequently ‘the split, the disjunction between the empirical and what we have called the literary, or poetic self, means precisely that the judgements and pronouncements emitted in romantic poetry cannot be transferred to the world of empirical experience’ (Romanticism, p. 26). The ironic, linguistic subject cannot assist the inauthentic, world-bound self, as this would be a betrayal of the ironic mode. It can know this inauthenticity but never overcome it. Furthermore, Baudelaire describes the ironic, twofold self as invariably heading for a fall. To know the inauthenticity of the empirical self is not the same as to be authentic. The ironic self may laugh at the empirical one, but this is a false feeling of pride or superiority, as ‘nature can at all times treat him as if he were a thing and remind him of his factitiousness’ (Blindness, p. 214). There are clear echoes here of the cosmic irony described

212

Notes

by Friedrich Schlegel. Thus the only possibility de Man sees of being ‘a subject whose insight is no longer in doubt and who is no longer vulnerable to irony’, or of having ‘the point of view of a unified self that fully recognizes a past condition as one of error and [sees] things as they actually are’ (Blindness, p. 224), is to write proleptically of one’s own death and speak, as it were, from beyond the grave. 17 This is a fine example of procatalepsis: anticipating an objection and preventing it. 18 This is, of course, one of de Man’s more eccentric assertions: the trivium may have been prevalent for hundreds of years until about the fifteenth century, but today it is scarcely ‘the most general of linguistic models’; indeed I suspect that many readers of ‘The Resistance to Theory’ will have encountered the trivium and quadrivium (arithmetic, geometry, astronomy, music) there for the first time.

Chapter 7 1 See Jakobson (1971, 1981), Benveniste (1971a, 1971b, 1974a, 1974b), Lyons (1978, ch. 15), Levinson (1983, ch. 2), and Kerbrat-Orecchioni (1980). 2 See Lindsay Waters’ ‘Introduction’ to Critical Writings (de Man, 1989). 3 This aspect of de Man’s thought is treated at length in Norris (1988). 4 In a series of lectures delivered shortly before his death, de Man took issue with Schiller’s claim that an aesthetic education can produce in individuals an aesthetic state that will enable them to live in harmony with their fellow human beings in an aesthetic State. He persuasively links Schiller’s advocacy of pure form [Gestalt] and aesthetic semblance [Schein] with the remarkable proposition to be found in Kleist’s short story ‘Über das Marionettentheater’ that formal perfection can best be achieved by mechanisation and mutilation. He describes Schiller’s argument as a grievous misreading of the Critique of Judgement, but shows how Schiller was in turn misread and appropriated by Goebbels. In MacKenzie (1993), I argue that the aesthetic state Schiller recommended need not necessarily lead to an aesthetic State like that of Nazi Germany. The parallels de Man draws between the (Schillerian) mechanisation and formalisation described in Kleist’s story about dancing marionettes and the practice of the Third Reich have no necessary relation with de Man’s account of the mechanical nature of language: even though fascism can be partially redescribed in aesthetic terms, its essence remains power rather than language or form. Whatever Anschluss of aesthetics may have occurred in the Third Reich, the notion of the aesthetic is not as inherently dangerous as de Man seeks to make out. 5 T. S. Eliot (1951, p. 288) famously – or rather, notoriously – suggested that a ‘dissociation of sensibility’ occurred in English literature during the seventeenth century, around the time of the Civil War, after which thought and language no longer coincided with sensory experience. 6 In his final essays, de Man returned to the theme he first developed in the mid-1950s, using Kant’s account of the sublime as one more example of the impossibility of capturing the phenomenality of experience in words. In The Critique of Judgement, Kant argued that the perceptual form of beautiful objects

Notes 213 harmonises the free play of the faculties of imagination and understanding. Pleasure in perceptual form is therefore (in theory) universal, necessary and communicable. It is also disinterested, and distinguishable from both moral judgements, and cognitive judgements aspiring to truth and knowledge. The sublime, on the contrary, is not a product of intrinsic perceptual form, but a state of mind produced by contemplation of the immensity of mountains, ravines, storms, waterfalls, and so on. It is relative and extrinsic, a matter of either numerical or dynamic greatness. The mathematical sublime concerns quantity or absolute magnitude rather than quality. It is so great as to seem immeasurable. It transposes the natural to the level of the supernatural, causing perception to give way to imagination, and understanding to give way to reason: in de Man’s succinct definition, ‘You can apprehend, to a certain point, but after a certain point you can’t comprehend what you apprehend’ (Resistance, p. 95). The mathematical sublime can be conceptualised by means of number, but the numbers are so large as to have no phenomenal equivalences, and are therefore strictly unimaginable and inaccessible to the senses. The mathematical sublime is thus a loss of the symbolic, a failure to make the infinite apparent, or to represent by sensory means the inventive articulation of which the mind is capable. This failure is overcome in the dynamic sublime, which is so powerful as to be overwhelming. It presents a threat to understanding: we are astonished to find that our faculties, including imagination, are unable to grasp or represent the totality of what they encounter. Reason or cognition transforms astonishment into tranquil detachment or admiration, by establishing the superiority of the mind over nature, thereby making the repellent attractive and the painful pleasurable. The sublime, that is to say, is not a property of nature but a purely inward experience or consciousness, independent from sensory experience (or, rather, übersinnlich, beyond the senses), and incapable of representation. 7 This is by no means the only occasion on which de Man takes on the entire collectivity of readers and critics. He suggests, for example, that ‘perhaps we have not yet begun to read [Nietzsche] properly. In the case of major authors, this is never a simple task. There are likely to be long periods of continual misinterpretation’ (De Man, 1974, pp. 49–50). Similarly, ‘the history of Rousseau interpretation is particularly rich in this respect, both in the diversity of the tactics employed to make him say something different from what he actually said, and in the convergence of these misreadings toward a definite configuration of meanings’ as if there were ‘a concerted effort to misrepresent his thought’ (Blindness, p. 112). From a pragmatist point of view, of course, a convergence of misreadings toward a definite configuration of meanings during any particular period is the meaning of the writer in question for that period, and there is no question of history requiring the rigorous rhetorical reader to come to its rescue. 8 Although he twice mentions ‘literature’ in these sentences, de Man seems to be suggesting that all language is fictional, as no form of language functions according to principles that are like those of the phenomenal world. De Man succinctly glosses the concept of fiction in ‘Excuses (Confessions)’, such that it ‘has nothing to do with representation but is the absence of any link between utterance and a referent, regardless of whether this link be causal, encoded, or governed by any other conceivable relationship that could lend itself to

214

9 10

11 12 13

Notes systematization’ (Allegories, p. 292). Pragmatists, of course, tend to be untroubled by the problem of fictional reference. Richard Rorty, for example, simply extends Dewey’s concept of ‘warranted assertibility’: Sherlock Holmes did not exist, but it is quite acceptable for his creator, and his readers, to assert that he lived on Baker Street. Rorty (1982, p. 127) says, ‘This view is amiably simple-minded, completely unhelpful if one hopes that light will be shed on the nature of fictions by analytic philosophy, and, I think, correct.’ Both here and in what follows, I draw on Soll (1969, pp. 98ff). Since writing ‘Wittgenstein on Aesthetic Responses’ (MacKenzie, 1987a) which endorsed this argument – indeed since reading Relevance – I have ceased to deny the recurrent experience of not being able to put my thoughts into words. On the issue of thoughts and words, see ‘Mentalese’, ch. 3 of Pinker (1995). In fact, the standard French translation uses viser rather than vouloir dire for meinen or mean (Hegel, Encyclopédie, 1970, p. 287). I take these three phrases about the ‘ownership’ of language from Holquist (1981). See also Clark and Holquist (1984, pp. 11–13). Cf. Wittgenstein (1958, para. 102): ‘understanding a rule is at the same time understanding how to apply it’.

Chapter 8 1 Although de Man writes that Rousseau ‘tells the entire story all over again’ (Allegories, p. 282) in the fourth Promenade, he in fact only refers to ‘le criminel mensonge dont la pauvre Marion fut la victime’, without going into details. Yet even the mention of this episode is of interest, given that Rousseau finished Book II of the Confessions by writing ‘Voilà ce que j’avais à dire sur cet article. Qu’il me soit permis de n’en reparler jamais’. 2 Thomas Trezise, the French translator of Allegories, dismisses the addition of the negative particle with a footnote: ‘De Man insère un “ne” pléonastique devant “craignisse”’ (Allégories de la lecture, 1989, p. 335). However, this is disingenuous at best, dishonest at worst. French grammar allows the addition of a pleonastic ne in subjunctive clauses with craindre and other expressions of fearing (redouter, avoir peur, de peur que, etc.), so that, for example, one can say either Je crains qu’il vienne or Je crains qu’il ne vienne for ‘I fear he’ll come’. But describing the ne as pleonastic or redundant means precisely that it serves no function; to turn the phrase into a negative the full negation ne . . . pas is required, as in Je crains qu’il ne vienne pas (‘I fear he won’t come’). But a negative particle in English cannot be pleonastic, and de Man’s English translation explicitly reads ‘if I did not fear to excuse myself’, thereby turning Rousseau’s positive phrase into a negative. De Man also mistranslates ‘en ce qui est conforme’, which means ‘concerning’ rather than ‘by means of’. 3 A possible reason for Rousseau’s insistence that he will excuse himself, and for the ‘tension’ de Man says ‘language reveals . . . in the expression: craindre de m’excuser’ (Allegories, p. 286), is the fact that in general, in French as in English, when apologising (as opposed to making excuses) one should normally ask others to do the excusing. Although English speakers say ‘Excuse me’

Notes 215 and not ‘I excuse myself,’ both are possible in French, but many Frenchspeaking parents correct their children: ‘on ne dit pas je m’excuse, mais excusez-moi’. 4 Rousseau (1959a, pp. 1273–4). The episode of the stolen ribbon is on pp. 84–7. In his ‘Introduction’ to a volume of Keats’s poetry published in 1966, de Man says that the late work, particularly The Fall of Hyperion, reveals that the poet succeeded in closing the gap between his moral stance and his private self, thereby achieving the same ‘genuine self-insight’ as Wordsworth and Rousseau and, at times, Blake and Coleridge. The moral stance was an admirable generosity and sympathy for the suffering of mankind, while ‘consciousness of self’ for de Man means an understanding of the torments of our temporal human contingency. The Romantic poets’ ‘genuine self-insight’ is ‘the first and necessary step toward moral judgement’, which distinguishes them from the rest of us, whose ‘sense of selfhood hardly ever rises above selfjustification’ (Writings, p. 197). Here de Man seems fully to accept Rousseau’s claim in the Confessions that his universal moral sense had its origin in his being falsely accused in childhood of breaking Mlle Lambercier’s comb. A decade later, in ‘Excuses’, he argues that Rousseau too, even in his so-called Confessions, rarely rises above self-justification. 5 This cartoon leads Fish to the conclusion that ‘there is no epistemological difference between direct and mediated communications because, in a fundamental sense, all communications are mediated. That is, communications of every kind are characterized by exactly the same conditions – the necessity of interpretive work, the unavoidability of perspective, and the construction by acts of interpretation of that which supposedly grounds interpretation, intentions, characters, and pieces of the world’ (Fish, 1989, pp. 43–4). This presumably cancels out Fish’s earlier argument (1981, p. 204) that ‘intentions are available to everyone who invokes the proper (publicly known and agreed upon) procedures, and . . . anyone who invokes those procedures (knowing that they will be recognized as such) takes responsibility for having that intention. Were it otherwise, then the consequences would be disastrous . . . formulas like “I’m sorry” and “thank you” would not be accepted as expressions of regret and gratitude unless it could be proven, by some independent text, that the speaker was actually so disposed.’ 6 One can hardly ascribe to Rousseau an awareness of unconscious impulses. As Marcel Raymond writes in a note to the fourth Promenade, ‘Pour Rousseau, la connaissance de soi n’est pas un problème, mais une donnée. D’un contact existentiel avec soi-même naît une intuition immédiate de soi: “passant ma vie avec moi je dois me connaître,” telle est l’évidence dont Rousseau fait part à Malesherbes le 4 janvier 1762’ (Rousseau, 1959b, p. 1788n). Although de Man’s account of the randomness of language leaves little room for psychoanalytic motivation, Barbara Johnson has analysed an intriguing personal moment in his final series of lectures, in which, while arguing that language is material and impersonal, he threw in a lot of intriguing immaterial personal remarks. In ‘The Task of the Translator’ (in Resistance), in which he explicitly states the thesis that language is inhuman, de Man discusses different connotations of bread, Brot, brood and pain, and states that he gets ‘very upset’ if he has to think that brood in Flemish (his mother tongue) and pain in French are the same thing. Questioned about this section of his lecture, de

216

7

8

9

10

Notes Man becomes, in Johnson’s words, ‘very close to total incoherence’. She points out that ‘his nervousness suggests that he sees it as a slip’. Unlike de Man, Johnson sees loss of control over language as distinctly human: ‘What de Man’s categories of human and inhuman seem to lack is a concept of the unconscious. Though he may have reasons to feel it with particular acuity, he is not alone in experiencing the approach to the mother tongue – or perhaps to the mother as such – as a threat of a loss of control. Indeed, even without assigning an unconscious meaning to the example (seeing in it de Man’s relation to his mother, or to Belgium), one would have to say that de Man’s riff on pain and Brot sounds suspiciously like a process of free association, a rather exuberant glide along what Lacan would call the signifying chain’ (Johnson, 1990a, p. 15). In the section of ‘Shelley Disfigured’ in which he quotes Shelley’s Rousseau describing himself as being ‘between desire and shame / Suspended’, de Man goes on to assert that ‘the divergence between words and deeds . . . seems to be suspended in Rousseau’s work’ (Rhetoric, p. 102), that ‘words cannot be isolated from the deeds they perform’ (p. 102), and that ‘Rousseau has overcome the discrepancy of action and intention that tears apart the historical world, and he has done so because his words have acquired the power of actions as well as of the will’ (p. 103). This argument – which is deconstructed as the essay continues – is not the same as the argument of ‘Excuses (Confessions)’. There are, of course, people who confess to things they have not done, in which case the confession – or the excuse, in the sense of apology rather than justification – or the speech act may be said to generate the guilt. But the speech act, in turn, is not the result of language performing as a machine, but is generated by mental compulsions (in some cases, it is often argued, ‘the train of psychic consequences’ of a Catholic education). All of de Man’s rediscovered wartime articles were published in de Man (1988), with English translations by Ortwin de Graef of the pieces written in Flemish. See also Werner Hamacher, Neil Hertz and Thomas Keenan (1989), a volume containing thirty-six pieces by academics responding to both de Man’s rediscovered articles and the newspaper coverage they received. The first three quotations are from Le Soir (18 March, 14 October and 28 October 1941); the translations are either adapted from various pieces in Hamacher et al. (1989) or are my own. The last quotation is from Het Vlaamsche Land, 31 May–1 June 1942. In his first postwar essay, ‘Montaigne and Transcendence’ (1953), de Man writes that ‘The wretched myths that surround us . . . must appeal to the most factitious loyalties – those to race and nation – in order to gain any vitality at all’ (Writings, p. 10), while in ‘The Literature of Nihilism’ (1966), he writes of the murder of German civilisation by the Nazis, and claims that nineteenth-century German thought had a great awareness of the historical forces that could bring about such catastrophes as Nazism – even though ‘national categories applied to literary and philosophical matters always tend to miss the mark’ (Writings, p. 164). He also writes that ‘the discrepancy between intellectual values and actual behavior has rarely been so baffling’ as in the case of the Third Reich, and that ‘the Nazis received little support from German writers and intellectuals and were not very eager to enlist them in their ranks’ (p. 163).

Notes 217 11 Balfour also states, however, that ‘even if a patient reading of de Man’s texts of this period reveals a contradictory and sometimes subversive program, their inclusion in Le Soir and Het Vlaamsche Land signals participation in a heinous political program for which there can be no good excuse’ (Hamacher et al., 1989, p. 7). 12 This figure comes from Els de Bens (in Hamacher et al., 1989). 13 In MacKenzie (1995), I contrast de Man’s accounts of the self, language and history with Richard Rorty’s pragmatist (and Freudian) account of human nature and human language, taking as an example the texts of the so-called ‘revisionists’ (I prefer ‘negationists’) who deny the existence of the Nazi gaschambers. I argue that the revisionists, far from being victims of the inhuman nature of language and the textual nature of history, are deliberately misreading and misusing language for very specific ends, and that contrary to de Man’s implicit logic, we can and must hold them responsible for their actions.

Chapter 9 1 In a splendid piece of overinterpretation – ‘GlieDErMANn: Defacement as Autobiography’ – William Flesch (1990) has analysed these two sentences. He remarks on the impersonal tone of de Man’s essays, and suggests that because ‘the impression he sought to give was one of pure intelligence, uncolored by empirical life . . . moments of autobiography . . . would be as precious as they are rare’ (Flesch, 1990, p. 243). Into this autobiographical aside in a discussion of Kleist’s ‘Über das Marionettentheater,’ a short story originally published in the Berliner Abendblätter, Flesch reads references to de Man’s wartime writings for Le Soir, Swiss neutrality, the death of de Man’s uncle Hendrik in a street accident in Switzerland, the death of de Man’s brother Hendrik in 1936, the suicide of de Man’s mother Magdelena and a great deal more besides. Flesch’s essay is a dazzling example of how one can interpret a text given time, biographical information and particular thematic premises. 2 Although there is no evidence that Saussure was aware of Durkheim’s work, there are clear similarities between his account of a shared, identical language, and the French sociologist’s anti-individualism or collectivism, with its stress on the ‘objective reality of social facts’ which must be considered ‘as things’. Durkheim (1964, ch. 2) described law, conventions and proverbs as definite, objective, social (or collective) realities; Saussure went even further, describing individual word meanings as objectively ascertainable, identical, collective representations. This conception of language recalls Chomsky’s (1965) startling idealisation at the beginning of Aspects of the Theory of Syntax such that ‘linguistic theory is concerned primarily with an ideal speaker–listener, in a completely homogeneous speech community, who knows its language perfectly and is unaffected by limitations, distractions, shifts of attention and interest, and errors (random or characteristic) in applying his knowledge of the language in actual performance’ (Chomsky, 1965, p. 3). As Steven Levinson (1983, p. 3) wrote in the preface to his book Pragmatics, American pragmatics was essentially ‘a remedial discipline

218

3

4

5

6

7 8

Notes

born, or re-born, of the starkly limited scope of Chomskyan linguistics’. But Chomsky, of course, has since extended his conception of language. For example, he has written, ‘I am using “language” to refer to an individual phenomenon, a system represented in the mind/brain of a particular individual. If we could investigate in sufficient detail, we would find that no two individuals share exactly the same language in this sense, even identical twins who grew up in the same social environment’ (Chomsky, 1988, p. 36), and ‘our goals, beliefs, expectations, and so forth clearly enter into our decision to use the rules in one way or another, and principles of rational inference and the like may also play a role in these decisions. This is true not only of what we decide to say but how we decide to say it, and similar factors enter at some level into determining how we understand what we hear’ (Chomsky, 1986, p. 261). There is something deeply ironic – or perhaps merely amusing – about the fact that de Man, proclaiming the inhuman nature of language, accidentally left out (for whatever reason) a crucial negative (restored by whoever transcribed the lecture), thereby unwittingly inverting the meaning of the clause – in a lecture in which he dryly points out how the French and English translators of Benjamin’s essay add a negative at one crucial place, and drop one at another. This is a case of a reader adding a necessary negative particle – as de Man thought he was doing in his reading of Rousseau’s Confessions (see Chapter 7). Amusingly, in my Robert & Collins French/English dictionary, bátard is given the symbol that means ‘has no equivalent in the target language and is therefore untranslatable’. Hence ‘the nearest cultural equivalent is given’, which turns out to be ‘Vienna roll’. A further example of an untranslatable word, of course, is Aufgabe, which means both ‘task’ and ‘to give up’. Benjamin’s account of literal translation recalls Mark Twain’s (1986) joke Jumping Frog: In English, Then in French, Then Clawed Back into a Civilized Language Once More by Patient, Unremunerated Toil: in response to a negative French appreciation of American humour, including his own, Twain transliterated a French translation of his story ‘The Celebrated Jumping Frog of Calveras County’ back into English, so as to be able to say, see, no wonder they don’t find my writing funny, look what they’ve done to it. Nicolas Boileau, from Satire II, and Art Poétique I, in Oeuvres Complètes (1966, pp. 18 and 157). These verses are both quoted by Milman Parry in ‘The Traditional Epithet in Homer’, in Adam Parry (1971, p. 133). See Levy (1969, pp. 103ff), and Gutt (1991, pp. 106–12). In ‘Phenomenality and Materiality in Kant’, however, de Man gives an engaging example of a philosophical argument that is, he claims, determined by a phonetic accident – Kant’s notion of the dynamic sublime: ‘Just try to translate one single somewhat complex sentence of Kant, or just consider what the efforts of entirely competent translators have produced, and you will soon notice how decisively determining the play of the letter and of the syllable, the way of saying (Art des Sagens) as opposed to what is being said (das Gesagte) – to quote Walter Benjamin – is in this most inconspicuous of stylists. Is not the persuasiveness of the entire passage on the recovery of the imagination’s tranquillity after the shock of sublime surprise based, not so much on the little play acted out by the senses, but on the proximity between

Notes 219 the German words for surprise and admiration, Verwunderung and Bewunderung? And are we not made to assent to the more than paradoxical but truly aporetic incompatibility between the failure of the imagination to grasp magnitude with what becomes, in the experience of the sublime, the success of this same imagination as an agent of reason, are we not made to assent to this because of a constant, and finally bewildering alternation of the two terms, Angemessen(heit) and Unangemessen(heit), to the point where one can no longer tell them apart? The bottom line, in Kant as well as in Hegel, is the prosaic materiality of the letter and no degree of obfuscation or ideology can transform this materiality into the phenomenal cognition of aesthetic judgement’ (Aesthetic, pp. 89–90). Unfortunately there is absolutely no way we could ever know whether this intriguing idea – that the German language, in the form of two cognate abstract nouns, is the begetter of at least one of the puzzling arguments in the Critique of Judgement – is true. 9 I will not expound on ‘Wordsworth and the Victorians’ and ‘Autobiography as De-Facement’, with their many invocations of disfiguration, and the attribution to Wordsworth of a ‘near-obsessive concern with mutilation’ (Rhetoric, p. 73), as a mere summary of their many remarkable (not to say grotesque) assertions, associations, substitutions, transitions and transformations would take several pages, and Bialostosky (1992) has already carefully undertaken this. 10 I have chosen to relegate to a footnote the elided half-sentence in this quote, which reads, ‘Besides attesting to the now well-supported view that de Man suffered from a kind of multiple personality disorder . . . ’ V. N. Vološinov (1973) analysed the active relations of messages involved in reported speech, and suggested that there are always two mingled tendencies – commenting and retorting – although one is usually dominant. Currie shows that in de Man’s case, things are rather more complex. 11 De Man’s page references are to the Gallimard Oeuvres Complètes; the translations are his own.

Chapter 10 1 Bate (1997) reveals that Empson, originally a maths student, was fully aware of Heisenberg and Schrödinger and the new physics, and suggests that he brought literary criticism into the twentieth century, superseding the eighteenth and nineteenth centuries’ need to choose between alternative meanings with a quantum-inspired logic of both/and. 2 There has been little argument about the ‘my’ and ‘I’: although critics more often refer to ‘the speaker’ than to ‘Wordsworth’, they usually go on to refer to other poems by Wordsworth and to his biography, thereby making it quite clear that they take the speaker of ‘the Lucy poems’ to represent the poet himself. 3 Interpretations based on punctuation, of course, rather contradict the notions of the poetic speaker and the lyric form. 4 Despite these remarks about the illegitimacy of making ‘she’ in line three refer to ‘Lucy’, Caraher agrees that for ‘she’ in the second stanza, it is necessary to

220

5

6

7

8

9

10

11

12

13

14

Notes go to ‘Strange fits of passion’ and ‘She dwelt among th’untrodden ways’ for clarification. This reading has been put forward more than once in the past twenty-five years, including by Peggy Kamuf (1987), who takes pains to deconstruct Knapp and Michaels’ suggestion that the author of ‘A slumber’ was half a dozen men in white lab coats in a small submarine. Thomas McFarland (1992, p. 143) has presented a persuasive case for considering both Rivers, and the proud and indignant man who built the seat in the yew-tree and retreated into solitude in Wordsworth’s first poem in Lyrical Ballads, as rejected versions of Wordsworth’s own self. Herbert Hartman (1934) assembled eight previously suggested candidates for Lucy, but like Matlak chooses Dorothy. Caraher offers a ninth possibility: Margaret (Peggy) Hutchinson, the younger sister of Wordsworth’s wife, who died of consumption in 1796. In his earliest commentary on ‘A slumber’, Hartman described Lucy as ‘a boundary being, nature sprite and human, yet not quite either . . . an intermediate modality of consciousness’ (1964, p. 158); Ferguson simply takes this dematerialization to the extreme. 1805 Prelude (V, 415). In the version of this poem in the Lyrical Ballads, the boy ‘died when he was ten years old’ (Wordsworth, 1992, p. 141); in the 1850 Prelude this becomes ‘ere he was full twelve years old’ (V, 390). In The Excursion, for example, the Wanderer is described as possessing ‘wondrous skill / To blend with knowledge of the years to come, /Human, or such as lie beyond the grave’ (I, 431–33, in Wordsworth, 1949). De Man’s various readings of ‘The Boy of Winander’ probably need to be considered in conjunction with his accounts of Wordsworth’s Essays on Epitaphs and various passages in The Prelude, in ‘Autobiography as Defacement’ and ‘Wordsworth and the Victorians’ (both in Rhetoric). See Bialostosky’s (1992) dissection of these essays, and Neil Hertz’s essay ‘Lurid Figures’ (1989). Tom Stoppard, for one, has suggested in more than one interview that there should be a typeface (rather than a mark of punctuation) called ironics: perhaps backward-sloping italics would do the trick. It is also worth mentioning that Miller has subsequently put forward a very similar, and uncontentious, paraphrase of ‘A slumber’, designed to demonstrate the claim that a key element of narrative is ‘personification, or, more accurately and technically stated, prosopopeia’ (Miller, 1990, p. 75). Although Miller doesn’t deal with the other Lucy poems, the justification for describing Lucy as a virgin child would seem to come from ‘Three years she grew in sun and shower’, rather than ‘A slumber’. Although the name Lucy does not appear in this poem, Caraher points out that earlier Lucys exist in poems by Tickell, Lyttleton, Edward Moore, Chatterton, William Collins and Samuel Rogers, which led Herbert Hartman (1934, p. 141) to suggest that ‘the name itself, it would seem, had become a neo-Arcadian commonplace, an eighteenth-century elegiac fixture’.

References Abrams, M. H., ‘A Colloquy on Recent Critical Theories’, in Doing Things with Texts, ed. Michael Fischer (New York: W. W. Norton, 1989a). Abrams, M. H., ‘Construing and Deconstructing’, in Doing Things with Texts, ed. Michael Fischer (New York: W. W. Norton, 1989b). Austin, J. L., How to Do Things with Words (Oxford: Clarendon Press, 1962). Austin, J. L., Philosophical Papers, 3rd edn, ed. J. O. Umson and G. J. Warnock (Oxford: Oxford University Press, 1979). Bahti, Timothy, ‘Telephonic Crossroads: the Reversal and the Double Cross’, in Werner Hamacher, Neil Hertz and Thomas Keenan (eds), Responses: On Paul de Man’s Wartime Journalism (Lincoln: University of Nebraska Press, 1989). Bakhtin, Mikhail, ‘Discourse in the Novel’ (1934–5), in The Dialogic Imagination: Four Essays by M. M. Bakhtin, ed. Michael Holquist, trans. Caryl Emerson and Michael Holquist (Austin: University of Texas Press, 1981). Bakhtin, Mikhail, Problems of Dostoevsky’s Poetics (1929/1973), ed. and trans. Caryl Emerson (Minneapolis: University of Minnesota Press, 1984). Bakhtin, Mikhail, Speech Genres and Other Late Essays, trans. Vern W. McGee, ed. Caryl Emerson and Michael Holquist (Austin: University of Texas Press, 1987). Balfour, Ian, ‘“Difficult Reading”: De Man’s Itineraries’, in Werner Hamacher, Neil Hertz and Thomas Keenan (eds), Responses: On Paul de Man’s Wartime Journalism (Lincoln: University of Nebraska Press, 1989). Barthes, Roland, ‘The Death of the Author’, in Image – Music – Text, ed. and trans. Stephen Heath (London: Fontana, 1977). Bate, Jonathan, The Genius of Shakespeare (London: Picador, 1997). Bateson, F. W., English Poetry: A Critical Introduction (London: Longman Green, 1950). Baudelaire, Charles, ‘De l’essence du rire’ (1855), in Oeuvres Complètes, vol. 2, ed. Claude Pichois (Paris: Gallimard, 1976). Benjamin, Walter, ‘The Task of the Translator’ (1921), in Illuminations, trans. Harry Zohn (New York: Schocken, 1969). Benveniste, Emile, ‘The Nature of Pronouns’, in Problems in General Linguistics, vol. 1, trans. M. E. Meek (Coral Gables, FL: University of Miami Press, 1971a). Benveniste, Emile, ‘Man in Language’, in Problems in General Linguistics, vol. 1, trans. M. E. Meek (Coral Gables, FL: University of Miami Press, 1971b). Benveniste, Emile, ‘Le langage et l’expérience humaine’, in Problèmes de linguistique générale, vol. 2 (Paris: Gallimard, 1974a). Benveniste, Emile, ‘L’appareil formel de l’enonciation’, in Problèmes de linguistique générale, vol. 2 (Paris: Gallimard, 1974b). Bialostosky, Don H., Wordsworth, Dialogics and the Practice of Criticism (Cambridge: Cambridge University Press, 1992). Bloom, Harold, The Visionary Company: A Reading of English Romantic Poetry, rev. edn (Ithaca, NY: Cornell University Press, 1971). Boileau, Nicolas, Satire II (1663), in Oeuvres Complètes (Paris: Gallimard, 1966a). 221

222

References

Boileau, Nicolas, Art Poétique (1674), in Oeuvres Complètes (Paris: Gallimard, 1966b). Booth, Wayne, A Rhetoric of Irony (Chicago: University of Chicago Press, 1974). Borges, Jorge Luis, ‘The Analytical Language of John Wilkins’, trans. Ruth L. C. Simms, in Other Inquisitions, 1937–1952 (Austin: University of Texas Press, 1993). Burke, Kenneth, A Rhetoric of Motives (1950; Berkeley and Los Angeles: University of California Press, 1969). Butler, Samuel, Hudibras (1663) (Cambridge: Cambridge University Press, 1905). Caraher, Brian G., Wordsworth’s ‘Slumber’ and the Problematics of Reading (University Park, PA: Pennsylvania State University Press, 1991). Carston, Robyn, ‘Enrichment and Loosening: Complementary Processes in Deriving the Proposition Expressed’, in UCL Working Papers in Linguistics, 8 (1996), 61–88. Carston, Robyn, Thoughts and Utterances: The Pragmatics of Explicit Communication (Oxford: Blackwell, 2002). Cavell, Stanley, ‘The Politics of Interpretation’, in Themes Out of School: Effects and Causes (San Francisco, CA: North Point Press, 1984). Chomsky, Noam, Aspects of the Theory of Syntax (Cambridge, MA: MIT Press, 1965). Chomsky, Noam, Knowledge of Language: Its Nature, Origin and Use (New York: Praeger, 1986). Chomsky, Noam, Language and Problems of Knowledge (Cambridge, MA: MIT Press, 1988). Clark, Katerina and Michael Holquist, Mikhail Bakhtin (Cambridge, MA: Harvard University Press, 1984). Colinet, Edouard, ‘Paul de Man and the Cercle du Libre Examen’, in Werner Hamacher, Neil Hertz and Thomas Keenan (eds), Responses: On Paul de Man’s Wartime Journalism (Lincoln: University of Nebraska Press, 1989). Condillac, Etienne, Essai sur l’origine des connaissances humaines (1746) (Paris: Galilée, 1973). Corngold, Stanley, ‘Error in Paul de Man’, Critical Inquiry, 8 (1982), 489–507. Craig, George, ‘Of Power, Pigs and the Human Body’, TLS, 19 May (1995), 23. Crosman, Robert, ‘Do Readers Make Meaning?’, in Inge Crosman and Susan Sulieman (eds), The Reader in the Text (Princeton, NJ: Princeton University Press, 1980). Culler, Jonathan, The Pursuit of Signs: Semiotics, Literature, Deconstruction (London: Routledge, 1981). Culler, Jonathan, ‘Changes in the Study of Lyric’, in Chaviva Hosek and Patricia Parker (eds), Lyric Poetry: Beyond New Criticism (Ithaca, NY: Cornell University Press, 1985). Currie, Mark, ‘The Voices of Paul de Man’, Language and Literature, 3 (1993), 183–96. Davies, Hugh Sykes, ‘Another New Poem by Wordsworth’, Essays in Criticism, 15 (April 1965), 135–61. Dawkins, Richard, The Selfish Gene (Oxford: Oxford University Press, 1976). Dawkins, Richard, The Extended Phenotype (Oxford: Oxford University Press, 1982). Dawkins, Richard, River Out of Eden (London: Weidenfeld and Nicolson, 1995). De Bens, Els, ‘Paul de Man and the Collaborationist Press’, in Werner Hamacher, Neil Hertz and Thomas Keenan (eds), Responses: On Paul de Man’s Wartime Journalism (Lincoln: University of Nebraska Press, 1989).

References

223

De Graef, Ortwin, ‘Silence to be Observed: a Trial for Paul de Man’s Inexcusable Confessions’, Yale Review of Criticism, 3:2 (1990), 205–23. De Graef, Ortwin, Serenity in Crisis: A Preface to Paul de Man, 1939–1960 (Lincoln: University of Nebraska Press, 1993). De Graef, Ortwin, Titanic Light: Paul de Man’s Post-Romanticism, 1960–1969 (Lincoln: University of Nebraska Press, 1995). De Man, Paul, ‘Nietzsche’s Theory of Rhetoric’, Symposium, 28:1 (1974), 33–51. De Man, Paul, Allegories of Reading: Figural Language in Rousseau, Nietzsche, Rilke and Proust (New Haven, CT: Yale University Press, 1979). De Man, Paul, Blindness and Insight: Essays in the Rhetoric of Contemporary Criticism, 2nd edn (London: Methuen, 1983). De Man, Paul, The Rhetoric of Romanticism (New York: Columbia University Press, 1984). De Man, Paul, The Resistance to Theory, ed. Wlad Godzich (Minneapolis: University of Minnesota Press, 1986). De Man, Paul, Wartime Journalism, 1939–1943, ed. Werner Hamacher, Neil Hertz and Thomas Keenan (Lincoln: University of Nebraska Press, 1988). De Man, Paul, Allégories de la lecture, trans. Thomas Trezise (Paris: Galilée, 1989). De Man, Paul, Critical Writings (1953–1978), ed. Lindsay Waters (Minneapolis: University of Minnesota Press, 1989). De Man, Paul, Romanticism and Contemporary Criticism: The Gauss Seminar and Other Papers, ed. E. S. Burt, Kevin Newmark and Andrzej Warminski (Baltimore, MD: Johns Hopkins University Press, 1993). De Man, Paul, Aesthetic Ideology, ed. Andrzej Warminski (Minneapolis: University of Minnesota Press, 1996). Derrida, Jacques, Of Grammatology, trans. Gayatri Chakravorty Spivak (Baltimore: Johns Hopkins University Press, 1976). Derrida, Jacques, Dissemination, trans. Barbara Johnson (Chicago: University of Chicago Press, 1981a). Derrida, Jacques, Positions, trans. Alan Bass (Chicago: University of Chicago Press, 1981b). Derrida, Jacques, ‘In Memoriam’, in The Lesson of Paul de Man: Yale French Studies, 69 (1985), 13–16. Derrida, Jacques, ‘Signature Event Context’, in Limited Inc., ed. Gerald Graff (Evanston, IL: Northwestern University Press, 1988). De Saussure, Ferdinand, Course in General Linguistics (1915), trans. Wade Baskin (London: Fontana, 1974). Dilthey, Wilhelm, ‘The Rise of Hermeneutics’ (1900), trans. Fredric Jameson, NLH, 3 (1972), 229–44. Durkheim, Emile, The Rules of Sociological Method (1895), ed. George E. G. Catlin, trans. Sarah A. Solavay and John H. Mueller (New York: Free Press, 1964). Eco, Umberto, A Theory of Semiotics (London: Macmillan, 1977). Eco, Umberto, The Role of the Reader (Bloomington: Indiana University Press, 1979). Eliot, T. S., ‘The Metaphysical Poets’, in Selected Essays (London: Faber, 1951). Empson, William, Seven Types of Ambiguity (1930) (London: Penguin, 1972). Ferguson, Frances, Wordsworth: Language as Counter-Spirit (New Haven, CT: Yale University Press, 1977). Fish, Stanley, Is there a Text in this Class? The Authority of Interpretive Communities (Cambridge, MA: Harvard University Press, 1981).

224

References

Fish, Stanley, Doing What Comes Naturally: Change, Rhetoric and the Practice of Theory in Literary and Legal Studies (Oxford: Clarendon Press, 1989). Flesch, William, ‘GlieDErMANn: Defacement as Autobiography’, in Colloquium Helveticum, 11/12 (1990), 241–57. Fodor, Jerry, The Modularity of Mind (Cambridge, MA: MIT Press, 1983). Foucault, Michel, The Order of Things (New York: Vintage, 1973). Foucault, Michel, ‘What is an Author?’, in Josué Harari (ed.), Textual Strategies: Perspectives in Post-Structuralist Criticism (Ithaca, NY: Cornell University Press, 1979). Fowler, Alasdair, ‘A New Theory of Communication’, London Review of Books, 30 (March 1989), 16–17. Friedman, Milton, ‘The Methodology of Positive Economics’, in Essays in Positive Economics (Chicago: University of Chicago Press, 1953). Frye, Northrop, The Anatomy of Criticism (Princeton, NJ: Princeton University Press, 1957). Gadamer, Hans-Georg, Truth and Method, 2nd edn, trans. Garrett Barden and John Cumming (New York: Crossroad, 1982). Gasché, Rodolphe, The Wild Card of Reading: On Paul de Man (Cambridge, MA: Harvard University Press, 1998). Gazdar, Gerald, Pragmatics: Implicature, Presupposition and Logical Form (New York: Academic Press, 1979). Geuss, Raymond, ‘A Response to Paul de Man’, Critical Inquiry, 10 (1983), 375–82. Gibbs, Raymond W. Jr, The Poetics of Mind: Figurative Thought, Language and Understanding (Cambridge: Cambridge University Press, 1994). Godzich, Wlad, The Culture of Literacy (Cambridge, MA: Harvard University Press, 1996). Greenblatt, Stephen J. (ed.), Allegory and Representation: Selected Papers from the English Institute 5 (Baltimore, MD: Johns Hopkins University Press, 1981). Grice, H. P., ‘Meaning’ (1957), in Studies in the Way of Words (Cambridge, MA: Harvard University Press, 1989a). Grice, H. P., ‘Logic and Conversation’ (1967/1987), in Studies in the Way of Words (Cambridge, MA: Harvard University Press, 1989b). Gutt, Ernst-August, Translation and Relevance: Cognition and Context (Oxford: Blackwell, 1991). Habermas, Jürgen, Legitimation Crisis, trans. Thomas McCarthy (Boston, MA: Beacon Press, 1975). Hamacher, Werner, Neil Hertz and Thomas Keenan (eds), Responses: On Paul de Man’s Wartime Journalism (Lincoln: University of Nebraska Press, 1989). Hartman, Geoffrey, Wordsworth’s Poetry, 1787–1814 (New Haven, CT: Yale University Press, 1964). Hartman, Geoffrey, The Unremarkable Wordsworth (Minneapolis: University of Minnesota Press, 1987). Hartman, Herbert, ‘Wordsworth’s Lucy Poems’, PMLA, 49 (1934), 134–42. Hegel, G. W. F., The Phenomenology of Mind (1807), trans. J. B. Baillie, 2nd edn (London: Allen and Unwin, 1949). Hegel, G. W. F., Encyclopédie des sciences philosophiques, vol. 1: La science de la logique, trans. Bernard Bourgeois (Paris: Vrin, 1970). Hegel, G. W. F., Enzyklopädie der philosophischen Wissenschaften: Erster Teil, Die Wissenschaft der Logik (1817), Werke, vol. 8 (Frankfurt: Suhrkamp, 1970).

References

225

Hegel, G. W. F., Logic (1817), trans. William Wallace (Oxford: Clarendon Press, 1975). Heidegger, Martin, ‘A Dialogue on Language’, in On the Way to Language, trans. Peter D. Hertz (New York: Harper and Row, 1971a). Heidegger, Martin, ‘The Nature of Language’, in On the Way to Language, trans. Peter D. Hertz (New York: Harper and Row, 1971b). Heidegger, Martin, ‘Language’, in Poetry, Language, Thought, trans. Albert Hofstadter (New York: Harper and Row, 1971c). Heidegger, Martin, ‘Poetically Man Dwells’, in Poetry, Language, Thought, trans. Albert Hofstadter (New York: Harper and Row, 1971d). Heidegger, Martin, Being and Time (1927), trans. John Macquarrie and Edward Robinson (Oxford: Blackwell, 1978). Hertz, Neil, ‘Lurid Figures’, in Lindsay Waters and Wlad Godzich (eds), Reading de Man Reading (Minneapolis: University of Minnesota Press, 1989). Hertz, Neil, ‘More Lurid Figures’, Diacritics, 20:3 (1990), 2–49. Hirsch, David H., The Deconstruction of Literature: Criticism after Auschwitz (Hanover: Brown University Press/University Press of New England, 1991). Hirsch, E. D., Validity in Interpretation (New Haven, CT: Yale University Press, 1967). Hirsch, E. D., Aims of Interpretation (Chicago: University of Chicago Press, 1974). Hirsch, E. D., ‘Meaning and Significance Reinterpreted’, Critical Inquiry, 11:2 (1984), 202–25. Holland, Norman, Poems in Persons (New York: W. W. Norton, 1973). Holland, Norman, 5 Readers Reading (New Haven, CT: Yale University Press, 1975). Holland, Norman, ‘Literary Interpretation and Three Phases of Psychoanalysis’, Critical Inquiry, 3 (1976), 221–33. Holland, Norman, ‘Unity Identity Text Self’, in Jane P. Tompkins (ed.), Reader– Response Criticism (Baltimore, MD: Johns Hopkins University Press, 1980a). Holland, Norman, ‘Re-Covering the Purloined Letter’, in Susan R. Sulieman and Inge Crosman (eds), The Reader in the Text (Princeton, NJ: Princeton University Press, 1980b). Holland, Norman, The I (New Haven, CT: Yale University Press, 1985). Holland, Norman, The Critical I (New York: Columbia University Press, 1992). Hollander, John, Melodious Guile: Fictive Pattern in Poetic Language (New Haven, CT: Yale University Press, 1988). Holquist, Michael, ‘The Politics of Representation’, in Stephen J. Greenblatt (ed.), Allegory and Representation: Selected Papers from the English Institute 5 (Baltimore, MD: Johns Hopkins University Press, 1981). Iser, Wolfgang, The Act of Reading (Baltimore, MD: Johns Hopkins University Press, 1978). Jakobson, Roman, ‘Shifters, Verbal Categories and the Russian Verb’, in Selected Writings, vol. 2 (The Hague: Mouton, 1971). Jakobson, Roman, ‘Linguistics and Poetics’, in Selected Writings, vol. 3 (The Hague: Mouton, 1981). Jakobson, Roman and Claude Lévi-Strauss, ‘“Les Chats” de Charles Baudelaire’, in Jakobson, Selected Writings, vol. 3 (The Hague: Mouton, 1981). Jauss, Hans Robert, Toward an Aesthetic of Reception, trans. Timothy Bakhti (Minneapolis: University of Minnesota Press, 1982).

226

References

Johnson, Barbara, ‘Rigorous Unreliability’, in A World of Difference (Baltimore, MD: Johns Hopkins University Press, 1989). Johnson, Barbara, ‘Poison or Remedy? Paul de Man as Pharmakon’, in Colloquium Helveticum, 11/12 (1990a), 7–20. Johnson, Barbara, ‘Writing’, in Frank Lentricchia and Thomas McLaughlin (eds), Critical Terms for Literary Study (Chicago: University of Chicago Press, 1990b). Juhl, P. D., Interpretation: An Essay on the Philosophy of Literary Criticism (Princeton, NJ: Princeton University Press, 1980). Kamuf, Peggy, ‘Floating Authorship’, Diacritics, 16:4 (1987), 3–13. Kant, Immanuel, Critique of Judgement (1790), trans. J. C. Meredith (Oxford: Oxford University Press, 1963). Kerbrat-Orecchioni, Catherine, L’Énonciation de la subjectivité dans le langage (Paris: Armand Colin, 1980). Kermode, Frank, Forms of Attention (Chicago: University of Chicago Press, 1985). Klein, Melanie, ‘A Contribution to the Psychogenesis of Manic-Depressive States’, in Love, Guilt and Reparation and Other Works: The Writings of Melanie Klein, vol. 2 (London: Hogarth Press, 1975a). Klein, Melanie, ‘Mourning and its Relation to Manic-Depressive States’, in Love, Guilt and Reparation and Other Works: The Writings of Melanie Klein, vol. 2 (London: Hogarth Press, 1975b). Klein, Melanie, ‘Notes on Some Schizoid Mechanisms’, in Envy and Gratitude and Other Works: The Writings of Melanie Klein, vol. 3 (London: Hogarth Press, 1975c). Klein, Richard, ‘DeMan’s Resistances: a Contribution to the Future Science of DeManology’, in Werner Hamacher, Neil Hertz and Thomas Keenan (eds), Responses: On Paul de Man’s Wartime Journalism (Lincoln: University of Nebraska Press, 1989). Knapp, Stephen and Walter Benn Michaels, ‘Against Theory’, in W. J. T. Mitchell (ed.), Against Theory: Literary Studies and the New Pragmatism (Chicago: University of Chicago Press, 1986). Kojève, Alexander, Introduction à la lecture de Hegel (Paris: Gallimard, 1947). Kuhn, T. S., The Essential Tension (Chicago: University of Chicago Press, 1977). Lacan, Jacques, ‘Le Séminaire sur “La Lettre volée”’, in Écrits (Paris: Seuil, 1966). Lakoff, George and Mark Johnson, Metaphors We Live By (Chicago: University of Chicago Press, 1980). Lakoff, George and Mark Johnson, Philosophy in the Flesh: The Embodiment of Mind and its Challenge to Western Philosophy (New York: Basic Books, 1999). Lakoff, George and Mark Turner, More than Cool Reason: A Field Guide to Poetic Metaphor (Chicago: University of Chicago Press, 1989). Leech, Geoffrey and Mick Short, Style in Fiction: A Linguistic Introduction to English Fictional Prose (London: Longman, 1981). Levinson, Steven, Pragmatics (Cambridge: Cambridge University Press, 1983). Levy, Jiry, Die literarische Übersetzung: Theorie einer Kunstgattung (Frankfurt: Athenäum, 1969). Locke, David, An Essay Concerning Human Understanding (1690) (Oxford: Clarendon Press, 1987). Lyons, John, Semantics, 2 vols (Cambridge: Cambridge University Press, 1978). Lyotard, Jean-François, The Postmodern Condition: A Report on Knowledge (Minneapolis: University of Minnesota Press, 1984).

References

227

Lyotard, Jean-François, The Differend: Phrases in Dispute (Minneapolis: University of Minnesota Press, 1988). MacKenzie, Ian, ‘Gadamer’s Hermeneutics and the Uses of Forgery’, Journal of Aesthetics and Art Criticism, 44 (1986), 41–8. MacKenzie, Ian, ‘Wittgenstein and Aesthetic Responses’, Philosophy and Literature, 11 (1987a), 92–103. MacKenzie, Ian, ‘Theories of Reading: Carvers and Modellers’, American Imago, 44 (1987b), 535–44. MacKenzie, Ian, ‘Author, Text, Reader and Community’, Journal of Literary Semantics, 19 (1990), 116–29. MacKenzie, Ian, ‘Marx on Value: Irony, Dialogue and the Aesthetic Ideology’, Southern Review, 24 (1991), 151–64. MacKenzie, Ian, ‘Terrible Beauty: Paul de Man’s Retreat from the Aesthetic’, Journal of Aesthetics and Art Criticism, 51 (1993), 551–60. MacKenzie, Ian, ‘Pragmatism, Rhetoric and History’, Poetics Today, 16 (1995), 283–99. MacKenzie, Ian, ‘Institutionalized Utterances, Literature and Language Teaching’, Language and Literature, 9 (2000), 61–78. Mailloux, Steven, Interpretive Conventions: The Reader in the Study of American Fiction (Ithaca, NY: Cornell University Press, 1982). Mäki, Uskali, ‘Diagnosing McCloskey’, Journal of Economic Literature, 33 (1995), 1300–18. Martin, Robert, ‘Irony and the Universe of Belief’, Lingua, 87 (1992), 77–90. Marx, Karl, Capital: A Critique of Political Economy, vol. 1 (1867) 3rd edn, trans. Samuel Moore and Edward Aveling, ed. Frederick Engels (London: Lawrence and Wishart, 1954). Matlak, Richard E., ‘Wordsworth’s Lucy Poems in Psychobiographical Context’, PMLA, 95 (1978), 46–65. Matlak, Richard E., ‘A Psychobiographical Approach to Teaching Wordsworth’s Goslar Poetry’, in Spencer Hall and Jonathan Ramsey (eds), Approaches to Teaching Wordsworth’s Poetry (New York: Modern Language Association, 1986). McCloskey, Donald N., The Rhetoric of Economics (Wisconsin: University of Wisconsin Press, 1985). McCloskey, Donald N., If You’re So Smart: The Narrative of Economic Expertise (Chicago: University of Chicago Press, 1990). McCloskey, Donald N., Knowledge and Persuasion in Economics (Cambridge: Cambridge University Press, 1994). McFarland, Thomas, William Wordsworth: Intensity and Achievement (Oxford: Clarendon Press, 1992). Miller, J. Hillis, The Ethics of Reading: Kant, de Man, Eliot, Trollope, James and Benjamin (New York: Columbia University Press, 1987). Miller, J. Hillis, ‘Narrative’, in Frank Lentricchia and Thomas McLaughlin (eds), Critical Terms for Literary Study (Chicago: University of Chicago Press, 1990). Miller, J. Hillis, ‘The Critic as Host’, in Theory Now and Then (London: Harvester Press, 1991a). Miller, J. Hillis, ‘On Edge: The Crossways of Contemporary Criticism’, in Theory Now and Then (London: Harvester Press, 1991b). Morris, Charles W., ‘Foundations of the Theory of Signs’ (1938), in Writings on the General Theory of Signs (The Hague: Mouton, 1971).

228

References

Moynihan, Robert, ‘Interview with Paul de Man’, The Yale Review, 73 (1984), 576–602. Nietzsche, Friedrich, ‘Rhetorik’ (1872–73), in Gesammelte Werke, vol. 5 (Munich: Musarion Verlag, 1922). Nietzsche, Friedrich, ‘Cours sur la rhétorique’, ed. and trans. Philippe LacoueLabarthe and Jean-Luc Nancy, Poètique, 5 (1970), 99–130. Nietzsche, Friedrich, ‘Description of Ancient Rhetoric, 1872–73’, in Friedrich Nietzsche on Rhetoric and Language, ed. and trans. Sander Gilman, Carole Blair and David J. Parent (New York: Oxford University Press, 1989a). Nietzsche, Friedrich, ‘On Truth and Lying in an Extra-Moral Sense’, in Friedrich Nietzsche on Rhetoric and Language, ed. and trans. Sander Gilman, Carole Blair and David J. Parent (New York: Oxford University Press, 1989b). Nietzsche, Friedrich, Ecce Homo (1908), trans. R. J. Hollingdale (London: Penguin, 1992). Nietzsche, Friedrich, On the Genealogy of Morality (1887), trans. Maudmarie Clark and Alan J. Svensen (Indianapolis: Hackett, 1998). Norris, Christopher, Paul de Man: Deconstruction and the Critique of the Aesthetic Ideology (London: Routledge, 1988). Ong, Walter J., ‘Text and Interpretation: Mark and After’, in John Miles Foley (ed.), Oral Tradition in Literature: Interpretation in Context (Columbia: University of Missouri Press, 1986). Parkinson, Thomas, W. B. Yeats: The Later Poetry (Berkeley and Los Angeles: California University Press, 1971). Parry, Adam (ed.), The Making of Homeric Verse: The Collected Papers of Milman Parry (Oxford: Clarendon Press, 1971). Pascal, Blaise, ‘Réflexions sur la géométrie en général: de l’esprit géométrique et de l’Art de persuader’ (1658), in Oeuvres Complètes, ed. Louis Lafuma (Paris: Seuil, 1963). Peirce, Charles Sanders, ‘Phenomenology’, in Collected Papers, vol. 1, ed. Charles Hartsthorne and Paul Weiss (Cambridge, MA: Harvard University Press, 1931–58). Pierce, Charles Sanders, ‘Speculative Grammar’, in Collected Papers, vol. 2, ed. Charles Hartsthorne and Paul Weiss (Cambridge, MA: Harvard University Press, 1931–58). Peirce, Charles Sanders, ‘A Survey of Pragmatism’, in Collected Papers, vol. 5, ed. Charles Hartsthorne and Paul Weiss (Cambridge, MA: Harvard University Press, 1931–58). Pilkington, Adrian, Poetic Effects (Amsterdam and Philadelphia: John Benjamins, 2000). Pilkington, Adrian, Barbara MacMahon and Billy Clark, ‘Looking for an Argument: a Response to Green’, Language and Literature, 6 (1997), 139–48. Pinker, Steven, The Language Instinct (London: Penguin, 1995). Poirier, Richard, Poetry and Pragmatism (London: Faber, 1992). Poulet, Georges, ‘Criticism and the Experience of Interiority’, in Jane P. Tompkins (ed.), Reader–Response Criticism (Baltimore, MD: Johns Hopkins University Press, 1980). Pratt, Mary Louise, Toward a Speech Act Theory of Literary Discourse (Bloomington: Indiana University Press, 1977). Rabinowitz, Peter, ‘Truth in Fiction: A Re-examination of Audiences’, Critical Inquiry, 4 (1977), 121–41.

References

229

Rabinowitz, Peter, Beyond Reading (Ithaca, NY: Cornell University Press, 1987). Rajan, Tillotama, The Supplement of Reading (Ithaca, NY: Cornell University Press, 1990). Reichler, Claude, ‘Comparaison n’est pas raison: Paul de Man lecteur de Rousseau’, La Nouvelle Revue Française, 445 (1990), 58–68. Richards, I. A., Principles of Literary Criticism (London: Kegan Paul, 1926). Riffaterre, Michael, ‘Describing Poetic Structures: Two Approaches to Baudelaire’s “Les Chats”’, in Jane P. Tompkins (ed.), Reader–Response Criticism (Baltimore, MD: Johns Hopkins University Press, 1980). Rorty, Richard, Philosophy and the Mirror of Nature (Princeton, NJ: Princeton University Press, 1979). Rorty, Richard, Consequences of Pragmatism: Essays 1972–80 (Minneapolis: University of Minnesota Press, 1982). Rorty, Richard, Objectivity, Relativism and Truth: Philosophical Papers, vol. 1 (Cambridge: Cambridge University Press, 1991). Rotman, Brian, Signifying Nothing: The Semiotics of Zero (Basingstoke: Macmillan, 1987). Rousseau, Jean-Jacques, Confessions (1782), in Oeuvres Complètes de Jean-Jacques Rousseau, vol. 1, ed. Bernard Gagnebin and Marcel Raymond (Paris: Gallimard, 1959a). Rousseau, Jean-Jacques, Rêveries du promeneur solitaire (1782), in Oeuvres Complètes vol. 1, ed. Bernard Gagnebin and Marcel Raymond (Paris: Gallimard, 1959b). Rousseau, Jean-Jacques, Julie ou la Nouvelle Héloïse (1761), in Oeuvres Complètes, vol. 2 (Paris: Gallimard, 1961). Rousseau, Jean-Jacques, Discours sur l’origine et des fondements de l’inégalité (1754), in Oeuvres Complètes, vol. 3 (Paris: Gallimard, 1964). Rousseau, Jean-Jacques, Essai sur l’origine des langues (1781), ed. Charles Porset (Bordeaux: Ducros, 1968). Rumrich, John P., Milton Unbound (Cambridge: Cambridge University Press, 1996). Schiller, Friedrich, On the Aesthetic Education of Man (1794), ed. Elizabeth M. Wilkinson and L. A. Willoughby (Oxford: Oxford University Press, 1967). Schlegel, Friedrich, Lucinde and the Fragments (1799), trans. Peter Firchow (Minneapolis: University of Minnesota Press, 1971). Schleiermacher, Fr D. E., ‘The Hermeneutics: Outline of the 1819 Lectures’, trans. Jan Wocjik and Roland Haas, NLH, 10 (1978), 1–16. Scholes, Robert, Textual Power (New Haven, CT: Yale University Press, 1985). Scholes, Robert, Protocols of Reading (New Haven, CT: Yale University Press, 1989). Shelley, Percy Bysshe, ‘A Defence of Poetry’ (1821), in Complete Works, vol. 3, ed. Roger Ingpen and Walter E. Peck (London: Ernest Benn, 1965a). Shelley, Percy Bysshe, ‘The Triumph of Life’, in Complete Works, vol. 4, ed. Roger Ingpen and Walter E. Peck (London: Ernest Benn, 1965b). Soll, Ivan, An Introduction to Hegel’s Metaphysics (Chicago: University of Chicago Press, 1969). Sperber, Dan, On Anthropological Knowledge (Cambridge: Cambridge University Press, 1985). Sperber, Dan, Explaining Culture: A Naturalistic Approach (Oxford: Blackwell, 1996).

230

References

Sperber, Dan, ‘Metarepresentations in an Evolutionary Perspective’, in Dan Sperber (ed.), Metarepresentations: A Multidisciplinary Perspective (Oxford: Oxford University Press, 2000). Sperber, Dan and Deirdre Wilson, Relevance: Communication and Cognition (Oxford: Blackwell, 1986/1995). Starobinski, Jean, Les mots sous les mots: Les anagrammes de Ferdinand de Saussure (Paris: Gallimard, 1971). Steiner, Peter, ‘Russian Formalism’, in Raman Selden (ed.), The Cambridge History of Literary Criticism, vol. 8: From Formalism to Poststructuralism (Cambridge: Cambridge University Press, 1995). Stokes, Adrian, The Critical Writings of Adrian Stokes, vol. 3, ed. Lawrence Gowing (London: Thames and Hudson, 1978). Stoppard, Tom, Jumpers (London: Faber and Faber, 1972). Strauss, Leo, Persecution and the Art of Writing (Westport: Greenwood Press, 1973). Twain, Mark, Jumping Frog: In English then in French then Clawed Back into a Civilized Language Once More by Patient Unremunerated Toil (1875) (New York: Dover, 1986). Vickers, Brian, In Defence of Rhetoric (Oxford: Clarendon Press, 1988). Vickers, Brian, ‘De Man’s Schismatizing of Rhetoric’, in S. Ijsseling and G. Vervaecke (eds), Renaissances of Rhetoric (Leuven: Leuven University Press, 1994). Vološinov, V. N., Marxism and the Philosophy of Language (1929), trans. L. Matejka and I. R. Titunik (New York: Seminar Press, 1973). Vonnegut, Kurt, Breakfast of Champions (New York: Dell, 1973). Warminski, Andrzej, ‘Epigraphs’, in Werner Hamacher, Neil Hertz and Thomas Keenan (eds), Responses: On Paul de Man’s Wartime Journalism (Lincoln: University of Nebraska Press, 1989). Waters, Lindsay, ‘Paul de Man: a Sketch of Two Generations’, in Werner Hamacher, Neil Hertz and Thomas Keenan (eds), Responses: On Paul de Man’s Wartime Journalism (Lincoln: University of Nebraska Press, 1989). Weaver, Warren, ‘Recent Contributions to the Mathematical Theory of Communication’, in Claude C. Shannon and Warren Weaver, The Mathematical Theory of Communication (Urbana: University of Illinois Press, 1949). Wilson, Deirdre, Slave of the Passions (London: Picador, 1991). Wilson, Deirdre and Dan Sperber, ‘Representation and Relevance’, in Ruth M. Kempson (ed.), Mental Representations: The Interface between Language and Reality (Cambridge: Cambridge University Press, 1988). Wilson, Deirdre and Dan Sperber, ‘Inference and Implicature’, in Steven Davis (ed.), Pragmatics: A Reader (Oxford: Oxford University Press, 1991a). Wilson, Deirdre and Dan Sperber, ‘Pragmatics and Modularity’, in Steven Davis (ed.), Pragmatics: A Reader (Oxford: Oxford University Press, 1991b). Wilson, Deirdre and Dan Sperber, ‘On Verbal Irony’, Lingua, 87 (1992), 53–76. Wittgenstein, Ludwig, Philosophical Investigations, 2nd edn, trans. G. E. M. Anscombe (Oxford: Blackwell, 1958). Wittgenstein, Ludwig, Philosophical Grammar, ed. R. Rhees, trans. A. Kenny (Oxford: Blackwell, 1974). Wolff, Robert Paul, Moneybags Must Be So Lucky: On the Literary Structure of Capital (Amherst: University of Massachusetts Press, 1988). Wordsworth, William, The Poetical Works of William Wordsworth, 5 vols, ed. Ernest de Selincourt and Helen Darbyshire (Oxford: Clarendon Press, 1940–9).

References

231

Wordsworth, William, The Prelude: A Parallel Text (1805, 1850), ed. J. C. Maxwell (New Haven and London: Yale University Press, 1981). Wordsworth, William, Lyrical Ballads and Other Poems, 1797–1800, ed. James Butler and Karen Green (Ithaca, NY: Cornell University Press, 1992). Ziolkowski, Theodore, ‘The Uses and Abuses of Romanticism’, Sewanee Review, 95 (1987), 276–87.

Index Abrams, M. H., 177, 191, 193–4 aesthetics, 90, 92–5, 115, 118, 208, 210–12, 214, 219 allegory, 68, 98, 111, 137, 142, 158–9, 171, 189, 192, 199–200, 211 application, 7, 9, 46, 107, 126–7 Aristotle, 1, 8, 62, 66–7, 75, 96, 173, 204 Austin, J. L., 28, 63–4, 83, 131–3, 210 Bahti, Timothy, 149 Bakhtin, Mikhail, 38, 54, 57, 129, 148, 201–5 Balfour, Ian, 149, 217 Barthes, Roland, 32, 103 Bate, Jonathan, 179, 219 Bateson, F. W., 177–9, 182, 185, 190, 195 Baumgarten, Alexander, 93 Baudelaire, Charles, 58, 108, 110, 211 Benjamin, Walter, 3, 160–6, 176, 196, 198, 218 Benveniste, Emile, 9, 107, 125, 139, 212 Bergson, Henri, 52 Betti, Emilio, 127 Bialostosky, Don H., 14, 170, 189, 219–20 Blackmur, R. P., 210 Blake, William, 215 Bloom, Harold, 36, 188 Boileau, Nicolas, 166–7, 218 Bohr, Niels, 63 Booth, Wayne, 100 Borges, Jorge Luis, 208 Brooks, Cleanth, 176, 178–9, 182, 193, 195 Brower, Reuben, 210 Bultmann, Rudolf, 35 Burke, Kenneth, 204, 210 Butler, Samuel, 64 canonicity, 7, 53–5, 187, 205, 209 Caraher, Brian G., 180–2, 185–6, 193, 219–20

Carston, Robyn, 209 catachresis, 76, 80, 82, 97–8, 112–14, 160, 208 Cavell, Stanley, 88, 90, 210 Chomsky, Noam, 37, 98, 217–18 Cicero, 8, 63 Clark, Billy, 177 Clark, Katerina, 203, 214 codes, 1, 3, 5, 8, 13–14, 16–18, 21–3, 26, 28–9, 36, 38–9, 42, 47–8, 55–9, 82–3, 102, 105–6, 113, 121, 128–9, 138, 145, 149, 153, 156–7, 194, 206, 212 cognition, 1–8, 10–11, 19, 25, 31, 41, 43, 47–8, 53, 65, 97, 102, 104, 116–17, 129–30, 143, 150, 159–60, 165, 184, 196–200, 202–4, 213, 219 cognitive effect, 1, 19–20, 31–2, 42, 47, 51, 53, 194 cognitive environment, 6, 20–3, 30, 34, 36, 40, 47, 57, 187, 193 Coleridge, Samuel Taylor, 109, 181, 184–5, 215 Colinet, Edouard, 148 communication, 1–2, 5–8, 16–48, 50, 52–3, 55–6, 64, 85, 93, 102, 122, 135, 150, 153, 155, 164, 184, 194–7, 199, 201, 203, 205, 209, 213, 215 communicative intention, 19, 36, 47–8, 60–1, 84, 195, 199 conceptualization, 65–71, 74, 82, 112, 124, 159, 172–4, 207, 213 Condillac, Etienne, 79–80 context, 1–6, 9–10, 13, 16, 18–25, 29–37, 39–40, 43–7, 50–7, 61, 85, 87, 102, 105, 113, 125, 129, 134, 139, 156, 162, 166, 169, 172, 178–9, 184–5, 187, 192, 194, 196, 202–3, 207–8 co-operative principle, 21 Copernicus, 13 232

Index Corngold, Stanley, 208 Craig, George, 11–12 Crosman, Robert, 204 Culler, Jonathan, 59, 155, 206 Currie, Mark, 171–5, 219 Davies, Hugh Sykes, 180, 182, 186 Dawkins, Richard, 42, 44 De Bens, Els, 217 decoding see codes deconstruction, 1, 2, 5, 10, 12–13, 15, 48, 53, 55–7, 61, 105, 112, 125, 134, 142, 156, 160, 171, 175, 190–1, 193–4, 199, 220 De Graef, Ortwin, 146, 149, 199, 208, 216 deixis, 5, 9, 17, 107–8, 118, 122–6, 139, 146, 149, 180 De Jouvenal, Bertrand, 147 De Man, Hendrik, 148, 217 De Man, Paul, 1–12, 14–15, 38, 41, 58, 61, 64–70, 74–82, 84–125, 127–53, 157–62, 164, 166–7, 169–77, 188–91, 193–201, 203–4, 206–20 Aesthetic Ideology, 2, 65, 74–5, 77–81, 98–101, 117–20, 122–4, 131, 145, 219 Allegories of Reading, 2, 12, 65–8, 74, 86–7, 89–90, 94–8, 112–14, 127–30, 131–42, 158–9, 169, 172–5, 199, 208–9, 214 Blindness and Insight, 2, 12, 41, 84–5, 91, 104, 111, 115, 158–9, 188–9, 199–200, 209, 211, 213 Critical Writings (1953–1978), 108–9, 188, 212, 215–17 ‘Nietzsche’s Theory of Rhetoric’, 99–100 The Resistance to Theory, 9, 12, 15, 80, 90–4, 101–6, 115–18, 149, 160–1, 164, 176, 201, 203, 209, 211, 213, 215 The Rhetoric of Romanticism, 4, 11, 91, 94–5, 98, 145–6, 152–3, 167, 169, 209, 216, 219–20 Romanticism and Contemporary Criticism, 12, 85, 109–10, 157–8, 189

233

Derrida, Jacques, 2, 4, 5, 89, 154–5, 159, 194, 198, 199 descriptive representation, 27, 70 Dewey, John, 151, 210, 214 dialogism, 57, 125, 170–1, 202–5 Dilthey, Wilhelm, 6, 33–4, 36, 39, 61 disambiguation, 17–18, 21, 51, 83, 182 Duhem, Pierre, 199 Durkheim, Emile, 217 Eco, Umberto, 38–9 Eikhenbaum, Boris, 106 Eliot, T. S., 109, 212 empiricism, 63, 121, 207 Emerson, Caryl, 203 Emerson, Ralph Waldo, 210 Empson, William, 24, 41, 179, 210, 219 encoding see codes encyclopaedic entries, 20, 23, 25, 30–1, 37, 51, 83, 178, 203, 209 enrichment, 21, 51, 82–3, 209 explicature, 22–3, 36, 39, 46–7, 51, 61, 90, 106, 166, 195, 198, 209 Ferguson, Frances, 186–8, 220 Fernandez, Ramon, 137 Feyerabend, Paul, 199 Fichte, Johann Gottlieb, 144–5 fiction, 33, 39, 105, 116, 132, 138–51, 169, 171–2, 200, 212, 214 figurative meaning, 18, 26, 47, 53, 55, 65–6, 69, 79, 81, 89, 96–7, 102, 113, 117, 121, 142, 169, 173, 192–5, 197, 208 figures, 2, 8, 45, 47, 62, 67, 79, 86, 90–9, 102, 113, 129, 136, 141, 154, 158, 169–70, 194, 201, 208, 210–11 first (or cognitive) principle of relevance, 8, 19, 31, 200 Fish, Stanley, 50–2, 77, 97–8, 135, 202, 215 Flesch, William, 217 Fodor, Jerry, 37, 203 Foucault, Michel, 32–3, 154, 201–2, 207–8 Fowler, Alasdair, 1, 13, 55, 201 free indirect speech, 171–3 Frege, Gottlob, 99

234

Index

Freud, Sigmund, 137–8, 182–3, 185 Friedman, Milton, 70–1 Frost, Robert, 166, 210 Frye, Northrop, 206

human subject, 11, 35, 45, 75, 93, 108–11, 118–26, 132, 139, 141, 149–75, 204, 211–12; see also self Husserl, Edmund, 34–5, 161

Gadamer, Hans-Georg, 6, 9, 35, 45, 126, 199, 202 Gasché, Rodolphe, 82, 104 Gazdar, Gerald, 206 Gerber, Gustave, 65, 95 Geuss, Raymond, 123–4 Gibbs, Raymond W. Jr, 97, 200 Girard, René, 157–8 Godzich, Wlad, 9–10, 118, 149–50 grammar, 3, 5, 9, 17, 21, 32, 45, 64–5, 86, 88, 98–9, 102, 106–7, 113, 127–9, 131–2, 139, 141–3, 145–6, 164, 167, 181–2, 187, 193–5, 203, 205, 210–11, 214 Greenblatt, Stephen J., 160 Grice, H. P., 18, 21, 25, 28, 178, 195 guilt, 4, 10, 117, 132–47, 185, 207, 216 Gutt, Ernst-August, 39, 218

ideology, 105, 117, 151, 154, 205, 208, 210, 219 illocutionary force, 5, 18, 51, 105, 127, 132, 135 implicature, 3, 5, 15, 18, 21–5, 28–9, 36, 39–47, 53, 55, 61, 105–6, 166, 177–82, 187, 194–5, 197–8, 209; see also weak implicature inference, 1–3, 5–6, 8–9, 13–19, 21–3, 30, 32, 36, 40, 43–4, 47, 55–7, 60–1, 83, 85, 88, 105–6, 117, 122, 144, 153–4, 157, 165–6, 179, 182, 187, 194, 196–200, 203, 218 informative intention, 19, 22–3, 30, 33, 36, 45, 47, 50, 61, 162, 178, 184, 196, 198, 202 inhuman language, 2, 4, 6, 12, 32, 61, 64, 157, 161–2, 176, 198, 215–18 intention, 2, 7, 10, 12–13, 15–19, 22–3, 25, 30–1, 33, 36, 44–50, 53, 56, 60–1, 64, 81–9, 95, 100–1, 105–6, 109, 123, 126–7, 130, 133, 141, 143, 155, 157–8, 160–3, 165–6, 170, 176–8, 184, 187, 190, 194–200, 202, 209, 215–16 interpretation, 1–4, 7–8, 11–12, 14, 16–39, 42–58, 61, 82–93, 102, 104, 108, 123, 126–7, 129, 135, 137, 140, 144, 151, 166, 176–87, 190, 193–9, 201–4, 206, 213, 215, 217, 219 interpretative representation, 2–3, 85, 88, 111, 117, 165, 194, 196 irony, 2–8, 14, 25–8, 72–3, 86–7, 97–101, 105–6, 144–5, 149, 158, 177–9, 182, 191–3, 195, 197, 198, 200–1, 208, 210–12, 220 Iser, Wolfgang, 38–9, 200

Habermas, Jürgen, 12, 64, 200 Hartman, Geoffrey, 177, 179, 188, 193, 220 Hartman, Herbert, 220 Hegel, G. W. F., 9–11, 65, 108, 115, 140, 144, 146, 170, 175, 197, 208, 219 Encyclopedia, 9, 107, 118–25, 214 Lectures on Aesthetics, 118 Phenomenology of Mind, 108, 110, 120 Heidegger, Martin, 3, 35, 78, 82, 111, 130, 164–5, 208 Heisenberg, Werner, 63, 219 hermeneutics, 6, 7, 9, 28, 29–61, 90–2, 126–7, 200, 209, 211 Hertz, Neil, 127, 170, 216, 220 Hirsch, David H., 148 Hirsch, E. D., 32, 34, 36, 61, 126–7, 156, 176, 178–9, 187, 202, 204 Hölderlin, Friedrich, 108, 111, 161–2, 164, 189, 199 Holland, Norman, 48–51, 154–7, 177, 182–4, 189–90, 193 Hollander, John, 90, 210 Holquist, Michael, 203, 214

Jakobson, Roman, 9, 12, 58–9, 66, 103, 107, 115, 166, 169, 173, 212 James, William, 151, 210 Jauss, Hans Robert, 210–11

Index Johnson, Barbara, 13, 55–6, 59, 215–16 Johnson, Mark, 211 Juhl, P. D., 176, 178 Kamuf, Peggy, 220 Kant, Immanuel, 63, 80, 93, 109, 131, 208, 212, 218–19 Keats, John, 94–5, 102, 108, 215 Kerbrat-Orecchioni, Catherine, 212 Kermode, Frank, 7, 53–5, 116, 205–6 Kierkegaard, Sören, 44, 100, 200 Klein, Melanie, 206–7 Klein, Richard, 148 Kleist, Heinrich von, 140–1, 144–5, 212, 217 Knapp, Stephen, 176, 178, 220 Kojève, Alexander, 108 Kuhn, T. S., 14, 199 Lacan, Jacques, 136–7, 183, 216 Lakatos, Imre, 199 Lakoff, George, 211 Leech, Geoffrey, 171 Lentricchia, Frank, 205 Levinson, Steven, 212, 217 Lévi-Strauss, Claude, 30, 58–9 Levy, Jiry, 218 lexical entry, 20, 209 literariness, 93, 103–4, 115, 117 Locke, David, 76–9, 207–8 logical entry, 20, 209 loosening, 17, 82–3, 111, 209 Lyons, John, 77, 212 Lyotard, Jean-François, 13 McCloskey, Donald N., 62–4, 199–200, 207 McFarland, Thomas, 220 MacKenzie, Ian, 39, 50, 202, 208, 212, 217 MacMahon, Barbara, 177 Mailloux, Steven, 38–9 Mäki, Uskali, 200 Mallarmé, Stéphane, 110, 128 manifest relevance, 23, 48, 50, 57, 106, 144, 190, 195, 198, 202 Markiewicz, Henryk, 115 Martin, Robert, 200

235

Marx, Karl, 72–3, 117 materiality, 2, 5–6, 13, 58–9, 92–4, 115, 129, 131, 166–7, 195, 219 Matlak, Richard E., 177, 184–5, 220 maximal relevance, 7–8, 11, 20, 31, 53 mechanical language, 2, 4, 10, 12, 32, 61, 64, 130–2, 140–5, 152, 167, 169, 196–7, 212 metaphor, 2, 5, 8, 23–8, 39, 41, 53, 62–83, 90, 92–4, 97–8, 105, 109–10, 112–15, 119, 136–7, 141–3, 150, 155–6, 166, 173–5, 179, 186, 195, 197–8, 201–2, 207–9 Michaels, Walter Benn, 176, 178, 220 Miller, J. Hillis, 56–8, 146, 177, 190–4, 206, 220 Milton, John, 47 Morgenstern, Christian, 168 Morris, Charles W., 4 Morson, Gary Saul, 203 Moynihan, Robert, 99–101 New Critics, 156, 176, 178, 210 Nietzsche, Friedrich, 8, 11, 34, 70, 82–3, 89, 95, 112, 131, 140, 144, 146, 150, 171–2, 174–5, 197, 209, 213 ‘Description of Ancient Rhetoric’, 8, 65, 95–8, 208 ‘On Truth and Lying’, 8, 65, 69–74, 94–5 nonliteral meaning, 17, 83, 197 Norris, Christopher, 212 numbers, 68–9, 74–6, 82, 128, 213 Ong, Walter J., 204 optimal relevance, 6–8, 19–23, 26, 30–1, 36, 40, 47–8, 50, 53, 55–6, 61, 106, 140, 178–9, 185, 194–5, 198 ostensive-inferential communication, 8, 18, 19, 21–2, 31, 61 paranomasia, 94, 115–16, 121, 201 Parkinson, Thomas, 210 Parry, Milman, 33–4, 201, 218 Pascal, Blaise, 74–6

236

Index

Peirce, Charles Sanders, 62, 77, 99, 206, 210 performatives, 95, 100, 105, 126, 129, 131–2, 135, 142, 144, 146, 194, 210 persuasion, 8, 62–5, 93–7, 99, 132, 158, 204 Pilkington, Adrian, 177, 200 Pinker, Steven, 214 Plato, 2, 79, 93, 95–6, 193 Poirier, Richard, 88, 210 Poe, Edgar Allan, 136 poetic effects, 3, 22–5, 40–1, 53, 179, 182, 187, 198, 200 poetics, 90–3, 99, 187, 211 post-Saussurean linguistics, 5, 10, 121, 130, 151, 153–4, 156 Poulet, Georges, 6, 40–1, 61, 160, 199 pragmatics, 1–5, 8–28, 30, 73, 82–3, 88, 97, 105, 107, 113, 126, 129, 152, 157, 162, 164, 177, 196, 198–9, 203, 206, 217 pragmatism, 150, 199, 210, 213–14, 217 Pratt, Mary Louise, 133 prejudgements, 34–6, 40, 126 processing effort, 1, 19–20, 31–2, 47, 51–3, 97, 185, 194 propositional attitude, 5, 17–18, 40, 47, 105, 153, 197 propositional form, 17, 23, 27–8, 39–40, 43–4, 153, 182, 197, 201, 204, 209 Proust, Marcel, 169, 171–2 psychoanalytic criticism, 137, 154, 182–5, 206–7, 215 Quine, Willard van Orman, 199 Quintilian, 62, 96 Rabinowitz, Peter, 39 Rajan, Tillotama, 45–6, 205 Raymond, Marcel, 135–7, 215 Reichler, Claude, 66 relevance, 1–3, 5–9, 11, 13, 16–61, 82, 85, 88, 97, 105–7, 111, 121–2, 129, 143, 153, 157, 165–6, 169, 177–80, 182, 184–5, 187, 193–200, 202, 205, 211, 214

reference, 2, 4–5, 8–10, 15, 17–18, 21, 40, 43, 46, 51, 61, 65–6, 69, 74–6, 78, 81, 83, 86, 90–1, 94–7, 99, 102–3, 105, 107–8, 112–18, 120, 123, 125–32, 135, 138–41, 143–4, 149, 151, 153, 155, 169, 173, 180–2, 194, 197, 200, 208, 213–14, 217 representations, 2, 3, 6, 13, 16–20, 22, 26–7, 29, 33, 37–8, 41–4, 80–1, 85, 88, 111, 117, 165, 194, 196, 200, 203–4, 213, 217–18 rhetoric, 2–11, 13, 15, 45, 62–106, 112–15, 123, 129, 131, 137–8, 141, 144, 158–61, 173, 189, 191, 193–8, 200, 204, 207–11, 213 rhetorical tetrad, 62–3 Richards, I. A., 6, 41, 99, 110 Riffaterre, Michael, 58–9 Rilke, Rainer Maria, 108, 112, 171 Rorty, Richard, 47–8, 124, 151, 214, 217 Rotman, Brian, 208 Rousseau, Jean-Jacques, 8, 10–11, 70, 111–12, 114, 128–9, 131–44, 171, 175, 199, 208, 213–16, 218 Confessions, 10, 126, 132–8, 143–4, 169, 214–15 Essai sur l’origine des langues, 68, 114, 159–60 Julie ou la Nouvelle Héloïse, 12, 112, 114 Rêveries du promeneur solitaire, 132–4, 138–42, 214–15 Second Discourse, 8, 64–9, 113–14, 131, 171–5 Social Contract, 126, 133 Rumrich, John P., 199 Saussure, Ferdinand de, 5, 10, 12, 57–8, 74, 77, 115, 126, 152–5, 217 Schelling, Friedrich Wilhelm, 109 Schiller, Friedrich von, 93, 109, 212 Schlegel, Friedrich, 100–1, 144–5, 197, 200, 212 Schleiermacher, Fr D. E., 6–7, 33–4, 36, 39, 45–6, 61 Scholes, Robert, 45

Index Schrödinger, Erwin, 219 second (or communicative) principle of relevance, 8, 19, 31, 200 self, 5, 12, 38, 40, 45, 48–50, 85, 99–100, 108–10, 154–60, 182–3, 206, 211–12, 215; see also human subject semantic representation, 2, 9, 16–18, 23, 39–40, 43, 85, 90, 153, 182, 194, 197, 200, 213 Shakespeare, William, 24–5, 36, 41, 54, 116, 126–7 Shelley, Percy Bysshe, 45, 54, 67, 91, 136, 140, 167, 169, 205, 216 Short, Mick, 171 signification, 1–2, 4–5, 10–11, 13–14, 16, 41–2, 46, 61, 74–7, 84–6, 88–91, 93–4, 99, 103, 106–7, 109, 111, 115–23, 125–6, 129–30, 132, 136, 138–45, 149, 151–8, 160, 162–3, 166–9, 186, 192, 194, 196–7 Soll, Ivan, 214 speech acts, 131, 133, 171, 173, 199, 205, 216 Sperber, Dan, 1, 3, 6–8, 11, 16–33, 37, 41–4, 52, 105, 179, 187, 193–6, 200–1, 203–4, 206 Starobinski, Jean, 58 Stein, Gertrude, 210 Stevens, Wallace, 98, 210 Stevin, Simon, 75 Stokes, Adrian, 60–1, 206–7 Stoppard, Tom, 13, 140, 220 Strauss, Leo, 149 symbol, 32, 80–1, 109–11, 117–21, 123, 128, 158, 186, 189–90

237

trivium, 102, 212 tropes, 2–9, 23, 26, 28, 32, 40, 57–8, 61–108, 115, 117, 131–2, 140, 142, 145, 153, 158, 169, 192–3, 195, 197, 211 Turner, Mark, 211 Twain, Mark, 218 Vickers, Brian, 89, 99, 208, 210–11 Vološinov, V. N., 38, 77, 203, 219 Vonnegut, Kurt, 152 Warminski, Andrzej, 117–18 Waters, Lindsay, 148, 212 weak implicature, 23–5, 29, 40–1, 43, 53, 55, 177, 179, 182, 187 Weaver, Warren, 59 Wilson, Deirdre, 1, 3, 7–8, 11, 16–32, 41, 43, 50–3, 82, 105, 121, 140, 179–80, 182, 184, 187, 193–8, 201, 203, 206 Wittgenstein, Ludwig, 12–13, 60, 77–8, 121, 214 Wolff, Robert Paul, 208 Wordsworth, William, 6, 15, 80, 111, 140, 170, 181, 215, 219–20 ‘A slumber did my spirit seal’, 11, 176–95, 220 Lucy poems, 180, 184–8, 192, 219–20 The Prelude, 170, 177, 180, 184–5, 189, 191, 220 ‘Tintern Abbey’, 178 Yeats, William Butler, 89–90, 108 zero, 74–6, 208 Ziolkowski, Theodore, 146, 208