02 Discourse Section 1 24/1/07 15:43 Page 3

1
Language in use

A sample of language
Given a sample of language, there are all kinds of things we can
say about it. Take, for example, a familiar public notice:
KEEP OFF THE GRASS

This, to begin with, is something in English, as distinct from

French or Arabic or Chinese or any other language. It consists of
four words, all in capital letters, and all, we might more expertly
add, monosyllabic. If we have been fortunate enough to have had
some instruction in linguistics, we might then go on point out that
the words combine to form a grammatical unit called a sentence,
and a sentence furthermore of an imperative as distinct from a
declarative or interrogative kind consisting of two main
constituents. The first is a verb phrase consisting of the two words
keep off, the second a noun phrase which itself consists of two
constituents, a definite article the and a noun grass. Noting these
grammatical features, we might think up a number of other
sentences that seem to be structured in the same way: Put out the
light, for example, or Turn off the tap, only to realize perhaps that
appearances are deceptive and that these are actually not quite the
same, but interestingly different. For these two structures can also
take the form of the alternative sequences Put the light out, Turn
the tap off, but Keep the grass off will not do. So examining the
properties of this sample of ours as a sentence might lead us into a
fascinating excursion into the mysteries of grammatical analysis.
But although linguists might delight in examining our sample
in this way, this is not the kind of thing people would customarily
do. Languages are traditionally recorded for us in analytic terms:
language in use 3
02 Discourse Section 1 24/1/07 15:43 Page 4

grammars display the range of possible structural combinations

in sentences, and dictionaries provide us with the meanings of
words separated out and listed in alphabetic order. These can be
said to represent the encoded resources of form and meaning that
speakers of a particular language know and draw upon intuitively
when they use it. But they do not correspond with how speakers
actually experience it as use. When they come across a public
notice, they do not see it as a sample of language and analyse it
into its formal constituents. They take note of it only to the extent
that they recognize its purpose, as something not to analyse but to
act upon. In other words they treat it as a text.

What is a text?
A text can be defined as an actual use of language, as distinct from
a sentence which is an abstract unit of linguistic analysis. We
identify a piece of language as a text as soon as we recognize that
it has been produced for a communicative purpose. But we can
identify a text as a purposeful use of language without necessarily
being able to interpret just what is meant by it. It is a fairly
common experience to come across texts in an unknown
language which we nevertheless recognize as public notices, food
labels, menus, or operating instructions, and to be frustrated by
the inability to understand them. Clearly we would generally
need to know the language a text is in to be able to interpret it. But
this is not the only condition on interpretation. We may know
what the language means but still not understand what is meant
by its use in a particular text.
Consider again the public notice ‘keep off the grass’. We
may know well enough what the word grass denotes (and should
we be in any doubt we can consult a dictionary to find out). But
what the word denotes is not the same as knowing what it is
meant to refer to when it occurs here in the phrase the grass. The
definite article the signals that what is being referred to is a matter
of shared knowledge. The grass. But which grass? Obviously, one
might say, the grass in the vicinity of the notice. So what we do is
to establish reference by relating the text to the context in which it
is located. But then the question arises as to how far this vicinity is
meant to extend. Does the grass refer just to the particular patch

4 survey
02 Discourse Section 1 24/1/07 15:43 Page 5

where the notice is placed, or to other patches nearby as well, or

to the whole park? The range of reference is not specified in the
language itself. We make assumptions about what it is on the
basis of what we know about public notices of this kind and how
they are conventionally meant to be understood. In other words
we relate the text not only to the actual situational context in
which we find it, but to the abstract cultural context of what we
know to be conventional.
And by relating text to context we infer not only what the
notice refers to, but also what its purpose is. We recognize that it
is intended as a prohibition, although whether we choose to pay
any attention to it is another matter—and one we shall be taking
up later.
The same point can be made about other notices we come
across in daily life. Thus we recognize that the texts ‘handle
with care’ or ‘this side up’ refer to a container on which
they are written and function as requests, that ‘wet paint’
refers to some surface in the immediate vicinity that has been
newly painted, and functions as a warning. Similarly, when we
see the label ‘keep away from children’ on a medicine
bottle, we take this as a specific warning in reference to the partic-
ular contents of the bottle, rather than, say, as a piece of general
advice to keep clear of young people at all times. When we come
across notices and labels, then, we make sense of them by relating
the language to the immediate perceptual context where they are
located, and to the conceptual context of our knowledge of how
such texts are designed to function. We cannot make sense out of
them simply by focusing on the language itself. In the case of
simple texts like notices and labels, establishing the language-
context connections is usually a fairly straightforward matter.
With other texts, even apparently simple ones, making such con-
nections is not so easy, as anybody who has had the experience of
assembling furniture from a set of instructions is likely to testify.

Text and discourse

The simple texts we have been considering so far all serve an
obvious utilitarian purpose: notices, labels, instructions are
designed to be directly acted upon and to get things done. But of

language in use 5
02 Discourse Section 1 24/1/07 15:43 Page 6

course not all texts are so simple in form or so straightforward in

function. Although, as we have seen, not all texts extend beyond
the sentence, a great many of them do: travel guides, information
leaflets, newspaper articles, interviews, speeches, reports, poems,
and so on. Some of these have an obvious utility function but
others are meant to serve a range of different social purposes:
to give information, express a point of view, shape opinion,
provide entertainment, and so on. These functions, furthermore,
are frequently combined in complex ways: a travel guide, for
example, may provide information, but is also designed to
promote the attractions it describes; and what is presented as a
factual account in a newspaper article will usually reflect, and
promote, a particular point of view.
Whether simple or complex, all texts are uses of language
which are produced with the intention to refer to something for
some purpose. We identify a stretch of language as a text when we
recognize this intention, and there are times when the intention
is made explicit as when a text is labelled as a notice, or
instructions, or report or proclamation. But recognizing a text is
not the same as realizing its meaning. You may not know what is
being referred to in a particular text, or in part of a text; or you
may know full well what is being referred to, but fail to see what
communicative purpose lies behind the reference. In the case of
simple texts, like public notices, it will be a straightforward
matter to match up intention with interpretation, but in the case
of more complex ones, like newspaper articles, such matching
can, as we shall see later, prove to be highly problematic.
People produce texts to get a message across, to express ideas
and beliefs, to explain something, to get other people to do certain
things or to think in a certain way, and so on. We can refer to this
complex of communicative purposes as the discourse that
underlies the text and motivates its production in the first place.
But at the receiving end readers or listeners then have to make
meaning out of the text to make it a communicative reality. In
other words, they have to interpret the text as a discourse that
makes sense to them. Texts, in this view, do not contain meaning,
but are used to mediate it across discourses. Sometimes, of course,
as with the notices we have been considering, the mediation is
relatively straightforward: what the text means to the reader will

6 survey
02 Discourse Section 1 24/1/07 15:43 Page 7

generally match up with what the producer of the text meant by

it. Obviously we must generally assume that texts will serve to
mediate some convergence between discourses, or otherwise no
communication would take place at all, but, as we shall see, the
degree of convergence varies a good deal. As we all know from
our own experience, no matter how explicitly we think we have
textualized what we want to say, there is always the possibility
that it will be interpreted otherwise.
So the term discourse is taken here to refer both to what a text
producer meant by a text and what a text means to the receiver.
Of course what somebody means by producing a particular text
may well relate to broader issues of what social and ideological
values they subscribe to, and another way of thinking of discourse
is indeed to focus on such broader issues and look at how texts
can be used to express, and impose, certain ways of thinking
about the world. This is something we shall return to later in the
book (in Chapter 7).

Spoken and written text

For the moment, the point to be made is that texts are the percep-
tible traces of the process, not itself open to direct perception, of
mediating a message. In conversation, these traces are typically
fragmented and ephemeral, and disappear as soon as they are
produced to serve their immediate discourse purpose. They can,
of course, be recorded, but do not need to be, and usually are not.
Thus, participants in spoken interaction produce and process
text as they go along and there is no need for it to be retained as a
record for it to mediate their discourse, and this mediation is regu-
lated on-line to negotiate whatever convergence between inten-
tion and interpretation is required for the purpose. Written text,
on the other hand, is not jointly constructed and construed on-
line in this way. It is typically designed and recorded unilaterally
in the act of production by one of the participants, the writer, as a
completed expression of the intended message. The text is then
taken up and interpreted as a separate process. The mediation,
therefore, is displaced and delayed and this obviously will often
make a convergence between intention and interpretation more
difficult to achieve.

language in use 7