Discourse Analysis, Advanced Learners
and the Cambridge CPE Exam
by Alex Case
Even a cursory listen to native speakers having a natural
informal conversation shows clearly that people do not speak
in a series of grammatically formed sentences. At the same
time, it is also obvious that an extended text, whether spoken
or written, is not a random collection of unconnected sentences
(NB. 'text' is used in this essay to refer to any piece of
language used for analysis). Historically, the grammar-translation
approach was guilty of treating language as both of the above.
Parts of audio-lingualism was very much the same, although
at least students did have the benefit of responding to questions
and taking part in extended, if artificially created, dialogues,
although neither the students nor the practitioners were expected
to involve themselves in analysis of the structures involved.
Discourse Analysis is, very generally, an attempt tackle the
two very important points in the first two sentences above,
i.e. to look at language at the 'beyond sentence' level. Because
of this, since the coining of the name in the early fifties,
and its gradual development as a science in its own right
since then, it has been of vital interest for the ELT profession.
own interest in this area comes from two distinct sources.
My own scientific background, specifically in Physics, has
made react somewhat against the 'eclectic approach' as it
is sometimes interpreted, as relying too much on intuition-
whereas many of the great scientific theories such as those
of relativity are very much counter-intuitive. This has given
me a great interest in more theoretical approaches to language
and to 'action research' in my own classroom. This has come
together with a realisation that many of my students' problems
with language and precisely those of 'discourse'.
essay aims to examine the methods and conclusions of Discourse
Analysis, and to examine the relevance of these for high-level
students, and more specifically those studying for the Cambridge
CPE. These will each be discussed in turn.
The first approach of Discourse Analysis looks at the fact
that the natural unit of a conversation is not the sentence,
and asks what it could be. The answer is said to be (1) 'the
turn'. This itself leads to the question of how speakers know
when and how to take turns. There seems to be no one simple
answer, but lack of this knowledge can be considered a reason
why non-native speakers find it difficult to join conversations
with native speakers.
next point of analysis is to see if any patterns emerge in
the language used in these turns. By looking at the functions
of the language used, familiar patterns of response appear
such as 'statement of opinion' followed by 'agreement/ disagreement'.
The idea of each piece of language used having a function
is very familiar in the world of English teaching, and teachers
are used to teaching from textbooks that have at least in
part a functional syllabus. What is not so obvious is that
there is no single function for a single piece of language
such as the traditional sentence stems. The question is then,
how do native speakers interpret the language correctly? One
theory is based on the assumption that people are 'co-operating
in order to communicate' (2). Another similar theory is based
on the concept of 'politeness' (1).
more things shared by the people communicating through a piece
of discourse are a context and a shared knowledge of the world.
Lack of knowledge of these through the use of extracts removed
from their original context or the lack of shared knowledge
a language learner can suffer from can make understanding
difficult, if not impossible.
approach is to look at the 'beyond sentence grammar' i.e.
the formal links between sentences. Formal links are known
as cohesive devices, and can be subdivided as below:
expressions, such as 'it' and 'that', which refer to other
things defined in different parts of the sentence or text.
This can be sub-divided into anaphoric reference, where the
thing is defined earlier, and cataphoric reference where it
is defined later. In a similar manner, we can avoid the repetition
of whole stretches of language by substitution by words such
as 'do' and 'so'. At other times we can miss the known part
out, known as ellipsis. One pattern that often appears in
texts is the use of similar tenses over long stretches of
language. Another is lexical chains, such as chains of synonyms
used to refer to the same thing. Finally, links are maintained
between sentences and clauses by the use of conjunctions.
as well as the formal links between sentences and paragraphs
extended texts often seem to have an overall form, for example
the well known 'introduction, points for, points against,
conclusion' structure for a discursive essay.
Discourse analysis in the CPE exam
There are three main questions:
1. What examples of discourse discussed above are found in
the texts in the exam?
2. Are these tested in the questions?
3. Do these things need teaching, and if so how?
Despite the name, Section A is a multiple choice test of vocabulary
using unconnected sentences, and hence cannot be considered
discourse . Section B, however, contains three extended written
texts on which students must answer multiple choice questions.
The first two passages are always 'non-fiction aimed at the
general educated reader' (all quotes from Cambridge CPE Handbook
(3) until stated otherwise) and the third from a 'novel or
literary work'. A detailed analysis of a past paper is included
in appendix 1.
the results given, looking at the text as a piece of discourse
rather than as individual sentences and pieces of vocabulary
could be directly used to fully or partially answer four of
the six questions given. Most surprisingly for me, these included
lexical sets and the overall structure of the text. This result
has made me re-think the importance of looking at texts as
discourse in my classes. For example, books such as Advanced
Masterclass (4) and Distinction (5) contain exercises along
the lines of 'find all the words to do with.....' and I have
used these in class, but have always considered them more
as vocabulary extension, and very much as an optional extra
for actual reading comprehension. In fact, looking at lexical
chains seems to aid comprehension. In a similar way, I will
consider using more activities such as labelling paragraphs
by purpose, putting jumbled texts into order etc.
next obvious question is whether students have any problems
with these questions, and therefore how much time should be
spent on practice of these points of discourse. My general
experience is that by the level of proficiency most students
tend to have a good instinctive grasp of these points, most
especially referring words and ellipsis, and would be more
likely to fail on not knowing the meaning of 'fortuitous'
in question 30, whereas at FCE level it is often worth spending
more time on these with activities such as placing missing
sentences back into texts.
Candidates might have to write a description, a discursive
essay, a narrative, a letter, a report or a short article.
as students at this level have already reached a level of
fluent communication, another factor comes into play- that
of producing a 'negative impact' in a receiver; due to clumsy
style, flat speech, lack of accuracy etc. Lack of knowledge
of features of discourse such as formality and overall format
of different genres of writing can be a large factor in this.
This effect is generally strongest in writing with a very
fixed and well known format- most especially formal letters,
but also reports, discursive essays and articles. The format
of an article in English, for example, tends to be to give
all the important, new information at the beginning- filling
in background, history and details later on. This format and
that of discursive essays are less applicable to other languages
than might be assumed. I have found that French students,
for example, can often produce almost unreadable discursive
essays simply because of this fact.
As well as these general matters, writers need to consider
the target audience. In terms of the exam, the discursive
essay and the description suffer from this not being given.
An overall format for the description also seems to be a good
example of where a student's own common sense, when guided
by the teacher if necessary, is a better guide than native
speaker models, and I have in fact found this to work well.
3- Use of English
Question 1, which is a cloze exercise with a whole text, tests
referring expressions very thoroughly. In the June 1992 exam,
6 of the 20 questions were actually referring words. As in
the Reading paper, however, I would not expect to find students
having the most problems with these. Even the handbook suggests
that 'this part of the paper should not be over practised,
as it is unlikely to raise learners' language awareness'.
In contrast, Questions 2 and 4 (key word transformations)
, are a continual fund of unknown and often useful structures
and fixed expressions for the students, despite the fact that
they are single sentences without context.
B is the summary. The text from 1992 has again been analysed
for referring words, ellipsis etc. (see appendix 1). The comprehension
questions seem deliberately designed to test meaning in context
e.g. What is 'this litany' and why is it 'unruffled'? Again,
however, students less seem to lack the ability to guess meaning
from context than the vocabulary with which to rephrase them
in their own words, something with which a native speaker
might also be expected to struggle. A teacher can help students
by 'guiding' them as much as possible to an answer, and I
have found that throwing students into a task such as this
without lots of preparation can quickly ruin self-confidence.
This leads on to the task of summarising one particular aspect
of the text in 60-90 words. The given aim is to test 'clarity,
coherence and conciseness' of the piece of writing. How this
does so in a way the writing paper does not remains unstated.
It can be a good opportunity for the teacher to concentrate
on conjunctions, however, as the students are not expected
to come up with their own opinions. Good practice for this
can be giving the full summary in short, unconnected sentences
and asking the students to join them into one five sentence
page 2 of 2
the lesson plan
to the articles index