At the same time, there is very little inspiration from the many other approaches to context in linguistics and

especially in anthropology, sociology or social psychology (see below), at least in the analysis of the context. There are relations with sociolinguistics (especially Bernstein's), sociology and critical discourse analysis, among other directions, but these do not primarily focus on the improvement of the theory of context. In light of for instance what has been done in the ethnography of speaking and in the social psychology of episodes (see below) it is striking that SF for so long has been content with a rather heterogeneous set of very general and vague notions, without doing systematic research on the properties of the social situation of communicative events. Although they are occasionally mentioned as part of one of the three categories, one wonders why there is no systematic analysis of, for instance, such categories as social domain, setting, time, place or direction, of the many institutional constraints on discourse. The same is a fortiori true for lack of attention to the many types of 'mental' aspects of the social situation that are relevant for text or talk, such as purposes or aims, and especially knowledge, a notion that is very seldom used in SF analyses of context, at least not as a cognitive notion or as knowledge of individual speakers (for SFL, undoubtedly, knowledge is a social notion, in the sense that knowledge is somehow "in" or "of" society, and not in the minds of people). In other words, the SF concept of context is not only inadequate for the reasons mentioned above, but it is also basically incomplete: very important categories are missing. The same is true for reflections on the internal structures of contexts. And finally, despite the main claims of a functional theory of language, there is strictly speaking no explicit theoretical framework that describes and explains how social structures of contexts can affect language or discourse structures, and vice versa. The relationship is merely postulated, but there is no explicit interface that shows how language users actually are able to adapt their discourse to their social environment. Like in much of classical sociolinguistics, this relationship in SF is not an explanatory one, but a correlational, descriptive (probabilistic) one. (p. 341-343) FRAGMENTS OF A THEORY OF CONTEXT (T. van Dijk 2004:348-354) A fully-fledged theory of context is a complex, multidisciplinary theory of the structure of social situations and communicative events and how their relevant properties are related to the structures and strategies of text and talk (for early more formal, but rather reductive formulations of this theory, see e.g., Van Dijk, 1972, 1977). I shall only highlight some of the dimensions of such a theory, and in this paper largely ignore the earlier research done on context in linguistics, anthropology, and social psychology as referred to above. Despite this earlier work, we still lack a more or less explicit theory of context. Indeed, until today, there is not a single monograph on context. Contexts as mental models The main thesis of my theory of context is that contexts should not be defined in terms of some kind of social situation in which discourse takes place, but rather as a mental representation, or model, constructed by the speech participants of or about such a situation (for details, see Van Dijk 1999). Social situations as such, as well as their properties, cannot directly influence how people write, speak or understand talk or text. Gender, age, roles, group membership or power of participants, among many other traditional properties of the situation of communicative events, can be relevant for discourse only when participants attend to them, and construct them as such. This observation is consistent with an ethnomethodological and discursive psychological approach to context. However, my proposal suggests that such `constructs' are not just abstract or vaguely "in between" participants, but defined in terms of mental models, and only thus able to function as the necessary link between social situations and discourse. According to contemporary psychology, mental models are representations of actions or events in Episodic Memory, which is part of Long Term Memory. In Episodic Memory (sometimes also called Autobiographical Memory), represent and store their (interpreted) personal experiences, including the ways they interpret the events they read or hear about, witness or participate in themselves. Thus, MPs debating about a recent ethnic conflict do so

on the basis of their personal interpretation of such a conflict, as represented in their mental model of that conflict. Each MP will have his or her own mental model (interpretation) of this conflict (for details of the notion of mental model, see, e.g., Johnson-Laird 1983; Van Dijk & Kintsch 1993; Van Oostendorp & Goldman 1999). Models are not only personal, but also have an important social dimension. What MPs construct also depends on their general, cultural knowledge about conflicts and ethnic groups, as well as on their socially shared attitudes and ideologies about such conflicts or ethnic groups. That is, mental models of different people may sometimes be very much alike. However, despite these social dimensions, each model as a whole is subjective and unique (for the current communicative situation) because it necessarily also features the personal experiences, opinions, or autobiographical associations of MPs about such a conflict. The same is true for mental models that participants construe of a very special class of events, namely the communicative event in which they are now taking part. These are also personal, and unique for each partcipant if only because of their different autobiographical experiences as well as the different current perspective and interests and at the same time have a social dimension. We may conceive of context models as explaining the crucial `pragmatic' notion of relevance: They define what for the discourse participants is now relevant in the social situation (see also Sperber & Wilson 1986). Without a conception of the communicative event as represented by a context model, participants are unable to adequately contribute to ongoing discourse. They would be unable to produce and understand speech acts, would be unable to adapt topics, lexical items, style and rhetoric to the current social event, and they would not even be able to tell what the recipients already know, so that they do not even know what content to express in the first place. Indeed, without context models, adequate, contextually sensitive discourse is impossible. In other words, contexts are not 'out there, but in here': They are mental constructs of participants; they are individually variable interpretations of the ongoing social situation. Thus, they may be biased, feature personal opinions, and for these reasons also embody the opinions of the participants as members of groups. Indeed, a feminist and male chauvinist in conversation are likely to have rather different context models, as do a liberal and a conservative, a professor and a student, and a doctor and a patient talking together. Indeed, biased or incomplete context models are the source of profound communicative and interactional conflicts. In other words, just like mental models of events talked about, also context models may be ideologically biased. Thus, MPs not only may express biased beliefs about immigrants, but may also exhibit such beliefs in their interaction and discourse with immigrants or with MPs of other political parties. It should be emphasized that context models are not static mental representations, but dynamic structures. They are ongoingly constructed, updated and reconstructed. They change with each change in (the interpretation) of the situation, if only because of the ongoing changes of discourse itself (one of the components of context). For instance, if nothing else, the discourse will dynamically change the knowledge the participants have about the knowledge of the other. But also the ongoing action, the participant roles, aims and other beliefs may change during interaction. That is, in all ongoing interaction and hence also in ongoing conversation, as well as during reading, language participants maintain a dynamically changing model that allows them to flexibly plan, understand, memorize and adapt their discourse to other participants and other aspects of the ongoing event.

Examples of parliamentary context and discourse (353-354) In other words, also due to the socially shared nature of our knowledge about language, discourse and communication, and because of the routine nature of everyday context building (a special case of making models of our daily experiences), MPs need not invent or build their context models from scratch. Despite the variations of the social/political situation, as well as the personally different experiences of MPs, much of their personal models should consist of a more or less fixed schema that can be applied now and again in the interpretation of each session of Parliament. This activation of a known schema is strategic, and similar to the activation of knowledge about discourse genres. However, such a ready-made schema can quickly be

adapted to specific circumstances. In light of the general theoretical remarks made above about contexts defined as mental models of communicative situations and events, and some more informal remarks about parliamentary contexts, let us now try to become more systematic and deal in more detail with the hypothetical categories of the parliamentary context schema. To test such hypotheses directly, we would need cognitive methods to assess the structure of mental models, but since these models generally involve forms of discourse (induced in the laboratory), I shall assume that these context categories may also be exhibited in parliamentary talk. This is not merely a (p. 354) methodological decision, but a theoretical one: contexts are defined in terms of relevance, and hence we must assume that its categories are constructed in such a way that they monitor specific structures of discourse. This does not mean that context categories are always explicitly formulated and attended to (as is the case for goals or knowledge), but by definition they are needed to describe and explain at least some properties of discourse. One of the many assumptions of a theory of mental models is that such models are (hierarchically) organized: important categories on top, and secondary categories lower in the hierarchical schema. In our analysis, we shall first deal with the respective categories, and then make assumptions on their ordering and organization in the model schema. Micro and macro categories Another form of organization pertains to the level of categorization. As we have seen, some categories may be called macro categories because they are defined in more global societal structures, whereas the more traditional situational categories of face-to-face interaction belong to the micro level of analysis and understanding. Note that it would be a fallacy to assume, as is often the case in contemporary SFL, microsociological or ethnomethodological approaches, that the micro level of situated action is more 'concrete' or more 'observable' than macro-structural categories. In our theory, as well as in everyday experience and understanding, both levels are constructs, and hence represented in mental models.Microlevel categories. Local actions: questioning and answering in Parliamentary debates: Setting: place and time Current action: speaking in Parliament Participants: analyze this category in terms of different kind of roles: - Communicative roles (various producer and recipient roles) - Interaction roles (friend, enemy, opponent, etc.) - Social roles (e.g., based on gender, class, ethnicity, profession, organization, etc.). (p. 359) Goals: defend government, attack Labour, discredit refugees Knowledge: general: on immigrants, financial issues; political: on legislation, policies, etc. (p. 368)

INTRODUCTION At the beginning of the course, we agreed that we are not interested in the study of language in terms of grammar. We are only interested in language defined as discourse. The term of discourse may designate a certain hypostasis of language such as youth discourse or media discourse. In the case of this usage, the term discourse may be ambiguous, because it may designate both the system that produces a category of texts, as well as the category itself: communist discourse refers both to the totality of texts produced by communists, and to the political system that led to their production (Maingueneau 2007: 59-60). The notion of discourse is so much used because it reflects an essential change of the way we conceive language. In its turn, the change was determined by the various trends that have appeared within the realm of human studies; these trends are currently labeled pragmatics. Being more than a doctrine, pragmatics represents a particular way of perceiving verbal communication. According to Maingueneau (2007), discourse has the following main characteristics: Discourse supposes an organization/structure that surpasses the level of the sentence (phrase in French) This size does not mean that a series of words of a certain lengh will automatically make up discourse. It refers to the fact that discourse refers to another level of organization, different from the sentence. A prohibitive notice such as No smoking is not a clause (grammar minimal unit), but it is discourse. As a trans-phrastic (or better-said under-phrastic) unit, the notice is subject to organizational rules produced and observed by a well-determined social group. A dialogue, an argumentation are different types of texts, having different structures and lengths. In this respect, a news story will differ from a dissertation. Discourse is oriented Discourse is not only considered to be oriented because it is conceived in keeping with the expectations of an audience. It is also oriented towards an end: unfolds in time, following a direction in a linear way. The speaker/writer guides the reader/hearer through discourse markers: the letter may anticipate what will happen in discourse (we will see that..; we will come back to that), may amend the previous text (or rather; I should have mentioned that) or make digressions (besides; on the other hand). These discourse words refer to the Tehnici de redactare in limba englez 44 structure of the text. They may also concern the order of the ideas exposed in the text (first, second, then, consequently, etc.). Besides orienting the audience, discourse (better said metadiscourse) may include the authors comments on her/his own speech content: Paul finds himself, if we can say so, in utter filthiness. Or: Rosalie (what a name!) loves Alfred. (Maingueneaus examples 2007: 61). This linear development unfolds in different conditions and depends on the existence of only one text producer, who controls discourse from the beginning to the end (monologal discourse), or of the authors inclusion in interaction, where she/he could be interrupted or deviated by an interlocutor (dialogal discourse). In oral interaction, it often happens that we cannot catch some words, we ask for clarifications, in keeping with the others reactions.

Discourse is a form of action Speaking is a form of acting on the other. It is not only a way of representing the world. The issues of speech acts started to be approached in the 1960s by philosophers such as J. L. Austin (How to Do Things with Words, 1962) and, later, by J. R. Searle (Speech Acts, 1969). They showed that any utterance stands for an act (to promise, to suggest, to state, to ask, etc.) that is aimed at changing a state of fact. At a higher level, these elementary acts are integrated into discourse belonging to a particular genre (a brochure, televised news programmes, etc.) that is intended to produce a modification with the addressees. Discourse is interactive Verbal activity is actually an inter-activity which involves at least two partners, whose trace left in the utterance is the pronominal couple I YOU of the verbal exchange. The most obvious manifestation of interactivity is oral interaction, a conversation, in which the two locators coordinate their utterances, communicate according to the partners attitude and instantly perceive the effects that their words have on the other. Besides conversations, there are numerous forms of oral communication that do not seem interactive; it may be the case of someone who delivers a lecture or of a radio anchor. This is also true in the case of the written text where the addressee is not physically present. Some researchers would consider that it is only oral exchanges that represent the authentic usage of language, while the other usages would be degraded forms of speech. Maingueneau (2007: 45 Tehnici de redactare in limba englez 63) suggests that we should not take fundamental interactivity of discourse for oral interaction. Any utterance, even produced in the absence of the addressee pertains to the constitutive interactivity of language (dialogism); it is an explicit or implicit exchange with other virtual or real interlocutors; it always supposes the presence of another entity approached by the speaker/writer and in relation to which the letter constructs her/his discourse. From this perspective, a conversation is not considered discourse par excellence, but only one form of manifestation even if the most important one of discourse interactivity. If we agree that discourse is interactive and that it mobilizes two partners, it is irrelevant to call an interlocutor addressee; it would mean that communication is unidirectional and that it is the expression of the thinking of only one utterer who speaks to a passive addressee. Consequently, following Antoine Culioli, we could give up the term of addressee, in favor of the more suitable term co-utterer (coenonciateur in French). The two partners of discourse will be designated by the plural term coutterers (coenonciateurs in French). Discourse is contextualized We cannot say that discourse intervenes in a context, as if context were a mere framework, a setting; in fact discourse can only be contextualized. It was already emphasized in Unit 1, that an utterance cannot be assigned a (pragmatic) meaning without a context. The same literal utterance, in two distinct contexts, will compose two distinct discourses. Moreover, discourse contributes to the configuration of its context

that it may alter during its unfolding. For instance, two partners may talk like friends (symmetrical communication), and after a couple of minutes, they may engage in new relationship, as one of them starts speaking like a doctor, while the other answers questions like a patient (institutional discourse is characterized by asymmetrical communication). Discourse is taken in charge by a subject There is no discourse without a subject, an I that becomes the source of personal temporal and spatial reference and who signals her/his attitude towards what s/he says and towards the interlocutor (the phenomenon of modalization). The subject mainly indicates who is responsible for what is said. An elementary utterance such as Its raining is stated as a fact by the utterer who guarantees its truth. Tehnici de redactare in limba englez 46 But the subject could equally modulate the degree of adhesion to the truth value of the statement: It may be raining. The subject could make somebody else responsible for the utterance: Paul says its raining. S/he could comment on her/his words: Frankly speaking, its raining. The subject could even show that s/he pretends to be in charge of the utterance, as it is the case of ironic statements. (Examples supplied by Maingueneau 2007: 64) Discourse is regulated by norms Verbal activity is inscribed in the vast institution of speech. Like any human behaviour, it is guided by specific norms. Every speech act implies norms or premises to be met before it is performed. A seemingly simple act, such as a question is conditioned by the fact that the questioner should not know the answer to the question. The questioner should be interested in the answer. S/he should believe it is only her/his interlocutor who could supply it. Fundamentally, no speech act may be performed without some justification of the way it is presented. This justification pertains to the exercise of speech. Discourse is included in interdiscourse Discourse acquires meaning within a universe formed by other discourses in relation to which it is identified. In order to interpret the shortest utterance, it has to be related to all kinds of discourses that we comment, parody or cite. Every discourse genre has its own modality of treating the numberless interdiscursive relations: a philosophy text book will not cite in the same way and the same sources as a sales agent. The mere fact of including a discourse in a genre (lecture, news programme) signifies it is associated to an infinite set of discourses belonging to the same genre.

2.2. DISCOURSE (UTTERANCE) AND TEXT

UTTERANCE As it was explained in Unit 1, an utterance is the product of the act of uttering. It stands for the verbal trace of the process of uttering. The size of the utterance is irrelevant: it may be made up of a few words or a whole book. 47 Tehnici de redactare in limba englez Some linguists define the utterance as the elementary unit of verbal communication, a series of words endowed with meaning and syntactically independent. For instance, Leon is ill, Oh!, What a girl!, Paul! are utterances of various types.

Other linguists oppose the utterance to the sentence/clause (phrase in French) considering that a clause is taken out of context. There are numerous utterances that result from the same clause when it is interpreted in context. That is why, the example No smoking! is a sentence/clause in the absence of a context and is an utterance if it is written with red capital letters in the waiting room of a hospital. If the same clause is written in paint on the wall of a house, it will have a different interpretation. The term utterance is also used to designate a verbal sequence that stands for a complete communication unit and belongs to a particular discourse genre. An utterance is related to the communicative intention of its discourse genre. A TV news programme as an utterance/discourse is conceived with the goal of informing the audience on daily events, while an advertisement is conceived with the goal of persuading consumers to buy. TEXT When discourse forms a whole, a coherent entity, we can speak about texts. The linguistic branch which studies coherence is called textual linguistics. The term of text is currently assigned to verbal products, both oral and written that are structured so as to last, to be repeated and to circulate far from their original context. We should use the term both for literary texts, juridical texts and for a conversation.

2.3. DISCOURSE, TEXT AND TALK

We subject the following text to your attention. It makes a useful synthesis of definitions and explanations of these terms in specialized literature. According to Phillips et al. (2004) discourse [is the] analytic framework to better understand how institutions are produced and maintained. We argue that discourse analysis provides a coherent framework for the investigation of institutionalization. Accordingly, we develop a discursive model of institutionalization that highlights the Tehnici de redactare in limba englez 48 relationship between discourse and social action through the production and consumption of texts. We argue that the tendency among institutional theorists has been to define the concept of institution in terms of patterns of action, whereas we believe institutions are constituted through discourse and that it is not action per se that provides the basis for institutionalization but, rather, the texts that describe and communicate those actions. It is primarily through texts that information about actions is widely distributed and comes to influence the actions of others. Institutions, therefore, can be understood as products of the discursive activity that influences actions. Using discourse analysis, we are therefore able to develop a model of institutionalization that shows the conditions under which institutionalization processes are most likely to occur. In this article we make a number of contributions. First, we develop a model that identifies the microprocesses whereby individual actors affect the discursive realm through the production of texts, as well as the processes through which discourses provide the socially constituted, self-regulating mechanisms that enact institutions and shape individual behavior. Second, in using discourse analysis (e.g.,

Fairclough, 1992; Parker, 1992), we highlight an alternative understanding of social construction to that of Berger and Luckmann (1966) that is better able to explain the production of the types of institutions that feature in most institutional research. Third, our model provides a methodological contribution: it can be readily connected to the sophisticated techniques developed in discourse analysis for analyzing the (635) social dynamics of language and meaning techniques that make it possible to complement the study of institutional effects with empirical studies of how institutionalization processes actually occur. Finally, our paper illustrates the contributions that studies of language, especially the use of discourse analysis, can make to the study of organizing. Despite the implicit concern with language and texts in organizational research since the 1950s (e.g., Dalton, 1959), in linguistic approaches scholars have so far found it difficult to engage with contemporary mainstream management theorizing. Our paper not only shows the ways in which discourse analysis connects with institutional theory a well-accepted body of literature in organization theorybut also how institutional theory can benefit from a linguistic perspective. We present our arguments in three sections. First, we provide an overview of discourse analysis, highlighting several ideas that we believe are of significant value in understanding institutions. Next, we develop a discursive model of institutionalization. We integrate concepts from discourse analysis and institutional theory to construct a model of the relationships among action, texts, discourse, and institutions. Finally, 49 Tehnici de redactare in limba englez we discuss the implications of this model for the study of institutional fields and institutional entrepreneurship, as well as for the study of language in and around organizations. DISCOURSE ANALYSIS Like many other terms in social science, discourse and discourse analysis are used in a variety of ways in different bodies of literature (van Dijk, 1997a). In a general sense, discourse refers to practices of writing and talking (e.g., Woodilla, 1998). Such a broad definition, however, is not very useful for our purposes. Instead, we draw on Parkers definition of a discourse as a system of statements which constructs an object (1992: 5). Discourse rules in certain ways of talking about a topic, defining an ac-ceptable and intelligible way to talk, write or conduct oneself and also rules out, limits and restricts other ways of talking, of conducting ourselves in relation to the topic or constructing knowledge about it (Hall, 2001: 72). In other words, discourses do not just describe things; they do things (Potter & Wetherell, 1987: 6) through the way they make sense of the world for its inhabitants, giving it meanings that generate particular experiences and practices (Fairclough, 1992; van Dijk, 1997b). Discourses, put simply, are structured collections of meaningful texts (Parker, 1992). In using the term text, we refer not just to written transcriptions but to any kind of symbolic expression requiring a physical medium and permitting of permanent storage (Taylor & Van Every, 1993: 109). For a text to be generated, it must be inscribed spoken, written, or depicted in some waythus taking on material form

and becoming accessible to others (Taylor, Cooren, Giroux, & Robichaud, 1996: 7). Talk is therefore also a kind of text (Fairclough, 1995; van Dijk, 1997a), and, in fact, the texts that make up discourses may take a variety of forms, including written documents, verbal reports, artwork, spoken words, pictures, symbols, buildings, and other artifacts (e.g., Fairclough, 1995; Grant, Keenoy, & Oswick, 1998; Taylor et al., 1996; Wood & Kroger, 2000). Discourses cannot be studied directlythey can only be explored by examining the texts that constitute them (Fairclough, 1992; Parker, 1992). Accordingly, discourse analysis involves the systematic study of textsincluding their production, dissemination, and consumptionin order to explore the relationship between discourse and social reality. The centrality of the text provides a focal point for data collection, one that is relatively easy to access and is amenable to systematic analysis (Phillips & Hardy, 2002; van Dijk, 1997b). Discourse analysis does not, however, simply focus on individual or isolated texts, because social reality does not depend on individual texts but, rather, on bodies of texts. Discourse analysis therefore involves analysis of collections of texts, the ways they are made meaningful through their links to other texts, the ways in which they draw on different discourses, how and to whom they are disseminated, the methods of their production, and the Tehnici de redactare in limba englez 50 manner in which they are received and consumed (Fairclough, 1992; Phillips & Hardy, 2002; van Dijk, 1997a,b). Discourse analysis has proven a useful theoretical framework for understanding the social production of organizational and interorganizational phenomena (e.g., Alvesson & Karreman, 2000; Grant et al., 1998; Hardy & Phillips, 1999; Morgan & Sturdy, 2000; Mumby & Clair, 1997; Phillips & Hardy, 1997, 2002; Putnam & Fairhurst, 2001). Discourse analysts explore how the so-(636)cially produced ideas and objects that comprise organizations, institutions, and the social world in general are created and maintained through the relationships among discourse, text, and action. Accordingly, discourse analysis involves not just practices of data collection and analysis, but also a set of metatheoretical and theoretical assumptions and a body of research claims and studies (Wood & Kroger, 2000: x) that not only emphasizes the importance of linguistic processes but also underscores language as fundamental to the construction of social reality (Chia, 1996; Gergen, 1999; Phillips & Hardy, 2002). Discourse analysts have adopted a variety of approaches that range from micro analyses, such as linguistics, conversation analysis, and narrative analysis, through ethnographic and ethnomethodological approaches to the more macro study of discourse associated with Foucault (for different categorizations of approaches to discourse analysis, see Alvesson & Karreman, 2000; Jaworski & Coupland, 1999; Phillips & Hardy, 2002; Putnam & Fairhurst, 2001; Wetherell, 2001; Woodilla, 1998). The approach we develop here is a form of critical discourse analysis (e.g., Fairclough, 1992, 1995; Fairclough & Wodak, 1997; van Dijk, 1993, 1996). We draw on Foucaults work in arguing that the social world and the relations of power that characterize it are determined by the discursive formations that exist at a moment in time. Critical discourse analysts argue, however, that regardless of how complete they may appear, discourses, in fact, are always the subject

of some degree of struggle (Grant et al., 1998). They are, therefore, never completely cohesive and never able to determine social reality totally. Instead, a substantial space exists within which agents can act self-interestedly and work toward discursive change in ways that privilege their interests and goals (Mumby & Clair, 1997). Hence, there is always the possibility that actors can influence discourses through the production and dissemination of texts (Fairclough, 1992). In summary, we assume that there is a mutually constitutive relationship among discourse, text, and action: the meanings of discourses are shared and social, emanating out of actors actions in producing texts; at the same time, discourse gives meaning to these actions, thereby constituting the social world (Phillips & Hardy, 2002). These relationships provide the basis for a set of methods of data collection and analysis that can be used to explore the multifaceted processes 51 Tehnici de redactare in limba englez through which social entities, such as organizations and institutions, emerge (Phillips & Hardy, 2002; Wood & Kroger, 2000). (637)__