Although Saussure stressed the importance of the relationship of signs to each other, one of the weaknesses of structuralist semiotics is the tendency to treat individual texts as discrete, closed-off entities and to focus exclusively on internal structures. Even where texts are studied as a 'corpus' (a unified collection), the overall generic structures tend themselves to be treated as strictly bounded. The structuralist's first analytical task is often described as being to delimit the boundaries of the system (what is to be included and what excluded), which is logistically understandable but ontologically problematic. Even remaining within the structuralist paradigm, we may note that codes transcend structures. The semiotic notion of intertextuality introduced by Julia Kristeva is associated primarily with poststructuralist theorists. Kristeva referred to texts in terms of two axes: a horizontal axis connecting the author and reader of a text, and a vertical axis, which connects the text to other texts (Kristeva 1980, 69). Uniting these two axes are shared codes: every text and every reading depends on prior codes. Kristeva declared that 'every text is from the outset under the jurisdiction of other discourses which impose a universe on it' (cited in Culler 1981, 105). She argued that rather than confining our attention to the structure of a text we should study its 'structuration' (how the structure came into being). This involved siting it 'within the totality of previous or synchronic texts' of which it was a 'transformation' (Le texte du roman, cited by Coward & Ellis 1977, 52).
Intertextuality refers to far more than the 'influences' of writers on each other. For structuralists, language has powers which not only exceed individual control but also determine subjectivity. Structuralists sought to counter what they saw as a deep-rooted bias in literary and aesthetic thought which emphasized the uniqueness of both texts and authors (Sturrock 1986, 87). The ideology of individualism (with its associated concepts of authorial 'originality', 'creativity' and 'expressiveness') is a post-Renaissance legacy which reached its peak in Romanticism but which still dominates popular discourse. 'Authorship' was a historical invention. Concepts such as 'authorship' and 'plagiarism' did not exist in the Middle Ages. 'Before 1500 or thereabouts people did not attach the same importance to ascertaining the precise identity of the author of a book they were reading or quoting as we do now' (Goldschmidt 1943, 88). Saussure emphasized that language is a system which pre-exists the individual speaker. For structuralists and poststructuralists alike we are (to use the stock Althusserian formulation) 'always already' positioned by semiotic systems - and most clearly by language. Contemporary theorists have referred to the subject as being spoken by language. Barthes declares that 'it is language which speaks, not the author; to write is... to reach the point where only language acts, "performs", and not "me"' (Barthes 1977, 143). When writers write they are also written. To communicate we must utilize existing concepts and conventions. Consequently, whilst our intention to communicate and what we intend to communicate are both important to us as individuals, meaning cannot be reduced to authorial 'intention'. To define meaning in terms of authorial intention is the so-called 'intentional fallacy' identified by W K Wimsatt and M C Beardsley of the 'New Critical' tendency in literary criticism (Wimsatt & Beardsley 1954). We may, for instance, communicate things without being aware of doing so. As Michael de Montaigne wrote in 1580, 'the work, by its own force and fortune, may second the workman, and sometimes out-strip him, beyond his invention and knowledge' (Essays, trans. Charles Cotton: 'Of the art of conferring' III, 8). Furthermore, in conforming to any of the conventions of our medium, we act as a medium for perpetuating such conventions.
Theorists of intertextuality problematize the status of 'authorship', treating the writer of a text as the orchestrator of what Roland Barthes refers to as the 'already-written' rather than as its originator (Barthes 1974, 21). 'A text is... a multidimensional space in which a variety of writings, none of them original, blend and clash. The text is a tissue of quotations... The writer can only imitate a gesture that is always anterior, never original. His only power is to mix writings, to counter the ones with the others, in such a way as never to rest on any one of them' (Barthes 1977, 146). In his book S/Z, Barthes deconstructed Balzac's short story Sarrasine, seeking to 'de-originate' the text - to demonstrate that it reflects many voices, not just that of Balzac (Barthes 1974). It would be pure idealism to regard Balzac as 'expressing himself' in language since we do not precede language but are produced by it. For Barthes, writing did not involve an instrumental process of recording pre-formed thoughts and feelings (working from signified to signifier) but was a matter of working with the signifiers and letting the signifieds take care of themselves (Chandler 1995, 60ff). Claude Lévi-Strauss declared that: 'I don't have the feeling that I write my books, I have the feeling that my books get written through me... I never had, and still do not have, the perception of feeling my personal identity. I appear to myself as the place where something is going on, but there is no "I", no "me"' (cited in Wiseman & Groves 2000, 173).
One of the founding texts of semiotics, the Cours de linguistique générale, itself problematizes the status of authorship. Whilst the text published by Payot in Paris bears the name of Ferdinand de Saussure as its author, it was in fact not the work of Saussure at all. Saussure died in 1913 without leaving any detailed outline of his theories on general linguistics or on what he called semiology. The Cours was first published posthumously in 1916 and was assembled by Charles Bally and Albert Sechehaye ('with the collaboration of Albert Riedlinger') on the basis of the notes which had been taken by at least seven students, together with a few personal notes which had been written by Saussure himself. The students' notes referred to three separate courses on general linguistics which Saussure had taught at the University of Geneva over the period of 1906-1911. Saussure thus neither wrote nor read the book which bears his name, although we continually imply that he did by attaching his name to it. It is hardly surprising that various contradictions and inconsistencies and a lack of cohesion in the text have often been noted. Indeed, some commentators have suggested that the Cours does not always offer 'a faithful reflection' of Saussure's ideas - a hardly unproblematic notion (Saussure 1983, xii). On top of all this, English readers have two competing translations of the Cours (Saussure 1974; Saussure 1983). Each translation is, of course, a re-authoring. No 'neutral' translation is possible, since languages involve different value systems - as is noted in the Cours itself. Nor can specialist translators be expected to be entirely disinterested. Furthermore, anyone who treats the Cours as a founding text in semiotics does so by effectively 'rewriting' it, since its treatment of semiology is fragmentary. Finally, we are hardly short of commentaries to bring both this foundational text and us as readers into line with the interpreter's own theories (e.g. Harris 1987; Thibault 1997).
This rather extreme but important example thus serves to highlight that every reading is always a rewriting. It is by no means an isolated example. The first critique of the ideas outlined in the Cours was in a book on Marxism and the Philosophy of Language which was published in Russian in 1929 under the name Valentin Voloshinov, but it has subsequently been claimed that this book had in fact been written by Mikhail Bakhtin, and the authorship of this text is still contested (Morris 1994, 1). Readers, in any case, construct authors. They perform a kind of amateur archeology, reconstructing them from textual shards whilst at the same time feeling able to say about anyone whose writings they have read, 'I know her (or him)'. The reader's 'Roland Barthes' (for example) never existed. If one had total access to everything he had ever written throughout his life it would be marked by contradiction. The best we can do to reduce such contradictions is to construct yet more authors, such as 'the early Barthes' and 'the later Barthes'. Barthes died in 1981, but every invocation of his name creates another Barthes.
In 1968 Barthes announced 'the death of the author' and 'the birth of the reader', declaring that 'a text's unity lies not in its origin but in its destination' (Barthes 1977, 148). The framing of texts by other texts has implications not only for their writers but also for their readers. Fredric Jameson argued that 'texts come before us as the always-already-read; we apprehend them through the sedimented layers of previous interpretations, or - if the text is brand-new - through the sedimented reading habits and categories developed by those inherited interpretive traditions' (cited in Rodowick 1994, 286, where it was, with delicious irony in this context, cited from Tony Bennett). A famous text has a history of readings. 'All literary works... are "rewritten", if only unconsciously, by the societies which read them' (Eagleton 1983, 12). No-one today - even for the first time - can read a famous novel or poem, look at a famous painting, drawing or sculpture, listen to a famous piece of music or watch a famous play or film without being conscious of the contexts in which the text had been reproduced, drawn upon, alluded to, parodied and so on. Such contexts constitute a primary frame which the reader cannot avoid drawing upon in interpreting the text.
The concept of intertextuality reminds us that each text exists in relation to others. In fact, texts owe more to other texts than to their own makers. Michel Foucault declared that:
Texts are framed by others in many ways. Most obvious are formal frames: a television programme, for instance, may be part of a series and part of a genre (such as soap or sitcom). Our understanding of any individual text relates to such framings. Texts provide contexts within which other texts may be created and interpreted. The art historian Ernst Gombrich goes further, arguing that all art, however 'naturalistic' is 'a manipulation of vocabulary' rather than a reflection of the world (Gombrich 1982, 70, 78, 100). Texts draw upon multiple codes from wider contexts - both textual and social. The assignment of a text to a genre provides the interpreter of the text with a key intertextual framework. Genre theory is an important field in its own right, and genre theorists do not necessarily embrace semiotics. Within semiotics genres can be seen as sign systems or codes - conventionalized but dynamic structures. Each example of a genre utilises conventions which link it to other members of that genre. Such conventions are at their most obvious in 'spoof' versions of the genre. But intertextuality is also reflected in the fluidity of genre boundaries and in the blurring of genres and their functions which is reflected in such recent coinages as 'advertorials', 'infomercials', 'edutainment', 'docudrama' and 'faction' (a blend of 'fact' and 'fiction').
The debts of a text to other texts are seldom acknowledged (other than in the scholarly apparatus of academic writing). This serves to further the mythology of authorial 'originality'. However, some texts allude directly to each other - as in 'remakes' of films, extra-diegetic references to the media in the animated cartoon The Simpsons, and many amusing contemporary TV ads (in the UK, perhaps most notably in the ads for Boddington's beer). This is a particularly self-conscious form of intertextuality: it credits its audience with the necessary experience to make sense of such allusions and offers them the pleasure of recognition. By alluding to other texts and other media this practice reminds us that we are in a mediated reality, so it can also be seen as an 'alienatory' mode which runs counter to the dominant 'realist' tradition which focuses on persuading the audience to believe in the on-going reality of the narrative. It appeals to the pleasures of critical detachment rather than of emotional involvement.
In order to make sense of the Absolut vodka advertisement shown here you need to know what to look for. Such expectations are established by reference to one's previous experience in looking at related advertisements in an extended series. Once we know that we are looking for the shape of the bottle, it is easier to perceive it here. Modern visual advertisements make extensive use of intertextuality in this way. Sometimes there is no direct reference to the product at all. Instant identification of the appropriate interpretative code serves to identify the interpreter of the advertisement as a member of an exclusive club, with each act of interpretation serving to renew one's membership.
Links also cross the boundaries of formal frames, for instance, in sharing topics with treatments within other genres (the theme of war is found in a range of genres such action-adventure film, documentary, news, current affairs). Some genres are shared by several media: the genres of soap, game show and phone-in are found on both television and radio; the genre of the news report is found on TV, radio and in newspapers; the advertisement appears in all mass media forms. Texts in the genre of the trailer are directly tied to specific texts within or outside the same medium. The genre of the programme listing exists within the medium of print (listings magazines, newspapers) to support the media of TV, radio and film. TV soaps generate substantial coverage in popular newspapers, magazines and books; the 'magazine' format was adopted by TV and radio. And so on.
The notion of intertextuality problematizes the idea of a text having boundaries and questions the dichotomy of 'inside' and 'outside': where does a text 'begin' and 'end'? What is 'text' and what is 'context'? The medium of television highlights this issue: it is productive to think of television in terms of a concept which Raymond Williams called 'flow' rather than as a series of discrete texts. Much the same applies to the World Wide Web, where hypertext links on a page can link it directly to many others. However, texts in any medium can be thought of in similar terms. The boundaries of texts are permeable. Each text exists within a vast 'society of texts' in various genres and media: no text is an island entire of itself. A useful semiotic technique is comparison and contrast between differing treatments of similar themes (or similar treatments of different themes), within or between different genres or media.
Whilst the term intertextuality would normally be used to refer to allusions to other texts, a related kind of allusion is what might be called 'intratextuality' - involving internal relations within the text. Within a single code (e.g. a photographic code) these would be simply syntagmatic relationships (e.g. the relationship of the image of one person to another within the same photograph). However, a text may involve several codes: a newspaper photograph, for instance, may have a caption (indeed, such an example serves to remind us that what we may choose to regard as a discrete 'text' for analysis lacks clearcut boundaries: the notion of intertextuality emphasizes that texts have contexts).
Roland Barthes introduced the concept of anchorage (Barthes 1977, 38ff). Linguistic elements can serve to 'anchor' (or constrain) the preferred readings of an image: 'to fix the floating chain of signifieds' (ibid., 39). Barthes introduced this concept of textual anchorage primarily in relation to advertisements, but it applies of course to other genres such as captioned photographs, maps, narrated television and film documentaries, and cartoons and comics ('comic books' to North Americans) with their speech and thought 'balloons'. Barthes argued that the principal function of anchorage was ideological (ibid., 40). This is perhaps most obvious when photographs are used in contexts such as newspapers. Photograph captions typically present themselves as neutral labels for what self-evidently exists in the depicted world whilst actually serving to define the terms of reference and point-of-view from which it is to be seen (Chaplin 1994, 270). For instance, 'It is a very common practice for the captions to news photographs to tell us, in words, exactly how the subject's expression ought to be read' (Hall 1981, 229). You may check your daily newspaper to verify this claim. Such textual anchorages can have a more subversive function, however. For instance, in the 1970s, the photographer Victor Burgin exhibited posters in the form of images appropriated from print advertisements together with his own printed text which ran counter to the intended meaning of the original ads.
Barthes used the term relay to describe text/image relationships which were 'complementary', instancing cartoons, comic strips and narrative film (ibid., 41). He did not coin a term for 'the paradoxical case where the image is constructed according to the text' (ibid., 40). Even if it were true in the 1950s and early 1960s that the verbal text was primary in the relation between texts and images, in contemporary society visual images have acquired far more importance in contexts such as advertising, so that what he called 'relay' is far more common. There are also many instances where the 'illustrative use' of an image provides anchorage for ambiguous text - as in assembly instructions for flat-pack furniture (note that when we talk about 'illustrating' and 'captioning' we logocentrically imply the primacy of verbal text over images). Awareness of the importance of intertextuality should lead us to examine the functions of those images and written or spoken text used in close association within a text not only in terms of their respective codes, but in terms of their overall rhetorical orchestration. Evelyn Goldsmith has produced a useful review of empirical research into the relationship between associated texts and images (Goldsmith 1984).
In media such as film, television and the worldwide web, multiple codes are involved. As the film theorist Christian Metz put it, codes 'are not... added to one another, or juxtaposed in just any manner; they are organized, articulated in terms of one another in accordance with a certain order, they contract unilateral hierarchies... Thus a veritable system of intercodical relations is generated which is itself, in some sort, another code' (Metz 1974, 242). The interaction of film and soundtrack in chart music videos offers a good example of the dynamic nature of their modes of relationship and patterns of relative dominance. The codes involved in such textual systems clearly cannot be considered in isolation: the dynamic patterns of dominance between them contribute to the generation of meaning. Nor need they be assumed to be always in complete accord with each other - indeed, the interplay of codes may be particularly revealing of incoherences, ambiguities, contradictions and omissions which may offer the interpreter scope for deconstructing the text.
The relationships between codes within a genre may shift over time, as William Leiss and his colleagues note:
In all, the effect was to make the commercial message more ambiguous; a 'reading' of it depended on relating elements in the ad's internal structure to each other, as well as drawing in references from the external world. (Leiss et al. 1990, 199)
Claude Lévi-Strauss's notion of the bricoleur who creates improvised structures by appropriating pre-existing materials which are ready-to-hand is now fairly well-known within cultural studies (Lévi-Strauss 1974, 16-33, 35-6, 150n; cf. Lévi-Strauss 1964). Lévi-Strauss saw 'mythical thought' as 'a kind of bricolage' (Lévi-Strauss 1974, 17): 'it builds ideological castles out of the debris of what was once a social discourse' (ibid., 21n): The bricoleur works with signs, constructing new arrangements by adopting existing signifieds as signifiers and 'speaking' 'through the medium of things' - by the choices made from 'limited possibilities' (ibid., 20, 21). 'The first aspect of bricolage is... to construct a system of paradigms with the fragments of syntagmatic chains', leading in turn to new syntagms (ibid., 150n). 'Authorship' could be seen in similar terms. Lévi-Strauss certainly saw artistic creation as in part a dialogue with the materials (ibid., 18, 27, 29). Logically (following Quintilian), the practice of bricolage can be seen as operating through several key transformations: addition, deletion, substitution and transposition (Nöth 1990, 341).
Gerard Genette proposed the term 'transtextuality' as a more inclusive term than 'intertextuality' (Genette 1997). He listed five subtypes:
To such a list, computer-based hypertextuality should be added: text which can take the reader directly to other texts (regardless of authorship or location). This kind of intertextuality disrupts the conventional 'linearity' of texts. Reading such texts is seldom a question of following standard sequences predetermined by their authors.
It may be useful to consider the issue of 'degrees of intertextuality'. Would the 'most intertextual' text be an indistinguishable copy of another text, or would that have gone beyond what it means to be intertextual? Would the 'most intratextual' text be one which approached the impossible goal of referring only to itself? Even if no specific text is referred to, texts are written within genres and use language in ways which their authors have seldom 'invented'. Intertextuality does not seem to be simply a continuum on a single dimension and there does not seem to be a consensus about what dimensions we should be looking for. Intertextuality is not a feature of the text alone but of the 'contract' which reading it forges between its author(s) and reader(s). Since the dominant mode of producing texts seems to involve masking their debts, reflexivity seems to be an important issue - we need to consider how marked the intertextuality is. Some defining features of intertextuality might include the following:
Confounding the realist agenda that 'art imitates life,' intertextuality suggests that art imitates art. Oscar Wilde (typically) took this notion further, declaring provocatively that 'life imitates art'. Texts are instrumental not only in the construction of other texts but in the construction of experiences. Much of what we 'know' about the world is derived from what we have read in books, newspapers and magazines, from what we have seen in the cinema and on television and from what we have heard on the radio. Life is thus lived through texts and framed by texts to a greater extent than we are normally aware of. As Scott Lash observes, 'We are living in a society in which our perception is directed almost as often to representations as it is to "reality"' (Lash 1990, 24). Intertextuality blurs the boundaries not only between texts but between texts and the world of lived experience. Indeed, we may argue that we know no pre-textual experience. The world as we know it is merely its current representation.