We shall be unable to make any progress in addressing the problem of the canon if we ignore the relevant theoretical considerations. If we start from a simple, monistic concept of The Canon, we will either condemn ourselves to reaching dubious and cheap conclusions or at least be forced to move on and to refine, qualify, and modify those conclusions.

The constraints of this paper mean that this clearing of the ground in advance of mapping the canon must be left to another occasion. However, I will try to summarize the problems faced by anyone undertaking this task. Although such separation may seem a little artificial, I shall address independently the canon, and the fields, institutions, and segments of society in which it manifests itself.

First, let us identify some possible conceptions of the canon. A fairly common view of it is as a body of carefully selected texts. In this connection it is important to explore what these texts are, and how they came together to form the canon. It can also be asked what common features - that is, their content, origin, and structure - are to be found in these texts, presupposing and insisting that our object of study is one thing and that things, products, or text-objects are what is in question. According to the "Great Works" conception, the canon is made up of ready-made products, a collection of examples, a sum of respectable texts which count as cornerstones of our culture and of our tradition.

One could label this conception of the canon the "canon-as-texts" standpoint; for the scholar working along these lines, it would be enough to identify which texts comprise a given canon - that is, what the "privileged" texts of a period are - and then seek to confirm his intuition, resulting in a list of works designated as the canon of this or that era or of this or that society, stratum, group, and so on. It would not be fair to dismiss this line of inquiry too quickly. It belongs to a certain positivistic pattern of thought, insofar as it seeks data, orders them, and then draws, in a strict logical framework, conclusions. It restricts itself to a controlled body of evidence, without referring to anything outside it. This methodology may produce results interesting from other, even anti-positivistic points of view, but it would be misleading to say that it is a prerequisite of other forms of inquiry since it presupposes a stage of study in which no preconceptions, prejudices, ideologies, theories, or even concepts are involved. However, it may be acknowledged that a mapping of what has been regarded as a canon may furnish raw materials for further analysis; reviews of this kind at least provide us with a view of the canon as produced by a professional in the field.

However, it is important to note that this approach is not widely practised, at least in its pure form. Even the most conservative, "scholastic" formulations refer to something outside - in fact, transcending them - the texts in question: tradition, values, ethics, aesthetic quality, and the like. The latter constitute a reference to a certain rule beyond the objects. It is extremely difficult, to say the least, to differentiate between "textual" and "linguistic" conceptions; that is, "textual" delimitations imply (often explicitly) a rule governing the selection of the texts.

The other - if it can really be said to be distinct - main conception of the canon may be dubbed the "canon-as-'langue'" standpoint. It is no less problematic than the first, but it does give rise to different questions. The nature of the canon can be treated as being similar to what Saussure called 'langue', in the sense that it embodies or manifests "common knowledge". Let us further suppose that there exists some sort of "literary competence", a knowledge which enables the "native speaker" of a culture to recognize literary texts as such. It is possible to argue that the canon, a selected set of "great works", is part of this knowledge. Consequently, if this knowledge can be said to correspond to the linguistic "langue", the canon would be part of a literary "langue".

If this is the case the canon does not belong to the sphere of realization or performance, but to that of competence or the "system". To this extent, we cannot do without a canon. This is confirmed by the fact that change in the canon is not autonomous, but rather due to influences external to literature.

The canon can also be regarded as an entrenched or institutionalised variety of interpretation. What changes is a particular set of interpretative assumptions. In tandem with this - or perhaps as a consequence of it - the texts picked out as valuable, as suitable objects of analysis or education, also change. This brings us back to what we said about the "canon-as-texts" conception: the selection of texts worth studying, the body of literature to be interpreted is itself the result of (explicit or implicit) rules, theoretical considerations, and value preferences which, in turn, can be described as a system beyond or within the objects in question, which, as a consequence, are no longer objects-in-themselves, but products of subjectivity, preformed by interpretation.

The "canon-as-langue" conception is sometimes formed in opposition to a conception of the canon as a list of texts canonical in their own right. What it opposes is the idea that it is even possible for a text to be canonical by virtue of its properties. In this connection, we may suggest that there may be another conception which deviates from the (structuralist) langue/parole distinction: the canon is neither a set of elements nor an abstract system, but a product of conventionalised acts, and canon formation is the performance of a - particularly potent - act. The creation of the canon is originally and archetypically the task of the "priest", the professional interpreter of "the Law".

Whatever it may be, the canon is certainly not something given or eternal, whose origin and nature cannot be sought. Even if they are not interested in the development or origin of a canon, most studies agree that it can be traced back not only in time, but also synchronically, to its conception.

But who makes, changes, and maintains the canon? The answer, naturally, is that it is the function of some groups, who may be called professionals, or people in the culture 'whose power is obtained through their control over literary institutions', as Zohar Shavit puts it.1 The least which can be said of the relation between the canon and literary professionals is that the practice of the latter affects the former in one way or another. Literary professionals are often characterized by their preoccupation with the canon and, conversely, the canon is almost exclusively approached as the principal field of a specific group of people, namely, literary professionals. I have already illustrated the important role played by literary professionals in forming the canon - and, to be sure, that of the canon in establishing the status of these professionals - in terms of the paradigmatic case of canonization, the delimitation of the texts which belong to Scripture. This is a practise of a highly professionalised nature and has serious consequences for numerous communities.

It seems reasonable to differentiate between the unconditional authority and the more restricted and ephemeral power which literary professionals exert. Partly, of course, the difference is only a matter of degree. But both forms of influence presuppose institutions, power, and groups. Furthermore, we may hope to specify exactly what groups and what powers take part in the formation of the canon and how they do so.

The relation of literary professionals to the canon is not confined to the fact that they constitute a more or less institutionalised group of "cultural workers" responsible for forming and maintaining a distinct class of texts. This special class of texts offers them a basis of existence as professionals. It therefore constitutes a field of operations for them as well as a field of rewards.

However, even if the community of professionals seems to have a decisive influence in the process of forming the canon, we must repeatedly ask whether it is a single thing which is formed? Why should we regard a canon formed by a special community of professionals as the only important one? No doubt, the best documented and best known canons (whatever that may mean) are of this kind: they are passed on to the next generation, taught in schools, respected above all others, and represent a set of texts and conventions which is regarded as a prerequisite if one is to be entitled to call oneself educated. However, it is clear that what - and how - people read is far from identical with what the professionals (or some of them) regard as canonical. For instance, the "great works" taught in schools do not have to be identical with those most revered by the professionals; at least since the Romantic period, a section of literary professionals has represented a counter-canon, to which institutional education has appeared as a symbol of conservatism. Although the curriculum for school education is formed by the professional powers-that-be, those not in power often challenge this sort of canon. On the other hand, it must be supposed that particular communities create their own canons. Those involved in education (both pupils and teachers) do not regard the obligatory readings as the most valuable works, and what and how they read most is quite different from that canon.

The problem is that this canon is hidden, fragmentary, implicit, and extremely hard to trace. Synchronically, we are aware of its existence, but rarely do we care to describe it; diachronically we may have some data at our disposal, mostly supplied by professionals, but very few direct references. Another problem concerns the number of non-professional canons (and, accordingly, how many communities) should be taken into account. It may be misleading to speak of a particular non-professional canon as a single whole; there may be several sub-communities, and, correspondingly, sub-canons and counter-canons. Theoretical speculation must circumscribe, prior to any research, the possible range of canons which should be taken into account.

Needless to say, this range will differ according to culture, historical period, and even nation. We may speculate that a small country (or a language spoken by few people) must insist much more rigidly on its sacred texts and their sanctified (canonised) interpretation. On the other hand, a small country is much more exposed to external influences, especially if it has powerful and "great" nations as neighbours.

This leads us to the second complex of issues: what are we doing when we investigate canons? What sort of data, evidence, or clues can we expect to find? What is the method of canonical research?

First, what we hope to find is texts. (This will sooner or later raise the problem of interpretation, but I shall put it aside for the moment.) I am not talking about the "great works" included in the canon. A text will not display any sign of being canonical, no matter how closely we investigate it; canonicity is a matter of interpretations, institutions, communities, powers, traditions, and so on. The texts in question are those surrounding the canonical texts; the documents of its reception, its interpretations, the traces of its path to canonicity.

Thus, the first thing we could look at might be what we might call interpretations, and interpretations of all kinds: authors" comments and diaries, literary criticism, school textbooks, school papers, theatre performances, prefaces and editorial comments, blurbs, advertisements, all sorts of references.

Secondly, these texts (canonical and meta-canonical) appear somewhere, in particular contexts. Another field of research would be to map all possible sources of such texts: what journals, newspapers, textbooks, TV and radio channels, or other forms of mediation might play a role in the circulation of a work and its interpretation? What do the elementary and secondary school and the university curricula look like? How and where were the teachers trained? What other forms of opinion-forming were available?

A third area of interest is institutions. The sites where these texts appear all belong to institutions: the system of education, the system of literary production, circulation and reception, including literary schools, trends and groups, salons and movements; book publishing, along with its participants (censors, editors, marketing personnel, and so on), the book market, libraries, and readers" clubs.

All these elements require thorough examination, and we should always bear in mind what we have already said about the concept of the canon in general, that is, that there are always several canons, and that a canon is not simply a set of texts. Thus, we must proceed to a fourth area, that of assessing the relevant conventions of interpretation. This seems to be an even more difficult task. It is not enough, clearly, to map all possible meta-canonical texts, their context, and their institutions; we also have to interpret the interpretations, to evaluate the evaluations.

This is a rather delicate and difficult business. What we should do in canonical research is to survey the language the meta-canonical texts use; their value preferences, their hidden interpretative presuppositions, their political, ideological, or personal motives, and, of course, to pick out those which may count in the reception history of the work in question. I am not talking only about literary criticism; such facts as the most popular books in public libraries also require careful investigation. Although not of the same nature as the investigation of critical texts, the whole structure of, say, public library acquisitions and lending policies, and which sections are closed or open to the public should be studied. Similarly, a curriculum as it stands cannot be regarded as useful raw data for canonical investigation. It may well be that some works appear on some lists merely for political reasons, but in fact are never taught. It is also possible that certain works are regarded as typical or highly controversial, but never as canonical.

We must therefore interpret the meta-canonical texts, their contexts, and the institutions which make use of them. Of course, we always have some intuition of what counts, in a given period and in a given culture, as canonical. However, it may happen that the temporal distance is too great and it turns out that the canon was somewhat different from what we supposed. It may turn out that other, lesser known works belong to that canon, which have since dropped out of the tradition. Or it may happen that the reasons for a work becoming canonical are quite different from what we supposed. Literary works may have different functions in different periods, while preserving their canonical status. This seems especially true in cases of "small" literatures: poems or epics which once served nationalistic ends, or the struggle against alien invaders, may remain canonical, although their interpretation or evaluation is now quite different.

To summarise, we cannot even attempt canonical research until we face a number of important questions. What are we looking for? Is it one thing or more? Is it a set of texts or something more? Where should we look for it? Only in reference books, textbooks, and curricula? Or should we take into account all the corresponding interpretations, along with the channels and institutions mediating the texts and meta-texts? My conclusion is that all these questions must be answered.

*

I have argued that a canon is not merely a set of texts, but a range of socially preferred interpretations of texts; that is, what is taken as canonical is a text together with its interpretations, and so this interpretation is also taken as canonical. To repeat, 'the canon can also be regarded as an entrenched or institutionalised variety of interpretation. What changes is a particular set of interpretative assumptions. In tandem with this - or perhaps as a consequence of it - the texts picked out as valuable, as suitable objects of analysis or education also change'. This would explain the fact that new texts can be incorporated into the canon: a canon is not fixed, but can be extended, enriched, and modified.

This brings us to an important point. Because a text is part of the canon together with its interpretations, other texts which in some way resemble the first can be interpreted much more easily and so incorporated into the canon.

A clear example of this is the patriotic poetry of the nineteenth century. Adherents of this trend would be more likely to become part of the canon than those who ignore or turn against it. This may be one of the causes of epigonism. The issue must be much more complicated than that, however. Even if texts become canonical together with their interpretations, not just any canonical texts generate a set of canonical interpretations. The interpretive tradition of Balzac would promote the inclusion of minor Hungarian Realist writers in the Hungarian canon, but not vice versa; we often read second-rate poetry in accordance with the conventions used for the interpretation of great, canonized poetry, but we do not extend our interpretations of second-rate poetry to the canonized variety. It may be that there are levels of canonicity, depending on the corresponding canonical interpretation.

Further, the history of canonical interpretative conventions is far from being the same as that of canonical texts. Canonical interpretations (or canons of interpretation) may prove to be much more long-lasting or conservative than the texts themselves. It may well be that some texts are no longer read while their interpretations still influence the interpretation of later texts. We must then ask what makes interpretations survive while their corresponding texts fade away.

The powers which stand behind these canonical interpretations, underpinning and changing them, must also be examined. While we may hope to find the particular critics, interpreters, and institutions responsible for the canonization of particular canonical texts, it would be more difficult to trace the formation of a canonical interpretation. As already mentioned, some texts may preserve their canonical position even if their interpretation is modified, but we must ask why some canonical interpretations or canonized strategies of interpretation preserve their status.

Canonical interpretations could be viewed as higher or even as great narratives. They recount how an encounter between the text and the reader should take place. A canonical interpretation is a general scenario which may take particular forms in particular cases. Let me give some examples. I will outline two canonized interpretations, or rather two patterns of canonized interpretation. Neither of them is connected to a specific literary text, but to a set of texts. The first example is the canonical reception of Realist narrative; the second is the canonical position of folk art in "high" literature.

In the case of the Realist novel, the canonical interpretation (that is, the narrative describing - or rather prescribing - the text/reader relationship) goes something like this: the uninformed reader turns to the text in order to gain information about society (or the history of a society), to obtain an insight into the hidden motives of social actors (or historical actors). The text, as a good, reliable, canonical Realist text, fulfils these expectations, and, moreover, offers some patterns of behaviour or role models. The role of the reader is to look through the text; the text is transparent for the reader. It can either be exhausted, or at least its pool of meanings is rather restricted. It is, to borrow Roland Barthes"s term, "lisible" [readable].

Think of the reception of Realist works in the second half of the nineteenth century and in the twentieth century. The reflection of real life, providing an account of social strata, and unmasking hidden motives have long been the most important elements of the interpretation of Realist narratives. What count as the cornerstones of interpretation are the denotative and ideological functions of the text (rather than, say, textual characteristics or intertextual relations). The novelists of the second half of the nineteenth and the first half of the twentieth century were praised for their ability to show the true history or present state of society, because of their illustrative capacity and their reliance on facts and reality. This pattern has been modified over time: Realist narrative has been regarded as a modern chronicle, a reliable account of the present or the past, and an essentially objective form of literature (as opposed to the subjectivity of lyric poetry). Narrative literature is read, according to this interpretation, in order to gain access to the "facts of life", as well as to draw some ethical conclusions from the lives of the people depicted in the narrative. A narrative, then, can be included in the (Realist) canon if and only if it can be read or interpreted along these lines, that is, if it complies with the canonized interpretations of Realist narrative.

Later, in post-war Hungarian literary history for instance, Realism became a magical label which served as a tool to legitimise Romantic, Classicist, or other texts. If the traits (or traces) of Realism could be detected in a literary work, it deserved its position in the canon. Following Engels" remarks on Balzac, Realists were regarded as inherently and perhaps unconsciously revolutionaries. On the other hand, new works had to be read as Realist in order to accord with the ruling canon. There has also been a tendency for modern novels, whether those of the 1920s or those of the 1960s, to be subsumed under Realism: this was at least partly motivated by ideological, even political considerations. If the literary historians, those in charge of the defence and maintenance of the canon, wish to include outstanding works which would otherwise be left out, a label like Realism is fairly accommodating.

This picture may seem like a caricature, but the point is that the vulnerability or fallibility of a great narrative, the historically transitory character of a canonical interpretation, becomes tangible only when a competing interpretation emerges. For instance, the most canonical figure of Realist narrative, Balzac, whose interpretation seemed to govern all other interpretations of the Realist novel, was radically reinterpreted by Roland Barthes in S/Z. One of the consequences of this reinterpretation was that, although Realism preserved its canonical position, the whole scenario of the desired interpretation was rewritten. The Realist novel was no longer transparent, but was shown to have a subtle textual structure, and its representational nature was not its most important feature. Consequently, the structure of the canon was modified: perhaps some of the "lisible" works will drop out, whereas others (earlier regarded as traditional Realist works) will prove to be "scriptible" [writable], to use Barthes" complementary term.

Let me turn to my second example. In the last two centuries, and especially in some Central and East European literatures, popular art became a major source of "high" or canonized literature. More specifically, from the age of Romanticism (and perhaps earlier) onwards, the importance of folk song and folk tale in literature increased dramatically. That is, a specific interpretation of folk literature served as a point of reference for a number of influential writers. One of these interpretations - or rather a general pattern of interpretation - was to take folk literature as representing a counter-culture, as if folk songs or folk tales expressed, fundamentally and essentially, a gesture of resistance.

What is the canonical interpretation corresponding to the poetry of folk origin? What is the scenario (or narrative) of the encounter of the text and its reader?

First, the reader must realize that what he faces has an intertextual relationship to what he believes is folk poetry. It is thus presupposed that the reader has at his disposal a sort of repertoire of popular culture: specific rhyme patterns, repetitions, parallelisms, meter, thematic structure, genre rules, and the like. The reader must interpret this as a hallmark of "popularity". Moreover, he will suppose that since folk literature is, by its nature, essentially anti-totalitarian or subversive, even if it has a revolutionary character, texts which refer back to this source will also be, by their very nature, subversive. For instance, not only did one of the major Hungarian poets of the nineteenth century, Sándor Petőfi, become an emblem of the harmony of poetry and revolutionary thought, but, following the scenario of his canonical interpretation, the literary reception of popular literature became an expression of progressive and patriotic attitudes. A slight reference to folk poetry rang the bell of national resistance or solidarity with the poor.

This tradition of interpretation became canonical up to the twentieth century, maybe even to the present day. Turning to folk literature and entering into all sorts of intertextual relations with it was perhaps not a sine qua non for canonicity, but it definitely made it easier for a text to be connected to the great tradition and so to become part of the canon.

To take an early example, the great Hungarian poet Endre Ady once used some lines from an old Hungarian folk song. The folk song concerned a peacock which flies onto the roof of the county council building, as a result of which the poor prisoners are freed. For Ady, it became a symbol of political freedom; in this interpretation (that is, in the poem which itself is an interpretation of the original subtext of the folk song), he followed the tradition of regarding the poor prisoner, the outlaw, even the bandit as an emblematic figure of freedom - something like cowboys for Americans - and of regarding folk ballads and folk songs as the most genuine expressions of people"s desire to live freely. It is not accidental that this was one of the few poems by Ady which became famous in Hungary in an arrangement for choir by Kodály.

Ady was a Modern poet, and this poem is more or less unique. There are very few references in his work to folk poetry. Nevertheless, his poem, along with Kodály"s choral setting, has a solid place in the canon because it strengthens the canonical interpretation. Still, there may come another interpretation, a counter-canonical one, which may take the view that poems which make use of folk poetry are much more interesting in respect of their structural, rhythmic, or poetic character than in respect of their ideological character. Or there may be attempts to show that this vision of folk poetry can be called into question, that folk poetry is not always and not in itself subversive or anti-authoritarian. These reinterpretations would perhaps change, once again, the canonical scene, although the main figures and main works would retain their position.

Finally, it is worth reflecting on the fact that the most obvious examples of canonized interpretations are those of the nineteenth century and perhaps the Realism of the twentieth. That is, when one looks for well established and widely known conventions of reading, operations which enable new works to take their place in the canon, avant-garde, modernist, or even medieval conventions of reading never come to the fore. There may therefore be a canonical hierarchy even among the scenarios or narratives of interpretation. There may be a canon of canonical interpretations, and some of these may prove to be more powerful than others.