Jan Ifversen

Text, Discourse, Concept: Approaches to Textual Analysis

Foucauldian discourse analysis and Begriffsgechichte can be fruitfully combined to develop a textual analysis,
which takes into account both pragmatic and semantic dimensions of language By Jan Ifversen As a historian by training I was not taught to work with texts and textual analysis1, but with documents and source criticism. Although historians do place documents within communicative contexts, they are primarily treated as monuments of past voices. Source criticism is not really concerned with establishing how texts produced meaning, but rather with answering questions about who said what and why. In later years, I have been interested in combining the advantages of source criticism with textual analysis.2 Source criticism works from the premise that the historians material demands a sort of authenticity: the texts must bear witness to past events; they must contain a claim of testimony. Paul Ricur has a remarkable formulation of the relation between testimony and representation:
I was here. This claim is that of transmitted memory. Its a word said publicly to somebody else who receives the testimony and in some cases writes it down.3

It is the task of source criticism to establish this claim. Textual analysis, on the other hand is concerned with the linguistic forms of past representations. It must get to grips with the representational chain that links memory to testimony and testimony to writing. In this paper, I shall present some approaches to a textual analysis of historical documents. I will touch upon aspects within textual analysis that particularly concern the work with historical material. But I will not here take up the highly interesting question of the relation between documentary analysis i.e. the confirmation of authenticity and textual analysis. In line with the conference theme, Discourse Theory and Practice, my aim is limited to presenting approaches to textual analysis inspired by discourse analysis. Alongside this presentation, I shall also highlight some of the problems involved in working within a textual horizon.

Constructivism tout court Although not a dominant trend within the discipline of history, a discussion of textuality has been going on for some years now. Louis Montrose, one of the leading figures within what has been called new historicism,4 launched the slogan that we need to study the textuality of history and the historicity of texts.5 Obviously, this slogan leaves a lot of questions unanswered: What is meant by text? How does history manifest itself in texts? What is the relation between the two parts of the chiasm? More than a specific theory or paradigm, the slogan pointed out a certain room for discussion, which raised new questions to historians about the particular role of texts. To recognize the role of textuality means accepting a basic constructivist premise. Such a premise might be construed in broad terms. Here is a formulation by Heidegger: Only that which has been conceived can be seen; but that which has been conceived is that which has been invented.6 Admittedly, in this quotation Heidegger speaks of concepts, not language as such. But my main point is just to stress that, for Heidegger, the relation between sensation and conceptualization is based on invention. From a more Kantian point of view, Ernst Cassirer, Heideggers opponent in the heated philosophical debates in Weimar Germany, has given an even more radical touch to constructivism: Whatever has been fixed by a name, henceforth is not only real, but is Reality.7 Before we take these statements to be expressions of radical ontological claims of everything being formed by language, we have to remember that experience (for Heidegger) and consciousness (for Cassirer) played an important role. Obviously, the constructivist premise contains ontological claims about the relation between language (or representation or conceptualization) and reality. But I do not intend to engage in an ontological discussion. It is enough for me to emphasize the inevitable role of
The term new historicism was coined by Stephen Greenblatt, professor of English, as the designation of new scholarly interest in combining formal textual analysis of poststructuralist descent with a study of the historicity of texts. For a discussion of new historicism, see H. Aram Veeser, The New Historicism, London, Routledge 1989. 5 Louis Montrose, Professing the Renaissance: The Poetics and Politics of Culture, in Veeser, op.cit. p. 20 6 Martin Heidegger, Aus der Erfahrung des Denkens (1954), quoted from Edmundo OGorman, The Invention of America, Greenwood Press, Westport, Conn. 1972, p. 73. 7 Ernst Cassirer, Language and Myth, New York, Dover Books 1946, p.58. 60
4

The following article is a slightly modified version of a paper given at the research seminar Discourse Theory and Practice organized by SPIRIT, Aalborg University, at Gl. Vr Slot, 26-28 September 2002. 2 I have discussed the relation between textual analysis and source criticism at greater length in Jan Ifversen, Tekster er kilder og kilder er tekster: Kildekritik og historisk tekstanalyse, in Den Jyske Historiker 88, 2000, pp.149-174. 3 Paul Ricur, Humanities between Science and Art. Centre for Cultural Research, University of Aarhus 1999, p.7.

KONTUR nr. 7 2003

Jan Ifversen

language in the formation of human reality. My own entrance into constructivism was via Berger & Luckmanns now classical work The Social Construction of Reality from 1966. I did not have serious difficulties in accepting statements like the following:
The common objectivations of everyday life are maintained primarily by linguistic signification. Everyday life is, above all, life with and by means of the language I share with my fellowmen.8

6) Which dimensions of language are we interested in?

But the aspect of their work I found most interesting was their focus on the social character of language. Language or, more broadly speaking, representation was thus seen as an inherent part of the social, not as something separate from or external to society. Here there was no need for referring to a consciousness operating behind language. Language was an element of the admittedly unconscious social operations that made society appear as an external reality. I could therefore subscribe to another of their slogans: Society is a human product. Society is an objective reality. Man is a social product.9 Questions However, this broad constructivist premise needs refinement. The claim that objectivations are products of linguistic signification does not say much about the way language operates. That is why we must combine the constructivist premise with a theory of language. Discourse theory is one such theory. There seems, however, to be a tendency among people discussing discourse theory to prioritize the grand constructivist claims and to rehearse huge ontological debates, confronting realism and nominalism on the question of the relation between language and reality. It is not that I find these debates uninteresting we need to come to terms with our constructivism and nominalism but I also think it is necessary to take up challenges of a more methodological nature and discuss how we approach texts from a linguistic point of view. The following questions come up when I venture into textual analysis:
1) How does our constructivism relate to language? What kind of language theory do we subscribe to? 2) What do we mean by a text? How do we account for the text as a meaningful unity? 3) At what linguistic level do we want to work? Textual or supra-textual? 4) How do we describe supra-textual units? (discourses, semantic fields, ideologies) 5) How do we understand the relation between text and context? How do language practices affect social practices? How does the context determine what can be said?

What do we mean by text? These are some of the questions I will try to answer in the following. I shall leave the first question unanswered for the moment and return to it when I introduce the different linguistic approaches I work with. The second question I shall only touch briefly upon. At a semantic level, a text can be said to constitute a certain unity of meaning, which contains sequences of sentences (other unities are morphemes, lexemes, syntagma and sentences). The linguist M.A.K. Halliday has defined text as a semantic unit containing specific textual components, which makes it internally cohesive and functioning as a whole as the relevant environment for the operation of the theme and information system.10 In Hallidays functional approach to semantics, the textual component determines the channels and modes through which a message (or a theme) is transmitted. To put it somewhat differently, you could say that the textual component (of the text) is what tells us about the kind of text we encounter. I do not intend to say much about the description of the internal meaning of a text. The material form of the text is important for the rendering of cohesion. The material form of the book, the letter or the newspaper article, for instance, tells us that we have a textual unity. When it comes to the semiotic form, we will have to demonstrate how a text constitutes an autonomous, meaningful order. Different descriptive terms can be used, depending on the type of analysis to be conducted. Within a semantic analysis, the description of the internal relations between the different segments of meaning (lexemes, words), for instance the syntagmatic and paradigmatic relation, are important. Roughly speaking, syntagmatic relations are relating linguistic entities as well syntactically as semantically (horizontal relation), whereas paradigmatic relations designate those entities that are only related semantically (vertical relation). The latter relation concerns the existence of synonyms, antonyms, homonyms (the same sound/signifier, but different meanings) etc. Within a narrative analysis, on the other hand, the focus is directed at the relation between the different roles (actants) in making something happen in the text (the plot). Of course, narrative analysis does not exclude a semantic approach. It could thus be of interest to examine the different terms attached to the various actants. Another type of analysis, which has become quite influential recently, is rhetorical or stylistic analysis where the focus is on the figures (tropes) and argumentative forms in the text aimed at producing an effective and beautiful language.

8 Peter L. Berger and Thomas Luckmann, The Social Construction of Reality, Garden City, New York, Anchor Books, 1966, p.51. 9 ibid. p.79.

10 M.A.K. Halliday , Language as social semiotic : the social interpretation of language and meaning, London, Edward Arnold 1978, p.136.

KONTUR nr. 7 2003

61

Jan Ifversen

At what linguistic level do we want to work? I am interested in describing the production of meaning in single texts. But, I am not analysing single texts. I am interested in the way specific objects, themes and concepts are produced and reproduced in a field made up of several texts. I therefore need to find out first how we can define a level above the single text, i.e. a supratextual level; and secondly, how we can analyse the relation between individual texts. Concerning the first, we must choose a term for the level above the text whether genre, discourse, semantic field or ideology and furthermore, we need a way of describing the meaningful unities at this level. We will also have, however, to decide how these units are expressed in the single texts. To take an example: how does a text convey the sense that it belongs to the genre of detective stories? What semantic and discursive features determine its membership of this genre? The introduction of a higher-level thus involves a dual perspective: from the text to the higher level, for instance the discourse, and from the discourse to the text. Fig. 1: A dual perspective DISCOURSE

role of the supra-textual level in an moment. But first I want to consider another non-textual level that we often refer to as context. How do we understand the relation between text and context? I have mentioned that part of the supra-textual level might be considered as context. This part is often termed co-text, the term context being reserved to the extra-linguistic situation. The question of the context can be approached in several ways. We need to consider how to describe the relation between text and context from a textual point of view as well as from a contextual point of view. We must equally decide about the range of the context. Are we focusing on the immediate communicative situation? Or are we interested in larger institutional frameworks or even societal structures?12 It is well known that classical, structural linguistics were only interested in language as an autonomous system of meaning. Language in use Saussures parole was referred to the domain of sociology and psychology. A basic premise of structural linguistics is the sharp separation between what a linguistic unity means and what it stands for. In the words of the linguist, Stephen Ullmann:
(T)here is no direct relation between words and the things they stand for: the word symbolizes a thought which in turn refers to the feature or event we are talking about.13

Text Text .

Text

But no direct relation does not mean no relation at all. The problem involved here is about the referentiality of language. Maybe the most fundamental aspect in the study of language is the recognition of its tripartite structure, as can be illustrated by Ogden & Richards famous triangle.14 Fig. 2: Ogden & Richards triangle

This dual perspective presupposes that the textual and the supra-textual levels can be delimited from each other. There has been a tendency within French semiotics to expand the notion of text to a higher level in order to point to the unfinished and unstable character of language as such.11 This expanded notion of text has served to focus on the work of language in textual analysis, but in my view the deconstruction of textual limits is difficult to reconcile with a notion of source that is bound to a certain material form. The interrelation of texts is often rendered as intertextuality, by which we understand the ways that a specific text manifests traces of other texts. But intertextuality is also part of a supra-textual level. Genres or discourses thus include particular texts and ways of citing them. The bible is an important element of a Christian discourse. To quote the bible would be an important feature of this discourse. I shall return to the

MEANING

signification

denotation

WORD

REFERENCE

11 Se for instance Roland Barthes, Thorie du texte, in Encyclopedia Universalis France, 1989, Tome 15, pp.996-1000.

12 Norman Fairclough has developed a version of discourse analysis that intends to cover as well the immediate communicative situation as the large social structures determining all communication, see Norman Fairclough, Language and Power, Longman London 1989. 13 Stephen Ullmann, Semantics : An Introduction to the Science of Meaning, Oxford, Blackwell, 1962, p.56). 14 The Meaning of Meaning: a Study of the Influence of Language upon Thought and of the Science of Symbolism, London. 1923

KONTUR nr. 7 2003

62

Jan Ifversen

Signification indicates the internal relation between the form (the signifier) and the mental image or thought linked to a specific form (Saussures signified), whereas denotation (or the referential function) is the relation between word (or sign) and an object in the world, i.e. the extra-linguistic dimension (which can be an imaginary object as in the case of a dragon). The introduction of this triangle might seem trivial. I only want to emphasize that the extra-linguistic is always present in language and that we therefore through language have access to reality. However, the question of referentiality can be taken a step further. It does not only concern the way objects are denoted in language, but also the ways in which the individual speaker or author manages denotation. Within this pragmatic view of language, the focus is on the linguistic traces of the communicative context. Spoken language is, of course, full of such indexical or deictic traces. When I address the addressee of my talk as you, I dont need to specify who I mean. The immediate context in which my utterance is located determines whom I mean with you. But written texts also contain traces that the historian will use to reconstruct the communicative situation. Still, referentiality and traces of communication are notions that belong to an approach which is mainly interested in looking at the outside from the point of view of the language. I shall not try to reverse the point of view completely that would invalidate my basic premise but rather to introduce a more nuanced view of the outside. Referentiality is an operation that takes place inside language. If we, instead of referentiality, think of effect we move towards the context. By effect I have in mind the social consequences that an utterance might have. Effect can be studied linguistically as in the case of the so-called speech acts. In speech act theory, language or more precisely the perlocutionary aspect - is directly viewed as action, and the limits between linguistic action and other forms of action is conflated. In rhetorical analysis, the focus is on the deliberate production of effects on the audience. A further way of paving the way for a more contextual view of the text is to let its meaning be dependent on an act of interpretation. Fairclough has proposed to analyse the meaning of a text as a result of an interpretation of textual cues what others have called invited inferences - in specific situational and institutional contexts.15 The context is here the action of the recipients (and their context). Perhaps we can even take the contextual view a bit further by saying that texts respond to challenges from particular events or situations. Response is not to be understood as mirroring external practices, but - to quote Stephen Greenblatt - as that which renders something sufficiently notable to be represented.16 We might also define challenge as a sort of problem that raises questions in need of answer. Greenblatt relates

texts to a historical context. He sees a given text as a result of an interplay between three different elements: intention, genre (his term for the supra-textual level) and historical situation. The relation between these three elements is described in the following way:
Neither intention nor genre can be reduced to [the] historical situation: a given genre () may have great difficulty accommodating a particular representational object, and artistic intention has an arsenal of strategies () designed to differentiate it from the surrounding world.17

The historical situation is that past context from where all sorts of incidents produced by a myriad of interactions arise. Incidents only become represented objects we could perhaps speak of events - by being represented in texts. Ricoeur would probably say that only some incidents would be testified. Not all incidents will be represented everywhere and in the same way. There is an option of genre, and perhaps even of artistic strategy. We will later have to come to terms with intention. For the moment it is enough to point to the triangular relation between text, supra-text and context. Fig. 3: The relation between text, supratext and context TEXT

SUPRA-TEXT (discourse, genre)

CONTEXT (situation, institution)

The point is only that the three elements are to be viewed as independent factors, and also that the text makes a choice of discourse and of context. If we were not to recognize this choice we would end in a pure contextualism; that is, a claim of a causal link going from the context to the text. Texts would thus be reduced to simple products of the context. If we, on the other hand, eliminated any contextual view we would end in a textual absolutism where reference would be arbitrary and communication something external to the text. Which dimensions of language are we interested in? Let me now say something about various textual approaches to the triangular relation between text, supra-text and context. Such approaches might differ
17

Fairclough, op.cit. pp.141-146. Stephen Greenblatt , Learning to Curse. Essays in Early Modern Culture. London, Routledge, 1990, p.112.
16

15

ibid. p.112. 63

KONTUR nr. 7 2003

Jan Ifversen

quite extensively. They might differ in how radically they formulate their constructivist claim. Conceptual history or Begriffsgeschichte that I shall return to in a moment recognizes that (w)ithout common concepts there is no society, and above all, no political field of action, but at the same time it claims that our concepts are founded in politico-social systems that are far more complex than would be indicated by treating them simply as linguistic communities organized around specific key concepts.18 According to Reinhard Koselleck, the analysis of these systems constitutes a separate task fulfilled by what he calls social history (Sozialgeschichte). Where Koselleck is eager to mark out the limits of language, others like Ernesto Laclau and Chantal Mouffe well known proponents of discourse analysis advocate a far more radical version of constructivism:
What is denied is not that such objects [earthquakes or falling bricks] exist externally to thought, but the rather different assertion that they could constitute themselves as objects outside of any discursive condition of emergence.19

Textual approaches might also vary as to their general understanding of language. Koselleck deals with language from a classical semantic point of view where concept and word designate the process of signification. But when he introduces the term basic concept (Grundbegriff) and a host of other concept categories (battle concepts, concepts of expectation etc.) he moves away from traditional semantics (and is often criticized for confusing the meaning of concept). Discourse analysis, on the other hand, works with the broader notion of discourse. Although there is a whole industry involved in defining this notion, what one can probably say without stirring up discussions is that discourse designates a totality of statements formed by a given configuration. Within traditional linguistics, discourse only meant language in use as opposed to the language system. In discourse analysis, these two different levels have been abolished. Discourses are viewed as having a systemic character (they have an internal order) with their own practice (Michel Foucault, for instance, talks about the discursive practice). Discourse analysis is therefore not only confined to semantic analysis, it also includes elements related to language use. Naturally, concepts and discourses are not the only entries to text and language. One might instead choose to focus on narrative structures, argumentative structures or actional structures (speech acts). The latter approach has been

rather influential in recent years, mainly through the works of Quentin Skinner.20 Textual approaches might also differ with regard to their scope. Some approaches limit themselves to a micro level, which makes possible a close textual analysis of few texts including a narrow definition of context. Other approaches like Begriffsgeschichte and discourse analysis work on a macro-level where concepts or discourses are analysed in many different texts and related to huge temporal (e.g. the modern period) or spatial contexts (e.g. Western Europe, Germany). This means that the analysis is performed at the supra-textual level where a textual corpus consisting of many texts is used to demonstrate different patterns and configurations. At this level, there is no interest in analysing the choices made by a particular text within a textual universe. Finally, some approaches are more spatially oriented, that is, more focused on a synchronic level; others are more oriented towards the problems of temporalization and change. The first is typical of (some versions of) discourse analysis with its focus on fractions, dislocations and collisions of discursive formations, but also, its very critical stance towards any philosophy of history points towards synchronic analysis. Begriffsgeschichte, on the other hand, is mainly interested in the temporal stratification of meaning, and the temporal tensions (Koselleck) involved in the use of concepts. Discourse analysis I shall limit myself to a discussion of two textual approaches, Begriffsgeschichte and discourse analysis, and try to show how, in my view, they can supplement each other. Different varieties of discourse analysis have been highly popular in recent years. Some versions like the one introduced by Chantal & Mouffe - are very general in nature and seem to function mostly as weapons in an epistemological debate. Others I see as more analytically refined. In the following, I shall introduce a version inspired by Michel Foucault.21 As I have already mentioned, the term discourse designates specific patterns or rules of distributrion (Foucault) within a given universe a corpus of statements. According to Foucault, these rules can be analysed by means of four different devices. The most important is to identify how a discourse turns something into an object that can be classified, explained, acted upon, institutionalized etc. The constructivist premise is clear: Objects do not exist as something statements refer to. Psychiatry, for instance, only became an object in the 19th century through the way that different prior and separate statements became related to each other. To analyse discourses means to be attentive to the way delimitations operate. Discourses are modes of
20 See the recent publication of Skinners methodological essays in Quentin Skinner, Visions of Politics, vol. 1: Regarding Methods, Cambridge, Cambridge U.P. 2002. 21 Foucault primarily presents his theory and method in Michel Foucault, Larchologie du savoir, Paris, Gallimard 1969.

Reinhard Koselleck, Begriffgeschichte and Social History, in R.Koselleck, Futures Past: On the Semantics of Historical Time, Cambridge, MIT Press, 1985 p.74. 19 Ernesto Laclau and Chantal Mouffe, Hegemony and Socialist Strategy: Towards A Radical Democratic Politics, New York, Verso, 1985, p.108.

18

KONTUR nr. 7 2003

64

Jan Ifversen

delimitation within a larger field of statements. The appearance of an object thus indicates a process of delimitation from other objects. A discourse on immigration a discourse that makes immigration and immigrants an object - must demonstrate that immigration is different from, lets say, tourism. But discourses do not only work within a negative mode of demonstrating what the object is not. They also have to specify what the object is about. Delimitation and specification can be observed in different textual dimensions. Foucault proposes three such dimensions:
1) Conceptual architecture 2) Semantic macro-areas 3) Positions.

treat these areas as completely separate, they tell us that national identity relates to time, to cultural space, to political community and to others. However, contrary to Wodak, I choose to maintain a division of labour between concepts and semantic macro-areas. In my view, semantic macro-areas can be analysed as the relation between a key concept and various side concepts (Nebenbegriffe). Ad 3) Perhaps one of the most interesting features in the analytical framework proposed by Foucault has to do with the organization of positions, and specially subject positions. This notion points to the distribution of roles and speaker positions within the discourse. The subject position designates the whole gamut of authorization and legitimation attached to the distribution of roles. In a medical discourse, the position of the doctor indicates a right to speak and to use a certain vocabulary, whereas the position of the patient excludes the speakers from certain parts of it. Foucault links the distribution of roles to what he calls the formation of enunciative modalities (modalits nonciatives). Enunciation is a term used within the so-called pragmatic approach to language to describe the communicative features inherent in language: Enunciation displays itself in those features which point to a situation of communication. It is first of all the position of enunciation (position dnonciation) which specify the agent and the action of communication. Foucault himself does not refer directly to the linguistic description of enunciation, but he points in this direction. The analysis of position can perhaps also gain from an incorporation of insights from Hallidays functional semantics. Halliday points to the interpersonal component of meaning as the component that produces the social interaction in a text. This meaning is played out in what he calls the tenor of discourse, which refers to who is taking part, to the nature of the participants, their statuses and roles, and to their mutual relationships.23 The interpersonal meaning is focused on the doing in language, not on the content. In my view, this pragmatic dimension of language can provide a more solid linguistic foundation to Foucaults positions. The analysis of positions will thus take into account the relation between the direct speaker positions (with their indexical markers) in the text and the positions formed at the level of discourse.

Ad 1) Discourses may be identified by a specific terminology, as is found in scientific discourses. But a discourse is also characterized by the way it combines various concepts, and how certain concepts are attributed a central role. In an analysis of the conceptual architecture of a discourse, it is also important to observe how series of concepts exist side by side, and how concepts imported from other discursive formations are retransformed. Foucaults very structuralist way of thinking reveals itself in the prominence he gives to the relational aspect the combinations, series and networks - that is formed through a discourse. Ad 2) The conceptual architecture of a discourse contributes to the formation of some basic themes or theories. In general, Foucault is not terribly precise in defining his analytical tools. This is certainly true for his notion of theme or theory. As examples of themes or theories he mentions the idea of an original first language for the 18th century discourse on language, or the theory of evolution for the discourse on nature, also in the 18th century. The aim of introducing the notion of theme/theory is to highlight that the same concepts, or the same objects, might operate in different thematic settings. They only obtain their particular function when related to a specific theme/theory. For Foucault, the theme/theory seems to perform the final closing of the discourse, or what he calls the strategic choice of the discourse. To be less ambitious (and perhaps more precise) than Foucault, I will simply choose to see themes or theories as semantic macro-areas or topics. In this I am inspired by Ruth Wodak, who has done an analysis of the discursive construction of Austrian identity, in which she treats the construction of the following macro-areas under the general heading of content: the narrative of a collective political history, common culture, a collective present and future, a national body and the difference between those inside and those outside.22 Although it might be difficult to
Rudolf de Cilla, Martin Reisigl and Ruth Wodak, The Discursive Construction of National Identities, in Discourse&Society vol. 10, 2, pp.149-73. Within critical discourse analysis, content is linked to the way that textual representations of the world are coded in the vocabulary. Content covers all the semantic aspects of world KONTUR nr. 7 2003
22

representation. The relation between content and the experiential component of language has been worked out by M.A.K. Halliday. 23 M.A.K Halliday & Ruqaiya HasanRusan, Language, Context and Text: Aspects of Language in a Social-Semiotic Perspective, Oxford, Oxford U.P. 1989. 65

Jan Ifversen

Fig. 4: Positions in text and discourse

OBJECTS

Fig. 5: Levels of discourse

POSITIONS

DISCOURSE

TEXT speaker positions

STATEMENTS IN TEXTS

SUBJECT POSITIONS
DISCOURSE

To be true to Foucaults analytical practice the four elements the object, the conceptual architecture, the themes/theories and the distribution of positions must all be placed within a functional perspective, where the question is to find out what function a discourse performs within different practices.24 Discourses have effects on different practices. An economic discourse that is, a discourse in which economy is established as an object of study certainly has effects on political decisions made by governments or by other agents. But discourses also become practical when they are being institutionalized. Certain instances or institutions secure delimitations of an object. A medical discourse has its institutions such as the hospital with its medical corps. Positions become institutionalized. Scientific institutions control concepts and theories. It is sometimes difficult to distinguish between discourses and institutions within a Foucauldian discourse analysis. Institutions operate through discourses when they act. Perhaps the only difference is that the notion of institution includes a larger spectrum of practices and effects. The levels of discourse Foucaults analytical model needs refinement. A lot of efforts have been directed towards refining Foucault either sociologically (by focusing on institutions) or linguistically (by making a more direct link between discourse and text). Before trying to take a step in the latter direction, I will have to say something about the different levels of analysis presented by Foucault. His analysis works at two levels simultaneously. The first level goes from the materiality of the discourse in the form of the different existing statements the basic unity for Foucault - to the discourse. Let us call this the intra-discursive level or textual level. The second level which we might call the inter-discursive level or in the words of Foucault the discursive economy relates the discourse in question to other discourses and practices in order to examine the processes of delimitation. I have tried to illustrate the play of levels in figure 5:

DISCOURSES AND PRACTICES

CONCEPTS

THEMES/THEORIES

The analysis of a discourse is thus two-edged: it is based on the materiality of (a given set of) statements, and it has to be confronted with statements, which can be identified in other discourses. The focus of the analysis can either be on statements limited to certain texts where the aim is to show how they draw on one or more discourses. This implies, however, a prior analysis of discursive formations. The focus can also be on the strategies of delimitation within a larger landscape of discourses. Within this larger focus, the work of single texts can hardly be observed. Begriffsgeschichte As I said, Foucauldian discourse analysis needs to be refined methodologically, not least as regards its description of language. In the following, I shall concentrate on developing a more refined semantic analysis. To this purpose I turn to Begriffsgeschichte. The meeting points between discourse analysis and Begriffsgeschichte are concepts and conceptual architecture. Reinhard Koselleck, the founder of this approach, defines the concept in two different ways.25 On the one hand, it indicates the meaning or the signified in the process of signification. Concepts are thus linked to words (which is the only semantic entity in Kosellecks linguistic theory). On the other hand, according to Koselleck, the concepts acquire an additional layer of meaning from their use. This pragmatic dimension is important for the understanding of Begriffsgeschichte. The focus is primarily on the role that words and concepts perform in situations of contestation. Either the words themselves are contested (direct conflicts over semantics), or they play a crucial part in a conflict. But it is also this pragmatic perspective which, from a methodological point of view, is the least developed in
25 I have discussed Begriffsgeschichte at greater length in Jan Ifversen, Om den tyske begrebshistorie, in Politologiske Studier 6. rg., nr.1, Maj 2003, pp.18-34. References to Kosellecks work can be found here.

24

Foucault, op.cit. p.90.

KONTUR nr. 7 2003

66

Jan Ifversen

Begriffsgeschichte. Koselleck has difficulties in describing precisely how communicative contexts affect the meaning of words. I shall limit myself to looking at the semantic analysis of the relation between concepts and words. This relation can take several forms. Koselleck has only been interested in two types of relations, one where a concept appears in different words (an onomasiological relation). The concept of state might be signified by such words as sovereignty, power and territory. An analysis of this type of relation means to identify words that bear the same meaning or belong to the same type of meaning. From this it follows that concepts have a broader range of meaning than words. The latter can be given very precise definitions, such as those given in dictionaries. Concepts are therefore to be treated as clusters of meaning. The other relation between word and concept the so-called semasiological relation deals with homonymy or polysemy, that is, the multiple meaning of single words, for instance estate. Although polysemy is important for understanding the mechanisms of transfers of meaning, the primary relation of interest for Begriffsgeschichte is the onomasiological one. Concepts point to clusters of meaning that play a central, formative role in the language uses involved in the social changes towards modern society. These changes characterize the so-called Sattelzeit Kosellecks term for the period of radical change between roughly 1750 and 1850. Part of the conceptual analysis is to identify the basic concepts Grundbegriffe in the language of the political and social struggles of the Sattelzeit. The analysis of language use is thus important for identifying our basic concepts. Basic concepts do not get their meaning in isolation. Criticism has been raised against Begriffsgeschichte for focusing only on isolated words. Koselleck has, however, proposed to work with combinations of words within semantic fields. This work has been taken much further by some of his students, in particular by Rolf Reichardt and Hans-Jorgen Lsebrink, the editors of the never-ending Handbuch politisch-sozialer Grundbegriffe in Frankreich.26 There is a clear connection between the onomasiological relation of words, concepts and semantic fields. The latter points to the way that basic concepts, or key concepts, get their meaning from neighbouring concepts. The task of the analysis is to delimit these fields of related concepts. The figures below taken from my own research gives an idea of how to delimit a semantic field.

Fig: 6: Culture as a semantic field

customs nation

civility language CULTURE character

art

Fig. 7: State as a semantic field

territory

borders

constitution

STATE

government

citizenship

welfare

If we take the idea of semantic fields to be the analytical counterpart of the conceptual architecture proposed by Foucault, it is possible to imagine a combination of Begriffsgeschichte and discourse analysis. The analysis of semantic fields is thus a way of identifying the semantic level of a discourse. It will also be possible to show how discourses delimit themselves semantically by analysing the role of counter-concepts (the antonyms at the word level) in a given semantic field. Semantics, positions and strategies an example But it will also be possible to expand the analysis of concepts to include positions (and interpersonal meaning). Let me end by mentioning how I have analysed the meaning of positions and positioning in speeches held in the National Assembly of the French revolution of 1789.27 Here I shall concentrate on the way political speakers locate themselves within a democratico-revolutionary discourse. At a macro-level, we can observe the discursive formation and institutionalization of a democracy. A central element in this discourse is the representation of the people. The emergence of democracy, however, takes place in a situation that might be interpreted as revolutionary. A central feature in a discourse on revolution is the temporal or processual aspect: the aim of the revolution is always deferred more or less to a future. Another dimension of revolutionary discourse is the sharpening of the dividing lines between friend and enemy. Now,
27 Jan Ifversen, Om magt, demokrati og diskurs: Diskuteret I lyset af den franske revolution, vol. 2, Begrebshistoriske Studier, Aarhus Universitet 1997.

26

For an introduction to the approach used in the Handbuch, see Rolf Reichardt, Einleitung, in Rolf Reichardt und Eberhard Schmitt (hrgs.), Handbuch politisch-sozialer Grundbegriffe in Frankreich 1680-1820, Heft 1/2, Mnchen, Oldenbourg, pp.139148.

KONTUR nr. 7 2003

67

Jan Ifversen

how can we analyse the ways in which revolutionary politicians position themselves vis--vis others in a democratic institution such as the National Assembly? At the level of discourse, we have certain fixed positions: there are the purely symbolic positions of the people and the enemy, there are the positions of the legitimate representative and his legitimate opponent. The purely symbolic positions can only appear as represented in the speech of the institutionalized agents. But, at the same time, for speech to be democratic it is necessary that it refers back to the people. The people is thus represented in the speech of the representatives. The enemy is given the same symbolic status in the revolutionary discourse. The interplay between the various positions can be analysed at different levels and in different fields of language (semantics, grammar). At the level of discourse, the role of the different positions and their semantics can be observed. How is, for instance, the relation between opponent and enemy established? What are the terms used to designate the symbolic position of the people (nation, virtuous people etc.), of the enemy (aristocracy, counter-revolutionaries, traitors etc.), of the various groupings (true representatives of the people, Jacobins, sectarians)? If we include grammar in our analysis, we can also observe how the different speaker positions are expressed indexically, for instance in the distribution of personal pronouns (I/us, you, he/they). The relations between the different levels of analysis and the linguistic fields might be summarized in the following figure: Fig. 8: Positions in discourse and text DISCOURSE
POSITIONS Symbolic positions PEOPLE ENEMY

observe how the different speaker positions are distributed within the available semantics of positions. We might even go as far as asking how the different speakers make use of the semantics and grammar to position themselves politically in the debate. But this would raise the question of agency, intention and rhetorical strategy, which is yet another story. Concluding remarks Historians must come to terms with texts. Most documents that historians make use of are texts. And texts are linguistic representations of reality. Historians, therefore, must engage in textual analysis. There are, however, many approaches to textual analysis. Some are more contextualist, others more formalist. Some approaches tend to work at a micro-level, where the role and meaning of the singular text becomes important. Other approaches are oriented towards a macro-level where many texts are studied in order to make wide-ranging claims about a certain period or a certain society. In the present article, I have raised some questions concerning texts and textual analysis that I view as important for a historian venturing into this field. These questions relate to classical areas within linguistic theory, such as the relation between text and context, the relation between various dimensions of language, the relation between a narrow and a broad perspective on texts. To provide some answers to these questions, I have been inspired by existing theories and analytical practices within what, in general, could be termed linguistic constructivism; that is, a form of constructivism based on a deep understanding of the role of language in representing reality. Although I treat a range of very diverse approaches, in providing answers I have primarily been inspired by Foucauldian discourse analyse and Begriffsgeschichte. I believe that the two approaches can be fruitfully combined to develop a textual analysis, which takes into consideration both a pragmatic dimension and a semantic dimension of language. Foucauldian discourse analysis is focused on the role of symbolic positions in the formation of discourses. The study of positions can be enriched by adding a more communicative focus, which includes the actual speaker positions, for instance, the speakers in a political speech. But the study of these positions can also gain from a more thorough semantic analysis as it has been developed in Begriffsgeschichte. The analysis of the role of, let us say, people, could thus combine a focus on its symbolic status in a democratic discourse with a view of the semantic field within which the meaning of people become stable at a given moment. Constructivism, deconstruction and like theories have recently become very popular among scholars within the social sciences and the humanities. There has, however, been a tendency to let the discussion of these theories be confined within epistemological battlefields where wide-ranging claims on the nature of language and reality are stirred up. Although I consider that it is important to be conscious of the theoretical claims beneath various analytical approaches to textual
68

Nation, truth aristocracy SEMANTICS ........................................................................................................................... Institutionalised positions REPRESENTATIVE OPPONENT

TEXT Symbolic positions PEOPLE ENEMY

........................................................................................................................... Institutionalised positions Speaker positions REPRESENTATIVE OPPONENT

I/us

you/you

he/they/them

GRAMMAR

The goal of the analysis is thus to demonstrate how different speaker positions relate to the different semantic options in the text (and how these options are formed by the discourse). It will also be of interest to
KONTUR nr. 7 2003

Jan Ifversen

analysis, I also think that it is time to move on to discuss the analytical practices in more detail. By trying to raise questions and provide some answers, I hope I have contributed to move the discussion in the direction of analysis. Jan Ifversen is assoc.professor at the Department of European Studies, University of Aarhus.

KONTUR nr. 7 2003

69