Unit 1 - Introduction: Text, Context and Schema

To help people use discourse more effectively, we must first understand its normal uses.
Robert de Beaugrande 1997: 1

The purpose of this Unit is to provide an introduction to the topic of written discourse analysis and to get you thinking about some of the key issues involved. I hope that by the end of the Unit, you will have had some new thoughts about:

• what text is and why it is worth our while to study it
• what is meant by the authenticity of a text and why it is important
• how and why texts can be interfered with
• what a corpus is and how big it needs to be
• what context is and how it is important in relation to text
• who Firth and Malinowski were and why they were important (but this last I
will require you to find out yourself)
• what a schema is
• how inference and background knowledge contribute to coherence

Since these issues are central to the whole module, there is no suggestion that by the end of this Unit you will have discovered everything you need to know about these interesting questions. They will continue to be addressed throughout the module.

Why study written discourse analysis?
Or - Monks, Magicians, Martians and Big Macs
It would be a task of mammoth proportions to list all the ways in which language plays a part in the day-to-day life of a society or indeed of any individual in that society. Only a hermit bereft of all printed matter and entirely lacking artificial means of communication and recording - telephone, radio, TV, computer, tape recorder, and so on - could be expected to make a nil return. And even such lonely souls probably talk to themselves or to some Higher Being. Trappist monks, who have taken a vow of silence, read their breviaries and Bibles and make notes, and some probably write grocery lists and orders for religious books. For those of us who live in more interactive communities the range of linguistic activity is enormous. In a British television advertisement for a miracle language-teaching course, a popular entertainer - a magician or illusionist, appropriately enough - said something like: 'Language is not very complicated really. It's just a lot of words. So to learn a language all you need to do is to learn a lot of new words every day. And we have a method for enabling you to do this.' I have no idea how well this represented the course he was selling, but it didn’t inspire me to sign up, even though I would have welcomed a simple route to a fluent command of Spanish or Italian.

Of course, in a sense it is true that language is 'just a lot of words', and you can get a long way in a foreign country with a set of vocabulary items plus an array of gestures and a lot of good will on both sides. But the sort of communication that this restricts you to falls far short of the optimal. When a basic knowledge of the grammar of the target language is added to the vocabulary store, the situation is very different. Instead of communication at a level which is little better than that of gesture, you can attempt to express quite complex ideas with some degree of success. (See Dave Willis’s The Lexical Syllabus to see how far one academic thinks you can go with a concentration on lexis.)

In fact, the separation of grammar from vocabulary is a great over-simplification and possibly dangerously misleading. Grammatical regularities do not exist independently of words, but rather within words and in the relationship between words. Grammar regulates how we construct words and how we link them together in hierarchical combinations to express quite complex thoughts which are way beyond the mere naming of objects. Knowing how to use a word in a given language means knowing, amongst many other things, its grammar, which forms it can take, which structures it can occur in, and which other words and structures it can co-occur with. For this reason, following the practice of Michael Halliday, I will from now on speak of lexico-grammar rather than implying that grammar and vocabulary are independent separable components of language.

So would the course salesman have been more correct if he had said 'just a lot of words and some grammar'? Again, to some degree, yes. But is 'knowing a language' really simply a matter of knowing the lexico-grammar? Only in the most reductionist sense of the expression 'knowing a language', and only in a sense which is not particularly useful for learners or teachers of English.

David Bowie once memorably asked in song, 'Is there life on Mars?' Suppose the answer is: ‘Yes, there are life forms on Mars. They have no knowledge of human modes of communication, but by some alien means they have been able to internalise the grammar, including semantics (meaning), phonology and vocabulary of English’. Suppose they are then able to take on human form and turn up in an English-language environment. Although they would be able to produce perfectly grammatical English sentences, they would not get very far without being spotted. Comedy science-fiction has played on these truths in a somewhat random way; one example is the American TV series, Third Rock from the Sun, where the humour (?humor) largely resides in the fact that the alien beings constantly try to imitate Earth behaviour and usually succeed only in bewildering the real Earth people.

Why would these imagined Martians get caught out? Because, of course, knowing how to use English involves far more than knowing what constitutes a grammatical sentence in English. To get away with its1 nefarious schemes, our Martian would need to use the grammatical sentences in a manner appropriate to the circumstances. It would need to be able to put its sentences together in such a way that they seemed to an Earthling speaker/ writer of English to 'make sense'. Having no knowledge of how speakers of English put sentences together in a coherent way, or what linguistic devices they use to signal relations between sentences, it would produce sequences of apparently unrelated sentences. And having no feeling for stylistic variation, it might order hamburgers in language more suited to a business letter. It might conceivably say to the attendant at Macdonald's:

'Dear Sir, Thanking you for the prompt delivery of our previous order of today's date of a strawberry milk shake and a cappuccino, we wish to request an order of two Big Macs and fries without ketchup. We remain, Yours faithfully.'2 (constructed)

Suffice it to say that this kind of inappropriate behaviour would lead to prompt unmasking or, possibly, physical assault. Here I end this brief detour into science fantasy.

Being human, your students have all kinds of advantages over alien life forms when it comes to functioning in English. But they also have to do more than simply learn to produce and understand grammatical sentences. They need to be able to produce and understand text. And they need to be able to produce and understand text that is appropriate to the particular situation in which they find themselves. Not all the knowledge (beyond the lexico-grammar) that is required can be carried over from one language to another. French texts differ from English ones in than just grammar and vocabulary, and Japanese texts differ even more. Therefore, as a teacher, you need to know a great deal about the characteristics of English texts, and more specifically about the kinds of texts that figure - or will figure in the future - in your students' lives.

Linguists (and this category, too, now embraces you) need to study text because a text is a manifestation of language. The totality of texts constitutes language in the same way that the totality of human beings constitutes humanity. We might be sceptical of the claims of a model of plant biology that had no place for considering those plants which actually occur and we should be similarly dubious about any linguistic theory that has no place for the consideration of real instances of language usage.

De Beaugrande (1997) starts with a very ambitious statement about text and discourse analysis. Just before the statement already quoted at the head of this Unit, he writes:

The top goal of the science of text and discourse proposed here is to support the freedom of access to knowledge and society through discourse. This goal has become enormously urgent in our 'modernizing' world, where social progress demands that the increasingly diverse social classes and cultures develop more co-operative practices for sharing knowledge and negotiating social roles; and discourse must surely be our central modality for doing so. de Beaugrande 1997: 1

This is a commendable goal and I hope that we will be able to reach it. As teachers, we have a duty to initiate our students in the discourse practices of our disciplines. For language teachers, this is a considerable and complex task and, as de Beaugrande points out, before we can help others, we must ourselves understand what is going on.

What is text?

Text is something that happens, in the form of talking or writing, listening or reading. When we analyse it, we analyse the product of this process, and the term 'text' is usually taken as referring to the product... Halliday 1994: 311

In lay usage (i.e. non-specialist usage), the term text is generally applied exclusively to written material and sometimes more specifically to a course book, for example: a teacher might ask her students to bring their 'texts' to the next lesson. However, when we talk about text as linguists, we are using it with a much broader meaning. For us, and henceforth that includes you, dear Participant, text means any stretch of language in use on which we choose to focus; it can be of any length and spoken or written. In this sense, the huge novel War and Peace is a text. Milton's' sonnet On his Blindness is a text. Willis and Willis's Challenge and Change is a text. What you are reading now is a text. But so is a bill, a receipt, an advertisement, a road sign saying Halt, a note on a door reading Closed. A university lecture is a text, as is the verbal exchange that takes place when you buy something, or the exchange of greetings: 'Hello, there!' 'Hi! or a single cry of Help!

We may speak of a complete text to refer to the whole of the language event (for example, a whole research paper, an entire letter, an entire book, a complete lecture); or we may speak of a text fragment (a paragraph from a book, five minutes of a one hour lecture, and so on). But the distinction between a text and a text-fragment is not very precise, and often the simple term text is applied to any piece of actual language regardless of its completeness.

Further, the term text may be applied to the ongoing discourse process (the sales transaction as it occurs, the lecture as it is being given, etc.) or to a written or electronic record of the event (a transcript or a tape-recording of the lecture).


There is considerable variation in how terms such as text and discourse are used in linguistics. Sometimes the terminological variation signals important conceptual distinctions, but often it does not and terminological debates are usually of little interest. Stubbs 1996: 4

Some writers make a distinction between text and discourse and some don't; unfortunately those who do aren’t always in agreement about what the distinction actually is. I myself usually use the term discourse when I am speaking about the communicative process and text when I am talking about the product. However, it has to be accepted that terminology in general is not yet very fixed in our field, and so some degree of uncertainty is just something we have to learn to live with.

Authentic text

By text, I mean an instance of language in use, either spoken or written: a piece of language behaviour which has occurred naturally, without the intervention of the linguist. This excludes examples of language that have been invented by a linguist merely to illustrate a point in linguistic theory. Stubbs 1996: 4

The description of text given so far presupposes authenticity. In other words, we normally expect a text to be authentic, that it was originally produced as part of a piece of communication and not invented for some exemplificatory purpose. Thus, if I wanted to give you an example of a set of minutes, I could use either of the following:

(i) a text which was produced as a record of a meeting which actually occurred.

(ii) a text which I wrote myself based on imaginary events, but drawing on my experience of many hours spent writing and/or reading real minutes.

I would class the first of these as authentic and the second as inauthentic (or invented or constructed or artificial or simulated, etc.). In the unlikely event of my choosing the second option, I would feel obliged to state clearly that this was constructed, artificial data that I had concocted myself rather than a real set of minutes. I would also be very wary of making any generalisations on the basis of this second type of construct since it is not a real instance of what it purports to be. Of course, my artificial minutes might successfully simulate the real thing, but I could not be sure of this, and I would prefer to use an authentic set.

Just pause for a moment, take any textbook you are use for teaching and choose any ten texts - what proportion of them are authentic according to my definition?

In fact, teachers, materials writers and others are often tempted to use artificial data for understandable reasons. They might, for example, feel that their students lack the necessary linguistic skills to tackle the real thing and so they offer something simpler or simplified. They might believe that this serves their pedagogic purposes, but it is a risky strategy. Risky, because it is very difficult to simulate real text; one risk you run is of teaching an artificial, fake English. If you want your students one day to read or write real minutes, then expose them to real minutes. If you want them to read or write real history books then expose them to text from real history books. If you want them to listen to physics lectures, expose them to data from real physics lectures. It goes without saying that it is equally true that analysts must look at real texts and not concoct something for themselves or use the artificial concoctions of others (unless they have very relevant reason to do so, such as investigating the degree of resemblance and deviation of concoctions from the real thing).

Some academics, notably Widdowson, have tried to justify their own dubious practice by arguing that an authentic text is no longer authentic once it has been taken out of its original environment and presented in a classroom for a different, pedagogic purpose. So, they then argue, as there is no such thing as authenticity in the classroom or in teaching materials, in the sense in which I have been using it, we might as well write our own texts for the classroom. These would then have a different kind of authenticity conferred on them by the fact of being language learning texts. Thus, they argue, any sample of language that serves a useful purpose is authentic in this sense. Widdowson implemented these views by editing a series of ESP books, the Focus series, which, critics have argued fail because of the lack of commitment to authentic text - or rather because of a commitment to an idiosyncratic notion of authenticity.

My own view is that, although there is some truth in the claim that taking a text out of its original setting changes its status, there is still a crucial distinction between such texts and an artificially contrived text. This difference is frequently evident in the fabric of the text itself: the language of a simulated text - the lexicogrammar, the patterns of discourse, and so on - is seldom like that of the real thing. So our insistence on using texts that were originally produced in an actual interactive event (be it written or spoken) is a practical consideration rather than some fanciful preoccupation with the notion of authenticity itself.

Tampering with texts
There are various ways in which educators, publishers and others may try to make written text more readily accessible for student readers. And there are other reasons for changing text, too. First of all, they may select texts that are intrinsically easy to read - or rather that are at a level of difficulty with which a given set of students can cope without undue puzzlement. If the texts are appropriate to the needs and interests of the students, this is arguably the optimal situation.

There are various methods for measuring the so-called readability of texts, which attempt to identify the relative difficulty in terms of the reading age norms of native speaker/readers, for example. Texts can be graded according to the normal reading age at which they can be comprehended, and reading schemes exploit these methods by offering progressively more difficult texts in the form of books or cards. Such readability measures are often applied to specially written or doctored texts as discussed below.

A second way is to write texts from scratch that conform to predetermined lexicogrammatical constraints. We can call these controlled texts. People who write books for children usually work on fairly loose intuitive lines to produce language that they feel children of the target group will find accessible. But the huge world-wide market for English as a second language has led many publishers to pursue a policy of setting explicit linguistic criteria for newly written books, readers as they are confusingly called. The editors of these books may specify a particular set of vocabulary, certain grammatical structures and other criteria for controlling the degree of sentence complexity. Writers must then work within these constraints. Do you use, or have you ever used, such readers?

A third option is the simplified text, the result of rewriting an existing text according to similar constraints. This is a very popular option for publishers who, for example, produce as part of a series of graded readers simplified versions of classic novels such as Tom Sawyer or Robinson Crusoe. In effect, these versions are a retelling of the same basic plot, usually much more briefly, as well as in a simpler linguistic form, than the original.

A fourth option is the abridged text, a text that has been changed only by removing parts. In other words only part of the original text remains, but what is left is still in the form in which it was originally written. Thus the language remains totally authentic, but, having lost some of its linguistic context, it will paradoxically not be exactly as it was when it was produced originally as a complete authentic text.

Task 1.1
Take something you have written previously, perhaps an assignment. Take a 4 paragraph section from somewhere near the middle. This is by definition an extract from an authentic text. Now, imagine you were going to try to communicate just that content to an audience – how would you want to change the textual encoding? You would most likely want to make changes at least to the opening and closing sentences.

Obviously, texts can be abridged in varying degrees. There may be many reasons for wanting to make a text shorter: economy of production costs, physical convenience, limitations of space. Or the motivation may be to make the text more easily processible. Because of this last aim, abridgement can be seen to have something in common with vocabulary and structure control and with simplification. As I indicated above, simplified texts too are often much shorter than the original, but the term abridged is usually reserved for texts that have simply been cut.

When a text has been altered simply to remove taboo words and concepts, it is said to have been bowdlerized. Thomas Bowdler (1745-1825) was a Scottish medical doctor who published a 'family version' of Shakespeare's plays with all the ‘naughty’ bits cut out. Such texts are sometimes described as censored since it is official or self-appointed censors who impose such alterations. British TV broadcasts of films are sometimes advertised as 'edited for strong language and nudity'. That is a form of bowdlerization - a rather special case of the role of context of culture in text. It is curious that non-linguists refer to taboo words as strong language, and sometimes as just language; less surprising, perhaps, is the term bad language.

By definition, all these forms of simplifying or modifying produce something that is different from the authentic original. For most purposes in discourse analysis and teaching, and perhaps most obviously in English for Specific Purposes, as I have already said, authentic texts are preferable to those that have been interfered with.

It is possible, of course, for a re-telling to be a valued text in its own right. In the field of literature, Charles & Mary Lamb's Tales from Shakespeare is a work of children's literature in its own right. But it is not Shakespeare. Shakespeare himself borrowed most of his plots, although he did for the most part greatly improve on the originals, but the greatness of Shakespeare's work is not in the plots but in the language. And you would not teach people to write history essays by giving them Shakespeare's history plays to read. In any case, this kind of re-telling of stories is part and parcel of the narrative tradition and is a far cry from the kind of tinkering that we have doubts about. Similarly, a book such as Bloor and Bloor's The Functional Analysis of English is not a doctored version of any of Halliday's texts, but a new text expounding Halliday's ideas from a particular point of view - their own and their students'.

There is nothing intrinsically wrong with simplified readers or other alternative versions of stories (or ideas), but there is a pedagogic risk attached. To repeat: this is that students may be exposed to an artificial EFL variety of English and be shielded from the kind of language that they need or will need later. The advocacy of authenticity is not a religious dogma; it is based on common-sense about what students need to learn and what claims researchers are justified in making.

Sometimes authentic texts can be modified without seriously affecting their authenticity. For example in the sample of legal text given in the next Unit, I have changed the names of people and also the street and town names, to prevent identification. In doing so, I have not seriously undermined the authenticity of the text for the purposes of analysis, although, in a real legal context, changing the names in such a document might be a heinous crime. Such ethical issues are less liable to arise (though they are not absolutely ruled out) when dealing with texts which are already in the public domain: published articles, books, advertisements, and so on); but here the question of legal copyright may be an issue.

What is more rarely justifiable is adjustment to fit the convenience of the example. Even here, though, there may be occasions where a case can be made for emendation - so long as it is not done surreptitiously.

Data and corpus
For discourse analysts, texts constitute potential data. Data are the phenomena under investigation or the phenomena that provide evidence for the claims that the analyst makes. Thus, the research process in which the analyst is engaged is the investigation of texts. Usually, this involves the putting together of a collection or corpus of texts from which the data are to be selected.

Question: Is size important?
At the present time, the use of the term corpus analysis or corpus linguistics usually implies computational analysis of some kind, but corpora were used in linguistic analysis long before electronic computers, and a corpus can be a very small sample of text, which could be easily analysed manually. Some corpus linguists set great store by the size of their corpus, arguing that only a collection of many millions of words can provide a valid basis for useful generalisations about language use, but the fact is that the necessary size of a corpus depends on the type of investigation being carried out; that is to say, it is a question of what you are looking for. If you are interested in ‘of’, which is the second most frequent word in English and makes up roughly 3% of all the words in a given text, you need a very much smaller collection of texts than if you want to investigate ‘man’ which despite being the 150th most frequent word occurs roughly once in every 2000 words, or ‘presumption’ which occurs roughly three times per million words. Of course, it might be rash to make sweeping generalisations on the basis of a small sample, but it is also true that not all the questions we ask about texts can be answered by electronic surveys of huge corpora or by the use of statistical procedures. Computational methods have allowed significant developments in the study of language, but some kinds of truth can be better observed through a local analysis and some questions require judgements that computers cannot make.

One reason for using massive corpora is the desire to make generalisations about the English language as a whole. For example, a major advocate of huge corpora and a key figure in the creation of the massive Bank of English corpus in Birmingham, Sinclair (1991) offers interesting observations about the word of in English on the basis of its frequency of occurrence in certain grammatical structure types. If Sinclair had examined only a few thousand words, we might say: 'Come on, John! How do we know that this is typical of English in general? Perhaps another few thousand words might give a different result.' Even with a million words, we might say something of the sort, and so Sinclair opts for a multi-million word corpus.
Similarly, to make generalisations about 'English', we need to have a corpus that represents an enormous range of text types, varying in subject matter, purpose of production, degree of formality, context, etc. - in fact, as many different varieties of English as we can lay our hands on. Hence, Sinclair uses a corpus that includes as wide a range of sources as possible. But other goals and other circumstances might lead to differently structured corpora.

Task 1.2
Suppose you have been asked to teach the writing of minutes in English to a group of trainee secretaries in a foreign branch of an English company. Time is short. Would you base your teaching on an analysis of:

(a) the entire multi-million word Cobuild corpus
(b) a large corpus of minutes from various sources in the English-speaking world
(c) a corpus of minutes produced in the company in question

Note: This is based on an actual teaching situation I was told about.

If your answer was (c), I agree with you. If your answer was (a) or (b), I wonder how you reached your conclusion. I think (a) is far less plausible than (b), although, of course, there is scope for a lot of 'ifs' and 'buts' here. You might, for example, argue that you have a moral duty to inform your students of a wide variety of ways of writing minutes in case they want to work for another firm some day; therefore you chose (b). I can sympathise with that argument. Or perhaps you wished to teach more than just minute-writing and so chose (a) - and to hell with the capitalists who are footing the bill. Again, this has a certain appeal but there are a few practical drawbacks. Do you wish to be employed again? Will the secretaries lose their jobs? I think that, by and large, the relation between your hypothetical goals and the choices are relatively clear.

I concede that the choice here is not exactly a question of the size of the corpus, but rather of its focus, but it has a bearing on the issue of size since macro-corpora are less likely to be highly focused. And, in any case, the point has been made that various types of corpora are suited to various purposes. And to go back to a previous point, I am sure that whatever you did, you would not wish to use simplified minutes with half the features of real minutes removed.

If we are interested in the nature of some particular variety, say courtroom discourse or medical research articles, then we would be well advised to focus on a corpus of such items rather than a sweeping selection of entirely unrelated data. It may well be that the results we get from a comprehensive collection of text will be different from those obtained from a carefully selected set. We might, of course, sometimes wish to compare our variety-specific results with more general ones, and then access to a large corpus will be necessary, but the primary interest will be in the narrow corpus. For most pedagogic purposes, it is the narrow corpus that is the most enlightening; certainly for people engaged in ESP. Of course, if you have access to a macro-corpus that allows selection on the basis of text type, then you can extract the set of data relevant for your purposes and ignore the rest, but in this case your corpus is the specialised section that you have selected and not the macro-corpus itself. A useful small corpus may, for example, consist of a mere dozen or so articles, or abstracts, or subject textbooks, or business letters, or transcripts of lessons.

So, the short answer to the question: 'Is size important?' is 'Not always. It depends what you are trying to do.' However, in spite of all this, the term corpus does sound a little grand and for some people does connote considerable bulk, so you might be wise to avoid using it in public if you have analysed only five business letters - even though they do technically constitute a corpus. I myself have used the term to refer to a dozen or so articles, but I am aware of the risk I run of being criticised by size-fixated corpus linguists. Of course, I believe that what I am doing in looking at a few articles in the way I do is as valid a way of doing discourse analysis as carrying out a computer study of a multi-million word corpus. Not better but as good.

Even when a computer is an appropriate tool for text analysis (and indisputably it very often is), it is the questions that the analyst asks and the quality of the deductions drawn from the results of the analysis, that determine the value of the investigation. In the words of the old computational proverb: 'Garbage in; garbage out.'

I hope that I have already more than hinted at the fact that text is not created in a vacuum. It is created - indeed it is part of - a context. The notion of context is central to the study of discourse. People sometimes complain that a given utterance attributed to them (by the Press, for example, or in a court of law) was misinterpreted, because it was 'taken out of context'. By this they may mean one of two things: (a) that the rest of what they said has been ignored or (b) that the circumstances in which the utterance was made and all the paraphernalia of presuppositions, etc., have been ignored. In either case, the complainant is appealing to the indisputable view that the sense of an utterance is not inherent in the words and grammar alone, but is crucially affected by contextual factors. Context in the first sense we can call co-text; the second can be labelled context of situation. A major aspect of context of situation is sometimes labelled context of culture. Some people treat this as separate from the context of situation, but it seems to make more sense to see it as an integral part of it.

At the micro-level, a stretch of language under consideration can be seen to fit into the context of its surrounding text. The surrounding text is the co-text. The sense of a chunk of language - a few words or a paragraph - is in part dependent on words and paragraphs around it; these constitute the co-text of the chunk in focus. The co-text of the Unit you are now reading is made up of the other Units comprising this module. Some of the meaning of this Unit is inherent in its positioning as part of the module as a whole, on the fact that is the first of a series of such units, that they resemble it in format, and so on.

The context of situation is made up of all the phenomena which affect the discourse. In face-to-face interaction, the context of situation includes the immediate and wider environment in which the text actually occurs, like the classroom in the case of a teaching discourse, the shop or market in a sales transaction, the workshop in the case of a discussion about a gearbox replacement.

It may be that the physical setting of the discourse is not germane to the nature of the text itself. If you discuss gearbox replacement while on top of a mountain, the precise fact of the altitude may have little bearing on the discourse (on the other hand, it might), but the fact that there is no engine present is likely to be very significant. In addition to the physical location, there is the location in time of the event: time in history, time of the year, time of day may all play a determining role.

The interactants also play a part in the context of situation. The people who are discussing gearbox replacement, their ages, nationalities, gender and especially their social roles on this occasion (for example, mechanic and car-owner; apprentice mechanic and skilled mechanic; teacher and student; two non-expert car-owners; friends or strangers) may all be significant. In the case of written text the situation is more complex as the writer writes for an imagined reader to whom s/he attributes certain knowledge and certain ignorance, but the text is processed only by real readers who may differ considerably from the imagined and may have more or less difficulty understanding the text. As I begin this module I am assuming certain things and telling you others, but I am conscious that for some readers I may be telling you what you already know and/or assuming things of which you are ignorant, as in the throw-away reference to the ‘paraphernalia of presuppositions’ a short time ago.

Every immediate situation is located in a cultural context. The context of culture is an intricate complex of various social phenomena involving historical and geographical settings but also more general aspects like the field of the activity: education, medicine, provision of goods and services in exchange for money. Car maintenance discourse in a highly hierarchical society may be different from that which takes place in a relatively egalitarian society. Classroom discourse takes place within a wider cultural context of, say, university education or secondary school education, or slightly more specifically African university education, or Kenyan University education. The discipline in question also plays a part in the context of culture: thus a physics lecture takes place within the cultural practices and traditions of the field of physics at large as well as in a particular education system or institution.

Much of the credit for the emphasis on the role of context in language can be attributed to two significant figures in the history of linguistics: Firth and Malinowski. Rather than repeat facts which I have written up elsewhere, I will ask you to read Bloor and Bloor 2004, 244-246 (1995, 248-50), now, before proceeding to the next section.

De Beaugrande (1997) posits a set of criteria for textuality, well known from earlier publications, including De Beaugrande and Dressler (1981). These are listed in the box below with my own paraphrase of his explanation:

cohesion: the relation between forms and patterns
coherence: the way meanings are understood
intentionality: what text producers intend, mean to achieve
informativity: the extent to which the text tells you what you don't already know
situationality: the relation between the text-event and the situation in which it occurs
intertextuality: the relation between this text and other texts

In this module, we shall not be giving all these issues equal attention or necessarily discussing them under these headings, but they will all figure to a greater or lesser degree in the ensuing units.

Optimally, we might ask two things of a contextual model of variation in discourse: (a) that given a text, we should be able to say something about the context of situation that produced it; and (b) that given a context of situation, we should be able to predict the type of text which it generates.

I think that there is little doubt that we can meet the first criterion with a reasonable degree of satisfaction. I have rather cautiously said 'say something about', but I think that in the vast majority of cases, we can say a great deal. This is not to say that we can always, without exception, state precisely the circumstances which produced the text. No test of a theory could ask that. But with the right expertise, we should be able to deduce a great deal of information about the context of situation from the lexicogrammatical form and other relevant features (for example, in the case of written text: layout, accompanying illustrations, and so on; in the case of a sound recording: intonation, timing, presence of echo, and so on).

It is matter for debate how far we can predict from a knowledge of the situation the type of language that will ensue. However, it might be argued that, if we knew enough about how discourse works, we would be able to predict with reasonable statistical success. The notion of context of situation is bound up with the notion of social institutions. All forms of social activity can be seen as in a sense institutional. Weddings, funerals, trials are obviously institutional activities, but for the sociologist, ethnographer or discourse analyst, so are university lectures, buying and selling, making a will, banking transactions, writing minutes, joke-telling, eating in a restaurant, writing a letter of complaint.

In every situation, there is scope for variation from the norm and for idiosyncratic behaviour, and the degree of detail that can be predicted varies from situation to situation: the more obviously 'institutional' the situation, by and large, the more predictable the language. For example, for a given narrowly-specified society, we can reasonably predict the sort of text produced in a marriage ceremony. Given the context of a Catholic church in Britain, say, or a mosque, or a registry office, we might even be able to specify much of what will be uttered because weddings make considerable use of prescribed and ritualised ‘written-to-be-spoken language. More generally, we will expect the discourse to have certain characteristics: predominantly spoken channel, initiation of verbal exchanges by the presiding official (priest, Imam, local government official), responses from the marrying couple, some declarations of intent/promises, and so on.

These matters are pursued in more detail in subsequent units.

Task 1.3
What can you deduce about the context of situation of the following texts?

1 Warning. Customers are advised that videoscan closed circuit television
is in operation with video-recording.

2 I'll need a 19 gauge needle, IV tubing and a preptic swab.

3 SKYE lounger. Grey lacquered tubular steel frame.
Light brown buffalo leather cover padded with foam and fibres.
W68, L160, H98 cm, seat height 35 cm £295
Take it home today. See page 218 for details.

4 E 180
This VHS videocassette is
designed exclusively for use
with video recorders bearing
the VHS mark.


5 Williams cleared over Senna death

(Note: the physical format of these texts is not identical to the original but is approximate)

Coherence, inference and schemata

Coherence is no joke

Flaherty is in the bar. O'Reilly says to him, 'Pat, your glass is empty. Would you like another?' Flaherty replies: 'And why would I be wanting two empty glasses?'

There are thousands of jokes based on the idea of misunderstanding ambiguous or vague utterances. The above is one of them. (The sub-class of joke is the Irish/Kerryman/Polish/ policeman/Portuguese/Sikh/Swedish/etc. joke, which can be found worldwide.) It didn't have a lot of life in it, but let's kill it completely. The joke lies in Flaherty's failure/refusal to pick up the obvious meaning of the question, namely whether he would like another drink - another glass, not another empty glass. Out of context, the two sentences that make up O'Reilly's utterance can plausibly be interpreted in the way Flaherty takes them, but knowing what we know about behaviour in bars and what would constitute a likely question in the circumstances, Flaherty's interpretation is ridiculous. Flaherty's interpretation imposes on O'Reilly's question a kind of incoherence - or at least a lack of appropriateness to the situation. In making the utterance, O'Reilly assumes that Flaherty will bring to the utterance the knowledge that would permit only one interpretation: the right one in the circumstances.

When someone in a pub says, 'Would you like another?', it normally constitutes an offer of a drink. We bring our background knowledge into play, and we do not take it to mean another chair, another barman, another overcoat, even though all these things (chair, barman, overcoat) are present. But suppose you are in a bar and you have no chair and there are a few empty chairs on the other side of the room. You cross the room and take a chair and a person sitting next to another empty chair says 'Would you like another?' You assume that s/he is enquiring whether you need another chair, not offering you a drink. And the speaker knows that you will make that leap of understanding. That is how people communicate.

What I am saying is that in making sense of any sentence/utterance, we don't rely exclusively on our knowledge of the words and grammar of the language; we also have to use what we know of the situation and of the world and things and people and the way they function. I think that this section also serves to demonstrate that explaining the joke largely deprives it of its humorous effect. This is because jokes often depend on some element being implicit.

Looking for Al
This question of what we expect is quite important in the way that we process text (or indeed most other things in our experience). You must have experienced situations where you join a group of people engaged in conversation. You listen to a couple of exchanges and you remain mystified as to what they are talking about. So you say: 'What are you talking about?' One of the conversationalists says, 'We're talking about Al Pacino' - or whatever. And you say. 'Oh, right.' And then you can join in the conversation. Without that clue to the topic, you are lost. You know the vocabulary; you know the grammar; but you don't really know what the conversation means. You cannot fit it into any framework that enables you to make sense of it.

Of course, that is the unusual case. Usually, you do pick up the topic without any help, and you provide your own frame of reference, which, with luck, more or less coincides with that of the other people present. Possibly this is harder to do in a foreign language because it is harder to recognise the subtle clues, but it is still no mean feat in your own. Of course, if you don't know who Al Pacino is, then you are not much wiser, and you still cannot participate very effectively. You say, 'Is he an applied linguist?' They say, 'No, you moron, he's a film actor. Where have you been living?' If you don't ask, but proceed on the assumption that they are discussing an Applied Linguist and say, 'Well, I'm having a hard time with Henry Widdowson myself', there is a serious breakdown: a pragmatic breakdown, to use a technical term.

The hypothetical exchanges that I have just outlined serve to illustrate the importance of background knowledge in human interaction. The reason that you were bewildered by the conversation at the outset is that talk (i.e. the oral production of text) proceeds on the basis of presuppositions regarding the things being talked about. Its coherence depends on a lot of assumptions made by speaker/writer and hearer/reader. That is another reason why a Martian or a computer would have an enormous job talking like a human; almost every utterance involves a colossal freight of background knowledge: not just such things as who Al Pacino is, but also such facts as that films have actors, that some actors are considered to be better at their job than others, that some are considered more physically attractive than others, and so on. Other conversations might take it for granted that a car runs on petrol, that houses have doors, that the only way to avoid growing old is to die young (though people sometimes forget that one).

The coherence of a text depends not only on what the text actually says, but also on what inferences the reader/listener makes. Text producers (speakers or writers) automatically assume that text processors (hearers or readers) bring a great deal of prior knowledge with them to the text. Without such assumptions being made, it would be virtually impossible to communicate. Imagine how laborious it would be if you had to spell out all the background information for everything uttered.

When someone asks: 'What time is it?' and the person addressed says: 'The ferry has just left', the success of the interaction depends on all kinds of mutual knowledge. If both people know that the ferry always leaves at six o'clock, then the inference is that it is just after six o'clock. But that is an extreme case.

It is easy to agree that more money is the key to meeting the government's education aims, but we have to face facts: the public purse is not bottomless.

The above statement presupposes that the reader will know the following (among other) things: education costs money; public money is spent on education; the government administers public money; public money comes from taxes; money helps to maintain high standards. If a speaker or a writer over-estimates the background knowledge of the hearer or reader, there will be some degree of break-down in communication; if the writer under-estimates, the text will be tedious for the hearer or reader.

Schema theory
One attempt to present a model of background knowledge is schema theory. The singular term is schema; the plural is schemata. When you tune in correctly to the conversation about Al Pacino, outlined above, you could be said to activate a film actor schema: a mental construct into which you can attempt to fit aspects of the present conversation. This entails knowing the things about film actors mentioned above, and many more. Everyone's knowledge will differ, at least in the details, but with sufficient common ground communication is possible.

If you know about Al already, you might even be said at a very local level to activate an Al Pacino schema. This would probably include such elements as: male, small, dark, animated, American, Italian ancestry, starred in The Godfather films; it might also include more esoteric facts such as New Yorker, Shakespeare enthusiast, appeared in Dog Day Afternoon, Revolution, Looking for Richard, Scent of a Woman, Donny Brasco, and so on. It might also include evaluative elements: talented actor/genius/ham; handsome/sexy/ugly/; short; thin; etc. Any or all of these could be presupposed in the course of a conversation. As you can see, I myself have a fairly detailed schema for Al Pacino, and so I could join in easily. But even with only the film actor schema activated, you could probably make sense of most of the conversation, and not lose your credibility as a conversationalist on this occasion.

What I have just said seems to suggest that each person has a different schema in his or her mind, and this seems very plausible. But there must be enough similarity, enough information common to everyone's schemata, to enable us to communicate. Obviously, if your schema and mine were different in every respect, effective communication on the topic would be impossible; we would simply misunderstand each other.

What we are talking about here is what Carrell (1988: 101) describes as: 'the role of pre-existing knowledge structures in providing information left implicit in text’. But it is not just a single schema that needs to be activated for any situation. The background knowledge presupposed in almost any text - written or spoken - is enormous. Let's take a real bit of text from the journal English Language Teaching. Some contextual (co-text) detail: the text appears on a new page below the title Action and condition in the post-elementary classroom in a larger bold font and also below the name Sherrill Howard Pochieca. The stretch of text cited here is an extract from a longer stretch all printed in italic, which in turn occurs before a much longer non-italicised stretch that continues for nearly six pages.

TEXT 1.1
This article proposes that a distinction between 'action verbs' and 'condition verbs' can be very useful for post-elementary learners who have trouble choosing correct verb forms. By facilitating a more functional approach to the tense system, the distinction can contribute to a better understanding of the appropriateness of target structures.

This text could be said to exploit a language-learning schema. (It also involves a 'research article schema', and all the things that go with that, which is text-oriented rather than content-oriented - I shall focus on content here. The text-oriented schema is discussed again in Unit 8.) This includes or interacts with a 'grammar schema'. Certain elements are built into the schema (or schemata) and so they don't need to be spelled out by the writer. They include:

• people usually learn foreign languages by studying
• verbs can be rightly or wrongly chosen by learners
• learners can be classified into developmental stages
• learners need to choose tenses (when they speak/write)
• some verb functions are difficult for learners
• there are different approaches to teaching the tense system
• structures can be targeted
• teachers teach learners
• teachers target structures
• teachers make choices about how to teach/what to target/etc.
• languages have grammar
• grammar has categories, which include verbs
• verbs have functions
• verbs can be classified into different types according to function
• verbs have tenses
• tenses can be viewed as a system
• etc.

Note that although learners are mentioned, teachers are not. Yet we can reasonably assume a teacher (or someone in a teacher-type role: course-writer, materials-writer, etc.). In fact, we can say that our socially-induced language-learning schema involves a teacher. Of course, people learn languages without teachers and so, whereas learner is an obligatory element, teacher is usually but not necessarily taken for granted. We could call this a default item. In computing, a default item is one that is present unless you specify that you don't want it. In a language-learning schema, we might assume the existence of a teacher unless we are told there isn't one: e.g. by the use of the term self-tuition or some other indicator of a no-teacher situation.

The word schema, like most words, is used with several different meanings. In the sense (more or less) in which I am using it here, it originates in the field of psychology in the 1930s (Bartlett 1932), although it has been traced back to the 18th Century German philosopher Kant. It came back into academic prominence with work on artificial intelligence (AI). AI is the interface between psychology and computational science. It is concerned with such issues as human interaction with the computer, making computers talk or think like humans and also with shedding light on the workings of the human mind by computer simulations and modelling. Computers are very good at some things that people find difficult, like calculating the sum of a huge list of very large numbers. But computers have enormous difficulty with things that people find easy. For example, even a very young child knows that a cup is a still a cup when it is seen from a different point of view or turned upside down; computers have difficulty with things like that. (They are improving though, whilst we are not improving at all.) So it might be argued - and indeed it is - that computer modelling is not the best way to shed light on the workings of the human mind, but luckily we don't need to pursue that sort of question here.

The term schema theory tends to be used as a blanket term to include work that uses other terms and concepts for related ideas, such as scenario, script, frame. It is not really necessary for you to pay much attention to the fine points of difference between these terms. The theory was not initially concerned with language, but more with mental representations of the material world: how do we recognise something as a house and something else as a cup and yet another thing as a horse when instances of these things vary so much? Do we have a picture of a typical house in our minds? Do we have a list of attributes that we tick off and if they are all there say: Yes, that's a house? Psychologists tried to build up models of what we store in our minds. The suggestion is that we match what we experience with some mentally stored information and in this way make sense of our environment. The schema is the mental framework or pattern.

The term frame has been used for a kind of proposed pattern for such as house. It has obligatory and optional features: a roof might be obligatory; walls and door might also; but windows might be optional; probably a porch or a patio would need to be optional. Some items are default items; that is, we assume they are there unless told otherwise. One question that needs to be answered is: how is it that a person who has learned to recognise a house in one culture (e.g. as having a roof, doors, windows, several rooms devoted to different activities; and so on) also recognises as a house a structure on stilts without doors or windows and not divided into rooms?

A roof seems to be the most basic requirement, but we can still recognise a building without a roof - a defective house perhaps but still a house. However, the presence of a roof on a house is so much part of our concept that we would feel obliged to mention the lack of a roof, if we talked about it. If I gave you directions, for example, and said, 'Walk down the road until you come to a house', I would be pretty sure to say 'without a roof', and, if I didn't, when you got there you would be surprised and perhaps even doubtful about whether this was the house I meant. In Britain, I think that a door would also be assumed, but there are parts of the world where it might not be.

So when I write about a house, I don't need to say, 'And by the way it had a roof'. The existence of the roof is taken for granted, unless I explicitly mention that there wasn't one. This has important effects on the way we talk and write. Once the 'house' schema or frame has been activated, the roof is part of the picture, as it were.

The term script (by analogy with a film script) has been used for mental representations of various human activities. The best known is the restaurant script. The hypothesis is that when we think about eating in a restaurant - or even just hear the word restaurant - we call up a stored representation involving food, waiters, tables, chairs, etc. We do not need to be told that these things are present because we take them for granted.

Look at the following text. It comes from a Do-It-Yourself book on home improvements.

TEXT 1.2
Every now and again somebody hits the headlines by building something his local council doesn't like and then, after a long legal struggle, being ordered to pull it down again. Prudent householders will avoid such confrontations, if only because lawyers cost even more than builders; but if you are self-sufficient as well as prudent you must check carefully before you begin putting brick on brick, and even sometimes before dipping brush into paintpot.

To understand this it is not enough to know the vocabulary and grammar of English. We might say that to make full sense of the text we need to activate a building permission schema (or script or scenario); we need to know that in some societies there are social constraints on what you can do or have done to the structure of a house that you own, and that these constraints are enforceable by law. The authority that deals with these matters in Britain is the local council. If they object to changes in your house they can take you to court and you may need lawyers to represent you. You may have to pay these lawyers. This text embeds, as it were, what we might call a home improvement schema or scenario, which assumes building and painting. Building assumes the presence of bricks (or other building materials) and painting assumes paint and brushes. So the script would include as characters: householders, the local council, lawyers: Props include: bricks, paint, paintbrush. Proceeding on the assumption that the reader has these schemata, the writer leaves out such things as the connection between the council not liking something and a long legal struggle. An additional set of assumptions underlies Every now and again somebody hits the headlines, which requires the activation of a newspaper schema. It also requires us not to interpret it literally.

This text might well be incoherent for a reader with none of this background knowledge. At least, it would be hard to understand. When I, as a linguist with a professional background in English teaching, try to read a specialist text about nuclear physics, it is unlikely that it will have the coherence for me that it has for a nuclear physicist. To put it another way: I am unlikely to be able to impose upon it the coherence which s/he can impose upon it. Even if they might not put it in the same terms, writers know this and act accordingly. Nuclear physicists do not write in the same way for their peer group as for a nonspecialist readership. They assume different background knowledge, different schemata, a different potential for making inferences. Of course, some people do this better than others, but everyone makes some effort in this direction. (And I hope I am making the right assumptions as far as you are concerned.)

Inference obviously plays a major role in our processing of text. It is one of the key ways in which coherence is achieved. We might say that coherence is given to the text, at least in large part, by the listener or reader, whose contribution is as crucial as that of the writer/speaker.

Task 1.4
1 Explain the inference involved in making sense of the following utterance (one speaker only):
Can you buy me a coffee? I left my coat in the classroom.

2 Explain the following joke.
Caller: I'd like an ambulance urgently, please, Mrs Smith is about to
have a baby.
Hospital: Is this her first baby?
Caller: No, this is her husband.

3 What kind of basic schema or scenario might be activated by the following:
(i) Do you have a table for two near the window?
(ii) I'd like to cash a cheque, please.
(iii) Swab, please, nurse! Scalpel!
(iv) Just the one suitcase? Did you pack it yourself? And has it been with you
since you packed it?

(v) The Giancano family had locked up the action in Orleans and Jefferson
parishes in Prohibition. Their sanction and charter came from the Chicago
commission, of course, and no other crime family ever tried to intrude.

Background knowledge and the processing of headlines
Headlines seem to cover the gamut of informative potential from excessively explicit to very cryptic. In the first instance, they seem to almost make the reading of the rest of the text redundant; in the second they seem to be deliberately obscure, perhaps in order to compel the reader to continue reading beyond the headline. (The grammar of headlines is very distinctive, too. This topic is dealt with in more detail in Unit 8 of the GLE module.)

But the obscurity of the headline (as of any text) is not just a quality of the text itself; it varies with the reader. Inevitably, much of the material in this course is culturally biased since the sources are predominantly British. This is something I am constantly aware of and try to mitigate. With some kinds of material this is much less significant than with others. With headlines, it can be very significant.

The processing of headlines often demands more than a knowledge of English, and, even more than a knowledge of the register of newspaper headlines and the practices of newspapers in Britain. It often depends as well on prior background knowledge of the events, persons and circumstances referred to in the headlines. Thus, some course participants - regardless of country of origin - will find de-coding these headlines more difficult than will others. For example, those who are unfamiliar with UK football will miss meanings that are apparent to the football fans who read the papers in Britain. The Sun headline and subhead below illustrate this:


Little Euro joy

This heads a report on Aston Villa's 2-0 victory over Steaua Bucharest in an early round of the 1997 European cup; the match took place just down the road from your alma mater, Aston University. As Neil Custis of the Sun tells it: 'ASTON VILLA roared into the UEFA Cup quarter-finals on a night of high drama.') Tayls you win' is a pun: a play on the expression used when tossing a coin and the name of one of the goal-scorers, Ian Taylor. The caption to an impressive action shot of Taylor kicking the ball is TAYLOR-MADE, another pun.) Processing this requires detailed knowledge of current football and some familiarity with the names of the people involved. But, of course, this headline is no easier for a British native speaker of English who is not a football fan than it is for someone who lives in another country.

The nominal group Little Euro joy is not the negative item it appears to be to the uninitiated, but a reference to Brian Little, Aston Villa's manager at the time. Thus, fully spelt out in 'non-headline' English, it means something like Brian Little's joy about the European Cup competition. The second sentence of the text says: 'Brian Little's gallant warriors booted out the Romanians amid incredible tension in this third round tie at Villa Park.' Thus the co-text gives the reader a chance to correct any false impression and re-process the headline, but the informed reader will probably have decoded the intended meaning the first time.

Proper names can have different significance depending on the time and place: the word City in a headline in a national - or in a London local - paper usually refers to the City of London; that is, not literally the city of London itself, or even the region of London known as the City (within the ancient boundaries), but to the Stock Exchange and the Banks: the financial powers. In the Birmingham Evening Mail, the same word usually refers to the city of Birmingham: so the phrase City man means a man from Birmingham. However, in local newspapers in Manchester, Stoke and Bristol, city usually refers to the football team. (To spell out the inference that my last statement requires: there are teams called Stoke City, Manchester City, Bristol City etc. You might deduce this even if you did not know already, but it would be hard work without the help of some awareness of football teams and their naming practices.)

Even the time gap between publication and the time when you read the headlines will have an effect on how easy it is to process them. Newspapers by their very nature deal primarily with immediate events. People and organizations that are in the news today may be forgotten in a few weeks.

This question of how the reader's background knowledge contributes to understanding is a crucial factor in all text processing. As a final task what can you say about the following headlines?

Task 1.5
See if you can explain the connections and what the following headlines are about. What background knowledge is assumed? Some or all may be incomprehensible to non-Brits - or even to Brits, but try anyway.

(1) In Penny Lane there is an
adman selling motor-cars

Volkswagen in £6m bid for
Beatles songs to plug Beetles

(2) Let them eat cakes - with
compulsory folic acid
Gloss: This headline precedes an item about a possible government policy of adding supplements to all bread and cakes to prevent neural tube defects in new-born babies. (A science news item in a national paper)

(3) PM's fury at 'tacky Diana death industry'

Keys to tasks

Task 1.3
Obviously, many different correct answers are possible, but you might have mentioned the following:

1 Physical setting: Store or supermarket. Written notice, large size for easy reading, displayed in prominent place e.g. at high level. Context of culture: retail sales, large, relatively impersonal company (not one-to-one small shop) with acceptance of shop-lifting as a serious possibility. Participants: store management to customers. Impersonal style. Co-text: no predictable precise co-text though other texts in the same environment indicating location of goods, special offers, prices, etc.

2 Physical setting: probably hospital, operating theatre, surgery, or clinic or vicinity thereof. Participants: doctor to assistant, e.g. nurse, dresser or junior doctor. (Note: IV means 'intra-venous'). Doctor about to perform some form of surgery or demonstrate surgical technique. Relatively urgent situation. Spoken discourse (or a written representation of it).

All sorts of alternatives are possible, of course. It might all be taking place in someone's kitchen. Or the speaker might be an amateur about to operate because no doctor is available, etc. Or it could be an actor in a drama about hospitals. But the features of the text are still determined by the institutional context of situation outlined above. Of course, the whole thing could be a discussion of a hypothetical situation; e.g. a discussion of how to perform this process, but in that case the imagined situation would be similar and this is what in part determines the discourse. Similar things could be said of all these texts. To say that a situation can be imitated or imagined is not to say that situations do not determine text.

3 Physical location of the text: part of furniture catalogue (note reference to page 218) for actual warehouse or store (Note: Take it home today). Sales culture; promotion of sale. (Note the similarity to advertising discourse). Participants: management to customers. Written. Predictable co-text: similar information about other items of furniture; possibly illustrations.

4 Physical location of the text: part of the information on a video-cassette box. Manufacturer to user. Written. Context of culture: includes sales of goods, ownership of patents.

5 Physical location of text: Newspaper headline. Context of culture: Press accounts of legal trials, alleged crimes, etc. Editorial journalists to general news reading public. If you recognise the names, you might also identify it as being about alleged crime in motor-racing.

(Don't worry too much if your answers were not quite the same as mine. They may be as good or better. Besides, we go further into this type of analysis in subsequent units.)

Task 1.4
1 We need to infer that the speaker's money is in the coat and so s/he cannot buy the coffee.

2 The speaker from the hospital means this to refer to the baby that is about to appear on the scene; the father implausibly assumes that it means himself (as in Who is this speaking? Incidentally, this an Australian joke. Brits would usually say that rather than this in this situation and so the misunderstanding could not arise.

3 (i) Restaurant. (ii) Bank. (iii) Medical operating theatre. (iv) Airport: checking-in.
(v) American gangster schema/Mafia schema.

Task 1.5
1 A news item about the German/multinational car manufacturing company, Volkswagen, trying to buy songs by the long defunct British pop group, the Beatles, to advertise their Beetle car. One of the Beatles 'songs is 'Penny Lane' and it has similar lyrics: 'adman' is substituted for 'barber' and 'selling motor-cars' substituted for 'showing photographs'. The sub-head helps to explain, but you still need to know roughly the identity of Volkswagen, Beatles and Beetles. You need to activate a motor-car sales/advertising schema and a pop song schema.

2. Without reading on, it is unlikely that any reader could make much of just the headline here. But I guessed it was something to do with food supplements from the mention of 'compulsory folic acid'; this triggered for me a 'nutrition' schema since I knew that folic acid is an important vitamin. 'Let them eat cakes' is a near-quotation of the supposed comment of Marie-Antoinette, Queen of France, shortly before the Revolution, in response to being told that the poor had no bread. It adds little to the meaning here, though knowing the story makes us feel better retrospectively about Marie-Antoinette having her head cut off - not that this has any relevance at all to the folic acid story. The usual quote is ‘Let them eat cake’, but if she said it, she said it in French, or maybe, since she was an Austrian, in German. None of this is at all apposite to this particular text, but it is part of my Marie Antoinette schema or French Revolution schema which was activated by the quotation and which might be relevant on another occasion.

3. At the time of publication (1998), this issue of the commercial spin-off of the accidental death of Diana, Princess of Wales, (sales of memorabilia, mugs, pictures, records, etc.) was still an occasional news item though this was six months after the incident. The editors could assume that most readers would know that 'PM' meant the British Prime Minister (otherwise the headline would have specified 'of Sri Lanka' or whatever) and that the PM at the time was Tony Blair; also of course that 'Diana' meant that particular Diana and not someone else of the same name. Also 'Diana death industry' was a familiar schema at the time.

Bartlett, F.C. (1932). Remembering: An Experimental and Social Study. Cambridge: Cambridge University Press.
Bloor T & Bloor M 2004 The Functional Analysis of English, 2nd edition London & New York: Arnold, Chapter 1; Chapter 12, sections 12.7 and 12.9.
Carrell P L & Eisterhold J C 1988 Schema theory and ESL reading pedagogy. In Carrell P, Devine J & Eskey D (eds.) Interactive Approaches to Second Language Reading, (also any other chapters in the same book)
de Beaugrande R 1997 New Foundations for a Science of Text and Discourse: Cognition, Communication, and the Freedom of Access to Knowledge and Society Norwood, New Jersey: Ablex. Chapter 1 only
de Beaugrande R and Dressler W 1981 Introduction to text linguistics. London; New York: Longman.
Halliday M A K. 1994 Introduction to Functional Grammar 2nd edition London.: Arnold.
Sinclair, J McH 1991 Corpus, Concordance, Collocation, Oxford: OUP
Stubbs M 1996 Text and Corpus Analysis Oxford UK & Cambridge, Mass. USA: Blackwell. Chapter 1
Willis D 1990 The Lexical Syllabus: A New Approach to Language Teaching, London: Collins COBUILD.