Toward a Semiotics of Canon

GEORGE AICHELE

I understand a canon to be a collection or list of authoritative writings accepted by some community of readers. The canon identifies the accepted texts and fixes the written contents of those texts. To a large degree, canon controls and maintains the transmission of the selected texts, and therefore canon is a profoundly semiotic phenomenon. Canonical status is not something that is intrinsic to any text, nor can any text demonstrate its own canonicity. The establishment of a canon is the attempt of some group of the text's readers to clarify a text's meaning and to achieve narrative completeness through a metatextual and intertextual commentary-that is, to create a text that can explain itself.

Because it is a semiotic phenomenon, canon is ideological. Like all ideologies, the canon does not reflect some external, objective reality more or less well, but rather it creates reality, or better, it provides the filter or lens through which individuals and communities perceive reality. Ideologies arise from the fact that we humans are finite, fragile beings, and our knowledge of anything is at best partial and limited. We have no ideology-free knowledge of reality; there is no escaping from ideology. Even the ideal of objective scientific or historical knowledge is an ideological product. Hence the biblical canon and the claims made on its behalf by believers do not describe actual states of affairs but rather are expressions of desire (Zizek 1994: 296-331)-desire for what we want the canon to be.

I understand ideology, as it applies to the understanding of texts, to take the form of intertextuality. Insofar as texts are ideological, ideology is intertext (see also Aichele 1996, ch. 7; 1997, ch. 3). According to Julia Kristeva, intertextuality is the 'intersection of textual surfaces rather than a point (a fixed meaning)' (Kristeva 1980: 65, emphasis original). Intertextuality is the 'transposition of one (or more) system(s) of signs into another' (Kristeva 1984: 59-60). The reader's ideology consists of an intertextual network or web of texts, constantly shifting, growing, and diminishing as the reader both reads new texts and forgets old ones. A text is any group (or system) of signs, in any medium. Every text is read in the light of a great many texts that the reader has already read. These other texts are also re-interpreted in light of the new text. This is an important dimension of intertextuality: intertextuality is not something extra added to the juxtaposed texts, but something that emerges between the texts, in the way that they are juxtaposed. The individual reader stands at the junctures of an indefinite web or network of texts; the individual is that web or network. The reader is in this light no more than a point at which many texts intersect one another (see also the introduction to Aichele and Phillips 1995). Intertextuality is thus not something that readers create, but rather it is something that creates readers. The texts in the intertextual web resonate, interfere with, or otherwise play off of one another in various and complex ways, and ideology is this intertextual weaving of texts.

Normally the reader's ideology is unconscious or at best semi-conscious. Nevertheless the intertextual web deeply influences the reading brought to any new text. No meaningful reading (or writing) is possible without some intertextual context. The intertext enables the reader to:

- recognize the text as a text, that is, a signifying object as opposed to a bunch of meaningless marks;

- determine the limits of the text, where it begins and ends;

- whether or how it refers to extratextual reality (the meta-genres of history and fiction);

- identify the cultural and linguistic codes in terms of which the text should be read-the correct genre, as well as any 'microcontextual variations' (Rabkin 1976: 36) within that genre and establish the correct meaning of the text through the use of those codes.

A canon is a comprehensive, exclusive, and authoritative intertext, and thus it both is produced by and reinforces an ideology. This ideology is not in the individual texts themselves but rather in the ways in which the canonized texts are juxtaposed with one another (and with non-canonical texts) and held together in the reading practice of a community of readers. The canon intensifies the ideological process and makes the appropriate intertextuality explicit.

The purpose of a canon is to obscure and replace the physical text itself- both texts that are included and texts that are excluded-with something else, a signified or at least a signifying potential that far surpasses any of the included texts (Gamble 1985: 14, 75). Canon is a form of the Peircean interpretant, and in fact it is what C.S. Peirce called the 'final interpretant' (1992: xxxvi-xxxvii). Canon serves as a metalanguage-that is, a language that explains the meaning of another language, in this case the language of the various biblical texts (Aichele 1997: 127ff.). It performs what Roman Jakobson called the metalingual function, which is to identify the proper codes for understanding of the text (Jakobson 1987: 69). As a metalanguage, the canon legitimizes certain interpretations of the Bible and delegitimizes others. Canon is also intensely connotative. In a famous essay, Roland Barthes (1988: 173-78) argued that all advertising messages, despite their evident denotative differences, have only one connotative message, namely, 'Buy this product!' This connotation is the 'bottom line' of the advertisement, and all other meaning is subservient to it.

We must not suppose that the second message (of connotation) is 'hidden' beneath the first (of denotation); quire the contrary: what we immediately perceive ... is the advertising character of the message, its second signified... : the second message is not surreptitious (Barthes 1988: 17)

Likewise the canon of the Bible, throughout its length and depth and despite its numerous denotative incoherences, discrepancies, and contradictions, has only one connotative message, namely, 'This is Absolute Truth. This is God speaking.' All other messages in the Bible as a canon are understood to refer to this 'second signified'. In this sense we might also say that canon is 'mythic' or mythographic (Barr 1983: 169), although the biblical canon contains much more than stories properly categorized as myths.

Because of its ideological function, a canon is also a social, economic, and political phenomenon. The canon defines (in part) the reading community's identity, and the community claims ownership of the canonized texts. Canon is in effect an ancient form of copyright-or rather, copyright is a modern 'translation' of canon-protecting the texts from the distorted readings of 'outsiders'. Canon identifies the listed texts as valuable, better than the other, noncanonical texts. Canon also measures the limits of the individual reader's power over and possession of the text, as an 'insider'. The reader is not free to read the text in any way that s/he wants; s/he is bound to the canonical intertext.

Canon freezes the text, fixing its physical form and its relation to other texts, both canonical and non-canonical. The eternal, perfect canonical text is of course an ideal, not the historical actuality, because no specific text, whether ancient manuscript or modern edition, is the definitive version' of the book, and revised and corrected editions and translations are always possible (and probable). However, that which was frozen cannot become too frigid, or else it will die. The canon must continue to speak to the reading community; its meaning must retain some flexibility, even when it is believed to be Eternal Truth. If the books are no longer perceived as relevant to its readers, the canon may become a venerated relic, like much of the Old Testament in many Christian churches today, but it will no longer be read. As a metatext, canon provides its own intertextual commentary, but it also requires extra-canonical commentary to keep it 'alive' and authoritative, to acknowledge its validity and its relevance to the contemporary world. There is no canon without commentary.

A canon may be defined, in part at least, by its opposition to that which it is not, namely, the non-canonical. The canon is itself defined through the exclusion of non-canonical texts. Non-canonical texts fall across a range of possibilities. There are different ways of being non-canonical. At one extreme are texts that are rejected from the canonical list because they are heretical or blasphemous-that is, unacceptable in some way to the reading community. These texts threaten the identity of the community. They are not merely excluded from the canon, but they are also forbidden to the community. These texts are regarded as dangerous and evil. At the other extreme are texts to which the reading community is indifferent. The community may know of these texts but they do not read them, not because the texts are forbidden but because the community is not interested in them. Between these extremes are a wide variety of other texts that may be acceptable and even valuable to the community but are finally regarded as non-authoritative. These writings are not forbidden, nor are they regarded as dangerous or evil. Included in this group are the supplementary commentaries mentioned above. These commentaries are essential to the canon itself, as canon, but they are not themselves canonical. Paradoxically, all of the non-canonical writings derive their truth, values and meanings from the canonical list even as they create the canon by their exclusion from it. In other words, the truth of a non-canonical text is thought of as inherently different than the truth of a canonical text, even if both texts are stating the same message. The reference of a canonical text is different than that of a noncanonical one. The non-canonical text may be true, but it is not True. It does not share in the connotative univocity of the canonical message.

The canon provides a textual frame that works in two related ways (see also Aichele 1985: 18ff.). First, through this canon-frame the reading community identifies certain texts as essential to its identity. Canon defines the identity of the group, by drawing a line around a group of texts that is in some way associated with the group's beliefs and values. The canonized texts are both necessary and sufficient to the community's definition of itself. Through the canon-frame the reading community identifies itself in relation to these texts. James Barr argues that this identity formation role can be overstated: canons do not create communities, but rather communities create canons (Barr 1983: 41-43). Nevertheless, correct identification of the canon is one of several necessary conditions for membership in the community. The canon draws a line around the group, determining through its acceptance or rejection who is 'in' and who is 'out' of the group. Canonical lists are always conservative, even reactionary, and never innovative. The question of canon is therefore a question of value, an ethical question. It frames the act of discovering or receiving or creating a meaningful world. In choosing one text and rejecting another, we create meaning. For the reading community, all understanding and interpretation must be referred to this act of choice.

Second, interpretation is not merely something to be applied upon or added to the canonical text. Understanding begins with the very choice of a text and its placement in the larger canonical grouping. The canon frames a content within a context-as I have already said, it is a metatext. Every canon implies an ontology. The reader's understanding of space and time, the forms of human existence, is governed by this selection of canonical texts, and thus every canon is also at least implicitly a cosmology. The canon is about reality: it is metaphysical. Upon being chosen, that is, constituted as the canonical text, the words refer to a world which becomes our world, for we have chosen it. The selection of texts intends a world, or better, the reader intends the world through the selected texts. In constituting the canon as a world, we also constitute the reader as ourselves, as the 'other side' of the same act (cf. Phillips 1980: 135). The framed texts establish a world, most often in the form of a story, and they locate the community in that world.

The reader is not likely, of course, consciously to choose a meaning. Ideology rarely works that way. What we choose is usually a story. Human beings cannot live without a story of some sort-that is, without a myth, without a world. Yet there is no simple identity between any story and its meaning. We all do not find the same meaning in the 'same' story. These differences reflect ideological differences between readers, for no two readers are likely to share exactly the same intertextual web. These differences may be small or large. In extreme cases it might even be argued that several readers are not really reading the same story, even though the physical text is the same.

Human consciousness and experience are always existentially and conceptually prior to the meaningless 'stuff (hule) that constitutes the material aspect of the signifier. Nevertheless, without that material stuff there is no text and there can be no intertext. Upon being chosen by the reader this stuff becomes significant-it becomes my story and this world-yet as such this material stuff can never be directly perceived. In Regis Debray's terminology, the hule of the signifier is 'hyperphenomenal' (Debray 1996: 51). Hyletic matter can only be encountered in a text as the signifier's resistance to meaning.2 The story is both more and less than any meaning ascribed to it, and it carries in the materiality of its signifiers the potential to negate any meaning that the reader may choose. Thus there is always tension between any text and the canon of which it may be a part.

Every canon is both cosmological and ontological, and therefore canon is metaphysical, onto-theological. But the metaphysical question also cuts the other way: a canon creates reality, a world. Meaning is a function of metaphysics. This is as true of secular literary canons as it is of religious ones. Within this framework, 'belief appears in the canonizing act that selects one group of texts and not some other group. Canon is a function of belief. As we have just seen, canon is ideological. A canon is like a story-indeed, a canon is a story of stories, or meta-story. To tell a story is to choose some matters and to reject others. What is not told is, in a sense, just as crucial to the story's identity and meaning as what is told. Aristotle claimed that every story requires a beginning, a middle, and an end (Aristotle 1967: 30, ¤1450b, 26-34). Any story can only tell some things, and never every thing; otherwise it ceases to be this story rather than some other one. Therefore, all stories are artificial-that is, fictional. The same is true for canon.

However, canon is also unlike story. Every story is inherently incomplete, dotted with 'spots of indeterminacy' (Ingarden 1973: 68, 102, 237) that must be 'concretized' by the reader, often unconsciously, always intertextually, in order for the story to have any meaning at all. No story can specify in advance the codes in terms of which it should be read, and thus there is always the possibility that different readers will apply different codes in order to decipher the same story. Thus we disagree about the meanings of stories. No story can conclusively explain itself because every story is referentially incomplete (see Aichele 1985, ch. 3). The story's reference is always at least somewhat indirect or interrupted. No story can interpret itself, and no story is read entirely by itself. In order for the story's meaning to be understood, it must be supplemented from 'outside' of the story. This supplementation is provided by the reader's interpretation. Because no two readers share exactly the same intertext, the story is inevitably a matter of disagreement among its readers.

If the story is relatively unimportant-that is, outside of the canon-this disagreement does not bother us at all: 'You have your opinion and I have mine'. It is rather like a difference of taste. But if the story is an important one, one that in some way defines who we are, that answers basic questions, then our disagreement is very troubling. We do not believe in texts themselves, but in particular ways of reading texts If my way of reading an important text is challenged by other readers, then my belief is threatened. Canon both identifies which texts are important or authoritative, and it also provides a hermeneutical context in which the important texts can be read correctly. From the standpoint of the believing community, canon provides the only proper context in which the texts may be read. Canon specifies the proper codes in terms of which the text should be read, and thus it identifies a range of acceptable interpretations. The desire for canon is the desire for a text that interprets itself.

This is the ideal of canon-the desire that produces canon-and not the actual reality of canon. No text can interpret itself. Different readers and communities of readers do in fact read the 'same' canonized texts in very different ways. The biblical canon is both unable to satisfy the reader's desire for understanding of the texts and unable to prevent readers from bringing to the texts a wide variety of presuppositions and assumptions-that is, ideological intertexts-not authorized by the canon. Nevertheless, the canon does have very powerful, if never entirely successful, effects on how the texts are read. Can there be any doubt that readers would read the gospel of Mark differently if it were excluded from the New Testament, or that readers would read the gospel of Thomas differently if it were included in the New Testament? (Barr 1983: 44-45) In effect the New Testament canon 'quotes' the entire story of Mark, making it part of a much larger story, and thus it reduces or even eliminates the referential ambiguity that characterizes that text. The canon in effect provides in advance the intertext in terms of which we can agree on the meaning of the important text. The canon completes the inherent incompleteness of the story. The concept of canon arises from the need to end the endless demand for a meaningful written text by completing the uncompleteable story.

The concept of canon is related to but also distinct from the older and vaguer concept of 'the scriptures' (Barton 1986: 28). The scriptures are texts that are authoritative but not necessarily exclusive. Like the canon, the scriptures also provide an intertextual frame, but unlike the canon, they have no clearly defined limit. They are not 'closed'. We may agree that some writings are scripture without agreeing about others. 'The scriptures' are not necessarily clearly defined for those who cite them; hence New Testament references to Old Testament 'scriptures' do not imply the existence of a canon. The pre-canonical scriptures intertextually complement or play against one another in fluid and indefinite ways. Hence even though the words 'scriptures' and 'writings' are synonymous for the most part, nevertheless a distinction must be made between the scriptures and any other writings. The denotation is identical but the connotations differ.

The question of canon arises in the form of disputes over which writings are (or which reading community possesses) the genuine scriptures, the authoritative writings. Once the canon has been established, it must be clearly defined for all who recognize it. The canonizing of scriptures separates 'the canon of scripture' from all other texts, including commentaries and traditions of interpretation with which the canonized texts had formerly been intermingled. Canonizing makes the scriptures comprehensive and exclusive. Canon fixes the text, identifying its proper extent and order. The canonical collection is necessarily all-or-nothing. This does not rule out what Barr calls a 'stepped canon' (Barr 1983: 41). Some texts may be more central (such as the Torah in the Jewish scriptures) and other texts more marginal (such as the Writings). In these cases, the canon must be definitive about which texts are central and which ones are marginal. Changes in 'status' are not permitted. Despite this, it is frequently not easy to distinguish between the meaning of 'scripture' and the meaning of 'canon'. By the time the Christian and Jewish canons have been established, and perhaps even prior to that point, any significant difference between the word 'canon' and the word 'scripture' has disappeared; 'canon' has either replaced 'scripture', or else the meaning of 'scripture' has changed and become synonymous with 'canon'.

The list of writings recognized as 'scripture' in Judaism is defined differently than it is in Christianity, although it is identical in extent (but not in order) to the Protestant Christian Old Testament. The concept of canon is itself fundamentally a Christian one (Barton 1986: 63), and it is doubtful that Jews understand the concept of canon in the same way that Christians do. The difference between the Christian and Jewish views of scripture correlates to what Kurt and Barbara Aland call the difference between the 'Greek' and the 'oriental' views of writing.3

Greeks and orientals view the written word differently. For orientals the very letter had a sanctity of its own. The Hebrew text of the Old Testament, like the text of the Quran, is alike in all manuscripts (except for unintentional errors). For Greeks it was the message contained that was sacred. (Aland and Aland 1987: 286)

The 'Greek' view of writing holds that the universe consists of a set of extratextual beings to which language points with greater or lesser reliability. Language is at best an image of extratextual reality, and that reality is more important than the words that express it. In other words, the message that the biblical texts contain is more important than the physical stuff of the texts. The text of both the Old Testament and the New Testament is ultimately unimportant to Christians; what counts for them is how the message conveyed by the Old Testament texts relates to the message of the New Testament texts. The New Testament as a metatextual collection rewrites and spiritualizes not only its own texts but those of the Old Testament as well. The canon exists in order to make this rewriting possible.

For what the Alands call the 'oriental' view of writing, the universe itself is a text, upon which every being depends. Reality is language, and it cannot be separated from language. As Barr argues, 'for Judaism as it has developed, the Torah is not primarily a report of or witness to salvific events ... {but} it is the divinely given text, essential for ritual worship and for the establishment of legal norms' (1983:100, emphasis original). The signified message is not superior to the textual signifier, as it is in Christianity. A canon, in the Christian sense of the word, is unthinkable in this context except, as John Barton suggests, as the Jewish refusal of Christian appropriation of the scriptures. Jewish interpretation and exegesis of the biblical texts has long been open to the polysemy of the scriptures, and to the intertextuality within them and between them and other writings. Within rabbinic thought there is a potential for confidence in the face of the intertextual abyss, the possibility of what Jose Faur (1986: 142) calls a 'literal theology' and the recognition of what Jacques Derrida calls 'the circumcision of the word' (1986: 346).

The Bible does not describe itself as a canon. Texts such as Deuteronomy 4.2 ('You shall not add to the word which I command you, nor take from it; that you may keep the commandments of the LORD your God which I command you'), 2 Timothy 3.16-17 ('All scripture is inspired by God and profitable for teaching, for reproof, for correction, and for training in righteousness, that the man of God may be complete, equipped for every good work'), or Revelation 22.18-19 ('I warn every one who hears the words of the prophecy of this book: if any one adds to them, God will add to him the plagues described in this book, and if any one takes away from the words of the book of this prophecy, God will take away his share in the tree of life and in the holy city, which are described in this book') do not unambiguously assert the canonicity of the entire collection. No writing can by itself demonstrate its own authority or even its own truthfulness. If I write here the sentence, 'Every word in this book is the authoritative Word of God', that does not increase in any way the value of this book. If this book is not the Word of God, then that sentence is also not the Word of God.

The word 'canon' (Greek kanon) is used only a few times in the Bible (Micah 7.4 [LXX]; 2 Corinthians 10.13, 15, 16; Galatians 6.16; cf. Judith 13.6; 4 Maccabees 7.21), and always in the general sense of 'rule' or 'standard'. However, '[i]t is important to observe that "canon" in the sense of "canon of scripture" ... appears not to derive from the sense "rule, standard,"' but rather the word 'derives from the familiar Greek sense, as used of a table of figures or the like' (Barr 1983: 49, n. 1). Barton seems to contest Barr's claim: 'If the word "canon" is to be used at all, then it should probably be in the sense in which the term was sometimes used in the early Church, to denote a "norm" or regulative standard rather than a closed body of texts' (Barton 1986: 63). However, Barton's statement appears in the context of an argument that there was no canon ('closed body of texts') early in the common era. The use of the word as a technical term to describe an accepted and exclusive list of authoritative writings for Jews or Christians does not appear until the third century CE in the writings of Origen (Graham 1987: 193, n. 13). Even then neither the present-day Jewish nor Christian canons have been conclusively 'closed'. Only the concept of canon, or the desire for a canon, has appeared.

The 'original' and first canon is the Christian canon, which serves as the prototype for all others. Strictly speaking, however, Christianity does not have one canon but two. The New Testament presents the dilemma of a double canon, a problem which has deeply influenced the development of Christian theology, but which Christians have never adequately faced. In effect there can never be more than one canon, because no group of readers can ascribe complete authority to two distinct canons. Canons are by definition totalizing and exclusive. One canon must dominate and subsume the other. Just as, according to Christian belief, the four canonical gospels must refer to one true story,4 despite their evident differences both of meaning and of the written word, so both the Old and the New Testaments must refer to the same thing, and the new must be the proper translation of the old. Both the New Testament and the Old Testament writings are appropriated for Christian reading by their canonization.

Once the scriptures are transformed into the canon, it becomes the highly ideological metatext of the Bible understood as a unified whole. As I have already suggested, the Bible becomes a single message with a single meaning. As a result, when the writings included in the biblical canon refer to each other, in effect they also refer to themselves. This self-referentiality creates a semiotic feedback loop in which the text's meaning is 'clarified'-that is, restricted. The semiotic function of 'the scriptures' (or related phrases) appears in the intertextual tension between two texts, where it enables the first text, the scripture, to signify in the message of the second text, which becomes in effect a commentary on the scripture and controls its meaning. The second text, in effect, colonizes and appropriates the first one. Therefore, passages in which the scriptures are cited, either explicitly or implicitly, are important points at which intertextual tension between the two Christian canons, or between any specific text and the canon, come to light.

A canon may define the whole, and the same parts may figure in different canons. For Christian commentators the Psalms belong to a whole different from the whole to which they belong for Jewish commentators; they may agree that there are messianic psalms, but the plain sense of such psalms must be different for each, since the whole text for the Christian shows the fulfillment of the messianic promises. (Kermode 198G: 181-82; cf. Barnstone 1993: 163)

It is quite probable that if there had been no New Testament, there would have been neither the Christian Old Testament nor the canonical Jewish Bible either. Barton suggests that Judaism probably would not have established a canon except for the need to resist the reinterpretation and appropriation of Jewish writings into the Christian canon (1986: 63).5 When the New Testament writings refer to the 'scriptures', sometimes using that word (Gk. graphe) and sometimes using other terminology such as 'the law and the prophets' or 'Moses' or 'it is written', they often cite the Jewish scriptures. There are also texts within the Jewish scriptures that in effect comment on other texts within those scriptures, and thus the 'paradox' (Fishbane 1986: 25, 28) of scripture that is also commentary is there also for Jews as well, but within a single collection. The fact that the New Testament adds a second list of authoritative writings raises the problem to a higher level. More precisely, the New Testament canon transforms the Jewish scriptures into the Christian Old Testament-a Christian canon.

This case is exactly analogous to that of Pierre Menard, the fictional hero of a story by Jorge Luis Borges (1962). Menard, a twentieth-century Frenchman, decides to write a book that will 'coincide-word for word and line for line' (1962: 49) with the Don Quixote of the seventeenth-century Spanish author Miguel de Cervantes. It will not be a translation, or a paraphrase, or an 'updating' of the story, but an exactly identical story. 'To be, in some way, Cervantes and to arrive at Don Quixote seemed to him less arduous-and consequently less interesting-than to continue being Pierre Menard and to arrive at Don Quixote through the experiences of Pierre Menard' (Borges 1962: 49). In a letter to the story's narrator, Menard says,

My solitary game is governed by two polar laws. The first permits me to attempt variants of a formal and psychological nature; the second obliges me to sacrifice them to the 'original' text and irrefutably to rationalize this annihilation. (1962:51)

Menard does not complete his task; in fact, we are told that all that he produced was two chapters and part of a third. Nevertheless, the nameless narrator confesses, 'I often imagine that he finished it and that I am reading Don Quixote-the entire work-as if Menard had conceived it' (Borges 1962: SO). The narrator then compares Menard's text to Cervantes's:

the fragmentary Don Quixote of Menard is more subtle than that of Cervantes. The latter indulges in a rather coarse opposition between tales of knighthood and the meager, provincial reality of his country; Menard chooses as 'reality' the land of Carmen during the century of Lepanto and Lope.... The text of Cervantes and that of Menard are verbally identical, but the second is almost infinitely richer (Borges 1962: 51, 52).

Borges satirizes a whole complex of literary assumptions in this amusing story: the value of authorial intention and the social context in which the text is written, the importance of the 'original' text in interpretation, the function of influence and imitation in literary history. Nevertheless, what Menard does to Cervantes's book is in effect what Christian appropriation of the Jewish scriptures has done to them. The Jewish scriptures are the writings of people for whom the messiah has not yet come, for whom the messiah is at best a vague figure. The Old Testament consists of Christian writings-not writings by Christians but writings for Christians-in which the messiah of the New Testament has been announced and a meaning-filled context for his life is provided. Likewise the New Testament consists of Christian writings, regardless of whom their historical authors might have been. In order to transform the Jewish scriptures into the Old Testament a rewriting of the texts had to occur-not the act of a Pierre Menard but something every bit as absurd and grandiose. In both cases this rewriting is not a physical act that transforms or replaces the material hule of the signifier, but it does change the signifier anyway, because it changes the signification.

Therefore the Christian New Testament is not merely a second canon-it never could be, for it is fundamentally incomplete without the Old Testament-but it is also, and primarily, a meta-canon. Already in the second century CE, before there was any canon, the Christian heretic Marcion understood this problem. Marcion rejected the Jewish scriptures (which at that point were still fairly fluid, except for the Torah) and replaced them with his own collection of Pauline letters and a version of the gospel of Luke, from which 'Jewish' material had been removed. Marcion also rejected the four gospels, insisting on just one.6 Whether or not Marcion's collection of texts was truly a canon is doubtful. Other early collections of epistles and of gospels were apparently already in process of forming. Harry Gamble denies that Christianity developed the New Testament as a reaction to Marcion's collection (1985: 60). Nevertheless, Marcion added impetus to the Christian canonizing process. Marcion's challenge to his fellow Christians made canon thinkable. The need for a canon began to be noted by early Christians, perhaps in part as a defensive reaction against Marcion's collection of texts. The emerging Christian mainstream rejected Marcion's rejection of the Jewish scriptures, and they claimed both an Old Testament and a New Testament for themselves. It was this double claim that led eventually (so Barton suggests) to the Jewish formation of a canon.

It is not clear exactly when the process of canonization was completed in any of the believing communities, but that may well have taken many more years-possibly not until the fourth century CE or later. It is probably not coincidental that this period was also the time in which the emerging 'catholic' Christianity became the state religion of the Roman Empire. As the religion as a whole became socially acceptable and even desirable, its leaders and institutions became powerful. Heretical and heterodox individuals, groups, and tendencies were suppressed. The emperor Constantine pressured the churches to 'standardize' their various versions of the scriptures, and if this standardization is not already a canonization it is at least another powerful push in that direction The effect of Constantine on the Christian canon was much the same as that in a much earlier period of the Persian empire and its policies on the exiles returning from Babylon to Judah. This led in effect to the 'canonization' of the Torah (Deist 1995: 71-72; cf. Berquist 1996). Both Ferdinand Deist and Jon Berquist use the term 'canon' throughout their discussions of this phenomenon and so I repeat it here. Nevertheless, I accept Barton's argument that the word 'canon', as well as the concept, is not appropriate for these texts at this point in history (1986: 63). Even if the 'canon of Torah' is closed around the time of Ezra, the rest of the Jewish canon remains undetermined, and an undetermined canon is not really a canon (Barton 1986: 28).

As we have already said, canon is a form of intertextuality, but canon raises intertextuality to a higher degree. A canon both makes understanding of its included texts possible, by providing an authoritative intertext, and it establishes a 'correct' reading for each text. Likewise the New Testament meta-canon explains and effectively absorbs the other canon, for example, by 'fulfilling' it. Another of Borges's writings is instructive here. In 'Kafka and His Precursors' (1964), Borges holds that Franz Kafka's stories and novels in effect create the works of his precursors. '[A]fter frequenting [Kafka's] pages a bit, I came to think that I could recognize his voice, or his practices, in texts from diverse literatures and periods" (Borges 1964: 199). These 'precursors' include texts by Zeno, Han Yu, Kierkegaard, Browning, Bloy, and Dunsany.

Borges claims that 'not all of [these texts} resemble each other.... In each of these texts we find Kafka's idiosyncrasy to a greater or lesser degree, but if Kafka had never written a line, we would not perceive this quality; in other words, it would not exist.' (1964:201) Kafka's texts cause us to read these precursors in a different way.7 The precursors' works are changed by their relation to Kafka's works, a relation that is established retroactively from the reading of Kafka's writings. Borges does not assert any conscious intention to do this on Kafka's part and indeed, it seems quite possible that such a relation could exist between two or more works even if the respective authors were entirely unaware of each other. In other words, historical 'influence' as that is usually understood is not a crucial factor in intertextuality.

The New Testament creates the Old Testament as that which is to be fulfilled in much the same way that Kafka creates his precursors. However, Kafka's writings are not a canon, nor are the writings of all of his precursors. These writings may appear in one or another secular literary canon, but that is another matter. What is irony and paradox in relation to a single writer's oeuvre is contradiction in relation to a canon such as the Bible. Therefore the Christian double canon presents another problem. In Jewish tradition, commentary or 'oral Torah' cannot also be scripture or "written Torah," much less canon (Faur 1986: 16, 103, 133). One way of signifying this distinction in early Judaism was through the difference between codex and scroll. The Torah and eventually the rest of the Jewish scriptures were written on scrolls. Jewish communities continued to use the scroll form for their scriptures (especially texts to be used liturgically) for hundreds of years after the codex form was introduced. When the Mishnaic traditions were first written down, codex 'notebooks' were used for the transcription. 'The employment of the note-book was the most suitable way of indicating that [the rabbis] were writing Oral Law for private, or unofficial use, and not for publication' (Saul Lieberman, quoted in Resnick 1992: 11, note 51). In other words, the codex form signified 'non-scriptural'. However, the early Christians wrote both the Old and the New Testaments in codex form.

While the initial impulse to employ the codex in the primitive Christian community may have arisen from a sincere desire to avoid transgressing the law, an entirely different psychology would have been at work in the gentile Christian world. There it may have been especially in order to demonstrate that the community is no longer bound by the law that the codex was received as the vehicle for Christian sacred texts. What originally may have been an expression of submission to Jewish tradition in another setting became its opposite: an expression of disregard, if not contempt for, the Law. By the time the Church had become a largely gentile community-that is, by about the middle of the second century-Christianity had disavowed the use of the roll for biblical literature. (Resnick 1992: 12)

Canon separates commentary from scripture; thus no book can be both commentary and canonical scripture. This is an important ideological function of canon. Of course, every text is in some ways a commentary on other texts, for no text is ever read or written in a textual vacuum. The texts of both the Jewish scriptures and the New Testament were written and copied and re-copied in ancient intertextual milieux in which they were echoing yet other texts, including each other. As Borges's essay suggests, intertextuality extends far beyond the conscious 'influence' of one text on another text's author. As we have seen, one thing canon does is to distinguish between those texts that deserve commentary, and all others. When a text functions as 'canonical', it no longer functions as a 'commentary'. To read that text as a commentary would be to decanonize it. The New Testament creates the Old Testament as the canonical texts that deserve its own commentary. But in so doing, the New Testament would seem to rule out itself as a canon. If the New Testament is a commentary on the Old Testament scriptures, then it cannot be canonical scripture in its own right.

Barr disputes the idea that the New Testament is a commentary on the Old Testament and claims that the reverse description would be more accurate:

The business of the New Testament is not primarily to tell what the Old really means, but to declare a new substance which for the Old was not there, although it was understood that it had prophesied its future coming.... It is more correct to say that the Old Testament was used to interpret the situations and events of the New. (1983:70)

This does not resolve the problem, since the Old Testament also cannot be both commentary and canon. In either case the point is that Christians give the New Testament a privileged position in relation to the Old Testament. Neither testament of the Christian Bible can be both a commentary on the other testament and a canon in its own right. This contradiction accounts for many of the intertextual tensions between the two Christian Testaments.

This in turn leads to yet another factor to be considered in the relation between the Old and the New Testaments, which we have already hinted at: the question of translation. Disputes over canon are disputes over texts, but they are also disputes over translation: is the translated text still 'scripture', still authoritative? The various Jewish legends about the writing of the Septuagint, the Greek translation of the Hebrew scriptures, reflect a profound anxiety over this matter. Likewise, Christians remain uncomfortable over the apparent difference in language between the New Testament texts, written originally in Greek, and the language of Jesus and his earliest followers, which was most probably Aramaic. The New Testament writings appear 'always already' translated. The alternative is that the words of Jesus are hopelessly lost. Furthermore, in their citations of the Jewish scriptures, the New Testament texts quote the Septuagint, not the Hebrew text (Aland and Aland 1987: 52), and so the Old Testament also is canonized in translation. The Old Testament was never the 'Hebrew scriptures'! In order to hold that the New Testament is canonical, Christians have to affirm the possibility of perfect translation. The theological consequences of this belief are considerable.8 Christian affirmation of translatability is apparent in the theory of 'dynamic equivalence' that now dominates the study and practice of Christian Bible translation.

In order to accept the New Testament as canonical, Christianity had to adopt a fundamentally different understanding of translation and language, and hence of scripture and canon, than did Judaism.

While the collections of 'Old Testament' and 'Written Torah' might more or less consist of the same contents-they clearly do not function in the same way vis a vis their applications in 'New Testament' or the 'Oral Torah'.... [I]n Judaism the locus of God's revelation is in the text alone. Scripture itself is the revelation. Among early Christians (and among later ones, I suspect, too), the revelation is something extra-textual, the events in the life, death, and resurrection of Jesus Christ, for which 'scripture' is primarily a testimony, an affadavit that it happened. (Brumberg-Kraus 1995)

In choosing the 'Greek', logocentric view of language, Christianity made a fundamental theological move-an ideological move-which redefined the nature of the scriptures. The Christians chose to understand the scriptures as a uniform, authoritative, and perfectly translatable message, an ideal content that could be readily and completely separated from the physical aspects of the texts in which that message was contained. It may be argued that this ideological move was unavoidable, given the early history of the Christian movement and especially the sort of texts that eventually became the Christian scriptures-that is, given the Christian need to treat those texts as scripture.

To have opted otherwise would have required a very different sort of Christian faith. The alternative would have required a fundamental sacrifice in the Christian understanding of language and of narrative, equal to or perhaps even greater than the failure to understand the significance of a double canon. '[T]he very renunciation of the pleasure of the text, understood as story and about bodies, is itself a turning from corporeal pleasure to spiritual contemplation' (Boyarin 1992: 481). This sacrifice may have had important repercussions for Christian attitudes towards Jews, Muslims, and others, for according to the Christian ideology the Bible must be brimming with clear, coherent meaning. How could any willing reader not understand the message of the Bible? Might these views of scripture and of canon also have played their part in the emergence and the long sorry history of Christian anti-semitism?

* An earlier form of the following article was presented at the inaugural meeting of the Bible and Critical Theory Seminar, Newman College, Melbourne University, 2 July, 1998.

1. This may or may not be the same as what Barr (1983) calls the 'final form' of the text.

2. Several cases of this are described in Aichele 1996.

3. See also Dan 1986: 128 and Kermode 1986: 184-85. The Alands' distinction between two fundamentally different views of writing is an important one, although the association with 'Greeks' and 'orientals' seems racist.

4. The inclusion of multiple gospels does not encourage diversity but rather subjects diversity to an emphasis on a greater unity. See Gamble 1985: 24-35, 88.

5. Although Resnick 1992 does not actually say as much, his overall argument is congenial to this view.

6. Gamble (1985: 27) suggests that in the second century CE this may have been 'common practice'. In any case it distinguishes Marcion's collection from the eventual Christian collection.

7. This is somewhat reminiscent of T.S. Eliot's essay, 'Tradition and the Individual Talent' (1964: 3-11), and indeed Borges includes a footnote to Eliot. See also Fishbane 1986: 34. Funk 1975 uses Borges's essay to support readings of Jesus's parables.

8. These are explored at some length in Aichele 1997, especially ch. 1. See also the essays in Bailey and Pippin 199G, especially R.P. Carroll's 'Cultural Encroachment and Bible Translation: Observations on Elements of Violence, Race, and Class in the Production of Bibles in Translation' (pp. 39-53).

Aichele, George (1985) The Limits of Story. Chico: Scholars Press.

-(1996) Jesus Framed. Biblical Limits; London: Routledge.

-(1997) Sign, Text, Scripture: Semiotics and the Bible. Interventions; Sheffield: Sheffield Academic Press

Aichele, G. and G.A. Phillips, eds. (1995) Semeia 69/70. Atlanta: Scholars Press.

Aland, K. and B. Aland (1987) The Text of the New Testament. Trans. E.F. Rhodes; Grand Rapids/Leiden: Eerdmanns/Brill.

Aristotle (1967) Poetics. Trans G. Else; Ann Arbor: University of Michigan Press.

Bailey, R.C. and T. Pippin, eds. (1996) Semeia 76. Atlanta: Scholars Press.

Barnstone, W. (1993) The Poetics of Translation: History, Theory, Practice. New Haven: Yale University Press.

Barr, J. (1983) Holy Scripture: Canon, Authority, Criticism. Philadelphia: Westminster Press.

Barthes, R. (1988) The Semiotic Challenge. Trans. R. Howard; New York: Hill and Wang.

Barton, J. (1986) Oracles of God: Perceptions of Ancient Prophecy in Israel After the Exile. London: Darton, Longman and Todd.

Berquist, J.L. (1996) 'Postcolonialism and Imperial Motives for Canonization', Semeia 7S: 15-35.

Borges, J.L. (1962) 'Pierre Menard, Author of Don Quixote', in A. Kerrigan (trans. and ed.) Ficciones. New York: Grove Press.

-(1964) 'Kafka and His Precursors' (trans. J.E. Irby), in D.A. Yates and J.E. Irby (eds.), Labyrinths. New York: New Directions.

Boyarin, D. (1992) "'This We Know to Be the Carnal Israel": Circumcision and the Erotic Life of God and Israel', Critical Inquiry 18: 474-505.

Brumberg-Kraus, J. (1995) 'Re: Canon', email post to IOUDAIOS-L [http://ccat.sas.upenn. edu/ioudaios/}, 19 October, 1995.

Dan, J. (1986) 'Midrash and the Dawn of Kabbalah', in G.H. Hartman and S. Budick (eds.), Midrash and Literature. New Haven: Yale University Press.

Debray, R. (1996) Media Manifestos. Trans. E. Rauth; London: Verso.

Deist, F. (1995) 'Canonical Literature. Some Ideology-Critical Observations', in PJ. Nel and D.J. van den Berg (eds.), Concepts of Textuality and Religious Texts. Acta Academica Supplementum, 1; Bloemfonteine: UOVS-SASOL Biblioteek.

Derrida, J. (1986) 'Shibboleth', in G.H. Hartman and S. Budick (eds.), Midrash and Literature. New Haven: Yale University Press.

Eliot, T.S. (1964) Selected Essays. New York: Harcourt, Brace, and World.

Faur, J. (1986) Golden Doves with Silver Dots. Bloomington: Indiana University Press.

Fishbane, M. (1986) 'Inner Biblical Exegesis: Types and Strategies of Interpretation in Ancient Israel', in G.H. Hartman and S. Budick (eds.), Midrash and Literature. New Haven: Yale University Press.

Funk, R.W. (1975) Jesus as Precursor. Missoula: Scholars Press.

Gamble, H.Y. (1985) The New Testament Canon: Its Making and Meaning. Philadelphia: Fortress Press.

Graham, W.A. (1987) Beyond the Written Word. Cambridge: Cambridge University Press

Ingarden, R. (1973) The Literary Work of Art. Trans. G.G. Grabowicz; Evanston: Northwestern University Press.

Jakobson, R. (1987) Language and Literature. Ed. K. Pomorska and S. Rudy; Cambridge, MA: Belknap Press.

Kermode, F. (1986) 'The Plain Sense of Things', in G.H. Hartman and S. Budick (eds.), Midrash and Literature. New Haven: Yale University Press.

Kristeva, J. (1980) Desire in Language. Trans. T. Gora, A. Jardine and L.S. Roudiez; New York: Columbia University Press.

-(1984) Revolution in Poetic Language. Trans. L.S. Roudiez; New York: Columbia University Press

Peirce, C.S. (1992) The Essential Peirce: Selected Philosophical Writings. 1. 1867-1893. Ed. N. Houser and C. Kloesel; Bloomington: Indiana University Press.

Phillips, G.A. (1980) 'Structure and Narration, An Enunciative View', Semeia 18: 131-35.

Rabkin, E.S. (1976) The Fantastic in Literature. Princeton: Princeton University Press.

Resnick, I.M. (1992) 'The Codex in Early Jewish and Christian Communities', Journal of Religious History 17: 1 - 17.

Zizek, S., ed. (1994) Mapping Ideology. London: Verso.