Fixing meaning

Fixing meaning Intertextuality, inference and the horizon of the publishable

Rachel malik

What is reading? Recent attempts to characterize it have conceded, and in many cases celebrated, its elusiveness as an experience. In Michel De Certeauʼs words, reading, unlike writing, takes no measures against the erosion of time (one forgets oneself and also forgets), it does not keep what it acquires, or it does so poorly, and each of the places through which it passes is a repetition of the lost paradise. [1]

Reading fails or escapes ʻordinaryʼ measure and record: what we acquire is largely lost, left behind, forgotten. The work of de Certeau himself and other cultural historians such as Roger Chartier has given substance to the ephemerality of reading. [2] The elaborations of modes of reading (aloud or silently, intensively or extensively, privately or collectively), of the social and psychic character of the reader, of the significations of design and typography have made valuable contributions to rendering reading rhetorical. In congruent work on other media (the new audience studies is one example) it is again the subject and her practices that emerge most strongly. [3] What remains elusive is the character of reading as a general process.

De Certeau assumes that reading, like writing (echoing Barthes), is an intertextual process, a work of textual transformation, a remaking of the text read. And across many fields of cultural study this is the presumption. But again, there is little attempt to specify exactly what this process is. ʻNot decodingʼ is the general answer, or not decoding in any simple singular sense. Yet the notion of en-de-coding – never definitive or determining – still overshadows the understanding of the reading process. Popular formulations, such as not decoding, recoding, resignification attest as much, and so in their different ways do the various attempts to introduce into discourse an uncodable dimension, of which Kristevaʼs semiotic is one example. There seems to be neither need nor will to conceive reading as also some other kind of process.

The derivation of intertextual reading from intertextual accounts of writing or production creates another problem. De Certeau naturally twins reading and writing: reading is the hazy inverse of writing. More commonly, the two collapse into each other: writing is ʻonlyʼ re-reading, reading is (re-)writing. Barthes insists on their likeness, though never (quite) their identity, but in many contemporary accounts the distinction has all but disappeared. Intertextuality has become almost exclusively associated (positively or negatively) with the fundamental unfixity of meaning and the freeing of the reader (now ʻactiveʼ, now a ʻproducerʼ) from the determinations of Romantic authority and structuralist logic, while the text itself dissolves into innumerable and ever-shifting contexts. [4]

What I want to propose here is an account of reading that takes intertextuality as the condition of language and signification, but formulates it to include a process very different from en-de-coding, inferencing. Inferencing, which treats the text as evidence for interpretation rather than the instantiation of meaning, is central to explaining the interpretation of the like-but-unlike texts and genres that constitute the intertextual text. But as a concept inferencing is also suspect, part of the tradition of Anglo-American pragmatics that is highly (and rightly) questionable, that after all includes Austin and Searle; and, for many, a tradition exploded by Derridean writing and/or citationality. [5] However, I will argue that inferential pragmatics identifies key weaknesses in intertextual accounts, whilst intertextual theories can transform the concept of inference as it is understood in pragmatics. My first two sections sketch the central axioms of intertextual and inferential theories, each in their own discourse, so foregrounding the conceptual mismatch between them: inference, as conceived in pragmatics, cannot simply make good the failings of intertextuality. Therefore the third and fourth sections stage a critical confrontation between them, in which each is examined through the lens of the other. Inferential theories disturb some of the central assumptions of intertextual accounts and force questions about the interpretative process that are usually ignored. Intertextual theories transform the field and concepts that pragmatics operates within, including inference. The account of reading that emerges assumes as its condition multiple interpretative possibilities, a position shared by inferencing and intertextuality, but it focuses on the ways in which the processes of production and reception order and delimit them, on how meanings are fixed (where signification is understood as dynamic between fixing and unfixing). This distinctive focus on fixing contests the assumptions of both traditions, identifying a shared problem. Neither pragmatics nor intertextual theories adequately considers how the processes of textual production fix meanings. Whilst studies of film, television and new media have taken the institutions and processes of production as part of their object, the same is not true of the book and publishing. Barthes dismantles Romantic authorship, but the author is only one figure and writing (in the narrow sense of composition) only one of the processes that constitutes the Work. [6] A Romantic understanding of textual production persists: composition remains the privileged process that orders all others. These are the usual terms in which the relations between ʻwritingʼ and ʻpublishingʼ are understood, implicitly or explicitly: other production processes (editing, design, marketing, production) exist to make the text public. These processes may modify, improve, diminish or destroy it, but the priority and precession of writing are never troubled – a weak concept of publishing as publication. In my final section, I will propose an alternative conception of publishing, as preceding and constitutive of the contingencies of both writing and reading.


ʻIntertextualityʼ is frequently banalized (comfortingly reclaimed as allusion or ʻthe study of sourcesʼ) or hyperbolized (the text is always a radical transformation of its pre-texts, always fundamentally fragmentary). Kristevaʼs neologism both builds on and transforms Bakhtinʼs dialogic, which in turn builds on and transforms Voloshinovʼs concepts of multiaccentual sign and verbal interaction. Voloshinovʼs critique of Romantic and ʻabstract-objectivistʼ models finds its counter in the multiaccentual sign which articulates a movement of convergence (users share the same language) and divergence (their social interests are differential, conflictual) in all language use: ʻdifferently oriented [class] accents intersect in every ideological signʼ. [7]

This movement is paralleled in the complex historical rhythms of verbal interaction: the ʻutteranceʼ (ʻhowever weighty in itselfʼ) is always part of a complex chain that responds to and anticipates (confirms and/or contests) others. Bakhtin retains and extends the problematic figuring of print as speech and sound (ʻpolyphonyʼ, ʻorchestrationʼ, ʻdouble-voiced discourseʼ), but also stabilizes and transforms the multiaccentual sign into an intersection of languages that instantiate the heteroglossia. [8] The dissonant dynamic of Voloshinovʼs model is reconfigured in the relations between heteroglossia and all the attempts to posit and enforce a unitary language. Working within the contradiction of language as both mine and the otherʼs, the dialogic word becomes the dynamic condition of language and meaning, always positioned between and participating in the relations between this particular use and others, previous and prospective. Kristeva draws out the full implications of the text as process and production (ʻproductivityʼ) and the role of a constitutive textual context. Her focus on the speaking subject – a psychic subject who is constituted in and against language – aligns subjectivity with textuality, as co-constitutive processes. Her first formulation of intertextuality in Le Texte du Roman marks a distinctive displacement of the dialogic, definitively formulating the speaking subject within the text and giving a new emphasis to the relations between textual practices. The text is a permutation of texts, no longer voices. [9] The fifteenthcentury proto-novel she discusses incorporates a wide range of texts and genres: moral precepts, Latin citations, epic and courtly love poetry, and blazons or street cries. These have particular meanings and values within the General Text (Culture), but their conjunction in a new space produces new meanings: texts and genres are permuted, resignified.

Here, in contrast with contemporary accounts of intertextual reception, there is no collapsing of writing into reading. Writing is clearly a process that includes reading, most explicitly in Kristeva, and in Bakhtinʼs account of evaluation within meaning production. But there is no dissolution of the one into the other. Within this tradition there is also a commitment to explaining the processes of textual production. Most importantly, these accounts develop a strong concept of textual context. Text and context are dynamically bound within history and society as a whole. The text is at once inseparable from context, present within it and constitutive of it, and distinct: it permutes what it configures. Further, the complex chains of verbal interaction, the heteroglossia and the General Text all articulate the contestation between dominant and subordinate social forces, making contexts multiple and socially conflictual. This emphasis on ʻthe text of history and societyʼ and conflictual social relations inscribes an important difference between this lineage and the apparently similar concept(s) of iterability and citationality, which are neither the production nor the effect of conflictual social relations and socially contested meanings. [10] The character of history is also distinctive. In Voloshinov, Bakhtin and Kristeva the sense emerges – often implicit and never fully explicated – that historical meaning is temporally complex. [11] Voloshinovʼs distinction between immediate and ʻbroaderʼ contexts, Bakhtinʼs accounts of the historical forms of the dialogic and the heteroglossia, and Kristevaʼs text, composed of contemporary and anterior signifying practices, open up the possibility of theorizing text and context as historically complex: meanings can endure. This contests the localizing tendencies of historicism, where text and reader are conceived in terms of an absolute otherness and context is always and only change. Whilst contemporary theories of intertextual reception tend to focus almost exclusively on a process of unfixing, in Voloshinov, Bakhtin and Kristeva meaning is always understood also as a process of fixing. Bakhtinʼs account of the poetic and monologic and Kristevaʼs account of the relations between semiotic and symbolic instance how within this lineage meaning is always a dynamic between fixing and unfixing.

Intertextual theories of production do not share the core problems in intertextual accounts of reception, but there are inherent difficulties facing any attempt to formulate intertextual reading. Kristevaʼs account of the novel and its signifying logic is shaped by the gram, the ʻsignʼ finally set loose and surpassed, and by a specifically modernist writing. As with Bakhtin, aesthetic preferences and historically distinctive rhetorical strategies become conflated with the general conditions of language and meaning. The ambiguous status of Bakhtinʼs ʻdialogicʼ as condition of language and as specific cultural and political value is a topos. [12] But beyond this, there is a contemporary tendency to formulate dialogism or intertextuality in general in the terms of a small set of its historical varieties, where the rendering of multiple ʻvoicesʼ or languages is at its most concentrated and explicit. Barthesʼ specifically modernist writing exposes the limited plural of realism in S/Z, but also effaces narrative (rendered as a code of potentially paired actions – opening and possibly closing – that seemingly have little or no cumulative or connective effect). [13] A further move treats these privileged varieties not as dissident practices but as the empirical reality of all signifying practice(s). Fiskeʼs formulation of the mass broadcast text as producerly (the televisual equivalent of Barthesʼ writerly) is one instance. [14]

Radical transformation, intertextual ʻdensityʼ, and explicit and extensive fragmentation are better understood as elective forms of a general condition of intertextual variation. In many, perhaps the majority of texts, intersection and permutation are ordered by a dominant signifying practice (usually a genre). Gothic is a frequent visitor to nineteenth-century realist novels, but the uncertain moment of the fantastic usually give way rapidly to a reading that understands the Gothic signifier in realist terms. This suggests the need for a stronger (intertextual) concept of narrative and, more generally, for a reassessment of genre. This occupies a problematic place in intertextual theories which challenge any formulation that posits a one-to-one relation between text and genre. Bakhtin and Kristevaʼs analyses suggest a radical redefinition of historical ʻgenresʼ (such as the novel), now conceived as specific permutations or of languages (varieties, registers, genres). But Bakhtin most explicitly is also in flight from the implications of genre as an institutionalized conjunction of signifying possibilities. ʻWe speak only in definite genresʼ, he says, nevertheless resisting the stability of situation and subject that this formulation presumes. [15]

Pragmatics, inference and relevance

Pragmatics is predicated on the differences and discrepancies between linguistic form and the meanings of utterances; between sentences composed exclusively of linguistic properties and the meanings produced by them in specific situations. Various questions follow. Are these differences contingent or constitutive features of communication and how might we explain them? If utterance meaning is not wholly encoded in the utterance, is it encoded elsewhere, for example in features of the situation? If so, how do hearers decode such meanings, which must involve non-linguistic knowledge? [16] Are there, then, ʻpragmaticʼ rules or principles governing utterance interpretation? But, on the other hand, is non-linguistic meaning encoded at all? Are there other processes that might govern utterance interpretation? Are these specific to communication or more generally cognitive? The constitutive gap between coded meaning and interpretation is formally familiar to much of contemporary cultural and literary theory but it does not necessarily follow that interpretation is fundamentally unconstrained, ʻopenʼ and various, as is so frequently assumed. How meaning is disambiguated, determined or resolved is the central question.

The relations between language form and language use furnish the materials for a wide range of topics and programmes in philosophy and linguistics. But within this field, it is possible to delineate a strong pragmatics, pledged to describing and explaining communication, in which accounts of inferencing are central. Within this tradition, the gap between form and meaning is constitutive. Austinʼs and Searleʼs speech acts and Paul Griceʼs writings on non-natural meaning and implicature are two canonical examples. In inferential accounts, utterances provide evidence for an interpretation, and in strong accounts such as Dan Sperber and Deirdre Wilsonʼs Relevance Theory – the centrepiece of this discussion – no more than this. This clearly gives distinctive content to the other concepts it presumes and formulates. Context, defined as the knowledge that speaker–hearers use in communication, specifies and determines the utterance; and its deployment is central to any account of the interpretative process. If the utterance is defined as radically underdetermined by its linguistic meaning (as it is in Relevance), the range of possible interpretations necessarily expands, and any account of the process must specify how a particular set of interpretations is selected from a wider range. This usually leads to a strong distinction between utterance meaning and speaker meaning. For Grice, as for Sperber and Wilson, the goal is to explain speaker meaning: the difference between ʻthe meaning in general of a “sign”ʼ and ʻwhat a particular speaker or writer means by a sign on a particular occasionʼ. [17] The speaking or interpreting subject of pragmatics is polarized in familiar terms: s/he may be always-already social, but in the tradition of strong pragmatics I am exploring s/he is fundamentally an individual.

Inferential accounts of communication treat utterances as evidence for meaning (rather than the encoded instantiation of it), evidence which is put together with other evidence (contexts) as ʻpremissesʼ in order to derive conclusions, better known as interpretations. Grice, treated by many as the founder of inferential approaches, coined the term ʻimplicatureʼ to capture the difference between logical and natural languages. Implicature has a relation to the verb imply but is distinguishable from the counterpart logical category of ʻinferringʼ. [18] Pragmatics follows Grice in arguing that inferencing cannot be characterized as a formal deductive procedure and is more akin to induction. But there is a fundamental difference between formal logic as such and pragmatic inference, where the ʻpremissesʼ that the utterance and context supply are routinely subject to radical disjunctures of form and meaning. At the same time, accounts of inferencing are everywhere shot through with the like–unlikeness of logic and communication. For Grice, the nonsequiturs, irrelevancies, redundancies and ellipses that characterize ordinary conversation are, in the main, only apparently disorderly. In truth, they are evidence of a principle, the Cooperative Principle, which underlies ordinary conversation:

Make your conversational contribution such as is required, at the stage at which it occurs, by the accepted purpose or direction of the talk exchange in which you are engaged. [19]

This overarching principle governs four maxims that connect the speaker with the spoken. These pertain to the quantity of information the speaker supplies (neither too much nor too little), its truthfulness (ʻtry to make your contribution one that is trueʼ), its relevance (ʻbe relevantʼ), and how it is said (the speaker should avoid obscurity, ambiguity and be brief and orderly). [20] Despite the imperative form, Grice is well aware that speakers do not always or often behave in this way. The maxims are not an elaboration of the Principle: it is the Principle itself that, actually and not ideally, governs conversation. What a speaker says may be literally untrue or irrelevant, but the hearer assumes that the Cooperative Principle is still operating and produces an interpretation in accordance with it:

Tony: Whereʼs Richard?

Helen: Thereʼs a silver Vespa parked outside Markʼs.

Helenʼs reply does not directly answer Tonyʼs question and seems irrelevant. But if Tony assumes that Helen is being Cooperative, he will access the knowledge that Richard has a silver Vespa (which Helen assumes he knows) and produce the implicature that Richard may well be at Markʼs. There is no conventional (or coded relation) between Helenʼs utterance and Tonyʼs interpretation (conclusion). The interpretative process is inferential. Griceʼs definition of conversational implicatures is structured by a contrast with a definition of logical deduction. A conversational implicature can be cancelled (unlike a logical inference); it can be true even though the utterance which proposes it is false (and vice versa).

Whilst Grice is suspect even within the limits of post-analytic philosophy, his influence on pragmatics has been considerable, and a wide range of modifications and developments have been proposed. Most of these have operated broadly within a Gricean paradigm, but Sperber and Wilson, whilst acknowledging Griceʼs contribution, broke radically with his model, in Relevance: Communication and Cognition, a book which has excited considerable interest, much of it highly critical, since it appeared in 1986.21 The book is characterized by a rhetorical boldness – both in the force of its own claims and in its outright dismissal of many of the sacred cows of pragmatics. More than fifteen years later, Relevance theory remains highly controversial: the answer to communication, cognition and ʻeverythingʼ, or the emperorʼs new clothes. Relevance is the focus here because Sperber and Wilson claim that inferencing is the central process in all interpretation, whilst most other accounts, Griceʼs included, view it as contingent. The theory aims at a full explanation of all types of verbal communication and considers a wide range of utterance types, ʻparasiticʼ or ʻliteraryʼ as well as ʻordinaryʼ, in a bid to demonstrate its power. Relevance therefore provides a single strong model of inferential interpretation through and against which intertextual accounts can be read.

Most simply, Relevance proposes that one process (a specific type of inferencing) and one principle (relevance) can explain our understanding of anything from the smell of gas to a complex metaphor. [22] Semiotics usually treats language as a model for other signifying systems, but Sperber and Wilson classify linguistic communication within a taxonomy of cognition that foregrounds its distinctiveness. Unlike other phenomena which may be cognized, it is intentional and ostensive; unlike other forms of deliberate and ostensive ʻstimuliʼ that communicate, it deploys a code. However, verbal communication involves more than the decoding of a linguistic signal: there is a gap between the semantic representations yielded by decoding and the thoughts communicated by them which en-de-coding cannot explain. Utterances are always interpretations of thoughts, and the hearerʼs interpretation is always an interpretation of an interpretation. Communication can at best guarantee a similarity of representations between communicator and audience and ʻfailures … are to be expectedʼ. [23] This proximate and contingent characterization of communication challenges existing accounts of inferencing in their predominantly Gricean mode. There, inferencing is contingent and subordinate to decoding, which also serves as a model: rules and premisses are conceived as shared and applied. Sperber and Wilson contest this formulation. Conceived as evidence or premiss, an utterance has a multiplicity of possible conclusions (interpretations). How are some and not others produced as interpretations; why does interpretation ʻstopʼ? For Sperber and Wilson, it is the conclusions that the speaker intends that the hearer must identify.

The rather punishing technical detail of Sperber and Wilsonʼs characterization of inference is not relevant here, but a few points require noting. Inferencing is ʻless a logical process than a form of suitably constrained guessworkʼ. [24] It is a general thought process that is spontaneous, nearly instantaneous and unconscious; it is constrained by speed and, in the case of verbal communication, the helpfulness of the source. The process should be judged in terms of success or failure and not validity or invalidity, a move in line with Austin and Searle and against Grice. Whilst not deductive as a whole, it does make use of deductive rules which effect an economy of information storage.

An account of the rules of inference, however explicit, does not explain why individuals process information or how they reach one set of conclusions rather than another. It is relevance that governs the inferential and interpretative process. But ʻrelevanceʼ bears little or no resemblance to any ordinary sense of the word. For Sperber and Wilson, humans are efficient information-processing devices whose overarching aim is to improve their knowledge of the world. Relevance governs the efficient humanʼs cognitive relations with the world: something is relevant in so far as the cognitive gains justify the ʻcostsʼ. From this standpoint, old information is not worth processing and completely new information requires too much expenditure for too little achievement. But new information proc-essed in relation to old can be cost-effective, giving rise to modifications of context or ʻcontextual effectsʼ which are used to measure achievement–efficiency relations and define relevance: ʻother things being equal, the greater the contextual effects, the greater the relevanceʼ. The first Principle of Relevance states: ʻHuman cognition tends to be geared towards the maximization of relevance.ʼ The Second or Communicative Principle is grounded in the first: ʻEvery act of ostensive communication communicates a presumption of its own optimal relevance.ʼ [25] The utterance has to be relevant ʻenoughʼ to be worth the effort of processing, but no more than this. Sperber and Wilson distinguish themselves from Grice very explicitly. ʻUnderstanding and being understoodʼ is the only ʻcommon goalʼ that speaker–hearers share. Griceʼs maxims are a set of norms which communicator and audience must know and follow. Relevance is not a general principle but a particular presumption which is communicated by and about every particular act of communication. It is not simply ʻfollowedʼ and could not be violated. [26]

Interpretation is a process of hypothesis construction. Which of the many possible assumptions communicable by the utterance or text – Sperber and Wilson make no distinction – are actually intended by the communicator? Relevance governs every aspect of the process. The choice of contexts is determined by relevance and not, as we might intuitively assume, the other way round. Context, the knowledges that hearers deploy in interpretation, is not fixed but a ʻvariableʼ. There is an immediately given context – those assumptions mobilized or/and produced by the previous utterance interpretation – but this is merely an initial context which can be variously developed. Accessibility of context is likewise determined by the effects–effort ratio of relevance. As soon as an interpretation or ʻconclusionʼ confirming the initial presumption of relevance is produced, the process of hypothesis construction stops.

Decoding subserves inferencing. An initial decoding of the utterance produces a number of semantic representations that correspond to all its possible senses. But these are ʻincomplete logical forms i.e. at best fragmentary representations of thoughtsʼ, and require completion into a fully propositional form. This process is inferential and includes reference assignment, disambiguation and the specification of vague terms. Sperber and Wilson coin the term ʻexplicatureʼ to capture the result of this process. An explicature is an explicitly communicated assumption ʻthat leads to an overall interpretation which is consistent with the principle of Relevanceʼ. From Grice they retain the term ʻimplicatureʼ, which is any assumption which is intentionally but not explicitly communicated. The speaker assumes the hearer can access certain assumptions, use these as a context in which the explicatures of the utterance are processed, and derive particular conclusions:

Peter: Would you drive a Mercedes?

Mary: I wouldnʼt drive ANY expensive car. [27]

The main explicature of Maryʼs response does not directly answer Peterʼs question. But it does allow Peter to access encyclopaedic knowledge about cars, including expensive ones, which includes the information that a Mercedes is an expensive car. This could yield the implicature that Mary wouldnʼt drive a Mercedes and, perhaps, not a Rolls Royce either, and that she disapproves of ostentatious displays of wealth.

Relevance extends the scope of pragmatic enquiry to the zone of what is explicitly ʻsaidʼ, the interpretation of which can no longer be assumed. What is said is radically more underdetermined than pragmatics generally and Grice in particular acknowledge. As Robyn Carston, a key proponent of Relevance, puts it: ʻnot only does linguistic meaning underdetermine what is meant and what is said underdetermine what is meant but ʻ[l]inguistic meaning underdetermines what is saidʼ. [28]

Inference reads intertextuality

Inferential theories expose the intellectual inertia at the heart of intertextual theories of reception: the treatment of reading as rewriting, an inability to think beyond en-de-coding, however critically it is reconfigured, or to formulate explicitly how contexts operate in the interpretative process. By contrast, pragmatics understands the processes of production and interpretation as clearly distinct. Sperber and Wilsonʼs characterization of communication as ostensive–inferential crystallizes the difference: utterance production involves the making manifest of an intention to communicate; interpretation involves the identification of a communicative and informative intention. These distinct processes are united by relevance as principle. My point is not that Relevance is definitive, far from it, but that it formulates interpretative procedures so rigorously. Inferential interpretation is always an active process of production in very specific terms. The procedures of reference assignment, disambiguation and enrichment ʻadd toʼ the utterance as evidence. ʻNewʼ information or knowledge, not coded in the utterance or its explicatures, is produced by conjoining the ʻexplicatedʼ evidence of the utterance with contextual assumptions to produce implicatures. If inferencing is admitted into the process, it follows that in an important sense the active productions of reading are banal, because they are always-already present in the process: any attempt to formulate critical or dissonant reading must start from this assumption, rather than treating it as a contingency.

Intertextuality formulates the meanings of a text as simultaneously underdetermined – its meaning are never self-contained or fully resident within it – and overdetermined – the text and its meanings are an intersection and permutation of multiple signifying practices. Codes are rendered in one sense less stable – strict repetition or identity is impossible – but it is still coding, best understood (if imperfectly) as recoding which shapes meaning, even though its semantic effects are multiple, potentially contradictory and unpredictable. [29] Yet there are processes that not even a reconfigured decoding can explain. How do readers interpret the relations between texts within a text, specifically the differential and competing values they inscribe? How do readers constitute the relations of similarity and difference between the text being read and other texts? In the film Wall Street, the corporate trader–raider Gordon Gecko shares his credo with a rapt audience of shareholders: ʻGreed is goodʼ. One of the senses of this utterance is that Gecko believes what he is saying. His previous statements and actions are immediately available contexts that leave no room for irony. But the utterance means more than this. Gecko is also implying that his audience shares his belief, though they do not wish to acknowledge it openly, and that they should go public. These are implicatures and are not coded into the utterance. They rely on the viewer understanding that what Gecko says is controversial and that it overturns a familiar moral doxa which is present and displaced in his evangelistic discourse. Such implicatures are proposed by the intertextual character of the utterance. This suggests an intimate relation between (at least) a certain class of implicatures and intertextual interpretation, but also that some implicatures are properly speaking utterances – or, better, texts – and should be treated as such. The hearer or reader mobilizes or constructs an intertextual relation between two or more utterances. The oppositional or contesting utterance that is a central category in Bakhtin, Kristeva and Barthes is a ʻnegationʼ (understood here in a discursive rather than a grammatical or logical sense) of another utterance, which is proposed as a strong textual context for constructing implicatures.

Intertextuality, in redefining both text and its contexts (of both production and reception) has blurred their boundaries, making it increasingly difficult to identify the text that is read either as concept or/and as empirical object. The textualization of context, perhaps currently most visible within new historicism, and the textualization of the reading subject as a ʻsiteʼ of textual knowledges and their transformation are certainly valuable but have also led to an impasse in accounts of the interpretative process. This is nowhere more acute than in ʻnewʼ audience studies. Is it possible to distinguish between text and context? (And should we even want to?) For some, this blurring of the boundaries is a welcome last nail in the coffin of textual determinism. In the words of Lawrence Grossberg,

Not only is every media event mediated by other texts, but itʼs almost impossible to know what constitutes the bounded text which might be interpreted or which is actually consumed. [30]

And there are those, for example David Morley, who are unhappy about the slippage and ʻthis new emphasis on intertextualityʼ, and concerned lest text – as concept – be ʻdissolved into its readingsʼ. Grossbergʼs formulation of intertextuality transposes an abstract definition – intertextuality as the ontological condition of text and textuality – to the plane of the concrete: the particular text and the (particular) readerʼs interpretation, which Morley repeats in the negative. Morley also seems to identify intertextuality with an excess of polysemy, an uncritical concept which apparently cuts loose both textuality and reader from social relations. [30] Strong inferential accounts focus a particular set of questions about text–context relations. If all utterances are to be understood as evidence from which hearers derive ʻconclusionsʼ (interpretations), then the process must specify how and why certain knowledges, or contexts, are mobilized in the interpretative process and others are not, why certain interpretations and not others are produced, and why the interpretative process ʻstopsʼ – given the multiplicity of inferences that can be derived from any utterance (the last is central to Relevance). Sperber and Wilsonʼs account of the process of context selection and use is dynamic, ʻopen to choices and revisions throughout the comprehension processʼ. [32] The interpretation of contestational utterances is a clear instance of this dynamic aspect: the implicature that makes possible the classification of the utterance as a contestation is constructed during the interpretative process.

At the most general level, Relevance provides an account of how text can and does ʻbecomeʼ context via the interpretative processes of explicature and implicature, but the two always remain distinguishable. Sperber and Wilson also draw attention to the status of the knowledge deployed in interpretation: the relative ʻstrengthʼ or ʻweaknessʼ with which assumptions are held by the hearer. Utterances may strengthen or weaken, to the extent of contradicting and cancelling, certain contextually accessible assumptions. This is suggestive for thinking about the relations between contexts within specific acts of reading, central to intertextual interpretation. For example, a plurality of markers of a particular genre can make possible not only the mobilization of various assumptions such as ʻthis is a romanceʼ and so forth but, because of their ʻdensityʼ, strengthen that context. A genre shift can act to mobilize another context and may weaken an existing one. Sperber and Wilson are also specifically interested in the differential ʻaccessibilityʼ of contexts within a given situation of utterance (accessibility formulated in terms of the effects–effort ratio of Relevance. But even if we bracket their formula, ʻaccessibilityʼ has a general salience for any attempt to theorize interpretation as intertextual. If we equate the totality of a readerʼs textual knowledge with context, we are indeed faced with the question of the text ʻunboundʼ; but if instead we insist on the need to hierarchize a readerʼs knowledges, asking which are most, more, quite, less likely to become contexts within a particular situation of reading, we specify contingency as a gradient rather than a single term opposed to a factitious necessity.

Intertextuality reads inference

I have so far sought to represent inferential accounts within their own terms, indicating only where Relevance appears to depart from the axioms of a language philosophy that we now know first and best through a Derridean intertext. The overarching concept of ʻcommunicationʼ, inference itself, raising the spectre of a logical model of natural languages, the absence of textuality as a fundamental explanatory concept, and interpretation understood as resolution, where intention is not reasserted but simply and pre-critically assumed – these are the most familiar markers of this discourse. The ʻplain speakingʼ of Sperber and Wilson, with its echoes of Grice and Oxford philosophy, and the cognitive lexicon of Relevance, where utterances are ʻstimuliʼ and human beings are ʻefficient information-processing devicesʼ, can easily be dismissed as tendentious. I am trying to show that intertextual interpretation can only be adequately understood if it encompasses inferencing, but also that the concept as it exists in Relevance (and in other less extreme accounts) demands substantive critical reconstruction.

Both Grice and Sperber/Wilson acknowledge only two conditions of communication: a language (conceived as a code – a single grammar – or a set of conventions according with majority usage) and its individual users. There is nothing else. This concept of language, which is implicitly a uniform, inclusive national language, is fatally vulnerable to a Bakhtinian critique, in the light of which it must be reformulated as heteroglossia. There is no single system or set of agreed uses, only multiple and conflicting modes of signification. Second, there is no place here for genre, register and other kinds of ʻcodeʼ, understood as textual practices which shape utterance meaning and the interpretative process, even down to its smallest units. The disavowal of the textual is common in Anglo-American pragmatics. Griceʼs maxim of manner is the marker of this within his theory, the place where the how of meaning is confined and permitted to have an effect. [33] Relevance seems to be different: there is no zero degree of style, no naive distinction between form and content. ʻStyleʼ is a natural property of utterance arising ʻin the pursuit of relevanceʼ. But this manoeuvre masks the same resistance to the textual, generally evidenced in the kinds of short examples explanatory pragmatics prefers. In this way, genre and other codes are rendered invisible, though not always, as this example from Relevance, contrasting three possible responses to Peterʼs question, unwittingly attests: (a) Peter: Is Jack a good sailor? (b) Mary: Yes, he is.(c) Mary: ALL the English are good sailors.(d) Mary: Heʼs English. [34]

Sperber and Wilson never consider that the phrase ʻa good sailorʼ is potentially ambiguous (not only ʻgood at the skill of sailingʼ but ʻdoesnʼt get seasickʼ) – surprising given the general pragmatic attention to ambiguity and its frequent deliberate use. This is not a case of oversight: there is a code at work backgrounding the ambiguity. It is not of a kind that Sperber and Wilson would acknowledge, but it is a code nonetheless, one that conjoins nationalities with particular skills or attributes, positive or negative: Americans have no sense of irony, Italians are sentimental, and the English are… This cancels the seasickness sense of ʻgood sailorʼ: it is not an adequately strong attribute. Further, the name Jack mobilizes the figure of ʻJolly Jack Tarʼ, the personification of English seafaring excellence, and strengthens the first sense further. The ways in which textuality both produces and fixes meaning are disavowed. The affirmative in Maryʼs last two replies is certainly an inference, but one governed by textual relations.

The Relevance understanding of both context and the subject is shaped by the same liberal discourse that resists the textual as the inscription of the constitutive sociality of language. It is a discourse shared with much pragmatic thinking, but as in so many things Relevance goes one further. Communication is conceived as a coincidence of self-interest: the speaker wishes to communicate, the hearer wants to recognize her communicative intention. Inferential theories are naturally predisposed to assume that knowledge is significantly shared by speaker-hearers, who therefore converge on premisses and conclusions. This is Griceʼs implicit position, but it is not available to Sperber and Wilson, who insist on the idiosyncrasy of individual knowledge. ʻIdiosyncrasyʼ is the concept that is supposed to register the complexity of knowledge relations, while keeping culture and the social from making too much difference. But Relevance is not consistent in its treatment of individuals as predominantly idiosyncratic. It minimizes the possibilities of knowledge differentials and variations by dwelling on the special case of speaker–hearers who are alwaysalready intimates: a couple at a party, a picnic, a walk in the country and so on. This pervasive intimacy banalizes the social and cultural heterogeneity of the knowledges mobilized in interpretation. A speaker simply knows that the hearer knows what osso bucco is (this is a real example) and can make a pretty reasonable guess that s/he has read Sense and Sensibility (so is this). The interpretative impact of socially and culturally specific knowledges is therefore virtually invisible in accounts of the interpretative process. Peter and Mary, the imagined suppliers of examples for Relevance, are a very particular kind of subject. Mary is a lawyer, Peter is a surgeon. Both of them cook; they enjoy Italian food and have a favourite Italian restaurant. They take walks in the country, holiday in rural France and can ʻget byʼ in French. They read Jane Austen and are familiar with the Romantic valorization of nature. Peter and Mary are not ʻeverypersonʼ any more than they inhabit a neutral ʻeverydayʼ. They are bourgeois subjects whose cultural knowledges and values are explicitly English, and again, in a class-specific sense, European. Through them, social specificity is naturalized, and the knowledges and codes of a particular class fraction are displaced as determinants of interpretation by a process that naturalizes similarity. [35] Voluntary and equal relationships between speaker–hearers are offered as a synecdoche for the totality of speaker–hearer relations.

Inferencing needs to be situated within an intertextual model where context and subject are constitutively social and textual, and social relations are conflictual. Within intertextual theories, the concept of culture as competing sets of signifying practices opens up the possibility of registering and examining the complexities of social and cultural identity in a highly nuanced way. But this complexity cannot be reduced to idiosyncrasy. Inferencing as process has to be configured within a constitutively contestational culture: one that is always dialogic, always producing difference and antagonism as well as commonality. Relevance conceives the overarching cognitive task as the adding of more, more accurate information that is more easily retrievable and above all consistent. [36] But as social and – Kristevaʼs distinctive contribution – psychic subjects, we may and do resist knowledge in a multiplicity of ways, including knowledge that conflicts with or contradicts what we know. The repeated processing of ʻold informationʼ (which Sperber and Wilson cannot see the point of processing) should come as no surprise either. Nor can Relevance handle the commonplace that we hold all kinds of assumptions that conflict with or contradict one another. The relations between ʻassumptionsʼ – or better ʻstatementsʼ – are not governed by a singular set of logical rules but by the discourses that constitute them: any statement is always-already ordered in terms of its relations with others. [37] What is important are the boundaries between discourses, the rules by which statements are included or excluded and ordered as series. Within such a framework, there is every reason why subjects will entertain ʻassumptionsʼ that conflict or contradict one another without resolution. And every reason why subjects will converge on certain patterns of inference and interpretation and diverge on others. Convergence is not the effect of a naturally shared knowledge, any more than divergence is evidence of mere technical failure.

Relevance is also committed to the pragmatic common sense that interpretation is a process of reso-lution, a process that appears to be in contradiction with intertextual theories. Yet if inferencing is, as I am arguing, central to (intertextual) interpretation, the question of how and why interpretation ʻstopsʼ or continues cannot be ignored. For Sperber and Wilson, the conjunction of intention and relevance secures resolution: interpretation stops when the speakerʼs informative intentions have been identified, which in turn coincides with sufficient contextual effects. This reliance on intention has been challenged from within pragmatics, but the charge is usually one of extremism: intention as such is left undisturbed. As hearers and readers, we can and do ascribe intentions to writers and speakers, but these cannot be separated from the text as pragmatics generally supposes. A plethora of signifying systems constitute the intentions that texts inscribe. The intentionalism of Relevance derives from its individualism but also serves to control the anarchic interpretative consequences of that basic commitment. In conditions of radical textual instability, the notion of an originating subject can underwrite the possibility of communication. Intention is the substitute for the constraining force of social relations inscribed in the conditions of communication – in the text, in the subject, in the situations and practices of reading. All these modalities delimit and order the interpretative process, suggesting why it might stop or indeed continue, but they are unacknowledged by Relevance. Communication is a not a subset, however specialized, of cognition; rather, cognition is a mode of ʻcommunicationʼ.

Inference is indeed central to interpretation, which cannot be conceived exclusively as some critically reconstituted form of en-de-coding. The concept is clearly necessary to understanding how readers interpret the competing and differential values accorded to the texts within the text being read, and the like–unlike relations between these texts and others. However, inferencing is not governed by a singular logic and a singular pragmatic (or cognitive) principle. Patterns of inference are shaped by genres and discourses, not ʻpremissesʼ and ʻconclusionsʼ. But if no pattern of inference, no interpretation, is necessary in this sense, nevertheless contingency is graduated. Some interpretations are highly likely, others barely possible, and a range of possibilities lies in between. Interpretative processes require situation within a model which takes account not only of the well-furrowed ground of reading knowledges, practices and situations, but of the processes and practices of textual production. Grice and Sperber/Wilson, in common with much explanatory pragmatics, focus on speech, Bakhtin and Kristeva (and Barthes) on writing. But this predictable divergence masks a common underlying repression: print, or, to avoid the suggestion of a ʻneutralʼ medium, publishing as a constituent in both the production and interpretation of meaning.

Shared problems and the horizon of the publishable

Grice and Sperber/Wilson are only doing what comes naturally to pragmatics. The canonical speech situation, positing two speaker–hearers who are co-temporal, co-spatial and co-present, is the default condition and paradigm of communication and is justified in familiar terms. Speech (in the terms defined by the canonical situation of utterance) precedes ʻwritingʼ and presumably all other media; it is the ʻparentʼ of all other forms. Speech, or more specifically conversation, is both natural and normal: it is the everyday, spontaneous, mode of communicative practice to which all other kinds can be opposed. In pragmatics, there is no understanding of speech, writing and print as historical categories and practices whose constitutive relations are now radically interdependent. Print technology made possible, and its development within specific conditions made actual, a process of linguistic standardization that has had massive impact on the practices of speech and writing. There are many forms – the novel and the newspaper, to take the least contentious examples – as well as a plethora of genres, which are precisely the consequent of this conjunction of technology and historical conditions. Intertextual accounts of meaning expose the fundamental flaws of the canonical speech situation: there are never only two ʻspeaker-hearersʼ, even in conversation. By the same token, utterances cannot be punctually timed or mapped and their meanings elicited according to the local co-ordinates of the ʻnowʼ and the ʻhereʼ. The meaning of any utterance is only explicable in relation to the utterances that precede and might succeed it, binding it to a complex chain of ʻthensʼ and ʻtheresʼ.

Beyond this, there is a profound naivety in the categories of co-temporality, co-spatiality and copresence, which are all objectifications of a presupposed commonality between speaker–hearers and their communicative ʻwillingʼ. Co-temporal and co-spatial speaker–hearers may have radically different understandings of the spatio-temporal location in which they converge. The complex of social and cultural modalities which shape identity – including politics, class, gender, generation, religion, a sense (or sense of loss) of ʻhomeʼ, with all the complexity that migration introduces – intersect with one another and with a set of dominant histories and geographies. Living in England, I cannot avoid ʻroyal timeʼ any more than the Christian time in which it is imbricated. This is the ʻbaselineʼ complexity of face-to-face communication, which the canonical speech situation evades by displacing its assumptions of commonality onto the ʻnoncontroversialʼ coordinates of the co-temporal and the co-spatial.

Theories of intertextuality have congruent problems. Bakhtinʼs work on ʻspeechʼ genres makes some general distinctions between speech and writing, and Kristeva is interested in the transposition of the spoken into the written, but in neither do print and publishing emerge as specific technologies, processes and practices. [38] Barthes perhaps does most to shatter the private autonomy of writing, insisting on the text as a process everywhere scored by history and society. Here, the book is at once conceived, correctly, as an attempt to fix meaning and challenged as a set of boundaries, foreclosing its possibilities. Barthesʼ ʻworkʼ is aligned with the book, ʻoccupying a part of the space of books (in a library for example)ʼ or in bookshops, catalogues and exam syllabuses. [39] Text, as method, cuts across works, ʻcannot stop on a library shelfʼ. For Derrida,

The idea of the book which always refers to a natural totality is profoundly alien to the sense of writing…. If I distinguish the text from the book, I shall say that the destruction of the book, as it is now under way in all domains, denudes the surface of the text. [40]

The reality, resisted here but not effaced, is that the production of meaning is always and simultaneously a movement both towards and away from its fixing, however provisional and unstable: text after all ʻcuts acrossʼ, divides, even where it does not respect canonical boundaries. But this fixing which must be part of intertextuality, of writing, is usually neglected or refused. The book, reconceived, is one of the sites where these relations between fixing and unfixing can be explored. The book is no longer imagined as boundary and stasis that seeks to constrain the ever-multiplying volume and movement of language, but as a site where publishing processes – ʻwritingʼ, editing, design, marketing, production – intersect and conflict. Barthes approvingly cites Mallarmé, ʻwanting the audience to produce the bookʼ, but authors donʼt write books: composition is only one process. [41] Barthes insists, quite rightly, that we rethink the temporality that governs the relations between writer and text. But the temporal relations which govern how we think about writing and publishing also require rethinking. Publishing, or rather the horizon of the publishable, precedes and constitutes both what can be written and what can be read.

The book has become a renewed site of theoretical interest over the last twenty or so years – within cultural history, literary studies and also in the narrower fields of book history and textual studies, specifically in debates about the theory and practice of textual editing. Alongside these is Gérard Genetteʼs Paratexts, the only sustained attempt to think the book, or better the edition, as an intertextual apparatus. [42] The limits of these engagements are instructive. Genette offers a richly nuanced account of the pragmatic function of everything from titles, signatures and prefaces to formats, series, epigraphs and notes, but his basic distinction between text and paratext marks his preestablished limit. Whilst the paratexts of the edition operate as explicit reading contexts that orient and adapt the text for different readerships, the text itself is ʻdumbʼ. [43] Residual Romanticism creates an ideal separation between text and the market: the paratext is not only threshold but boundary. Within this formula, publishing is necessary modification or ʻadaptationʼ (his preferred lexeme), acting on a text that always precedes it: publishing is publication.

This absence of process is not a feature of the ʻnewʼ textual studies, where the ʻnewʼ signifies both new historicism and, more usually, a generic poststructuralism. Where once critical editing practices were underpinned by an anxious narrative of textual decline and ʻcorruptionʼ – the lamentable but inevitable wear and tear of reproduction and circulation – the territory is now occupied by the topoi of unfixity and contingency. [44] Thus, a symptomatic critique of traditional scholarly editing may expose the variational crux (is Hamletʼs flesh too ʻsolidʼ or too ʻsullied?) as the carefully managed moment which acknowledges as atypical and aberrant what is in fact the everyday condition of meaning and interpretation. Alternatively, textual editing seeks to find ways of marking contingency in editions that represent the text as ongoing process, foregrounding the apparatus and the editorʼs status as ʻreaderʼ (rather than agent of purgation). Such work is valuable in that it foregrounds the constitutive role of editing practices and their historicity. But textual studies cannot suggest a general form for the book and publishing, replicating as it does the common sense of intertextual accounts of reception: the binary of necessity versus a monolithic contingency. The acknowledgement that editing practices delimit interpretative possibilities is quickly passed over in favour of the contingencies of editorial practice as represented in different editions of the same text and their variable signifying effects. More specifically, the focus of textual studies (on literature, fiction, the publishing histories of individual texts and critical editing) is too narrow a ground to work on. Jerome McGann exemplifies this, at once stressing the ʻsocial and institutional conditionsʼ under which texts are produced and reproduced and asserting that texts ʻalways stand within an editorial horizonʼ (my emphasis), which he then glosses as the horizon of the textʼs production and reproduction. [45] Thus editing is mistaken for publishing as a whole.

It is the horizon of the publishable – what it is plausible to publish in a given context – that constitutes texts, writerships and readerships. The horizon of the publishable is neither singular nor autonomously defined by the industry. The publishable is defined by the relations between publishing and other institutions – commercial, legal, political, educational, cultural – and most obviously other media. Within publishing itself there is a range of horizons. Some correspond to particular genres or categories, others cut across these – celebrity publishing for example. Within this formulation, particular publishing categories can be specified according to the horizons which distinguish them: the role of educational institutions and discourses in childrenʼs publishing, the relations between publishers and museums and archives in illustrated art books, between academic publishing and higher education, and between romantic fiction and public libraries. Others have a more extensive range and force. The publishing horizon of many genres, from biography and autobiography through many categories of popular and literary fiction, is increasingly shaped by the horizon of possibilities of other media. The publishable must be malleable in a range of media forms: versionable as adaptation, extractable and/or abridgeable for newspapers and magazines. The horizon of the publishable is not the utilitarian counter to Romantic or post-Romantic formulations of the creative process. It is not the logic of the industry orienting the writerʼs work for the market. Writing is not opposed to publishing: composition is one of its processes. It has its own specificity but it is not of a different order. Composition does not precede publishing. This is not merely an effect of contemporary cultural production, where processes are frequently co-temporal and an apparent logic of sequence is reversed: books are marketed and publicized before they are composed, or at least completed. More fundamentally, the horizon of the publishable constitutes what it is possible to write and, significantly, how it is written, marketed, edited, designed and produced. Publishing always precedes ʻwritingʼ.

Publishing processes are always instantiated as particular practices and their relations. The editing practices of classics are very different from those of contemporary fiction, travel memoirs from guidebooks, and so on. In literary and intellectual publishing, the writer usually has greater freedom and sanction, and the compositional process a more constitutive role, than probably in any other mode (it is paradoxical that critiques of authorial intention have focused so attentively on the atypical). Illustrated books as a general category make a particular conjunction of design and production the dominant. Celebrity publishing subordinates composition, editing, design and production to marketing and publicity. This suggests a stronger, more stable concept of genre than intertextual theories usually admit. Genres – institutionalized conjunctions of signifying possibilities – are constituted in the relations between the totality of production practices, where one is usually dominant (and their relations with the other institutions that together define the horizon of the publishable). Practices may cohere or conflict, and the publishing category and text may be constituted in divergent or contradictory ways. Thus the contemporary editing practices that shape the literary classic define the literary text as both transparent and opaque. A familiar notion of the literary – a zone of settled meanings and values – intersects with a range of discourses that privilege accessibility and presume opacity. The increasingly elaborate editorial apparatus of classics proposes interpretative and reading possibilities shaped by both these definitions, but there is a resolution (of sorts) on the side of transparency. Endnotes, above all, secure the boundaries of the literary as the un-noted. Notes offer the contextual enrichments of scholarly hindsight, and recuperate the meanings of forgotten and marginal ʻtechnicalʼ languages and ʻdialectsʼ, but they do not trespass on the ground of the literary, which remains unmarked and legible – because, essentially, ʻclassicʼ.

So conceived, the edition is no longer merely the material form that the reader encounters but a site where publishing practices intersect. These co-constitute the text within networks of intertextual relations or (inter)textual contexts that propose how we should read. The reading practices proposed by the edition often pull in different directions, and the tensions may be decisive – textual contexts and interpretative possibilities may be radically divergent and conflictual. But the tensions may also be, often are, trivial, insignificant. Interpretative possibilities are many and rich, but interpretations are for the most part few and sparse. The formal echo with pragmatics is only that. Intertextuality constitutes a multiplication and transformation of interpretative possibilities that are all but invisible to pragmatics, a transformation that renders a recast inferencing all the more important. The richness of interpretative possibility does not wither away in a defining moment of pragmatic resolution. But ʻtextsʼ – and reading – are constrained, fixed, by the horizon of the publishable in ways that neither pragmatics nor intertextual accounts acknowledge.


1. ^ Michel de Certeau, The Practice of Everyday Life, trans.

S. Rendall, University of California Press, Berkley, 1984, p. 174.

2. ^ See for example, Roger Chartier, The Order of Books: Readers, Authors and Libraries in Europe between the Fourteenth and Eighteenth Centuries, trans. L. Cochrane,

Polity Press, Cambridge, 1994; Robert Darnton, ʻFirst Steps towards a History of Readingʼ, Australian Journal of French Studies, vol. 23, no. 1, 1986; D.F. Mackenzie, Bibliography and the Sociology of Texts, Panizzi Lectures, 1985, British Library, London, 1986.

3. ^ See, for example, Jon Cruz and Justin Lewis, eds, Viewing, Reading, Listening: Audiences and Cultural Reception, Westview Press, Boulder CO, 1994; Henry Jenkins, Textual Poachers, Routledge, London, 1992; Roger Silverstone, Television and Everyday Life, Routledge, London, 1994.

4. ^ A concentrated early formulation of this position, which implicitly or explicitly informs much thinking in cultural studies is Tony Bennett and Janet Woollacott, Bond and Beyond: The Political Career of a Popular Hero, Macmillan, London, 1987.

5. ^ Jacques Derrida, ʻSignature Event Contextʼ, trans. A.

Bass, in A Derrida Reader: Between the Blinds, ed.

P. Kamuf, Harvester Wheatsheaf, Hemel Hempstead, 1991.

6. ^ Roland Barthes, ʻThe Death of the Authorʼ and ʻFrom Work to Textʼ, in Image Music Text, selected and trans.

S. Heath, Fontana, London, 1977.

7. ^ V.N. Voloshinov, Marxism and the Philosophy of Language, trans. L. Matejka and I.R. Titunik, Harvard University Press, Cambridge MA, 1986, p. 23.

8. ^ Mikail Bakhtin, ʻDiscourse in the Novelʼ, in The Dialogic Imagination: Four Essays, ed. M. Holquist, trans.

M. Holquist and C. Emerson, University of Texas Press,

Austin, 1981, pp. 262–72. See also Ken Hirschkop, ʻIntroduction: Bakhtin and Cultural Theoryʼ, in Ken Hirschkop and David Shepherd, eds, Bakhtin and Cultural Theory, Manchester University Press, Manchester, 1989, on the unresolved place of an individuating concept of intonation in Bakhtin.

9. ^ Julia Kristeva, Le Texte du Roman, Mouton, The Hague, 1970; ʻWord Dialogue Novelʼ and ʻThe Ethics of Linguisticsʼ, in Desire in Language: A Semiotic Approach to Literature and Art, ed. L. Roudiez, trans.

L. Roudiez, T. Gora and A. Jardine, Blackwell, Oxford, 1981; ʻThe System and the Speaking Subjectʼ, in The Kristeva Reader, ed. T. Moi, Blackwell, Oxford, 1986.

10. ^ Derrida, ʻSignature Event Contextʼ, pp. 102–3.

11. ^ Complexly historical in the terms formulated by Louis Althusser, most explicitly in ʻThe Object of Capitalʼ, chapters 4 and 5 of Althusser and Étienne Balibarʼs Reading Capital, trans. B. Brewster, New Left Books,

London, 1977.

12. ^ See, for example, Tony Crowley, ʻBakhtin and the History of Languageʼ, in Hirschkop and Shepherd, eds, Bakhtin and Cultural Theory, on the deprivileging of historical context in Bakhtin; and Ken Hirschkop on the way ʻnovelnessʼ is built into Bakhtinʼs dialogic from the start, in Bakhtin: An Aesthetic for Democracy, Oxford University Press, Oxford, 1999.

13. ^ R. Barthes, S/Z, trans. R. Miller, Blackwell, Oxford, 1990, p. 19.

14. ^ John Fiske, ʻThe Moments of Television: Neither the Text nor the Audienceʼ in E. Seiter et al., eds, Remote Control: Television, Audiences and Cultural Power, Routledge, London, 1989.

15. ^ M. Bakhtin, ʻThe Problem of Speech Genresʼ, in Speech Genres and Other Late Essays, ed. C. Emerson and M.

Holquist, trans. V. McGee, University of Texas Press,

Austin, p. 78.

16. ^ The contrast is with knowledge of language: syntactic, semantic and, potentially, pragmatic. ʻNon-linguisticʼ knowledge makes no explicit claim about how other types of knowledge (logical, cultural etc.) are represented and is not to be confused with pre-linguistic or pre-discursive knowledge.

17. ^ Paul Grice, ʻMeaningʼ, in Studies in the Way of Words, Harvard University Press, Cambridge MA, 1989, pp. 216–17.

18. ^ Grice, ʻLogic and Conversationʼ, in Studies in the Way of Words, pp. 21–5.

19. ^ Ibid., p. 26.

20. ^ Ibid., pp. 26–7.

21. ^ Dan Sperber and Deirdre Wilson, Relevance: Communication and Cognition, Blackwell, Oxford, 1986. A second edition was published in 1995. The important addition is a ʻPostfaceʼ which sketches developments in Relevance theory and proposes certain revisions, specifically to the Principle of Relevance itself, noted below. On critical responses, see in particular Stephen Levinson, ʻA Review of Relevanceʼ, Journal of Linguistics 25, 1989, pp. 455–72; and Mary Talbot and Peter Mey, ʻComputation and the Soulʼ, Semiotica, vol. 72, no. 3/4, 1988, pp. 291–339.

22. ^ Sperber and Wilson never capitalize relevance when they are discussing its operation. They do capitalize it when they are discussing the First and Second Principles of Relevance. I have followed their notation in both cases.

Therefore when I speak of ʻRelevanceʼ, I am referring to the theory as a whole.

23. ^ Sperber and Wilson, Relevance, p. 45.

24. ^ Ibid., p. 69.

25. ^ Ibid., pp. 260, 270 (Postface).

26. ^ Ibid., pp. 268 (Postface), 162.

27. ^ Ibid., pp. 193, 184, 182, 194.

28. ^ Robyn Carston, Pragmatics and the Explicit–Implicit Distinction, Ph.D. thesis, London University, 1998, p. 19.

29. ^ She goes on to define novelistic enunciation as an inferential process by which different types of language are drawn together. This formulation is certainly suggestive, but it is too vague to function as a definition of intertextuality as an inferential operation and also seems to C O U R A G E Some make it a virtue.

F E A R Some make it a weapon.

S H A M E To some it’s a burden.

We make it an issue.

A special limited time offer toreceive Social ResearchNow you can subscribe to Social Research with this special limited time offer and save up to $48 off the single issue rate. This is a rare opportunity to share the unique and valuable insights of each issue of this quarterly journal, which has been lauded as “an invaluable resource for anyone interested in the modern world.” — Timothy Garton AshTo order your subscription, please send check or money order for $48 for two years or $36 for a 1 year subscription to:

Social Research, New School University, 65 Fifth Avenue, room 344, NY, NY 10003, or call 212-229-5776

This offer is for new subscribers only and expires 10/31/04. Discounts based on single issue rates. Institutions, agents and booksellers should call 212-229-5776 for special pricing on subscriptions. All overseas orders add $5 per year for postage.

www.socres.orgJust three of the subjects explored in upcoming issues of

Social Research.Does terrorism ever achieve its goals? Is courage always a virtue? Can shame be an effective tool of social discipline? Questions as timely as today’s news and as timeless as civilization itself are discussed by leading scholars from around the world in the pages of Social Research. A respected periodical since 1934, Social Research will challenge you with provocative, original, engaging — and relevant issues, four times per year.

An affirmative action/equal opportunity institution.

SP04AD SR RadicalPhil 1/15/04 3:53 PM Page 1claim the inferential process as something specific to the novel. See ʻThe Bounded Textʼ in Desire in Language, p. 36.

30. ^ Lawrence Grossberg, ʻThe In-difference of TVʼ, Screen, vol. 28, no. 2, 1987, pp. 28–45, p. 33.

31. ^ David Morley, Television, Audiences and Cultural Studies, Routledge, London, 1992, pp. 27–8.

32. ^ Relevance, p. 137.

33. ^ Manner is the only maxim which generates detachable implicatures. The other maxims (the ʻcontentʼ maxims) generate non-detachable ones. This assumes that ʻit is not possible to find another way of saying the same thing (or approximately the same thing) which simply lacks the implicatureʼ (ʻFurther Notes on Logic and Conversationʼ, p. 43). Grice seems to assume that that it is possible to say ʻthe same thingʼ in a different way without impacting on the implicatures produced – the effect of his naive form/content distinction, which makes manner the special place where form matters.

34. ^ Sperber and Wilson, Relevance, p. 219.

35. ^ Griceʼs examples also presuppose a shared social and cultural world, but if anything it is narrower than Sperber and Wilsonʼs: professional, gossipy and frequently sexist – ʻX is meeting a woman tonightʼ is his first example of a Generalized Conversational Implicature in ʻLogic and Conversationʼ (Studies in the Way of Words, p. 37).

The world of the Oxford Common Room, perhaps, the ʻcommonʼ of which shares something with his notions of shared or common knowledge.

36. ^ Sperber and Wilson, Relevance, p. 47.

37. ^ Michel Foucault, The Archaeology of Knowledge, trans.

A. Sheridan-Smith, Tavistock, London, 1972; and ʻThe Order of Discourseʼ in R. Young, ed., Untying the Text: A Post-structuralist Reader, Routledge, London, 1981.

38. ^ Kristevaʼs analysis of the way that the proto-novel resignifies the blazon – a spoken or indeed shouted genre – which is transposed or translated into ʻlaudatory descriptionʼ is one clear instance of this interest. Le Texte Du Roman also thematizes ʻthe coming of the bookʼ in a more general argument about secularization.

39. ^ Barthes, ʻFrom Work to Textʼ, pp. 156–7.

40. ^ Jacques Derrida, Of Grammatology, trans. G. Chakravorty Spivak, Johns Hopkins University Press, Baltimore, corrected edition, 1997, p. 18.

41. ^ Barthes, ʻFrom Work to Textʼ, p. 163.

42. ^ Gérard Genette, Paratexts: Thresholds of Interpretation, trans. J. Lewin, Cambridge University Press, Cambridge, 1997.

43. ^ Genette, Palimpsestes: La Littérature au Second Degré, Éditions du Seuil, Paris, 1982.

44. ^ On the ʻnewʼ editing, see George Bornstein and Ralph G. Williams, eds, Palimpsest: Editorial Theory in the Humanities, University of Michigan Press, Ann Arbor, 1993; and Philip Cohen ed., Devils and Angels: Textual Editing and Literary Theory, University of Virginia Press, Charlottesville, 1991.

45. ^ Jerome McGann, ʻLiterary Pragmatics and the Editorial Horizonʼ, in Cohen, ed., Devils and Angels.