1.1. Introduction: Language as both, meaning and misunderstanding...........................1

1.2. A new theory of language meaning...............2

1.3. What is "discourse"?...........................2

1.4. Discourse and semantics........................4

1.4.1. Discourse as systems of order......................5

1.4.2. Semantics as discourse analysis....................6

1.5. The dimensions of discourse....................7

1.5.1. Structural organization............................7

1.5.2. Cognitive representation of discourse..............8

1.5.3. Signs and society..................................9

2. DISCOURSE MEANING........................10

2.1. Structural meaning............................10

2.2. Meaning beyond the level of a sentence........11

2.3. Language as discourse.........................12

2.3.1. Subjective position...............................12

2.3.2. Text, coherence and other institutions.....13

3. APPLICATION OF SEMANTIC DISCOURSE ANALYSIS......15

3.1. Explanation of discourse practices............15

3.2. A "tool-box"..................................17

3.2.1. Changes in the order and the discoursive treatment of a subject.........17

3.2.2. Elucidation of discourse strategies........18

3.2.3. Material conditions of discourse ..........19

3.3. Summary.......................................20

Bibliography.

This is a common experience: We read a line by Shakespeare, a friend talks to us allusively, we are asked to comment on a poem by e. e. Cummings. In each case we usually feel a bit uncomfortable because there is a gap between the text and our understanding of it. This gap appears in the question "What does the author mean by that?"

Obviously meaning is a very important aspect of language. Scientifically this field is the subject of the study of semantics. Traditional linguistic theory presupposed that proper language (as it is performed, i. e. parole) has an intended meaning and that the recipient has only to understand that meaning by means of interpretation to make communication work.

This concept assumes that the purpose of language is communication. In that case language is only a kind of code system, a carrier of information, which is capable of transmitting any intended meaning whatsoever. This concept seems to work quite well as regards words, for instance. We are sure that we know what is meant by words like "horse", "book" or "bottle". This is studied by morphology and lexicography, which are branches of semantics.

One could proceed from this starting point and explain the meaning of a sentence, too. Sentence meaning would depend, then, on a) the meaning of the single words and b) the way they are connected (syntax). Noam Chomsky has produced quite an elaborate theory to account for the meaning and the correctness of a sentence (transformational grammar).

All this presupposes, as has already been suggested, that language (as a code system) is transparent. It is quite obvious that this is not the case with language in use (parole) as it is often opaque, ambiguous or diffuse, and we often make the experience of misunderstandings. Besides, as my examples at the beginning of this chapter have shown, men do rarely consent to one distinctive meaning of a given text. Obviously the question of "meaning" and of "which method elucidates meaning" is not as easy as traditional linguistics might imply.

The question dealt with in this paper will be: Is there a linguistic theory which is capable of elucidating the meaning of all kinds of texts in all their different communicative situations (i.e. purpose, style, time, place, length, language, medium and so on)?

I will suggest that we have to invent a new dimension of language which is far more comprehensive than a sentence or even larger than a text. We can call it the dimension of discourse, which has already been introduced by a number of scientists (e. g. J. Habermas, M. Foucault, A. Easthope).

The semantic theory I am looking for is a kind of discourse science. Therefore I will first attempt to clarify the meaning of the term "discourse". Then I will examine to what extent we can assign meaning to this dimension of language and I will draw up a list of the respective sciences that are involved when analysing discourse.

In the second chapter I will check if conventional structuralism is able to give an account of discourse meaning. As this is not fully the case, it is necessary to introduce some new methods of linguistic investigation. This will lead to a concept which puts the notion of discourse into a larger framework of social and communicative sciences. Finally, I will try to derive methods of linguistic research which can be useful tools in literary criticism.

First it has to be made clear what is meant by the term "discourse".

Historically, there is a great deal of confusion about its definition. The term has been derived from French discours whose basic meaning is "talk". In linguistics it is often used in the sense of "a sequence of utterances":

But some scientists have assigned a more concrete meaning, which implies and presupposes more, to the term: Habermas claimed that discourse is (or, at least, ought to be) a kind of free and fair form of linguistic communication whose purpose is the finding of truth. Foucault used the term "discourse" to describe a level of language in which, in his opinion, language is organized in a way different from all traditional linguistic theory, but which is, nonetheless, essential for linguistic and philosophical research. This is the concept of discourse which I will largely adopt.

"Discourse... is a term which specifies the way that sentences form a consecutive order, take part in a whole which is homogeneous as well as heterogeneous". Therefore, discourse is a level of language use which is superordinate to sentences and texts.

As discourse is concerned with "the continuing exchange of meaning by means of language" it must not be confused with either of Chomsky's categories. It is neither confined to language in its actual utterances (performance, parole) nor is it only a question of language as a code system and as a system of communicative conventions (competence, langue). It contains elements of both, and yet, it goes beyond this distinction since it is the study of language use, regardless of rules or single speaking subjects, but regarding all possible and real utterances, in the past and, as far as predictable, in the future. Every string of utterances or printed letters is part of the universe of general human linguistic discourse. But, if people are able to make sense of utterances, language users must at least believe that utterances have a relation to common knowledge. To make language meaningful there have to be some reliable rules, there have to be recent forms of language use to which any future realization of language can be related. The most accepted set of rules is grammar. But the same goes for other forms of use, for example with respect to the subject dealt with or the choice of sign system and communicative channel (difference between speech and writing, the qualities of the characters in written or printed language, accent and qualities of the voice in spoken language).

Thus, we may infer a definition of what is to be meant by the term "discourse".

It is important to stress that this "string of utterances" need not be a string of words that one author has put in this or that order. The investigation of the discourse of teacher-pupil-interaction, for instance, will make use of as many utterances of that type as possible, regardless of "to whom they may be ascribed".

Here it becomes clear that, to a large extent, discourse is concerned with the production of meaning. One could insist on asking the transcendental question of what makes the production of meaning possible at all. But, though this question is undoubtedly interesting and very important, I will point to reality which gives us innumerable examples of this phenomenon (perlocution is a clear case of communication at work!).

Why can we expect discourse analysis to be helpful in semantics? What is a "semantic theory of discourse"?

First, discourse analysis puts the focus on the production of meaning, like pragmatics, and not on static and predetermined meaning, like a logic-oriented semantics, or structural analysis (in its strict interpretation).

Second, discourse is a structure which enables structuural analysis to go beyond the given temporal order of utterances. Structural analysis works perfectly within the limits of a given text. But how to cope with relations between several texts, perhaps from different ages, by different authors, of different cultures?- Texts with, nonetheless, obviously common features, texts, which exert certain influences on each other, at least as far as their interpretation is concerned. How can we check that two utterances deal with the same (or similar) subjects?

Traditionally, this is done by means of hermeneutics. But the severe criticism hermeneutics has suffered in this century has proved that there are a lot of metaphysical implications in hermeneutics: it presupposes that there are pre-given and, as it were, unified subjects (authors) whose intention is to communicate and that there is a clear and intended meaning (idea) in the message which is to be communicated. Besides, hermeneutics treats language as if it were transparent, i. e. that it is capable of neutrally transmitting any possible meaning. But it is debatable whether there is really "meaning" totally independent from the code system and from the way the code is used. After all, the traditional concept of meaning is, at least, inadequate to account for all aspects of the exchange of meaning.

Discourse analysis tries to avoid these presumptions by treating language as a series of utterances, coherent as well as discontinuous, and by treating communication that works as only one phenomenon of language use, besides, for instance, misunderstandings and phatic communication. Discourse analysis examines a certain discourse, e. g. pedagogical discourse, metaphysical discourse, discourse of narratives, discourse of men-women-interaction, power discourse, discourse of political song-texts and so on. It collects utterances belonging to the respective discourse. Once one has established a distinctive discourse it is possible to elucidate its internal organization and the way it functions: there may be significant changes in the order of a discourse in time, there may be distinct exclusions of certain utterances (utterances which do not appear in the respective discourse but which, logically, belong to it) in one discourse or there can be important interrelations and exchanges between two discourses, e. g. one discourse disappears as a distinctive one when it has been subsumed under another, or a new discourse is being created in order to prevent ideas from being connected with existing discourses.

Obviously the way language is distributed in discourse and to discourses, and the way these innumerable discourses are internally organized is more than merely formal. It is part of the meaning of utterances in so far as it influences the meaning-potential that can be created and exchanged. Distribution, organization, change and metamorphosis, functions of institutions and forms of language use in certain discourses- all these are features of the level of discourse.

The meaning of an utterance or a string of utterances is a function of the discourses in which it occurs and of the place and the function it is assigned in these discourses. This is the central thesis of the theory which is supposed to account for the real, possible and conceivable meanings of language and also for the production of these meanings.

To make this semantic theory of discourse work and to make it applicable to literary texts or to utterances in everyday life it is necessary to explicate the inferences of the definition of chapter 1.2. and of the thesis claimed above. It has already become clear that discourse theory leads towards a new understanding of semantics: We cannot look for a final and invariable meaning of language anymore. Instead, semantics becomes a study of discourse organization, of the creation or production of meaning.

Discourse analysis takes into account the whole communicative event including all conditions and functions which enable its taking place. For that reason the science of discourse analysis needs a rather comprehensive supply of methods. Just as language and its usage are subject to substantial changes all the time, the methods involved here will alter, too. But first we ought to provide a large variety of single sciences, each of which elucidates a certain aspect of discourse. As examples, I will introduce three central fields of study which must be regarded by discourse analysis: structuralism, cognitive sciences and semiotics. Beside these, there are other sciences which could be integrated into a concept of discourse analysis: study of communication (electronic media, printed media, distribution of informational sources and so on), pragmatics, psycholinguistics (of which cognitive sciences form a part), logic, sociolinguistics, social sciences...

Given a piece of spoken or written language as a text we can analyse its internal organization, its structures. "Structural descriptions characterize discourse at several levels or dimensions of analysis and in terms of many different units, categories, schematic patterns or relations". One can find a starting point for that kind of analysis in the traditional structuralism (in Chomsky's terminology): phonology describes the composition of oral utterances from phonemes, morphology describes the composition of words from morphemes, syntax explains the structure of a sentence and textlinguistics scrutinizes the internal structure of a text and the formal and thematic relations of texts to each other, narrativics examines the way stories are told, e. g. distribution of information. It is important to be aware of the difference between this kind of structural analysis and the traditional grammatical terminology (i. e. "case grammar") since we are concerned with the semantics of discourse. Geoffrey Leech has attempted to introduce a semantic theory of sentences which operates on the structural level.

Therefore Leech introduces componential analysis, which aims at finding out "meaning units" that together form a logical structure of meaning, i. e. the meaning of a sentence. "The conclusion is that the semantic unit within which componential analysis applies is not only smaller than a sentence, but is potentially larger than a word". Leech finds this unit in the concept of "predications" which consist of arguments and predicates. His "componential analysis" is a mixture of predicative logic and generative grammar. As it is strictly structure-oriented it cannot suffice for accounting for discourse meaning or meaning in terms of discourse analysis. But it is one form of semantic analysis, derived from structuralism, which is, if it is integrated into a larger, more comprehensive and more critical theory, a helpful tool.

One aspect of language which structural analysis totally ignores is that language (parole) is usually not produced by a computer or other "perfect" brains, but by human beings in certain social situations. Language production, and with it meaning production, depends on cognitive processes.

It is impossible to either produce or understand discourse whose structural complexitiy goes beyond the capacity of the human brain's language processing facilities. In a discussion, for example, a participant has only a certain degree of "overall-view": the whole state of affairs, the whole previous discourse cannot be represented by a speaker. Therefore his understanding of a present utterance and his reply to that will be within the limits of that phenomenon which Van Dijk calls "short-term memory capacity".

Another important aspect can be found in the strategies that a speaker conceives during a talk or while writing a text. These strategies are influenced by a lot of situative factors like the kind of people he is talking/writing to, the attitude he wants to pretend, the interest he takes in the subject and so on.

Here a hint to psychoanalytic criticism is also possible: Utterances, as Freud has already pointed out, may be influenced by the subconscious; something that has not been intended, but whose mentioning is paradigmatically possible, is uttered involuntarily ("Freudian slip").

Therefore, in addition to structural organization, we must take into account that discourse is also organised with regard to these variables: cognitive representation, situative consciousness, semantic strategies, influence of the subconscious.

A third important dimension of discourse has been explored by the science of semiotics. Discourse, on its most immediate and material level, is encoded in signs: spoken language uses the phonological sign system, written or printed language consists of letters taken from the alphabetical sign system. Semiotics, then, examines the nature of these signs and the way in which they work. Recent studies by Derrida have shown that primarily, signs do not refer to objects (or concepts, for that matter) in the real world, but to each other. That is, a sign has a certain meaning (reference to a concept or an object) because it is not any other sign from the paradigmatic axis. The phoneme /b/ of the word /bit/ refers first to the other possible phonemes, like /h/, / / or /p/, which, in front of /it/ form minimal pairs with /bit/. A contrast, a difference between two signs is necessary if a sign is to denote anything distinctive at all. This gives substantial reason to the preliminary answer "no" that I gave to the question of whether the linguistic sign system is a transparent vehicle for meanings. Obviously there is a pre-given structure of relations in the sign system which lies in the materiality of signs. The idealistic part of a sign, reference and the "thing" referred to, is subordinate to the sign as matter.

Moreover, the linguistic sign system, like any code to which public has access, is a social system. Society and its language use has considerable influence on the system itself (e. g. creation of new words or sentence structures, the establishment and breaking of taboos, the invention of new recording systems for language (writing, printing, copying, recording of speech, digital media). To do justice to the sociosemiotic nature of discourse semiotics has to be a study of signs in their relation to society. Signs, in this context, are not only single symbols like a letter or a phoneme. A text, for instance, is also a kind of sign which operates within discourse and which has a specific semantic function.

Now we have established the outlines of a discourse theory which is supposed to excel the traditional conceptions of semantics this theory can be put to the test.

In this second chapter I am going to examine the levels on which discourse may function as an exchange of meaning, as communication. I will describe meaning production on three different levels: 1) structure up to the level of a sentence (or utterance, in spoken language), 2) structure beyond this level, up to the level of a text, a book or any other conceivable linguistic aggregate, and 3) the organization of language as a whole in dimensions of discourse, i. e. the discursive order which any language as a system as well as any real utterance owns.

First, we shall take a look at the structural organization of language. A sentence, or an oral utterance, is a combination of smaller units. A choice out of the possible different phonemes/letters is syntagmatically combined with other choices from that level and together they can form, for instance, a morpheme. Thus, there is a paradigmatic collection of possible morphemes, which again may be syntagmatically combined to words, words, then, can form a sentence or utterance.

From this point of view the meaning of a sentence is primarily a function of the meaning of its constituents. Logically, the meaning of an utterance is the sum or the relation of several semantic units (e. g. Leech's 'predications', cf. chapter 1.3.1. of this paper). From the distinctive combination of sounds or letters in words, and of these words in a sentence, we may therefore derive the basic features of a semantic characterization of that sentece. Moreover, there are other structural features to be regarded, like stress and intonation in spoken language, and visual arrangement in printed language.

But the same sentence may have different meanings, different discourse positions, in variable contexts: The sentence "Mary took his hand and led him to the door" has quite different functions in a detective story (Mary as police officer, "he" as a criminal), in a love story or in a narrative where he is simply blind. Yet, there are common features (somebody is led by somebody else) which function in different discourses.

In addition, there are sentences whose meanings in discourse are totally different from the pure grammatical and lexical sense, e. g. the German expression "Gehen Sie doch dahin, wo der Pfeffer wächst!" is not a polite piece of advice for a disoriented person, but a curse, an insult. This seems to be more a question of shared social and linguistic knowledge (maybe the term "discourse knowledge" might be adequate), not of grammatical and lexical rules.

Obviously, "meaning", as a communicative phenomenon, cannot be divided into independent and single meaning units. Sentences/utterances form a text. Texts, again, are combined in fanciful ways: different texts are put together in a newspaper, a magazine, in a book of selected texts, in a library or bookshop where many books containing different texts are placed side by side.

The game of structural composition can be continued also beyond the level of a sentence. Sentences also form a paradigmatic axis from which selected signs are combined to a syntagmatic axis. The internal organization of a text can be described by the grammatical and deictical phenomenon of "cohesion" (Different statements refer to the same persons, objects, time, place...). "Coherence" describes the same thing on the semantic level: The "plot" "behind" the real text is coherent if it has a kind of internal logic. Such relations also occur between different texts, which is called "intertextuality". The traditional idea is that one may derive meaning from the meaning-potential of sentences and the way they have been put together, provided that there is a certain degree of cohesion and coherence.

After all we get two poles between which structural organization works: the smallest unit of signs (phonemes/letters) and their largest conceivable syntagmatic arrangement (in the end: all actual human linguistic discourse, in the past and in the future). The meaning of a certain syntagm is deliberate in the sense that we cannot derive the meaning of a word from the meaning of its letters, since letters do not have any immediate meaning, apart from their differences between each other.

It has become clear that one variable in the complex phenomenon of discourse meaning is its structural organization. But there are other more opaque aspects of discourse which play a role in meaning production.

To come back to my example from the previous chapter ("Mary took his hand and led him to the door"), a further problem is to find out who speaks these words. Is it a witness, a poet, an author who collects stories told by other people... Does the speaker have a personal relationship to the persons mentioned in the sentence? Who are the people mentioned? - All these questions can not be accounted for by simply analysing the structure of the sentence.

Each language user has a certain position in discourse. First of all there are factors like social position and means of communication which inevitably create a posiion from which somebody uses language. Second, as soon as somebody starts to write or to speak an ideological position is established. The language user becomes unavoidably the subject, the bearer of certain concepts, of beliefs, of ideologies. Such a position, as Easthope has shown, is not autonomous, it cannot exist before language, or without language. One cannot be in such a position without starting to speak.

From a social point of view the position created by any speaker depends on his social, cognitive and linguistic competence. Halliday relates social competence and social situations to the scope in which meaning production may take place. He introduces the concept of "register", which "is the meaning potential that is accessible in a given social context". Therefore, in a given social communicative (or discourse) situation, certain meanings cannot be produced if they are out of the participants' reach, i. e. out of register. This restricts the positions constructed by language use and distributes the power of discourse control.

These subjective positions, and the means by which they are created (e. g. concepts like religious "belief", "Marxism", certain verbs like "to think", "to assume", "to know"), which enable language to construct subjective position, contain a great deal of "meaning", since they reveal the way in which certain discourses function.

A lot of ideas we usually accept as if they were God-given can be treated as institutions, as practices which make discourse possible and which, at the same time, give a primary order to discourse.

Coherence, for instance, is a property of discourse which makes texts more interpretable.

The essential terms are: overall control (upon discourse), constraint, whose lack causes confusion, since the meaning goes astray. Thus, global coherence is a presupposition for the genesis of meaning (in the traditional, static sense), but it is also a means to constrain discourse, to keep it within the boundaries of "rationality".

"Text" is another institution which turns out to be rather meaningful since it creates a new dimension of signs. The concept of text allows interpreters to treat a given text as one sign. Through global coherence, a text is supposed to have one distinctive subject, to have one more or less clear meaning, to be univocal.

Concepts like "coherence" or "text" serve as means of exclusion: They prevent certain discourses to appear as text or as sensible language at all. Before textlinguistics accepted that everyday speech (and not only language as "literature") also contains "texts", the term was restricted to pieces of language which had been intended to be "texts" (an essay or a novel, for instance, but certainly not a dialogue at school or the minutes of a business meeting).

The order of discourse that I introduced as characteristic of language as discourse in chapter 1 also turns out to function in a certain way in language use. Michel Foucault describes some structures of order that have become fixed due to their useful function for those users who are in control of that discoursive orders.

The organization of knowledge in academic life may serve as an example. The formation of different "fields of studies", of a number of "faculties" and "sciences" will influence the ideas that can be developed in them. Foucault commits himself to a historical analysis of what kind of study used to be allowed to be "biological" during the 19th century. He proves that the "truth" of scientific claims in the study of biology depends on the respective historic ideas of what biology is about, on the academic practices in biology, which are expressed and structured by means of language.

Thus, semantic analysis ought to include an examination of these discourse requirements, of its practices and fixed orders, in order to make explicit the tacit presumptions which lie behind or which are overprinted by the really expressed utterances.

By making these concepts and their function explicit we learn about the conditions of meaning productions and about the nature of the boundaries that limit the scope within meaning production takes place.

Now we have established a rather comprehensive theory of the semantics of language as discourse what use does it have in applied linguistics or textual (e. g. literary) criticism?

First of all, a discourse theory of semantics leads to a criticism of traditional concepts like "communication", "text", "author" or "intention". These concepts have been discovered to be institutions which have a specific function in discourse. Discourse analysis, applied as a study of meaning, requires new methods: These institutions have to be made explicit as presuppositions of the production of meaning. Meaning must be seen as a function of language use, of discourse practices, of exertion of control or even power.

As an example of explication of discourse practices I will examine which function the idea of "author" has in the production of meaning. A text must be written by someone, we assume, and as a consequence the person who said or wrote something has to be responsible for the meaning of what he has written. In traditional terms, a text needs to have a subjective position that is filled by something we usually call an "author". But, from what has been said so far, it is quite clear that not the physically existing person can be resposible but only the subject that is being created as a "speaker" by the text. A text must be read to be understood at all, in case there is no implicit and totally independent meaning in the texts themselves, which I tried to prove to be false at the beginning of this paper. Thus, the constitution of a speaker of text depends not only on the text itself, but also of its being read, being read and re-read again. "Any text... is constantly read and re-read in different ways - by different people, by the same people at different times in their lives, by different people at different periods in history. The meaning of a text is always produced in a process of reading." This productivity of the reader is usually ignored and replaced by a static meaning which is supposed to be "in" the text, by the "intention" of "the author". "Conventional criticism accepts the notion of the author as unquestionable and pre-given in order to be able to define how the text should be read". In his discourse theory of poetry Easthope criticizes this tradition:

In order to get to know more about the institution of "author" and the way it functions it is helpful to view into history. It is particularly interesting to look at the etymology of the word "author". Latin auctor means both "writer, somebody who wrote a text" (Schriftsteller, Berichterstatter) and "witness, somebody who is authorized, who is an authority" (Urheber, Zeuge, Vorbild, Wortführer). In English there is the derivation "author-ity" which denotes exactly what we assume a subjective discourse position to be. Foucault has thoroughly investigated this phenomenon:

The practices with which we treat certain discourses have to undergo the same criticism. The ideas of "narrator" and "author" are only a part of a larger literary practice. If it is claimed that something is, let us say, a poetical text, this text is placed in a specific literary, poetical discourse, to which belong other poems, secondary material like commentary and certain forms of language use like rhyme scheme or verse structure. "To say that a text has meaning as literature is to relate it specifically to a literary universe of discourse as distinct from others, and thus to interpret it in terms of literary norms and assumptions about the nature of meaning". The same goes for concepts like "novel", "play", "informative or topical texts", "philosophical text" and so on.

In order to make semantic analysis work in applied textual criticism I will sketch out three strategies of analysis which are results of the semantic theory I have supported in chapter 2.

As "meaningful" discourse always has a certain order it is useful for textual criticism to examine the way in which certain discoursive fields are organized in language. Many words that we consider as signs for "real objects" are only the material expression of concepts, their points of condensation, which found the respective discourse on that concept. Words like "love", "family", "beauty", "childhood" and so on do not seem to have a counterpart in the "real world". They are concepts in certain discourses.

Often these concepts change, for example during the last fifteen years the German idea of Flüchtling (refugee) has slowly been replaced by the term Asylant, which has certain negative connotations due to the etymology and the history of the word. Change of terminology and discourse order can be considered an expression of concept-rebuilding. And, the other way round, discourse order builds up concepts in human understanding.

Thus, in a given text, semantics should thoroughly observe the words that are used, their adequance, and the concepts they involve. Besides, it is revealing to bring into discussion the words or concpets which could be used but are actually avoided. Analysis of concepts and of concept change in history is an important strategy in semantic discourse analysis.

In this discourse fragment, which could take place on every street, there are some strategies in the use of concepts and in the establishing of a scope of meaning.

The native person's answer begins with a strategy of time-gaining by repeating a part of the question. At the same time the repetition stresses the conceptual opposition of "foreigners" and "people 'who live here'". The next statement is frank and, presumably, honest: "they don't work", which is not more than a supposition by the speaker. Either unconsciously or even consciously the speaker knows that such direct reproaches or prejudices against foreigners usually put people who use them in a light of negative human attitudes. A person using them is supposed to be jealous, hostile and prejudicial. In order to maintain his attitude ("they don't work" has been uttered and will not be taken back) and, at the same time, prevent himself from being ascribed these negative features, he adds a qualification after having filled a gap by repetition: "well, don't work, (here we can imagine a short break, ed.) they just mess around with cars and sell them". This qualification expresses more or less the same as his first plain statement, but avoids the plain racism, since he admits that foreigners do something (i. e. "mess around with cars") but they do not work properly (i. e. according to the concept presupposed by the noun "work").

In this kind of analysis classical semantics as well as cognitive analysis (cf. 1.3.2.) are involved. Discourse strategies are neither a part merely of linguistic competence (there are no clear rules for their generation) nor of linguistic performance (the concepts which make them possible are part of the linguistic system and not fully at the speaker's disposition). Strategies of meaning-building and of either extending or narrowing the scope of possible interpretation are part of the general meaning-construction in discourse, and should therefore be examined by semantic analysis.

Another aspect of discourse which has scarcely been regarded yet by semantics is the sign system decoding and the medium transmitting the actual message. In opposition to the thesis that code systems are neutral and capable of transmitting any conceivable message, I have shown that the chosen sign system limits the space in which meaning can be created. In the case of language we have to analyze the linguistic sign system on different levels (letters of the alphapetic systemm, lexicology, phonology, texts as signs...). "Discourse is linguistically determined, in that it follows the laws of its own material nature, its materiality". Moreover, the system which is transmitting the encoded message again limits the scope of meaning-construction: Oral language can only reach a limited number of recipients. Its repetition is possible only as far as the capacity of memory reaches, and the memory will fade in time. In contrast, printed books or even digital data as electric impulses can be multiplied to an unlimited extent. This will influence the scope of possible interpretation (10 million people will certainly produce more varied interpretations than a dozen people ever could), and that, again, will influence the speaker's/writer's semantic strategies (since he/she usually knows how many recipients there are, and this allows anticipation of the possible interpretation). Semantic analysis has to take notice of the material conditions of discourse such as code system, system of transmission, multiplication and recording and means of repetion.

"The means of represantation is not a neutral vehicle that could equally be used to convey some other ideological signified but is already "shaped" for ideology and therefore itself ideological".

In the first chapter I asked the question whether it is really communication that constitutes the main essence of language. With a hint to misunderstandings, multiple interpretations and the problems arising from "interpretation" in the sense of hermeneutics I claimed the thesis that there are at least several meanings in any piece of language, that language has "meaning" on far more levels than just the communicative one.

I proposed to replace the idea of language, divided in code system (langue) and realization of that code system (parole), by the concept of discourse. With this substitution, we leave behind a lot of presuppositions, prejudices, as it were: The central events in language in use are not necessarily meaning, interpretation and understanding (but establishing, ordering, changing and using discourse); we need not assume that there are pre-given subjects who speak and understand, or pre-given objects to which language refers (but we can examine how subjective positions and concepts of objects, to which language refers, are created by discourse) and it is possible to reject the concept of a clear and unified meaning (and offer a variety of "interpretations", of "understandings" and "implications" instead).

Now, what does a discourse theory of semantics consist of? Its foundation is the rejection of classical communicative theories and the establishing of the more comprehensive idea of discourse. The method involved is primarily discourse analysis, with its similar techniques, like deconstruction. Discourse analysis offers a variety of tools which can be used in a semantic analysis of discourse. One of these is the examination of institutions enabling meaning: a "text" should be regarded as a "discourse fragment", the concept of "author" ought to be replaced by the subjective position created by the discourse itself or by the social institution which is legally resposible for a discourse fragment, and a "reader" could be a "meaning-establisher", as somebody who fixes a meaning.

As a result, meaning ought to be regarded as the establishment of a scope, in which interpretation usually roams without aim or plan since there is no fixed and everlasting meaning within this scope.

Bibliography

Easthope, Anthony: Poetry as Discourse (London, 1983).

Foucault, Michel: Die Ordnung des Diskurses (Frankfurt, 1991).

Foucault, Michel: "Was ist ein Autor", in: ------, Schriften zur Literatur (Frankfurt, 1988), p. 7-31.

Halliday, M. A. K.: Language as Social Semiotic (London, 1978).

Leech, Geoffrey: Semantics. The Study of Meaning (Harmondsworth, 1990 [11974]).

Swan, Michael: Practical English Usage (Oxford, 1991 [11980]).

Van Dijk, Teun A.: "Introduction: Levels and Dimensions of Discourse Analysis", in: ------(ed.): Handbook of Discourse Analysis (London, 1985), Vol. 2, p. 1-11.

Van Dijk, Teun A.: "Semantic Discourse Analysis", in:

------ (ed.): Handbook of Discourse Analysis (London, 1985), Vol. 2, p. 103-136.