Context · Web viewThe notion of ‘situation model’ was used (e.g. in *Van Dijk & Kintsch, 1983), but as referring to events talked about, and hence rather as a referential-semantic

Teun A. van Dijk, Context. Towards a Multidisciplinary Theory.Volume 1, Language, Discourse and Cognition

Chapter 1

Introduction

On Tuesday, March 18, 2003, British Prime Minister Tony Blair held a speech in the House of Commons, defending a motion allowing British military action against Iraq “because of its continuing non-compliance with Security Council Resolutions”. He began his speech as follows:

At the outset, I say that it is right that the House debate this issue and pass judgment. That is the democracy that is our right, but that others struggle for in vain. Again, I say that I do not disrespect the views in opposition to mine. This is a tough choice indeed, but it is also a stark one: to stand British troops down now and turn back, or to hold firm to the course that we have set. I believe passionately that we must hold firm to that course. The question most often posed is not "Why does it matter?" but "Why does it matter so much?" Here we are, the Government, with their most serious test, their majority at risk, the first Cabinet resignation over an issue of policy, the main parties internally divided, people who agree on everything else?

[Hon. Members: "The main parties?"]

Ah, yes, of course. The Liberal Democrats - unified, as ever, in opportunism and error.

[Interruption.]

For the Members of Parliament present and for us readers and analysts to understand this fragment – as transcribed in the official Hansard record – it is obviously crucial to know English grammar and the rules of discourse. At the same time, such understanding requires large amounts of ‘knowledge about the world’, e.g., about democracy or British troops, and implicitly in this fragment, about Iraq. We thus understand, among many other things, that the speaker is defending sending troops to Iraq to bring democracy, and presupposing, again among many other things, that Iraq is not a democracy and that troops (war, etc.) can bring democracy.

van Dijk, Context, Vol. I, Chapter 1 1

Such understanding, based on grammar, discourse rules and world knowledge, is however only partial. Especially the Members of Parliament (MPs) also understand that and why such an intervention is appropriate in this debate and in parliament, what its functions are, and what speaker Tony Blair is now doing apart from speaking, meaning and referring to, e.g., British troops. That is, they not only understand the ‘text’ of Blair’s discourse, but also its context. They know that Tony Blair is now speaker but also speaking as Prime Minister and as leader of the current British Government; that he is now addressing them as MPs and party members; that he is intending to defend the current Iraq policy of his government; that when referring to ‘The House’ he deictically refers to ‘this’ House of Commons of which they are members and where he is now speaking; that he is mocking the Liberal Democrats for their alleged opportunism; and much more.

By understanding the combined text-in-context of this speech the MPs – and we as readers of the Hansard report -- understand what this speech really is about, namely a specific way of ‘doing politics’ by means of participating in parliamentary debates. Through our knowledge of the political context of this speech, we know that this speech is not only grammatical English and meaningful, but also appropriate in the current situation of a parliamentary debate and understandable as part of the political process of parliamentary decision making and legislation. In sum, we understand the political ‘point’ of this speech.

As analysts we know that the MPs understand Blair’s speech (more or less) in this way not only because we do so, given our knowledge of politics, parliamentary debates, the UK and current world history, but also because Blair as well as the MPs variously express, presuppose and signal such ‘contextual’ understandings, both in this as well as in later parts of this debate (see the analysis in Volume 2, Chapter 5). For instance, in this fragment Blair uses several deictic expressions that explicitly refer to how he understands the current context of his speech, by including the referents of ‘I’, ‘the House’, ‘this issue’, ‘our right’, ‘I say’, ‘the course we have set’, ‘here we are, the Government’, ‘the main parties’, that is, referring to the current situation and himself as speaker, his function as Prime Minister, parliament, British political parties, current policy, and so on. In their later interventions, the MPs also display such contextual understanding, in this fragment for instance by critically questioning Blair’s reference to the main parties while ‘forgetting’ the Liberal Democrats. That is, these MPs show that they have a different ongoing definition of the relevant communicative situation, and the ironical reaction of Tony Blair again shows that he understands this alternative construction of the context of the MPs, by making it explicit as an afterthought: the presence of the Liberal Democrats as a party in the House – and the


debate. In other words, their ‘pragmatic’ understanding of Blair’s speech involves ‘contextualizing’ it, that is, making inferences about his definition of the communicative situation — a definition with which they may not agree.

We see that producing and understanding text and talk crucially involves what we informally have called the ‘context’ of this speech, involving such categories as participant identities and roles, place, time, institution, political actions, and political knowledge, among other components.

More detailed analysis almost surely will require a more refined analysis of this fragment and its context, such as the fact that Blair’s ironical remark about the Liberal Democrats presupposes that they are part of the opposition and not of the government party or parties. This is not a semantic presupposition or implication, however, as when supporting troops presupposes that the UK has troops and that the UK is engaging in military action, but rather some kind of ‘pragmatic’ or ‘contextual’ presupposition based on political knowledge about the current political interaction in the debate. We also see that this fragment not only contains a question and a reply, but that the question may be heard as a challenge to Blair and as response to this challenge, possibly hearable as ‘doing irony’. Although also such an interactional analysis of this fragment may and should be refined, it does not provide sufficient insight into what is going on without further analysis of relevant context properties, such as the relation between Tony Blair as Prime Minister and members of the Labour Party and his opinion about and opposition against the Liberal Democrats. Without such a contextualized understanding we do not know that the interruption of the MPs is not merely a question, or even a critique, but also a form of political opposition if the speakers are members of the opposition. It is only through such a political understanding of the relevant context that Blair’s response can be heard as ironical, and hence as a relevant political attack against the Liberal Democrats. In other words, to understand this fragment as an interaction, i.e., to understand what Blair is actually doing, both the participant MPs as well as we as analysts need to construct an appropriate context for this fragment.

From this example and my brief analytical comments we may also conclude that ‘contextual’ analysis of discourse goes beyond grammatical, ‘textual’ and interactional analysis or understanding. Similarly, this analysis goes beyond the usual ‘cognitive’ analysis. Not only do we need to make explicit the knowledge of the world that sustains ‘semantic’ understanding of this fragment. We also need the more specific political knowledge required to construct a relevant context for this fragment and hence to understand its political meaning as an appropriate contribution to a parliamentary debate and the political process in the UK.


In other words, understanding discourse means understanding text/talk-in-context. Hence, discourse analysis and conversation analysis need to make explicit what contexts are and how exactly the relations between contexts and text or talk are to be analyzed in ways that explain how language users do so.

What is ‘context’?

Both in everyday conversation as well as in scholarly discourse, we frequently use general notions, such as ‘language’, ‘discourse’, ‘action’, ‘mind’, ‘knowledge’, ‘society’ or ‘power’, but we have a hard time defining them more or less satisfactorily. This often means that we are dealing with fundamental notions that need complex theories, if not whole disciplines, to account for their properties. At the same time, we usually have specialized fields of philosophy dealing with such concepts.

The same is true for the notion of ‘context’ as I have used it above, and which is also quite often used in many types of text and talk. Perhaps slightly more formal than related concepts, such as ‘situation’, ‘circumstances’, or ‘environment’, we use the notion of ‘context’ whenever we want to indicate that some phenomenon, event, action or discourse needs to be seen or studied in relationship to its environment, that is, its ‘surrounding’ conditions and consequences. That is, we not only describe but especially also explain the occurrence or properties of events in terms of aspects of their context.

Thus, when informally referring to the ‘context’ of Tony Blair’s speech, we may roughly summarize such a context with the description ‘the parliamentary debate in the UK House of Commons on March 18, 2003’. Especially much later, however, we might also define the context of Blair’s speech in broader terms, such as the ‘debates about the war in Iraq’ or ‘UK’s Foreign Policy’. That is, contexts come in different sizes or scopes, may be more or less micro, or more or less macro, and metaphorically speaking seem to be concentric circles of influence or effect of some state of affairs, event or discourse.

Also, there seems to be a mutual relationship of conditional influence between events and their contexts. The broader context of Blair’s (or more generally British) foreign policy – such as relationships with the USA – no doubt explains many aspects of the current parliamentary debate as well as Tony Blair’s speech. And conversely, the current debate and speech in turn contribute to this very foreign policy of the UK. Text and talk not only are constituents of (or even produced by) their contexts, but also appear to be


constitutive of their contexts: When addressing parliament about military action in Iraq, Tony Blair is also setting or defining UK foreign policy.

In the media, in the humanities and in the social sciences, the notion of ‘context’ is frequently used in order to place or explain things. One puts or sees things ‘in perspective’ or in their ‘proper context’, and we are often urged not to take or describe things ‘out of their context’. News report schemata in the press typically have a special Context category that places current events in their political, social or historical context (*Van Dijk, 1988a).

We may conclude from this informal characterization of the notion of ‘context’ that we do not properly understand complex things without understanding their context. This is also true for parliamentary speeches. We would hardly understand large parts and the political ‘point’ of Blair’s speech if we did not know that he was defending his Iraq policy in the British House of Commons. Much of the ‘content’ of this speech on Iraq could be (and has been) used by other speakers on other occasions, but obviously with very different functions while uttered in different situations. In this situation of the parliamentary debate, only Blair as Prime Minister – as well as some others allowed by the rules and the Speaker of the House -- may open the debate, present motions, and do other political things. And vice versa: What he says, and how he says it, may not always be appropriate in other situations. It is not likely that during a family dispute at home Tony Blair will say something like “I do not disrespect the views in opposition to mine”. Apparently, contexts also control discourse style, such as this formal use of the rhetorical negated antonym (litotes) and his choice of lexical items (e.g., ‘in opposition to mine’ instead of ‘dissident’). In other words, since Blair knows the specific contextual constraints of the parliamentary debates in the UK, he is able to formulate the content and style of his speech in accordance with such constraints.

‘Context’ in the humanities and social sciences

Literature, semiotics and the arts

In the study of literature and the arts, at various moments of history, scholars were urged to study works of art and their structures ‘in their own right’, and to ignore the social context or psychological conditions of the author. Eventually, such ‘isolationist’ or ‘autonomous’ positions (l’art pour l’art, formalism, New Criticism, close reading, etc., *Bell-Villada, 1996; *Gibbons, 1979; *Erlich,


1965) were rejected in favor of a more ‘contextual’ approach that accounts for many properties of works of art in terms of psychological, social, cultural or historical ‘circumstances’. This does not mean that we should be less precise and systematic in describing the structures of a poem or a novel, but our understanding is surely more complete when we are able to describe and also explain many more properties of such literary texts in terms of their various contexts. Contextualization is a fundamental part of our understanding of human conduct in general, and of literature and other texts and talk in particular. Indeed, con-texts are called that way, because etymologically they come with ‘texts’.

Similar observations may be made for the emergence of the new cross-discipline of semiotics in the 1960s, one of the paradigms of the structuralist movement in the humanities (see, among a vast number of other introductions, *Eco, 1978). Largely based on abstract concepts of ‘signs’ as applied to other forms of discourse and communication, e.g., in literature, narrative, film, dance, the arts or design, and inspired by the structuralist linguistic ideas of Saussure, Jakobson, Hjelmslev, Martinet, Barthes, Greimas, and others, few semiotic studies paid attention to social or cultural contexts.

Towards the 1990s, with the emergence of more explicit social semiotics and the critical analysis of multimodal messages that semiotics took a more social direction of research (*Van Leeuwen, 2005).

Linguistics

The same is true, as we shall see in more detail later (see Chapters 2 and 7), for the study of language. One does not need much historical knowledge of linguistics to know that the discipline for decades was limited to a ‘formalist’, ‘structuralist’ or ‘transformational’ study of signs, sounds, words, sentences, meanings or speech acts (see, e.g., the chapters in *Aronoff, 2003). In such studies lip service tends to be paid, if at all, and typically in introductory chapters only, to the fact that language and language use are of course social phenomena, and need to be studied in their social and cultural contexts. Few linguistic schools, originally interested only in grammar, have explored the role of context, except systemic and other functional approaches, to which we shall turn in Chapter 2 — see for instance the work of Givón (see, e.g., *Givón, 2005).

We have to wait until the late 1960s to witness the emergence of new interdisciplines such as pragmatics, psycholinguistics, sociolinguistics and the ethnography of speaking that began to provide some insight into the cognitive, and especially the social and


cultural ‘contexts’ of language and language use (see references in later chapters).

Thus, at the boundary of linguistics and philosophy, the study of speech acts, implicatures and conversational postulates (*Austin, 1962; *Grice, 1975; *Searle, 1969) for the first time not only emphasized the role social action in language use, but also an account of the (formal) contextual conditions of the appropriateness of utterances, as one of the characteristics of the new cross-discipline of pragmatics. It is also in this framework that the notion of ‘context’ received analysis in its own right (see, e.g., *Stalnaker, 1999).

Susan Ervin-Tripp, one of the pioneers of sociolinguistics, has been among those linguists who most emphatically advocated the explicit study of context, while criticizing the lack of context in earlier studies:

The omission of context from linguistic accounts has occurred because some linguists have considered contextual structure to be too chaotic, too idiosyncratic, to be characterized systematically. When linguists began to identify variable rules (*Labov, 1969), the separation of the variable from the obligatory or categorial was obvious and unavoidable. Variationists have gradually introduced context into their analyses. What we are now beginning to do is use contrasts in linguistic features, including those that are variable, as our guideposts for identifying both the structure of conversation and the structure of context, indeed the immediate social structure for speakers. Linguistic features can tell us what are natural human categories for context. Such an approach can at last systematize the domain of context (*Ervin-Tripp, 1996: 35).

Discourse analysis

Also the emerging discourse studies of the 1960s brought important new ideas to the study of language and communication (*Van Dijk, 1985, 1997). However, many of its first contributions were also rather structuralist and formal. Early text grammars often emulated generative sentence grammars (*Van Dijk, 1972), although with attempts to incorporate a formal account of context (*Van Dijk, 1977). Narrative and argumentation studies generally followed a formal paradigm, and seldom used more contextual approaches. And cognitive studies of text processing later offered insight into what could be called the ‘cognitive context’ of discourse, but — with some exceptions — would do so itself in terms of a socially isolated mind (*Van Dijk & Kintsch, 1983).


These first discourse analyses made one step ahead in the direction of an account of context, but mostly limited such a context to the verbal context or ‘co-text’ (*Petöfi, 1971) for units of language or language use. Many studies of ‘context’, both in linguistics as well as in other more formal approaches, still limit the notion of context to this ‘verbal context’ of previous sentences, propositions, utterances or turns of conversation.

We had to wait until the end of the 1970s and the early 1980s, with the arrival of critical linguistics (*Fowler, Hodge, Kress, & Trew, 1979) and critical discourse analysis, before discourse structures were more systematically studied in their social, historical and cultural contexts – something typically done in for instance the ethnography of speaking from the start (*Bauman & Scherzer, 1974). At the same time the study of gender, race and power became prominent aims of the study of language and discourse, thereby introducing fundamental aspects of societal contexts of language use and interaction. With the slow merger of different approaches to discourse in linguistics, literature, psychology and the social sciences, we thus at the- same time witness an increasing interest in aspects of text-context relationships.

Sociology

Also in sociology the end of the 1960s brought renewal by adding an important qualitative and microsociological dimension to the study of society by focusing on the details of situated interaction in general, and of conversation in particular (see, e.g., *Button, 1991). However, these early ‘ethnomethodological’ studies in many ways followed the same pattern as linguistics, by initially focusing more on the formal structures of interaction and conversation, such as the rules of turn-taking than on their social ‘situatedness’ (*Sacks, Schegloff & Jefferson, 1974). Later the methodological strictures of conversation analysis were somewhat loosened (or simply ignored) in order to more explicitly place the structures and strategies of conversation and interaction in their societal, institutional or cultural ‘context’ (for an early collection in this new direction of conversation analysis, see, e.g., *Boden & Zimmerman, 1991; and many other references in Chapter 5). Since the late 1990s we thus find increasing attention for context also in conversation analysis and related approaches to the study of language use and interaction (see also the special issue edited by Karen *Tracy, 1998).

Ethnography and anthropology

If there is one discipline that by definition should be an exception to this general trend of the humanities and the social


sciences to focus on formal properties first and deal with situations, context or environmental factors later, it is anthropology. In a way this is true as long as it dealt with the general, broader study of culture, and it is obviously also true for most ethnographic studies of discourse, which by definition are not limited to an account of discourse alone.

However, remarkably parallel to the other disciplines mentioned above, and in fact often preceding and influencing them, also modern anthropology has been going through important structuralist and formalist phases. In the 1960s, thus, the systematic study of folktales and myths in anthropology (e. g., by Lévi-Strauss; see *Lévi-Strauss, 1963) in many ways became the paradigm for the structuralism in the new discipline of semiotics and related studies, first in Europe and later also in the USA and elsewhere.

At the same time, ethnography in the USA made an original contribution in the 1960s by focusing on the detailed study of ‘communicative events’ and the ‘communicative competence’ of the members of a community (*Bauman & Sherzer, 1974; *Saville-Troike, 2002). In this paradigm Dell Hymes, its founder, formulated his well-known SPEAKING grid as a summary of the contextual factors of communicative events (*Hymes, 1972), one of the earliest more explicit accounts of the structures of context. Although this formulation was quite programmatic for the ethnography of speaking, this hardly led to a systematic exploration of these contextual factors of language use and discourse.

These developments in anthropology were initially closely related to those in linguistics and other social sciences. As is also the case in the disciplines mentioned above, we had to wait a decade for these ethnographic studies to take a more ‘contextual’ turn, introducing notions such as ‘recontextualization’ on the one hand, for instance in the work of Gumperz and others (*Gumperz, 1982a, 1982a), and dimensions such as identity, power, social structure, or ethnic relationships, on the other hand (see, e.g., the contributions in *Duranti, 2001). As we shall see in more detail in Chapter 6, linguistic anthropology thus became (again) one of the leading disciplines, this time because of several scholars and studies explicitly dealing with context, such as Hymes, Gumperz, Duranti and Hanks, among others.

Psychology

Psychology traditionally focused on people’s individual ‘behavior’ and later on their ‘minds’, and much less on ‘context’ beyond the experimental conditions of the laboratory — in which ‘context’ factors appear mostly as independent variables, such as the gender or age of the experimental subjects. Again, this has been true


for much of behaviorist and then cognitive psychology until the 1980s, and remains true for much mainstream psychology today, even in ‘social’ psychology. As always, there are notable exceptions, such as the work of F. C. Bartlett and Herbert Clark, to which we shall turn in Chapter 3.

In the last decades interest in the role of context in discourse processing has been growing rapidly in cognitive psychology, but whereas the social approaches to discourse have largely ignored the cognitive nature of context understanding, most cognitive psychologists have paid little attention to the sociolinguistic approaches to contextualization. Even those interested in discourse, generally focused on discourse structures, meaning and the nature of their interpretation in ‘situation models’ in memory, rather than on the role of context (and its memory representation) in production and understanding.

The study of ‘social cognition’ in modern social psychology seemed to provide the necessary social context to the study of cognition, but was generally limited to the study of formalist mental schemata and laboratory experiments that were hardly different from those in individual psychology (*Augoustinos & Walker, 1995). Indeed, until recently it was hard to find a reference to a book on society or culture in mainstream social psychology. Only since the 1980s do we witness the development towards a broader, ‘societal’ and ‘critical’ orientation to the study of minds, knowledge, persons, groups, or attitudes, on the one hand, and a more discursive, interactionist approach to social psychology, on the other hand (of many studies, see, e.g., *Resnick, Levine & Teasley, 1991; and further references in Chapter 4).

Computer Science and Artificial Intelligence

Interestingly, there is more work on context in formal approaches in computer science, Artificial Intelligence and the area of Natural Language Processing than in psychology. These approaches aim to account in formal terms for discourse interpretation, e.g., of pronouns, deictic expressions, verb tenses, presuppositions, knowledge accumulation, and many other properties of discourse that need context modeling (see, e.g., *Akman, Bouqet, Thomason, Young, 2001; *Iwańska, & Zadrozny, 1997). This work is related to work in formal grammar, logic and philosophy, originally inspired by Richard *Montague (1974), and Hans *Kamp (see *Kamp & Reyle, 1993). Although often called formal pragmatics, most of this work focuses on semantics, that is on how to interpret discourse expressions in terms of (formally represented) contexts, rather than on their appropriateness. This formal approach to contexts is also the only direction of research that represents context as models, as I shall also do, but then not as


formal models, but as mental models. Also, these scholars are the only ones who organize a bi-annual conference on context. In these paradigms, contexts were often reduced to sets of propositions (see, also *Sperber & Wilson, 1995), for instance, but hardly analyzed in their own right beyond some obvious parameters as time, place and shared knowledge (Common Ground) of the participants, as we also know from psychology (see, also *Clark, 1996).

Conclusion

We see that most of the humanities and the social sciences have shown a very similar development between the 1960s and 1980s, namely from a formal study of sentences, discourses, speech acts, interaction or communicative events or mental processing, on the one hand, to more socially or contextually sensitive approaches, on the other hand. During the 1990s, in most contemporary discourse studies, sociolinguistics, social psychology, ethnography, formal linguistics and AI, “context” and “contextualization” became key concepts.

Given these developments, one would expect that at present the notion of context not only would have been used profusely in many disciplines, but also that many articles and monographs would have been specifically dedicated to this notion. Nothing is less true, however. There are many articles and books that feature the notion of context in their titles or descriptors, but often these publications do not study context per se, but simply take it for granted.

There are articles, edited books and special journal issues that do study the notion of context more explicitly (see, e. g., *Auer & Luzio, 1992; *Duranti & Goodwin, 1992; *Fetzer, 2004; *Leckie-Tarry, 1995; *Owen, 1997; *Tracy, 1998, and other references in the next chapters), but so far there is not a single monograph that offers an integrated theory of the notion of context in the humanities and the social sciences. It is the aim of this book to offer just such an integrated, multidisciplinary theory.

The everyday uses of ‘context’

Before we deal with the notion of ‘context’ in a more systematic and explicit way in the chapters that follow, we need to describe and delimit it in a more informal way. In order to do this, let us begin with a brief look at some uses of the word ‘context’. Later, we present a more systematic study of the uses of ‘context’ in various corpora of text and talk.

1. A search with the Google search machine on the internet produces the staggering amount of 15,900,000


webpages that mention the notion of ‘context’ on August 27, 2002, nearly the double amount two years later: 27.100.000 pages, on October 8, 2004, about 60 million pages on April 2, 2005, and a staggering number of about 704,000,000 pages on April 23, 2006.

2. In Collins/Cobuild Wordbank’s English corpus of 56 million words, the word ‘context’ appears 1642 times (in 2002), that is, once every 34,104 words. Just for comparison, the apparently more common word ‘situation’ appears 7655 times, and ‘environment’ 4369 times. Significant collocates (words that occur close to it) for ‘context’ are: ‘social’, ‘historical’, ‘wider’, ‘cultural’, ‘broader’, ‘European’, ‘family’, ‘modern’, ‘contemporary’, ‘international’ and ‘global’ – apart from the obvious words such as definite and indefinite articles, demonstratives and prepositions such as ‘in’ or ‘within’, or verbs such as ‘put’, ‘taken’ or ‘seen’.

3. Dictionaries list basically two meanings of the word ‘context’, namely verbal context and conditions and circumstances, as the following in Webster’s Encyclopedic Unabridged (1996 edition):

1. the parts of a written or spoken statement that precede or follow a specific word or passage, usually influencing its meaning or effect: You have misinterpreted my sentence because you took it out of context.

2. the set of circumstances or facts that surround a particular event, situation, etc.

Merriam-Webster’s Collegiate dictionary uses the term ‘interrelated conditions’ for the second meaning, and mentions ‘environment’ or ‘setting’ as synonyms.

The Spanish dictionary of the Academy (DRAE) also lists these two basic meanings.

4. A quick glance of 10 uses of ‘context’ during one month on the opinion pages of the Washington Post, June, 1993, according to Lexis-Nexis database), produces passages in which the meaning of ‘context’ can best be paraphrased as follows (we shall later examine such examples in more detail):

a. political situationb. on the occasion of (‘in the context of’)c. in the perspective ofd. against the background ofe. in the situation of.


No use of ‘context’ in the linguistic sense of ‘co-text’ occurs in these data.

Similarly, a look at the uses of the Spanish word ‘contexto’, as it occurred in the leading Spanish newspaper El País provides the following meanings:

a. on the occasion ofb. in a situation ofc. economic situationd. discursive environmente. political situationf. cultural situation

These first observations of the everyday uses of the term ‘context’ show several things. First of all, although ‘context’ is used in millions of webpages, it is used less than words with related meanings, such as ‘situation’ or ‘environment’: In the bank of English it is used only once every 34,000 words or so; in the opinion articles of the Washington Post of June 1993, together 177,535 words (in the Nexis-Lexis file), the word ‘context’ occurs only 11 times.

Secondly, the dictionaries basically list two meanings, namely that of verbal context, and that of social, political, economic or historical situation or circumstances or in relation to geographical extension, as in “international context”. In both cases, the idea is that the context somehow influences a word, passage, meaning or event.

Thirdly, the actual uses in the press, both in the USA and Spain, favor the second dictionary meaning of ‘situation’ or ‘occasion’ and closely related meanings, such as ‘perspective’, and so on. Because of the opinions and news in the press, the meaning of ‘context’ is especially related to social, political, financial, and cultural backgrounds. In general, then, the everyday uses of ‘context’ imply that some thing (event, action) is related to a given situation, conditions, circumstances or background. In terms of the well-known metaphors used in Gestalt Theory, we would say that context is like the ‘ground’ for the ‘figure’ focused upon.

There are some uses that are more idiosyncratic, as in the example, “people who place their life in a context of meaning and purpose”(Washington Post, June 11, 1993), where ‘context’ means something vague as ‘having’, ‘against a background of’ or ‘with the purpose of’, which also imply the meaning ‘being related to’.

As a first exploration of the everyday uses of ‘context’ this will have to do. It is obviously a word that is preferred in more formal, often written modes of discourse, and especially used in discourse fragments that want to explain some thing in relation to other, ‘broader’ things, such as conditions and circumstances or a complex environment. Indeed, also the dictionary definitions emphasize this


explanatory dimension of ‘context’ in the sense that the meaning or significance of words or other events somehow depend on the context. Hence the quite normative constraint, even mentioned as an example in Webster’s, that one should not take or cite people’s words “out of context”, because that may alter their meaning. In that sense, the word ‘context’ correctly signals that it is a verbal or social environment of ‘text’, that is, con-text ‘comes with’ or ‘goes with’ the text. Or otherwise, etymologically, that it is something that is “interwoven” with something else (from Latin con-tegěre).

Although the more linguistic and discursive notions of context will be prominently analyzed in this book, it should be emphasized that in everyday uses the notion of context is much more general. The majority of uses both in the media and in the social sciences emphasize that some event or action, now focused on, should be seen or explained in relation to another one, or rather a complex of other events or situation, that form its conditions.

A corpus study of “context” in scientific titles

In order to further explore the use of the scientific notion of context, we did a corpus study of all 3428 English book titles (in 2002) in the Library of Congress and all 5104 article titles in the Social Science Citation Index (between 1993 and 2001) that had the word ‘context’ in them.

Our analyses (not reported here) of the (co-)occurrence of “context” in English book titles are consistent with those of articles: ‘context’ is typically used as a summary term for temporal, geographical and sociocultural situations, factors or variables that impinge on something else. The notion of “verbal context” (co-text) hardly occurs outside of linguistics. Often the notion barely has meaning and only vaguely indicates some relationship between a phenomenon under study or focus and something else, as in “context of change” or “context of crisis”, or the influence of context is mentioned without mentioning what kind of context is meant.

This brief summary of a preliminary corpus study of the ‘scientific’ uses of context, as summarized in the titles of articles and books in English (and hence, in principle, a prominent concept in their semantic macrostructure) suggests that many social phenomena are not studied autonomously, ‘as such’, but in relationship to some kind of influencing geographical, historical, sociocultural or organizational setting or environment, thereby also limiting the scope of the study. Studying interaction, poverty, AIDS, management, among a large number of other phenomena in society is generally impossible when done in general terms, and books and especially articles need to do so by limiting the scope of the study to a specific period, country, culture, neighborhood, or organization.


Towards a new theory of context

In this book, I shall not further explore the uses of the term “context” in various types of discourse or communicative situations, but design elements of a theoretical framework for a concept of ‘context’ that can be used in theories of language, discourse, cognition, interaction, society, politics and culture. Before we deal with the details of such a theory in the next chapters, and before we define context in language, cognition, society and culture, respectively, let me briefly summarize some of its main tenets. We do so first without providing relevant references to other work, which will be given in the next chapters.

Contexts as subjective mental models of communicative situations

Taking into account, but departing from other approaches to context, I shall first of all propose that we should not define this linguistic or communicational concept of context in terms of some kind of objective, social, political or cultural situation of talk or text. Rather, I shall argue that contexts are to be defined in terms of the ongoing (inter)subjective understandings, interpretations or definitions of participants of such situations as communicative environments of their production or understanding of talk or text.

In more technical terms, this means that contexts are members’ constructs. Although such concepts of context have occasionally been formulated before (see below for references) I have made ‘subjective definitions of situations’ more explicit as a specific type of dynamic mental model, which I call context model (for the first time in some detail in *Van Dijk, 1998, but already briefly introduced and defined in *Van Dijk, 1981, 1987, and in *Van Dijk & Kintsch, 1983). Such a model represents or constructs the relevant parameters of some social situation found to be relevant by the language users for the production or comprehension of (a fragment of) discourse. Conversely, discourse may also contribute to the ongoing change of such context models in interaction and communication.

In other words, social, political, economic or cultural situations and their properties may very well influence discourse and conduct, but they do so only indirectly, namely through their interpretations or definitions by the participants. This may seem trivial to most psychologists, but is a fundamental principle largely ignored – or even denied – by more ‘social’ approaches to discourse and context (see the contributions to the special of Discourse Studies on this topic: *Van Dijk, 2006).


The notion of context defined in the theoretical framework of this book is different from my first, more formal, analyses of context in Text and Context (*Van Dijk, 1977), where context is also briefly defined in terms of models, but in abstract terms, and not as mental models. Also in this earlier study a distinction is made between contexts as a construct of ‘facts’ that determine the appropriateness of utterances, on the one hand, and socially real ‘communicative situations’, on the other hand.

Subjective, intersubjective and social contexts

The definition of contexts as mental models, that is, as subjective participant constructs, of course does not imply that participants in interaction have a totally idiosyncratic understanding of the communicative situation. As members of the same epistemic or discourse communities they share categories, rules and other principles for the interpretation of events and situations, and hence also of communicative situations. This means that for all practical purposes large part of their context models will overlap with those of other participants, and thus define the necessary intersubjectivity of interaction and communication.

It should be emphasized that from the perspective of the participants, such a definition of the communicative situation may well be based on ‘objective facts’, and not be a ‘mere’ subjective construct: No one in British Parliamentary has any doubts that Tony Blair, when addressing the MPs during his Iraq speech, does so as Prime Minister, for example. And no one of the MPs doubts that they are in fact MPs. They may have doubt about other things, such as what Tony Blair’s exact goals are, but many of the aspects of the current communicative context are as ‘objective’ for them as their own existence. That is, we should carefully distinguish between our constructivist theoretical position, on the one hand, which emphasizes the (inter) subjectivity of situation definitions by the participants, and the self-experiences of the participants, on the other hand. We do so precisely because we want to emphasize that the way participants act in communicative situations is as they define them (as ‘true’ or ‘objective’), that is, according to their own model of the situation, and not according to the definition of the situation by the theorist or analyst.

We shall see later that the context model as a unique personal experience (and thus as part of ‘autobiographical’ episodic memory) is construed on the basis of more general, socioculturally shared knowledge, norms, values and — for specific groups — ideologies, that is, some kind of (social) Common Ground.

However, the intersubjectivity of (parts of) context models or their shared sociocultural basis do not imply the identity of such models or a ‘shared’ or ‘dyadic’ model. Each participant in the


interaction and communicative situation has her or his own self-definition, body, location, perspective, personal experiences, opinions, and so on, and these define the uniqueness of each context model for each moment in each interaction. In an ongoing interaction a recipient cannot possibly have the exact same context model of the speaker, if only because a recipient may (not yet) know the intentions of the speaker, as well as the way the speaker models the recipient. Thus, the participants ‘see’ the ‘same situation’ from different perspectives, and they have by definition different communicative roles, namely one as speaker, and the other as recipient. These differences between the context models of the participants are multiply signaled in interaction, not only by errors, but also by requests for clarification: ”What do you mean?”, or “Are you threatening me!”, or “Yes, you told me that already”, and so on. They also appear in the different interpretations (referents) of the deictic expressions used by the participants.

Contexts are both uniquely ‘embodied’, presupposing the bodily position and space of each participant, e.g., as indexed by deictic expressions, as well as uniquely ‘minded’, that is featuring the current perception, view, perspective, opinions and personal experiences and history of each participant. This is also one of the reasons why different speakers in the ‘same situation’ do not exactly speak in the same way. If the definition of the situation would be (fully) shared by participants, this would also rule out any mistakes, deviance or other forms of personal variation and change of socially shared rules. Among many other effects of the assumption of identical situation definitions, there would be no communicative conflict, and no language change.

And last but not least, if contexts would be fully shared, the very interaction and discourse would be pointless, because in that case the recipient would already know or feel everything the speakers knows or feels.

In other words, there are many fundamental cognitive and interactional reasons why contexts must be unique and subjective, and hence to be accounted for in terms of mental models. At the same time, however, the social nature of interaction or communication also presupposes that many aspects of these context models are shared, and that participants know much about how other participants understand the current communicative situation and they usually do (but need not) orient to such shared aspects of the communicative situation. Hence models might be said to ‘overlap’, or have some kind of ‘common ground’, but they are not identical or (fully) shared.

What is shared with all competent members of the same community is the sociocultural basis of these models, that is social representations such as knowledge, norms, values and ideologies.


These are ‘applied’ in the formation of each individual model, but again, these applications may again be unique or personal.

In sum, contexts as mental models are personal and unique, but have intersubjective and social dimensions based on socioculturally shared cognitions, including general rules of appropriateness.

Contexts are dynamic models

Contexts are not static, but dynamic (*van Dijk, 1977: 191f). They are constructed for each new situation and then ongoingly updated and adapted to (the subjective interpretation of the) current constraints of the situation, including the immediately preceding discourse and interaction. That is, contexts develop ‘ongoingly’ and ‘on line’, that is, in parallel with interaction and thought. Actually, especially in formal, planned talk, before speaking, participants often already have some vague idea or plan (that is, a provisional outline for a context model) about the situation (setting, participants, aims, etc.) in which they are going to speak, as we already observed for the hypothetic context model of Tony Blair: also because of earlier context models of speeches in parliament, large part of the current model is already pre-programmed. That is, the context model is not construed from scratch at the moment talk begins, but usually will further specify and adapt a model that was already designed before. And since context models are a specific form of models of everyday experience, even spontaneous encounters already have some aspects of the model in place, such as the Setting and one or more participants.

Context model schemas Contexts, just like text and talk, are not unorganized sets of

mental units (propositions, etc.) of some kind, but ordered: they have some kind of structure (*Van Dijk, 1981). Without such order they would not be able to function adequately in the fast processes of on-line discourse production, comprehension and interaction, which require processing of vast amounts of information in milliseconds (*Van Dijk & Kintsch, 1983). That is, we assume that context models consist of a general schema of contextual categories that represent the usually relevant social conditions of discourse.

Traditionally, setting (time, place), and categories for various participant characteristics such as gender, class, ethnicity, age, social roles, status, power, goals and (mutual) knowledge have informally been used as properties of social situations that may influence discourse. Such schemas are socially acquired and quite general for each culture, even when in actual discourse unique models are formed with such schemas – much in the way we are able


to make new, unique sentences and discourses with a more general grammar and rules of discourse. In interaction, social members gradually learn to infer which types of social conditions may be relevant for variable discourse structures. For instance, in many languages, the selection of pronouns or honorifics is – indirectly -- conditioned by such social attributes of actors as age, class or status taken to be relevant categories of context of talk.

At the same time, such context model schemas need to be flexible and strategically adaptable to (changing, unique) social situations. Thus, in one situation gender or age may be relevant, but such relevance may also be variable: sometimes gender is more relevant than in other situations.

Similarly, these social situational categories are not acting alone, but are coordinated and their influence may be concurrent or opposed. Thus, whereas perceived age in many cultures may tend to be construed as a category that controls more polite, respectful or honorific uses of languages (in greetings, pronouns, terms of address, noun morphology, verb forms, and so on), it may combine with conditions such as familiarity or intimacy as types of social relations between speech participants, and then lead to different discourse forms. Hence, one of the major tasks of a theory of context is to formulate the rules or strategies that coordinate the selection and influence of social categories of context models. Indeed, how much of the specific news report style of a black, female, socialist journalist from New York may be explained in terms of these combined, but variably relevant, identities as represented in the context models of such a woman?

The pragmatic functions of context models

The fundamental function of context models is to make sure that participants are able to produce and understand text or talk as being ‘appropriate’ in the current communicative situation. In this sense, a theory of context would be one of the aims of a pragmatic account of discourse. It explains how language users adapt their discursive interaction to the current cognitive and sociocultural ‘environments’. Such a theory also makes explicit the usual felicity conditions of illocutionary acts and the appropriateness conditions of politeness and other dimensions of interaction. Hence, an explicit theory of context at the same time provides a solid basis for various approaches in pragmatics.

Context models as interface between society and discourse

I shall later describe in more detail the cognitive, social and cultural properties of this notion of context defined as the subjective mental models of social situations as construed by the participants in


communicative events. However, it is important to repeat that contexts – at least in a theory of discourse, interaction and communication -- are not conceptualized as something “objectively” out-there, but as a socially based and interactionally situated personal interpretation or definition of the relevant aspects of the surrounding social situation.

This does not mean that I propose to reduce the sociocultural constraints of language use or communication to cognitive representations, but only that the way such constraints operate on discourse is necessarily mentally mediated. That is, context models are the indispensable interface between society and discourse.

Our current insights into the cognitive basis of discourse production and understanding are inconsistent with the (still) dominant, but very vague, view that assumes a direct influence of social structure or situations on discourse structure. Indeed, such a view can only survive because it does not make explicit how discourse is actually produced, but rather sees discourse as magically emerging as talk or text in the social situation — if it analyzes situations at all. In sum, my theory of discourse and context needs to be configured within a triangle linking Discourse and Society via Cognition. Only in this way are we able to model not only language users as social actors who talk, but also think before and while they talk, and are continuously busy understanding other participants, actions and other aspects of the communicative environment. Context models, thus, are part of a class of very general sociocognitive devices that regulate how humans are able to act appropriately in social situations.

One of the many advantages of such a ‘mediated’ account of context is that although they are partly interpersonally shared and hence allow communication, contexts are by definition (at least slightly) different for all participants. This means that also the production and comprehension of discourse controlled by such context models is both socially constrained as well as individually variable, as is the case for all human action and experience.

Contextual variation is an important aspect of communication. We know from everyday experience that people usually produce and understand discourses in different ways, even when their social ‘background’ and the current situation of interaction are the ‘same.’ We must conclude that we cannot simply equate context with social situation, because in that case people in the same social ‘context’ would produce and understand discourses in the same way. And we know they don’t.

Contexts as everyday experience

This also means that besides the relevant social constraints of situations, language users also routinely integrate (earlier) personal


experiences, perspectives or opinions as elements of their context models. That is, contexts are subjectively produced as parts of the everyday lives of the participants as human beings, subjects, group members or social actors. Indeed, we shall see later that context models are not some strange theoretical invention, but basically a specific case of what we call our daily experiences. Also such experiences are subjective mental models, as stored in what is called ‘episodic’ (personal, autobiographical) memory. The interpretation of communication episodes in terms of context models is just one of these daily ‘experience models’.

Against reductionist accounts of contexts

The concept of ‘context’ as proposed in this book is consistent with several ‘constructionist’ approaches, namely when defined as a product of the subjective constructs or definitions of participants in interaction who are actively and ongoingly understanding the relevant aspects of the ongoing situation. Unlike these approaches, however, I do not leave such notions as ‘understanding’, ‘definition’ or ‘construct’ undefined, and I do not reduce them to actual expressions in discourse.

Rather, I define these notions in terms of the theoretically based, explicit notions of contemporary model theory in psychology. Understanding, interpretation, definition and construction, as intended here, are mental phenomena and hence need a theory of cognition. That such understandings, definitions or constructs also may (though need not explicitly) be expressed or ‘performed’ in discourse requires an account in terms of discourse, but does not make these notions ‘discursive’.

Unlike discursive psychology I do not reduce ‘mental’ notions to ‘discourse’ notions, even when it is of course very interesting to also study how people talk about or use mental notions in their discourse. On the contrary, as a discourse analyst I claim that many aspects of discourse should be accounted for both discursive or interactional terms as well as in cognitive, social, cultural or political terms, as we also have seen in the brief analysis of a fragment of Tony Blair’s speech in parliament. Although already questionable for grammar, an ‘autonomous’ approach to talk and text is even less feasible, and would yield very superficial and fragmentary analyses. The elementary notions of local and global coherence already require a cognitive approach in terms of mental models and knowledge, and we have shown that without a contextual analysis the very point — for the participants — of a talk or text cannot be assessed. And although I define contexts in terms of mental models, we shall see in the respective chapters of this book that they need to be based on social psychological, social, cultural and political notions.


Contexts vs. texts

Context is one of the notions that should not be reduced to text or talk. Contexts are called ‘contexts’ precisely because they are not ‘texts’ – although such ‘texts’ may be or become parts of contexts of course. Indeed, context models and their properties remain largely implicit and presupposed. They influence talk and text in indirect ways that only under specific circumstances (problems, errors, misunderstandings) are being made explicit in talk and text itself.

Contexts are signaled or indexed rather than expressed. Their properties often need to be inferred from structures and variations of discourse as used in different social situations, and this is what both recipients and analysts do.

We cannot observe contexts directly. We postulate or infer them on the basis of discourse or human conduct, as is the case for many other properties of language and communication, such as categories, rules, norms, or grammars. On the other hand, they are not merely theoretical constructs of analysts, but ‘real’ mental phenomena, more or less consciously experienced by language users. Indeed, as properties of the a social situation of text and talk they are ‘real’ for people because they are real in their consequences for text and talk. If necessary, people are or can be made explicitly ‘aware’ of their communicative surroundings.

Relevance

One of the notions that needs to be further examined in such a model theory of context, is that of relevance, namely when contexts are defined or construed in terms of what language users find relevant in the social situation. Indeed, my theory of context at the same time should provide a contribution to a theory of relevance, with which it is related (*Sperber & Wilson, 1995). I shall examine this notion more closely in Chapter 3 on context and cognition, since ‘relevance’ is a cognitive notion defining the way language users analyze, understand and represent their social environment.

Macro and micro contexts

Another aspect of context theory to be examined is the well-known level or scope difference of social situations, i.e., the difference between local or micro contexts and global and macro contexts, respectively. Such notions would apply to, respectively, everyday, face-to-face interactions and situations, on the one hand, and societal structure involving groups, institutions, power relations, on the other hand, as we saw in the example of Tony Blair’s speech in the British House of Commons.


More specifically we need to know how models of social macrostructures – as is also the case in discourse processing at the semantic level – may influence modeling of local microstructures such as immediate or ‘proximate’ contexts, and vice versa.

Most of the more informal and social scientific uses of ‘context’ we briefly examined above refer to ‘macro’ contexts, that is, more global societal or political situations, and not to the local, face-to-face situations of everyday conversation.

Also in sociolinguistics, reference to ‘the social context’ usually involves societal structures, groups or categories of people, institutions and so on; class, gender, age, institution, etc. In other words, we need to introduce special terminological distinctions to avoid confusing these different aspects of social situations or environments.

Hence I shall provisionally use the familiar distinctions between global and macro on the one hand and local and micro on the other hand, and only the latter will also be called ‘situation’. Note though that when used as a technical term, I shall only use the notion of ‘context’ as defined above, namely as a subjective mental model of the local or global social environment of interaction or communication. And when I technically refer to the ‘social context’, this means ‘the social context as defined by a language user’.

Throughout this book, however, I shall focus more on the context modeling of local (micro) situations than on the way language users at the same time model their (often backgrounded) understanding of relevant social (macro) structure — for instance the role of the institution of parliament or of British foreign policy in Tony Blair’s context model when he addressed the House of Commons.

Contexts as the relative ‘Center of My World’.

Contexts are crucially egocentric. They are defined by a set of parameters that include a Setting that is the spatiotemporal hic et nunc of the ongoing act of speaking or writing, of Ego as speaker or listener, of other participants whom I now address, or listen to, as well as of the ongoing social actions I am now engaging in with specific aims and purposes, and on the basis of what I now know and believe.

The properties of this ‘egocentric’ nature of contexts define the conditions of the many different deictic expressions of language, such as personal pronouns, demonstratives, verb tenses, motion verbs, prepositions, expressions of politeness and deference, and so on. It is not only from my own point of view or perspective that I am speaking or writing, but also describe events and the world, thus


giving rise to a frame of reference that is essentially relative to me, as the speaker. That is, in our culture as well as in many others, we are able to describe, direct others and orient ourselves in space in relation to where we are now speaking. Other cultures prominently or only use absolute descriptions, such as ‘north of’ the tree’ instead of ‘before’ or ‘after’ the tree (*Levinson, 2003) and hence must have rather different context model schemas.

The orientation of the speaker in an (embodied) space has metaphorical extensions for other forms of ‘placing’ ourselves and others, for instance when we describe political orientation as being ‘on the left’ or ‘the extreme right’, expressions metonymically derived from the placement of political parties in parliament. The contextual relativity of the use of such expressions obviously also depends on where we ‘stand’ ourselves, politically speaking, that is on our context models.

Similar observations hold for temporal deixis. Disregarding for a moment the complex relations between time and tense, one of the functions of the present tense in English is to index the time or period of speaking, also referred to by temporal adverbs such as ‘now’. And it is with respect to that moment of time, as defined in the context model of the speaker, that other temporal adverbial expressions, such as ‘yesterday’, ‘tomorrow’, ‘just’, ‘recently’, ‘in a moment’, as well as expressions such as ‘past’, ‘present’ and ‘future’ are interpreted.

As we have metaphorical extensions of spatial deixis, we also have metaphorical extension of time, e.g., when we describe people or objects as ‘old fashioned’ or ‘modern’, ‘backward’, ‘traditional’ or ‘progressive’ (*Fabian, 1983). In other words, contextual Setting ‘places’ speakers both spatiotemporally as well as ideologically.

And finally, such relative contextual coordinates and frames of reference also apply to the identities and the group membership of the speaker and the other participants or people talked about, namely as members of ingroups and outgroups, as Us vs. Them.

In sum, the definition of the relevant parameters of context represent how speakers perceive, represent and describe themselves and others, as well as events and actions relative to their current place, time, perspective, group membership, knowledge and beliefs. They do so both semantically for the reference and interpretation of deictic expressions, as well as pragmatically to adapt discourse appropriately to the current communicative situation. Thus, expressions of deference and politeness, among many others, depend on the social ‘position’ speakers occupy in relationship to recipients — both as to social ‘place’ and relationships of status and power, as well as to social ‘time’ and relationships of age.

And finally, if we take speaker’s knowledge and beliefs as elements of context models, we again find the same fundamental principle of contextual relativity: Speakers make assertions on the


basis of what they now know, what they know at each point of a conversation about the knowledge of the recipients and on the basis of the shared knowledge of their community. Similarly, they express current personal opinions which however also depend on the attitudes and ideologies of the group(s) with which they now identify. Thus, the parameters of what participants now construe as relevant of the communicative event are at the same time the egocentric point of orientation for the spatiotemporal, cognitive, social, political and cultural coordinates of their semantic and pragmatic models of communication and interaction, and hence of their discourse.

We see how making contexts explicit as subjective models of communicative situations provides an analytical framework for the account of a vast number of properties of discourse, and especially of the semantic relativity and pragmatic appropriateness of language use.

The general idea of context as mental models of social members has many dimensions that will be elaborated in the respective chapters of this book. Below I summarize some of these dimensions as an outline for further analysis that will be provided in the next chapters – as well as the extensive bibliographical references about each dimension of the new multidisciplinary theory of context.

Semantics vs. pragmatics of context

We have seen that contexts provide the basis for the semantics of deictic expressions as well as for the pragmatics of communicative appropriateness. For theoretical and analytical reasons these different dimensions of a theory of context need to be carefully distinguished.

When speakers explicitly or implicitly refer to properties of their immediate ‘environment’, such as place, space, time, themselves or other participants, and so on, contexts provide the models for the ‘situation semantics’ of deictic expressions (*Barwise & Perry, 1983).

We have suggested above that formal work on context, also in Artificial Intelligence, is limited to such contextual interpretations of expressions, such as pronouns, demonstratives, time and place adverbials, verb tenses, and so on. Strictly speaking such studies belong to the realm of discourse semantics: they are about (relative) truth values, reference and interpretation. The only difference with other (formal or informal) semantics is that interpretations are specified with respect to the current situation of speaking, and not more generally to (for instance) possible worlds or even more generally and more abstractly.

This ‘semantic’ approach to context will not be a prominent topic in this book, also because it has been extensively explored in


other studies on deixis and relative or situation semantics in various disciplines (*Akman, et al, 2001; *Barwise & Perry, 1983; *Hanks, 1992; *Jarvella & Klein, 1982; *Levinson, 1993, 2003).

The focus of this book will be on the ‘pragmatics’ of context, that is, on those properties of communicative situations that are (defined as) relevant for the appropriateness of text and talk and that control many more than just the deictic expressions in discourse, namely its style and register, its expressions of politeness and deference, speaker roles, topic selection, speech acts, and so on. We have seen that these properties of contexts are not usually referred to, for instance by deictic expressions, but rather presupposed in ‘normal’ situations, and hence precisely not referred to, but only indexed or signaled, unless there is some kind of ‘trouble’ or special circumstances. These properties are for instance current time, place and space; current speakers and recipients and their various communicative and social roles, identities and relations; the current social acts being accomplished; the current intentions and goals of such actions; or the knowledge and other beliefs that are expressed or presupposed. Each of these categories, and more, will be considered in much detail in the following chapters.

Among a host of other things, context pragmatics describes and explains why it is usually inappropriate to request information we already have, to issue commands to superiors or loved ones, to use formal pronouns when speaking to our friends, why a speech in parliament is more formal than a dinner conversation, why we address children in a different way than the elderly, why debates in parliament have political implications, and so on.

Hence, the fundamental difference between semantics and pragmatics is not defined in terms of whether or not they are relative to the communicative situation (both are), but what kind of dimension of language use we are studying, namely reference (denotation, extension, truth, interpretation, etc.) and appropriateness (adequacy, fit, relevance, etc.), respectively. I shall come back to the notion of ‘appropriateness’ below.

It is not always easy to distinguish these two different levels of analysis and they sometimes seem to overlap in actual discourse, because the same expression may have a semantic and a pragmatic (contextual) function. The fundamentally deictic pronoun I refers to the specific person who is now speaking (when not citing or speaking for others), say Mary, which is a referential-semantic function. But its use as a deictic-expression-with-such-reference is (more) appropriate in a communicative situation in which the person who uses such an expression also (knows she) now has the communicative role of the Speaker. Indeed, although the expression “Mary has a headache” is obviously true when said by Mary (but also by others who now know about her headache) when she has a headache and is referring to herself, it is usually inappropriate when


uttered by Mary herself, since in that situation Mary would normally be expected to say “I have a headache”, which obviously cannot be used by others speaking about Mary’s headache.

Not using ‘I’ when self-referring to the current speaker or writer is appropriate only in specific formal situations, such as the authorial ‘we’ in this book, or when Tony Blair refers to himself as the Prime Minister, thus making salient his political identity or establishing distance between personal and formal responsibilities (see, e.g., *Martín Rojo & Van Dijk, 1997, about a speech of the Spanish Minister of the Interior legitimizing the forced expulsion of African immigrants).

Although the distinctions made above will often allow us to distinguish between ‘semantic’ and ‘pragmatic’ properties of language use or discourse, the criteria used are hardly explicit, so that confusion cannot be ruled out. Thus, if we take context-dependency of discourse structures (including semantic structures) as a criterion, a ‘pragmatic’ account would include all non context-free structures at all levels of discourse, viz., sound structure (e.g., intonation), morphology, syntax and the lexicon, pronouns, deictic expressions, meaning (from global topics to local word meanings), implicatures, politeness phenomena, style, speech acts, and so on — that is, all properties of discourse that can vary across communicative situations. In that case we would then have to distinguish between pragmatic phonology, pragmatic syntax, pragmatic semantics, pragmatic stylistics, etc. Pragmatics in that conception is ‘transversal’ through all levels of discourse, and hence rather a dimension than a level of description. And apart from the other level-specific evaluation criteria (well-formedness, grammaticality, meaningfulness, reference, truth, etc.) for the description of discourse structures, the pragmatic dimension at the same time allows discourses (or specific discourse structures) to be evaluated with respect to their contextual appropriateness. Such an approach would define a broad conception of pragmatics.

On the other hand, a narrow conception of pragmatics would define a specific level of analysis, namely that of action, besides the levels of form (expressions: sounds, syntax, lexicon) and meaning (or reference), typically (but not exclusively) so in the study of speech acts or illocution, whose felicity conditions are always contextual, e.g., the intentions, knowledge, beliefs or social positions of speakers. However in that case also the typical analysis of interaction in for instance conversation analysis would be a type of pragmatics.

To add to the confusion, even this action-based concept of pragmatics may become ‘transversal’ when we also speak of locutionary act, propositional act, stylistic act, etc. But in that case such ‘acts’ are rather aspects or levels of more the comprehensive act of a speech act or of another social act accomplished by language


use. In this case, a contextual approach to discourse provides the basis for the felicity conditions of speech acts, but does not collapse pragmatics with contextual discourse study or ‘broad’ pragmatics.

This study focuses on the general nature and role of context in the study of discourse, and in that sense it is a study in a ‘broad’ discourse pragmatics, as defined above, even when I in general prefer to limit the concept of pragmatics to the description of the action dimension of language use.

Appropriateness

The different components of grammar or a theory of language are traditionally associated with abstract criteria that summarize the specific goal or explanandum of that component. Thus, syntax is usually associated with the notion of well-formedness (or grammaticalness), as is also the case for formal languages. Similarly, intentional (meaning) semantics is traditionally associated with the notion of meaningfulness, and extensional semantics with that of truth (or events, situations, etc., depending on the kind of semantics).

Likewise, as explained above, the theory of speech acts, and more generally pragmatics, is often based on the concept of appropriateness (for detailed discussion, see *Fetzer, 2004; see also *Van Dijk, 1981). That is, we may form a well-formed, meaningful and true sentence, but the actual use of such a sentence in a specific situation may not be acceptable, for other than grammatical reasons, e.g., while violating communicative, interactional or social norms. For instance, as we have seen above, and as we know from pragmatic studies, it is often inappropriate to assert what we know our recipients to know already, to ask what we know the recipients do not know, to issue commands to superiors or friends, to use informal words in a very formal situation, to be too polite to close friends or impolite to strangers or superiors, and so on.

This notion of appropriateness is often used with its everyday, commonsense meaning of social propriety, that is, defined in terms of social norms, rules, criteria or other conditions of interaction. However, as is the case for similar general concepts, it is fuzzy and not without ambiguity. The examples given above are traditional ones to which the notion of appropriateness is applied — for instance in a theory of speech acts.

But what about, e.g., not responding to a question or request, ignoring the topic introduced by a speaker, breaking off a conversation without properly ‘ending’ it, constantly interrupting a


speaker, lying, speaking half-truths, shouting to people who are close or whispering when one cannot be heard, tell a personal story in a formal news report, participate in a job interview or an oral exam and say nothing, and so on?

The ways to break some kind of rule, convention, norm, expectation, regularity, or tradition are nearly endless, and many of them are found to be ‘inappropriate’ – even when grammatically speaking there is no rule that is broken. From the last examples we see that not satisfying conditions of speech acts, disregarding maxims (e.g., to speak the truth, the whole truth and nothing but the truth, as the legal formula tells people who declare in court), or breaking rules of conversation (e.g., of interruptions or ending conversations) may be found (more or less) inappropriate. But these are quite diverse forms of rule breaking, and again quite different from breaking the formal style of a news report with an informal personal experience.

Hence, as is the case for well-formedness, meaningfulness and truth, we need a proper theoretical framework that provides an explicit basis for decisions of appropriateness, and a theory of context should provide such a framework. This would mean that appropriateness is understood as a criterion for the social (situational and interactional) acceptability of discourse (or parts of discourse) as defined in terms of any of the properties of its context, much in the same way as the grammaticalness of an utterance is defined in terms of the rules of the grammar. However, since our notion of context is a relative construct, that is, defined in terms of the participant definitions of the communicative situation, also appropriateness is relative to the participants: A discourse is appropriate-for-the recipient if it meets the appropriateness conditions of the context as it is now being constructed by the recipient. For instance, if the recipient construes himself of superior social status with respect to the speaker, and the speaker does not signal such a relationship in his selection of, e.g., pronouns or other deference markers, then the discourse may be found inappropriate by the recipient. Obviously, such need not be the case for the speaker, who may not construe the recipient in such a way in his own context model. Such differences of context models may then give rise to an interactional conflict if the recipient chooses not to disregard such (attributed) inappropriateness, for instance because he deems that the inappropriateness is not due to an unintentional error.

This example also suggest that actual judgments of inappropriateness are generally made by recipients, assuming that speakers normally engage in talk that they would define as appropriate with respect to their own definitions of the situation — except in those situations in which they intentionally break the rules, for instance in order to insult someone, or to impose a definition of the situation that they know is not shared by the recipient. We also


see that whereas context definitions are subjective, the rules and other conventions of discourse and interaction are social and hence shared in a given community of discourse. It is also on this basis that recipients may criticize the speaker — if they assume that the speaker should have the same definition of the communicative situation as they have. Violation of the rules in that case may thus be perceived as intentional and hence be interpreted for instance as an insult or challenge of authority, as a form of carelessness, or as an unintentional error due to ignorance or lack of awareness about a property of the situation, such as the actual role or status of the recipient or the current knowledge of the recipient.

In other words, intentional inappropriateness presupposes that the speaker knows the context definition of the recipient as well as the appropriateness rules of the community, but has different aims than those conventionally associated with the utterance. Some of these may traditionally be described in terms of pragmatic implicatures, but more broadly should be characterized in terms of the relations between the context models of the participants, for instance when speakers want to impose or persuade recipients to adopt their own definition of the situation, such as the definition of the social relationship between the participants. The theory of context should make such conditions explicit, not only by formulating the possible structures of context models in some community, but also the rules that relate such structures with discourse structures.

Types of contexts and genres

As we do for discourse, also contexts may be classified as different types. In fact, as we see more often throughout this book, a theory of discourse genres is in fact often (based on) a theory of context types. We have seen that what defines the parliamentary debate as a genre is less what the participants say and how they say it, than who they are, what roles they play, what intentions they have, what political actions they engage in, and in which institution they do so.

Thus, genres, contexts, communicative events or practices can be classified in many ways, e.g., by spheres (public, private), mode (spoken, written, multimedia, etc.), main social domain (Politics, Media, Education, etc.), institution or organization (Parliament, University, shop), participant roles and relations (Doctor-Patient, Prime Minister-Members of Parliament), goals (impart or require knowledge, advice, service, etc.), or (inter)actions (decision making, governing, etc.), among other dimensions that may be taken as so many proposals for categories of a formal context schema.

Of course, at higher or lower levels further theoretical notions may be developed to make the typology, and hence the theory of context and its social embedding, more explicit. Thus, domains may


be further grouped into realms that organize collective decision making, action and control (Politics, Law, Administration, etc.), a symbolic realm for the exchange of knowledge and beliefs (Media, Education, Science, Religion, etc.), a production realm (fabrication of commodities), and a service realm (Health, Agencies, etc). At the lower level, genres, communicative events or social situations may be further classified in terms of such subtypes of actions as getting knowledge about persons (interviews, interrogations), exchanging scientific knowledge (congresses, papers, etc.), controlling people’s actions (commands, arrests, instructions, manuals, etc.), and so on.

Towards a theory of social situations

It has been emphasized that contexts are not some kind of (objective) social situation influencing text or talk, but subjective definitions of such situations, made explicit here as mental models stored in episodic memory. Thus, the theoretical focus is on how participants understand and represent the communicative situations in which they participate and how such an interpretation influences their discourse production or comprehension, respectively.

However, although we focus on the understanding the situational parameters of communicative events, obviously understanding situations, and especially our own role in them as participants, is a more general and fundamental ability of human beings. Throughout this study, I shall therefore not only examine the sociocognitive aspects of understanding communicative events, but more generally examine the structures of social situations, because obviously understanding such situations is not only a subjective process of participants, but based on socially and culturally shared knowledge about social situations in general. That is, as was traditionally done in several disciplines, we may well take social situations as a fundamental notion in the social sciences, besides such other fundamental notions as interaction or institution, among others. Thus, we shall examine what scholars in social psychology, sociology and anthropology have said about social situations, and we shall see that despite the fact that several leading scholars found this concept fundamental, there is surprisingly little theorizing on situations as such in these disciplines, compared for instance with a notion such as ‘interaction’. It is generally recognized we should study ‘situated interaction’, but the ‘situated’ aspect often receives short shrift.

Besides a general theory of context, this study may therefore also be seen as trying to provide some further elements for a general theory of social situations as a basis for such a theory of context. It is also in this way that a theory of context is integrated into broader theoretical frameworks in the social sciences.


Let us now spell out, still quite informally, what components a multidisciplinary theory of context should have.

The (socio)linguistic and discourse analytical components

In linguistics it is a standard assumption that language use and discourse vary as a function of context. A (socio)linguistic approach describes which social conditions (e.g., age, gender, class, social role, etc.) or what social situations (e.g., a parliamentary debate of a visit to one’s doctor) typically control which discourse structures and variations.

Although our theory says that such control is mediated by context models, empirical research of language use in different social situations will be necessary to be able to make assumptions about how such social situations are perceived and interpreted (or ignored) by language users.

That is, individual language users may have personal, subjective interpretations of social situations, but that does not mean that there are not shared, social definitions of social situations that count as ‘given’ or as ‘objective’ in a specific community. For instance, we may assume that all MPs in the case of Tony Blair’s speech construct the current Location and Institutional Setting as the UK Parliament or House of Commons — and anyone who doesn’t — and would say so — most likely would be treated as deviant at best, if not as a ‘loony.’ It is these socially shared, consensual constructions of social situations that are taken to be ‘objective’ for a specific group or community, a social analysis of situations will reveal what most participants most likely will construct as part of their mental models. Hence, we also need a sociolinguistic approach that (only) has focused on these ‘objective’ aspects of communicative events.

As is the case for the other disciplines mentioned here, we can only partly discuss and integrate the relevant aspects of such a — vast — sociolinguistic approach to discourse, and I shall do so by focusing on the socially conditioned discourse structures, rather than on those traditionally studied in sociolinguistics, such as sound structures, syntax and the lexicon of isolated words and sentences. The same is true for the account of style and register, also to be defined in terms of context dependent variation of discourse.

That such linguistic approaches to context are relevant is obvious from the fact that the use of the very notion of context-dependent linguistic variation presupposes we know what linguistic structures are ‘obligatory’, and hence context-free, and which ones may vary with context. Trivially, since articles precede nouns in English, and do so in all possible contexts, such structures are hardly interesting for a theory of context-discourse relationships.


We shall see (in Chapter 7) that many — if not most — most studies of context-dependent discourse variation have been carried out for the role of gender, often within a feminist paradigm. In many ways, these studies are paradigmatic for a contextual approach to the study of language use, also because few properties of context are as pervasive and influential — both in grammar as well as at other levels of discourse structures — as the gender of the participants of interaction. On the other hand, we have as yet few systematic studies of the role of class in discourse variation.

It should be repeated (again) that unlike in sociolinguistic studies our approach does not direct link these social ‘variables’ (gender, class, etc.) to variations of discourse, but only through the subjective definitions of the participants, that is, context models. This guarantees first that our account allows for personal variation and that it is non-deterministic (being a woman does not ‘cause’ women to speak ‘as a woman’).

The cognitive component

As suggested before, and as we shall see in more detail in Chapter 3, it is a well-known assumption in psychology that people conceptualize their ongoing actions, experiences and environments, and represent these as models in episodic memory. However, much of the current work on discourse processing did not explicitly introduce the notion of context in theories of production and comprehension. The notion of ‘situation model’ was used (e.g. in *Van Dijk & Kintsch, 1983), but as referring to events talked about, and hence rather as a referential-semantic than as a pragmatic notion referring to the communicative situation.

If contexts are defined as mental models, we need to be much more specific about their form, representation, construction, activation in memory, and their strategic role in discourse processing. Thus, we need to specify how context models take control over much of the discourse production and understanding processes, and how they thus influence the actual mechanisms of discourse. That is, for the selection or interpretation of genre, intonation, lexical items, syntax, pronouns, topics, presuppositions, speech acts or interactional moves, among many other discourse properties, language users permanently need to take into account the constraints of their context models and update these models. As is the case for grammar, also some aspects of discourse processing appear to be context-free, and we therefore need to carefully distinguish which aspects of production and comprehension are independent of communicative mental models (contexts) in episodic memory. Another aspect we need to attend to is how and why


participants process the ‘same’ situation differently, so as to show that context models are really unique.

Context understanding is a very complex cognitive task, which needs detailed further analysis. We should find out how the (schematic) structures of contexts in a culture are acquired and used in the real time of everyday interaction. Language users are apparently able to reduce the vast complexity of a social environment to a manageable context model in fractions of seconds. They do so many times a day, so they have become experts in the strategic construction of such models – although they of course still make mistakes. To describe such processes is a daunting task, because this requires a cognitive theory of pragmatics and language use.

Yet, it is precisely at this point where contexts – as context models – are interesting, namely in the way they are able to influence the production and comprehension of discourse. As has been repeatedly emphasized above, it is a fundamental assumption of this book and my other work that social situations cannot possibly influence discourse directly. This is only possible when social actors, as language users, interpret or construct such a social situation in a certain way, and it is this construct, this mental model, that plays a fundamental role in text processing and language use. Only through an account in terms of the cognitive interface of subjective mental models are we able to describe and explain why people often speak inappropriately, why there are conflicts of understanding, and why different people also speak differently, even in the same situation. Hence we need to say much more about contemporary mental model theory in order to understand the role of context models in processing.

Social situations of communication as subjectively represented in context models are not limited to constructions of the social characteristics of participants, e.g., their communicative or social roles, their identity or relations. It is crucial also to include in the model the ‘cognitive’ properties of the participants, such as their intentions or goals, their knowledge and maybe also their attitudes and ideologies, norms and values.

Especially relevant in this case is the role of knowledge – and the way language users represent the knowledge of the other participants. We touch upon the thorny philosophical issue of Other Minds. All discourse processing constantly needs to evaluate and model what recipients (already) know. We shall see later that we need a special theory of a Knowledge Device or K-Device to describe and explain the fundamental role of mutual knowledge and knowledge assumptions in discourse and interaction. This fragment of the theory of context alone requires a rather complex framework of cognitive, social and cultural ideas about the nature of new and


shared knowledge, knowledge communities, intercultural communication and common ground, among other notions.

Among a host of other cognitive aspects of context models, we need to pay special attention to the nature of intentions (as well as plans, goals and purposes) — one of the crucial, but controversial, notions in the theory of action and obviously part of a representation that controls interaction in general, and discourse in particular.

If context models are not merely theoretical devices, but ‘real’ mental representations, they should also have neurological counterparts in the brain. In this study, I shall have very little to say about these neurological aspects of context models. However, we may assume that if the brain structures that are the basis of mental models are damaged, for instance as a result of Alzheimer disease or accidents, we should be able to observe impaired processing of communicative situations. We shall briefly need to examine some of the relevant literature on pragmatic disorders to see which ones might be explained in terms of the neurological basis of context models.

We have seen that at the border of cognitive and computer science, scholars have increasingly paid attention to formal analyses, modeling and computer programs that simulate context, for instance in (automatic) language and discourse production and comprehension. As usual such approaches tend to be abstract and limited to a few dimensions of context (e.g., the role of time or speaker perspective in talk). Where relevant, I hope to integrate the main results of this work in our mental model approach to context, and at the same time I hope that conversely our theory may provide further support and extensions for such formal approaches.

The social psychological component

Since our general framework precisely tries to integrate cognitive and social dimensions of discourse and contexts, the theory also needs to feature components that are typically developed in social psychology. For instance, notions such as ‘situation’ or ‘episode’, needed in the definition of social situations of communicative events, have received more attention in social psychology than in other disciplines, and these proposals need to be critically examined as a basis for a theory of context.

The same is true for many other aspects of discourse and communication that are traditionally accounted for in a social-psychological perspective, such as interaction, social identity, social norms and values, shared knowledge, social representations, opinions, attitudes, ideologies, face-keeping and impression formation, and so on. That is, many of the categories traditionally proposed as components of context, such as Settings, and especially


the social identities and relations of the participants, and their shared beliefs, belong to the classical domain of social psychology.

Although the uniqueness of context models also requires accounts of general cognitive psychology, e.g., in terms of mental models in episodic memory, participants also use language as members of groups and social or cultural communities. Hence we need to account of their socially shared beliefs (knowledge, ideologies, norms, values) as well, and how these in turn provide the necessary information for the more specific, personal context models.

As a result of an arbitrary division of psychology, knowledge is typically studied in cognitive psychology, whereas opinion, attitudes, norms, values and ideologies are studied in social psychology. And since context models also feature personal opinions of participants about other participants and the current action, we also need to account for their roots in shared social attitudes and ideologies.

The sociological component

Although contexts as defined here should not be confounded with the social situation of a communicative event, this does not mean that they have nothing to do with each other. I have to stress again that my point is not to reduce social contexts to cognitive constructs, or to deny social ‘reality’ as it is experienced by social actors. Rather, I emphasize that the way such a social reality impinges on how we act or speak should be accounted for in terms of participants’ constructions or interpretations of such social reality, that is, in terms of mental models of situations or circumstances.

We have seen that these personal mental constructs of social situations are not personal idiosyncrasies or used in a social vacuum. The way people (learn to) interpret communicative events and their social environment is of course socially organized, negotiated, and subjected to social norms. Families, peer groups, media and schools are involved in the interactional and discursive acquisition of relevant context categories. Ongoing evaluations of appropriateness in everyday discourse are the typical means by which members learn to understand and construct contexts. For instance, children in many societies are being taught that when addressing “older people” they need to use more polite forms of address or pronominalization, and thus they learn that ‘age’ of participants may be one of the relevant categories of communicative situations.

This means that whatever autobiographical or situational idiosyncrasies people may have in their context models, appropriate interaction presuppose that context model schemata must be socially based. It is also in this way that people can use such shared schema


categories and well-known strategies in the frequent daily activity of interpreting discourse environments.

Indeed, the very notion of appropriateness is a social notion. Members of a group or community need to learn its conditions, and these are formulated in terms of the structures of context models — such as when, where and as what to say to whom in which way. This is so not only for the conditions of speech acts, genres, styles and other context-based properties of everyday talk, but also for special, professional, institutional or other group-based forms of interaction, as is the case for speaking in parliament, as we have seen for the speech of Tony Blair.

A specific context model is strictly unique and personal, because it is the specific interpretation of a communicative event by one person with her or his own unique biography of personal experiences. But the general structure of context models, its categories, that is, the context model schema, is socially shared and learned in a given society or culture, and the same is true for the cultural variations of such schemas.

Social sanctions apply when the context model of a language user is inappropriate, for instance when no politeness is discursively expressed in a social situation in which conventional context categories would require this, or when a specific genre of public speech (e.g. a news report or a lecture) makes use of a totally inadequate style.

In sum, we need to examine which societal constraints contribute to the social acquisition and uses of the categories of context model schemas, that is, which categories are relevant in interaction. That gender, class, age, status or power often influence how people talk, rather than, for instance, their weight or the color of their clothes, obviously also has social conditions and functions that require sociological analysis that need to be taken into account in a theory of context. We have seen that for cognitive reasons context models cannot be too complex. Since however social situations may be very complex, we need a sociological theory that helps explain why and how social members select those categories of communicative situations that are interactionally relevant.

We have informally characterized contexts as participants’ definitions of situations. This is a classical notion from sociology, and we therefore need to examine how contemporary sociology has contributed to the analysis of this notion of ‘situation’. Macrosociology typically ignores the micro level of social situations. However, microsociologies such as ethnomethodology and conversation analysis, focus on only one aspect of such situations, namely interaction, neglecting other aspects of situations, such as settings, actors, their identities and relations or their intentions, knowledge or beliefs. We shall therefore need to examine what further sociological analysis will be necessary to understand the role


of, e.g., time, place and space in interaction, as well as social roles and relations between participants.

Similarly, we need to deal with another classical debate in sociology namely how to relate macro and micro levels, structure with interaction or agency, because we have assumed that contexts typically must feature both micro and macro dimensions: Tony Blair can only appropriately speak in parliament when he locally manages the interaction with MPs and at the same time he must know and apply the general rules of parliament, the constraints of the institution and role of Prime Minister, British foreign policy and other ‘macro’ aspects of the situation of parliamentary debates as represented in the context models of all participants.

There have been many attempts in current sociology to integrate such micro (agency) and macro (structure) dimensions of the social order. Ignoring the sociocognitive account in terms of mental models and social representations, however, such has been done in traditionally vague and impressionistic notions such as the ‘consciousness’ of social actors.

Thus, much of the debate about whether or not ‘contexts’ should be involved in a theory of language use has raged in conversation analysis. As we shall see in more detail in Chapter 5, leading conversation analysts have emphasized the necessity of an autonomous approach to the study of conversation, in which accounts of social context only should be given when such context is ‘oriented to’ by the participants, and when such an orientation is also ‘procedurally consequential’ for talk. However, such criteria are quite vague and we shall therefore need to examine whether they are relevant and how they can be made more explicit.

The anthropological component

Of course the same arguments hold for the anthropological component of a theory of context. We just suggested that for each culture adequately participating in communicative events not only presupposes knowing the rules and strategies of discourse, but also the relevant context categories of such social situations.

The anthropological component of a theory of context should, among other things, provide insight into the cultural specificity and variations of context models and their categories. For instance, in one society, age or kinship of the recipient may be a more relevant category in context models than in other societies, and thus be associated with other forms and strategies of politeness, deference or forms of address. Also, since Time and Space are organized differently in different cultures, also the components of the Setting of context models may be culturally variable. Indeed, as we shall see, the very notion of ‘context’ itself may not be a folk notion in all cultures.


In Volume 2, Chapter 4, we shall therefore begin with the first scholarly definitions of context in anthropology, namely those proposed by Dell Hymes in the 1960s, and review the subsequent ethnographies of communication that examined variable social conditions of talk. Next we examine several contemporary studies, especially in linguistic anthropology, dealing with relevant dimensions of a theory of context and their application in ethnographic studies.

Context in other disciplines

The analysis of context is not limited to the disciplines of the humanities and social sciences mentioned above. Of the thousands of books that have the word ‘context’ in their title or bibliographical descriptors, many deal with other phenomena and disciplines.

Indeed, one might say that context is not just a concept or category studied in many disciplines, in each of which it has a slightly different meaning and different implications. Rather, we may speak of contextualism, that is, of a movement, perspective or kind of theory that for each discipline is contrasted with context-free, abstract, structuralist, formalist, autonomous, isolated, or other ‘introvert’ ways of studying phenomena. Thus, contextualism in many disciplines implies that phenomena must always be studied in relation to a situation or environment, as we have seen for language and discourse studies above.

Thus, in philosophy, and especially in epistemology, contextualism breaks with a theory of knowledge in terms of context-free, absolute truth in which knowledge is traditionally defined as justified true beliefs. Contextualist epistemology conceptualizes a more realistic and commonsense notion of knowledge (*Blaauw, 2005; *Brendel & Jäger, 2005: *Preyer, & Peter, 2005). It emphasizes that truth of beliefs may vary with social situations: what is true in one context, for some people, may not be true in another, so that also knowledge may contextually differ. We shall come back to this philosophical concept of context, and especially relate it with a theory of knowledge.

Closely related to the other studies in the social sciences dealing with discourse is the interest in context in the field of communication studies. Largely profiled after traditional social psychological research, such interest in context generally focuses on context dimensions as independent variables influencing communication messages, on the one hand, or on the “effects” of (mass media or persuasive) messages on people, on the other hand. There are however some publications that show a more explicit interest in the study of context in communication, such as the edited book by *Owen (1977), published by a publisher that apparently focuses on context: the Context Press in Reno, Nevada. In his


introduction James Owen especially highlights the 1942 study of Stephen Pepper on World Hypotheses: One of these world views (besides ‘mechanism’, ‘formism’, and ‘organism’) is ‘contextualism’. The root metaphor of this contextualism is the ‘historic event’ or the ‘act’ which is alive in the current setting; these events in the real world are being experienced in a novel way by each individual; the goal of the contextualist is understanding, a process that is personal and situational. As is the case for other edited studies on context, also in this book several of the other articles only have tangential relations to a theory of context but rather pursue the respective research directions of the authors. One of these studies, by Gary Cronkhite (on the cognitive representation of rhetorical situations), relevant for my own approach, will be further referred to in Chapter 3. Several of the articles in the book highlight the relation of this kind of contextualism with pragmatism in philosophy, and with the Coordinated Management of Meaning (CMM) approach in communication studies, advocated by Pearce, Cronen and associates, for whom contexts are “not found things, but […] interpretive achievements,” situated, consisting of systemic relations and elaborated and transformed in interaction (*Shailor, 1997: 97-98).

Also in biology (*Smocovitis, 1996), physics (*Kitchener, 1988), and the other sciences, there are developments that emphasize that forms of life or physical events need to be studied in their respective contexts. A more detailed study of these approaches is beyond the scope of this book, but they should be seen as an intellectual manifestation of the same kind of metatheoretical concern, namely that we better understand phenomena when we explicitly link them to their environments.

Towards an integrated theory

Much of what has been said above is not new to (at least some) scholars in the respective disciplines mentioned. Also under different names, several aspects of contexts, situations and environments, have been studied in much detail in psychology and the social sciences.

In linguistics the study of grammar has increasingly become context-sensitive, and with the introduction of the pragmatic studies of speech acts, conversational maxims and politeness, among other phenomena, contextual conditions have become the basis for the definition of appropriateness. Sociolinguistics by definition has been interested in the ways language use varies with aspects of the social situations, such as gender, age or class.

Psychologists have examined the perception, comprehension and representation of settings, objects and people for decades, lately also in terms of cognitive model theories, and also applied such insights to discourse processing. They have paid much attention to


the role of knowledge and common ground in language use, one of the objects of analysis also studied in computer science, Artificial Intelligence, formal grammar and related approaches.

Sociologists, especially those in the analysis of conversation, have not only studied interaction, but also the situations, participants, roles, and institutional settings of conduct. And anthropology has done much of the above but then focusing on how such is (variably) the case in different cultures. Indeed, the ethnography of speaking might be seen as the first direction in modern discourse studies in which whole communicative events were analyzed, and not only what people said and how they said it.

In other words, the disciplines we deal with in this book already have offered important contributions to an integrated theory of context. Yet, although most of these disciplines have been using the notion of ‘context’ for a long time, none of them provided an explicit theory of the relevant categories and structures of communicative situations and how these systematically influence language use defined as discourse processing and understanding. The prevalent social approaches ignored or denied the relevance of cognitive processing and representation of social environments of talk and text — an interface that is trivial for psychologists, even when we still do not know all details of such context models and their role in language use. On the other hand, also because of their focus on laboratory experiments, much cognitive psychology ignored the social dimensions of discourse processing and comprehension — which again is an obvious premise in the social sciences.

Thus, although much in this book is hardly new for many of the disciplines mentioned, its overall aim is to present a global framework for an integrated theory of context. It wants to bring together results from different disciplines, with a concrete proposal for the definition of context in terms of the sociocognitive construct of ‘context models’ as an interface between social situations and structures, on the one hand, and individual language users and their discourses on the other hand.

Context and Critical Discourse Studies

Within discourse studies, a contextual approach is specifically relevant also for a critical perspective on text and talk. The principal aim of critical discourse studies (CDS) has always been to study discourse as a means of power, power abuse or resistance in society (*Fairclough, 1995; *Fairclough & Wodak, 1997; *Van Dijk, 1993). Critical studies by definition feature many situational elements that have been shown to be relevant in talk and text, such as gender, ethnicity, class, power, status, social roles, knowledge and ideologies. In this respect, CDS has been one of the main movements opposed to more autonomous and formalistic approaches to


discourse. A theory of context, therefore, provides insight into the foundations of CDS. It explains how social and political conditions at various levels of social structure and interaction can be manifested, (re)produced and resisted by text and talk.

For instance, in our brief analytical comments on Tony Blair’s speech, we have assumed that he was manipulating the House of Commons into legitimating his decision to go to war in Iraq, This assumption turned out to be confirmed by the political facts as they have been revealed since the invasion of Iraq: Bush and Blair already had taken the decision to invade Iraq at least one year earlier. If we want to critically analyze a speech as an instance of manipulation, however, a mere discursive or interactional analysis will not do. To define interaction we need contextual categories, such as the intentions and the interests of the participants, so as to be able to interpret what Blair says and does as part of the political of legitimation and manipulation (*Van Dijk, 2006). For many other notions that are crucial in a critical approach to language and discourse, such as the identity and power relations of the participants, thus, a contextual approach is essential. In that sense, context theory is at the heart of critical discourse studies.

A socio-cognitive approach

It has become obvious in the preceding pages that this study will attempt to integrate different approaches from various disciplines. More specifically, I have advocated, also in my earlier work, a socio-cognitive approach that bridges the unfortunate gap between the cognitive and the social sciences in the study of discourse.

This means, first of all, that I shall be critical of psychological studies that study the ‘minds’ of individual language users while ignoring both social cognition shared by language users as group members as well as the ways such mental representations are shaped, changed and function in situated social interaction and communication.

Secondly, I shall repeatedly criticize the many contemporary approaches to discourse that limit their analysis to a ‘mindless’ study of conversational interaction. Thus, the field of discourse is vast and comprises much more than informal or institutional talk. A theory of context should obviously also apply to the production, comprehension and functions of the myriad of texts and documents that define contemporary societies. And even more crucially, the study of action and interaction itself is incomplete without a detailed analysis of the cognitive dimensions of human conduct, featuring such notions as meaning, understanding, intentions, plans, goals, knowledge, beliefs, norms, values, intersubjectivity, common ground,


and so on. Instead of taking such notions for granted, they require analysis in their own right as fundamental dimensions of interaction and discourse. Of course, one may strategically ‘bracket’ such an analysis, or leave it to psychologists, and see how far one gets by a pure analysis of action sequences themselves. In actual practice, however, such an analysis admits many ‘cognitive’ notions through the backdoor, such as recipient design, interpretation (‘hearing as’), knowledge and inferences. Any study that not just temporarily brackets a cognitive analysis, but explicitly denies its relevance or even its possibility, would strictly speaking be limiting itself to a study of ‘conduct’ that can only be qualified as a contemporary version of behaviorism — and which for that reason I shall call ‘interactionism’. Fortunately, also in conversation analysis such a radical denial of the relevance of the cognitive underpinnings of interaction are becoming increasingly rare.

A socio-cognitive approach to discourse explicitly integrates both the cognitive and social dimensions of text and talk, and is a crucial condition for an explicit theory of context. We have shown, and will see in more detail later, that on the one hand contexts are (inter)subjective mental constructs of discourse and communication participants, and that on the other hand such contexts are shaped and transformed during social interaction and within institutional, social and cultural situations whose contingent relevance they define for the participants. Mind and embodied conduct go hand in hand in situated discourse, and any approach that denies the relevance of the complementary dimension may be called reductionist. The current interdisciplinary nature of the studies of cognition and interaction no longer needs such reductionist approaches.

Major and minor myths

Part of the critical aspects of this study is to occasionally expose major and minor myths in the study of discourse and context. We are all familiar with examples of notions and ideas that became tremendously popular for some time, but on closer analysis appeared to be hardly more substantial than the emperor’s clothes of the well-known tale.

Thus, the major myth we already said we hope to expose is that contexts as traditionally conceived — namely as objective parameters of social situations (gender, class, etc.) — directly influence, or are construed by text and talk. We shall show that contexts are construed by participants, and that hence discourses shape, and are shaped by, such (inter)subjective participant constructs. Discourses do not ‘contextualize’, people do.

Secondly, we shall repeatedly question the possibility of adequate discourse and conversation analysis without an explicit context-analytical component. Instead of taking context features for


granted, they should be made explicit. Without them, an analysis of text or talk is at best fragmentary and never explanatory.

Thirdly, contexts cannot be reduced to discourse or interaction, but need to be studied in their own right, featuring both cognitive and social dimensions.

Finally, the concepts used in a theory of context should each be critically examined for their clarity, precision and relevance. Thus, we shall encounter several notions that (still) are quite popular in contemporary discourse studies in general and in the account of context in particular, but that on closer examination are simply too vague for comfort. For instance, in Chapter 2 of this volume, I shall critically examine the Systemic-Functional study of context and its often repeated components of ‘field’, ‘tenor’ and ‘mode’.

Data and observation

Discourse and conversation analyses are usually based on examples drawn from discourse corpora. But what are the data of context studies? Obviously we cannot ‘cite’ fragments of contexts. We can only postulate contexts, and do so by drawing inferences from discourse, and by applying (instantiating) our socially shared knowledge about communicative situations.

This is also one of the reasons why some scholars, for instance in ‘discursive psychology’ prefer to reduce this kind of analysis to discourse, which they define as an observable form of interaction (see, e.g., *Edwards & Potter, 1992).

This is however a problematic argument, because also discourses cannot be directly observed: in a strict empirical sense we only have sounds or marks on paper or a computer screen. Discourses are (defined in terms of) abstract structures (lexical, syntactic, semantic, narrative, etc.), or in terms of meanings or actions attributed to them by language users. Yet, we are able to cite (tokens of) text and talk, and together with their interpretations which we assign to them as language users and analysts, we are able to study their ‘underlying’ structures.

As is the case for action and interaction, this makes discourses semi-observable: that is we can observe their concrete manifestations (body movements, sounds, etc.) and interpret these in terms of the rules and units of a natural language, a semiotic system, and our knowledge about the world in general, and about action and discourse in particular.

But, what about contexts? As we have seen above, contexts defined as mental models are not visible at all. They are interpretive or abstract constructs of participants, observers or analysts. Of course, in one sense of observation, we can observe manifestations of fragments of social situations that are used as the basis for the construction of context as mental models. We ‘see’ the person with


whom we are now talking, but it’s a long inferential way between such an observation and the contextual category concepts of ‘participant’, ‘interlocutor’ or ‘recipient’. Similarly, we may ‘see’ that the person is a woman, given our knowledge of the world, and schemata about what women usually look like, and so on, but from that ‘observation’ to determining the role of ‘gender’ in talk and text it is again quite a distance. The same is true for other social roles, such as professor or friend. Setting elements such as time and place cannot be observed as such either: they are abstract notions of which only some aspects are in some sense ‘observable’, for instance the walls of a room defined as a ‘classroom’ or ‘newsroom’.

However, these observable dimensions are not very interesting at all for a theory of context, except in a cognitive theory of perception and understanding. In other words, we cannot directly observe, and hence cannot ‘cite’ context fragments as such.

What we can do, at most, is cite fragments of discourse, as we have done at the beginning of this chapter when we cited a fragment of Tony Blair’s speech in the U.K. House of Commons, and reconstruct a plausible context for it so as to describe and explain why that speech is also politically meaningful.

Our evidence for such interpretations however is not observational, but sociocognitive: these interpretations are shared by others, such as the MPs themselves, or other knowledgeable language users or analysts, and such understandings can be expressed in discourse so that they become shared.

Thus ‘reading’ of contexts is no different, basically, from readings of text and talk in discourse and conversation analysis: They presuppose members’ knowledge and interpretations. For participants and knowledgeable analysts alike, there is no fundamental difference between understanding of words, phrases or sentences and their (non observable) intended referents, on the one and, and understanding that Tony Blair is speaking at a specific moment (now) and in a specific place (here in Parliament), as the Prime Minister and as leader of Labour, among other identities, and addressing (us) MPs, also in various identities, and with specific intentions, and so on. Participants need to know what ‘the House’ refers to, and ‘British troops’, and such (deictic) expressions presuppose the same kind of sociocultural knowledge as the knowledge of the participants of the current communicative situation.

In fact, where they are not constructed or modified by text and talk, contexts are presupposed to be known and shared by participants — and hence need not be ‘observed’ at all. Moreover, discourse itself multiply signals fragments of context, which will help recipients reconstruct the specific context models of the speaker.

In sum, contexts are assessed on the basis of (a) general, shared social-cultural knowledge, (b) previous experiences (and


context models) of participants, (c) ongoing perception and understanding of relevant dimensions of the social situation and (d) indexical expressions of current, ongoing discourse, including non-verbal aspects (gestures, etc.).

Obviously, unless they are also participants, analysts seldom have the same data as the participants — they do not have access to the details of previous experiences of participants or ongoing perceptions — so they generally need to rely on shared historical or sociocultural knowledge of social events and situations combined with what is indexed in talk or text itself.

It is in this way that we are able to provide a reliable contextual analysis of the speech of Tony Blair, an interpretation that can be consensually shared with others who have such knowledge and hence become an ‘objective’ procedure for analysis. If in addition we can find manifestations of the influence of such a hypothetical context in the speech itself, then we have additional evidence that such a contextual reading is probably a correct reconstruction of the context model of the speaker.

Terminology

Above and in the rest of this book, we have used many different terms that somehow are related to ‘context’: situation, environment, and so on. Although of course terminology may be quite arbitrary and a question of taste, we need to be at least more or less consistent, in line with our theoretical perspective, so let us briefly comment on the terms we shall use in this book. I shall come back to this terminological discussion in the next chapters and especially also in Chapter 7, when we discuss the relations between context and discourse.

Situations and episodes

Above as well as throughout this book, the notion of (social) situation has been used as the more general term of contexts: informally contexts are types of social situation, namely social situations in which participants talk or write, listen or read. More specifically and theoretically, contexts are defined as subjective mental models of the way participants understand or construct the relevant characteristics of social situations. We shall see in Chapter 5 that social situations themselves are also social constructs, that is those aspects of the social environment that are relevant for interaction of social actors. In other words, contexts are special cases of social situations, namely communicative or discourse situations. More generally, social (or other) situations may be defined as fragments of possible worlds (*Barwise & Perry, 1983),


for instance defined in terms of time or place: what is going on at a specific moment at a specific time.

There is however a possible ground of confusion here. When we speak of ‘context’ in a more informal way (as in ‘text and context’), we usually refer to the (social) ‘circumstances’ of some discourse or interaction, and not to the discourse or interaction itself. However, if social or other situations are ‘fragments’ of some possible world, or the building blocks of society, we refer to complete units that include discourse or interaction, namely as their central elements.

This means that we would need an extra term to denote such a complete social situation, that is, a combination of interaction and all its relevant social conditions and consequences. One of the notions that would apply very well here is that of episode, also because it implies a spatiotemporal unit of the history of the world (like a war), or of our personal lives (like our youth, accident or interaction), of variable length. That is, an episode is a world fragment that consists of an event/interaction and the situation ‘in which’ it occurs. In that case, the notion of situation becomes less ambiguous and a better foundation for the study of context, because that aspect of an episode that ‘situates’ the event. This notion of episode has found its way in the term ‘episodic memory’, which is precisely the kind of memory where contexts are stored. Also some studies in social psychology have used the notion (see, e.g., *Forgas, 1979).

However, although the notion of ‘episode’ would be more adequate and allow us to define ‘situations’ in a more precise way, such a use of the term ‘situation’ will probably generate more problems than it solves. I shall therefore maintain the term ‘situation’ to refer to what I would call episodes: fragments of social reality of various sizes. This means that events, actions and hence discourse are parts of these situations, and if contexts are mental models of relevant aspects of such social situations, also text or talk are part of the context. This may seem strange, and inconsistent with the way we talk about the influence of context on discourse, but this is actually how we want it, because participants not only represent the conditions and consequences (the social ‘environment’) of their discourse, but also, reflexively the ongoing discourse itself. That is, they are (partially) aware of and reflexively monitor what they are now doing and saying, and how they do so, and hence must also represent this, namely as part of their context models.

It follows that if we speak of the relation between discourse structures and contexts, this will in practice often mean the way participants represent the rest of the social situation. Also, as we shall see, the ‘past parts’ of the ongoing discourse may thus become part of this context, for instance as part of the (new) knowledge now shared by the participants.


So, although we would need to introduce different terms for these different aspects of the situation, we shall maintain the general notions of situation and context, and where necessary use special words to refer to their parts or fragments, such as conditions and consequences, or the informal term ‘environment’. Obviously, these are merely terms: They still need to be founded on an explicit theory of such notions — about their general nature, boundaries, contents, structure and so on.

Apart from the terms situation and context, there are many others than are being used to refer to them or their parts, such as communicative event, encounter, activity, social practice, and so on. In the following chapters we occasionally shall use some of these terms, but the fundamental notion if that of (social) situation and its discursive counterpart: context. And we shall see that the informal notion corresponding to that of both situations and contexts is that of experience¸ that is the way people live social and communicative situations as part of their everyday lives.

Example: ‘Iraq’ or the discourses of war and peace

The main objective of this book is theoretical: to provide a multidisciplinary account of the notion of context within a broader theory of discourse. However, as has become obvious at the start of this chapter, it is quite useful to argue on the basis of examples: we cannot ‘cite’ contexts, and since contexts by definition come with ‘texts’ analyzing them makes sense only when we provide examples of text and talk, both as an illustration of the theory, and as an empirical warrant. We shall see below that in several directions of discourse and conversation analysis, contexts are only accounted for when they somehow ‘show’ in text or talk – if only to make sure that ‘contexts’ do not grow out of proportion and we need a theory of everything to describe them.

Throughout this book; I shall therefore make use of the example given at the beginning of this chapter, and later (in Chapter 8) analyze other fragments of the same debate about “Iraq.” In line with the broader framework of Critical Discourse Analysis (CDA), this analysis obviously also features a more critical approach to the kind of power abuse and manipulation engaged in by such leaders as Bush, Blair and Aznar.

“Iraq” here stands for a complex of themes organizing discourses about the war in Iraq following the invasion of Iraq by the U.S. army and its allies in March 2003 so as to oust the regime of Saddam Hussein and gain control over this crucial oil producing country in the Middle East. These discourses followed the devastating attack against the Twin Towers and the Pentagon on September 11, 2001, by members of El Qaeda, generally described as a ‘terrorist’ organization, although there have been more or less


public discourses about invading Iraq already since the Gulf War of 1991.

Apart from the discourses about “Iraq” by Bush, Blair and Aznar, a vast number of other, also alternative and opposed, discourses became part of the public domain, largely through the mass media, internet as well as meetings and demonstrations worldwide. These discourses need much further and more detailed analysis than we can offer with a few examples in this book, and need specialized study in many future monographs, for instance about the rhetoric, argumentations, and many other features of this kind of political text and talk. In this book, then, I shall limit myself to the ‘contextual’ analysis of one genre of this vast corpus, namely a parliamentary debate.

The organization of this book

There are many ways to organize a theoretical investigation of a complex notion such as ‘context’. It would have been ideal to do so according to the kinds of levels, dimensions or types of structures of context, e.g., beginning with time-space coordinates, types of participants and their properties, relations and cognitions, and so on.

However, more practically, I have followed a ‘disciplinary’ organization, by examining the actual and possible contributions of the humanities and the social sciences, beginning with linguistics in the next chapter, followed by cognitive and social psychology, sociology and anthropology. We thus are able to focus on the linguistic, cognitive, socio-cognitive, interactional, societal and cultural dimensions of context.

In Chapter 7 we then get back to the more specific relations between contexts, thus defined in various disciplines, and the structures of language use, and especially those of discourse. This will involve the integration of many results of sociolinguistic studies, for instance on the role of gender in talk and text. Obviously, as suggested before, one study cannot possibly review and integrate everything that has been written in several disciplines on the various aspects of social situations and how they impinge on language use. In that respect also the ‘sociolinguistic’ study in Chapter 7 needs to be quite limited.

Finally, Chapter 8 provides detailed analyses of fragments the parliamentary debate about “Iraq” in the UK House of Commons featuring — and testing— at least some of the notions that have been theoretically examined in the earlier chapter.

Prospects and limitations

The theoretical analysis of context as a fundamental notion in the humanities and the social sciences, as proposed in the respective


chapters of this book, is as complex as the very fundamental notions mentioned above and to be further analyzed in the following chapters. A single monograph and a single author cannot possibly examine all relevant aspects of the structures and functions of contexts in communicative events. In that respect, the following chapters only offer fragments of a future theory. By doing so from the perspective of different disciplines, I try to emphasize the multidisciplinary nature of such an enterprise.

I especially value the necessity of theoretical integration, of relating discourse analysis with cognitive, interactional and (other) social analyses. I shall show how social structures and situations influence interaction and discourse through specific kinds of mental models – contexts. We already are used to deal with language use as ‘text’ or ‘talk’ in terms of interaction or social practices. The next step is to make explicit the relations between such practices and the others aspects of the communicative events, such as social actors, their actions and minds, as well as the societal basis and background of these events. Thus, a theory of context is intended also as a contribution to the urgently necessary integration of the linguistic, cognitive, interactional, societal, political and cultural approaches to discourse. Obviously, as emphasized before, we need many more empirical studies, of many more genres, in order to be able to assess all relevant contextual categories, norms and rules for each culture. These will have to be supplied in further work.


Documents

Context · Web viewThe notion of ‘situation model’ was used (e.g. in *Van Dijk & Kintsch, 1983), but as referring to events talked about, and hence rather as a referential-semantic