12 Discourse analysis

There’s two types of favors, the big favor and the small favor. You can measure
the size of the favor by the pause that a person takes after they ask you to “Do
me a favor.” Small favor – small pause. “Can you do me a favor, hand me that
pencil.” No pause at all. Big favors are, “Could you do me a favor . . .” Eight
seconds go by. “Yeah? What?”
“. . . well.” The longer it takes them to get to it, the bigger the pain it’s going
to be.
Humans are the only species that do favors. Animals don’t do favors. A lizard
doesn’t go up to a cockroach and say, “Could you do me a favor and hold still,
I’d like to eat you alive.” That’s a big favor even with no pause.
Seinfeld (1993)
In the study of language, some of the most interesting observations are made,
not in terms of the components of language, but in terms of the way language
is used, even how pauses are used, as in Jerry Seinfeld’s commentary. We have
already considered some of the features of language in use when we discussed
pragmatics in the preceding chapter. We were, in effect, asking how it is that
language-users successfully interpret what other language-users intend to con-
vey. When we carry this investigation further and ask how we make sense of
what we read, how we can recognize well-constructed texts as opposed to those
that are jumbled or incoherent, how we understand speakers who communicate
more than they say, and how we successfully take part in that complex activity
called conversation, we are undertaking what is known as discourse analysis.
The word ‘discourse’ is usually defined as ‘language beyond the sentence’ and
the analysis of discourse is typically concerned with the study of language in
text and conversation.

Interpreting discourse
When we concentrate on the description of a particular language, we are nor-
mally concerned with the accurate representation of the forms and structures
used in that language. However, as language-users, we are capable of more than
simply recognizing correct versus incorrect forms and structures. We can cope
with fragments in newspaper headlines such as Trains collide, two die, and know
that what happened in the first part was the cause of what happened in the second
Discourse analysis 

part. We can also make sense of notices like No shoes, no service, on shop win-
dows in summer, understanding that a conditional relation exists between the
two parts (‘If you are wearing no shoes, you will receive no service’). We can
even cope with texts, written in English, which appear to break a lot of the rules
of the English language. The following example, provided by Eric Nelson, is
from an essay by a student learning English and contains all kinds of errors, yet
it can be understood.
My Town
My natal was in a small town, very close to Riyadh capital of Saudi Arabia.
The distant between my town and Riyadh 7 miles exactly. The name of this
Almasani that means in English Factories. It takes this name from the peopl’s
carrer. In my childhood I remmeber the people live. It was very simple. Most
the people was farmer.
This example may serve to illustrate a simple point about the way we react to
language that contains ungrammatical forms. Rather than simply reject the text
as ungrammatical, we try to make sense of it. That is, we attempt to arrive at a
reasonable interpretation of what the writer intended to convey. (Most people
say they understand the ‘My Town’ text quite easily.) It is this effort to interpret
(or to be interpreted), and how we accomplish it, that are the key elements
investigated in the study of discourse. To arrive at an interpretation, and to
make our messages interpretable, we certainly rely on what we know about
linguistic form and structure. But, as language-users, we have more knowledge
than that.

Cohesion
We know, for example, that texts must have a certain structure that depends on
factors quite different from those required in the structure of a single sentence.
Some of those factors are described in terms of cohesion, or the ties and con-
nections that exist within texts. A number of those types of cohesive ties can
be identified in the following paragraph.
My father once bought a Lincoln convertible. He did it by saving every penny
he could. That car would be worth a fortune nowadays. However, he sold it to
help pay for my college education. Sometimes I think I’d rather have the
convertible.
There are connections present here in the use of words to maintain reference
to the same people and things throughout: father – he – he – he; my – my
– I; Lincoln – it. There are connections between phrases such as: a Lincoln
convertible – that car – the convertible. There are more general connections
created by a number of terms that share a common element of meaning, such
as ‘money’ (bought – saving – penny – worth a fortune – sold – pay) and ‘time’
 The Study of Language

(once – nowadays – sometimes). There is also a connector (However) that marks

the relationship of what follows to what went before. The verb tenses in the first
four sentences are all in the past, creating a connection between those events,
and a different time is indicated by the present tense of the final sentence.
Analysis of these cohesive ties within a text gives us some insight into how
writers structure what they want to say and they may be crucial factors in
our judgments on whether something is well written or not. It has also been
noted that the conventions of cohesive structure differ from one language to
the next and may be one of the sources of difficulty encountered in translating
texts.
However, by itself, cohesion would not be sufficient to enable us to make
sense of what we read. It is quite easy to create a highly cohesive text that has a
lot of connections between the sentences, but is very difficult to interpret. Note
that the following text has connections such as Lincoln – the car, red – that
color, her – she, letters – a letter, and so on.

My father bought a Lincoln convertible. The car driven by the police was red.
That color doesn’t suit her. She consists of three letters. However, a letter isn’t
as fast as a telephone call.

It becomes clear from this type of example that the ‘connectedness’ we expe-
rience in our interpretation of normal texts is not simply based on connections
between the words. There must be some other factor that leads us to distinguish
connected texts that make sense from those that do not. This factor is usually
described as ‘coherence’.

Coherence
The key to the concept of coherence (‘everything fitting together well’) is not
something that exists in words or structures, but something that exists in people.
It is people who ‘make sense’ of what they read and hear. They try to arrive at
an interpretation that is in line with their experience of the way the world is.
Indeed, our ability to make sense of what we read is probably only a small part
of that general ability we have to make sense of what we perceive or experience
in the world. You may have found when you were reading the last example
text that you kept trying to make the text fit some situation or experience that
would accommodate all the details (involving a red car, a woman and a letter).
If you work at it long enough, you may indeed find a way to incorporate all
those disparate elements into a single coherent interpretation. In doing so, you
would necessarily be involved in a process of filling in a lot of gaps that exist in
the text. You would have to create meaningful connections that are not actually
expressed by the words and sentences. This process is not restricted to trying
Discourse analysis 

to understand ‘odd’ texts. In one way or another, it seems to be involved in our

interpretation of all discourse.
It is certainly present in the interpretation of casual conversation. We are
continually taking part in conversational interactions where a great deal of what
is meant is not actually present in what is said. Perhaps it is the ease with
which we ordinarily anticipate each other’s intentions that makes this whole
complex process seem so unremarkable. Here is a good example, adapted from
Widdowson (1978).
   : That’s the telephone
   : I’m in the bath
   : O.K.

There are certainly no cohesive ties within this fragment of discourse. How
does each of these people manage to make sense of what the other says? They
do use the information contained in the sentences expressed, but there must
be something else involved in the interpretation. It has been suggested that
exchanges of this type are best understood in terms of the conventional actions
performed by the speakers in such interactions. Drawing on concepts derived
from the study of speech acts (introduced in chapter 11), we can characterize
the brief conversation in the following way.
She makes a request of him to perform action.
He states reason why he cannot comply with request.
She undertakes to perform action.
If this is a reasonable analysis of what took place in the conversation, then it
is clear that language-users must have a lot of knowledge of how conversation
works that is not simply ‘linguistic’ knowledge. Trying to describe aspects of
that knowledge has been the focus of an increasing amount of research.

Speech events
In exploring what it is we know about taking part in conversation, or any other
speech event (e.g. debate, interview, various types of discussions), we quickly
realize that there is enormous variation in what people say and do in different
circumstances. In order to begin to describe the sources of that variation, we
would have to take account of a number of criteria. For example, we would have
to specify the roles of speaker and hearer (or hearers) and their relationship(s),
whether they were friends, strangers, men, women, young, old, of equal or
unequal status, and many other factors. All of these factors will have an influence
on what is said and how it is said. We would have to describe what the topic of
conversation was and in what setting it took place. Some of the effects of these
factors on the way language is used are explored in greater detail in chapters 19
 The Study of Language

and 20. Yet, even when we have described all these factors, we will still not
have analyzed the actual structure of the conversation itself. As language-users,
in a particular culture, we clearly have quite sophisticated knowledge of how
conversation works.

Conversation analysis
In simple terms, English conversation can be described as an activity in which,
for the most part, two or more people take turns at speaking. Typically, only one
person speaks at a time and there tends to be an avoidance of silence between
speaking turns. (This is not true in all situations or societies.) If more than one
participant tries to talk at the same time, one of them usually stops, as in the
following example, where A stops until B has finished.
 : Didn’t you [know wh-
: [But he must’ve been there by two
 : Yes but you knew where he was going

(A small square bracket [ is conventionally used to indicate a place where

simultaneous or overlapping speech occurs.)
For the most part, participants wait until one speaker indicates that he or she
has finished, usually by signaling a completion point. Speakers can mark their
turns as complete in a number of ways: by asking a question, for example, or by
pausing at the end of a completed syntactic structure like a phrase or sentence.
Other participants can indicate that they want to take the speaking turn, also in
a number of ways. They can start to make short sounds, usually repeated, while
the speaker is talking, and often use body shifts or facial expressions to signal
that they have something to say.

Turn-taking
Some of the most interesting research in this area has revealed different expec-
tations of conversational style and different strategies of participation in con-
versation. Some of these strategies seem to be the source of what is some-
times described by participants as ‘rudeness’ (if one speaker cuts in on another
speaker) or ‘shyness’ (if one speaker keeps waiting for an opportunity to take
a turn and none seems to occur). The participants characterized as ‘rude’ or
‘shy’ in this way may simply be adhering to slightly different conventions of
turn-taking.
One strategy, which may be overused by ‘long-winded’ speakers or those who
are used to ‘holding the floor’ (e.g. some politicians and professors), is designed
to avoid having normal completion points occur. We all use this strategy to some
extent, usually in situations where we have to work out what we are trying to
say while actually saying it. If the normal expectation is that completion points
Discourse analysis 

are marked by the end of a sentence and a pause, then one way to ‘keep the
turn’ is to avoid having those two markers occur together. That is, don’t pause
at the end of sentences; make your sentences run on by using connectors like
and, and then, so, but; place your pauses at points where the message is clearly
incomplete; and preferably ‘fill’ the pause with a hesitation marker such as er,
em, uh, ah.
In the following example, note how the pauses (marked by . . .) are placed
before and after verbs rather than at the end of sentences, making it difficult to
get a clear sense of what this person is saying until we hear the part after each
pause.
 : that’s their favorite restaurant because they . . . enjoy French food and when
they were . . . in France they couldn’t believe it that . . . you know that they
had . . . that they had had better meals back home

In the next example, speaker X produces filled pauses (with em, er, you know)
after having almost lost the turn at his first brief hesitation.
 : well that film really was . . . [wasn’t what he was good at
: [when di-
 : I mean his other . . em his later films were much more . . er really more in
the romantic style and that was more what what he was . . you know . . em
best at doing
 : so when did he make that one

These types of strategies, by themselves, should not be considered undesirable

or domineering. They are present in the conversational speech of most peo-
ple and they are part of what makes conversation work. We recognize these
subtle indicators as ways of organizing our turns and negotiating the intricate
business of social interaction via language. In fact, one of the most noticeable
features of conversational discourse is that it is generally very ‘co-operative’.
This observation has been formulated as a principle of conversation.

The co-operative principle

An underlying assumption in most conversational exchanges seems to be that the
participants are co-operating with each other. This principle, together with four
maxims that we expect our conversational partners to obey, was first described by
the philosopher Paul Grice. The co-operative principle is stated in the following
way: “Make your conversational contribution such as is required, at the stage
at which it occurs, by the accepted purpose or direction of the talk exchange in
which you are engaged” (Grice, 1975: 45). Supporting this principle are four
maxims, often called the ‘Gricean maxims’.
 The Study of Language

The quantity maxim: Make your contribution as informative as is required,

but not more, or less, than is required.
The quality maxim: Do not say that which you believe to be false or for which
you lack adequate evidence.
The relation maxim: Be relevant.
The manner maxim: Be clear, brief and orderly.
It is certainly true that, on occasion, we can experience conversational exchanges
in which the co-operative principle may not seem to be in operation. However,
this general description of the normal expectations we have in conversation
helps to explain a number of regular features in the way people say things. For
example, during their lunch break, one woman asks another how she likes the
sandwich she is eating and receives the following answer.

Oh, a sandwich is a sandwich.

In logical terms, this reply appears to have no communicative value since it states
something obvious and doesn’t seem to be informative at all. However, if the
woman is being co-operative and adhering to the quantity maxim about being
“as informative as is required”, then the listener must assume that her friend
is communicating something. Given the opportunity to evaluate the sandwich,
her friend has responded without an explicit evaluation, thereby implying that
she has no opinion, good or bad, to express. That is, her friend has essentially
communicated that the sandwich isn’t worth talking about.

Hedges
We use certain types of expressions, called hedges, to show that we are con-
cerned about following the maxims while being co-operative participants in
conversation. Hedges can be defined as words or phrases used to indicate that
we’re not really sure that what we’re saying is sufficiently correct or complete.
We can use sort of or kind of as hedges on the accuracy of our statements, as
in descriptions such as His hair was kind of long or The book cover is sort of
yellow (rather than It is yellow). These are examples of hedges on the quality
maxim. Other examples would include the expressions listed below that people
sometimes put at the beginning of their conversational contributions.
As far as I know . . . ,
Now, correct me if I’m wrong, but . . .
I’m not absolutely sure, but. . . .
We also take care to indicate that what we report is something we think or
feel (not know), is possible or likely (not certain), and may or could (not must)
happen. Hence the difference between saying Jackson is guilty and I think it’s
Discourse analysis 

possible that Jackson may be guilty. In the first version, we will be assumed to
have very good evidence for the statement.

Implicatures
When we try to analyze how hedges work, we usually talk about speakers
implying something that is not said. Similarly, in considering what the woman
meant by a sandwich is a sandwich, we decided that she was implying that
the sandwich wasn’t worth talking about. With the co-operative principle and
the maxims as guides, we can start to work out how people actually decide
that someone is ‘implying’ something in conversation. Consider the following
example.
    : Are you coming to the party tonight?
   : I’ve got an exam tomorrow.

On the face of it, Lara’s statement is not an answer to Carol’s question. Lara
doesn’t say Yes or No. Yet Carol will immediately interpret the statement as
meaning ‘No’ or ‘Probably not’. How can we account for this ability to grasp
one meaning from a sentence that, in a literal sense, means something else? It
seems to depend, at least partially, on the assumption that Lara is being relevant
and informative, adhering to the maxims of relation and quantity. (To appreciate
this point, try to imagine Carol’s reaction if Lara had said something like Roses
are red, you know.) Given that Lara’s original answer contains relevant informa-
tion, Carol can work out that ‘exam tomorrow’ conventionally involves ‘study
tonight’, and ‘study tonight’ precludes ‘party tonight’. Thus, Lara’s answer is
not simply a statement about tomorrow’s activities, it contains an implicature
(an additional conveyed meaning) concerning tonight’s activities.
It is noticeable that, in order to describe the conversational implicature
involved in Lara’s statement, we had to appeal to some background knowledge
(about exams, studying and partying) that must be shared by the conversational
participants. Investigating how we use our background knowledge to arrive at
interpretations of what we hear and read is a critical part of doing discourse
analysis.

Background knowledge
A particularly good example of the processes involved in using background
knowledge was provided by Sanford & Garrod (1981), who presented readers
with a short text, one sentence at a time. Their text begins with the following
two sentences.

John was on his way to school last Friday.

He was really worried about the math lesson.
 The Study of Language

Most people who are asked to read these sentences report that they think John
is probably a schoolboy. Since this piece of information is not directly stated
in the text, it must be an inference. Other inferences, for different readers, are
that John is walking or that he is on a bus. These inferences are clearly derived
from our conventional knowledge, in our culture, about ‘going to school’, and
no reader has ever suggested that John is swimming or on a boat, though both
are physically possible, if unlikely, interpretations.
An interesting aspect of the reported inferences is that they are treated as
likely or possible interpretations that readers will quickly abandon if they do
not fit in with some subsequent information. Here is the next sentence in the
text.
Last week he had been unable to control the class.
On encountering this sentence, most readers decide that John is, in fact, a teacher
and that he is not very happy. Many report that he is probably driving a car to
school. Then the next sentence is presented.
It was unfair of the math teacher to leave him in charge.
Suddenly, John reverts to his schoolboy status, and the inference that he is a
teacher is quickly abandoned. The final sentence of the text contains a surprise.
After all, it is not a normal part of a janitor’s duties.
This type of text and manner of presentation, one sentence at a time, is rather
artificial, of course. Yet the exercise involved does provide us with some insight
into the ways in which we ‘build’ interpretations of what we read by using a
lot more information than is presented in the words on the page. That is, we
actually create what the text is about, based on our expectations of what normally
happens. In attempting to describe this phenomenon, researchers often use the
concept of a ‘schema’ or a ‘script’.

Schemas and scripts

A schema is a general term for a conventional knowledge structure that exists in
memory. We were using our conventional knowledge of what a school classroom
is like, or a ‘classroom schema’, as we tried to make sense of the previous
example. We have many schemas (or schemata) that are used in the interpretation
of what we experience and what we hear or read about. If you hear someone
describe what happened during a visit to a supermarket, you don’t have to be
told what is normally found in a supermarket. You already have a ‘supermarket
schema’ (food displayed on shelves, arranged in aisles, shopping carts and
baskets, check-out counter, and other conventional features) as part of your
background knowledge.
Similar in many ways to a schema is a script. A script is essentially a dynamic
schema. That is, instead of the set of typical fixed features in a schema, a script
Discourse analysis 

has a series of conventional actions that take place. You have a script for ‘Going
to the dentist’ and another script for ‘Going to the movies’. We all have versions
of an ‘Eating in a restaurant’ script, which we can activate to make sense of the
following discourse.

Trying not to be out of the office for long, Suzy went into the nearest place, sat
down and ordered an avocado sandwich. It was quite crowded, but the service
was fast, so she left a good tip. Back in the office, things were not going well.

On the basis of our restaurant script, we would be able to say a number of things
about the scene and events briefly described in this short text. For example,
although the text doesn’t have this information, we would assume that Suzy
opened a door to get into the restaurant, that there were tables there, that she
ate the sandwich, then she paid for it, and so on. The fact that information
of this type can turn up in people’s attempts to remember the text is further
evidence of the existence of scripts. It is also a good indication of the fact that
our understanding of what we read doesn’t come directly from what words and
sentences are on the page, but the interpretations we create, in our minds, of
what we read.
Indeed, crucial information is sometimes omitted from important instructions
on the assumption that everybody knows the script. Think carefully about the
following instructions from a bottle of cough syrup.

Fill measure cup to line

and repeat every 2 to 3 hours.

No, you’ve not to just keep filling the measure cup every 2 to 3 hours. Nor
have you to rub the cough syrup on your neck or in your hair. You are
expected to know the script and drink the stuff from the measure cup every 2 or
3 hours.
Clearly, our understanding of what we read is not only based on what we
see on the page (language structures), but also on other things that we have in
mind (knowledge structures). To understand more about the connection between
these two things, we have to take a close look at the workings of the human
brain.

Study questions
1 What is the basic difference between cohesion and coherence?
2 How do speakers mark completion points at the end of a turn?
3 What are the names of the Gricean maxims?
4 What are hedges in discourse?
5 Which maxim does this speaker seem to be particularly careful about?

I may be mistaken, but I thought I saw a wedding ring on his finger.

6 In the study of discourse understanding, what are scripts?

 The Study of Language

Research tasks
A In the analysis of discourse, what is ‘intertextuality’?
B In conversation analysis, what is the difference between a ‘preferred’
response and a ‘dispreferred’ response? How would you characterize the
responses by ‘Her’ in these two examples?
   : How about going for some coffee?
   : Oh . . . eh . . . I’d love to . . . but you see . . . I . . . I’m supposed to get
this thing finished . . . you know

   : I think she’s really sexy.

   : Well . . . er . . . I’m not sure . . . you may be right . . . but you see . . .
other people probably don’t go for all that . . . you know . . . all that
make-up . . . so em sorry but I don’t think so
C Using what you know about the co-operative principle and maxims,
describe how or something is used (twice) in this extract from a
conversation between two women chatting about people they knew in high
school (Overstreet, 1999).
     : I can’t remember any ge- guys in our grade that were gay.
    : Larry Brown an’ an’ John Murphy. I – huh I dunno, I heard
John Murphy was dressed – was like a transvestite or something.
     : You’re kidding.
    : I – I dunno. That was a – an old rumor, I don’t even know if
it was true.
     : That’s funny.
    : Or cross-dresser or something
     : Larry – Larry Brown is gay?
D (i) Identify the main cohesive ties in this first paragraph of a novel.
(ii) What do you think ‘they’ were hitting?

Through the fence, between the curling flower spaces, I could see them
hitting. They were coming toward where the flag was and I went along the
fence. Luster was hunting in the grass by the flower tree. They took the flag
out, and they were hitting. Then they put the flag back and they went to the
table, and he hit and the other hit. They went on, and I went along the fence.
Luster came away from the flower tree and we went along the fence and
they stopped and we stopped and I looked through the fence while Luster
was hunting in the grass.
(From William Faulkner, The Sound and the Fury)

Discussion topics/projects
I In the study of discourse, a distinction is often made between ‘new
information’ (treated as new for the reader or listener) and ‘given
information’ (treated as already known by the reader or listener). Read
Discourse analysis 

through the following recipe for bread sauce and identify the ways in which
given information is presented. (Try to think carefully about carrying out the
instructions in the Method section and how many unmentioned things you
are assumed to have and use.)

Ingredients: 1 small onion 3 oz. fresh breadcrumbs

2 cloves 1 oz. butter
1 cup of milk Pepper and salt
Method: Peel the onion and push cloves into it. Simmer gently with the
milk and butter for at least twenty minutes. Remove the onion, pour
the milk over the breadcrumbs. Let this stand to thicken and reheat
before serving.

(For background reading, see chapter 5 of Brown & Yule, 1983.)

II According to Deborah Schiffrin, “the analysis of discourse markers is part
of the more general analysis of discourse coherence” (1987: 49). Looking at
the use of discourse markers (in bold) in the following extract from
conversation, do you think that they help to make this discourse more
coherent? If any of them were omitted, would it become less coherent?
Given these examples, how would you define discourse markers? Do you
think the word like (used twice here) should be treated as a discourse
marker?
I believe in that. Whatever’s gonna happen is gonna happen. I believe . . .
that . . . y’know it’s fate. It really is. Because eh my husband has a brother,
that was killed in an automobile accident, and at the same time there was
another fellow, in there, that walked away with not even a scratch on him.
And I really fee- I don’t feel y’can push fate, and I think a lot of people
do. But I feel that you were put here for so many, years or whatever the
case is, and that’s how it was meant to be. Because like when we got
married, we were supposed t’get married uh like about five months later.
My husband got a notice t’go into the service and we moved it up. And
my father died the week . . . after we got married. While we were on our
honeymoon. And I just felt, that move was meant to be, because if not, he
wouldn’t have been there. So eh y’know it just s- seems that that’s how
things work.
(For background reading, see chapter 3 of Schiffrin, 1987.)

Further reading
For another short introduction to discourse analysis, see chapter 9 of Finegan
(2004). Introductory textbooks are Cutting (2002) and Nunan (1993). Other
textbooks are Cameron (2001), Johnstone (2002) and Renkema (2004). More
detailed treatments can be found in Chafe (1994) and Schiffrin (1994). More
 The Study of Language

specifically, on cohesion, see Halliday & Hasan (1976), on conversational style,

see Tannen (1984), on conversation analysis, see Hutchby & Wooffitt (1998) or
Psathas (1995), on maxims and implicatures, see Grice (1989) and on schemas
and scripts, see chapter 7 of Brown & Yule (1983). Comprehensive reviews are
presented in Schiffrin et al. (2001) and Wetherell et al. (2001).