Chapter 1: Variation and Change in English

1.1 Introduction
Linguistics is concerned with the study of language, including theories of language as discussed in
Chapmans Thinking about Language: Theories of English (TL:TE) and the ways in which a
language is structured and patterned as in Jeffries Discovering Language: Describing English
(DL:DE). Language in this general sense can be theorised and described in a general and
abstract way, or we can be concerned with the categorisation and description of a particular
language, such as English. However, when we come to consider how any language such as
English is actually used in everyday life, then it becomes clear that far from being spoken and
written in exactly the same way by everybody, language is in fact tremendously varied. This
chapter considers different spoken varieties of English, and the extent to which the society and
communities in which we live affect the ways in which we speak and write.
1.2 begins with definitions of language, dialect, accent, variety and standard English,
and a consideration of the attitudes people have towards variation in language use. 1.3
outlines the history of the standardisation of English in England in order to illustrate why
dialectal variation persists in this region and throughout the United Kingdom as a whole.
This section also traces the origins of prejudicial attitudes towards variation that continue
to this day. By contrast, a consideration of the linguistic history of the USA shows how
the processes of standardisation have been very different from those in England. This
contrast explains why there is not the same degree of variation to be found in the USA as
in the UK, nor the same degree of prejudice towards non-standard varieties. These two
examples show how, although standardisation follows identifiable processes, the way in
which these processes are enacted in the case of individual language varieties varies a
great deal and depends upon a unique combination of social, economic, political,
geographic and historicalvariables.
1.4 discusses research undertaken into variation based upon two different
methodological approaches: firstly, focus upon the linguistic variable; secondly, focus
upon the social variable and social networks. Studies into linguistic variation of
phonology, morphology and syntax are the focus of traditional dialectology such as the
regionally based studies undertaken by The Survey of English Dialects (1962) and
Kortmann and Schneiders two-volume A Handbook of Varieties of English (2004).
Studies into the social variable are the focus of sociolinguistic dialectology, which also
focuses upon the linguistic variable, but also takes into account social issues such as
race, class and gender in relation to linguistic variation, such as those of Labov (1966,
1972, 1979) and Trudgill (1974, 1978). More recently, Milroy (1987) and Milroy and
Gordon (2003) have based their sociolinguistic studies upon the notion of social
networks, arguing that in addition to linguistic and social variables, attention should also
be paid to the communities and contexts within which speech occurs. 1.5 provides
students with guidance and advice on undertaking their own studies into variation and
change, whilst section 1.6 provides suggestions for further reading.
1.2 Language, dialect, accent and variety
1.2.1 Language
Let us consider the meaning of the term language and to what it refers. In DL:DE Jeffries makes
a distinction between language as a system and language use. A language system refers to an
idealised form of the language which is separate from how a language is actually used, and
DL:DE concentrates upon describing language as a system. In TL:TE Chapman considers
language from different theoretical perspectives: as a type of behaviour, as a state of mind and as
a form of communication. A sociolinguistic approach to the study of language considers language
as behaviour, particularly in taking account of the regional and social situations in which language
occurs, and the social as well as linguistic factors that affect how speakers relate to one another.
Consequently, a sociolinguistic approach to language behaviour, rather than being concerned with
language in a more general or abstract way, asks questions such as: what is a language? and
what is language for? Language is not just about communication, but also about identity, a
factor which is paramount in sociolinguistics.
Deciding which criteria to adopt for defining a language, however, is far from straightforward.
Take the example of the language called English. Who are the speakers of English? Are they the
people living in a particular country, England, where the language is spoken? One popular way of
deciding the boundary of a language and boundaries between languages is to consider their
geography. We generally assume that people living in a particular geographically defined country
speak the language associated with it: French in France, German in Germany and so on. English
by this definition is the language spoken by people living in England, Great Britain (England,
Scotland and Wales) and the United Kingdom (England, Scotland, Wales and Northern Ireland).
However, it is not always the case that people living in a geographically defined area all speak the
same language, or that one language is the exclusive property of a particular country. This is
certainly not the case with English, which is spoken not only in countries that make up the UK,
but is also widely used across many countries around the globe, including the United States.
Another factor which has to be taken into account is that English today is widely used in many
countries across the globe as the language of business, diplomacy, medicine and the internet.
We should also take account of the fact that there are many countries in the world which are not
monolingual: that is, they have not one nationally recognised language, but several. For example,
in Switzerland, there are three major languages: German, French and Italian. Switzerland
recognises itself as multi-lingual society whereas most of us would agree that Britain (with the
exception of Wales) unlike Switzerland, is a monolingual society in that everyone speaks
English. The same could also be said of other countries across the globe, such as the USA, where
English is the language of official communication, used in educational contexts, the law,
government, the media and so on. However, if we look at the actual languages spoken in areas
such as the UK, USA or Australia today, then they include many others besides English. Far from
being monolingual, these countries, including those that make up the UK, are actually
multilingual, with many inhabitants speaking languages other than English. So although the UK
has an official language, English, its inhabitants actually come from a vast range of language
backgrounds, making the UK linguistically diverse.
This is further complicated by the fact that one of the countries which makes up Great Britain and
the UK, the principality of Wales, has two officially recognised languages, Welsh as well as
English, and all its inhabitants are taught to be bilingual. This situation is similar to the one in
Canada, where people are taught to be bilingual in French and English. In the United States of
America, there is no officially recognised national language, and legislating to impose one is
forbidden by its constitution, although to all intents and purposes English functions as a national
language through its use in public institutions such as education, business and the law. What
these examples all illustrate is that what counts as a language then, is not only dependent upon
geography, but also upon history, politics and economics.
The association between language and nation or nationality is a very strong and powerful one. The
association between language and identity of all kinds, regional and social as well as national, is
also very powerful. The language, languages or varieties of a language that we speak form an
integral part of who we are, and attempts at imposing one language or variety of a language on the
population of a nation are often bound up with issues of power and ideology. The reasons why
one language or one variety of a language becomes associated with a particular nation are many
and varied, resulting from a combination of historical and social changes. Throughout history, one
of the first things an invading force of another country imposes upon the conquered people is its
language, particularly in terms of political, economic and educational institutions and suchlike.
For example, The Norman Conquest of 1066, the Roman invasion of the first century BC and the
altering of country boundaries in Eastern Europe post 1945 to form the United Soviet States of
Russia. What counts as the language of a country at any particular moment in time, therefore, is
not as simple and straightforward as it might at first seem. The term language is also a very
difficult, if not impossible, one to define linguistically, as the example in the following section
illustrates.
1.2.2 The Ebonics Debate
In December 1996 the Oakland School District Board in the American State of California
passed a resolution which gave official recognition to Ebonics, a separate language and
distinct from English. Ebonics is a compound word made up of from the two words
Ebony meaning black and phonics meaning sound, As a consequence, schools in the
Oakland District were required to recognise and accept Black pupils speech in the
classroom as part of a bilingual education program, so that pupils would be taught both in
their primary language, Ebonics, and in English. The impetus for adopting such a
resolution came from the persistently low educational achievements obtained by black
students in the district, who made up over fifty percent of the school population. Although
a local issue, the passing of this resolution quickly became national news and
precipitated a fierce debate across all the American States. Amongst the issues raised by
the Oakland resolution on Ebonics was whether or not black English could be shown to
be linguistically a separate language. The very raising of this issue immediately brought
to the fore another one, namely, the wider, more politically sensitive one of the nature of
the relationship between language and ethnicity, and between African-Americans and
Anglo-Americans in contemporary American society. At the heart of the debate was not,
as it tended to be presented in the press, whether one was for or against Ebonics, but
the far wider issue of equality: of equal access to education for all American citizens
regardless of ethnicity and through it, right of access to a full participating status in
American life regardless of class, ethnicity and gender (Clark 2001:237-252).
Tatalovich makes the point that whenever an opportunity arises in America such as that
provided by the Oakland Resolution to debate matters of language, ordinary people rise
to defend the English language against those who speak other tongues (1995:1). He
points out that the Oakland Resolution, in common with similar episodes throughout the
history of the United States, is symptomatic of the debate over whether the United
States should reflect a dominant English-speaking majoritarianism or encourage a
multilingual culture (1995:2). Consequently, for Tatalovich, controversies over language
such as those sparked by the Ebonics debate become not only linguistic conflicts but
also moral ones. Such controversy is further compounded by the fact that, although
English is by far the most common language spoken and used in most areas of
American public life, it has no official recognition as the national language of all
American states, nor indeed does any other language. Furthermore, unlike many other
major English speaking countries in the world, the US Federal government has not been
able to assert the dominance of English or legislate any kind of national language policy
through the education system, since neither language nor education are enshrined in its
constitution. One of the ways in which the United States gets around this is by the
importance it places on immigrants into the United States taking a test in citizenship,
which is in English.
Not surprisingly, the Ebonics debate found its way onto the agenda of the Linguistics Society of
America. In 1997, the society passed a resolution calling for the recognition of Ebonics, alongside
African-American Vernacular English (AAVE) and Vernacular Black English, to be recognised as
systematic and governed by linguistic rules. However, the society refused to be drawn upon the
issue of classification, on the grounds that the distinction between languages and dialects or
varieties is usually made more on social and political grounds than purely linguistic ones. It
argued that what was important from a linguistic and educational perspective was not whether
Ebonics or AAVE is called a language but that they, in common with other speech varieties, be
recognised as systematic and governed by linguistic rules. At the heart of the debate then,
according to the Society, was not the linguistic issue of what counts as a language, but more the
social and political ones which surround the establishment and maintenance of language
hierarchies.
If linguistics does not help us in defining the term language, then maybe another way of
defining language is in terms of sub-divisions or as a collection of mutually intelligible
dialects. In this way, we can talk about the southwest dialect of France, the Black
Country dialect of English, the Bavarian dialect of German and so on. So, for example,
English as a language includes not only its standardised form known as standard
English (see 1.3 below), but all other dialects which exist within the geographical
boundaries of England and elsewhere. However, mutual intelligibility as a criterion is not
very helpful, since different languages as well as dialects can be mutually intelligible. For
example, Norwegian, Swedish and Danish, though accepted as different languages, can
each be understood by the speakers of the other languages. Other factors concerning
intelligibility also have to be taken into account, such as the individuals degree of
exposure to a language, her/his educational background and a willingness to understand.
1.2.3 Dialect, accent and variety
One way of defining a language is as a group of dialects and accents which have a
certain number of forms and structures in common. Put simply, dialect refers to words
and syntactic structure, whereas accent refers to the sounds that speakers produce and
the intonation and pitch which accompanies sound. If a dialect describes the words and
syntactic structures used by one person or a group of speakers, then accent is the word
used to describe pronunciation, and the two often go hand in hand. For example, if
someone speaks in a regional dialect of English such as Scouse in the North West or
Black Country in the Midlands, then her/his pronunciation will also be particular to that
area. If you were to walk north from Lands End in Cornwall to John OGroats at the very
north of Scotland, you would hear different accents and dialects of English Geordie in
the North East, West Country in the South West and Cornish. This a known as a dialect
continuum or a chain of mutual intelligibility; that is, there is no distinct or complete
break from one dialect and accent to another, and speakers of geographically adjacent
dialects can understand one another. However, the cumulative effect of linguistic
differences is such that the greater the geographical separation, the greater the difficulty
of understanding what people say. Europe has many dialect continua, an example of
which is omance, stretching across the Iberian peninsula through France and parts of
Belgium down to the southern tip of Italy.
In addition to a purely linguistically descriptive dimension to accents and dialects, there
is also a social one. Chambers and Trudgill (1980:3) point out that dialects are commonly
viewed as:
substandard, low status, often rustic forms of language, generally associated with
the peasantry, the working class, or other groups lacking in prestige. DIALECT is
also a term which is often applied to other forms of language, particularly those
spoken in more isolated parts of the world, which have no written form. And dialects
are often regarded as some kind of (often erroneous) deviation from a norm as
aberrations of a correct or standard form of language.
Trudgill and Chambers found that people speaking with rural accents such as those of
Devon and Cornwall in South West England, for example, are typified as dim-witted but
trusting, whereas people speaking with urban ones such as Cockney in London are
typified as quick-witted but untrustworthy. By contrast, speakers of standard English
with a Received Pronunciation accent (see 1.3.1 below) are generally thought to be
more intelligent than speakers of other dialects and also superior morally as well as
socially.
Because of such negative connotations, linguists have come to prefer to use the term
variety when describing variation in language. This has none of the negative
connotations associated with the terms dialect and accent, and fits in with the idea of
descriptive linguistics: that is, basing descriptions of language upon actual use. It can
also be applied across a wider range of language features than the terms dialect and
accent. For example, we can talk of linguistic variation, historical variation, social
variation, geographic variation, stylistic variation and so on.
The main reason why there is so much geographical variation in English throughout the
United Kingdom and England especially, and throughout Europe in general, is historical.
Language operates across two dimensions simultaneously: the horizontal dimension of
space, also called the diachronic axis, and the vertical dimension of time, called the
synchronic axis. In addition, there is also a third, social dimension to be considered,
which accounts for variation between social classes and cuts across both. The reason
why, for example, there is so much variation in England today and so little in the USA is
historical. Similarly, the negative and prejudicial attitudes commonly held towards
linguistic variation, particularly in the UK, have their roots in history. Consequently, in
order to comprehend variation in English and attitudes held towards it, it is important to
consider the social history of English.
1.3 Standards of English
1.3.1 What is standard English?
The term standard English (SE) is the one most commonly used to label the language
English. It is the variety of English used in public life in England and other English
speaking countries, for example: in education, law, medicine and government.
Nowadays, it has no geographical boundary, and is used across the whole of England
and other English speaking countries. In England, it also has an accent associated with
it, known as Received Pronunciation: RP for short. Because of their origins and history,
SE and RP are closely associated with the language of the middle and upper classes in
English society, known variously as the Queens English or BBC English. The concept
of RP is a peculiarly English one, having no equivalent in any other part of the English
speaking world. For example, there is no US equivalent of RP. Some linguists have
argued that SE is best defined as the written form of English, on the grounds that
standard English is not a matter of pronunciation and is thus not tied to any particular
accent (See: Crystal 1995 and Trudgill 1999). Rather, it is a matter of grammar,
vocabulary and orthography, that is, spelling and print face, and not of pronunciation.
Nevertheless, linguists such as Stubbs (1986) argue that accent is involved in notions of
standardness, since people have an idea of what is and is not standard in
pronunciation. That is, the accent RP is widely regarded as the standard accent of
British English, just as standard English is widely regarded as the standard written form
of English, as used in education and other public institutions such as the law and
government.
Indeed, there is much disagreement amongst linguists as to whether or not SE can be
classed as dialect at all. Some, such as Trudgill and Chambers (1980) and Milroy (1987)
argue that it is, pointing out that all speakers speak at least one dialect, and that
standard English is as much a dialect as any other form of English. Consequently, some
speakers may have no other variety than SE, whilst others may have either a regional
variety and/or SE. Other linguists disagree, on the grounds that standard English differs
from other dialects in a number of ways, especially in the fact that it has its own writing
system. Because of this, they argue that dialects and the study of dialect should
concentrate upon speech. This position, however, ignores the fact that many non-
standard English dialects in England such as that found in the Black Country in the
English West Midlands or Geordie in the North East of England have an established
tradition of writing. Also, if standard English is not a dialect, then it is difficult to see what
else it could be.
As has already been mentioned, whilst regional accents and dialects go hand in hand, standard
English can be spoken with any accent. So, although it is possible (but not very likely) that
someone may speak the Geordie dialect with an RP accent, it is both possible and probable that
someone may speak standard English with a Geordie accent. Indeed, the BBC, once an important
gatekeeper in the maintenance of standard English and RP, has in recent years adopted a much
more liberal policy towards the use of regional accents amongst its presenters. When the BBC was
founded in the 1930s, all broadcasters spoke standard English with a uniform RP accent.
Nowadays, no such training is given, and presenters can be heard speaking standard English in a
variety of different regional accents. Indeed, attitudes towards the use of regional dialects and
accents in public life has altered so significantly since the 1930s that in 2005, the BBC launched a
national campaign celebrating and promoting regional English linguistic diversity called
Voices. This campaign, in addition to television and radio programmes, has used the internet and
phone-ins to undertake a national dialect survey on the use of regional dialects in England today.
Working together with dialectologists at the universities of Leeds and Cardiff, this initiative aims
to produce a comprehensive survey of contemporary English dialects and accents. This BBC
sponsored project takes issue with the correlation between RP and BBC English, and is fighting
hard to disassociate the BBC from any gate-keeping role in language use. It argues that the fact
that BBC news broadcasters spoke RP in the early days was more the product of a restricted
group from which BBC employees were drawn rather than as part of any deliberate policy. Even
so, employees outside that group, at the time, were given elocution lessons to ensure use of RP.
The project also questions the role of the BBC in championing a particular accent and thus
potentially holding back the tide of language change. It also takes issue with the fact that some
linguists have relabelled RP as BBC English, and called it such in pronunciation dictionaries for
foreign learners. However, if you call the accent normally used in BBC news broadcasts BBC
English, and that it is an example of RP, then by definition, the people the BBC employs as news
broadcasters are therefore RP speakers. This circularity in defining BBC English in relation to RP
and RP in relation to BBC English renders both concepts meaningless. It is also ironic that this is
happening at a time when the relationship between RP and so-called BBC English is more
logically viewed as a thing of the past.
Nevertheless, initiatives such as Voices aside, there is still a strong correlation in Britain
between dialect, accent and social class. The social dialect of SE is bound not by
geographical region, but by social class.
Fig 1: The triangle or pyramid model of the relationship between status and
accent (Trudgill 1983:29-30)
As figure 1 illustrates, there still exists a sliding scale of hierarchy in English dialects and
accents, with that of standard English and RP held to be the most prestigious, and those
of large urban conurbations cities such as Birmingham the least. Speakers of the most
marked form of RP appear at the apex of the triangle and tend to be accorded the
highest status, whereas speakers of marked regional pronunciation are accorded the
lowest. Even though London is the historical centre of standard English, nevertheless
geographically bound local London accents and dialects such as those associated with
the East End of London (Cockney) and Essex (Estuary English) are generally thought of
as less prestigious and open to ridicule. No wonder then, that the playwright George
Bernard Shaw wrote in 1916 that: it is impossible for an Englishman to open his mouth
without making some other Englishman despise him. Such prejudices have deep
historical roots going back to at least the eighteenth century and the period when the
English we have today began its processes of standardisation. In order to understand
current issues and debates surrounding accents and dialects, we have to consider the
ways in which English was standardised, and the social attitudes and prejudices upon
which standardisation was based.
1.3.2 A brief history of the standardisation of English in England
On any day of any week, we can open an English newspaper and find either articles on,
or letters about, the degeneration and corruption of the English language. In case we
might be tempted to think of such concerns as symptomatic of modern times, then it is
worth pointing out that they date back to at least the fifteenth century and a time when
English was beginning to be standardised. The concept of a standard, unified form of a
language is a relatively new one in historical terms. Very briefly, the need for a unified
form of a language arose during the late middle ages and the Renaissance period in
history for two main reasons: political and economic. Communication over greater areas
of land than before became a political necessity for governments, and an economic
necessity for trade. It was during this period that the concept of a national language
began to be linked to notions of national identity, as the nation states of modern day
Europe began to be formed.
Haugen (1996:97-111) identifies the process of standardisation as a four-stage one:
Stage one: Selection an existing dialect is chosen as the one to be standardised. In the case of
English, this was the Middle English dialect of the East Midlands.
Stage 2: Elaboration the chosen dialect is expanded and elaborated so that it can fulfil the
variety of different functions it is intended to serve. That is, the dialect is added to by adding
vocabulary and/or elaborating grammatical structures. In the case of English, the East Midlands
dialect was elaborated by the importation of words from French and Latin and its grammatical
structures elaborated by the adoption of Latinate constructions.
Stage 3: Codification the vocabulary and grammar is made explicit and written down. Once
codified in writing, then language becomes an object of consciousness in the material world, in
that it can be seen as well as heard, and thus fixed or pinned down. Changes to the language
are thus more open to regulation and control by its community of users, especially by the more
powerful sections of society.
Stage 4: Implementation the chosen variety is implemented throughout a community, usually
by becoming the language of instruction in education and being adopted by public institutions
such as the law, government and the media.
The reason why various regional dialects and accents exist in England today dates back to the late
Middle Ages and beyond. The reason why prejudices against their use persist is due to the theory
of language upon which the processes of standardisation was based. This was a
prescriptive theory, based upon an idealised form of language and how it ought to be used, rather
than upon descriptive theory, and how it actually is used. Prescriptive theories of language
immediately give rise to value judgements. From the late Middle Ages onwards, judgements on
the selection of a dialect were based upon notions of correctness, the best and the pure, which
linked language with morality, and by appealing to an idealised form of usage. In making the case
for the standardisation of English, it was argued that the best language was that spoken by the
best people and any other form was deemed corrupt and spoken by corrupt people.
The history of the English language can be divided into four periods: Old English (400ad to 1100
ad); Middle English (c.1100 ad to 1500 ad); Early Modern English (c.1500ad to 1800 ad) and
Modern English (from 1800). By the end of the Middle English period (1100 ad -1500 ad),
England had five main dialects, each with its own earlier history. The Old English dialect of
Northumbrian developed into Scots and Northern English, north of a line from the Humber to the
Lake District. The Mercian dialect of Old English became the West Midlands and East Midlands
dialects. Below the Rivers Thames and Severn the Southern dialect was spoken, while Kentish
remained in the far south east.
Fig 2: Middle English dialect map
At this time, all dialects of English were bound geographically. People of all social
backgrounds from the richest to the poorest living in these areas spoke and used the
dialect of the region, which included writing in it. Medieval authors lacked any notion of a
standard or prestige form of English, as one did simply not exist, and so they wrote in the
language of their birthplace. During the Middle Ages then, all English dialects had a
written equivalent. Questions of dialect versus standard, regional versus social
dialects which create so much debate today were simply not an issue then, because the
economic and political landscape was very different. Even after English was
standardised and its written form codified, local people continued to speak and indeed
write in their local dialect. Consequently, many regional dialects of todays English, such
as that of the Black Country in the West Midlands, for example, can trace their history
back to Middle English dialects and continue to use syntactic forms, vocabulary and
pronunciation that remain little altered since that time, particularly syntactically and
phonologically.
By the end of the Middle English period, English did not yet have an official, standard
form, in the sense of fixed or uniform, yet many Renaissance writers refer time and time
again to a variety of English they describe as the common dialecte or common
language. Common, that is, in the sense of the most usual, meaning a shared language
of usage and understanding. It is the identifying, elaborating, and codifying of this
common language which preoccupied much of the Early Modern English period, c1500
c1800. By the beginning of this time and not without contestation, the East Midlands
dialect had been selected as the one for elaboration and codification into the standard
form of English. The East Midlands region of England had come to exert tremendous
power intellectually, economically and politically in the processes which shaped the
emerging cultural identity of the English nation. However, unlike other European modern
nation states being formed around the same time, the selection of a dialect as the
standard form of a language did not happen in any planned, systematic or uniform way.
Whilst the Tudors, along with other reigning monarchies in Europe sought to standardise
language in the interest of national unity they did not, unlike in France for example,
establish an Academy to plan and regulate the processes of standardisation. During the
Renaissance period, from 1400-1700, two important factors affected the process of
standardisation: firstly, the invention of printing and secondly, increased mobility amongst
the aristocracy, both of which helped to promote one dialect over others. In addition, two
other noticeable features of change occurred in the English language. The first was the
borrowing of many words from Latin and Greek as well as Italian and Spanish. This came
about as a result of the emphasis on classical learning which spread across Europe from
the Italian States, with Latin and Greek becoming the languages of the civilised and
educated person. Between 1500 and 1650 over 10,000 new words were incorporated
into the emerging standard language, generally through the medium of writing with newly
coined and invented words, for example allurement, allusion, atmosphere, adding to those
borrowed from other languages such as bombast, chocolate, genteel, and shock.
The second change was in pronunciation, which shifted the pronunciation of vowels to sounds
that we more or less recognise today, known as the great vowel shift. Chaucers pronunciation of
name - /((((/ became Shakespeares /((((/ and /((((/ in todays RP accent. No one knows
precisely why such a shift in pronunciation happened, but one of the reasons given for it is the
speakers sense of their own prestige (see for example, Baugh and Cable (2002). At a time of
growing urbanisation in London and the forming of the modern class system, sensitivity had been
growing to accents which betrayed peoples origins. During the sixteenth century, people from
East Anglia and the Midlands moving towards London did not want to sound like country
bumpkins and so altered their accent to emulate or sound like others. As often happens, when
people consciously adjust their pronunciation to emulate the speech of others, they
hypercorrected their pronunciation, That is, in being consciously aware of their accent in a way
which requires imitation of another, a person often over-emphasises the sound they wish to make
for example, putting an h in before vowels as in whonder and whay.. However, these
hypercorrected accents gradually became an indicator of the rising wealthy merchants and
would themselves be copied or emulated by others. Thus, what was originally an accidental
innovation became prestigious and spread, although such a vowel shift did not occur in all
regional dialects, and accounts for some of the variation that continues to this day.
This new variety of English, standard English, though originally geographically rooted in the
Middle English area and dialect of East Midlands, rapidly became dissociated from its regional
origins. Indeed, the emerging form of standard English was viewed more in terms of a foreign
language, designed to function as a lingua franca (that is, a language which is used internationally
in trade or business), in the same way as Ancient Greek and Latin, the languages which became its
blueprint. The underlying theory of prescriptivism upon which the standardisation of English was
based, really took hold during this period, especially in deriving a grammatical structure. This was
based on the notion that the description of a language should abide by a set of rules prescribed for
it, rather than, as is the case with modern grammar, be derived from a description of how it
actually works. In England then, from the very beginning, the codification of English was marked
by the theoretical foundations of prescription which inextricably linked language, social class
and moral value (see TL:DE Part 1 4.1). Throughout Renaissance Europe, questions of language
were tied to questions of literary language, and linguistics as it existed then was the study of
language in general, particularly focused upon the study of poetic language. In Europe,
standardisation of language distinguished between literary, including poetic, language, and the
form of language used for transactional purposes. In England, by contrast, writing about literary
language doubled as a contemporary theory of English usage, as though the two styles of language
were, or should be, the same. It was the written, literary form of standard English rather than any
other that formed the basis for its codification. This entrenched even further the link between
standard English and social class, since its use demanded a degree a literacy which was not
available to the vast majority of the working population. Education thus played an important role
in the standardisation process, a role which continues to be important in its reproduction and
maintenance.
Having been selected and elaborated, standard English then began to be codified in dictionaries
and grammars. As social institutions such as the law and printing came to use English, so the first
dictionaries and grammars of English were written, with the purpose of identifying and
disseminating the best forms of English. The idea that the best English was that spoken by the
best people as defined by the nobility gave rise to the notion that therefore, anything else was
not only linguistically but also morally and socially inferior. As a result of standardisation, all
other dialects, Northern, Southern and Western, were treated as second class and ignored, unless
written as poetry. As living dialects however, they did not fossilise, but continued to be used by
the inhabitants of those areas. Towards the end of the Early Modern English period, arguments
about the establishment of an Academy in England of the kind established in France (1635) and
Italy (1582) continued. This idea found little public sympathy, since observers noticed that the
French still changed their language despite having an Academy. In England., language was
perceived as an embodiment of the spirit of individual liberty, and of resistance to central
regulation.
Nevertheless, this did not stop people from trying to pin down the language through the means of
dictionaries and grammars, which were first written with the prime aim of fixing the language
once and for all. In 1755, Samuel Johnson published his Dictionary with the aim of making a
dictionary by which the pronunciation of our language may be fixed and its attainment facilitated;
by which its purity may be preserved, its use ascertained and its duration lengthened. It may
seem odd to us now that there was ever a time when words did not have a standard spelling, but
before Johnson published his dictionary, this was the case. Words were spelt in several different
ways without much thought or anxiety attached to such practice as is to be found today;
Shakespeare, for example, signed his own name with five different spellings over the course of his
lifetime.
Johnsons dictionary was by no means exhaustive, and he made no apology for its selective
nature, leaving out as it does any terms to do with manufacturing, law, medicine and the physical
sciences on the grounds that: I could not visit caverns to learn the miners language, nor take a
voyage to perfect my skill in the dialect of navigation, nor visit the warehouses of merchants, and
shops of artificers, to gain the names of wares, tools and operations, of which no mention is
found in books. Johnsons principal source of both orthography (spelling) and definition for the
words he included in his dictionary was from written, literary texts.
By the beginning of the Modern English era, from 1800 onwards, England was a
linguistically divided nation. It was divided geographically by regional dialect and accent,
and socially by the growing rise and influence of standard English as a national
language, becoming the language not only of government and administration, but also of
a cultural elite. Paradoxically, as standard English became removed from its regional
origins, it served to increase rather than to decrease class differentiation. During the
nineteenth century, the use of standard English spread throughout the country and all
social classes, especially after 1870 when compulsory education was introduced in
England (1872 in Scotland). Learning to read and to write standard English was an
important part of the state school curriculum. Spoken as well as written standard English
was held up as a model of refinement and, with the expansion of the British Empire, a
carrier of political power world-wide.
During the nineteenth century, language, as much as science or any other discipline,
came to be an object of study among others. For Max Muller, the first professor of
Philology at Oxford, the object of such study was language in general rather than any
specific one. This made it possible to study language as a science, abstracted from the
contexts of its use, a study that was eventually called linguistics. Attitudes towards
language use were validated by claims to scientificness, fostering the notion that any
variation from standard English as a norm was deviant, and therefore improper and
incorrect. Such deviance was a manifestation not only of improper speech, but improper
behaviour. Thus, to speak the superior, refined language demonstrated that one
belonged to a superior class, and that speaking a vulgar language correspondingly
meant that one belonged to a vulgar one. Smith writes that:
The study of universal grammar at that time stipulated that languages were
fundamentally alike in that they represented the mind, and fundamentally different in
the quality of mind and civilisation that they represented...By dividing the population
into two extremes, ideas about language firmly distinguished those who were within
the civilised world from those who were entirely outside it
Smith (1984:2-3)
The grammar and dictionaries of English which came to be written and used in schools
during the first half of the twentieth century continued to be based upon a prescriptive
description of the grammar and vocabulary of Standard English. Grammar continued to
be thought of as a unified, universal concept that described any language, and the
degree to which a language fitted grammatical prescription was a marker of its
superiority over all others, which were therefore deemed imperfect and full of errors.
This concept remained unchallenged until the 1950s when the first modern grammars of
English came to be written, based upon description rather than prescription, a seismic
shift in linguistic grammatical theory that broke with ong-standing tradition.
In 1927 the influential phonetician Wyld argued that RP is superior, from the character
of its vowel sounds, to any other form of English, in beauty and clarity (Wyld 1927:606).
Appealing to aesthetics, however, is hardly a scientific basis upon which to argue
grounds for superiority. Wyld tried to prove scientifically that there was a hierarchy of
pronunciation, with RP at the top as the most a superior accent. His ideas were very
influential in the curriculum for English in elementary schools and the training of teachers
and still persist to this day. For example, when a national curriculum was re-introducedd
into English schools in the late 1980s, and revised shortly after in the early 1990s, and
one of the most controversial issues was around the teaching of standard English and
RP as part of the speaking and listening curriculum (see; Clark 2001).
Although RP is not exclusive to any social class, through its history it has social
associations with upper and upper-middle classes, whilst lower-middle and working
classes have tended to be associated with regional accents. In evaluations of speakers
whose speech differed solely in terms of accent, RP always comes out as having more
prestige, being more pleasant sounding, its speakers are viewed as more ambitious,
competent and better suited for higher status jobs than speakers with regional accents.
Some earlier studies though, rated RP speakers as less attractive: less sincere,
trustworthy, friendly, generous, kind, than speakers of other accents. The prestige
associated with RP is thus a social, not a linguistic, phenomenon and can change.
A further important point to note here is one that Aitchison (1981 & 2001) has made,
namely that change in language is neither degenerative nor progressive: it simply
happens and no amount of fixing can stop this occurring. The fact that English is not
stable across history is because it is a living language, and change is a necessary aspect
of this. Such a view has huge implications for the ways in which language variation and
change have previously been studied, and especially for the methods used to reconstruct
our linguistic past. Similarly, there is no way of predicting what the English language will
sound like in another five centuries, other than it will be different from what we have
today.
Far from dying out, dialects and accents continue to survive, and also to change, as do
our attitudes towards their use. Although it could be said that accents and dialects in
particular, are far more convergent and mutually intelligible than they were say two
centuries ago, there is still a great deal of variation. The processes of change governing
dialects, but more particularly accents, are difficult to determine, but, like the great vowel
shift, are more often than not a result of social change of some kind rather than through
any pattern of linguistic regularity (see 1.4 below) .
The brief outline of the social history of English in England above illustrates how the
processes of standardization were enacted in that language variety, and explains why
regional varieties in England persist to this day and where notions such as correctness
and purity of language come from. The next section considers the processes of
standardization in a different context, and illustrates how although the processes may be
the same, their realisation is very different.
1.3.3 A brief history of the standardisation of English in the USA
American English now dominates the world. It is often standard American English which is
taught to foreign learners around the globe, and it is not unusual to find people from the Pacific,
the Far East and even Europe speaking English with an American accent and using American
idioms. Since the Second World War and from the 1940s onwards, American English has
consolidated its position through its dominance of the entertainment industry and in recent
decades, through its dominance of information technology and the internet. American English,
like that of British standard English, owes a great deal to its social and cultural history.
The main feature of American English in contrast with standard British English is in lexical
variation and variation in pronunciation. There is very little, if any, syntactic variation. Many
words, such as moccasin, jazz, zucchini, bagel, and kosher, have been borrowed from the
languages of non-English speaking immigrants, as well as from the indigenous peoples. Some
recognisably British words have developed variant meanings such as mad meaning angry; are
you through? meaning are you finished? and bad or wicked meaning good. There is also a good
deal of cross-over between the two varieties, American English and British English, in that
American technical innovation has imported into British English words such as program,
telephone and typewriter. Other words express a culture and environment different from the
British with words such as lynch, blizzard, joy-ride, bayou, levy and prairie. So, in British
English, you drive a car, fill up with petrol, wipe your windscreen, check your bonnet and
boot are shut and drive down the road, dual carriage way or motorway overtaking lorries. In
American English, you drive an automobile, fill up with gas, wipe your windshield, check your
hood and trunk are shut, and drive down the freeway, expressway or divided highway, overtaking
trucks. In England people queue, whilst in America they stand in line. Phonetic variation does
exist across the USA as well, but given its relative youth as a country, this is shown not so much
by regional variation of the kind that exists in England. Rather, phonological variation exists on
divisions between urban and rural areas and between different ethnic groups.
The origins of American English date back to the seventeenth century and the colonization by
several disparate European nations of the New World. In 1620 the first English colonists landed
at Plymouth rock, fleeing religious persecution in England. By this time in England, standard
English had begun to emerge and these initial inhabitants of America adopted its use. Thus, it was
standard English which was selected at the first stage of the process of standardisation, rather than
any other variety. As American English developed, and as grammars of English came to be
written, its syntactic structure adopted that of British Standard English. The association between
linguistic variation and social class so prevalent in British English is therefore missing from
American English. In terms of the processes of standardisation then, American English followed
a very different course from that which occurred in England. Differences between British English
and American English are to be found more in its subsequent elaboration and codification than in
the processes of selection and implementation. The initial pilgrim settlers, in order to survive,
enlisted the aid of the native Indians, and, realising the importance of local Indian knowledge for
this survival, acquired Indian terminology which they then incorporated into their own
vocabulary. During this time, words such as moose, wigwam and naiack, meaning a point or
corner from which comes the expression that neck of the woods derives, were part of an
emerging vocabulary. Dillard (1976) estimates that of these initial Indian borrowings, some
17,000 survive within American English to this day.
The English settlers were not the only recent inhabitants on the new continent; the Dutch had
landed at the same time as the first pilgrims. They established colonies along the Hudson Valley,
near to present day New York, and the Germans also formed their own colonies near Philadelphia.
As time went on, the colonists expanded their horizons beyond the East coast and pioneers
penetrated further West and then South into the mainland, encountering colonies of French and
Spanish settlers as they went. As new and unfamiliar terrain and wildlife were encountered, terms
were needed to name them. The French settlers had already named most of the geography and
natural flora and fauna around the Saint Lawrence and Mississippi rivers, which the English
speakers absorbed into their own lexicon, incorporating such words as chowder, bayou, voyager,
apache and brave. The Spanish, who had occupied the Caribbean islands, South and Central
America, moved north through the Gulf of Mexico, also named the geography and natural flora
and fauna around these areas, and words such as alfalfa, pumpkin, canyon, chaparral, corral,
fiesta and stampede were similarly absorbed.
In 1804 two explorers, Lewis and Clark, embarked on a journey across America with the
ambitious aim of mapping and defining the inner territories, which involved labelling everything
that passed before them. Consequently, a new vocabulary distinct from that in Britain was formed
to meet with the demands of a new and different terrain and culture that Americans now inhabited.
Although numerous languages existed in America throughout the seventeenth and early eighteenth
centuries, and despite the anti-British sentiments expressed during the war of Independence (1775-
83), English became the medium used to unite speakers of different languages. In Jeffersons
Declaration of Independence, no official language was named as American, since this was seen as
contrary to the prevailing spirit of freedom and pluralism of the time and also in direct
confrontation with a constitution which advocated egalitarianism. To establish a national language
was seen as redolent of the monarchical policies which the New World had left behind. The
American constitution prevents prescription in two important aspects in the transmission of
national of culture and identity; in language, and in education. Nevertheless, although English has
never been or can be recognised as the official language of the USA, in practice, as the language
of public life, including government, the media and education, its use as the national language
prevails.
The alternative to a national language, it seemed, was a distinctive form of English that would
serve the functional aspects of communication, but at the same time represent American cultural
and national identity. Noah Webster proposed such an alternative, and he was arguably one of the
most important men in the development of the American language. As part of its elaboration,
American English had absorbed words and phrases from the many different cultures inhabiting the
continent. Webster proposed to codify the language through the writing of an American
dictionary. Webster was well aware of the need for a culturally independent society, not one that
was still influenced by Britain. The printed word was the most powerful medium of
communication at the time, and Webster was conscious of the fact that to disseminate and
propagate a new national culture, a new national book trade was needed (Simpson 1986:58). He
feared Americas dependence on foreign nations for the development of their manners and felt
that new American citizens should show their affiliations by buying American literature to
stimulate a national book trade, and to enhance American morality, politics, culture and fashion.
As Webster, quoted in Mencken (1921:12), says: Let us seize the present moment and establish a
national language as well as a national government. As an independent nation our honor requires
us to have a system of our own, in language as well as in government.
Webster resolved to make cultural distinctions between America and Britain, initially by devising
his own system of spellings, which he later changed to a modification rather than a whole-scale
change, realising the impossibility of implementing such a revision. He deplored the foreign-
influenced pronunciations and spellings evident in written English of the time, believing that the
British system of spelling was class-ridden and created divisions in society, since it demanded a
level of educated orthographic knowledge of the language and was thus inaccessible not only to
the lower classes, but also to foreign nationals. As a consequence of his disaffection with the
English aristocracy and its abstruse orthography, Webster resolved to introduce a system of
modified spellings that was more uniform in its sounds than those proposed by Johnson. For
example, he excised the u in the British English -our structure, so that words such as honour
became honor; changed the final spelling of re to er in words such as theater, center, the c
before e to an s in words such as defense, the suffix ise to ize; dropped the doubling of
consonants in words such as travelling and programme, and lost the silent e in words such as
program.
Websters modifications do not appear very radical, but by simplifying spelling he claimed to
have made it easier for a primarily agrarian population to learn to write and thus assist in the
development of an American culture and manner that would distance it from Europe. It would
also be easier for newcomers to learn. The modifications, by their very existence, challenged
British assumptions that the language belonged to them and that its development or change was
exclusively their province.
By the mid nineteenth century, cultural divergence and isolation, coupled with Websters
influence, had created a form of English sufficiently different from that used in Britain. However,
arguments continued on both sides of the Atlantic about the nature and purpose of this new form
of English. In Britain particularly, Americanism became a pejorative term, so much so that in
1842, Charles Dickens wrote in his American Notes that he was at a loss when asked by a waiter if
he wanted his food right away: I need not tell you that the prevailing grammar (in America) is
more than doubtful, and that the oddest idioms are received idioms. However, what were
generally thought of as aberrations were sometimes forms of English that had become outdated
or old fashioned in British use but had continued to be used in America. Americans, as colonists,
had continued using the standard English of the seventeenth century pilgrims and still used words
whose form or meanings had altered in British English. For example, the word sick used by
Americans to mean somebody who is ill in British use was widely used in Shakespeares time; the
past participle gotten used in America had been in use in Elizabethan times. Other terms, common
in Elizabethan England, were: fall for autumn; progress as a verb; mean for unpleasant; and
deck of cards for pack of cards.
By the end of the nineteenth century, American English, in addition to its verbal form, was also
establishing a literary canon, centred upon the works of writers such as James Fennimore Cooper
and Mark Twain who encapsulated the essence of frontier America. Ralph Emerson had declared
that now America was free from the literary constraints of English romantic poetry, a new
national American poet should come to the fore to represent the nations poetic aspirations. Walt
Whitman established himself as this poet with works such as Leaves of Grass (1855). Literary
prominence may seem insignificant when considering the development of a language, but the
establishment of a literary canon is indicative of linguistic recognition.
Throughout the twentieth century, the influence of American English upon the world, through the
influence of firstly the media and more recently information technology, has, transformed it into a
global language, used for economic and communicative purposes around the globe. This has
created a kind of reverse linguistic colonisation and a two-way traffic between the two different
varieties of English. British people today, particularly in speech, use a host of American terms
such as okay, goodbye (from the pilgrims abbreviation of God be with you) and hi. Other
items of American English adopted by the British are:
Lexis idioms
french-fries truck movie fly off the handle
snoop bar(pub) has been(adjective) give me the bottom line
teenager fan(sports) burger pass the buck
soap opera parking ticket download its a walk in the park
commute gameshow phoney it comes with the territory
babysitter downsize gimmick take the bull by the horns
gridlock hospitalize hopefully let the cat out of the bag
Numerous analyses have been undertaken into the lexical differences between the two varieties,
spawning a large dictionary industry. If there are any distinct characteristics that differentiate
American English from British English at present, then it has been in the formation of new
vocabulary in two areas: back formation and word blends. One of the consequences of these
processes is known as tall talk.
Back formation is a process in morphology whereby new terms are created by the removal of
affixes from an existing word. Most of these new creations are verbs. American English already
had many terms formed with the noun suffix ion. Where verbs did not exist, they were created by
deleting the suffix ion and adding an e, thereby verbalising nouns. The first noted example of
this process was the formation of locate from location in 1652, donate from donation (1795)
commute from commutation (1865) and electrocute from electrocution (1889). This process,
though as a zero derivation process, continues with recent additions such as housekeep, burgle,
shoplift and babysit.
Word blends refers to the blending of existing words. This process has always added to the word
stock of a language where previously a term did not exist, but is especially prevalent in American
English. Word blends are the merging of phonological elements from two or more words (see
DL:DE 4.2.3) , such as:
Cablegram from cable and telegram
Travelog from travel and monologue
Newscast from news and broadcast
Telecast from television and broadcast
Motel from motor and hotel
Brunch from breakfast and lunch
Although American in origin, most of these terms have now been assimilated into British
English.
Tall talk refers to the imagery and vivid expression which has been central to American linguistic
development, created through the use of, for example, back formation and word blending. During
frontier times, the nation was a rough and ready one, but frontiersmen, desiring to make the
ordinary seem grander than it actually was, elaborated and exaggerated their descriptions in a
form of vocabulary. spawning a host of bizarre creations in lexical terms as in ripsnitious,
absquatulate, flusticated, elegantiferously and teetotacuiously. Even so, tall talk prevails in
contemporary American English, such as the lexical terms discombobulating, rambunctious,
fantabulous and bodacious.
By far the greatest degree of variation between British English and American English is
in the phonological distinctions of accent. Although American accents are more
homogenous than British ones, nevertheless differences in accent are evident. Labov
(1997), identifies eight distinct American dialects across the whole of the United States.
Consequently, it is no wonder that American accents are distinctly more uniform than the
ones to be found in Britain, and is one of the reasons why American English is so much
more appealing to foreign learners than that of British English. The greatest difference in
the range of accents is in the number of vowel phonemes realised in British English and
American English. Both have the same number of consonant phonemes, which is 24,
whereas RP has 20 vowel phonemes, and General American (GA) has 16. This is
probably due to the fact that many American English accents are rhotic (see 1.4 and
2.3.4 DL:DE). This means that the phoneme /r/ is pronounced after vowels, where a
British accent may omit an orthographic r. Another distinctive area, among others, is
realised in intervocality, where the alveolar plosives /t/ and /d/ are, through a process of
voicing, seemingly merged together and consequently words such as ladder and
latter sound identical.
Although in the United States, unlike in the United Kingdom, the majority of regional
dialects are largely free from social stigma and prejudice, there is one notable exception,
that of African American English. The term African American English (AAE) is an
umbrella term for various descriptions known as African American Vernacular English
AAVE, Black English BE, Inner City English ICE and Ebonics (see 1.2.1). It is spoken by
a large population of Americans of African descent, as the historical discrimination
against African Americans has created social isolation which has intensified its isolation
from other American varieties. There are critics of the variety who equate its use with
inferior genetic intelligence and cultural deprivation, on the grounds that AAVE is a
deficient, or incomplete language, or who believe that different races speak differently
inherently. Such views have no linguistic foundation and it is as unscientific a view in
reference to AAVE as to standard American English, Russian or Chinese. Equally, a
childs race plays no part in the language a child learns to speak, since a white child
raised in an AAVE environment will learn to speak AAVE just as a black child raised in a
white middle-class American family will speak standard American English. Nevertheless,
the position of AAVE and its speakers in the United States, as with, to a less intensified
degree the speakers of regional dialects and accents in the UK, illustrate in two different
ways just how deeply entrenched attitudes and prejudices towards language use are
and the part that social and cultural history plays in the formation of such attitudes and
prejudices.
1.4. Studies in variation and change
1.4.1 Linguistic variation
Early twentieth century research into dialects and accents tended to concentrate
upon descriptions of linguistic variation, particularly lexis and phonology, compiling
linguistic atlases to show the different distributions of different dialect forms. The aim of
such research was to counter a mainstream view in historical linguistics held at the time
that all sound change was regular and had no exceptions. Dialectologists aimed to show
that linguistic change, far from being regular, was in fact irregular and did not affect all
sounds and/or words equally.. Early dialectologists were particularly interested in lexical
variation, and how different words were used to refer to the same thing in different
places. They did this by going out and collecting examples of speech (known as data)
from people in the regions in which they were interested, a process known as fieldwork.
The people who were interviewed are known as informants (see also 1.5.2 below).
One such study undertaken in England was the Survey of English Dialects (1962) by
Stan Ellis and Harold Orton, later edited by Harold Orton. The impetus for Ortons
survey, in addition to the reasons cited above, was his fear that as an agricultural way of
life was fast becoming mechanized, then lexical items associated with rural life would die
out, and he wanted to capture them before they did so. The fieldwork for the project was
undertaken during the 1950s and 1960s in over 300 rural communities throughout
England, and interviewed about a 1000 men. The criteria for selection was that at the
time of the survey they were aged 60 or above, with little or no formal education, still
lived in the area where they had been born and had not travelled much. This section of
the population was most likely to be speaking in dialects which had escaped the
processes of standardisation and thus still bore a resemblance to the Middle English
dialects as described in 1.3.2 above. Ortons survey was thus intended as a resource for
linguistic historians, who could investigate the phonology, lexis and syntax of Medieval
English. The survey has since been used to reconstruct spoken English in the times of
Shakespeare, Chaucer and even earlier and is still used a resource to this day.
However, the ways in which data was collected for the survey began to be severely
criticised from the 1960s onwards, both in terms of its methods and its sampling: that is,
the range of factors such as age, social class, geographical location such as urban and
rural, gender and ethnicity of the informants from which data was gathered. The main
method used was a long questionnaire, usually with one word answers to questions such
as You sweeten your tea with?, which were then transcribed phonetically by the survey
worker. Critics argued that one word answers were too divorced from everyday use to
provide an accurate account of how people actually used language. They also took issue
with the sampling, arguing that dialectology should not restrict itself to such a small
segment of the population who were old, rural and male, but should also be concerned
with young men and women, and urban as well as rural areas. As a result, traditional
dialectologists have made changes both to their methods and to their sampling of
informants. The invention firstly of sound recording and portable tape recorders and more
recently video recorders, digitisation and computer technology have revolutionised
research into dialectology. Most books now published in the field are accompanied by a
CD or reference to websites and electronic databases. Consequently, dialectologists
have developed more interactive techniques of data gathering such as those undertaken
by the Survey of Regional English based at the University of Leeds, also used as part of
a large-scale web-based project called Voices, sponsored by the BBC see also 1.3.1
above). For example, data collected as part of the Voices project has identified at least
ten alternative words or phrases for the verb to play truant. These are: skive, bunk off,
wag, skip, mitch, dog, hookey, twag, sag and nick off. Although most of these phrases
can be found across the United Kingdom, twag is specific to Hull and Doncaster, towns in
the North East of England, cap to Derby and Nottingham, cities in the Midlands area of
England, and skidge to Paisley in Northern Ireland.
Traditional dialectologists have also extended their sampling to include men and women
of all ages and social backgrounds within particular geographical areas, and urban areas
as well as rural ones They are much more careful these days to ensure that they have a
representative sample taken from categories of age, gender and social background from
a particular region in selecting their informants from whom they gather data and
subsequently analyse. Consequently, the kinds of surveys exemplified by the SED, far
from dying out, are still being compiled around the world. In addition to lexical variation,
traditional dialectologists have also been concerned with phonological variation. The
methods of data collection and sampling used here are the same as for lexical variation,
but the focus of analysis is upon sound rather than word. The basic conceptual tool here
is that of the linguistic variable, a term used to mean a linguistic item that has
identifiable variants. There are at least two basically different kinds of variation. The first
is in clearly distinct variation. For example, words such as laughing and going can be
pronounced as laughin and goin. The final sound of these two words is the linguistic
variable (ng) with its two variants [(] in laughing and [n] in goin. Another example of a
linguistic variable is in words like car and farm, which can be be given r-less
pronunciations: cah and fahm. Here, there is the linguistic variable (r) with two variants [r]
in car and ( (pronounced zero) as in fahm (see also 1.4.2 below). The second kind of
variation is the quantity of nasalisation, for example in a word such as bend. The vowel
e in this word is sometimes nasalized and sometimes not, and the amount or degree of
nasalization can be noticeably different. Here, we have the linguistic variable (e) and a
number of variants (see also 1.5 and DL:DE Chapter 3).
Linguists who have studied variation in this way have focused on a number of linguistic
variables, such as those outlined above. Studies of variation employing the linguistic
variable have also been used in studying grammatical variation in addition to
phonological. For example, the presence or absence of (s) in the third person singular:
she goes or she go; the form of the verb be in sentences such as shes glad, she be glad
and she glad and multiple negation, such as he dont mean to hurt nobody and I ay done
nothing (see .1.4.3. below).
Within sociolinguistic research that focuses upon phonology, one sound is selected as a
linguistic variable and used as the dependent variable: that is, a sound against which to
measure other sounds. Other factors such as social class, age, gender, region and
ethnic group are varied, and the dependable variable: that is, the variant, is compared
for each of these variables. In this way, the speech of older informants can be compared
with that of younger ones, and that of men to women, lower class to upper class, and so
on.
Sound recording has also made it possible to investigate patterns of stress and tone in
addition to those of sound when studying speech, known as prosodic features. English
accents vary tremendously in this respect. Wells (1999) points out that many Northern
dialects of English tend not to reduce vowels in unstressed Latinate prefixes (eg con-, ex-
) as much as do RP and other Southern-based varieties. Although relatively little work
has so far been done on dialect intonation, Wells (1999: 91) points out that certain British
accents including Birmingham, Liverpool, Newcastle and Glasgow appear to have
some tendency to use rising tones where most other accents have falling tones. Such
tendencies are noted, for example, by Biddulph (1986: 3), who suggests that West
Midland speech characteristically has a peculiar intonation involving terminal raising in
statements, as well as negative verbs (such as <wor> wasnt/werent) taking a markedly
high tone. Wells (1999: 93) also points out that the working-class accents of the West
Midlands (as well as Liverpool and some New York) characteristically have a
velarised voice quality (that is, with the centre of gravity of the tongue further back and
higher than for other accents
1.4.2 Social variation
In addition to descriptions based upon the linguistic variable which continue, throughout
the later half of the twentieth century, linguists began to turn their attention to the
relationship between language and social class, particularly in relation to accent
variation. One of the first of these was the American linguist William Labov. In order to
investigate differences between accents, Labov (1966), undertook research based upon
the notion of the linguistic variable. For example, one very noticeable linguistic variable
in accents of the English language is rhoticity. This refers to the pronunciation /r/ when it
occurs after a vowel (as in bar, sort, churn) and is usually known as post-vocalic r. How
[r] is pronounced after vowels differs across the English speaking world. For example,
RP is r-less or non-rhotic, whilst Scots English or Cornish is r-ful or rhotic. In the British
Isles, rhoticity differentiates Scottish, Irish, Cornish, West Country, rural East Anglian
and Northumbrian accents from most Midland and southern accents. It is also, in Britain,
regarded as a low-prestige feature, associated with backward rural areas. By contrast,
in most of the USA, rhoticity is prestigious and its absence stigmatised.
Labovs influential study, based in New York City, showed that if speakers are ranked
according to social class, then they can also be ranked in the same order by differences
in the use of certain linguistic variables. One of the most notable of these is the variable (
r ) after vowels in words such as lark and bar. Labov wanted to demonstrate that patterns
of variation existed in New York centred upon the use of the postvocalic /r/. Before
undertaking a large-scale survey, Labov did a pilot study; that is, a small-scale survey to
check and test the research methods used to ensure their reliability before embarking
upon a large-scale survey. For this pilot survey Labov chose three sites, three
department stores in Manhattan. His hypothesis was that sales assistants accents in
each of the three stores would largely reflect that of their typical customers, which in turn
he believed would exemplify patterns of variation across the City. The three stores were
Sakss Fifth Avenue, a high-status store, Macys, regarded as middle-class and middle-
priced, and Kleins, which sold cheaper items catering for poorer customers. Labov
pretended to be a customer, but first he checked what items were for sale on the fourth
floor of each store. He then asked sales people on different floors in each of the three
department stores the whereabouts of items he knew to be on the fourth floor. He chose
this particular floor because fourth floor is a phrase that contains two tokens of
postvocalic [r]. Having asked his question, such as Excuse me, where are the womens
shoes?, Labov pretended to be hard of hearing and followed listening to the answer with
a further Excuse me?. In this way, he obtained two more tokens given in a careful,
stressed way as the sales person repeated fourth floor. He then repeated the exercise
on the fourth floor itself, Labov asked sales assistants, Excuse me, which floor is this?
Once Labov had received his answers, he would move out of sight as quickly as he
could, and write down the pronunciation and details such as the gender, race and
approximate age of the assistant. All three department stores are very large, and in this
way Labov was able to gather data from 264 unsuspecting subjects. Multiplied by four
tokens per informant, this gave a collected total of over 1,000 tokens of the variable (r),
taken over six and a half hours in total. Analysis of the data confirmed Labovs
hypothesis, and showed patterns of variation in the use of postvocalic /r/ according to the
speech style, social class and linguistic context associated with each store. Of all the
employees, 62 percent at Saks, 51 per cent at Macys and 20 per cent at Kleins used [r]
in at least one of the four tokens. All groups showed an increase in the use of [r] in the
more deliberate repetition, and interestingly it was employees of Macys the middle-
status store, who showed the greatest increase. This seemed to indicate, as Labov later
commented (1972: 52): It would seem that r-pronunication is the norm at which a
majority of Macys employees aim, yet not the one they use most often. On the quieter
and often more expensive upper floors of Saks, the highest-ranking store, the
percentage of [r] was also much greater than on the much busier ground floor.
Labovs pilot study showed that /r/ could be studied systematically. In the full scale
survey that followed, a more representative sample of informants was used, by
undertaking sampling procedures similar to those employed by sociological surveys. He
divided his informants into six different groups, principally on basis of age and socio-
economic class, and his findings confirmed those of his pilot study, in that his upper-class
informants pronounced non-prevocalic /r/ most consistently, whereas by contrast, lower-
class informants pronounced it the least. In New York City postvocalic /r/ was associated
with high status and prestige. He argued that his research showed language change in
process, in that pronunciation of the postvocalic /r/ was spreading. Like other researchers
into pronunciation, Labov found that it was lower class women who were responsible for
the diffusion of /r/ pronunciation in the community, on the basis that they were particularly
conscious of its prestige value.
Like the sales assistants in middle-class Macys, Labovs full-scale study also discovered
that lower middle class speakers, in more formal speaking styles, showed a greater
increase in the use of [r], a phenomenon known as hypercorrection (see 1.3.2). These
speakers vocalisation of [r] went beyond that of the highest status group, in their
tendency to use a pronunciation considered to be appropriate for formal styles. This use
can be attributed to the position of the lower middle class in the class hierarchy, where its
members wish to distance themselves from the working class and to become more like
the upper class.
Trudgill (1974) replicated Labovs methodology in undertaking a study of non-standard
variants in the city of Norwich in the East of England. Like Labov, Trudgill aimed to describe
the norms of a whole city by detailed interviews with a sample of its population, in this case fifty
adults and ten schoolchildren. He analysed several linguistic variables of grammar and accent. For
example, in Norwich, as in other parts of England, there are two alternative forms for the third
person singular present tense: she runs, walks, skips and so on and the local dialect form with out
the s inflection: she run, walk, skip and so on. Trudgill found that there was a correlation
between social class and use of this variable. His findings across all variables confirmed those of
Labov, in that members of the lower working class used them the most often and that the lower
middle class produce relatively more of the prestigious forms than do members of the social group
immediately above them on the social scale.
Like Labov, Trudgill distinguished between overt and covert prestige. The issue of
prestige is generally an important but complicated one in sociolinguistics. Overt prestige
refers to the positive or negative assessment of a variants or a speech variety in
accordance with the dominant norms of educational institutions, public media and upper
middle-class speech. In Labovs New York City study, informants who made the highest
use of stigmatised forms showed the greatest tendency to stigmatise others for their use
of the same form. However, working class informants did not adopt middle class norms
and the stability of working class norms calls for other explanations than those
associated with prestige. Covert prestige refers to this set of opposing values implicit in
lower and working class lifestyles that do not appear in conventional subject-reaction
tests. Working class speech is thus seen as a mechanism for signalling participation in
local norms and values, whereas middle class speech reveals a concern for social status
and is thus a mechanism for solidarity. However, in his study, Trudgill also correlated
gender as well as age and social class, and found that gender played a significant part in
the use of variant forms . He found that within each social group, his female informants
consistently used less of the non-standard variation than men. They also had a tendency
to over-report their use of prestige forms whilst male informants were more likely to under-
report them. Women in Norwich seemed more responsive to the overt prestige of the
standard variety, while men seemed more responsive to the covert prestige of localised
Norwich speech. Thus, whilst Labovs study showed clear social stratification by status in
New York City over the variable (r), Trudgills study emphasises the difference gender
makes in terms of solidarity with a local variant norm. The covert prestige of the
vernacular is thus a counterbalancing force to the overt prestige of the standard variety.
Account has to be taken of the fact, however, that Trudgill wrongly charaterised women
by their husbands social class, for which he has been criticised. He concluded that
differences in attitude between the two countries reflects differences in class
consciousness at the time, particularly the lack of any militant class-consciousness in
the USA as compared with that of the UK, and the relative lack of embourgeoisement of
the British working class.
1.4.3 Variation and social networks
By the late twentieth century, studies undertaken into urban dialectology by linguists such as
Labov and Trudgill were challenged by work undertaken by the Milroys ([Link] 1987; J.
Milroy 1992). They took issue with the notion of the linguistic variable and variation theory in
general. A study undertaken in Belfast led to the conclusion that the relationship between
linguistic and social structures is not necessarily best examined by an exclusive exploration of
social variables such as age and class. Instead, they based their research upon the notion of social
networks. Rather than grouping speakers into predetermined categories such as social class, the
social network situates an individual within the sum of his or her relationships, both formal and
informal; with other people, such as family, friends, work colleagues; neighbourhood and
relations based upon ethnicity. They found that particular kinds of networks will either inhibit or
advance linguistic variation in a community.
For example, men in the Ballymacarett district of Belfast who worked in local shipyard with
dense social networks bound by strong ties were more likely to enforce the prevailing linguistic
norms and less likely to tolerate or encourage change. This differed from the women who, by
contrast, often travelled outside the area to work and had fewer local ties. This points to their
conclusion that dense, multiplex networks act as brakes on linguistic innovation, whilst weak ties
between individuals seem to make the spread of linguistic innovation, such as new pronunciation
or new words from network to network, easier. The Milroys argue that peripheral members of a
community who have ties with other communities (through work, study or friends) effectively
conduct or carry innovation into a community. Their conclusions complement Labovs arguments
about change in progress, but provide far more information about the site and possible
explanations of change.
Virtually all of the work undertaken in social and urban dialectology until this point had
concentrated upon variation in phonology. Milroy and Milroy (1993) sought to redress this
balance. The aim of their book was to raise the status of English regional dialects and to draw
attention to their syntactic and morphological characteristics. To give a morphological example,
in standard English, the past tense forms of break and know have been codified as broke and
know. Verb forms such as breaked and knowed, have been part of the English language for
centuries and still exist in regional dialects, but since they were not chosen as the past tense forms
for standard English, are no longer used in writing or educated speech. A syntactic example is
that of multiple negation: she never said nothing. This pattern was outlawed by nineteenth century
prescriptive grammarians, but was common in the middle English of Chaucers time, used by
Shakespeare and continues to be used by native speakers of English today as it remains a
grammatical feature of some dialects.
For example, in some English dialects thee, thy and thou are still to be heard instead of you.
Other dialects retain forms of the verb be which is different from that of standard
English, such as the Black Country dialect in England, centring upon the Midlands towns
of Wolverhampton, Dudley and Walsall. For example, in the affirmative present tense:
I am(am) I am (am)
You/Thou am (Yam) you are
S/he is s/he is
We am we are
You am (Yam) you are
Such variation, because of its high degree of dissimilarity and variation from standard
English grammar, has been interpreted as an indication of ignorance, lack of intelligence
and as a bastardised form of English. Such attitudes can be traced back to the ways in
which English was standardised in England, summarized in 1.3.2 above. However, the
Black Country dialect, far from being a bastardised form of English, has a history and
pedigree dating back to the Middle English dialect of the West Midlands. The Black
Country dialect is also noted for its highly contracted negative modal forms such as aint
(am not/isnt/arent, hasnt/havent ) and doe/doh/dow (doesn t/dont).
(adapted from Clark 2004:261)
On-going investigations into the Black Country variety (Clark, forthcoming) have shown
that syntactic forms discussed above are all still being widely used by men and women
from all kinds of social backgrounds and different ages, particularly in social networks
that are informal and localised, such as talking to friends and relatives. This is the case
even where speakers have been educated to a level that includes higher education. On
the whole, consistent use of most or all Black Country grammatical forms listed above is
most prevalent amongst older speakers of the fifty plus generation. Younger speakers,
especially adolescents and those of the twenty to thirty age group, tend to use them
less, mostly the verb be in its various tenses and its negative form. The use of dialect
amongst adolescents has been the focus of recent studies, particular Eckert (1988 & 1989)
and Kerswill & Williams 2000 & 2002). Eckert in her studies involving American adolescents,
suggests that they are linguistically the most innovative of all generations and, through their peer
groups, establish new norms that may diffuse to the wider community.
Kerswill and Williams investigated the role of adolescents in dialect levelling in three English
towns: two in the south, Reading and Milton Keynes, and one in the north, Hull. Dialect levelling
is a term used to describe changes in dialects that make them considerably more like standard
English in phonology, grammar and vocabulary.
Kerswill and Williams examined phonological and grammatical variables and compared of
teenage and elderly subjects, across middle and working classes and males and females. Their
study showed, for phonological features, convergence between the two southern towns, whereas
convergence between the southern towns and the northern one was limited to some consonantal
features (see also 1.4.1 above). Kerwill and Williams also tested Milroys claim (L. Milroy 1987
& [Link] 1992) that network characteristics rather than class which account for the more
levelled, standard speech of the middle classes than the working class. They found that, whilst the
Milton Keynes working class speakers were more linguistically levelled than their Reading
counterparts, they remained strongly non-standard and non-RP in grammar and pronunciation.
They ascribed this to a lack of social, as opposed to geographical, mobility and to the maintenance
of class-based cultural differences surrounding relationships with schools, literacy, authorities
and employers.
In the same study, Kerswill and Williams also investigated adolescents attitudes to language and
social groups. They found that when discussing aspects of teenage culture, language issues evoked
the strongest response. They expressed strong allegiance with their own peer group, framed in
terms of in-group and out-group labels, and usually with reference to a hierarchical class structure.
They also found that adolescents self-identifications and their linguistic use did not stand in a one-
to-one relationship: that is, what adolescents said about the way they spoke and how they actually
spoke themselves did not always match. This led them to conclude that adolescents are partly,
but not fully aware, of language change processes. However, they found a uniformity of attitudes
towards language issues in both northern and southern schools, suggesting that part of the
mechanism of levelling lies in these subjective parameters.
What all the studies summarised in this section show is that both linguistic variation in language
use and processes of language change are inextricably bound up with issues of class, gender, age
and culture. The study of variation and change, then, involves not only linguistic description, but
also consideration of social and cultural issues in accounting for variation and processes of
change.
1.5 Studying variation in English
The earlier sections in this chapter introduce issues in studying variation and change in
English. As they have shown, the focus of research in linguistic variation is upon
identifying differences in the actual production of language in one or more of the three
areas of pronunciation or accent, vocabulary and grammar. The focus of research into
language change has normally been across different generations. A further associated
area of research is into attitudes towards variation.
A purely linguistic description of variation in itself does not explain why variation occurs.
For such an account to take place, notice has to be taken of non-linguistic variables
which derive from the social context or networks within which linguistic variation takes
place.
What we mean by a variable is a feature which has a number of possible values, or in
the case of linguistic variables, we might say that they have a number of potential
realisations. Thus, the phoneme /t/ can be said to have a number of possible variants
(or allophones), including at least [(], [((], [((] and [(] (glottal stop). See chapter 2 of DLDE
for further information on phonological variation. Any investigation of the variable t,
therefore, would be looking to collect examples of possible /t/ phonemes, and see
whether there is any patterning in the occurrence of different variants, perhaps in relation
to syllable structure, word structure, or in the case of glottal stop, issues of formality of
setting or class of the speaker. In the case of non-linguistic or social variables, the
variants will be the different social classes informants may belong to, their age, their
geographical origins and so on. Thus, the variable itself would be known by its generic
name, such as social class and the variants would be lower-middle class, upper-working
class etc. Much of the discussion and advice that follows will refer to linguistic and non-
linguistic variables in the sense just defined.
1.5.1 Selecting a topic for research
The first thing to decide is the specific question you wish to address. For example, you
may wish to investigate variation:
> in a particular geographical area;
> in a particular ethnic group;
> between two neighbouring areas;
> across different age groups;
> across different social classes;
> across different ethnic groups;
> the same person in different social settings
> attitudes towards variation.
The second thing to decide when undertaking your own research is its focus:
Investigating variation in a particular geographical area will involve you in selecting
the area upon which you wish to concentrate, deciding upon the linguistic
variables you wish to investigate and selecting the nonlinguistic variables for your
sample of informants (see 1.5.2 below).
Investigating variation between two neighbouring areas is similar to the first
example above, but two regions are chosen instead of one. For example, you may
wish to investigate two neighbouring villages rather than one town or city, or two
neighbouring conurbations of a city.
Investigating variation in a particular ethnic group will involve you in selecting the
ethnic group you wish to investigate. If you want to add a social dimension to your
study, then you will need to consider the settings within which you wish to
undertake the investigation, for example, a domestic setting or in a workplace and
the impact this has upon language use.
Investigating variation across different age groups, genders, social classes and
ethnic groups will involve you in identifying the ages, social classes and/or ethnic
groups you wish to investigate and how you intend to restrict these non-linguistic
variables to make the project manageable and the results clear. Investigating
variation across different age groups, particularly if you include the oldest
generation through to children in your study, lends itself to investigating language
change over the time span of half a century or more.
Investigating attitudes towards variation should not be confused with
investigations of the linguistic variation itself, since how people think they speak
and how they actually speak can be very different. This kind of topic is usually
best undertaken by a questionnaire, and one which includes dialect recognition as
part of its design, or as a separate element (see 1.4 3 above and 1.5.3 below).
For all of the examples above except for the last, and for any study into linguistic
variation, once you have decided upon its focus and scope, you then need to decide on
the specific linguistic variable(s) you wish to investigate and any non-linguistic variables
too.
If you choose to investigate phonological or grammatical variation, then you will have to
decide whether you are going to choose specific variables to focus on, or whether you
are going to let any potential linguistic variable emerge from your data. Investigating
lexical variation will involve discussing situations where lexical variation is most likely to
occur, for example, topics relating to food and drink, or discussing word lists of some
kind.
Non-linguistic variables include factors such as geographical origin, age, gender, social
class, ethnicity and so on. The decisions to be made here involve selecting specific non-
linguistic variables and the extent to which you control each one. For example, choosing
informants who are all of a similar age, geographical location or gender means that
instead of investigating differences between informants with different backgrounds, you
will have provided a control for those variables which are therefore invariant. However,
the extent to which you control each of these non-linguistic variables will depend upon
the aim of your study. For example, a study looking at one age group ironically finds out
least about what is specific to that age group, because it has nothing to compare its
results with. This is why a study which considers linguistic variation in relation to age is
best undertaken from informants of different ages. From a practical point of view, though,
it is not sensible to investigate variation across different age groups if you do not know
many people from different age groups. If your study relates to geographical area,
which it most likely will, then you need to decide on the extent of that area and its
boundaries. Your access to it is also important. Equally, if you select gender as one of
your non-linguistic variables, then you must be sure that you know enough people of
either gender who would be willing to act as informants. Undertaking fieldwork is fraught
with enough pitfalls without the added complication of setting yourself unrealistic or
overambitious parameters of non-linguistic variation (see discussion on sampling in
1.5.2 below). For this reason, and also to avoid having to carry out overly complex
statistical testing, the suggested list of topics below confines itself to one or two non-
linguistic variables.
Once you have selected a topic and the focus for your research, then you are in a
position to design your project and collect the data required for your investigation.
1.5.2 Project design
Your project should be carefully designed to reflect your research question, and will usually fall
under the general heading of a quantitative or a qualitative study. The former aims to quantify the
number of variants of a linguistic variable, and sometimes also to plot these results against non-
linguistic variables. The latter considers the nature of linguistic features in context, and normally
explores a text or set of texts in more depth, often considering a range of variables, rather than
restricting the study to a narrow selection.
So, for example, a quantitative study may involve the investigation of h-dropping in different
generations of Londoners, and would need to obtain data from a roughly equal number of
different informants within a set of pre-determined age-ranges such as the following:
| |18-27 |35-44 |52- 60 |70+ |
|Number of | | | | |
|informants: |20 |18 |22 |16 |
We will look a little later at the issues involved in analysing data arising from a project like
this one. For now, let us consider a typical qualitative research project, in order for the
distinction to be clear. Whilst the h-dropping project focuses on a single variable, in this
case the pronunciation (or dropping) of the glottal fricative, /h/, a qualitative project might
wish to look more broadly at the phonology of a particular speaker or speakers who
share a similar background, such as geographical or ethnic origin. This would probably
involve a longer recording of individual speakers, rather than focussed data which aims
to elicit the particular variable, as in the case of the h-dropping study.
The precise nature of the data you collect is not entirely determined by whether your study is
quantitative or qualitative, though there are tendencies to use different methods of data collection
to produce either statistical or contextual results. We will consider both the different kinds of data
collection and also their correspondence to categories of research type in the next section. Before
we do so, however, we will introduce some very important principles of how to choose the
informants that you use for research into language variation, particularly, and most vitally, where
you are planning to carry out a statistical or quantitative study.
The way in which you choose informants or the principles of selection are known as sampling.
Sampling can be of three different kinds: Random, structured and opportunistic. If you wish to
establish the patterning of linguistic usage in a particular group, for example, the English
morphology of New Yorkers, you may decide to use random sampling to acquire your set of
informants. This use of the term random is a very precise and technical one, and does not relate
to the colloquial use of the word which may indicate an accidental set of occurrences or items.
The random sample in this more technical sense refers to set of informants (for our purposes) who
have been chosen from a well-defined population by the use of random number tables or computer-
generated random numbers. For the above project, then, the population would be the whole of the
population of New York city, and in order to be sure that your sample was representative, you
would need to choose people from the electoral register or phone book, or some other
comprehensive listing, using random numbers based on the total population, to make sure that
everyone has an equal chance of being picked in the sample. You will appreciate that this is not an
easy route to take, and you also have to build in certain factors, such as the reluctance of some
people to take part in surveys and interviews. However, it is possible to imagine a much smaller
population such as the whole first year of a University course, and still use the random sampling
technique, to make sure that you are not skewing the results in favour of, for example, people who
you know, because it is likely that they will not be representative of the whole population.
Although random sampling is the most reliable method of getting a representative sample, there is
another kind of sampling technique which suits many research projects on variation, particularly
where you wish to compare the linguistic usage of informants with different ages, social classes or
ethnic or geographical origins. This is known as structured sampling, and reflects the desire to
create a balanced sample, using non-linguistic variables to define the different groups of
informants. Thus, we may decide that we wish to study dialect words for small, individual
portions of bread, baked as a separate unit (i.e. not a slice of bread). The very many possibilities in
British English include bap, roll, breadcake, teacake and barmcake, and these are the variants that
the research would be investigating. In order to establish whether there are regional differences,
we may decide to take a sample of people in cities in the South West, South East, East and West
Midlands and the North East and North West regions of England say, Bristol, London,
Leicester, Wolverhampton, Newcastle and Manchester. These six samples will be made up of
those informants that we are able to get hold of and who are willing to be part of the project.
Whilst in an ideal world, they might also be chosen from those populations by random sampling,
the enormity of the task would probably preclude this. The important issue for structured
sampling, to make sure that the different categories of informant are able to be compared, is to
make sure that there are a minimum of 5, and preferably more, in each group. So, to return to the
h-dropping project discussed earlier, we have a minimum of 16 in the categories, and their results
are therefore comparable. If we were to add another social variable, for example, gender, to see
whether this linguistic feature is more marked in men or women, we would immediately double
the number of informants needed:
| |18-27 |35-44 |52- 60 |70+ |
|Number of | | | | |
|male |20 |18 |22 |16 |
|informants | | | | |
|Number of | | | | |
|female |22 |16 |20 |18 |
|informants | | | | |
The other issue that arises when more than one social variable is introduced as well as
the linguistic variable is the difficulty of analysing the results. For quantitative projects,
you would need to perform some fairly sophisticated (multivariate) statistical tests on
data of this kind, and before you do so, it is recommended that you read up on some of
the basic statistical principles (see section 1.6 for suggestions and Woods, Fletcher,
Hughes & Anderson 1986).
Apart from random and structured sampling, the other form of sampling is opportunistic, and as
long as you are not trying to make great generalised claims for your findings, and are thus
probably carrying out qualitative research, then this is fine. Opportunisitc sets of data will be
collected from friends, family and others with whom you are in contact.
1.5.3 Methods of data collection
A very common form of quantitative data collection is to use questionnaires or surveys. This is
because the same questionnaire or survey can be sent out to a large number of people to gather
information from as many informants as possible. Sound recordings of interviews and
transcriptions based upon dependant or pre-given variables can also be used as a quantitative
method, particularly if you are investigating phonological variables. For example, in Kerswill &
Williamss 2002 study summarised in 1.5.3 above, 100 adolescent participants were tape recorded
across the three towns drawn from two secondary schools in each area., totalling about ninety
hours of tape recording. This data was compared with recordings taken from four to six elderly
participants in each area from a working class background and who had lived in the area since
birth, plus data from the SED (see 1.4.2 above).
Qualitative methods of data collection are used in studies where more in-depth study is required,
normally gathered from a small sample. The most usual form of data collection here is interviews,
usually taped.. In Clarks 2004 study of the phonology of the English West Midlands, five
informants were tape recorded totalling two hours of tape recording, and the data was compared
with previously published findings such as Wells(1999) and Biddulph (1986). Comparison of this
kind thus adds a limited quantitative dimension to a study. Whereas questionnaires can be sent out
to a large number of people, interviews by their very nature are time-consuming, and most
research studies do not have the time or the resources to conduct large numbers of interviews.
Talking to people face to face also allows for flexibility in a way which is not possible with
questionnaires. It is important to remember that a method of data collection is not of itself either
quantitative or qualitative. Rather, the terms refer to the ways in which you use the methods and
to the size of your sample. For example, you could design a questionnaire to send to ten
informants, and then follow up the questionnaire with interviews, interviewing five of the ten
informants. Both methods of questionnaire and interview would be used in a qualitative way.
Similarly, you could design a questionnaire to send to five hundred informants, in which case you
would be using the method in a quantitative way. The difference between the two is not only
one of data collection, but also of aim. If your aim is to investigate variation across two or three
different regions, such as phonological variation in three towns such as Kerswill (2004) or across
the British Isles such as Widdowson & Upton (2006) or grammatical variation across the British
Isles as in Cheshire (1993), then quantitative methods are probably best. Equally, if your study
involves attitudes towards variation, then the more people who give you that information the
better, and questionnaires are also appropriate here. For example, a student of mine undertook a
study into attitudes towards local accents by employers. The focus of her study was call centres,
and she sent questionnaires to the Personnel departments of ten different national call centres, to
find out what account was taken of regional accents in determining employability.
If your aim is undertake an in-depth study of linguistic variation in a specific geographical
location, then qualitative methods can be the most appropriate. Thus, although Kortmann &
Scneiders (2004) two volume publication investigates linguistic variation across the world, each
chapter investigates a specific locality and the methods used in most cases were qualitative ones.
For example, a local brewery requested that research be undertaken into lexical variation for
words to do with beer and drinking which they could use in an advertising campaign, which was
targeted at men. They wanted to make sure that the dialect words and phrases they used were not
only authentic but also conformed to dialect spelling. This project involved students employing
qualitative methods of interviewing and recording, talking to men in local pubs about their
drinking habits.
The main methods of data collection you choose then, will depend upon your research question
and the focus of your study.
Structured sampling can also be used in qualitative studies, although the size and range of the
sample will not be sufficient for any worthwhile statistical analysis to be undertaken. For
example, the following table shows the distribution of informants for one qualitative study:
.
| |British Asian |British White |
|Male |1 |1 |
|Female |1 |2 |
The topic under investigation here is phonological variation between different ethnic
groups across both genders in a specific geographical location. The number of
informants is five in total, which is normally a minimum in each category for the purposes of
comparison, and the number of non-linguistic variables is two; gender and ethnicity. It is sensible
to restrict non-linguistic variables in a qualitative study of this kind to a maximum of two, since
this makes the focus of your study clear. Too many non-linguistic variables in such a small range
of informants could mean that it is difficult to draw any worthwhile conclusions.
Whether or not your sample is random or structured, it is important to bear in mind that five long
interviews or less are not really sufficient to provide you with enough data upon which to draw
general conclusions about social variability. Any conclusions that are drawn would have to be
provisional upon a larger study in order for them to be worthwhile. Thus, for the study into dialect
words and phrases for beer and drinking mentioned earlier, students interviewed informants in
pairs, and each pair of students interviewed five informants. The data they gathered provided
information on lexis, grammar and pronunciation, but the data gathered by each pair of students
was not sufficient to draw conclusions about how such information linked to social networks or
social class. However, the data gathered in this way by ten pairs of students produced data from
ten different interviews and fifty informants over a period of two weeks. In this case, sufficient
data has been produced for analysis and conclusions about social variability to be drawn. In
another study, a doctoral research student investigating variation of register in which social
variation was also a factor, recorded forty children aged between eight and twelve over a period of
eighteen months in six different primary schools. Data for a study involving social variability can
thus be gathered by various researchers interviewing informants at the same time, or by one
researcher interviewing and/or recording informants over a period of time, of about a year to
eighteen months.
The biographical information of your informants is a crucial source for providing the contextual
information within which your study is located. Consequently, it is important that you obtain
biographical details of each of your informants. This information will include not only standard
items such as name, age and address but also factors which may have a bearing on language use
such as level of education, length of time spent in the locality and the birth place of parents or
even grandparents. The best way of obtaining this information is to design a biographical sheet
with all the categories on it you wish informants to fill in, and ask them to complete it either at the
start or at the end of your interview. The advantage of this is that you have an easily available
written record to which you can refer. Alternatively, you could ask for the information as
questions at the start of an interview by way of an ice-breaker, although this will involve you in
additional form filling or transcription after the event.
Once you have established your variables and your sample, then you need to collect your data.
The main methods used for this purpose are questionnaires and/or recording of some kind.
Questionnaires
If the main method of your data collection is a questionnaire, then you will need to issue sufficient
questionnaires to ensure a return that is going to be of any use. As a rough guide, expect a fifty
percent return. People do not return questionnaires they are sent for a number of reasons,
including lack of time or lack of interest. This means that to obtain ten completed questionnaires,
you will need to send out twenty.
A written questionnaire may be used as a main source of data collection when investigating
grammar, lexis or attitudes towards language use, though it is worth considering the more formal
nature of the written language if you are trying to obtain information on spoken grammar or
informal lexis. It is important when designing a questionnaire that you are clear about the kinds of
questions you want to ask. Closed questions are ones which require a yes/no answer, semi-open or
multiple choice questions are ones which invite informants to choose from a restricted list of
options and open questions invite informants to give their own answers. For example, an
investigation into grammatical variation may include a question such as: Do you ever say I dont
know nothing? This question invites the answer yes or no, and is thus a closed question. A yes
answer will give you information on informants who use this expression, but a no answer does
not allow for identification of any other non-standard negative constructions such as I ay know
nothing. It is thus not a very useful tool for discovering the usage in this area. A semi-open
question would be one which says: tick the following expressions you are most likely to use
followed by a list of negative constructions, and including a final category of other which
captures anything you have not included. An open question would be one which invites
informants to give their own categories and descriptions such as: How do you say no to
something?. It is also important to avoid leading questions: that is, questions which invite a
specific response. For example, a question such as Do you think the Queen speaks good
English? is more likely than not lead to the answer yes. Note that non-standard grammatical
constructions may not be acknowledged by speakers anxious to be on their best linguistic
behaviour, so more indirect methods of discovering this information are useful. These would
include, for example, recordings of informants speaking in relatively informal surroundings.
To return to questionnaires, it is a good idea to have a mixture of all three kinds of questions,
since it is very difficult if not impossible to predict all possible answers to a closed or multiple-
choice question. The amount of time it takes to process a questionnaire, however, should also be
taken into account. Although closed questions limit the number of possible answers, they are the
quickest to process since all they require is an adding up of the responses to each option. Open
questions require more time, since there is potential here for every answer to be different.
Once you have designed your questionnaire, then you will need to pilot it before distributing it.
That is, you need to try it out on a few people to make sure the questions you ask are clear and
will get you the information you want. Piloting your questionnaire will allow you judge how
successful your design has been, and to make any subsequent changes or adjustments thrown up
by the pilot. For example, a questionnaire whose questions were regularly misunderstood by
informants in the pilot study, or where the questions turn out to be leading the informants more
than expected, should be redesigned and piloted again.
Recordings
If you are trying to obtain more natural spoken language, and a questionnaire does not suit your
project, you may need to record informants as your main source of data. This choice, unless you
have a great deal of money and many research assistants to help, will push your project in the
qualitative direction. However, it would in theory be possible to record a large number of
informants in the way described below, if resources were available.
To collect recordings which will generate sufficient data for analysis, you need to record
informants for interviews for up to one hour, depending upon the focus of your research.
Recordings of spoken data can be of different kinds. Different types of spoken data are:
> reading word lists or passages aloud;
> interviews between you and the informant(s);
> conversation between informants;
> conversation or interview about a written questionnaires or other form of written data such
as a spidergram (see 1.5.3 below). .
Reading aloud involves selecting material which all informants will read. This method of data
collection restricts you to identifying phonological variables, although the more formal nature of
writing may affect more naturally occurring speech. Even so, reading a passage could test, for
example, the pronunciation of /t/ in different contexts, including, for instance, syllable-final glottal
stops, and comparing this with the reading of a word list, which is more formal still, can give an
idea of the variation according to social context. Usually, where reading aloud is used as a form of
data collection, then it is accompanied by another form of data collection which produces more
naturally occurring data, such as interviews.
Interviews involve you in preparing questions in advance which you then ask of all your
informants. Interview questions usually fall into one of three categories: structured, semi-
structured and unstructured or open. Structured interview questions are ones where all the
questions asked of informants are the same, asked in the same order, with no room for any
additional or supplementary questions. Semi-structured interview questions are ones where most
of the questions are the same and asked in the same order, but with additional or supplementary
questions added. Unstructured interviews are ones where no questions are determined in advance
and conversation, rather than an interview, is left to take any direction. For example, a
structured interview will have, say, five questions which every informant is asked and has to be
asked, in the same order. For example, in the study students undertook on behalf of a local
brewery, then a structured interview was undertaken, asking all informants the questions: Is this
your local? What beer do you drink? How does x beer compare with others? What words or
phrases do you associate with drinking beer? With drinking x? The advantage of this kind of
interview is that it makes comparison across the questions for the purposes of analysis
straightforward, particularly if you are interviewing large numbers of informants. The
disadvanatage is that it does not allow for any interesting answers to be developed or pursued.
A semi-structured interview will have say, three main questions with supplementary questions
added designed to draw out further information and scope for digression within the framework of
the main questions, should it occur. For example, main questions can centre upon recent events
such as: Do you go to the cinema often? What was the last film you saw? What is your favourite
film? Each of these questions could have additional ones such as: do you go locally or elsewhere?
Do you enjoy films by that director and so on. The advantage of this kind of interview is that it
does allow for discussion to develop, but may make comparison of date other than the answers for
the main questions more time-consuming.
An unstructured interview has no pre-determined questions. The advantage is that discussion can
be more free-flowing and not tied to answering particular questions; the disadvantages are that
should the interview dry up, then the interviewer is thrown back on her/his own resources and the
data generated might not have sufficiently similar items or topics to make comparison possible.
For the purposes of sociolinguistic research, and depending upon your own specific research
question, the questions you prepare for interview will normally be semi-structured. This is so that
you can make sure any variables you wish to investigate, particularly phonological ones, are
covered by the answers to the questions you ask, which can then be compared with one another
for the purposes of identifying variation. Questions are there to help the flow of conversation and
keep it going, and semi-structured interviews allow for the addition of supplementary questions
to the main questions asked. They are intended to cover ground that the speaker does not
volunteer, and to ensure that all informants discuss roughly the same topic. For this reason, semi-
structured interviews are ones which will have the greatest degree of success in capturing the data
you wish to collect. Semi-structured interviews also allow for lines of conversation that develop
during the course of your conversation to be included in ways in which structured interviews do
not. Alternatively, depending upon your research focus, you may choose to conduct unstructured
interviews which are likely to resemble conversation more than undertaking an interview. Here,
the speech you capture may sound more natural than that of reading passages or interviews. By
their very nature, unstructured interviews are unpredictable, and you will need to be prepared to
cope with occasions such as conversation drying up or where one person dominates the
conversation to the exclusion of others. For occasions of this kind, it is a good idea to have some
questions ready such as when did you last go on holiday? in case you need them. The important
thing to remember above all else is to choose methods of data collection that are going to give you
the data that will best answer the research question you have chosen to investigate. The choice of
data collection for any study should always be justified in the write-up.
Recording and talking about speech make people more aware and conscious of it, and gives rise
to a phenomenon identified by Labov (1966) called the observers paradox. This refers to the
fact that it is impossible to observe someone behaving naturally without affecting what they do
or how they do it. This paradox can affect informants speech in two main ways. An interview
situation brings with it a certain degree of formality which affects speakers use of language.
Added to this, if the topic of conversation is the informants speech itself then this too makes
speakers more aware of what they are saying and even more likely to formalise their speech. For
this reason you need to think very carefully about the questions you are going to ask. Questions
which take the focus of attention away from speech onto other topics is always a good idea, such
as Labovs (1972) danger of death hypothesis: Were you ever in a situation where you thought
you were in serious danger?
You may get better results and minimise the observers paradox by interviewing people about
their views on a particular topic where the emphasis is upon something other than informants
speech. Another decision to be made is whether you interview your informants one at a time, or
in pairs. Interviewing them in pairs can take the focus of attention away from you as the
interviewer and allow for more varied dialogue and thus also minimise the observers paradox.
Investigating attitudes towards language use is particularly problematic area to study, not least
because peoples perceptions of how they or others speak are often at odds with actual use (see
Kerswill & Williams 2002). It is difficult to assess the reliability of a persons judgement of this
without recourse to other evidence, such as a dialect recognition experiment (Kerswill & Williams
2002). This involves playing taped extracts of samples of speech and asking informants to
identify their own community dialect. An experiment of this kind could also be written instead of
spoken. A study into attitudes works best if it has it is a focus, rather than the vague notion of
attitudes, such as attitudes towards employability as in the example given above. For the purposes
of reliability, triangulation is often employed in data collection: that is, more than one method of
data collection is used to cross-reference results across different methods. Thus, the usage
reported by informants could be compared with naturally occurring conversation (perhaps in
response to the danger of death question), so that the accuracy of self-reporting can be checked.
One final set of issues you need to consider before you begin to collect your data, especially taped
recordings, are those to do with ethics. These include issues about the ways in which you collect
your data, and how you obtain the informants consent. If your main method of data collection
involves questionnaires, then completing and particularly returning a questionnaire means that
the informant is happy for you to use the information provided. Ethical issues are more complex
when it comes to recording. The ways in which you are going to collect your data may be done in
an overt or covert way. An overt way is where the use of the recorder is obvious and your
informants are in no doubt that they are being recorded. They are also fully aware of the reasons
why they are being recorded and that the topic of the interview or conversation is the one which is
being researched. A covert way of recording is where you either collect your recording without
informants realising that you are doing so, or, you give a reason other than the real one as to why
you are collecting data.
Undertaking data collection in an overt way involves obtaining informants permission to
undertake recording before the recording itself takes place. If no permission is given, then no
recording takes place. Undertaking it in a covert way is more problematic. If you record
informants without their knowledge, then you need to obtain their permission after the event
before you can use the data. One way to get around the paradox is to give a reason other than the
real one to obtain that permission. For example, you wish to record informants speech to study
phonological variation and its relation to social class or regional identity, but you tell your
informants that you are conducting a survey into eating habits. Making the focus of the interview
and recording something other than the informants own use of language gives you more chance
of obtaining more naturally occurring data than would otherwise be the case. However, if the
informant is kept ignorant of the real reason why they are being recorded then this remains an
ethical consideration. A way of getting around this, is to tell the informant after the recording has
been completed the real reason why they have been recorded, and, in the light of this new
information, ask for their permission again.
1.5.4 Sound and video recording
Central to the data you collect will be your informants speech. The medium you use to record this
can be chosen from a wide range that includes an MP3 player, camcorder, laptop or digital audio
tape (DAT) in addition to the standard audio cassette and cassette recorder. For the purposes of
researching variation in English, and especially accent, sound recording is by far the most reliable
source of data, since this captures sound as it is actually used. Video recording has allowed an
extra dimension in data analysis, since it captures the context in which naturally occurring
language is used. However, because the focus of research into variation is usually the language
itself, and not, for example, body language or facial expression, then it is still usual to use a sound
recorder such as an audio cassette, or (increasingly) minidisk, as the main method of data
collection.
The kind of microphone you use is also important. Many cassette players have built-in
microphones, but the sound quality they capture can be very poor, and for this reason it is best to
have a separate microphone, and one which is omni-directional: that is, captures sound from
multiple directions. Since an interview can last up to an hour, it is best to place the microphone
between the informants rather than to hold it, since holding a microphone can be not only
obtrusive but also tiring. The best place for the microphone is to set it up so that it is spaced
equally between you and the informants, so that you are all speaking across the microphone. The
ideal place is on a level close to the informants mouths, which will allow for speech to be picked
up clearly by the microphone. Try not to position the microphone too close to the interviewees
mouths or you will get a distorted sound. Positioning the microphone roughly thirty centimetres
from the interviewees and positioning yourself the same distance next to them is best. That way
your questions can also be clearly heard. It is also vital that you check how the microphone works
and if it works in advance of your recording, and that you are absolutely sure you have it plugged
into the correct socket on the tape recorder. Also, you need to be aware of any peripheral noise
such as a fridge humming in the background, that might affect your recording.
Some recorders allow you to set a recording level to help with the avoidance of distortion. There
will be a meter which shows the level, plus a dial to adjust it. To set the level, ask your informants
some simple questions such as: What did you have for breakfast? or Where did you go on
holiday? whilst you adjust the setting. Other recorders may offer recording level options such as
high or low. Here you can ask your questions and play back to judge which is the right setting.
A further important factor to take into account is the location in which the recording is to take
place. The overriding aim of recording is to record speech which is as close to that used by
informants when the recorder is not there. You will need to be in a reasonably quiet location,
where there will not be too many interruptions, and in a setting where your informants feel
comfortable. For this reason, it is important to interview informants in a setting which is familiar,
for example a domestic setting such as their own home, or a social setting such as a caf, bar or
pub, or, if investigating younger children, a playground, either at school or close to home. For
example, if you are investigating lexical variation of words to do with beer and drinking, then the
most logical setting for such investigation is a pub. However, you need to record at a time when
the pub is not too noisy, or arrange with the landlord if you could have exclusive use of a room.
Similarly, if you wish to investigate phonological variation between people of different ages, then
a family gathering or social occasion such as a birthday party or family gathering is appropriate,
though it would need to be small enough to record individual conversations, but large enough to
generate suitable data.
You will also need to decide in advance and let your informants know how long the interview or
recording session is going to take. This will involve agreeing a start and end time for the
recoding session, which allows for an actual hour of recording time.
Depending upon the focus of your study, your recording may centre upon discussion of a written
document such as a spidergram, mentioned in 1.x. above. (For more information on this
technique, see: [Link]/voices). Here, informants are given a sheet of paper which asks
them to list alternative vocabulary items for certain groups of words. The spidergram may be
given to informants in advance to complete, with an interview being recorded subsequently to
discuss the items listed, or it can be completed as part of the interview. For example, this
technique would be a good one to use when interviewing men (and women) about drinking habits
and beer.
In order to play back recorded speech, it is important to have a recorder that will do the job you
want it to do. One with a built-in counter and speed control is especially useful. The counter will
let you pin point sections of tape easily, and a speed control will let you vary the speed of the
recording, which is particularly useful when it comes to transcribing. The development of digital
recording has allowed electronic resources such as sound files to be developed. Another resource
is electronic data bases such as the Newcastle electronic corpus of Tyneside English. This has
been developed into a major corpus of Tyneside English, an area of North Eastern England, with
orthographic and phonetic transcription, tagging and sound files (see: [Link]/necte).
Once you are ready to start recording, then note the date, time and location of recording, both on
the tape itself and on a sheet of paper. Make sure your questions and/or any other material you
may need are to hand, including preparation for follow-up questions and drying-up or the
unexpected. Something you will also have to bear in mind is that conversation, and the
circumstances in which it takes place, can be unpredictable. You have to be prepared to record
more than once and for greater lengths of time than you actually need in order to obtain sufficient
data for analysis. It is always far better to have too much than too little.
1.5.5 Transcribing data
Once you have completed your recordings, then your next task is transcription. Transcription is a
very time-consuming business, and is best done with a tape recorder that lets you slow down the
pace of the speech, so you can transcribe it more easily, or by using a digital recorder.
There are two main types of transcription: phonetic and orthographic. The kind of transcription
you undertake will depend upon your research question. For example, if you are undertaking a
phonological study, then you will need to transcribe phonetically using the IPA (see DL:DE
Chapter 1). The reason for using the IPA is because it allows for the notation of more sounds than
the alphabet allows, and can be used consistently to represent the same sounds whilst the alphabet
is not consistent in relation to sound.
There are two general types of phonetic transcription: narrow transcription and
broad transcription. Narrow transcription gives precise details of sound, encoding
information about the phonetic variations of the specific allophones in the utterance.
Broad transcription gives a basic idea of the sound and is usually a transcription of the
phonemes of an utterance. In dialectology, broad transcription is generally used. For
example:
Erm in about (.) erm twenty seven weeks I go away to cuba
(:? (? (?(v? (:? ?????? ???(? ??:?? ( ?(( (??? ?( ?v:?(
for two weeks er nd four days in Havana and ten days in a holiday
?( ?v: ??:?? (:?? ?(: ???? (? ?(?(?( (?? ??? ???? (? ( ?(????:
Resort erm its cost about two thousand pounds for the two of us (.)
???(:? (:? ??? ?(?? (?(v? ?v: (?(?(?? ??(??? ?( (? ?v: (? (?
This transcription is taken from a recording of an informant speaking in a regional English
accent, that of the Black Country, One of the characteristics of this accent is use of the
dipthongs, such as that of [(] previously identified by both Mathison (1999: 109-10) and
Clark (2004:262), whilst Wells (1982:539) points out that there is quite a degree of
phonetic variation. These findings are illustrated in the transcript above, where (
appears in (?(v?, ??? ??? ????????? a( in (?(?(?? and ??(???.
It is important that you listen to a passage several times before beginning your
transcription. This is so that you can be absolutely sure that you know exactly what a
speaker has said in order to transcribe it faithfully since you cannot, as in a written text,
rely upon a variety of interpretations.
Orthographic transcription uses the standard alphabet, in writing down verbatim what is said
(also see 2.8.3). This is usually set out like a play script, with speakers identified in the left hand
margin and the text of what they say next to their name. It is conventional not to use capital letters
or other punctuation in such transcription. It is also important that you transcribe what you
hear, including utterances not normally found in writing such as um, er, and so on, and resist
the temptation to modify what you write into written, grammatical and punctuated standard
English. For example, the following extract has been transcribed both phonetically and
orthographically:
Once you have decided which form of transcription you are going to use, then if this is relevant to
your study, you need to select the parts you wish to transcribe, Bearing in mind that phonetic
transcription takes longer than orthographic transcription, as a rule of thumb it takes between
seven to ten hours to transcribe one hour of recording, so you can see why it is so important to
select sections of recording for transcription, rather than to transcribe whole recordings.
The criteria you use for selecting sections for transcription will depend upon the focus of your
study. The focus will to a large extent determine the criteria, and it is a good idea to have a list of
those criteria when you listen to the recordings and make your selection. For example, if you are
intending to analyse differences in accent of a particular geographical area, then you need to select
the sections where your informants are most at ease, which is most likely to be once your
interview or conversation has become established, which would be after about ten minutes into
your recording. Depending upon the focal point of the study, you might decide to use a broader or
narrower transcription system (see: DLDE Chapter 1). If the focus of your study is lexis and/or
grammar, then may be less need to transcribe or to transcribe in short sections where the relevant
structures or lexical items are used.
1.5.6 Analysing data
Once you have transcribed your recordings, then you are in a position to analyse them.
The focus of your analysis will depend upon the focus of your study, but will almost
certainly concentrate upon identifying regular patterns of difference in accent,
vocabulary or grammar. Depending again upon the focus of your study, and especially if
you have chosen to investigate grammar, vocabulary or attitudes towards language use,
then you may have other data to analyse, such as questionnaires.
Identifying patterns of linguistic variation can be achieved by comparing your data with standard
English and RP. This will involve you in not only identifying patterns of difference in the speech
itself, but also taking account of the non-linguistic variables you have included in your study,
such as age, gender, social class and ethnicity which may account for the variation. Any
conclusion you may draw here as of necessity to be tentative, since you may not easily be able to
tell which variable is responsible for the linguistic variation unless you are in a position to
perform statistical tests on large amounts of quantitative data.
The first thing you need to do is to identify patterns of difference or variation in your data. For
example, the following transcript is taken from a study of the Walsall dialect in the West
Midlands area of England. The dialect of this area is known as the Black Country dialect. From
interviews of five white British informants aged between forty and fifty, of whom three were
female and two male and all of whom left school aged fifteen or sixteen with no formal
qualifications, the following features of non-standard grammar were identified:
1. J; yeah, he got me a job cos I worwell, I was working, but I changed jobs cos it was
nearer, worn it?
2. K: and her came back Sunday
3. M: thats got to be our jodhpurs wot we make
4. K: and then her come back Sunday
5. K: them chickens
6. J: Never touched it
7. K: and the zip was broke so he says
8. K: was that one yorn?
Once you have completed this first level of analysis, then you can undertake a second level,
which is to identify the grammatical forms and present them in a table, given below. Column A
describes the feature in grammatical terms, column B gives an example, and column B gives the
first occurrence of each feature from five transcripts, with subsequent occurrences listed.
|A Feature |B Example |C script and|
| | |line |
|1. Dialect use of preterite <to be> |<was> |1.3, 1.4 |
|<was> affirmative |<worn> | |
|<were> negative | | |
|2. Non-standard use of <her> would be <she> |<her> |3.2, 3.3 |
|in SE | | |
|3. <wot> for SE <that> (relative pronoun) |<what> |2.7 |
|4. Preterite identical to present tense |<come> |3.3, 5.5 |
|5. <them> for demonstrative <those> |<them> |4.1 |
|6. <never> as past tense negator |<never> |5.10, 5.12 |
|7. (adjectival) past participle identical to |<broke> |2.6 (2.8) |
|preterite: <broke> for SE <broken> | | |
|8. <yorn> for SE <yours> (2nd person |<yorn> |2.9 |
|possessive pronoun | | |
| | | |
A third level of analysis would involve identifying grammatical features which are specific
to a particular dialect, and those which are a more general feature of non-standard
dialects. For example, from the data given above, then dialect use of the 2
nd
person
possessive pronoun as in the last example yorn is a distinctive feature of the Black Country
dialect, identified in previous studies. Never in example 6 as a past tense indicator or past
participles being identical to preterites such as broke in example 7 are features to be found in
several non-standard dialects.
Having analysed your own data, then a comparison can be drawn with studies previously
published to draw conclusions and to see whether or not your data affirms or contradicts that
previously undertaken.
If your study involves researching attitudes towards language use, then your analysis will centre
upon identifying what these attitudes actually are and possibly, in addition, where these attitudes
come from. As 1.3.1 and 1.3.2 above has explained, attitudes people have towards language use,
including any prejudices, can be traced back to a period when people held very different beliefs
about language to the ones linguists hold today, but which still persist. Your study may also wish
to investigate the extent to which individuals choose a variety of language not only to express
ideas but also their own identity and to distinguish themselves from other groups of people.
Analysis of any questionnaire data will, in many cases, produce numerical results, particularly
where closed or multiple choice questions have been used. Note that there are some simple
statistical texts, such as the ?
2
test, which can quite quickly indicate whether raw scores and/or
percentage scores are significantly different from each other.
1.5.7 Writing up your study
The overall structure of your study will be different from that of a conventional essay. Generally it
will include the following elements:
Presentation of your hypothesis or hypotheses. This will include: a description of the
issue(s) at stake; the questions to be addressed; what is your hypothesis, and why it
matters; what type of data will be used ( phonological, lexical, syntactic or peoples
opinions); if appropriate, which variable (s) you will use, if any, and their possible
variants; description of the speech community to be examined in your project; what type
of analysis will be conducted and briefly, what conclusions you expect to draw. You
should also locate this discussion within the context of other studies already undertaken.
This will allow you to set your own hypothesis against the work of others, and provide a
context for your own hypothesis.
Description of your methodology. Your methodology will normally be drawn
from certain methodological and analytical frameworks, and in applying them the
researcher is making assumptions about theoretical issues. Description of your
methodology and methods should be detailed enough so that someone else could
replicate your study, and include information about: the informants and
investigators, and how many there were of each; an explanation of the methods
used to locate possible informants and agreeing their participation; any criteria
used for sampling; how the data was collected, including how it was recorded and
transcribed, and by what method; how many questions or items were elicited, and
when and where; how the results were analysed and if any statistical analysis was
involved. Your write-up should include a copy of any research instruments you
used, such as interview questions you asked, questionnaires, dialect recognition
experiments and completed biographical data sheets,
Presentation of your data. This is the place to describe the variant(s) or the attitudes you
are investigating. You will need to define the degree or envelope of variation for the
dependent linguistic variable(s) you are investigating. There are also other types of
variables that may affect the dependent variable you are investigating. These can be
grouped into three different kinds: social, such as age, gender and ethnicity; stylistic, such
as a casual conversation compared to a reading passage or linguistic, for example, the
position (of a variable) in a clause, the height of vowel. You will also need to define and
discuss each of the social factors relevant to your project, and provide a table or list that
summarises all the relevant linguistic variables and their possible variants, with examples
from your data, or a summary of your questionnaire responses. Your write-up should
include a copy of the transcript from which you have taken your examples or a copy of the
questionnaire. As mentioned in 1.x above, quantitative studies, even though they canot
rely on percentages, must use statistical testing for significance. ,
Discussion of what you found. This is the place to analyse your data (see also 1.5.5
above) and discuss any patterns you found, and how they correspond to your hypothesis.
What did you find out? Discuss each independent variable individually, commenting upon
how it correlates to each/the dependent variable, or each question of your questionnaire
and commenting upon what each one reveals.
Your results. This is the place to comment upon your analysis and what you have found
out. This section addresses questions such as: Do your results support your hypothesis? If
not, why not? Make sure you respond to everything you raised in the introduction. Were
there any surprises? How do your findings compare with any others? How might your
study have been improved? What did this investigation suggest for future research?
1.5.8 Sample projects
a) An Analysis of the Accent of (place).
Choose a geographical area and undertake an analysis of its accent in relation to your
own choice of dependable linguistic variable(s) and two non-linguistic variables. In
undertaking this study, you should choose a sample of at five least informants.
Methods of data collection here would normally be taped interviews or conversations.
You will also need to choose your linguistic variable to act as your dependable variable.
As suggested in 1.5.6 above, you would then need to outline the scope of your study,
why you chose this particular one and justify the research methods used in your write-up.
You will also need to give an overview of biographical information and transcribe
selections of your recordings into IPA, using a narrow or broad system of transcription.
Your analysis of the transcription would then identify the dependable variable(s) and the
scope, presented in tabular form. A comparison can then be made with other published
studies on the locality, to see whether or not they confirm what has already been found
or show something different in drawing your conclusions. Your transcripts would form an
appendix to your write-up, which would cross-refer to it.
b) Lexical change in (place) circa 1950 to the present day.
Undertake a study across three generations of the same family, all living in the same
area where they were all born. You should choose a sample of at least three informants,
one of whom can be yourself, to determine the extent of any lexical change and the
reasons for it.
Methods of data collection here would normally be taped interviews based upon a written
questionnaire of some kind or a spidergram. You would then need to outline the scope of
your study, why you chose this particular one and justify the research methods used in
your write-up. The biographical information you collect will be an important part of your
study, particularly as it relates to education, since it may allow you to draw conclusions
as to the reason for any change. You may not need to transcribe selections of your
recordings into IPA, but orthographically instead. As in sample project a), a comparison
can then be made with other published studies on the locality, or with data from
electronic resources, to see whether or not they confirm what has already been found or
show something different in drawing your conclusions. Your transcripts would form an
appendix to your write-up, which would cross-refer to it.
c) Grammatical variation in (place).
Methods of data collection here would be questionnaires, or contemporary writing in the
local dialect such as newspaper articles, poems and short stories. This would allow you
to collect more date more quickly than interviews upon which to base your analysis. In a
project of this kind then like b), you would not need to undertake a transcription, but pick
out instances of grammatical variation in orthographic notation. As in sample project a), a
comparison can then be made with other published studies on the locality, or with data
from electronic resources, to see whether or not they confirm what has already been
found or show something different in drawing your conclusions.
d) Undertake a study of attitudes towards variation in your peer group.
Methods of data collection here would be questionnaires, possibly followed up by interviews and
the use of dialect recognition experiments. The success of your project will depend very much
upon the design of your questionnaire, and the kinds of questions you ask in the dialect
recognition experiment. You will need to make the questions general enough in your
questionnaire, or with sufficient options to allow for the spectrum of possible views, but also
narrow enough or specifically focused to get at the reasons why people hold the views that they
do. For this reason, biographical information of your informants is as important in this kind of
study as for any of the others listed above. Similarly the questions you ask as part of your dialect
recognition experiment shoud allow for possible options, rather than a closed yes or no.
1.6 Where to find out more
Leith (1997) and Beal (2006) are good places to start for socio-historical accounts of
English. Views on standardisation, grammaticality and correctness can be found in Clark
(2001), Crowley (1991 and 2001), Honey (1983) and Milroy and Milroy (1991). Labov
(1978) and Wolfram & Schilling-Estes (2006) cover social and geographical variation in
English in the USA and Milroy (1987) in the UK, whilst Wells (1992), Hughes, Trudgill
and Watts (2005) and Widdowson & Upton (2006) give an account of geographical
variation in English in the UK and Kortmann and Closs Traugot (2004) across the
whole world. Trudgill (2002) and Kerswill & Williams (2002) cover aspects of
sociolinguistic variation and change. Montgomery (1995) and Wardhaugh (2006) are
good introductory books on sociolinguistics in general. Woods, Fletcher, Hughes &
Anderson (1986) provide an overview of linguistic statistical analysis.
Chapter 2: Pragmatics and Discourse
2.1 Introduction
This chapter examines issues and debates in the field of pragmatics and discourse: that
is, it examines how interaction shapes linguistic structures and the ways in which
communication involves more than the words which are actually spoken. Traditionally,
linguists have concentrated upon the formal structure and properties of what we hear as
sound or see as words on a page, that is the visible or aural aspect of language,
Language and word structure have been described in terms of syntax, morphology,
phonology and the meaning of individual words as semantics (for more about this, see:
DL:DE). Variation in language use, discussed in Chapter 1, has also traditionally
concentrated upon variation of these features, especially phonology. In addition,
linguists have been concerned with the history of these structures and their change.
However, there is another aspect or dimension to the study of language, which views
language as communication or language in the context of its use between people in
everyday settings: that is, language as it is actually used in communication. Recognition
of this, and a desire to formulate or generate rules of how communication, as opposed to
language, works, has led to new areas of linguistics being developed.
Discourse Analysis (DA) and Conversation Analysis (CA) are concerned with the
structure and management of discourse and conversation whilst Pragmatics is also
concerned with unspoken or implicit meanings in language. A further area, interactional
sociolinguistics, focuses upon the importance of cultural variations in the ways people
use and interpret discourse. However, the concerns of each of these areas overlap and
have much in common, particularly as all of them study the context in which speech
occurs, how it is made meaningful and its purpose or function. These aspects of
language analysis go beyond studying the linguistic structure of utterances to look more
closely at, for example: how conversation structure can be as meaningful as its content;
the social force of what is said and how the assumptions and world-view of speakers is
encoded, or embedded in speech - in what they say, or do not say. As Bonvillan
(1993:85) puts it: Understanding meaning is necessarily contextual, situating speech in
interpersonal and cultural contexts. All cultures provide rules for appropriate
communicative interaction, defining behaviours that should occur, that may occur and
that should not occur in given contexts.
In formal linguistics, the term structure applies or refers to the ways in which words
combine, the grammatical relations between them, and how they function within a
sentence, the sentence usually being the largest unit of analysis. In investigating
communication, pragmatics, Discourse analysis, conversation analysis and interactional
sociolinguistics are all concerned with structures beyond the level of a sentence up to the
longer stretches of often complete texts. This is because, when we communicate, we
rarely do so through single sentences alone, but usually communicate through a series of
exchanges between two or more people. Indeed, work undertaken in discourse analysis
and more recently in the grammar of speech such as that of Carter & McCarthy (2006),
has found that applying the term sentence to speech is something of a misnomer, since
the ways in which we structure speech differs significantly from the ways in which we
structure writing (see 2.7.1 below). The preferred term in pragmatics for a segment of
continuous speech is thus utterance or speech event.
2.2 Language, meaning and communication
As children growing up in families and/or communities, and as part of our socialisation,
we acquire the rules for appropriate communicative interaction. This is done both
informally at home and in our communities and formally at school or other public settings.
As children, learning to behave is often not only communicated through language, but
also includes how to use language itself. We are told how to act and what to say - or not -
in particular places and times or to particular people: for example, we learn to say thank
you when offered something, even if we do not particularly want to accept what is
offered: Would you like a biscuit? Say thank you to Aunty. We are also told how to modify
aspects of our behaviour: Dont shout/point its rude! You mustnt tell your sister she
looks fat in that dress, even if you think she does! Much of what we learn though, is not
so much through such direct modifying of our verbal behaviour, but comes from our own
observations of people around us: family, peers, our neighbours or strangers on the bus,
so that we learn the interactional norms appropriate to the society - or groups within
society - in which we live. Consequently, much of what we learn is implicit: that is, not
much is ever spelt out or explained as to why we are told to behave in the way that we
do, but absorbed as part and parcel of growing up in a particular community, culture and
society.
Sometimes we can also find ourselves living within two or maybe more communities or
cultures which require different cultural norms, or growing up in one then moving to
another. For example, if we happen to be the first generation of children born to
immigrant parents or have spent the majority of our childhood in an institutional setting,
such as a childrens home or a setting different from that into which we were born, such
as foster care, then we have to learn to acquire and understand the implicit norms of
spoken interaction within both sets of communities. Similarly, if you move to another
country at any time in your life, then you will also find that part of acclimatization, even if
the same language is spoken, it involves acquisition and understanding of implicit norms
for successful communication to take place. If you happen to have experience of such
different settings, then youre probably more aware than most of the unspoken
interactional norms which govern verbal behaviour. Underlying and unstated rules of
interaction which we normally use without thinking are most usually brought to the
surface either when they are violated when someone speaks inappropriately or when we
have experience of an unstated or underlying rule being appropriate in one cultural
setting but not another.
For example, in English culture it is part of the underlying rules of conversation that we
learn to listen to one another and to take our turn in conversation, that it is rude to
interrupt, and that when we are interrupted because somebody cannot wait to speak,
then we stop speaking. In other cultures, such as those of the Indian sub-continent, this
rule does not exist and it is acceptable for people to talk at the same time and/or to
interrupt one another whilst speaking. As communication across the world increases,
through increased global business, travel, economic and social mobility and increased
use of the internet, we may find ourselves more frequently in situations where norms
differ from the ones in which we grew up (see 2.6 below for more on this).
2.3 Language in context
Some sociolinguists have argued that in order to find out what are the relevant features of
different contexts or situational variation, speech behaviour needs to be studied and analysed in its
widest social and cultural context. The work of Gumperz, 1982a &1982b, Hymes (1972 & 1974)
and Gumperz and Hymes (1986) has been very influential in this area. Drawing upon research in
sociology and anthropology, Gumperz and Hymes proposed an ethnography of communication.
Ethnography is primarily a research method associated with sociology and anthropology,
concerned with the study of human behaviour in its ordinary day to day settings. An ethnography
of communication extends this notion to include verbal behaviour, language, on the grounds that
no verbal activity has any meaning unless it is viewed in the context of its situation. For Hymes
(1974), the communicative conduct of a community is the starting point for ethnographic analysis.
He identified several components of communication that require description of which the most
important are:
1. setting or context
2. participants, minimally speaker and addressee
3. topics and attitudes
4. goals.
Each of these different aspects can be studied independently of one another. However,
sociolinguists like Gumperz argue that a speech event is likely to include all of them if it is
to count as a speech event at all, so all of the components are interdependent and
therefore need to be studied. The extent to which any one aspect or component
dominates more than another depends upon a speakers assessment of the whole
situation which they are in and their judgement on how things are likely to turn out. For
example, speech events in formal contexts - such as a law court or lecture hall or
seminar room - often take place in specified settings amongst expected participants and
concern relatively fixed topics. At other times, all four aspects may take a more equal
share, especially in more informal settings. Consider the following example of how the
different factors interconnect, from a court of law:
1. A trial setting: Trials or hearings are communicative events that occur in particular
settings, namely courtrooms. The courtroom itself has a structural design separating
seating areas for various categories of participants and orienting them in relation to one
another.
2. Participants: Participants include judges, lawyers, defendants (and plaintiffs),
witnesses, jurors, spectators, and court officials. Each participants behaviour, including
verbal behaviour is conditioned by his/her role. The judge is singled out by his/her
seating position in front, usually on a raised platform, of the courtroom and special attire.
The judge clearly controls communicative behaviour of other participants, each of whom
has certain obligations to speak or not to speak. In fact, failure to speak when so directed
or failure to be silent otherwise are legally punishable offences (contempt of court). Only
judges, lawyers, and witnesses may speak publicly. Other participants (jurors, officers,
spectators) must remain silent. Specific discourse patterns are expected of each type of
participant. Lawyers may make introductory and concluding statements or ask questions.
Witnesses answer questions. Judges have greater latitude in the kinds of speech
appropriate for them; they can make statements, ask questions, and issue commands
and rulings.
3. Topics: Topics of discussion are rather rigidly defined. The communicative event is
about something and all speech behaviour must be relevant to that issue. Rights of
participants to introduce changes of topic are narrowly limited and permission to
incorporate extraneous speech must be asked of the judge (and granted). In lawyers
questioning of witnesses, it must be clear that questions are relevant to the central issue
or are a logical argument from it. Similarly, witnesses answers must be relevant to the
topic and relate directly to their own experience. Judges have some flexibility in topic
choice, but they too are limited by the overall focus.
4. Goals: Goals of participants vary according to their role in the proceedings. Individual
communicative behaviour is oriented toward achieving specific goals. Speakers choose
words, tone of voice, facial expression, gesture, and so on, to accomplish their purpose:
for example, the judge must appear impartial, lawyers speak and act aggressively,
defendants portray themselves as innocent, witnesses appear honest and reliable, and
jurors remain silent but convey interest in the speech and behaviour of others.
Adapted from Bonvillan (1993:85-6)
A trial in a court of law is probably at the most formal end of the scale of human
interaction. At the other end, though, even the most informal of settings are constrained
by cultural norms of ways of speaking and rights to speak. Take the following example
from an English familys table at an everyday dinner:
1. Setting: family meals are communicative events that occur in particular settings,
although these may vary: at a table, in front of the TV or on ones knee. At a family table,
the structural design is centred upon the table and its arrangement of seats.
2. Participants: participants are the members of the family and occasional guests. Each
participants behaviour is conditioned by his/her role, including verbal behaviour. Although the
setting is not as formal as a courtroom, nevertheless family members will have usual seats at the
table, with the mother, father and/or any other adult in parental authority occupying at least one of
the seats at the end of the table. The parent(s), step-parent(s) or adult with parental responsibility
will control the communicative behaviour of other family members, encouraging talk or
discouraging it as the case may be.
3. Topics: Topics of discussion are potentially limitless and can be introduced by
anybody. Most typically, they relate to the family members own daily experience - events
at work, school, or discussion of other extended family members and friends - to
discussion of TV programmes, computers and games or to current affairs that have
grabbed news headlines - a national election or a world disaster, for example. The
speech event is about something but not all speech behaviour must be relevant to the
issue. Rights of participants to introduce changes of topic are not limited and no
permission needs usually to be asked. Topics may range over a number of issues with all
members taking part in no particular order, but the overall pace and direction of the
conversation may be controlled by a dominant adult.
4. Goals: The goals of all the participants may vary; for teenagers, the goal may be to eat
the meal as quickly as possible and limit participation to a minimum in order to get back
to the TV or computer. For adults, it is a time to catch up with the days events and
maybe plan the next days activities. Above all, family meals are a social occasion where
members catch up with one anothers news.
Ethnographers of communication thus claim that by observing the variation that exists in
speaking styles and verbal behaviour, we can uncover and make evident underlying and
unstated rules of interaction. One of the problems of such an approach is that it tries to
do far too much all at once (see DL:DE section 1.0). It takes into account the full
complexity of grammar, personality, social structure and cultural patterns and treats them
as integral to the speech activity itself. Although not many sociolinguists would disagree
with this as an aim, there is simply too much to take into account if one considers all
aspects of human behaviour - social, cultural and linguistic - all together and all at the
same time. In addition, because such study is so context specific, and given the apparent
endless possibilities of situations, it is difficult to generalise rules across a variety of
contexts even in similar settings, unless for the most formal of settings such as the law
court example above. Such an approach lends itself well to the understanding of how
communication occurs in particular contexts, and might well lead to discovering
underlying and unstated rules of interaction for a particular situation, but it is doubtful how
far such a discovery could then be transferred to other situations or made into a
methodology itself. Some sociolinguists argue that an ethnography of communication is
too context bound for it to be of much use linguistically. Nevertheless, the approach
remains important because it highlights the central place context plays in communication.
(see also: TL:TE Part II Chapter 5).
Focusing upon the verbal aspect of a speech activity and uncovering the underlying or
unstated and implicit rules which govern communication at all levels, is what pragmatics
and discourse analysis are concerned with. In the example given above of the court of
law, communication in this context is made as explicit as possible with little opportunity
for ambiguity, misinterpretation or misunderstanding. Consequently, uncovering implicit
rules is much more straightforward than in less formal situations where the topics and
goals of the participants are not as orientated or explicit, and where there is greater
potential for misunderstanding and misinterpretation. The question those concerned with
pragmatics frequently ask themselves is: what patterns of unquestioned meaning
emerge from verbal communication?
Consider also the following examples:
Example 1:
Urszula: Whats the road number where we turn left?
John: Oh, weve got ages to go yet.
( conversation on 7 August 2004 between the author and her partner driving to a
party)
This conversation is between two people who live together and know one another well.
Urszula has asked a specific question, wanting to be prepared in good time for turning
left, but rather than answering the question, John interprets it as indicating anxiety and so
does not answer the question, but instead offers words to soothe this perceived anxiety.
Example 2:
A mother and her young son are out shopping, and the mother needs to use the toilet.
She takes her son with her into the Ladies, and uses a stall but discovers it has no paper.
She calls out to her son:
Jonny, go and see if the mens toilets have any toilet paper. He goes, comes back and
says: Yes, Mummy!
(True story told as a joke at a conference by Lesley Jeffries, July 04) .
In this case, the child has taken the mothers request literally, and not worked out the
illocutionary intent (see 2.4.2. below) of her sentence and if there is some get it for me
which the mother had left out. This issue of interpretation, of explaining the unspoken or
implied, is a central concern in pragmatics. As the examples above illustrate, often the
meaning of particular utterances or exchanges between people depends as much upon
the context of the speech and the purpose of the speakers as on the literal meanings
given by the utterances themselves. Trying to account for, the underlying rules is an
important aspect of pragmatics, where the focus is upon how utterances mean what they
say and the possibilities of interpretation within a given context.
For example, consider the phrase: Howre things going? The meaning potential of this
phrase is in theory open to any number of replies, but the potential is limited by the
situation, the speaker of the question and the relationship between the two participants.
For example, a standard opening between two acquaintances who accidentally meet in
the street and have not seen one another for a long time, especially when one of them
cannot quite remember when they last talked to the other, is Howre things going. Given
the situation, the meaning potential of this utterance is limited. It cannot be interpreted as
a request to walk the dog, to leap into bed or to go on holiday. In the particular context of
this occasion with its clear social constraints on the sorts of questions that could
appropriately be asked, it would not be reasonable to go into the minute details of
relationship or health problems and so on. Even excluding improbable interpretations,
there still remain a number of options open. One is, to take the question as a purely
phatic greeting, say OK thanks and move on, or take it as a question about a job, or a
general enquiry about family. However, if one changes the context in which the question
is asked, for example at a relationship guidance counselling session between a
counsellor and a couple, or in a doctors surgery between a doctor and a patient, then the
question takes on a very different meaning. Both speaker and hearer develop topics from
a range of possible utterance meanings and speaker meanings which are appropriate to
the circumstances and congenial to all participants.
It is also the case that people do not always or even usually say what they mean. For
example, in saying Its warm in here what the speaker may mean is Can you open a
window? People can mean something quite different from what their words say, or even
just the opposite. For example, if asked a friends opinion on her new hair cut I might say
It suits you when actually I think the opposite, that it does not suit her at all, but I dont
wish to hurt her feelings by telling here what I really think. Sections 2.3, 2.4 and 2.5
below describe key theories and methods that have emerged within the field of
pragmatics and discourse analysis, before going on to consider issues raised by
interactional sociolinguistics in more detail.
2.4 Language as action
Our understanding of underlying or unstated rules which govern communication owes a
great deal to a particular theory of language, known as speech act theory (for more
about this, see TL:TE Part II Chapter 5). Prior to the nineteen seventies, it was presumed
that spoken interaction was a relatively haphazard, chaotic way of using language which
was not subject to the same kind of linguistic norms or rules as writing was. Far from
being unstructured, conversation and spoken language does, indeed, have a unique
structure of its own. In addition to having underlying structures and rules, speech can
also function in a performative way; that is, speaking may not only be a way of
verbalising but also of performing actions. For example, Close the door verbalises a
wish to have a door closed, and also functions as a request to a second party who is
then expected to perform the action of closing the door.
Speech act theory began with the work of the philosopher John Austin (1962) whose
ideas were expanded by John Searle (1969) and developed by H.P. Grice (1975). This
theory is based upon the belief that language is often used to perform actions and on
how meaning and action relate to language. As Austin (1962:100) says: the words used
are to some extent to be explained by the context in which they are designed to be or
have actually been spoken in a linguistic interchange. It is this emphasis upon context
which underpins speech act theory, and which has made it far more suitable as a theory
upon which to base linguistic method and practice, rather than the all-encompassing
ethnographic theories taken from anthropology and sociology outlined above, which
attempt to cover every aspect of behaviour and interaction.. The section below gives a
brief summary of speech act theory. (For a more detailed explication, see Part II Chapter
5 of TL:TE.)
2.4.1 Speech Act Theory
Austin wished to move away from traditional approaches to the study of meaning in philosophy,
and particularly from the notion that the meaning of a sentence could be analysed in terms of
truth or falsity (see: DL:DE ii.6). As humans we communicate verbally and non-verbally.
Where communication is performed by words, Austin called the act a speech act, and argued
that speech acts performed three different acts at once: the locutionary act, the perlocutionary
act, and the illocutionary act.
- The locutionary act corresponds to the act of saying something. It involves uttering
certain noises in a particular grammatical construction with a more or less definite sense of
reference.
- The perlocutionary act is the act performed by saying something. It involves the effect
that the speaker has on her/his listener in uttering the sentence. These effects might be
upon the feelings, thoughts or actions of the hearer, even if the effect is not intended.
Examples of perlocutionary acts are such things as convincing, persuading, annoying,
amusing and so on.
- The illocutionary act is the act performed in saying something. It reflects the speakers
intent in uttering a sentence, and it is the illocutionary force of an utterance which
determines what illocutionary act is performed.
Austin argued that linguists should not limit themselves to the study of the literal meanings of
utterances (the locutionary act) but should also concern themselves with what communicative act
was intended (perlocutionary act) or accomplished (illocutionary act). We know what the sentence
I promise to be there at nine oclock exactly means, but how is it to be taken? As a promise? As a
threat? A teacher who says to a noisy disruptive class: Ill keep you in after the lesson, is
simultaneously producing three acts:
-locutionary: involving saying the sentence which means Ill make you stay in school later than
usual.
-illocutionary: threat.
-perlocutionary: silencing the students (or causing a revolt!).
For Austin, the illocutionary force is not part of the literal and linguistic meaning of what a
speaker says. Statements might in fact also act, for example, as requests. Its cold in here may
either function as a statement or be interpreted as a request to turn on some heating (see also
Thomas 1995:1 and 28-54). Sometimes, it is not possible to tell what is meant simply from the
form of words being used. Speakers, recognise illocutionary acts and/or the illocutionary force
behind them from, for example, our own previous experience (both linguistic and non-linguistic),
our knowledge of the speaker and from the context itself. However, we can also get it wrong,
since we do not always interpret illocutionary acts correctly. What is important to remember is
that there is no one-to-one relationship between the syntactic structure of an utterance and a
particular speech act. An illocutionary act of warning could involve a declarative sentence, a
statement, a command, a question or any other linguistic structure, given the right context. For
example:
Your t-shirt is on fire.
Watch out! Your t-shirts on fire.
Did you know your t-shirts on fire?
Why is your t-shirt on fire?
After Austins death, his work was pursued by Searle. Like Austin, he saw the importance
of recognising that when we say something to someone, very often we are not only
saying words but also inviting or performing action of some kind, which in some cases
can be legally binding. Searle picked up from where Austin had left off, focusing on
trying to explain how speakers intend a particular illocutionary act, and how hearers
recognise such acts. He proposed that some illocutionary acts can be performed by a
speaker if, and only if they fulfil the following conditions:
> some linguistic item (word, phrase, clause) is uttered
> which has some determinate characteristic (word order, stress, intonation, for
example),
> and the speakers utterance constitutes the relevant act
> because of some set of specific conditions which have to be satisfied on the
occasion of the utterance.
So, for example, we can do things like promising, warning, and so on by using certain
particular forms of words in particular circumstances. In addition to speech acts, Searle
also identified the phenomenon of indirect speech acts. For example, an utterance
such as Can you pass the salt please? is syntactically a question with the illocutionary
force of a request. Interpreting indirect speech acts correctly is a complex phenomenon,
that includes not only the identification of a set of conditions but also other factors such
as the participants own linguistic habits, cultural practices and present situation.
Following on from the work of Austin and Searle, the linguistic philosopher H.P Grice
took the theory of utterances one step further, by identifying underlying rules which
govern our management of conversation.
2.4.2 The co-operative principle
A major contribution to our understanding of how conversation is managed has been
Grices theory of conversational implicature. As a linguistic philosopher, Grice was
concerned with the relationship between logic and conversation. He made a distinction
between the logical meaning of given words and underlying broader interpretations
which he labelled implicatures, and which arise out of the rules and principles of
conversation (see DLDE:ii.6). The over-riding principle of all conversation according to
Grice (1975), is the principle of co-operation or co-operative principle (CP). The CP
assumes a tacit understanding between speakers to co-operate in conversation in a
meaningful way. The CP is formulated in Grices own words as:
[] a rough general principle which participants will be expected (ceteris paribus)to
observe, namely: make your conversational contribution such as it 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)
This rough general principle suggests that when we engage in verbal interaction, we
work on the assumption that this interaction operates according to a certain set of
expectations. Grice suggests that there are four categories (gruce, 1975: 45-46) which
people tacitly assume to be the norm in conversation, which he calls maxims of
conversation: that is, intuitive principles which are supposed to guide our conversational
interaction.
The four maxims or sub-principles of the CP are:
1. The maxim of quantity. This relates to the amount of information to be provided:
a) Make your contribution as informative as is required (for the current purpose of the
exchange).
b) Do not make your contribution more informative than is required.
2. The maxim of quality. This relates to the truth of the contribution:
a) Do not say what you believe to be false.
b) Do not say that for which you lack adequate evidence.
3. The maxim of relation relates to the relevance of the contribution:
a) be relevant.
4. The maxim of manner, relates to the manner in which the contribution is made:
a) be perspicuous, i.e.
i. Avoid obscurity.
ii Avoid ambiguity.
iii Be brief (avoid unnecessary prolixity).
iv Be orderly.
According to Grice then, we have a tacit, unspoken, general assumption or expectation
that our interlocutors when speaking (or writing) to us will, all other things being equal,
give us the degree of information required and no more or no less, tell us the truth, be
relevant to what has been said and say what they say in a clear, understandable manner
or way which is not antagonistic, ambiguous and so on, and that we will do the same by
return. Grice proposed that adherence to these maxims is the expected norm of
conversationalists, and much of our day to day conversation follows these maxims. Take
the following example, given in Thomas 1995:
Husband: where are the car keys?
Wife: Theyre on the table in the hall.
Here, the wife has given the right amount of information (quantity), truthfully (quality) ,
directly answered the question (relation) and clearly (manner). She has said exactly ehat
she meant to say, no more and no less, and with no implicature: that is, there is no
additional level of meaning, and there is no distinction between what she said and what
she meant. However, much conversation does in fact flouts these norms, giving rise to
conventional or standard conversational implicatures. A crucial feature of implicatures
is that they must be capable of being calculated by a hearer. Schiffrin (1994:195) sums
up Grices description of the process as follows:
To work out that a particular conversational implicature is present, the hearer will
rely on the following data:
(1) the conventional meanings of the words used, together with the identity of any
references that may be involved
(2) the CP and its maxims
(3) the context, linguistic or otherwise, of the utterance
(4) other items of background knowledge
(5) the fact (or supposed fact) that all relevant items falling under the previous
headings are available to both participants and both participants know or assume
this to be the case.
Given this basic process, implicatures are created by the speaker deliberately flouting
one or more of the maxims, and the hearer being aware of this violation of the CP. For
example, political language makes use of a great deal of implicature, making it harder for
the listener or reader to explicitly identify and thus to challenge - the basis of their
policies. Hoardings carrying advertising for the Conservative and Labour Parties for the
2005 General Election centred around a set of slogans, some of which flouted the maxim
of quantity by giving less information than would strictly be required. For the
Conservative Party, there was:
Are you thinking what were thinking?
For the Labour Party it was:
Are you remembering what were remembering?
The Conservative slogan depends on the flouting of the maxim of quantity. It is not made
explicit from the text what exactly it is that the Conservatives are thinking, but the force of
the slogan is to imply that the country is united, and in this case, the reader is expected
to share background knowledge that this unity is based on distrust of or disappointment
with the Labour government. Note that the convenience of this flouting of quantity is that
it enables readers to identify with the slogan, even if their own thoughts on Labour are
different from those identified in the last sentence. The gap in information, therefore, may
be filled in in the readers mind, by thoughts of how the Labour government is
ineffectual, lazy, arrogant, or whatever the particular voter thinks. The disadvantage, for
the Conservatives, is that the critical or resistant reader may choose to fill in the gap
with their own views, and have a mental response such as I dont know are you also
thinking how good and socially just the Labour government has been then?
The Labour slogan imitates the Conservative one by parallelism, using the same direct
address and phrasing as a question, and relying on intertextual referencing (see 3.7) to
build another implicature upon the earlier flouting, again depending on a lack of
information and thus flouting the maxim of quantity. In this case, the text itself does not
explain what it is that the Labour party is remembering, but background knowledge of
recent British political history would indicate that it is the years of Tory rule, including
those under Margaret Thatcher, and the sleaze of John Majors years in office, that are
being implied here. Again, a resistant reader may choose to sidestep the implicature and
think of what they considered to be good about the years of Conservative government,
but this is at odds with the intended, and compliant reading of the text.
Flouting the maxim of quality can be achieved in several ways. One way is to say
something that obviously does not represent what the speaker thinks. For example, if
an employer tells a prospective employee that theyd be happier elsewhere, it is a polite
and face- saving way of saying that they do not wish to employ them. Another way to
flout the maxim of quality is by exaggeration. I could eat a horse or Im starving or by
metaphors or euphemisms. For example, Shes such a wet blanket or Im just going to
wash my hands as a euphemism for Im going to urinate. The maxim can also be
flouted by irony and banter. Irony expresses a positive statement whilst implying a
negative one, whilst banter does the opposite: expresses a negative statement whilst
implying a positive one. An example of irony would be I love burnt toast in the mornings
but implying the opposite. The danger with both irony and banter is that it can offend if
hearers do not recognise the conversational implicature, and as a result take the words
literally.
Flouting the maxim of relation involves uttering statements which appear to have no
connection with one another, or expecting hearers to be able to work out for themselves
what the utterance does not say, and make the connection between the utterance they
hear and the one preceding it. For example:
John: So what did you think of Ashley?
Mary: His sisters a good laugh.
(authors data)
By not mentioning Ashley in her reply and apparently saying something irrelevant, Lucy
implies that she is not impressed by him. In technical (stylistic) terms, we can argue that
flouting the maxim of relation is often achieved by a lack of cohesion in the text. This
topic is covered in section 7.2 of DLDE and in chapter 3 in the current book. Note here,
however, that the maxims often overlap. Johns question is not only responded to
irrelevantly, but also fails to be answered, which is a flouting of the maxim of quality too
Flouting the maxim of manner concerns not what is said, but how it is said. Thus, the
intonation, modality, lexical connotation or other aspects of the style of a text, as well as
the level of detail, ordering of content and use of obscure language, may result in a clash
between the context-appropriate style, and that actually used. Thus the use of overly
formal or complex language in talking to students or trainees, may well amount to a
flouting of the maxim of manner, and could be interpreted variously as an attempt to
demonstrate how ignorant the hearers are in comparison with their lecturer or trainer, or
in a more friendly environment, and with the right intonation and body language, it could
be interpreted as humorous. An example of this sort, used to impress first year language
students with no previous linguistic knowledge, might include:
In order to explicate the precise mechanism for the articulation of utterances, we must
consider the egressive pulmonary airstream mechanism and the modification of the
egressive airstream as it encounters the articulatory features of the oral and nasal
cavities.
Though some of the technical terms used here might indeed be introduced to students,
they would normally be glossed or explained and would not be surrounded by
unnecessarily formal and long-winded (prolix) language that is seen here.
Grices theory of conversational implicature has been criticised on a number of counts.
One issue is with whether or not the CP should be interpreted as a system of social,
goal-sharing cooperation (see, for example, Pratt, 1977, Kiefer 1979, Watts 2003,) or
linguistic cooperation (Thomas 1995 and Leech and Thomas 1990). However, Grices
own writings (1975:48) clearly indicate that he does not intend the CP to be viewed as
social cooperation, since there are too many types of exchange (such as quarrelling, for
example) where issues of social cooperation and the CP do not match or coincide.
Linguistic, as opposed to social, goal sharing assumes that the only goal of a given
communication is the transmission of information, and that the CP operates solely to
allow for interlocutors to understand what is said or implied. It makes no claims about the
intentions of the speaker, good or otherwise, but to do with using language in such a way
that interlocutors understand what is said, presupposed or implied.
Another criticism of Grices model is that it applies to how conversation in Anglo-
American English is managed, and thus is open to criticism on the grounds that it does
not take account of different cultures, countries and communities who have different
ways of observing and expressing maxims for particular situations. For example, in
Britain if we say to someone: Ill come and see you tomorrow then we expect the
appointment to be kept and if we then do not turn up, this would be considered a
violation of the maxim of quality. In some cultures however, this is quite a normal way of
indicating lack of interest.
Nevertheless, although investigating conversation in cross-cultural contexts shows that
different conversational norms or conventions operate or apply in different cultural
contexts, the fact that norms can be identified at all points to the fact that they generally
need to be adhered to for effective or successful communication to happen.
Conversational norms such as the ones identified by Grice and others show how various
sentence meanings combine into discourse meaning and are integrated with context, in
much the same way that rules of sentence grammar allow word meanings to be
combined into sentence meaning.
A further criticism of the theories put forward by Austin, Searle and Grice by some
linguists is that, as linguistic philosophers, the research methods they used to arrive at
their conclusions involved solely introspection. In other words, their work was based
upon examining and theorizing ideas about the nature of interaction but without referring
to any actual real language use, or using real language data to inform their ideas (see:
DL:DE I.2). Even so, despite this limitation, the theoretical work undertaken by Austin,
Searle, Grice and others on the nature of conversation has provided linguists with a
theoretical foundation upon which to investigate the linguistic structure of naturally
occurring language. Refinements, correctives, elaborations or replacements of the CP
continue to be published (for example, Hawley, 2002 and Spencer-Oatey and Jiang
2003). Far from being unstructured and improvised, it is clear from speech act theory and
the CP that what appears at first to be so commonplace and trivial is on closer inspection
far more complex. Speech has a performative function, and our ability to take part
successfully in verbal exchanges of any kind depends upon knowing how to manage and
behave in conversation.
2.5 Discourse analysis
Discourse analysis (DA) is both an umbrella term used to discuss any form of discourse analysis
and also the term given to a specific method of analysis. As a specific method of analysis and in
its narrowest sense, it is the term given to discovering and specifying rules that are relevant for the
production of coherent discourse, now also called exchange structure (Cutting 2002:24-7). It is
concerned with how speakers combine utterances into broader speech units. This method of
analysis was first devised by the British linguists Sinclair and Coulthard (1975) and the
Birmingham School of Discourse Analysis. Drawing upon speech act theory (see 2.2. above)
Sinclair and Coulthard proposed a descriptive system to handle the structure of discourse within a
particular setting, that of the classroom.
2.5.1 The structure of talk
The conversational situation on which Sinclair and Coulthard based their work was the
formal, hierarchical setting of classroom talk between teachers and pupils. Their work
showed that a lesson, like any other example of talk, can be analysed sequentially, as a
series of turns or exchanges. During the course of a transaction the teacher would direct
pupils to perform a particular task, elicit some verbal contribution from the pupils to which
the pupils would respond and to which the teacher would make a further response. From
their data, Sinclair and Coulthard identified four main types of exchange: Directive (for
example: teacher: the first quiz is this); Check (for example: teacher: Finished Joan?)
Informative (for example teacher: here are some of the symbols used) and Elicitation (
for example: teacher; read us what youve written Joan).
The structure or composition of any of the four exchanges is determined by the function
the exchange performs. In the case of the informative exchange, the sequence only
requires the teachers contribution. In the case of directives and checks, however, pupil
response is an obligatory element. In the case of elicitiations, the structure allows for
follow-up turns. Consequently, the structure of the exchange can be described in terms
of component moves, of which there are three: initiating (I), responding (R ) and
follow-up (F).
Moves themselves can also be divided according to function. For example, in asking a
question of a whole class, the teacher might restrict the answer to one pupil by naming
them; follow-up moves can be segmented into accepting the response as relevant
(Thats right), evaluating the response (thats a good answer) or commenting upon it
(thats interesting). These various functional components of the moves of an exchange
are called acts.
For example:
Move Act
1. Teacher: Initiating (I): Read us what youve written Joan (elicit)
2. Pupil: Responding (R ): The cat sat on the rug (reply)
3. Teacher: Follow up (F): yes, thats right (accept
Pupil: I changed the last word (comment)
The Sinclair and Coulthard model of discourse analysis can be applied successfully to formal
exchanges which take place in a classroom, but it does not easily transfer to other situations,
particularly informal ones. A major criticism of the model is that it applies to a hierarchical,
formal and ritualistic form of discourse, with one person in a position of power over others, and
therefore ignores the underlying structures of social relationships that sustain power, authority and
hierarchy. Nevertheless, this structural approach is important because it was the first to show that
there is indeed an underlying, systematic, organisational structure to conversation which can be
categorised, and that there are ways of describing it.
When applied to conversation which does not involve hierarchy, such as a conversation between
two friends, then an even greater degree of interpretation and judgement is required than that
provided for by the Sinclair and Coulthard model. It also says nothing about the perlocutionary
force of language and the way language is used to convey humour, irony, threats and so on. It also
assumes that utterances have a single function when they may in fact have more than one. The
categorisation of moves also limits the length of an exchange to the three categories of I, R and F,
or subsequent variations such as R/I and F/I. Even so, the model as one of classroom transactions
still has limitations, not least because it applies best when the transactions are going according to
plan and does not easily allow for disruption. It is also centred very much on the teacher, in
which the teacher dominates with long turns and the students take short turns in response but do
not interrupt. Although such a style of interaction may still occur, there are other kinds of
classroom interaction such as where pupils work in pairs and groups, and exchanges between
teachers and pupils are generally more interactive.
Francis and Hunston (1992) presented a revised version of the Sinclair and Coulthard
model, which aimed to be more flexible and adaptable, and thus able to cope with a
wide variety of discourse situations. The data Francis and Hunston used was based upon
telephone conversations. This was chosen because of the lack of paralinguistic features
such as gesture, eye-gaze and so on, and because the interaction came complete with
rituals of greeting and leave-taking. In their model, the one to one correspondence
between exchange and move is called an exchange structure (e.s), and the range of
moves is extended from four to eight:
1. Framing: to mark boundaries in a conversation
2. Opening: to initiate, including greeting, or to close conversation, or any attempt to
impose structure upon the conversation in some way
3. Answering: indicating a willingness to take part in conversation
4. Eliciting: to inquire, prompt, clarify, elicit information
5. Informing; to offer information, concur, confirm or qualify
6. Acknowledging: to provide positive or negative follow-up
7. Directing: to request an immediate or future action
8. Behaving: to supply or defy a direction.
For example:
Move e.s exchange
B: Why I-its not floating at all eliciting I 1
A: No, its lying on the floor
Like any old balloon informing R
B: Its a bit strange you know acknowledging/ F
Informing I 2
A: Yeah interesting acknowledging R
A very useful exercise in beginning to understand the nature of conversation is to undertake an
analysis of your own, applying either or both of the discourse models outlined above, especially
the exchange structures, to see how far they are applicable, what they reveal about the structure of
the conversation and what their limitations are (see 2.8.5 below).
Although the Francis and Hunston model of exchange structure allows for some degree of
flexibility, it still remains very structural, focusing as it does upon the nature of the exchanges
and developing categories which can be applied to them. Consequently, it has fallen out of vogue
as attention has shifted more towards pragmatics and conversation analysis. In recent years,
however, the study of the structure of institutional discourse has shifted its focus onto studying the
relationship between language and power which has developed into new areas of study such as
Critical Discourse Analysis (see Chapter 4) and language and the law or forensic linguistics.
Nevertheless, exchange structure remains an important linguistic tool in discourse analysis.
Some discourse analyists such as Brown and Yule (1983), Hoey (1983) and Burton (1980) as
well as analysing conversations, have examined written texts. Burton has applied descriptive
frameworks developed for analysing exchanges in conversation to dramatic dialogue, and has
produced a model showing the ways that some types of modernist dramatic dialogue violates or
goes against discursive norms. Consequently, discourse analysis today has widened its field of
inquiry from a focus on exchange structure within a narrowly defined context such as that of the
classroom to discovering underlying structures of organisation across a whole spectrum of texts,
both spoken and written. Generally, however, the field of pragmatics and discourse is largely
concerned with speech, whilst that of stylistics, discussed in the next chapter, is predominantly
concerned with the writing, and especially, though not exclusively, literary writing.
2.6 Conversation analysis
Discourse analysis as outlined above is concerned with the structure of conversation and how
utterances fit together to constitute a discourse, providing a method for its analysis. Conversation
analysis is concerned with the management of conversation and how communication is achieved,
thereby providing a method of analysis for spoken interaction. Whereas discourse analysis comes
from a British tradition of linguistics, conversation analysis comes from an American tradition of
sociology.
Conversdation analysis also differs in its concept of conversation: discourse analysis starts with a
linguistically based theory based upon a patterning of units and fits conversation to the model,
thus viewing conversation as a product. Conversation analysis starts with the conversation itself
and lets the data determine its own categories of structure. Methods of conversation analysis
description and analysis are commonly applied to everyday conversation, although they have also
been used to study interaction in institutions and workplaces: medical interaction, business
settings, broadcast news and so on.
Cutting (2002:28) defines conversation as follows: Conversation is discourse mutually
constructed and negotiated in time between speakers; it is usually informal and unplanned. Cook
(1989:51) defines it as follows:
1. It is not primarily necessitated by a practical task
2. Any unequal power of participants is partially suspended
3. The number of participants is small
4. Turns are quite short
5. Talk is primarily for the participants not for an outside audience
Many linguists would disagree with the first item listed above, it is not primarily
necessitated by a practical task, citing the fact that even the most casual of
conversations has an interactional function, such as ascertaining at a party if future social
interaction with a stranger is desirable, or establishing contact between old friends and
planning future events. Other linguists, such as Fairclough (1989:12) would contest the
second item, any unequal power of participants is partially suspended on the grounds
that there is unequal power, in varying degrees, in all exchanges (see also Chapter 4).
However, although conversation analysis assumes that utterances have a contextual
relevance to one another, not all aspects of contexts are taken as equally relevant. In
practice, much of the work undertaken in conversation analysis has been concerned with
identifying structural features thrown up by conversation, concentrating upon a particular
methodology for discovering how social relations are achieved by participants. CA
focuses upon the conversation as the basis for the existence of patterns, distributions
and formation of rules which can then be applied to subsequent stretches of discourse.
The overall pattern or structure of face to face conversation varies a great deal,
depending upon a host of factors such as where the conversation takes place, the social
relationship between the participants and the topic of conversation. Nevertheless,
conversational analyists point out that speakers are constructing and negotiating their
conversation in time, and as they do, certain stretches of utterances or turns emerge
which shape and structure the conversation.
2.6.1 Turn-taking
Just as in discourse analysis, in conversation analysis, turn-taking is central to the
structure and management of conversation. It is the lynch-pin of analysis, and categories
associated with it have been identified which make it more applicable to analysing
informal conversation. In Anglo-American English, next speakers cannot be sure when
a current speaker is going to finish speaking, but will wait until the end of an utterance as
an indication that the turn may be complete. Stenstrm (1994:68) has identified that the
turn-taking system involves three basic strategies on the part of the speaker: taking the
turn, holding the turn and yielding the turn. A point in a conversation where a turn of
place is possible is called a transition relevance place or TRP. Although as listeners we
can never be sure when a speaker has finished talking, we learn to identify clues to
indicate the end of an utterance to indicate that the speaker is willing to yield their turn.
For example, in Britiah English these are phonological features such as low falling tones
ans syntactic ones such as the end of a mian clause. When speakers do not want to
wait until a TRP, then this is an interruption, whereas if listeners anticipate a turn being
completed, miscue their entry and come in before the end, then this is an overlap.
In the following extract, between a mother and her teenage son, an interruption is marked with //
and an overlap with =.
Mum: I want you to put // your dirty clothes in the washing basket
Harry: // I know what youre going to say. Alright =
Mum = then why dont you do it then?
Another aspect of turn-taking is the acceptable length of a pause between two turns. Each culture
seems to have an unspoken agreement about the acceptable length of a pause between two turns,
and as participants in a conversation we cannot but help to attribute meaning to a pause longer (or
shorter) than this norm. Consequently, analysts categorise pauses that carry meaning as
attributable silence. Consider the following exchange:
Lucy: Did you remember to buy the cheese?
John: (3) No
Lucy: Did you forget?
John: Afraid I did. Sorry
(authors data)
John pauses for three seconds before answering Lucys question, and she attributes this silence to
him trying to remember whether or not he did remember to buy the cheese. As participants in a
conversation, we do not usually feel comfortable when there is a pause or lull in a conversation;
which if it extends beyond about ten seconds, speakers tend to utter fillers such as um or well
then to get the conversation going again or to break the silence.
2.6.2 Adjacency pairs
Turn-taking provides a basic framework for the organisation of conversation since it allows for its
participants to interact with one another rather than simply speaking individually in a haphazard
and uncoordinated way. However, turn-taking on its own is not sufficient to guarantee that there is
purposeful and meaningful interaction. Take, for example, the quote from the playEquus used
earlier:
Dysart: Wont you sit down?
Is this your full name? Alan Strang?
And youre seventeen. Is that right? Seventeen..Well?
Alan: Double your pleasure
Double your fun
With Doublemint, Doublemint
Doublemint gum
Dysart: Now, lets see. You work in an electrical shop during the week. You live with
your parents and your fathers a printer. What sort of things does he print?
Alan: Double your pleasure
Double your fun
With Doublemint, Doublemint
Doublemint gum.
( Equus 1973. 1:3)
In this imagined exchange, the two parties have no problem coordinating an orderly
exchange of turns. However, the content of each turn is not cohesive with what went
before. Alans responses do not answer Dysarts questions. His manner of speaking is
intended to portray madness and he fails to provide a meaningful answer to the question
he has been asked. Flouting needs to be intentional and with an intentional effect. Mental
illness leads to a suspension of the norms. As Cameron (2001:95) points out: CA places
great emphasis on the idea that conversation is one thing after another: it is an activity
that unfolds in time, and what I say now must inevitably constrain what you can
meaningfully say next. It also looks for patterns in data so in this case, conversation
analysts might collect data from real situations with people in mental health settings, to
see whether Shaffer has got the patterning of the turns right: whatever right means in
a creative context. In the example above, Dysarts questions are a request for
information. Alans answers make the information requested sound straightforwardly
appropriate.
Thus, although turn-taking involves the participants in a conversation taking turns to speak, the
turn-taking itself is structured around utterances which occur one after the other, and in which the
second utterance is functionally dependant upon the first. This phenomenon uses the label
adjacency pairs to refer to a sequence of two utterances, adjacent to one another, produced by
different speakers and ordered as a first part and second part. The most obvious adjacency pairing
is that of question-answer. When we ask a question, we usually expect a response, known as a
preferred response. A question, therefore expects an answer, but can also be answered by a
question, which is a dispreferred response. Adjacency pairs are numerous, but the following is
an indicative list of the major cases:
a question has the preferred response of an answer
an offer an acceptance
an invitation an acceptance
an assessment an agreement
a proposal an agreement
a greeting a greeting
a complaint an apology
a blame a denial
(taken from Cutting 2002:30)
Dispreferred responses tend to be refusals and disagreements which can be interpreted as either
meaningful or rude. For example, in offering the greeting How are you? the response is Go
away!, then this would be a dispreferred response. Whether or not it is meaningful or rude would
depend upon the circumstances and context in which the exchange took place.
Consider the following exchange:
1. John: have you finished that yet?
2. Paul: no, Ive just got to get an email off to John Harrap.
1. John: oh good, well, when youve done, would you give me a hand?
2. Paul: (.) doing what?
1. John: something I have to get done for Monday. Itd be great if you
took a look at it.
2. Paul: (.) Ok. Give me five minutes.
(authors data)
In the first adjacency pair, John asks Paul a question, to which he replies (preferred
response). John then asks Paul a second question, who stall his reply by answering a
question with a question to find out what it is John wants (dispreferred response). John
then phrases his request as a type of proposal, to which Paul agrees. In CA terms, this
preference system refers to the patterning of responses, and not to the more usual
meaning of preference, where something is considered more desirable than something
else. Preferred responses are prompt and short whereas dispreferred ones tend to be
more hesitant and elaborate. Acceptances and agreements are typically structured one
way, and refusals and disagreements in another.
In conversation analysis, analysis of turn-taking has also established the idea of sequence in
conversation, of which four have been identified: opening and closing, pre-sequences and
insertions. Take the following example of a consultation between a doctor and a patient:
(Doctor leads patient into the consultation room, talking as he does so).
D: thats only the clutter in the (.) background
(.)
P: yeah
D:no problem
(.) 5
so (.) take your coat off
P: sure
(4.0) ((Doctor helps patient remove his
coat))
Thanks 10
D:dont think Philips got any clothes pegs in
here so uh (1.5) ((Doctor hangs up coat))
I dont usually use thi- sit down
P: fine
D:I dont usually use this room its erm (.) 15
Philips
(1.0)
P: yes, of course
D: anyway (.) going through the whole thing
(2.0) 20
youve changed your job(.) in effect
(.)
P: well (.) additional responsibility
(taken from Montgomery 1995:205-6)
In lines 1 to 10, the doctor and patient are engaging in an opening sequence centred upon getting
settled into the consultation room From line 12 onwards, insertions centre upon the doctor
checking the patients file. Since consultations are time specific, then the exchange ends with a
closing sequence. Cutting (2002) notes that British and North Americans preface closings with a
pre-closing sequence which can be long and drawn out. Rather than simply saying goodbye or
tarrah, markers are usually given to signal that a conversation is about to end, such as anyway,
Ive got to go, its been nice talking to youand so on.
Turn-taking strategies as described above appear to form an orderly system where one person
speaks at a time whilst another waits their turn. However, ordinary conversation does not always
work like this, since the person waiting to speak may decide not to wait their turn and butt in, or
lose concentration and/or interest and fail to respond, or speakers may lose their thread and so on.
In the normal course of events, as Montgomery (1995:194) points out, we expect our
conversations to proceed without such difficulties, because of the underlying patterns we have
absorbed that give rise to such expectations.
Even so, there is a growing body of research, which has questioned the universality of turn-taking,
and especially the one person speaks at a time principle upon which it is founded. This is
especially so in relation to the way men and women behave in conversation. Jennifer Coates, for
example, argues that the way turn-taking functions between women and between men is different,
based upon her analysis of a large corpus of informal talk between British women friends. She
found that turns were more usually jointly or co-constructed between these speakers than singly.
She views this kind of talk as highly cooperative, and concludes that it is more common amongst
groups of all female speakers than all male. A similar corpus of informal talk among men (Coates
1997), confirmed this conclusion, as she found that patterns corresponded more closely to the one
person at a time notion of turn-taking. Whilst Cameron (1997) supports Coates findings in an
analysis of the talk of young men friends.
Whilst Coates and Camerons studies have concentrated upon informal talk, Edelsky (1981)
provides evidence that supportive simultaneous speech can also happen in institutional contexts.
In an analysis of talk at business meetings, she distinguishes between two types of floor, F1 and
F2. F1 refers to the one speaker speaks at a time model of turn-taking, whereas F2 involves
more simultaneous speech which, according to Edelsky, enabled women to take the floor more
than was the case with F1. It seems then, that in both talk among friends and in some kinds of
institutional talk, it is possible for turn-taking to be organised other than on the one speaker
speaks at a time principle. As Cameron (2000:93) points out, the one at a time model is
presented as both normal and fundamental, and there is no obvious place within it for
simultaneous speech which is neither a violation nor a mistake, whereas it may be in fact, that it is
a normal feature of certain kinds of conversation, such as informal ones Studies undertaken in
interactional sociolinguistics and intercultural communication also show how norms associated
with turn-taking are cultural, rather than a linguistic, and different cultures have different turn-
taking norms (see 2.6 below).
In another study, Shaw (2000) examined the speech behaviour of male and female members of
the British Parliament in debates in the House of Commons. The House of Commons has arcane
rules which are highly codified regulating who may speak, when, and about what. One of these
rules is that in a debate, contributions can only be made by the MP who is standing up, and
contributions from the floor are not allowed and are illegal, although in practice many MPs
make such contributions, and the Speaker (effectively the chair) rarely censures them unless
called upon to do so by the MP holding the floor. The Speaker is the person responsible for
ensuring that this rule, among others, is observed. In an analysis of five debates, Shaw found that
male MPs were far more likely than women to make illegal interventions in other peoples
speeches, not because the latter were reluctant to intervene, but because they wished to avoid
illegal interruptions. By playing by the rules of the game more often than men, Shaw found that
women MPs were less likely to influence a debate than men, who were more inclined to exploit
rules. Shaw concludes from this that in this particular context, by being more inclined to play by
the rules, women MPs are disempowered by comparison with men.
When it comes to topic changes in informal talk, in their 1988 study of conversational shift
work, West and Garcia, investigating talk between men and women, found that men tended to
initiate unilateral topic changes in the course of womens turns-in-progress: that is, they would
change a topic completely whilst women were still talking about a previous one. These topic shifts
occurred most often when women were talking about affective (emotional) experiences.
Thomas (1995) argues that, whereas in other areas of linguistics such as grammar the linguist is
striving to make rules, pragmatics is concerned with probability rather than certainty. However,
there is a growing body of research which suggests that men and women acquire different norms
of turn-taking in both informal and institutional settings which, particularly in settings that involve
both men and women, see the balance of power and influence in mens speech behaviour rather
than womens. This issue is explored in more detail in Chapter 4. Conversation analysis also takes
nothing as predetermined, and views the turn-taking system and the relevance of particular actions
as always being governed by contingency. Contingency and probability is what conversation
analysis is all about. Coates notwithstanding, much conversation anbalysis research has addressed
overlap, interruption, collaborative completion, choral responses and the like. All of these
practices, rather than constituting evidence against the conversation analysis account of turn
taking, demonstrate instead the exploitation of the ways that turns at talk allow for the projection
of trajectories.
2.6.3 Presupposition
One important aspect of conversation is the amount of implicit background information
and common ground shared by the participants in a conversation. For example, take this
extract of conversation between two friends walking their dogs:
Lucy: Hows Lawrence?
Pamela: Hes OK thanks. Still hobbling but getting better slowly.
Lucy: Whens he back at school?
Pamela: Oh, next week some time
(authors data)
If Lucy were to clarify in completely linguistic terms what she means by Hows
Lawrence? then it might go something like this:
Assuming, Pamela, that you have normal hearing capabilities and a workable knowledge
of English, to understand me, I am addressing you. I guess that you, like me, do not want
to undertake this walk in silence but that we both want to interact socially by means of a
conversation. Since I know that your son has been in hospital recently for an operation
on his knee, it seems reasonable for me to start a conversation by enquiring about his
progress. So I am asking you Hows Lawrence? And I would appreciate it if you could
say something in reply to the question.
No matter how elaborate this attempt is, it still leaves much that is implicit. It could be for
example, that Lucys enquiry is asking for a more general update on Lawrences health
and well-being beyond the operation. Complete explicitness is thus impossible in what
we say and write, and language provides numerous conventionalized carriers of implicit
meaning, tools for linking explicit content to relevant aspects of background information.
One important tool is presupposition - that is, aspects of meaning that are presupposed,
understood or taken for granted for an utterance to make sense. Hows Lawrence? for
example, presupposes that both Lucy and Pamela know the Lawrence being referred to,
and furthermore that he is Pamelas son. This pragmatic presupposition cannot be
derived from the semantic base of the utterance; that is, the meaning of the words.
Rather, it resides in the shared conventions of language use rather than in the more
formal patterns of its logical structure. One concept of presupposition is that of
entailment. Entailment is the most literal component of the meaning of a sentence,
which remains the same or stable in whatever context it occurs. Take, for example, the
sentence:
The cat sat on the mat
This sentence entails a basic proposition p. indicated in italics:
P: the cat sat on the mat
The concept of entailment also extends to sentences or utterances which contain more
than one basic proposition. For example, the sentence:
Susan ate an apple and Peter ate a pear
Entails two propositions, expressed as p and q:
P: Susan ate an apple
q:Peter ate a pear
Assuming that the sentence expresses true claims, (that Susan did eat an apple and
Peter a pear), then it follows that p and q must also be true (see also 2.3.1 above and
TL:TE Part II 2.1) Entailment, then, can be defined more rigourously as:
A sentence S entails a proposition p if and only if in every possible circumstance where S
is true, p is true (Simpson 1993:123).
If someone says to you that the cat sat on the mat or Susan ate an apple and Peter ate a
pear, then you would expect the person to be committed to the truth of the entailment.
However, as Simpson (1993:123) points out, the concept of truth is a complex and
abstract one, the definition of which has exercised linguists and philosophers (see TL:TE
Part II 2.2). A great deal of effort has gone into explaining the conditions under which a
particular sentence maybe true or false. These truth-conditions form the foundation of
semantic description by specifying the components of meaning that give a sentence its
overall truth-value.
However, the concept of entailment, although an important aspect of sentence meaning,
does not provide a complete description of what utterances mean in a real situation of
use. Take the following example (from Simpson 1993:124):
Well darling the dog has ermstopped sleeping in its kennel.
Literally, this sentence means something like: the dog does not sleep in its kennel. If the
utterance as a whole expresses a true claim, then the entailment must also be true..
Presuppositions usually fall into one of two types: existential and logical. Existential
presuppositions are ones which state the existence of referents in the utterance, and are
triggered by definite noun phrases, in this case the dog. Logical presuppositions are
triggered by a number of possible features, including most commonly change of state
verbs such as stop, factive verbs such as regret or realise whose complement is
presupposed to be true, iterative adverbs such as again or still, which presuppose that
the process has happened before, cleft constructions such as it was John that let the dog
out, where the whole subordinate clause is presupposed to be true and comparators,
where the quality asserted is assumed to imply also, in some measure, to the referent, as
in This cake is richer than the last one, where the entailment is that the first cake is rich,
and the presupposition is that the second one was also rich in some degree. If the dog
has stopped sleeping in its kennel, then the logical presupposition arising from the
change of state verb, stop, is that it had previously been sleeping in the kennel.
Grammatical constructions that trigger a logical presupposition are known as triggers, of
which factive verbs is one. Others are: cleft sentences, such as: It was Ben who stole
the meat, which presupposes someone stole the meat; iteratives, such as: you cant
buy music cassettes any more, which presupposes that once you could and
comparators, such as: Peter is as sexist as Paul, which presupposes that Paul is sexist.
To return to the example of: well, darlingthe dog hasermstopped sleeping in its
kennel. The additional elements of well darling and erm give additional meaning to the
sentence. Use of an endearment marker such as darling usually indicates some
relationship of intimacy between the speaker and addressee. This information is not
relevant to its truth conditions, but forms part of the overall message. The particles well
and erm also lie outside truth conditions, but have an important function as hedges,
softening the impact of the message upon the addressee. Consider also the context in
which the utterance occurred, which was as part of an exchange between A and B:
A; What are all those hairs doing on the sofa?
B: Well, darlingthe dog has ermstopped sleeping in its kennel.
This contextualization of Bs utterance alters its meaning from the literal one of the dog
has stopped sleeping in its kennel to the inferred one that the dog is responsible for the
hairs on the sofa. Such a process of inferencing is triggered by the indirectness of Bs
answer to As question. The meaning which the exchange as a whole generates is a
combination of Bs indirectness and As inferencing work, a process known as
conversational implicature (see also 2.3.2 above).
The impossibility of being fully explicit in language lends itself to strategic exploitation.
For example, if asked why you did not make it on time for a meeting, you could answer I
didnt manage to get away. You may not have tried to get away, but by using the word
manage gives the impression that you did, a logical presupposition based on the factive
verb manage, which presupposes trying. You have not actually said that you tried to get
away, so no one can accuse you of lying. This is called presuppositional lying.
2.6.4 Politeness
Politeness in conversation refers to the choices people make in linguistic expressions
that show a friendly attitude towards speakers and listeners, including giving them space.
It lies not in the words and expressions themselves, but in their intended social meaning
and function. Brown and Levinson (1987) in their analysis of politeness, point to the key
aspect of face in social relationships. Face refers to self-image, and it is a universal
characteristic across all cultures that speakers should take account of their listeners
feelings, respect each others expectations regarding self-image and avoid face-
threatening acts (FTAs). Where FTAs are unavoidable, Brown and Levinson suggest
that the threat can be lessened through negative politeness. Negative politeness
respects the hearers negative face, which is recognising the need to be independent, not
to be imposed upon by others, to have freedom of action and generally to minimise
imposition. Alternatively, speakers can address the FTA by positive politeness. Postive
politeness attends to hearers interests, wants and needs. It respects the hearers
positive face, the need to be liked by others, to be treated as a member of a group and
so on.
One way of dealing with FTAs is to do them off record or on record. For example, if
you are in a supermarket looking for a particular produce and no assistant is to be found.
You would like some help, but do not wish to approach any other shopper directly since
you do not wish to interrupt their shopping and place yourself in a potential FTA. You
could ask for help indirectly, by saying in a voice loud enough for other shoppers to hear:
Where have they moved the breakfast cereals to? In this instance, you would be
speaking off record and using an indirect speech act (see 2.4 above) in the hope that a
fellow shopper will hear your question and respond to your request for help. The
illocutionary force will probably be understood by hearers, but they can choose to ignore
it without anyone losing face. Avoidance of FTA ia also achieved by flouting the maxim of
quantity (see 2.4.2), since you are not saying openly that you need help and thus not
making your contribution as informative as possible. Indirect speech acts and flouting
maxims of quantity in this way allows speakers to make requests, offers, invitations and
so on without addressing them to anyone in particular and thus avoiding any potential
FTA.
Alternatively, you have the option of approaching someone directly by going up to them
and saying: Excuse me, you couldnt tell me where theyve moved the breakfast cereals
to could you? This person then has to reply unless they want to be rude. Making a
request in such a direct way is called doing an FTA on record. On record FTAs are
direct speech acts. In this example, the speech act is phrased as a question as a
mitigating device, thereby giving the listener the option of saying no or of giving you the
help you need. Had you said to a fellow shopper tell me where the breakfast cereal is
then this would be a bald on record communicative act, with no mitigating devices, and
leaving the hearer with no option to answer or otherwise be thought of as uncooperative.
Consequently, this is the most face-threatening mode of action. It is also possible,
however, to orientate bald-on-record events to save the hearers face. For example, if
you say to someone at the meal table have another helping, then there is little risk of the
hearer feeling imposed upon. The degree to which such directness is face-threatening
depends upon how socially close we are to our hearers. The closer we are, then the
more direct we are likely to be; the more distant, then the more face-threatening such
directness will be.
Generally, speakers perform FTAs on record in ways which take account of face by face-
management. This can be done on record and using either negative or positive
politeness.
Negative politeness strategies, as defined above, pay attention to negative face.
Speakers use them to avoid intruding upon hearers territory and also give the hearer
options. For example, you couldnt tell me where the breakfast cereals are, could you?
Uses negative politeness by using a question in this way gives the opportunity to say no.
Other strategies commonly used in negative politeness include apology and hesitation.
Postive politeness strategies aim to save positive face, by appealing to friendship and
demonstrating solidarity. Asking where the breakfast cereals are with on-record postitive
polteness, for example, would mean approaching a shopper and saying something like:
excuse me, Id appreciate it if you could tell me whereabouts in the store you got that
packet of breakfast cereal. One of the main types of positive politeness strategy
identified by brown and Levinson (1987) is claiming common ground. Speakers can do
this by attending to hearers wants, needs and interests. For example I know you dont
like loud music, but this concert will be fun Ol. Mark and Chrisll be there and we havent
seen them for ages. This example uses many solidarity strategies, such as nicknames,
knowledge of personal information and seeking agreement.
Take the following example, which is an extract from a telephone conversation between
a boyfriend and girlfriend:
1. Mary: I (.) sometimes I do: think: I I suppose I have (.) I (.) I have felt that maybe you
2. James: were cross with me
3. Mary: cross with you darling
4. james: yeah: I I had =I had
[Link]: =I wasnt cross with you darling
6. James: mmm
7. Mary: well (.) cross with you about what darling
8. James: but you know I I kind of (0.1) but you know what Im like sort of thinking that:
=Im always
9:Mary: =right
10. Mary: thinking that Ive offended: and that: oh! No! I mean (.) Im not! I think (.) erm
(authors data)
Mary performs an on record FTA in turn 1, in that she is voicing a concern about James
behaviour which he may find potentially threatening. By turn 6 she has still failed to
provide an answer to the James question cross with you about what. In her responses,
Mary not only flouts the Gricean maxims of quantity, manner and relation, but she also
employs negative politeness and another politeness strategy, that of extensive hedging:
but you know I I kind of I think (.) erm. Both Cameron (2001: 211-2) and Brown (1980:
117) suggest that women may rate FTAs more severely than men. Marys behaviour in
the extract above bears this out, evidenced by her difficulty throughout to tell James of
his tendency to think that he is always right.
2.6.5 Context
Much is said within pragmatics and discourse analysis about the importance of studying
verbal behaviour not only as action and its management but also its context (see
[Link] and TL:TE Part II Chapter 5). Take the following example of two
undergraduate students, Claire and Peter, talking about a visit Peter had planned to
visit Claire at her parents house which had not turned out at all well:
Claire: hello, what happened to you?
Peter: (2) well, I got the train as far as Birmingham but then there was this massive
delay=
Claire:= what did you do? was there another train?
Peter: I had to go back to Leicester because it was impossible to go on. So then I
thought Id get the coach, but when I got to Leamington I didnt have your address.
Claire: oh no! why didnt you phone me?
Peter: my phone was out of battery, and Id hurt my ankle when I ran for the coach. Look,
its swollen like this.
Claire: poor you! so what did you do?
Peter: (1) um I got the train back to Leicester.
Claire: wait until I tell Debbie!
(authors data)
As with presupposition, there is a great deal of knowledge which the two students share
regarding the context of this situation. Presuppositions , though, are not really the shared
background, but an assumption of shared background, which may not in fact be shared.
Typically these are contexts of:
> situation what the speakers know about their physical and geographical
surroundings;
> background knowledge what the speakers know about one another and the
world;
> co-textual what the speakers know about what they have been saying.
The situational context in the example above, is Claires study bedroom at university
halls in Leicester. When Peter mentions his hurt ankle, he points to it and refers to it as
this. This referential quality of words such as this is known as deixis or
deictic expressions. Words which signal the use of deixis include personal pronouns (I,
you, we, they etc), demonstrative pronouns (this, that, these, those) and the article or
determiner the and adverbs of time and place. The more bound to a situation a
conversation is, then normally the more deictic expressions are used. (see 3.x for a
fuller discussion of deixis and TL:TE Part II 5.2).
Referring expressions and referent(s) are terms used to describe the identification of
something or someone and identifying the entity being referred to. For example, in the
words I shared a car with Frank and Michael, the first person singular pronoun I is a
referring expression that refers to the person speaking, who is known as the referent. In
the same way, Frank and Michael are referring expressions which refer to two people
named Frank and Michael, who are the referents.
The background knowledge speakers bring to a conversation is generally of two kinds:
cultural and interpersonal. In the conversation above, Claire and Peter share a cultural
knowledge of the geographical locations of particular English cities and towns and the
difficulties of traveling between the particular ones mentioned by public transport. Claire
is not overly surprised that Peter had difficulties traveling by train, because he undertook
his journey on a Saturday when trains are notoriously subject to delay and cancellation
without notice because of engineering works. The reference to a mobile phone battery
being dead is also another instance of shared knowledge, since Peter does not have to
explain what this means or why it meant he could not communicate with Claire. Talk
assuming shared knowledge of cultural context often presupposes or shows an
assumption of shared attitudes towards it. As 2.5.4 above discusses, speakers often
modify their expressions to accommodate those of other speakers in order to be
accepted and be seen as belonging to the same group.
Consideration of the situational context and shared knowledge is a central concern of
another aspect of pragmatics and discourse, namely interactional sociolinguistics, which
is discussed in more detail in 2.6 below.
Interpersonal knowledge refers to knowledge that has been acquired through previous
exchanges or activities and experiences, and can include personal, privileged knowledge
about a speaker. In the example above, both Claire and Peter know who Debbie is.
Peters interpersonal knowledge of Claire, however, does not extend to having previously
visited her at her parents and the situational context of the exchange leads one to
conclude that they are contemporaries at the University and their relationship dates from
them meeting at University rather than at some earlier time.
Finally, the co-textual context looks at both the context and the content of the text itself:
that is, the actual words spoken, and the ways in which the text uses lexical and
grammatical cohesion to express its meaning. This aspect of context is dealt with more
fully in section 3.4 of the next chapter.
The categories of turn-taking, adjacency pairs, presupposition, politeness and
consideration of context are all major methods of analysis not only in conversation
analysis but also in pragmatics in general. Although a useful tool for analysis, both
discourse analysis and conversation analysis have been criticised on the grounds that
their categories of description such as turn-taking and adjacency pairs described above
are not exhaustive, or how, in conversation analysis, TRPs (2.3.1 above) might be
recognised. As a method of analysis, both are qualitiative in nature; that is, they analyse
a finite set of segments of conversation in detail, and do not lend themselves easily to
quantitative analysis, that is, analysis of a large number of conversations. In conversation
analysis, methods of analysis derive categories from conversation itself, aiming to
provide descriptions of how conversation works. Since its description and analysis
applies to real-time activity, then there is a sense in which any attempt at description or
categorisation within any framework, conversation analysis or otherwise, can never be
exhaustive or claim to be so since it has to allow for the unpredictability of real-time
activity, and its capacity to generate new or different ways of interacting. Whilst the
categories of analysis within discourse analysis are finite, it has been criticized for its
hierarchical structure and for the fact that they derive from institutional settings which, it
is argued, make them less applicable to all naturally occurring communication. However,
there is a sense in which the ideological relations underlying all discourse means that all
discourse is inevitably hierarchical, and has led to the development of critical discourse
analysis, discussed in more detail in Chapter 4.
2.7 Language, action and interaction
One further approach to discourse analysis is known as interactional sociolinguistics,
which focuses upon the importance of small and subtle variations in the ways people use
and interpret discourse. The work of Gumperz (1982b) has been particularly influential in
this area. Practitioners of discourse analysis and conversation analysis have tended to
criticize interactional linguistics for being both too broad and too context specific as 2.2
above has discussed. However, although the main goal of interactional sociolinguistics is
not to describe the structure of discourse, which is the main goal of both discourse
analysis and conversation analysis, the two approaches are beginning to come together.
Sociolinguistics of the kind described in Chapter 1 centres upon the notion of the
linguistic variable in speakers grammar, vocabulary and pronunciation in identifying
different ways of saying the same thing (see 1.1. and 1.3). Interactional sociolinguistics
goes one step further, and takes the view that aspects of interaction such as rules of turn-
taking, conventions for signaling agreement and acknowledgement and the marking of
utterances as particular kinds of speech acts are also variables in that they are used
differently in different contexts or by different kinds of speakers. As discourse analysis
and conversation analysis have developed categories of description and analysis over
recent decades, this has enabled interactional sociolinguists to take into account both the
pragmatic and sociolinguistic aspects of interaction, which also includes consideration of
aspects of discourse and conversation analysis such as turn-taking, adjacency pairs and
sequences.
In addition, interactional sociolinguistics places importance upon the way in which
language is situated in particular circumstances in social life. It brings the situational
context and shared knowledge described in 2.5.5 above, and particularly about speakers,
their histories and their purpose in speaking, to the forefront. It also looks at grammar,
social structure and cultural patterns. Several studies have documented the complex
interplay of factors shaping turns and sequences in spoken interaction in a variety of
languages, showing how interactional features contribute to achieving particular linguistic
and communicative outcomes. This is particularly so with analysts who consider the
relationship between grammar and social interaction within the larger sphere of human
conduct and the organization of human life (e.g. Ochs, Schegloff and Thompson 1996).
Like practitioners of discourse analysis and conversation analysis, interactional
sociolinguists draw upon naturally occurring interactions for their data. Whilst discourse
analsysis and conversation analysis have traditionally focused upon native Anglo-
American speakers, many interactional sociolinguists follow in a tradition initiated by
Gumperz in focusing upon speakers from different cultural backgrounds. This has led to
a field of inquiry known as intercultural communication, sometimes also known as cross-
cultural communication, which tends to centre upon interaction in institutional settings
such as schools and businesses. Key features of such communication or, as is more
often the case miscommunication, are contextualization cues, crosstalk and uptalk.
Contextualization cues are signals which convey to the listener some quite complex
information about how we intend them to treat the message, conveyed through: prosody
(intonation, pitch and stress contrasts); paralinguistic clues (hesitation, pausing, contrasts
of speed and volume, simultaneous speech); or switching to a different language, dialect
style or register to signal an utterance as having particular significance. For example, in
Anglo-American English, the signaling of a question can come from rising intonation at
the end of a word. Just the word Gravy? spoken with a rising intonation indicates that
the speaker is asking whether or not a person wants gravy, presumably on their meal.
Gumperz (1982b) cites the example of a workplace in which white workers ate their
lunch in a canteen staffed by Asian workers. A customer was asked if he wanted gravy
on a meal which included meat by uttering the word gravy, not with a rising intonation
but a falling one, which made it sound more like a statement: this is gravy or like an
assertion: Im giving you gravy which was perceived as rude. But, in Indian varieties of
English, falling intonation has the same meaning as rising intonation in Anglo-British
varieties. Since neither group was explicitly aware of these differences, then
misunderstandings can occur.
Crosstalk refers to the assumptions we all make about the norms and conventions of
particular speech events, which may differ culturally. For example, in an interview extract
cited in Roberts et al (1992:131), a White interviewer and Asian interviewee have the
following exchange:
A: What have you been doing since you were made redundant?
B: Nothing
The interviewers question is a veiled invitation for the interviewee to reassure the
selectors that he has in fact been doing something since the redundancy since in the
culture to which the interviewer belongs, unemployment is seen as not only a misfortune
but also a moral failing. However, speaker B does not pick up on this hidden message,
and answers factually and truthfully, qualities considered interview-appropriate in his
culture. Such miscommunication can have potentially serious and damaging
consequences as, in this example, the interviewee could fail the interview due to failing to
pick up on hidden messages or agendas.
Uptalk refers to the intonation pattern thought to originate in Australian English, but
found in American English and increasingly in British English, especially among young
people, where intonation rises at the end of declarative sentences. For example, I went
shopping today? Since rising intonation in Anglo-American English is associated with
questions, then this pattern can irritate and puzzle other speakers, especially older
people. However, an example given by Cameron (2001:113) is that of a study
undertaken amongst young women who used uptalk. The focus of the study was to test
the assumptions that firstly, there is likely to be variation in the use of a socially and
linguistically meaningful feature: uptalk speakers would not use rising intonation on every
declarative sentence and secondly, that the pattern of variation would be structured. The
pattern uncovered by the study was to do with the status of the information being relayed
in an utterance: when the information given was new information considered important
by the speaker they would make it with rising intonation.
2.7.1 Grammar and interaction
Much of the work undertaken in pragmatics and discourse analysis discussed so far in
this chapter has centred upon identifying categories and structures of the way in which
speech is structured and managed, and has not been as concerned with the linguistic
structure of the utterances themselves. However, just as it was once assumed that
speech was unstructured and without pattern, so too the assumption that spoken speech
follows the grammatical rules of written language has been challenged. Grammars of
written English almost invariably use constructed sentences to illustrate their points.
However, once one moves away from constructed sentences to real language data,
especially that of speech and particularly conversation, then it becomes clear that the
categories designed to describe written speech do not transfer across very easily.
Features such as false starts, hesitations and reformulations are common in speech, but
grammars designed to describe written English prove inadequate to describe them.
Carter and McCarthy (1995 and 2006) and Douglas Biber et al (1999) for example, have
shown that the grammar of speech has many characteristics that are different from those
found in writing. Their studies have been made possible by the application of
computational techniques which allow for the processing of natural language data.
Computers are able to store large numbers of texts, and these collections are known as
corpora (the singular form is corpus) which can then be processed by using specialist
software. This has given rise to a whole new area in linguistic study, that of corpus
linguistics. Corpus analysis has allowed researchers to look in detail at recurring
patterns of use within corpora and how they relate to the context in which they is used
(see also 3.x).
Carter and McCarthys (1995) discuss the spoken data taken from a mini-corpus which
is part of a much larger one. This consisted of two and a half hours of transcribed talk
from a variety of male and female speakers of different ages and from different
backgrounds in a variety of different contexts: service encounters such as shopping,
talking to a bank manager; narratives such as retelling stories; casual conversation and
language in action; that is, the talk which accompanies tasks such as preparing a meal,
moving furniture and so on. They isolated four grammatical features that were common
across all types of speech: Ellipsis; left dislocation and topical information;
reinforcement and indirect speech.
Ellipsis is the leaving out or omission of parts which would otherwise be required in a
structure was found to be a common feature across the corpus. For example:
[At a dry-cleaners. <02> is leaving a pair of trousers for cleaning]
<01> Wednesday at four be okay
<02> Er yeah thats finejust check the pockets a minute
In this example, the initial will or would is missing from <01> and Ill is missing from the
second clause in <02>. Carter and McCarthy found that ellipsis was a feature of service
encounters, casual conversation and language in action, but was noticeably absent from
narratives. This is because the participants and processes of the story are normally
separated in time and place from the moment of telling.
Left dislocation refers to the phenomenon where items semantically co-referential
with the subject or object of the clause are positioned before the subject. For example:
<01> Well Sharon, where Im living, a friend of mine, shes got her railcard and.
In this example, a friend of mine is repeated as the subject pronoun she. Carter and
McCarthy argue that left-placed or fronted items of this kind are perfectly normal in
conversation, and especially in narratives: It is apparent that speakers use the available
slot to flag a variety of items of information that will be helpful to the listener in identifying
participants, in linking current topics to already mentioned ones, in reactiviating old
topics, and generally anchoring the discourse(1995:143) Carter and McCarthy term
the slot at the front of a clause used to carry topic-prominent items as topic.
Just as there is an available slot at the front of the clause called topic, so too, is there a
final slot, when all other core clause constituents have been exhausted, which speakers
often fill with different types of information. Tags often occupy this slot (such as: youre
daft, you are) as do other noun phrases. For example:
<01> Its lovely
<02> Good winter wine that
and:
<01> its very nice that road up through Skipton to the Dales [<02> Yeah] I cant
remember the names of the places.
Indirect speech is usually described as X said that Y where the reporting verb (for
example tell) is in the simple past tense. However, in Carter and McCarthys corpus,
examples of indirect speech had the reporting clause in past continuous. For example:
<01> Tom was saying they should have the heating on by about Wednesday.
Carter and McCarthy suggest that the past continuous seems to report the event of the
uttering, whereas the past simple gives more authority to the words uttered. The use of
the past continuous form of a reporting verb was found in every context except for the
narrative one, which may indicate that what someone said and what words they used are
considered equally important. In casual conversation, past continuous saying and telling
emphasise message content over its form, and to summarise or report entire
conversations rather than individual utterances.
The work of Biber et al (1999) has concentrated more upon types of structure
characteristic of speech than distinguishing different contexts at a finer level of detail.
They distinguish between the body of a speakers message which carries the main
content, and optional preceding elements, or prefaces and following elements or tags.
For example:
Preface: North and south London
Body: theyre two different worlds
Tags: arent they? In a way
Prefaces function as conversational launching devices which Biber et al call utterance
launchers and signal the beginning of a turn or an utterance. They include
fronting, noun phrase prefaces, discourse markers and overtures.
Fronting is a device used for the management of information, where word order is
capitalized upon to give prominence to one element in the immediate context. The
grammatical structure of a clause commonly follows the SVO/P/A word order of English
(see DL:DE xx and 3.x), which is the same for speech as for writing (disregarding
ellipsis). The ordering of 0/P/A S is known as topicalisation or fronting found in a
restrictive set of conditions. It gives prominence to one element in an immediate context,
and its use is thus restricted and not very common. For example:
A: You always remember numbers. Dont you? Car numbers and telehone numbers
and
B: Car numbers I remember more by the letters than the numbers.
Noun phrase prefaces occur in speech more frequently, dividing clause frames into two
chunks, where a pronoun co-refers to a prefaced noun phrase as in the following
examples:
This little shop its lovely
Those Marks and Spencers bags, can you see them all?
You know, the vase, did you see it?
In sentence grammar, such utterances would more usually be expressed as a single
clause with no co-referential pronoun, for example, This little shop is lovely.
Discourse markers are single words or phrases which signal that something is about to
be said and to gain the listeners attention. They are a particular feature of the grammar
of speech and are not normally found in writing. Words such as right, erm, well, oh,
okay, so and phrases such as so anyway, you see, look here, or in combination okay,
you see or well, so anyway.
Overtures are expressions which are longer than discourse markers and a more
explicit way of signaling a new direction in conversation, and the list is more open-ended.
For example: I would have thought has the pragmatic force of signalling a point of
disagreement; there again of adding a contrasting point to an argument and going back
to signalling a return to an earlier topic.
Whilst prefaces describe what comes before the main body of an utterance, tags
describe afterthoughts or qualify what has already been said. Biber et al identify four
different types of tags: retrospective comment clause, retrospective vagueness
hedges, question tags and noun phrase tags.
Retrospective comment clauses refers to the speaker adding a comment which modifies
the preceding clause. For example, Mm I wouldnt go into Amanda Close I dont
think whereas retrospective vagueness hedges signal that the speaker has some
doubt about what has just been said: For example, North and South London theyre two
different worlds arent they? In a way. Question tags have the pragmatic force of
turning a statement retrospectively into a question and an interactive function of wishing
to elicit a response. For example: You had a nice trip though yeah?.
Finally, noun phrase tags repeat and elaborate a preceding noun phrase, most
commonly with the aim of clarifying what might not at first have been clear. For example:
Hes had a blind up-a special blind that leads straight across the fanlight. They are
also the opposite of noun phrase tags, where a noun phrase links to a preceding
pronoun, for example: Oh, I reckon theyre lovely. I really do whippets. Biber et al
conclude that the main motivation for this device is the need to clarify retrospectively. In
speech, the use of a personal pronoun is most common, which the speaker then adds a
clausal unit with a clarifying noun phrase to make the reference clear.
Carter and McCarthys Cambridge Grammar of English (2006) draws upon extensive
corpus research in arriving at its descriptions and categories and identify nine basic
forms of spoken grammar. Like Biber, they too recognize the category of fronting, which
they call heads, and Bibers retrospective comment clauses are called tails. They also
recognise the category of discourse markers, and ellipsis, identified in their 1995
study. They add further categories of: deixis (could we just move that into this coner
here); modal expressions (I suppose it must be sort of difficult to phone or
whatever); adverbials (you know which one I mean probably); and spoken clause
structure which includes the previous 1995 categories of topical information and re-
inforcement and vague language (<SO1> do you think it is affected by your faith, like
you were saying you [<SO2> mm] have kind of moral standards or not, like hooliganising
and things I mean). (See DL:DE for explanation of grammatical terms).
Studies of the kind undertaken by Biber et al, Carter and McCarthy and others show that
not only does speech has its own grammar which is distinct in many ways from that of
writing, but that it can also be described and categorized into grammatical categories.
2.8 Studying speech
The earlier sections in this chapter outline and introduce issues in studying pragmatics
and discourse. As they have shown, the focus of research in pragmatics and discourse is
upon: a) identifying linguistic patterns in verbal behaviour; b) the management of
conversation and c) the nature of interaction as it occurs in everyday speech, taking
account of the contexts within which it occurs. Conversation analysis, and indeed
pragmatics in general, studies the order, organization and orderliness of social action,
and particularly as it is located in everyday interaction. Above all, it is concerned with
structures and ways in which order happens. The aim of research in pragmatics and
discourse then, is to provide analytic descriptions of the organization of spoken
interaction, to apply them to everyday speech and to analyse its implications on the
linguistic system itself.
2 .8.1 Selecting a topic for research
The first thing to decide when undertaking your own research into pragmatics and
discourse is the focus of your study. For example, you may wish to:
> replicate previous studies by taking new data and performing the same analysis;
> compare talk in particular settings, such as an institutional one and/or an informal
one;
> compare different kinds of talk, such as between men and women, among women or
men of a similar age or of different ages;
> describe something new, such as interactive or live TV or internet chat rooms;
> compare talk amongst people of different cultural backgrounds.
Replicating a previous study is perhaps the simplest kind of research project. It involves
you asking the same research question or questions and using the same methods as the
study you are going to replicate, but you collect your own data to analyse, to find out if
your analysis confirms the same results as the original study. Replication of study is in
fact an important aspect of research in pragmatics and discourse, since by nature this
kind of study tends to be qualitative and small scale with a large number of variables that
make it difficult to draw conclusions from any one study. Reliable generalizations about
the point at issue can only be made when similar results have been obtained across a
range of studies. Even if you replicate a study, there will still be differences between
your own and the original study. The informants (the term used to describe participants
you record, as in Chapter 1) in your study will almost certainly be different people from
those in the original study, even though they may share similar characteristics of age,
gender, ethnicity, class and so on.
One way in which you can usefully or consciously vary replication of a study is to ask the
same questions and use the same methods, but change one or more of the nonlinguistic
variables. For example, the women who took part in Coatess 1996 study of women
friends talk were predominantly middle-class. This study could be replicated using
working class women as informants, and the results of the two studies compared. The
question here would be to see whether patterns of friendly talk idenitifed by Coates apply
to women from a different social background.
Comparing talk in different settings is another common type of research project. Whereas
projects that replicate earlier work share largely the same research focus as the original
study, for this kind of study, you would need to be clear about this for yourself and decide
(in advance) on your own focus or research question. For example, in addition to those
given above you could:
- compare soap opera dialogue with spontaneous informal conversation, to see how far
soap opera dialogue resembles spontaneous speech;
- compare the speech of one person in their work setting as opposed to a more informal
and casual one, to see if the context of speech makes a difference.
Comparing different types of talk is also another common type of research project. As
with investigating talk in different settings, you would need to be clear what the focus or
research question you are investigating is going to be. For example, in addition to those
given above the you could:
- compare a group of elderly men or women with a group of young men or women from
similar social backgrounds to see if patterns of conversation vary with age;
- compare a mother talking with a daughter or son, or a father talking with a daughter or
son, to see if patterns of conversation vary according to gender. .
Describing something new is also a possible area for research. Technological advances
in TV broadcasting and the internet have both produced new interactive media genres,
and it would be interesting to see whether such genres observe patterns of use
previously identified or give rise to new ones.
Once you have chosen the focus of your research project, then it is important that you
read any relevant works which may help you to orientate your study within the body of
published research undertaken, in both books and journal articles. By reading around
you form a better picture of the field of enquiry in which you wish to locate your project,
such as what the arguments are about and what the questions are. Knowing this will help
you to formulate your own question more clearly and give you a picture of what findings
you might expect, and what would be unexpected. Another important aspect of reading
around your topic is that it gives you ideas and information regarding data collection.
Most studies will usually have a methods section where the author details the
procedures used in the study, such as how many informants, how long the recording was
for, and maybe even pitfalls to avoid!
It is important to take notice of the date of publication of any work, since that which is
most recent will be most up to date in developments in the field, and most likely refer
back to older references still being cited by researchers, which gives an idea of their
importance. The key to reading for the purposes of writing a research project, or indeed
any essay, is not the amount of reading that you do, but developing the ability to perceive
which reading is important to your own research question.
2.8.2 Choosing and collecting data
As with researching variation and change in English, the data you collect will usually be that of
naturally occurring speech. Research in pragmatics, as has been mentioned previously, is
qualitative in nature; that is, the methods it uses involve more time collecting and analysing data
from a small range of informants. This allows for a more detailed analysis than quantitative
methods using more standardized instruments such as questionnaires allow (see also 1.4.2).
In researching variation in spoken English, Chapter 1 described how interviews form a main
source of spoken data collection. However, for the purposes of research into discourse and
pragmatics, the aim of collecting data is different, in that it is to find out how some aspect(s) of
talk itself works in different contexts and settings. Interviews of any kind are generally structured
around interviewers asking questions on a predetermined topic with informants responding.
Consequently, the researcher ends up with a certain kind of talk to analyse, which may be suitable
for the purposes of analysing variation, but not for analysing naturally occurring speech. The
researcher is more likely to obtain data for the purposes of pragmatics analysis by taking the role
of an observer or eavesdropper rather than that of the interviewer. This sounds simple, since
people are talking all around us all the time; however, to capture that speech requires the use of a
recording device, and many of the circumstances in which we find ourselves day to day do not
lend themselves easily to recording. Since all talk is shaped by the context in which it occurs, then
where it is being observed and recorded is an important part of that context. Opting for a more
recording-friendly physical setting almost inevitably loses or contaminates the influence of more
common or usual contexts.
Recording natural conversation, particularly in domestic and social settings, is a different kind of
activity from recording interviews. Rather than preparing an interview, you could ask your
informants to discuss a particular topic, such as their views on something which is topical in the
news, or ask to record when a meal is taking place or a social event of some kind. With this latter
kind of activity, you need to be aware of noise levels and make sure that the event does not
involve loud background music. Recording talk in a more institutional setting can be easier from
the point of view of people talking, as speech is usually structured around a particular activity,
such as, in a school a lesson, in a university, a seminar or tutorial, in a doctors surgery the
greetings of patients at Reception and the doctor-patient interview or in a sales situation between a
salesperson and the potential buyer.
Although the purposes and contexts of recording data for researching pragmatics and discourse
are different from those for researching variation, the actual mechanics of the recording situation
are very similar, and discussed in 1.4.2. Briefly, these are: being aware of the observers paradox,
obtaining consent and the recording itself.
The observers paradox refers to how a researchers presence may affect other peoples behaviour,
which is a problem, since ideally, we would wish to observe how people behave when they are not
observed. One obvious way to get around this would be not to let people know that they are being
recorded, or let them know afterwards, but this raises ethical issues of truth and honesty. Coates in
her book Women Talk (1996) describes how she secretly recorded her friends talking at gatherings
at each others houses for over a year. Eventually, she told them what she had been doing, and to
her surprise they felt angry, exploited and betrayed, not because she had recorded them, but
because she had not told them. We may think recording people closest to us would be a relatively
easy thing to do, but the reactions you encounter to such a request may be surprising. Informing
people after you have recorded them may be an option, but one you need to think about carefully,
and to be aware of any potential consequences such as Coates experience should you choose to
do this.
The current ethical guidance is that it is not acceptable to fail to ask permission to record peoples
voices, but this raises the issue of exactly how much to tell them about the purpose of your
recording. As 1.4.2 pointed out, if you tell people that you wish to record their accent, then this
has the effect of making people more conscious and aware of it and results in them altering their
normal speech behaviour. One way to avoid this is to offer a secondary reason for recording, such
as gathering opinion on a particular issue or as part of a project into a topic of some kind.
However you decide to obtain consent, a straightforward issue is that of making them aware about
the overall purpose of your research. Not everyone is familiar with the conventions of academic
research and thus need to be made aware of the implications of consenting to be recorded. The
most important implication here is when or if you are recording private conversations. By
analysing the data and writing it up as an academic piece of work, you are effectively making that
conversation available in a much more public arena. Professional researchers aim to get their
studies published in books or journals which reach a wide audience, and your own work, although
not so publicly available, will still be read by tutors and external examiners. For this reason, it is
an academic convention that pseudonyms are used in the transcript and analytic comment. Your
informants might also feel more comfortable in taking part in recording if they know that this will
be the case. It is also worth noting that professionals are often interested in seeing results, and the
offer to give them a copy of the recording - or indeed the written up research project - can often
help in obtaining consent.
When it comes to recording itself, then the less obtrusive video, tape, or micro recorder you use
the better. Section 1.4.3 explains the mechanics of recording in more detail. One thing you will
also have to bear in mind is that conversation, and the circumstances in which it takes place, can
be unpredictable. You need to be prepared to record more than once and for greater lengths of
time than you originally think you need to. If you are one of the participants in the recording,
then you will have to take account of the observers paradox.
Once you have recorded your conversation, then you will need to select the part which you will
use in your analysis. This will usually be a stretch of dialogue lasting anything from about two to
five minutes, giving two to five pages of transcription. Depending upon the length of the
conversation, it is not usually necessary to transcribe all of it, since this is very time-consuming
and not necessary for the purposes of your analysis. The selections you make will be guided by
the research focus of your study. For example, if you wish to investigate FTAs in relation to
conversation between a male and a female, then it makes sense to choose extracts which feature
FTAs. Once you have made your selections, then you need to transcribe them.
2.8.3 Conventions of transcription
Because analysing discourse is concerned with the management and order of conversation and the
contexts within which it occurs, then transcribing speech uses ordinary orthographic script rather
than the phonetic fonts of the IPA (see 1.5.4 and DL:DE chapter 3). It is important that you
transcribe what you hear, including utterances not normally found in writing such as um, er, and
so on, and resist the temptation to modify what you write into written, grammatical and
punctuated English. It also helps if you line number your transcription down the right hand side,
for ease of reference. There is no one single set of transcription symbols which are used
uniformly, but a consensus of use has emerged. It is usual to set out the transcript like a play
script; that is, with the speakers name or initial in the margin and then the utterance after it. It is
also usual to leave out punctuation and use the method of indicating pauses described below.
Capital letters are normally either not used at all, or confined to proper nouns. They are not
usually used to mark the beginning of sentences, since spoken conversation is not subject to the
same rules and conventions of written language. Punctuation marks generally are used for
intonation rather than for grammar. Analysing discourse is a relatively new activity, and there is
no standard system for transcribing talk of the kind there is for transcribing accent such as the
IPA. Even so, certain conventions of transcription have become established. The list below gives
the most commonly used transcription notation:
Transcription notation.
(.) just noticeable pause
(0.3) (2.3) examples of exactly timed pauses, in seconds
.hh hh speakers in-breath and out-breath respectively
wo(h)rd laughter within a word
end. full stop (period) denotes falling, ending intonation
word? question mark depicts rising, questioning intonation
word pound signs enclose talk said in a smile voice
cu- a sharp cut-off of a prior word or sound
lo:ng stretching of a preceding sound
(word) transcribers guess at an unclear part of the tape
( ) unclear talk.
(- - - - ) unclear talk with each syllable represented as a dash
A: word= equal sign shows no discernible pause between two speakers
B: =word turns or, if between two sounds within a single speakers turn, shows that they run
together
word underlined sounds are louder
WORD capitals are louder still
>word word < inward arrows indicate faster speech, outward arrows slower
<word word>
[overlap]
Transcribing accurately and faithfully is a time-consuming and lengthy business, since you will be
concerned not only with the words that are spoken, but also with representing any speech markers
which occur such as oh, um, er; identifying the length of pauses; identifying amounts of
overlap and so on. Anyone coming new to transcription invariably underestimates the amount of
time an accurate transcription takes. Even if you have undertaken transcription activities
associated with sociolinguistics of the kind described in Chapter 1, the kind of transcription you
do for analysis here will be very different, since it seeks to capture all aspects of conversation as it
occurs in real time, rather than with the identification of whole, identifiable words.
One way of preparing you to realise what exactly is involved in transcription is to undertake a
small-scale activity with another person. Record between 10 and 15 minutes of conversation,
where you and your partner may or may not be one of the participants, in an informal setting such
as at home, or the communal space of a shared house or flat, (making sure you have permission to
record before you start). Once you have recorded, you and your partner need to listen to the tape
right through, select and agree the sections you are going to transcribe, which amount to between
2-3 minutes in total. This should give you between 3-5 pages of transcription which should be
sufficient for the purposes of future analysis. Agree the transcription conventions you are going to
use, and any additional ones that may be needed, then, independently write out your
transcriptions. Once you have made your transcript, compare it with that of your partner. Points of
comparison would be: whether or not either of you have included anything apart from words;
whether or not your transcripts are identical, and if not, where they are different. It is also worth
taking some time to think about features of talk that were hard to capture and why.
2.8.4 Analysing data
If you have prepared yourself well when deciding your research topic in reading the background
studies that pertain to your work, and identified a clear focus for what it is you wish to interrogate
your data for, then the task of analysing your data will be straightforward. For example, if you
have chosen to replicate a study, or changed a variant in the study, then the categories of analysis
will be the same as those used in the original piece of research. If you are designing your study
around any other of the topics of the kind listed in 2.7.1, you may decide to concentrate your
analysis upon the categories of the kind described in this chapter, such as Gricean pragmatics
outlined in 2.3; discourse analysis in 2.4; conversation analysis in 2.5 or interactional
sociolinguistics in 2.6. The preparatory reading you will have done will help you decide, since
reading previous studies on your chosen topic should give you an idea of which methods of
analysis best suit the transcription data you have collected and transcribed.
Students can be worried by the fact that the analysis they present with their transcription can be
the least time-consuming aspect of the project, whilst collecting, choosing and transcribing data is
very time consuming for an end result of three to four pages of transcription. The fact of the
matter is, that you cannot do your analysis without the transcription, and without it there would be
no project.
The aim of your analysis will usually be:
> to discover how far the model or categories of analysis you are applying can be identified
in your data;
> to what degree this has been successful;
> how far your own findings accord with published studies in the field.
Once you have analysed your results, then you will need to interpret them. That is, you will need
to consider your results in the light of your analysis and draw conclusions about any insights
your analysis and comparison of it with other published research have given you.
2.8.5 Sample projects
a) Choose a study to replicate such as Cameron (1997) or Coates 1997 discussed in 2.6
above.
This will involve you firstly: in recording and transcribing parts of conversation between
participants of a similar age, gender and background to the ones used in the original
study. Your own work may not involve such extensive recording as the original study, but
should be sufficiently long enough for comparison to take place. As a rough guide, half
an hour of conversation with transcription of sections totaling about five minutes should
be sufficient. Secondly, you will need to analyse your transcription according to the
categories used in the original study and thirdly, compare your analysis with that of the
original which involves interpreting it, drawing conclusions from it which may or may not
confirm those of the original study.
b) Compare talk in a particular setting, such as an institutional one and/or an informal
one, and analyse it according to either DA or CA categories of description.
This will involve you firstly: in recording conversation which takes part in your chosen
setting, probably about half an hour, and transcribing parts of it which total about five
minutes. Secondly, choose either DA or CA categories of analysis and apply them to
your transcription. Thirdly, discuss what your analysis reveals, and draw conclusions
from it, referring your conclusions to research already undertaken in this area. Since the
focus of this study is upon the setting, then your discussion will centre upon how aspects
of that setting affect verbal behaviour. You should also include an explanation as to why
you have chosen either DA or CA.
c) Compare and analyse different kinds of talk, such as; i) between men and women, ii)
among women or iii) men of a similar age or of different ages.
This study is very similar to b) above, and will follow the same process, except that the
focus will be more on the participants verbal behaviour and less on the effect of the
setting.
d) Describe interaction in live TV or internet chat rooms.
This study will involve you in either: i. recording data from TV or ii. printing off material
from chat rooms. Recording data from TV does not take as long to set up or to execute
as live recording, so you could record one or two programmes, depending upon their
length. Chatroom material needs to be long enough to select extracts, but not so long as
to make the data unwieldy. For a short study, 15-20 pages should be sufficient. You then
need to select extracts which you can analyse in detail, according to the methods and
categories of analysis you have chosen. Finally, you need to consider your data in terms
of what it reveals about the interaction and whether or not any conclusions can be drawn
as to how the context affects interaction.
e) Undertake an analysis of spoken grammar.
This study will involve you in analysng and transcribing conversation as for a), b) and c)
above. Instead of applying categories associated with either DA or CA, you will need to
analyse your data according to grammatical categories of spoken English. This will
involve you in explaining what these categories are, where they are found in your data
and how they affect the process of interaction.
2.9 Where to find out more
For other works on pragmatics and discourse, see Brown and Levinson (1987), Hewings
and Hewings (2005); Cutting (2002); Cameron (2001); Hutchby and Woofitt (1998);
Thomas (1995), Schiffrin (1994) Leech (1983) and Watts (2003). Chapters on pragmatics
and discourse also feature in general books on Applied Linguistics, such as that edited
by Davies and Elder (2004) and Schmidt (2002).
Chapter 3 Stylistics
3.1 Introduction
The first two chapters of this book have been concerned with spoken language. By
contrast, stylistics is concerned primarily with the linguistic analysis of written texts,
especially, but not exclusively, literary ones. In its earlier form, linguistics had been
concerned with spoken language, its history, acquisition and so on. Where it had
concerned itself with the written form of English, this was often through a consideration of
its formal features as expressed through vocabulary and grammar and not with literary
features of style. In a seminal and often quoted paper, Roman Jakobson (1960:377)
wrote that: a linguist deaf to the poetic function of language and a literary scholar
indifferent to linguistic problems and unconversant with linguistic methods are equally
flagrant anachronisms. That is, just as a linguist should take account of the creativity
and literariness of language, then a literary scholar should take account of the linguistic
structures through which that creativity and literariness is realised.
At the time, Jakobsons statement provided a landmark for the field of stylistics. Since
that time, many specifically linguistic investigations into literary texts have been
undertaken, often using categories of descriptive linguistics as their structuring
mechanism. In addition, aspects of phonology, especially metrics and sound patterning,
have been applied to poetic texts. Linguistic analysis of literature, however, involves not
only the identification of linguistic features, but also a consideration of how these are
interpreted, and the relationship between linguistic structures and their mental
processing. Consequently, in addition to more formal properties of style, there has also
been a parallel interest in the psychological and cognitive properties of style, particularly
in how readers respond to texts.
Many books or chapters on stylistics are organised around investigating the language of
particular textual genres, and especially the language of poetry, prose and plays. This
chapter takes a different approach, in that it is organised around key concepts or
linguistic categories in stylistics and organising principles of language as discourse. This
allows for consideration of stretches of text at the level of discourse and discussion of
stylistic tools across a variety of different text types and genres. 3.2 and 3.3 provide a
background and overview to the study of stylistics. 3.4, 3.5 and 3.6 discuss some major
aspects of textual organisation and patterning across stretches of text. 3.7 explores a
major organising principle of textual organisation, that of narrative, whilst 3.8 considers
representations of speech and thought in narrative and 3.9 the use of dialogue in drama.
These sections are by no means exhaustive of all methods and approaches to stylistics,
but cover a few in sufficient detail to give you a flavour of what is involved and to get you
started in your own analysis. 3.10 provides students with guidance and advice on
undertaking their own research into stylistics, whilst 3.11 provides suggestions for further
reading.
3.2 What is stylistics?
At its broadest, stylistics can be defined as the study of style of written -- predominantly
literary -- texts, as opposed to the investigation of the ideas that it may contain, which is
the focus of literary criticism. This is not to say that stylistics ignores ideas. Stylistics
looks mostly into HOW texts mean, whereas literary criticism focuses on WHAT they
convey. As Wales (2001:373) says: The goal of most stylistics is not simply to describe
the formal features of texts for their own sake, but in order to show their functional
significance for the interpretation of text; or in order to relate literary effects to linguistic
causes where these are felt to be relevant...
Although intuition and interpretative skills are important, stylistic analysis aims to avoid
vague and impressionistic judgements about the way formal features are used and
manipulated in texts. For their part, some literary critics take issue with what they think of
as an objective approach to the interpretation of literary texts. However, this objectivity
is an important feature of stylistics, since its methods of analysis aim to be, in Wales
words: methodical, systematic, empirical, analytical, coherent, accessible, retrievable
and consensual. Thus, stylistics is concerned both with identifying and examining the
linguistic organisation of a text, how that organisation interacts with its interpretation to
support a particular view or reading of the work under examination.
In addition to linguistics and literary criticism, in recent years stylistics has also been influenced
by cultural theory, which questions the assumption that interpretation or a particular view of a text
is fixed, stable, and the same for all readers. From the 1980s onwards, the work of linguists such
as Fowler (1986) and more recently Fairclough (1989 onwards) and van Dijk (1988) have added a
further dimension of interpretation, which is that interpretation does not happen in a vacuum or as
a solely individual activity but interacts with social and cultural experiences. Consequently,
stylistics can be said to be made up of three distinct but interrelated strands of inquiry, any of
which can independently form the primary focus of study, or lend themselves to viable
combination with either or both alternatives. These strands are: the formal, the cognitive and the
sociocultural.
A formal approach to the study of stylistics is concerned with the formal properties of a
text. It takes on board the analysis of words and other structures from a linguistic
perspective. This includes analysis of sounds, vocabulary, syntax of phrases and
clauses, and the textual organization of discourse. Within stylistics, analysing the formal
features of a text has the most developed conceptual vocabulary and frames of
reference. In the stylistics classroom, a common language or metalanguage has
evolved: that is, a set of linguistically derived terms for describing stylistic concepts.
These terms centre around the metaphorical concept of the stylisticians toolbox, and
includes the use of checklists of the kind offered by writers of textbooks in stylistics such
as Short (1988; 1996), McRae (1997), some of which are discussed more fully in 3.3
and 3.4 below.
A cognitive approach takes account of the points of contact between a text, other texts
and readers, including consideration of the function of a text, drawing upon literary,
cultural theory and psychological theory in addition to linguistic theory. In recent years,
stylistics has begun to draw upon work in cognitive psychology in attempting to
understand the point of contact between a text, other texts and readers. Thanks to
research in the field of pragmatics described in chapter 2, linguists have come to realise
that meaning is not stable and absolute, but often depends as much upon the processes
of interpretation undertaken by a reader or listener, as upon the actual linguistic
structures that are used. Stylistics can also consider the point of contact between the text
and the reader as an interactive, communicative act. It includes considerations such as
the ways in which writers draw attention to other texts, and studies how readers track
texts during the act of processing, discussed in 3.4.1 and 3.5. Consequently, those
working in the field of stylistics are increasingly coming to recognise the interactive
nature of roles played by the reader and the text in the activity of analysis and the
construction of an interpretation. The aim for stylisticians then, as well as for literary
critics, is to understand the reception of a text.
A sociocultural approach is concerned not only with the interaction between the text
and its readers, but also the socio-cultural contexts within which reading and writing take
place. Texts and their readers do not exist in isolation, but function within a wider social
and cultural context. This has led to account being taken of contextual factors such as
the cultural background of the reader, the circumstances in which the particular text is
read, and so on. Rather than concerning themselves exclusively with finding out what a
text means, stylisticians undertaking such an approach to analysis are more interested in
the systematic ways language is used to create texts which are similar or different from
one another, and link choices in texts to social and cultural contexts. The particular
concerns, experiences and political views which the reader brings to bear on the text will
obviously play a huge role in colouring her/his search for meaning in a text, and it is
essential that this influence is acknowledged when applying the objective criteria that are
deployed through the checklists of linguistic features contained within a text. Such a view
shifts the point of focus away from a static view of the text which exists in its own world
as a self-sufficient entity, towards one which is much more dynamic, cognitive,
intertextual and interpersonal. This third aspect of stylistic analysis can also include
taking account of unequal relationships in society and how texts mediate authority,
power and control. Viewed in this way, stylistic analysis can become embedded within a
framework of critical linguistics and critical discourse analysis (CDA), where explorations
of authority, power and inequality form a central motivation for the analysis. This
approach pays attention both to the formal features of the text and also to its reception
within a reading community in relation to ideology. This critical aspect or dimension of
textual analysis is discussed fully in chapter 4, whilst the current chapter concentrates its
discussion on the formal and cognitive approaches to textual analysis and interpretation.
A further issue in stylistics has been the range of texts which should be considered for
analysis. Stylistics was originally intended as a method of analysis for literary texts, but
as the whole notion of literary language, texts and textuality has become open to
question and become more fluid, then the range of texts open to stylistic analysis has
also broadened. Consequently, analysis extended beyond exclusively literary texts to
include, for example, newspaper articles, advertisements, and so on. Rather than
focusing on and identifying instances or features of literary language, it is more
appropriate to consider the concept of literariness , first coined by the Russian
Formalists in the early twentieth century. Literariness is a property of texts and contexts
which focuses upon and considers patterns of language in use, irrespective of their
acknowledged genres The examples used in this chapter then, come from a wide range
of texts, rather than being confined to exclusively literary ones.
3.3 Stylistics, levels of language and grammar
Any utterance or written piece of language is organised on several distinct levels of
language, each of which has an associated area of linguistic study. (see DL:DE Chapter
1 section 1.1 and chapters 4 and 5). Traditionally, the largest unit of grammatical analysis
has been that of the sentence. Texts are essentially made up of grammatical units of
analysis are ordered in a rank scale according to their size, from the largest down to the
smallest:
Sentence or clause complex
Clause
Phrase or group
Word
Morpheme
Phoneme
These levels of language from the phoneme through to the clause can be identified and
discussed individually in the stylistic analysis of any text. However, it is central to
stylistics and to our understanding of language and style that these levels are
interconnected. They depend upon each other and work together, representing multiple
and simultaneous linguistic operations in the planning and production of an utterance
(Simpson 2004:5). Stylistic analysis of texts then, at this formal level, considers each of
these levels and their interdependence upon one another, an interdependence which is
often most vividly apparent in poetry. Many textbooks on stylistics explain this
interdependence in detail (see, for example: Wright & Hope 1996; Short 1996).
Phonemes, the smallest meaning-changing units of sound (see: DL:DE 3.1), are often
exploited in poetry and advertising slogans, especially for their alliteration,
repetition, assonance and rhyme. Take the following example, from Old English
poetry:
Over breaking billows, with belly sail,
And foamy beak, like a flying bird
The ship sped on.
B is used alliteratively throughout the poem: breaking billows, belly, beak, bird; as is s in
sail and sped and f in foamy and flying. Alliterative syllables are usually strongly stressed
ones, so that they relate to the rhythmic pattern of the rhyme as in the following example,
taken from the medieval poem Sir Gawain and the Green Knight:
The snaw snitered ful snart, that snayped the wilde
(The snow came shivering down very bitterly, so that it nipped the wild animals)
Assonance is where a vowel sound is repeated in a word with different final consonants.
e.g. Break! Break! Break! On the cold grey stones, O Sea! For example, repetition of o in
cold and stones again as the single word O, stresses the poets anguish. Another
example of assonance is the advertising slogan: beans means Heinz. This repeats
the vowel sounds /ea/ which is then echoed in the assonance of /ei/.
The morpheme is the smallest unit in grammar often also said to be the smallest unit of
meaning in written language (see: DL:DE 4.0 and 4.1), whereas the phoneme is
commonly described as the smallest unit of meaning-changing sound. Certain
morphemes, called root morphemes, are individual words in their own right, whereas
others, such as prefixes and suffixes known as bound morphemes since they depend on
their meaning by being joined onto bound morphemes. For example, the word
teapots has three constituents: two root morphemes tea and pot plus the suffix or plural
morpheme s. One way in which morphemes can add to vocabulary is by creating newly
invented words. A common way of doing this is through compounds: that is, joining two
root morphemes together, such as teapot or mobile phone. Another way is to invent
completely new words, of which a famous examples of this is Lewis Carolls poem
Jabberwocky from the novel Through the Looking Glass:
Twas brillig and the slithy tothes,
Did gyre and gimble in the wabe;
All mimsy were the borogroves,
And the mome raths outgrabe.
"Beware the Jabberwock, my son,
The jaws that bite and claws that scratch
Beware the jubjub bird
And shun the frumious bandersnatch"
Here, English speakers can readily interpret invented words such as brillig gyre and
gimble as root morphemes, whilst tothes, borogroves and mome raths are shown by
their syntactic position to be bound by the plural s, and slth-y has two morphemes,
the final y being an adjectival one.
Morphemes combine into words which combine into phrases. These can be one word on
its own or a group of words, such as a noun phrase, verb phrase, adjective phrase
and adverb phrase. To this can be added the category of prepositional phrase which
is a noun phrase with a preposition in front. (see: DL:DE 5.3 ). For example, The cat is a
noun phrase made up of the determiner the and the noun cat. Jumped is a verb phrase
comprising the past partciple of the verb jump, whilst has jumped is a verb phrase made
up of two words, the auxiliary has and the verb jumped. Over the fence is a prepositional
phrase, formed by the preposition over plus a noun phrase the fence, itself made up of a
determiner and noun. Phrases combine elements of information into clauses. The main
difference between a phrase and a clause is that a clause contains a verb phrase in
addition to other phrases (see: DL:DE 5.3 for a more detailed explanation).
For the purposes of stylistic analysis, the clause is particularly important because it has
to include a verb, thus fulfilling several important functions of language such as
providing information on tense and grammatical mood: that is, whether a clause is
interrogative, imperative or declarative. In addition to the grammatical forms
summarised above, clauses can also be categorised according to their semantic
functions. The various phrase classes outlined above combine into five basic elements
of clause structure. These are:
subject (S),
predicator or verb (P),
object (O),
complement (C)
adverbial (A)
(see also DL:DE Chapter 5, , especially 5.4 for a fuller explanation).
In English, it is normal for the theme of the sentence, and the process, namely the
subject and the verb, to appear at the start of a sentence. Take, for example, this
sentence from F. Scott Fitzgeralds novel The Great Gatsby 1926&2000:8):
My family have been prominent, well-to-do people in this Middle-Western city ofr
three generations.
A SPOCA analysis of this sentence reveals the following:
Subject Predicator
My family have been
Complement
prominent, well-to-do people
Adverbial
in this Middle-Western city
Adverbial
for three generations
This sentence is comprised of five phrases:
my family (noun phrase);
have been (verb phrase);
prominent, well-to-do people (adjective phrase)
in this Middle-Western city (prepositional phrase).
for three generations (prepositional phrase).
The noun phrase my family functions as the subject of the sentence, whilst the adjective
phrase prominent, well-to-do people functions as a complement after the verb phrase
have been, which functions as a predicator. The predicator is the only place in a clause
where form and function have a one-to-one correspondence. Complements occur after
certain intensive verbs, of which be is the most common.
Not all clauses or sentences, however, fit such a standard pattern, particularly in poetry,
and this may have the effect of interrupting our expectations or bringing to the fore, and
therefore emphasising, information which would usually come later or, delaying
information we would normally expect at the beginning to the end. Another important
aspect of an analysis of this kind is the number of clauses that a sentence contains, and
whether they are independent, co-ordinate or dominant and dependent. In the example
above, there is one independent clause, but in the following extract from Wordsworths
poem Michael (1800, lines 10-114), we see the clause and sentence structures being
manipulated to serve the meaning of the poem:
Down from the ceiling, by the chimneys edge,
That in our ancient uncouth country style
With huge and black projection overbrowed
Large space beneath, as duly as the light
Of day grew dim the Housewife hung a lamp.
This sentence has two clauses: as duly as the light of day grew dim and the Housewife
hung a lamp. Both of these are preceded by a series of prepositional phrases which
function as adverbials. The main clause is the last one, the Housewife hung a lamp.
The subject of this clause is the housewife, who is also the main subject of the sentence.
Placing the main clause of a sentence at the end in this way delays the delivery of
information vital to the understanding of the sentence, making the reader aware of the
detail which comes before it in ways which we might not do if it came after the main
clause. The adverbials of the first part of the sentence work as a kind of camera, a filmic
panning which introduces the scene, moving from the wider perspective of the gloomy,
rustic country room to the housewife, who illuminates that gloom with a lamp. We may
also feel some sense of frustration, awaiting the main verb, which parallels the waiting of
those in the cottage as they anticipate the light illuminating their room (see; Jeffries
1993:68).
The examples above show how undertaking a clausal analysis of texts of this kind can
tell us a great deal not only about how the information given in them is ordered, but also
about the effect of such ordering. Traditionally, syntactic or grammatical analysis has
stopped with the clause as the largest element. However, patterns of language can be
identified across stretches of text that have extended the unit of analysis beyond that of
the clause or sentence. Any piece of writing, if it is to make sense at all, makes
connections within itself through the kinds of vocabulary and the syntactic structures it
uses to bond or connect its sentences together. Just as a random selection of words
does not in itself make a sentence, so too a random selection of sentences does not of
itself create a coherent text. Modern stylistics recognises that texts work as discourse,
that is, discourse in the linguistic sense of a stretch of language above the level of the
sentence.
3.4 Stylistic methods and categories of analysis
Analysing texts beyond the level of the sentence has been made possible by
developments in systemic-functional grammar. Within this grammar, patterns of syntactic
structure have been identified which go beyond the clause or sentence across sentences
in stretches of text and beyond the text, making direct links with the situational context of
the speaker or writer. Two of the most developed descriptions of such structures are
those concerned with cohesion and deixis (see also: DL:DE, chapter 7, especially 7.1 &
7.3 and, for deixis TL:TE Part II 5.2) Cohesion refers to the ways in which syntactic,
lexical and phonological features connect within and across sentences in a text.
Deixis refers to the ways in which some words and phrases shift their reference,
depending upon the speakers position in space and time and who says or writes them.
Cohesion thus performs a textual function, making explicit connections between the
sentences of a text, whilst deixis, though textually based, makes direct links with the
situational context of the speaker or writer. Cohesion and deixis then, exemplify ways
in which the text and the reader interact. Throughout 3.4 and 3.5, then, attention is
therefore paid to both formal and cognitive aspects of textual features.
3.4.1 Cohesion
In addition to subject matter, vocabulary and the syntactic structure of individual
sentences, a writer will normally use signals to make connections in and between
sentences. These include the ways in which sentences are sequenced, how one thing
leads to another, how certain feelings or events are implied and so on. These signals act
as markers of cohesion or cohesive ties in a text: that is, writing is held together not
only because of relationships between the ideas or action represented through lexis,
semantics or syntactic structure, but also through connecting forms in the lexis and
syntactic structure within and across the sentences themselves. Jeffries (2006) divides
cohesion into six different kinds: reference, substitution, ellipsis, conjunction and
lexical cohesion (see DL:DE Chapter 7 for further explanation of these terms). A
category also relevant to writing such as poetry and adverts is that of phonological
cohesion.
Take the following example:
When I first entered heaven, I thought everyone saw what I saw. That in everyones
heaven there were soccer goalposts in the distance and lumbering women throwing shot
put and javelin.
Alice Sebold Lovely Bones (2002:3).
References can be divided into two main categories: exophoric reference and
endophoric reference. Exophoric reference refers to the immediate situational context
in which the discourse is taking place. Examples of exophoric reference in the example
above are heaven, everyone, soccer, goalposts, women (who were lumbering).
Endophoric references can be sub-divided into anaphoric and cataphoric reference.
Anaphoric reference refers backwards to somebody or something that has already
been mentioned, usually by the use of personal or possessive pronouns. Once a referent
has been established, that is, the person or thing being referred to, then it is usually
replaced by a pronoun the next time it appears, unless using a pronoun makes the sense
unclear and ambiguous. The most common use of anaphora is where the pronoun refers
to a noun already mentioned in the same or preceding sentence. Take the following
examples:
Connie always had a foreboding of the hopelessness of her affair with Mick, as
people called him(1). Yet other men meant nothing to her (2). She was attached to
Clifford(3) . He wanted a good deal of her life and she gave it to him(4).
Lady Chatterleys Lover [Link] (1928& 2006:132)
The two characters Connie and Mick are named in the first sentence. Thereafter,
anaphoric referencing of him in sentence 1 and he and him in sentence 4 refer back to
Mick, whilst that of her in sentence 2, she in sentence 3 and her in sentence 4 refer back
to Connie..
Anaphoric referencing can also have the effect of implying a previous existence for the
characters in a text. Its use plunges us immediately into the world of the text, assuming
that we are familiar with the person for whom a pronoun stands when this could not
possibly be the case unless we were re-reading the text. For example, the following
extract, taken form the beginning of the first chapter of William Goldings novel The
Inheritors (1955), uses anaphora in this way:
He was struggling in every direction, he was the centre of the writhing and kicking
knot of his own body
Since this is the beginning of the novel, we do not know who he refers to other than that
he is a male, but by using the anaphoric reference instead of a noun, we are immediately
drawn into the middle of a narrative, instead of being led gently into it, and expect to
learn who he is in good time.
Cataphoric referencing is the opposite to anaphoric referencing, in that it refers
forward. Saying that something appears below in a text, for example, direct the reader
to something that is to be encountered, such as in the next chapter, we will examine this
theory in more detail. Cataphoric references delay more precise information, and are
important in creating an element of suspense. In the following example, cataphoric
referencing gradually reveals information about a particular woman, before naming her:
And slowly down the steps in her magnificent ballgown comes the young woman
of the moment we have all been waiting for, Princess Diana herself.
(BBC News 6 June 1986)
In this example, a pronoun her is used first, followed by a noun phrase, the young
woman, followed by the proper noun Princess Diana which is then followed, for added
emphasis, by a reflexive pronoun herself. Anaphoric referencing of this kind is commonly
used to build up suspense in, for example, thrillers and detective fiction.
The following lines taken from the beginning of Mark Twains novel The Adventures o f
Tom Sawyer ( 1910 & 1986:1) show different kinds of referencing at work:
"Whats gone with that boy, I wonder? You Tom!"
The old lady pulled her spectacles down and looked over them about the room;
then she put them up and looked out under them. She looked perplexed a moment
and said, not fiercely, but still loud enough for the furniture to hear, " Well, if I get
hold of you, Ill...
Before we even know his name or what is being referred to, the cataphoric reference that
signals that a person or thing is about to be mentioned; this is followed immediately by
the noun boy, followed by the pronoun you. It is not until we get to the end of the second
utterance that we get the proper noun Tom. Just as in the Princess Diana example
above, the cataphoric references delay precise information about the boy that the woman
is wondering about. These words are spoken by a woman who is not given a name, but
the exophoric reference the old lady. Since this is the opening of the novel, there is
nothing we can refer back to, to establish who she actually is. Similarly, we have nothing
to refer back to the room. Which old lady? Which room? After the first clause mentions
The old lady, she is referenced anaphorically as her and she, and her spectacles
as them. The use of anaphoric and cataphoric referencing in this extract link the world of
the text to the shared reality of the reader, drawing upon our experiences of what we
already know about the world, in this old ladies, rooms and young boys. In this way,
coherence is established between what is being described and the expectations and
shared knowledge which we, as readers, bring to a text.
In addition to using pronouns to refer outwards, backwards and forwards to something or
someone, referencing can also include repeating the noun or noun phrase that has been
used instead of replacing it with a pronoun at the beginning of clauses, phrases or
sentences.
Repetition re-reinforces description and emotional effects in narrative; in public speeches
and adverts it can be used to hammer home a point or a product, as in the following
examples:
Example 1.
The rain fell heavily on the roof, and pattered to the ground....The rain fell,
heavily, drearily. It was a night of tears.
(Little Dorrit, Charles Dickens Ch 17)
Example 2:
We are fighting for the rights of the little man...We are fighting, as we have always fought,
for the weak as well as the strong. We are fighting for great and good causes...
(Mrs Thatcher. Guardian 13Oct 1984)
In the examples above, repetition of the rain fell heavily in example 1 reinforces the
weight of the rain and its metaphorical association with tears, whilst the repetition of we
are fighting in example 2 emphasises the point that the Conservative politician is on the
side of the underdog and meeting resistance from other political parties.
As mentioned in Chapter 2, a common feature of cohesion in spoken utterances is
ellipsis, the name given to leaving out part of an utterance or a grammatical structure
which the listener or reader is assumed to readily understand from its context. For
example, predicators and sometimes also the following clause elements which have
already been used are often repeated in condensed, substituted forms (generally as a
form of the verb do or by the word so), or are left out altogether.
For example:
All trains go to the station. At least, most do.
The second sentence leaves out the unstated, obvious predicator go and adverbial the
station of the first sentence to avoid unecessary repetition.
He didnt fail. He might have done, if hed gone later.
This sentence leaves the main verb (failed) out of the verb phrase in the second
sentence, again to avoid repetition.
Because it leaves out information already given, ellipsis can help the listener or reader to
focus on new or important information. It is often used where economy of words is
needed, such as note-taking and personal newspaper adverts ( Wanted: Mothers Help.
Three children, 2, 4 and 7. Must drive) Omitting grammatical words such as determiners
and auxiliaries is also common in representing interior monologue in narratives,
suggesting a quick succession of thoughts or images as the following extract from
James Joyces novel Ulysses (1922) illustrates:
raised his eyes and met the stare of a bilious clock. Two. Pub clock five minutes fast.
Time going on. Hands moving. Two. Not yet.
In this example, Joyce uses ellipsis in a telegraphic way, omitting all words except for
the essential two rather than It was two oclock or It said two oclock, and omitting the
and was from the next sentence. The extract uses no referencing beyond his. Instead,
cohesion is achieved by cryptically describing the thoughts of one person from which
we, as the reader, have to infer far more than we are accustomed to make sense of the
text.
Where ellipsis is used to avoid unnecessary and tedious repetition, its use is typically
anaphoric. It is extremely common in everyday conversation which is far more context-
dependent than writing. One of the most striking differences between natural,
spontaneous conversation and written dialogue which aims to represent it is the use of
ellipsis. Dramatic dialogue tends to be far less elliptical than natural conversation, since
it lacks the degree of context dependence of ordinary speech (see 3.9 below). In written
texts which have to be explicit, such as legal contracts and advertising, there is far more
repetition and explicit referencing and subsequently far less ellipsis than in other types of
text.
Ellipsis can also be used in narratives and plays not only as grammatical omission or
substitution but also to signify passing of time, speeding up the action or pace of a
narrative by leaving out events assumed to have happened but not described or enacted,
either by leaving them out altogether or explicitly marking that they have happened.
Substitution works in a similar way to ellipsis, only rather than working by omission, one
word, most commonly a pronoun, is substituted for another word, phrase or clause. Other
items commonly used for substitution are:
One: I offered her a seat. She didnt want one.
Do: Did Frank take that letter? He might have done.
So/not: Do you need a lift? If so, wait for me; if not, Ill see you there.
Same: He chose the roast duck, Ill have the same.
Like ellipsis, substitution assumes much that is left out within the context in which it is
used, and, also like ellipsis, is a much more common feature of speech than writing. In
writing, its use is mainly in the writing of dialogue which aims to represent spontaneous
speech although again, like ellipsis, its use depends upon the context having been made
explicit so that its use can be understood by a listener or viewer.
Conjunctions (words like and, because, but) work in a way which is different from
reference, ellipsis and substitution, in that they do not search backwards or forwards for
their referent; rather they signal a relation between segments of a clause or phrase.
Conjunctions join clauses within a sentence which can lead to ellipsis being possible in
co-ordinating sentences. For example, Mary walked to the car and got in (to the
car) They also indicate that what follows in a sentence bears some relation to what has
already been said as well as grammatically joining the sentences together.
Within a text, conjunctions signal different types of relation between sentences. In the
example above, and connects the two parts of the sentence,, and marks the way that the
second clause follows the first, and how thew text as a whole is moving forward. Other
conjunctions which also perform such a function are so, therefore or hence Jeffries in
DL:DE 7.1.4 .notes that the signposting conjunctions provide falls into four semantic sets:
Additive: adding more information as in the example above, or in the following one: .
Shes intelligent. And shes very reliable.
Adversative: qualifying the information already given. For example, Ive lived here ten
years but havent ever heard of that pub.
Causal: giving a cause as to why something happened. For example, He caught a
cold because he fell in the river.
Continuative: signal further comment, For example: Well, he could be right.
Take, for example, the following extract, from Thomas Harriss novel Hannibal
(2000:251):
Clarice Starling was the last to know that Dr Lecter had killed again. After she hung up
the phone, she lay still for many minutes in the dark and her eyes stung for some reason
she did not understand, but she did not cry. From her pillow looking up, she could see
his face swarming in the dark. It was lecters old face, of course.
The first conjunction after is a causal one, telling us that Clarice lay still once she had
finished the phone call. The second conjunction and is additive, giving us more
information, whilst the third, but qualifies this information by saying that she does not give
into the temptation of crying. The fourth conjunction, of course, is continuative, adding
more comment and information on that given in the preceding sentence. Conjunctions
then, play a vital role in cohesion, logically connecting between parts of the text.
Lexical cohesion, as the term implies, describes ways in which items of vocabulary
relate to one another across clause and sentence boundaries to make a text coherent. It
refers to the part played by certain semantic relations between words to create textuality;
in other words, that which distinguishes sentences as a text as opposed to a random
sequence of unconnected sentences. The relations between vocabulary items in texts
are of two main kinds: reiteration and collocation.
Reiteration means either repeating the same word in a later section of the discourse or
else reasserting its meaning by exploiting lexical relations: that is, the stable, semantic
relationships that exist between words and which form the basis for descriptions or
definitions in a dictionary or group of words in a thesaurus (see DL:DE section 6.3).
These are of two main kinds: synonymy and hypomony. For example, zucchini and
courgette; bachelor and unmarried man are related by synonomy: they both refer to
exactly the same thing or state. Courgette and vegetable; mosquito and aeroplane are
both related by hyponymy: one is a superordinate in the family tree of another.
Consider this example of synonomy:
The meeting commenced at six thirty. But from the moment it began, it was clear all
was not well.
In this example, commence and began co-refer to the same thing in the real world, the
meeting. But this is not always the case:
The meeting commenced at six thirty; the storm began at eight.
In this second example, commence and began refer to separate events, although the
semantic relation of synonymy between them is being exploited stylistically here to create
humour or irony.
Now consider this example of hyponymy:
There was a fine old rocking chair that his father used to sit in, a desk where he wrote
his letters, a nest of small tables and a dark, imposing bookcase. Now all this furniture
was to be sold, and with it his own past.
Repeating ideas by direct lexical repetition is something we do not commonly do or
encounter, other than for a particular effect such as driving a point home. Instead, we
tend to vary items, in this case hyponyms of furniture, giving variation from sentence to
sentence which taken together build a mental picture of the kinds of furniture being
described. Such variation can add new dimensions and nuances to meaning, building up
an increasingly complex context. Every paraphrase of an earlier word brings with it its
own semantic connotations. The following newspaper report uses several types of lexical
cohesion:
Police toughened up their anti-protest tactics at Brightlingsea yesterday and
arrested 12 demonstrators who had defied their warnings against trying to block the
continuing export of live animals from the port in Essex.
Around 400 demonstrators, a lower turnout than the 1,000 anticipated by the
organisers, failed in their attempt to turn back three lorries containing around 1,500
sheep. Following a clear police warning that arrests would be made if the paths of
vehicles were blocked, the lorries containing the sheep began the final 400 yards
of their journey to the quayside at Brightlingsea. A line of police riot vans protected
the convoy, as they have through the months of protest.
Independent. 19th April 1995. p1 col 6
In the first line, arrested is being used as a hyponym of anti-protest tactics, or
synonomous with its; live animals is reiterated as the superordinate sheep which is also
repeated, making it clear that it is this particular live animal which is being discussed in
the article. Lorries and convoy are also used synonomously whilst the vans used by the
police is a reiteration to distinguish them from the vehicles used to transport the sheep.
Demonstrators is repeated; a more sensationalist style of reporting might have used
more lurid synonyms such as antivivisectionists or extended noun phrases such as:
extremist animal rights protestors.
Collocation describes the way in which certain words commonly (or uncommonly)
associate with one another in a semantic way over and above their syntactic ordering
(see DL:DE section 6.4.1). Some adjectives, for example, are used with some nouns
and not with others. The adjective beautiful collocates with the noun day more often than
with night, as do other adjectives such as sunny, warm and bright. The phrase a sunny
night is syntactically accurate but semantically rather suspect unless used in an ironic
way, since pretty and man do not normally collocate. Similarly, we would expect to see
the adjective shabby applied to the nouns clothes and treatment rather than to water or
to a baby. Some verbs regularly collocate with particular nouns, particularly ones
associated with animal and insects: bees buzz, dogs bark and geese quack; or transport:
drive a car and ride a bike. The reasons why some words have particular associations is
not all that clear. We do know that some words are more likely to combine with specific
items to form natural sounding combinations while others do not, even though they are
possible or understandable. For example, we call milk that has gone off sour, whereas
butter which has gone off is rancid and eggs rotten. All these adjectives describe
foodstuffs yet they are not interchangeable; we would think it odd for someone to
describe milk as rancid or butter as sour, yet these words essentially describe the same
process. Collocation can be exploited stylistically by collocating words that do not
normally go together, particularly in poetry. For example, the following two lines from T.S
Eliots poem Morning at the window:
I am aware of the damp souls of housemaids
Sprouting despondently at area gates
We do not normally collocate the adjective damp and verb sprouting with the noun soul,
or think of such an action as a despondent one. Damp and sprouting are terms more
normally collocated with the weather and/or gardening, rather than with spritiuality, which
might lead us to think of a souls growth in terms of gardening. Despondently also more
normally describes an emotion, thereby endowing souls with feeling.
In addition to the grammatical and lexical aspects of cohesion outlined above, there is
also phonological cohesion: that is, cohesion arising from sound. Sound patterns are
particularly important when it comes to writing verse of any kind. For example, in
languages where stress falls on the first syllable of every word, alliteration, where the first
letter of two or more words is the same, is often favoured as a poetic device.
Phonology can be a source of cohesion in a text, particularly through alliteration,
assonance and rhyme, all of which involve textual patterning created by the repetition of
same or similar sounds. Alliteration, assonance and rhyme are among the most obvious
and easiest ways a poem can be made phonologically cohesive and are often, therefore,
very superficial (see, for example, Short 1998; Jeffries 1993). Where these phonological
aspects form the overriding criteria for the structure of a poem, such as the use of rhyme
in commercial cards for birthdays, Christmas and other seasonal events, then its
meaning is often banal. A more complex kind of cohesion is that which is created
through the interaction of phonological patterns with semantic ones. Even though sounds
in themselves have no meaning (see Jeffries 1998 chapter 2), and the associations
between sounds and meanings in language are arbitrary and conventional, there are
ways of using sound so that it complements meaning. Take for example, the following
passage from Alexander Popes Essay on Criticism which was written in the eighteenth
century. The use of italics in the poem is an eighteenth century convention for
emphasis, shown in bold below, and nouns were also capitalised then.
True Ease in Writing comes from Art, not Chance,
As those move easiest who have learnd to dance.
Tis not enough no harshness gives Offence,
The Sound must seem an Eccho to the Sense.
Soft is the Strain when Zephyr gently blows,
And the smooth Stream in smoother Numbers flows;
But when loud Surges lash the sounding Shore,
The hoarse, rough Verse shoud like the Torrent roar.
When Ajax strives, some Rocks vast Weight to throw,
The Line too labours, and the Words move slow;
Not so, when swift Camilla scours the Plain,
Flies oer thunbending Corn, and skims along the Main.
Popes key advice here is that sound must seem an echo to the sense. The sound,
according to this advice, supports, rather than creates, meaning by triggering sound-
symbolism associations, with the syntax contributing further to the total effect. For
example, the word smooth in line 3 associates through alliteration with the str of stream,
like the everflowing, stable rhythm of a stream itself. In line 7, the word /((((/ cues us in to
the significance of low vowels, especially the diphthong /((/, intended to echo the roaring
of the waves. Such a phenomenon is stylistically known as iconicity, which explores the
nature of the relationship between meaning and sound.
One further form of phonological patterning is that of metre. Metre in this sense refers to
patterns of stress in lines of verse. Stress is a feature of language which we internalise
as we learn to speak it; metre is something which is imposed on language, where stress
patterns are made to fit a particular metre (see: DL:DE sections 2.5.3 and 3.3.1). Using
metre imposes a constraint upon language which we do not normally use in everyday
speech, particularly in a language such as English which is time-stressed (see: DL:DE:
Chapter 1). Many books on stylistics and the language of poetry (see for example,
Hobsbaum 1996), give thorough accounts of metrical stress in English, of which the
following is a summary.
The most common metrical types in English are the Iamb, the Trochee, the Anapest and
the Dactyl. An iambic foot has two syllables, of which the first is stressed less heavily
than the first, ( as in te-tum) Popes poem above is written using the iambic pentameter,
which is the most common metrical line in English poetry. This metre arranges words in
lines which make up five feet or ten syllables in alternating patterns of unstressed or
weak (w) syllables and stressed or strong (s) syllables:
The sound | must seem | an ec | cho to | the Sense
w s w s w s w s w s
The Trochee reverses the unstressed/stressed or weak/strong pattern to form a stressed/
unstressed strong/weak one:
And be| fore the |Sum mer | end ed
s w s w s w s w
Stood the | maize in | all its |beau ty
s w s w s w s w
The Anapest has three feet, made up of two unstressed or weak syllables followed by a
stressed or strong.
Not a sound | hath es caped |to thy ser | vants,
w w s w w s w w s w
of prayer | nor of praise
w s w w s
The Dactyl reverses the Anapaest, with one stressed or strong syllable followed by two
unstressed or weak ones:
Lulled by the | coil of his |cry stall ine | streams.
s s w s s w s s w s
As the last example illustrates, the actual number of feet or stresses in a line of poetry
does not always fall precisely into ten syllables in the iambic pentameter or two strong
and one weak in the Dactyl. In fact, the iambic pentameter, especially as used by
Shakespeare, can be anything from four to six iambic feet, eight to twelve syllables.
Using strict metre produces rhythm in a text, but a text does not have to be metric to be
rhythmic. Free verse and prose are both discourse types which by definition do not use a
fixed metrical scheme, but in both rhythm is often a source of cohesion and sound-sense
connections. Virginia Woolf, for example, is a writer whose prose is often very rhythmic
as the following extract from her novel Mrs Dalloway (1923& 1999: 2) shows:
What a lark! What a plunge! For it had always seemed to her, when, with a little
squeak of the hinges, which she could hear now, she had burst open the French
windows and plunged at Bourton into the open air. How fresh, how calm, stiller than this
of course, the air was in the early morning; like the flap of a wave; the kiss of a wave;
chill and sharp and yet (for a girl of eighteen as she then was) solemn, feeling as she did,
standing there at the open window, that something awful was about to happen....
Much of the rhythm in this passage comes from the repetition of similar stress patterns
either within a phrase or between two consecutive phrases. The first two sentences
What a lark! What a plunge! introduce repetition, the phrases being identical in their
syntax, syllable structure and stress and following the pattern of the Anapest; two weak
syllables followed by a strong one:
What a lark! What a plunge!
w w s w w s
The fourth sentence contains two pairs of parallel phrases: How fresh, how calm...; and
like the flap of a wave, the kiss of a wave, in which the similarity of stress pattern
accompanies virtually identical syntactic and syllabic structure. Although this is not verse,
some phrases within the prose partially use the iambic pentameter; for example: and
plunged at Bourton into the open air and : that something awful was about to happen.
These alternate rhythmic patterns and mirror images of phrases strongly associate with
one of the novels main themes: life as an alteration between a joyous lark and a
plunge into despair, and the search for an equilibrium between the two.
Taken together, these various cohesive elements create what is often called the texture
of a text: that is, the property of visibly holding it together as a connected entity, rather
than appearing to be a random or accidental sequence of sentences. Different types of
texts may use different kinds of cohesion to a greater or lesser extent, creating different
types of cohesion and texture. For example, conversations typically draw on material
which is shared or taken as given, because it can be retrieved from the immediately
surrounding situation and generally use a lot of pronouns and ellipsis. Fiction such as
novels and plays depend less upon situational references and tend to use referring
expressions to thing(s) which have already been mentioned in the text itself, rather than
to anything outside it. Reports will tend to use more co-reference, determiners and
conjunctions and less endophoric reference and ellipsis than novels.
For example, the continuation of a quote from Alice Seebolds Lovely Bones used
earlier;
When I first entered heaven, I thought everyone saw what I saw. That in everyones
heaven there were soccer goalposts in the distance and lumbering women throwing shot
put and javelin (1).
[]
After a few days in heaven, I realised that the javelin-throwers and the shot-putters and
the boys who played basketball on the cracked blacktop were all in their own version of
heaven(2).Theirs just fit with mine it didnt duplicate it precisely, but had a lot of the
same things going on inside (3).
I met Holly, who became my roommate, on the third day (4). She was sitting on the swing
set (5). (I didnt question that a high school had swing sets: thats what made it heaven
(6). And no flat-benched swings only bucket seats made out of hard black rubber that
cradled you and that you could bounce in a bit before swinging) (7). Holly sat reading a
book in a weird alphabet that I associated with the pork-fried rice my father brought home
from the Hop Fat Kitchen, a place Buckley loved the name of, loved so much he yelled
Hop Fat! at the top of his lungs (8).
An analysis of the stylistic range of cohesion in this passage would consider use of
referencing, both exophoric (heaven, everyone) and endophoric : What I saw in line 1
points forward to soccer goalposts in the distance and lumbering women throwing shot
put and javelin. There is repetition of heaven and swings, emphasising the importance of
both. It would also consider how conjunctions such as the temporal conjunction after and
the additive and in sentence 2 are used to logically connect parts of the text. If we look
more closely at sentence 8, then we can see that this sentence is made up of six
clauses. All of these are short and simple in structure, but are ordered in such a way as
to leave the reader with interpretative work to do, in order to understand what the
narrator is trying to say: instead of being told that Holly is Vietnamese, we are told she
is reading a book in a strange alphabet, which is then related to a Chinese takeaway the
narrator knew when she was alive, that then leads onto a reminiscence of her brother,
Buckley
3.5 Deixis and Fictional Worlds
Deixis refers to the ways in which words make direct links with the situational context of
the speaker or writer. (see also TL:TE Part II section 5.2 and DL:DE 7.2.2). Typically, as
Chapter 2 has discussed, speech often uses deixis, since we rely on context a great deal
in natural conversation. For example,
Come here and look at this mess.
Go over there and look at that mess.
These two statements point to two very different scenarios: one, where the mess is
physically close to the speaker, and the other one where it is further away, although both
are within sight. In plays or prose, as well as within poetry, deixis or deictic expressions
help to create and sustain the world of the play or narrative by referring to places, people,
times and events that have occurred within it. In writing, the use of deixis links the world
of the narrative or the poem with that of the reader. For example, in poetry, deixis can be
used to imply that the reader takes part in or watches a scene or events alongside the
poet:
That spot with spice-blooms must need be oerspread
Where such wealth to rot is run;
Blossoms pale and blue and red
There will shine full bright against the sun.
Hillman The Pearl (1360&1400 & 1961: lines 25-28)
In this example, through the shifting use of that and there, the poet is inviting us to share
his view,, and to read the poem as if we were standing beside him. This shifting of points
of reference, known as deictic shift, moves the point or points of focus from one place,
time, person or thing to another. In this way, deixis also extends the places, times,
people and things created in a text. Words which signal the use of deixis include
personal pronouns (I, you, we, they and so on), demonstrative determiners (this, that,
these, those and so on), the article or determiner the and adverbs of time and place.
Stockwell (2002:46) points out that understanding deixis involves mental processing that
is implied or inferred rather than made explicit by the linguistic structures of the text.
Various deictic fields or categories can be identified, including: perceptual, spatial,
temporal, and textual.
Perceptual deixis refers to participants in a text, including personal pronouns I, me, you,
they, it; demonstrative pronouns these/those; definite articles and definite reference such
as the woman and mental states such as seeing and believing. Third person pronouns
and names are more commonly viewed as part of reference, but Stockwell (2002:46)
argues that: taking cognition seriously means that reference is to a mental
representation is a socially located act and is therefore participatory and deictic.
Spatial deixis refers to expressions locating the deictic centre in a place, such as
adverbs here, there, nearby, far away; demonstratives this, that; locatives such as down
the hill, out of Africa and verbs of motion such as come and go.
Temporal deixis refers to expressions which locate the deictic centre in time, such as
temporal adverbs, today, yesterday, soon and so on; locatives such as in my childhood
and a week from now and verb tense and aspect which differentiate between the
speaker, the story and the receiver.
Textual deixis refers to expressions which foreground textuality, such as chapter
headings, titles, and any features which draw attention to the text itself or its production.
For example, take the following two paragraphs from George Elliots novel Mill on the
Floss (2002 & 1860). The first is the last paragraph of Volume 1 Chapter 1, and the
second is the beginning of the first paragraph of Volume 1 Chapter 2. The ways in which
these chapter headings appear is an instance of textual deixis. The novel is framed at the
beginning by the musings of an unnamed and unidentified narrator, who is the deictic
centre of the narrative. The end of Chapter 1 and the beginning of chapter 2 signal a
major deictic shift from the deictic centre of the narrator to the character of Mr Tulliver:
:
Ah, my arms are really benumbed. I have been pressing my elbows on the arms of
my chair and dreaming that I was standing on the bridge in front of Dorlcote Mill as
it looked one February afternoon many years ago. Before I dozed off, I was going to
tell you what Mr and Mrs Tulliver were talking about as they sat by the bright fire in
the left-hand parlour on that very afternoon I have been dreaming of.
What I want, you know, said Mr Tulliver, what I want, is to give Tom a good
eddication: an eddication asll be a bread to him. This was what I was thinking on
when I gave notice for him to leave th Academy at Ladyday. I mean to put him in a
downright good school at Midsummer.
The deictic shift to Mr Tulliver is preceded by the narrator directly addressing the reader
about what s/he is about to tell them. The first part is clearly deictically centred upon the
narrator as perceptual deixis, with first and second person pronouns appearing
throughout: I refers to the narrator and you to the reader. Further spatial deixis. Mental
states are also indicated in the nouns and verbs benumbed; pressing; dreaming; dozing
. Spatial relations are indicated by the chair in the room in which the narrator is sitting,
and beyond it by the bridge in the narrators dream and Mr and Mrs Tullivers parlour.
Temporal relations signalled by I was going to tell you signal a shift in the deictic centre
that occurs at the beginning of chapter 2 which is perceptual; from the narrator to Mr
Tulliver, which is temporal, in that the reported events take place at some unspecified
point back in time, and spatial, as the narrative moves from the narrators chair in an
unspecified room to Mr Tullivers left-hand parlour. As the novel progresses, the deictic
centre shifts between Mr Tulliver, his wife Mrs Tulliver, their two children Tom and
Maggie, and the narrator. Viewed in this way, deixis extends categories of deictic
reference from short stretches of text of two or three sentences to complete texts and
beyond, providing an account of how we as readers mentally process implied and
inferred fictional or text worlds we encounter in the texts that we read.
There are currently two textual meaning models based on the concept of worlds. The
first derives from Werths text world theory (1999) and is developed in Gavins (2006).
This theory aims to account for the conceptual space that links narrative levels by
proposing three worlds of discourse. The first is the immediate discourse world,
inhabited by an author and a reader . Understanding of this world by the reader is
dependent upon their knowledge and experience of what is being described in the
discourse world, which creates the second level of text world, and requires
understanding, memory and imagination as well as direct perception, Text worlds are
defined deictically and referentially, anchored or fixed to the world depicted in the
dicourse. For example, take the opening line of David Guttersons novel Our Lady of the
Forest (2003):
The girls errand in the forest that day was to gather chanterelle mushrooms in a bucket
to sell in the town at dusk.
Deictic references pick out spatial location (in) and temporal location (that) whilst
referential information identifies entities present in the text world (the girl; the forest;
chanteerlle mushrooms; a bucket; a town).
The third type of conceptual space identified by Werth is a sub-world. These occur when
a character projects thoughts and reflections, to create another space inside the text
world.
The second model of conceptual tracking during the course of reading a narrative is
Emmotts framework of narrative comprehension (1997). This framework aims to
account for the ways in which a reader can concentrate on one context in particular,
whilst holding on to ones previously encountered. She has identified two processes
called priming and binding. Priming refers to the process by which one contextual frame
becomes the main focus of attention for the reader, whilst binding refers to the way
episodic links between people and places are established in a text, and create a context
which is then monitored in the mind of the reader. Narrative strands that have been
bound into the story but been temporarily left alone by a narrator remain in a fictional
place until they are brought back in, or bound out. Text world theories such as these
aim to account for or emphasise the mental processing that goes on in the act of
reading, rather than upon textual representation. As Simpson (2004:92) points out,
finding a balance between the two is important, as a stylistic analysis can go too far in
either direction.
3.6 Similes and metaphors
Simile and metaphor, traditionally associated with literary analysis, are ways of
identifying patterns in language which are less to do with syntactic patterning and more
to do with semantics, by exploiting unusual collocations to achieve a particular effect ut
poetry or poetic language is not the only kind of language to exploit words in this way.
Much work in stylistics has been undertaken in the area of metaphor in recent years,
most of it influenced by social psychology which has informed cognitive linguistics and
building upon the seminal work of Lakoff and Johnson (1980). Traditionally, the study of
metaphor has restricted the texts under analysis to literary ones, but the publication of
Lakoff and Johnsons book and subsequent work in cognitive linguistics has
revolutionised the way we think about metaphor. Lakoff and Johnson and others working
in the field of cognitive psychology, have pointed out that far from being restricted to
literary and poetic language, metaphor pervades much of our everyday discourse and is
thus pervasive in all language use, both spoken and written. For a start, we are probably
all aware, as students of English or English Language, that meaning is not stable, and
can change. To create new ones, we often use existing words metaphorically. As the
word gets used more and more, new usages become absorbed into the language as
commonplace, traditionally known as dead metaphors: for example, foot of a bed, table
leg, foot of a page. Much of the terminology we now associate with electronic
communication - net, web, mouse - is metaphorical Their use is so commonplace and,
within traditional categories of metaphor, would be considered dead. Lakoff and
Johnson argue that far from being dead, the reason they are there is because metaphor
permeates the whole way in which we think about the world. They argue that metaphor is
not simply a conceptual category of one aspect of language, but fundamental to our very
thoughts and actions. The concepts that govern our thoughts are not just matters of
literal reason, but of allusion and analogy which also govern our everyday functioning
down to the most mundane details. Such metaphorical concepts structure what we
perceive, how we get around in the world, and how we relate to other people. Our
conceptual system thus plays a central role in defining our everyday realities. Since
communication is based upon the same conceptual system that we use in thinking and
acting, language is an important source of evidence for what that system is like.
Many everyday expressions appear to share underlying conceptual structures that are
understood and shared by groups of people. Grouped together, such common
expressions are called conceptual metaphors and are normally identified by being
written in capitals. Take the following example given by Stockwell (2002:110):
ARGUMENT IS WAR
Your claims are indefensible.
He attacked every weak point in my argument. His criticisms were right on target. I
demolished his argument.
Ive never won an argument with him. You disagree, Okay, shoot!
If you use that strategy, hell wipe you out. He shot down all of my arguments.
We do not just talk about arguments in terms of war, we win or lose them. There is no
physical battle, but a verbal one, and the structure of an argument: - attack, defence,
counterattack, etc - mirrors the movements of war. In this sense, the ARGUMENT IS
WAR metaphor one by which we live in Western culture; it structures the actions we
perform in arguing. It is not that arguments are a sub-species of war, but that the two
different things - verbal discourse and armed conflict - and different kinds of actions.
ARGUMENT are partially structured, understood, performed and talked about in terms of
WAR. The concept is metaphorically structured and, consequently, the language is
metaphorically structured. Such a way of talking about argument is far from being a
dead metaphor, but has become so ingrained in thought and language that it has
become the ordinary way of talking about war. Metaphors are possible as linguistic
expressions precisely because they are already embedded in a persons conceptual
system. Conceptual metaphors such as ARGUMENT IS WAR, LIFE IS A JOURNEY,
TIME IS MONEY and so on not only attempt to describe but also to structure a great deal
of our everyday discourse, and are thus not confined to literature.
Developments in cognitive stylistics build upon the work of literary critics such as I.A
Richards and cognitive theorists such as Lakoff and Johnson in proposing new models
for analysing metaphor. Richards (1925) categorised metaphor into three constituent
parts: tenor, vehicle and ground. Tenor is the subject of the metaphor, vehicle is the
terms in which the subject is expressed and ground is the common properties the other
two categories, tenor and vehicle, share. Take the following example from
Shakespeares Romeo and Juliet: Juliet is the sun.
Juliet = tenor
sun=vehicle
ground = common properties of warmth, beauty, life-affirming.
A cognitive approach would argue that in this example, we can only make the inference
of warmth, beauty and life-affirmation because our conceptual system sees the sun in
those terms and makes it possible to assign them to a person. We can thus make sense
of the metaphor because of how our perceptions are formed. Consequently, the
categories have been re-named to provide a different emphasis. Instead of the category
ground, in cognitive stylistics it has been re-named mapping. Ground is re-defined as
a mapping onto the properties between the two spaces or domains that are inferred:
warmth, beauty, life-affirming. Similarly, tenor is re-defined as target and vehicle as
source.
Juliet is the sun is thus analysed as:
Target= Juliet
Source =sun
Mapping= warmth, beauty, life-affirming.
However, it may be that a 21
st
century Western conceptual system sees the sun in
terms of warmth and life-affirmation, but such a cognitive approach assumes that
concepts are universal, or somehow acquired, with little or no reference to social and
cultural background, and that the association between the sun and life-affirmation is a
universal one. In Europe, we may well view the sun as warmth and life-affirming, but the
inhabitants of the Sahara desert or Australian bush may not. Similarly, in Japan the sun
is the symbol of the flag and of nationalism and in seventeenth century France it meant
something else yet again, generally political connotations of the divine right of the
monarchy, and specifically Louis X1Vth. The meaning of the word sun then, or, more
accurately, the properties associated with it, change or transfer its meaning from culture
to culture and from century to century within any given culture. Thus, the mapping we
infer is culturally, as well as cognitively, determined.
Consequently, although such a cognitive approach helps us to understand the
metaphoric nature of much of our discourse, and its reliance upon conceptual structures
being understood and shared by groups of people, it has not yet found a way of
accounting for why particular concepts are formed in particular ways. A cognitive
approach to analysis has yet to account for where our concepts come from or how they
originate, in ways that take account of shared cultural and social understanding in
addition to personal reaction.
3.7 Narrative
One possible basic definition of narrative is the choice of a specific linguistic technique or
techniques to report (mainly) past events. At its simplest, a narrative recounts a chain of cause
and effect which happens in space and time to a particular set of people. Even the
simplest narrative follows a pattern of development, with the listener or reader being
steered through the course of a tale by the teller or author which takes place in the two
dimensions of time and space. Generally, we are unaware that we are being guided, and
it is only when we begin to look more closely at the way in which language is structured
that we begin to realise that we are.
One important aspect of the way in which a narrative develops is not only the interplay of
cause and effect between events, but the way in which the story itself is narrated and the
way speech and thought are represented within it. Texts which tell a fictional story
usually describe the events, characters and setting within them from a particular
perspective, which may be the authors, the characters (or characters) or a combination
of both. However, providing a detailed and rigorous stylistic model of narrative has
proved extremely challenging, and a comprehensive one has yet to be developed. There
are, nevertheless, specific aspects of narrative which can inform a stylistic analysis and
models have been developed that have given rise to categories of stylistic analysis.
Two basic components of a narrative, stemming from the Russian Formalists (fabula and
szujet) are plot and discourse. Plot refers to the abstract story line of a narrative: what
you would say if asked to sum up what a narrative was about. Discourse refers to the
actual way in which or the manner in which the plot is narrated, including the use of
flashbacks, repetition and prevision which disrupt the basic chronology of the plot. Plots
can be realised through a variety of different media. For example, the novel Frankenstein
has been made into several film versions, all using material taken from the plot but
realised in different ways, and also written as a poem by the poet Liz Lockhead. Similarly
the play Romeo and Juliet can be performed as a ballet. Narrative discourse is thus not
confined to written narrative in poetry and prose, but also extends to other media such as
film, ballet, musicals and comic strips. Consequently, we can identify different textual
media for the analysis of narrative, which for stylistic analysis, is the written medium.
Simpson (2004) identifies five other basic aspects of medium in addition to the textual
one: sociolinguistic code; actions and events, points of view, textual structure and
intertextuality.
Sociolinguistic code locates a narrative in time and place by considering how language
reflects sociocultural context. It encompasses, for example, the varieties of dialects and
accents used in a narrative. As Chapter 1 has pointed out, it may be generally assumed
that standard English is not a dialect at all, but all speech and writing is spoken and
written in a dialect of some sort, whether it is standard or non-standard, prestigious or
low status. Although poetry has often been written in dialect, a common convention in
narratives is that the prose is written in standard English and only the characters speech
is represented in dialect such as in, for example, the novels of [Link] or
Elizabeth Gaskell; a model of representation which serves to emphasise the social class
divisions between the characters and gives us as readers, a flavour of the kind of society
within which the characters live. Although writing in dialect has a long history (see
Chapter 1), it has not usually formed part of mainstream writing, other than in poetry.
One such famous, contemporary example is the novel Trainspotting written by Irvine
Welsh, which is written entirely in Scottish dialect (1993:75)
The problem wi Begbie wis well, thirs that many problems wi Begbie. One ay the
things thit concerned us maist wis the fact thit ye couldnae really relax in his
company, especially if hed had a bevvy. Ah always felt thit a slight shift in the
cunts perception ay ye wid be sufficient tae change yir status fae great mate intae
persecuted victim.
Since non-standard dialects are associated with lower social class in England (see
Chapter 1), then its use by the narrator reinforces a class distinction between himself
and the establishment. Later in the novel, the protagonist appears before a magistrate,
and the usual ways in which narrative prose are written is reversed, in that the prose and
the narrators speech and thought is written in dialect, whilst the speech of the magistrate
in standard English:
- You stole the books from Waterstones bookshop, with the intention of selling them,
he states. Sell fuckin books. Ma fuckin erse.
- No, ah sais.
- (1993:165)
Actions, events and point of view are ways of considering the intersection of
narrative and character. Actions and events are described in Chapter 4 under the
heading of transitivity. Point of view, as the term implies, is the particular perspective
from which a narrative is told. Any written text has an author, known or unknown. Authors
of factual texts are usually at pains to separate themselves from what they are writing,
whereas in fiction the author can take a more active and positive role in telling the story.
The way this is usually done is by having a narrator. Much has been written about the
stylistics of narrators and narration (see, for example, ), the main points of which are
summarised below.
The degree to which writers make themselves known to readers as the narrator of the
story can vary, according to how much a writer chooses to directly intervene in the telling
of the story by choosing to be or create either a personal or impersonal narrator.
Telling the story itself from either narrative perspective involves further choices about
whether the narrator knows everything about the characters and events (authorial
omniscience) or has a restricted narrative perspective (authorial reportage). Both of
these perspectives can be referred to as authorial voice, though the narrator might be
different from the author. Taken altogether, the ways in which the narrative is told, with
either an impersonal or personal narrator, an omniscient or reporting authorial voice,
contribute to the particular perspective or point of view from which a story is told.
A personal narrator is one who intrudes into the story to address us, the readers,
directly, to make comments, pass judgement on or moralise about the characters and
events of the narrative, and may even appear as an I outside the story. Such an
intrusion was fairly common in earlier novels, such as Henry Fieldings Tom Jones. In
this novel, Fielding interrupts the telling of the story for whole chapters at a time whilst he
passes comment on the events that have just occurred. A very general characteristic of
the development of the novel is that this kind of personal intrusion has become used less
and less often within narrative so that when it does reappear in the postmodern novel,
such as in John Fowless novel The French Lieutenants Woman, its appearance is
startling and seems something new.
By contrast, an impersonal narrator is one who is less intrusive, and simply reports the
events of a story, without passing comment on them. They are much more likely to use
the third person in telling the story, that is, using characters names and the personal
pronouns he, she and they. Written narratives with an impersonal narrator are least likely
to use the second person, you or the first person I in telling their tale. It is perfectly
possible for both types of narrative reporting to be used within the same narrative. For
example, the 19th century novelist George Eliot generally writes as an impersonal
narrator style in her novels, but occasionally she addresses her readers directly,
changing to a personal narrative style. In the extract from The Mill on the Floss in 3.5
above, moves from the personal narrator in Chapter 1 to the impersonal narrator in
Chapter 2.
First person narratives, such as the example from Trainspotting above, tell the story from
one characters point of view, who may or may not be the author, as in autobiographical
novels such as [Link] Catcher in the Rye or fictional ones such as Bridget Jones
Diary . They are more common than second person narratives, that is, written in the
second person you, and restrict the point of view made possible within the narrative to
that of the I of the story.
Telling a tale also involves a writer in making choices about the extent to which details of
characters are to be portrayed. He or she may choose to be authorially omniscient, in
which case they deliberately enter their characters consciousness and are thus able to
tell us, the reader, about characters thoughts as well as their actions. The 19th century
writer Thomas Hardy, for example, uses a generally impersonal narrative style but is
authorially omniscient, able to reveal his characters innermost thoughts and feelings.
Alternatively, an author may choose to report events which are external to the characters
and leave their thoughts alone, as exemplified in many novels by the 20th century writer
Graham Greene. In novels of this kind, what characters think or feel is inferred entirely
from their speech and actions, rather than through descriptions of both thought and
speech, as is the case with authorial omniscience.
In addition to the voice of the author or narrator, a writer may also choose to tell the tale
from a particular character or characters perspective with which we as the reader are
invited to identify and who may also be the chief narrator of the tale. This character is
usually the first one to whom we are introduced in the book. Usually, narratives are told
from the point of view of the main character, and the events of the story are told as they
relate to him or her. Examples include novels such as Jane Eyre by Charlotte Bronte,
Emma by Jane Austen and The Passion of New Eve by Angela Carter. Other characters
appear in relation to the protagonist, and our opinion of them tends to be formed by that
of the main character. Novels written in the first person are generally told from one
characters point of view: the I who is telling the story.
Even though there may appear to be one main character in a story, a writer may also
choose to tell a tale from more than one perspective, moving from one characters
perspective to anothers. A novel such as Women in Love by D.H. Lawrence does just
this, moving between a description of the actions, thoughts and feelings of two sisters to
those of two men with whom they form relationships using an omniscient authorial voice.
Ian McEwans novel Enduring Love moves between a description of the actions thoughts
and feelings of two witnesses to a ballooning accident and the man involved in the
accident. Telling the tale from more than one characters perspective provides authors
with the opportunity to develop different narrative strands or sub-plots within the main
one, which may be used to keep the reader guessing, as in detective fiction, or in
suspense and thrillers, as well as in fiction more generally, for example in Little Dorrit by
Charles Dickens or The Man Who Made Husbands Jealous by Jilly Cooper.
Textual structure refers to the ways in which narrative units are arranged or combined
in a story. Several models of narrative structure exist (see, for example Propps
morphology of the folktale), of which one of the most influential has been William
Labovs 1972 framework of natural narrative (1972: 359-60). Labov derived the model
from stories told during the course of natural conversation from which he isolated six core
and recurrent features or categories Each one of these has a corresponding
hypothetical question that is addressed, a narrative function and linguistic form,
summarised below:
1. Abstract: The question here is: What has happened, or what is the story about?
the narrative function is a signal that the story is about to begin and the linguistic
form is often a one-sentence summary.
2. Orientation. The question here is: who or what is involved in the story? The narrative
function helps the reader to identify, for example, when and where the story takes place
and the linguistic form is often sentences describing participants, times and places, most
usually characterised by use of the past continuous tense and adjuncts of time, manner
and place.
3. Complicating action. The question here is: what happened? The narrative function is
to provide the what happened? part of the story and the linguistic form is usually
sentences ordered into narrative clauses with a verb in the simple past or present.
4. Resolution. The question here is: and what happened in the end? The narrative
function here is to bring the story to an end, and the linguistic form is normally expressed
as the last of the narrative clauses that began the complicating action.
5. Evaluation. The question here is: so what? or what else?. The narrative function here
is to make clear the point of the story, and the linguistic form includes evaluative
commentary, embedded speech, modal verbs, negatives and so on.
6. Coda. The question here is: how does it all end? or how does it relate to the here and
now? or supplies a moral or lesson learnt by the teller or protagonist. The narrative
function here is to signal the end of the story or to provide a bridge back to the present.
The linguistic form is often a generalised, timeless statement.
This model has worked equally well when applied to written narratives. For example,
there is currently running on British and American television a drama called House,
centring upon a cantankerous doctor of the same name. Each episode begins with a
prospective patient being admitted to hospital (abstract) for an uncomplicated procedure
or unconfirmed diagnosis (orientation). The illness, however, from which the patient is
suffering turns out to be far from straightforward and medically baffling (complicating
action). House and his team run through a series of options whilst the patient gets closer
and closer to deaths door (complicating action). House finally discovers the correct
diagnosis, usually at the eleventh hour (evaluation). This serves to re-inforce his skill as a
doctor (coda) which more than makes up for his boorish and non-cooperative stance with
patients. Toolan (1988 & ) and Simpson (2004) provide more detailed accounts of
stylistic approaches to narrative.
Intertextuality refers to the ways in which writing does not occur in a social and
historical vacuum, but echoes or alludes, either implicitly or explicitly, to other works. A
novel such as David Lodges Nice Work about a female lecturer specialising in the
nineteenth century Industrial novel, begins each chapter with a quotation from one such
nineteenth century Industrial novel. Angela Carters The Magic Toyshop makes a
stylistic allusion to a line by the Romantic poet Shelley, in the line: Look upon my works
ye mighty and beware with the original final word despair implicit as well as altered.
3.8 Presenting Speech and Thought
One way in which point of view is represented or can lead to a change of perspective
within a narrative is by the presentation of speech and thought. Speech and thought in
fictional texts are, by definition, imaginary, and although they may follow the rules of
natural conversation such as those discussed in chapter 2, there is nothing natural here.
Speech and thought in fictional texts are artificial, contrived and interwoven into a wider
creative structure where they always do more than just represent talk no matter how
natural it may appear to be.
Real time speech is very different to speech that is written down. One immediate and
important difference is they each use different mediums of expression: speech uses
sound and hearing whereas writing uses graphology and sight as the vehicle for words.
Spoken speech is also interactive, usually involving at least two people, which makes it a
dialogue and, depending on the situation, there can be no guarantee of topic when
speakers take turns in speaking (See Chapter 2). Where speech requires little or no
interaction, such as during a lecture, after dinner speech or parliamentary speech, then it
has usually been scripted and rehearsed and is more of the form of a monologue than
dialogue. Written dialogue or representation of speech and thought is interactive in a
completely different way from spoken speech. The interaction happens on two levels:
firstly, between the characters represented and secondly, between the reader and what
is read. A writer is also in control of what characters say compared to participants in
speech. Consequently, the frameworks or models for analysis that have been developed
to analyse written speech are very different from those used to analyse actual speech, as
discussed in chapter 2. By far the most influential of these has been the one developed
by Leech and Short (1981) , Short (1996) and Fludernik (1993).
There are five main ways in which speech or thought can be represented in narrative
which vary according to the way in which a characters speech or thought is represented.
They are: Direct, Free Direct, Indirect, Free Indirect and Narrative Report. Consider
this example:
Do you see her much she said, half-concerned.
In this sentence, there are two independent voices at work. One is the quotation which
gives the actual words spoken and is bound by speech marks (Do you see her much?).
The other is the rest of the sentence: a clause which reports information about the
quotation, usually who said it, with further information such as how and/or to whom given
to us by the narrator, known as a reporting clause: (she said, half-concernedly).
Reporting speech in this way is called direct speech (DS), where the actual words
spoken by a character are written, usually within a sentence that also gives us
information about the words and using punctuation marks to mark the actual words
spoken. The reporting verb (the word which tell us what was being done) may well tell us
a lot about the purpose, emotions and intentions of the utterance. It may also tell us
about the expression on a persons face and their emotional state. For example, notice
the difference between:
Go away! she screamed.
Go away, she said.
Adding an adverb or a prepositional phrase after the reporting verb as the following
examples do, are other ways of conveying more information about what is said.
Most of them were too young, she replied.
Most of them were too young, she replied sorrowfully.
Most of them were too young," she replied with disgust.
The reporting of direct thought (DT), is marked by the use of reporting verbs such as
wondered, thought, mused, and so on, as in the following example:
Does she still like me? she wondered.
Sometimes, writers may use dashes instead of conventional speech marks, or not use
them at all, depending on the particular effect they want to achieve, such as making
actual speech appear to be more like free indirect (see below) or to distinguish it from
thought. Throughout this century, writers have increasingly experimented with speech
representation, such as presenting speech directly without the accompanying
punctuation or reporting clause characteristic of direct speech. Such variations from the
representation of direct speech and thought are known as free direct (FDS and FDT): For
example:
She said I want to see the elephants
I want to see the elephants
In the second sentence, the reporting clause is left out, but the punctuation kept in.
Representing speech in this way tends to minimise the narrators role and foreground the
character and his or her speech. Consider this example:
She asked him whether he saw her much.
In this example, there is only one voice and one point of view. The narrator is using their
own words rather than the words that were actually used by the character and presenting
their version of them. There is a reporting clause (she asked him) and a reported
clause (whether he saw her much). The narrators point of view prevails and so no
speech marks are needed. Reporting speech or thought in this way is called indirect
speech (IDS or IDT). It usually has the effect of foregrounding the narrator rather than the
character who has spoken
Narrative report of speech or thought (NRS or NRT) involves a narrator reporting a
characters speech or thought without giving any indication of the actual words used.
Unlike the explicit models given above, NRS/T can be used to summarise whole
stretches of reported speech and thought.
The different ways in which writers represent speech and thought in writing is
summarised in the table below:
SPEECH THOUGHT
DIRECT
He said, Ill come back tomorrow. What will they say of me? she wondered.
FREE DIRECT
Am I too late? Was I too late?
INDIRECT
He said that he would return the next day She wondered what they. would say of
her.
FREE INDIRECT
He would come back tomorrow What would they say of her?
NARRATIVE REPORT
She told him about her evening She wondered about his love for her
Generally, the narrators role or function is to inform the reader of what is going on and/or
to evaluate a particular or whole situation for the benefit of the reader. The characters
role or function is usually to make claims, express doubts and desires, display all sorts of
emotions and evaluate themselves and/or others and/or a particular situation from their
individual point of view. Take the following example, from Virginia Woolfs novel Night
and Day:
Denhams one wish was to leave the house as soon as he could; but the elderly ladies
had risen, and were proposing to visit Mrs Hilbery in her bedroom, so that any move on
his part was impossible (1). At the same time, he wished to say something, but he knew
not what, to Katherine alone (2). She took her aunts upstairs, and returned, coming
towards him once more with an air of innocence and friendliness that amazed him (3).
(1919&1978:140)
This extract narrates Denhams free indirect thought which places us, as readers, in the
privileged position of reading Denhams mind and his reaction to the news he has just
heard: that Katherine is to marry William Rodney, which includes comments about the
movement of characters to and from the room.
We might consider or ask why have speech represented in narrative at all. It is perfectly
possible to recount a tale without using speech at all, but its use has become a common
feature of fiction, and part of what we expect writing which tells a story to include.
Generally, including speech and thought representation in writing does three things:
1. It interrupts the general flow of the narrative, slowing it down and concentrating
attention on a particular character, event, relationship and so on. In the extract above,
the flow of the narrative is interrupted to focus upon Denham and his thoughts
2. It develops and brings out relationships between characters; their personalities may be
revealed by what they say, their reaction to what is said and/or by what others say about
them. In the extract above, aspects of Denhams personality are revealed by his
thoughts.
3. It adds to the sense of social as well as physical background through personal
interactions which add further details associated with it, such as mannerisms, its
concerns, subject matter and so on. In the extract above, the setting, places the extract
not only in a particular place but also at a particular time in history and also places
Denham in terms of social class and their place in society.
3.9 Dialogue in drama
Most scripted speech, especially dialogue, aims to reproduce spontaneous, natural,
everyday speech. However, there are several important differences between actual
spoken speech and written speech intended to imitate it, which makes it difficult or
inappropriate to apply models of spoken discourse analysis to it as described in chapter
2:
- someone else has usually written the words that a different person speaks, unlike in
normal conversation where the speakers are the authors of their own words;
- we take part as active participants in any conversation with which we are involved,
whereas with scripted speech we take the part of onlookers who see and hear but do not
engage directly with the speakers, although we may be appealed to in a way which
invites response. Scripted speech implies an audience in much the same way as any
other form of written language which has implications for the pace at which something is
spoken, its phrasing and degree of implied context;
- scripting speech means that it is written in advance of being spoken and can be edited,
unlike everyday, ordinary conversation which cant.
- Scripted speech may look like spoken language written down, but if you compare a
transcript of a normal conversation with that of scripted speech then you will immediately
see many differences. In conversation, we usually take it in turns to speak and this is
known as turn-taking (see chapter 2). Often, though, we interrupt before someone has
finished their turn, whereas scripted speech normally allows each character to take their
turn. Scripted speech cannot assume such a shared understanding on the part of its
listeners. For example, in scripts of long running soap operas or series such as
Coronation St, Neighbours, Friends or The Simpsons, characters nearly always refer to
one another by name and explain details of events which in spontaneous conversation
would be taken for granted so that viewers new to the programme or who have not
watched it for a while can pick up the thread of the story. Scripted speech also tends to
be more organised, operating as it does within a narrative structure, usually being about
one particular thing, event or person at a time, whereas spontaneous speech can hop
about from one topic to another and back again in a seemingly haphazard way. Scripted
speech means that characters nearly always finish what they start to say without being
interrupted and talk to one person at a time, using hardly any of the normal features of
spontaneous conversation such as repetition, restructuring and fillers.
One very important difference between spontaneous and scripted speech is that the
content of scripted speech, particularly dialogic speech, is controlled by someone who is
not the originator of the words themselves, whereas the originator and controller of
content in spontaneous speech is the speakers themselves. You cannot walk away from
a script. Even though spoken conversation has an internal structure of its own, control of
it is shared in a way that of scripted speech is not.
Writers of scripted speech normally have control over both the content of what they write
and over how this is to be spoken. It is then up to the people speaking the words to
interpret these by adding tone of voice, expression, gesture etc. rather than to create the
words themselves. Scripted speech, then, can have two different audiences; the actors
or people who are to do the talking and the audience for whom they interpret the script.
When a script is written down, the words on the page, including any stage directions and
camera angles form part of the script as well as the actual words characters speak. What
we see or hear is actors interpretations of the writers words and accompanying
directions. Scripts, like any other kind of writing, can also be edited and changed. Once a
script has been written, the director may then choose to make his or her own alterations.
For example, a director may decide to cut certain lines that have been written, encourage
actors to improvise on the words present in the text or to change aspects of the location
a scene, such as changing it from indoors to out of doors and vice versa.
Scripted dialogue has to convey to its viewer or listener all the things normally described
through narration in writing. For example, what characters say has to convey emotion,
action and move on the plot as well as the words themselves. We do not have access to
the thoughts in characters heads as we do in novels, short stories or poems. All our
information about characters personalities, behaviour and actions comes from the words
which they say. An analysis of a script, then, will need to look very closely in ways which
the language spoken by characters as well as at that of stage directions gives us clues to
characters personalities and reasons for their actions as well as the actions themselves.
Consequently, although it is not always appropriate to apply pragmatic models to
dramatic dialogue, its representation shares a principle common to most models, which
is, that naturally occurring speech takes place within a context of use. Much of our
everyday conversation relies on a shared understanding of the context within which it is
taking place, and the more we have in common with the people we are talking to the
more assumptions we make. References to places, times, people and events will also
usually include a common understanding and refer to actual places, times, people and
events that make up our daily lives. The notion of context in drama can be divided into
three basic categories: physical, personal, and cognitive, to which can be added the
interactive and imaginary (see also 2.3 and 2.6.5).
Physical context, as the description implies, refers to the actual setting, which in drama
takes place usually on a stage of some sort with a set and props, however minimalistic.
Personal context refers to the social and personal relationships of the participants,
including social networks and group networks as well as the social status and distance
between participants. Cognitive context refers to the shared background knowledge
held by the participants, their world view, cultural knowledge and past experiences.
These three aspects of context, the physical, personal and cognitive, can be applied to
any face to face interaction, dramatic as well as natural. However, because this is drama,
there is also an interactive context. That is, what we as viewers and listeners of the
drama make of the unfolding dialogue, and how far it accords with our worlds views,
cultural knowledge and past experiences. There is also what can be called an
imaginary or creative aspect, where the participants on the stage and the audience
engage in the creation of a shared world both on the stage and beyond it. This requires
us to suspend our belief of what we see and hear in front of our eyes a stage with
people on it and become drawn into a world created by the participants which takes us
beyond the world of the stage which is entirely imaginary (see DL:DE 7.31 and TL:DE
Part II section 5.2). Deixis (see 3.5 above) plays an important part here, since it serves
not only as a way of binding the text together but also of extending the world on the
stage or set beyond itself to incorporate elements of the represented world beyond that
which we actually see. When a dialogue is being written, whether for radio, stage or
screen, any reference to a world outside the scene cannot actually exist, but we are
invited to believe in it as part of the process of engaging with the scenes that are being
acted out. Dialogue in scripts will also refer to places, times, people and events which
may happen outside the script. As the audience, we are drawn into the scripted world
both as it is presented to us through the actual dialogue and its setting as we experience
it and by extending beyond it to places, times, people and events which we do not
actually experience but which are referred to and which may have a direct bearing on the
action played out on stage or set. For example, at the beginning of Shakespeares play
Merchant of Venice,the merchant Antonio talks about the ships he has at sea. His friend,
Bassanio, wants to marry Portia, an heiress, but is in debt and needs money to set
himself up as a nobleman. Antonio tells Bassanio that his money is tied up at sea, but will
borrow money on his behalf. Thus, although we do not actually see the ships (or the sea)
on stage, nevertheless their voyage and its success are an integral part of the plot. When
it comes to the structure of the dialogue itself, then certain features can be identified,
such as structure and strategy. Like natural conversation, dramatic dialogue is
comprised of exchanges. Consequently, a structural analysis of dialogue can be applied,
based upon exchanges such as questions and answers, statements and
acknowledgements, commands and requests and so on. Burtons book (1980) is a
seminal work on the structure of dialogue. Burton uses a variety of different models in
conversation analysis and speech act theory (see Chapter 2) to investigate a number of
play texts to uncover patterns of dialogue which serve to delineate character. Through
her analysis of Pinters play The Dumb Waiter, she concludes that the unequal status of
characters is reflected in the dialogue. Take the following exchange as an example:
GUS: I want to ask you something.
BEN:[no response]
BEN:What are you doing out there?
GUS: Well I was just
BEN: What about tea?
GUS: Im just going to make it
BEN: Well, go on, make it
GUS: Yes, I will
(cited in Burton 1980:161-2)
In saying I want to ask you something, Gus attempts to initiate an exchange, letting Gus
know through the use of a metastatement, that is, a statement about language, which
lets Ben know what Gus is about to ask. The fact that Ben ignores this statement and
fails to provide the anticipated second half of the exchange do you? what is it or
suchlike, which immediately places Gus in a subordinate position to Gus. Ben then
initiates his own questions to which Ben responds, further reinforcing the inequality
between the two characters.
Culpeppers more recent study (2001) applies stylistic and pragmatic models of
discourse, focusing particularly on Shakespeares plays. Culpepper applies models
drawn from social psychology supplemented by ideas from cognitive linguistics, in
arguing his approach. Far from perceiving characters as if they were real people existing
in the real world, Culpepper points out that inferring characters from dialogue, or indeed
any text, relies in part on the cognitive structures and inferential mechanisms that the
audience or the reader has already developed for real-life people. Consequently, the
measure of success in characterisation is not how life-like such characterisation is but
how closely the characterisation resembles our own understanding of the particular
character being portrayed.
A strategic analysis considers the ways in which speakers communicative strategies are
sensitive to context, employing different types of utterances ranging from direct to
indirect and from polite to impolite. Dialogue which consistently violates our
understanding of such strategies, such as that to be found in the Theatre of the Absurd,
for example, often creates dialogue which is deviant, anti-realist or simply absurd, to
reinforce its central concern, which is the futility of human existence. In some plays,
however, violation is used to provide contrast between the world of the sane and insane.
Consider the following exchange, taken from Peter Shaffers play, Equus.
Dysart: So: did you have a good journey? I hope they gave you lunch at least. Not that
theres much to choose between a British Rail meal and one here.
Wont you sit down?
Is this your full name? Alan Strang?
And youre seventeen. Is that right? Seventeen..Well?
Alan: Double your pleasure
Double your fun
With Doublemint, Doublemint
Doublemint gum
(1975 1:3 )
Here, Dysart, as the doctor, is talking to Alan who has been brought in for a consultation
by his parents. Alans madness is signalled by his refusal to answer Dysarts questions
directly as a well-behaved and sane patient might, flouting the Gricean maxim of
relevance (see 2.4.2) and choosing instead to sing advertisement jingles by way of
answering Dysarts questions. In the lines which follow the extract, Dysart, by contrast,
observes the communicative strategies of consultation, even when Alan persists in
singing.
3.10 Studying stylistics
A stylistic approach to textual analysis usually involves consideration of three different
things. The first is deciding what it is you want to find out: your research question. The
second is the selection of the texts you are going to analyse, the third is selection of the
linguistic tools and any specific stylistics framework within which you are going to
undertake your analysis. Research in stylistics, just as much as in dialectology and
pragmatics, aims to be reliable, replicable and rigorous. Any of the models or frameworks
for analysis outlined above can, in theory, be applied to the stylistic analysis of any text.
However, there are certain methodologies and methods which apply themselves more
readily to some kinds of texts rather than others, summarised in 3.10.1 below.
3.10.1 Selecting a topic for study
The first thing to decide when undertaking your own research into stylistics is the focus of
your study. The data used in stylistic analysis, unlike that used in variation in English or
pragmatics, is normally that of written texts. It may be that the text selected for analysis
will be given to you, as will the focus of your analysis. If, however, you are given the
option of choosing your own text or texts to analyse, either for an essay or for an
extended piece of work, then your choice of text will depend to a large extent upon what
aspects of stylistic analysis you are most interested in investigating or, conversely,
choosing a text and then deciding which aspects of it to analyse. It is no good, for
example, choosing poetic texts or instruction booklets to undertake an analysis of
dramatic dialogue, since you will find relatively few (if any) examples; equally,
investigating speech and thought representation is best achieved through an analysis of
prose rather than poetry or drama, again because this is an aspect to be found more
often in prose than in any other genre. Some aspects of stylistic analysis such as those
concerned with narrative can be found across a variety of written genres, such as
newspaper reporting as well as prose fiction and spoken discourse. Consequently, it is
possible to choose texts for analysis across more than one genre. The methods of
analysis you choose will, to a certain degree, be determined by the type of text(s) you
choose, and the specific focus of your investigation. As has been emphasised throughout
this chapter, stylistics is eclectic in nature, and draws upon various research
methodologies associated with a number of areas of linguistic investigation including:
grammatical theory, literary theory, narrative theory, pragmatic theory and social
psychological theory. For example, if you look at vocabulary and word structure in
texts written for children, then you will draw upon methodologies associated with
phonology and grammar. If you wish to investigate differences and similarities of
narrative structure in detective fiction or any other kind of genre fiction, then you will draw
upon methodologies associated with discourse grammar and narratology. Or, if you wish
to investigate the structure of dialogue in drama, then you will draw upon methodologies
taken from pragmatics.
Undertaking research in stylistics, like variation and pragmatics, is a very practical activity
and the degree of your success rests in demonstrating your understanding of the
methodologies and methods it provides, by applying them successfully in your analysis.
For example, you may wish to investigate:
vocabulary and word structure in writing for children;
aspects of cohesion and coherence in poetry or a prose extract;
deixis and metaphor in poetry or an extract of prose fiction;
narrative structure in detective fiction or any other kind of genre fiction;
point of view and authorial voice in prose fiction;
the use of free direct and indirect speech and thought in Modernist novels;
the structure of dialogue in plays.
Once you have chosen the focus of your research project, then it is just as important for
stylistics as it is for research into variation or pragmatics that you read any relevant works
which may help you to orientate your study within the body of published research
undertaken, in both books and journal articles. By reading other works you form a better
picture of the field of enquiry in which you wish to locate your project, such as what the
arguments are about and what the questions are. Knowing this will help you to formulate
your own question more clearly and give you a picture of what findings you might expect,
and what would be unexpected. Another important aspect of reading around your topic is
that it gives you ideas and information regarding the number of texts you need to choose
for your own study and the length of any extracts.
It is important to take notice of the date of publication of any work, since that which is
most recent will be most up to date in developments in the field, and most likely refer
back to older references still being cited by researchers, which gives an idea of their
importance. The key to reading for the purposes of writing a research project, or indeed
any essay, is not the amount of reading that you do, but developing the ability to perceive
which reading is important to your own research question.
3.10.2 Choosing and collecting data
The data you choose to collect for stylistic analysis are written texts, and the most
pressing question here is how many texts to choose and collect and/or how many
extracts and of what length. Much work undertaken in stylistics is qualitative; that is, it
analyses a few texts in detail to demonstrate how a particular stylistic method works.
Consequently, the texts you choose have to be sufficiently lengthy to allow for a detailed
analysis, yet not so long that you either cannot complete the analysis or exceed the word
limit set on your study in order to complete it.
For example, if you wish to investigate vocabulary and word structure in writing for
children for a short study of essay length, then you could choose a complete story of
about 200 words, or extracts from two or three different stories of about 75 words each.
These extracts could be from stories by the same author or from one or more authors to
provide a contrastive dimension to your study. For a longer study, then as a rough guide
you would double this number. If you wish to study aspects of cohesion and coherence
or deixis and metaphor in poetry or a prose extract, then the lengths and numbers of
texts will be different depending upon choosing poetry or prose. Depending upon the
length of the poems you choose, then three poems of about 20-30 lines each would be
sufficient, or one 100 line poem or extract from a longer one. Your choice of prose
extracts will depend upon whether you wish to focus upon one extract from one text, or
one or more extracts from the same text or different ones. As a rough guide, extracts
which total about 600-800 words are probably sufficient for a short study of essay length.
If your study focuses upon aspects of narrative such as narrative structure and/ or point
of view and authorial voice, then the texts chosen for study here will be longer, and often
complete texts. If your study concentrates solely on narrative structure, then you can
choose four or five complete novels for an essay length piece of research, though there
may be some need for sampling, depending on the detail of your intended analysis. If
your study considers narrative structure and point of view and authorial voice, then two
novels plus short extracts of anything from 20 to 100 words from each is probably
sufficient for an essay length study. For a longer study, then more novels would be
necessary and longer or a greater number of extracts. Similarly, if you are investigating
the use of free direct and indirect speech and thought in novels of a particular period,
then for a short study you might choose one (or part of one if that novel is Joyces
Ulysses) and select extracts from it of again between 20 to 100 words or so or, for a
longer study, compare two or three novels. Depending upon the range of your data you
choose, then from your analysis it should be possible to generalise your findings.
An investigation into the structure of dialogue in plays will again involve you in choosing
at least one play for a short study, and, depending upon length, two or more for a longer
one. From these, you will need to select extracts for more detailed analysis or for worked
examples, with the length depending upon the point you wish to make, but usually
between 4-5 lines and 20-25.
Although much stylistic analysis is qualitative, one method of collecting and choosing
data which allows for analysis to be more quantitative is the use of corpora, which has
led to corpus approaches to stylistics being developed. Stylistic analysis is usually
undertaken on a small sample or range of texts, from which generalisations are made
and theories or models proposed. A corpus-based approach allows for that range to be
extended, by collecting texts into corpora and developing techniques for analysis. It has
alslo allowed for different approaches to stylistic analysis to be developed. Two main
approaches to stylistic use of corpus linguistics currently being undertaken are corpus
annotation and the analysis of collocation.
Corpus annotation involves investigating a particular linguistic feature by constructing a
corpus of texts and conducting a thorough analysis of this feature as it occurs within it.
The results of the analysis are normally inserted into the electronic version of the text as
tags or annotations. Such an analysis, because it is undertaken over a wide range of
texts, is thus arguably a more empirically sound procedure for discovering linguistic
phenomena, than choosing examples from which generalisations are made, and also
allows for a statistical analysis of frequency, distribution and so on. A corpus, once
annotated, is then available for studies which aim to replicate or further the research.
Such an approach, for example, was taken by Semino and Short (2004) in their work on
speech representation. This built upon the the system of classification for speech
presentation in the novel developed by Leech and Short (1981). A corpus constructed
of modern British English narrative texts was categorised, annotated and analysed in
order to test the theoretical model against real data. They found that not only did they
have to adapt the model in the light of their analysis, but also discovered new categories.
The second approach is to study literary effects in texts by using the evidence of
language norms in a reference corpus, an effect most commonly characterised by
deviance. A corpus can provide comparative information of what is normal and expected
in texts .and identify deviations from the norms of language use. For example, if a
particular word or phrase is thought to be exclusively literary, then its use can be
searched for in a corpus of non-literary texts to test this hypothesis. Following on from
the work of Firth and Sinclair, Louw (1993 and 2006) has developed a methodology for
analysing literary effects through his study of collocations, as has Zyngier (2005). This is
based upon the idea that certain words, phrases or constructions become associated
with certain types of meaning because of their regular co-occurrence. For example,
Sinclair (1987) discovered that the subjects of the phrasal verb set in are usually
unpleasant things, such as rigour mortis has set in; the rain had set in for the day. This
allows for unpleasantness to be evoked without using any other evaluative words or
phrases other than the phrasal verb. Louw argues that such a phenomena can only be
revealed computationally and is not generally accessible to our intuition. Only collocation,
he argues, can reveal the covert meanings of literary worlds and the feelings and
attitudes of those who create and inhabit them and unfettered by any condition related to
levels of language as outlined in 3.3 above.
Hoey (2005) has developed applications of collocation into a theory of lexical priming
which adds a cognitive dimension and also can be used to account for creativity in
language. This theory proposes that speakers and hearers associate meanings with
words not only because of their intrinsic meaning, but also because of the linguistic
contexts in which they are used to speaking and hearing them. In this way, words are
primed for certain uses and meanings. So, for example, set in has an intrinsic or primary
meaning of a process having started and continuing, but also a primed meaning of
unpleasantness. Corpora have made it possible to study the ways in which primings are
created, in ways that link corpora and the study of style with cognitive aspects of
language.
3.10.3 Analysing data
The methods that you use to analyse your data will depend upon the particular stylistic
features you wish to identify in your chosen text. Corpus methods notwithstanding, it is
still most common for stylistic analysis to be undertaken on a small, selective sample of
texts. For example, if you wish to undertake a stylistic analysis of a poem, then your
analysis will most probably be framed by the following investigations:
What can be said about:
1. the general patterns of grammar in the poem? This includes
identifying the general clause structure of the poem(s).
2. foregrounded patterns of grammar in the poem(s)? This includes
identifying sequences which are different from the basic grammatical
pattern.
3. sound and rhythm in the poem(s)? This includes identifying any
dominant metrical pattern, orthe use of free verse.
4. the way the poem(s) is/are set out on the page? This looks at the way in
which the words are set out on the page and any impact this may have
upon other levels of language.
5. vocabulary and word structure? This considers if there are any words or
word structures that are foregrounded; whether they deviate from the
norm, and if so, the ways in which they intersect with other levels of
language such as sound and metre.
6. simile, metaphor? This considers identifying any similes and metaphors
used, and the effect of such a use of language, including the use of
cognitive metaphors.
For example, for a short study, you could choose two poems by the same poet such as
Roald Dahls Revolting Rhymes or two poems, each by a different author, and apply
each of the methods above to each poem, to discover what, if anything, is similar or
different about them. Alternatively, you could contrast a poem from an earlier historical
period such as the seventeenth or eighteenth century with a late twentieth/early twenty
first one in order to draw some conclusions about any changes in structure and form that
may have occurred between the two. For a longer study, you would choose a greater
number of poems, say four or five of 100 to 150 lines.
If your study is concerned with prose fiction or narrative reporting, then your analysis will
most probably be framed by the following investigations:
1. What is the authorial voice? This involves considering in which person is
the narrative written, whether or not this changes, what kind of narration
is used omniscient or otherwise.
2. aspects of cohesion and deixis? This involves considering, for example,
how deixis and other aspects of cohesion are used in the text, the
ways in which it coheres, including the use of any imagery such as
similes and metaphors, and how they contribute to the creation of the
fictional world of the text.
3. point of view? This involves considering the perspective or perspectives
from which the narrative is written and the ways in which characters are
presented.
4. representation of speech and thought? This involves investigating the
ways in which speech and thought are used in the narrative and their
part in constructing characters and furthering narrative action.
5. narrative structure? This involves identifying aspects of a model of
narrative structure to see how far the narrative you are analysing
accords or differs from what is expected.
For example, for a short study you might wish to restrict your analysis to anything from
one to three of the aspects listed above as applied to two novels or examples of narrative
reporting, either by the same author or two different ones. For a longer study, you could
either broaden your analysis to include all five aspects, or widen your choice of novels or
narrative reporting to include more. If your study is concerned with dramatic texts, then
your analysis will most probably be framed by the following investigations:
In a dramatic text, what can be said about:
1. the context? physical, personal, cognitive, interactive and imaginary?
How is deixis used?
2. structure? what is the pattern of dialogue, and how is it structured in
terms of exchanges?
3. strategy? what conversational strategies are employed or violated?
4. characterisation? how do context, structure and strategy interrelate or
interweave to produce characterisation?
Again, the number of plays you choose and the length of extracts will depend upon the
length of your study. For a short study, extracts from one or two plays, either by the same
playwright or different ones would be sufficient, and you might wish to restrict your
analysis to anything from one to three of the aspects listed above. For a longer study,
you could either broaden your analysis to include all four aspects, or widen your choice
of plays to include more.
3.10.4 Sample projects
a) Undertake a stylistic analysis of a poem, in terms of: its clause structure; sound and
rhythm; graphology; vocabulary and word structure; cohesion and deixis and use of
simile and metaphor. For a longer study, you could compare and contrast two poems
from different periods of time, from the same poets or two different poets of the same age
or several poems by the same poet or of a particular period.
b) Undertake a stylistic analysis of a prose extract in terms of, for example: its clause
structure; authorial voice; aspects of cohesion and deixis and the use of any similes and
metaphors, and how they contribute to the creation of the fictional world of the text;
point of view; representation of speech and thought and narrative structure. For a more
in-depth or longer study, you could analyse these aspects in relation to a particular genre
such as detective or romance fiction, to novels by the same author, in order to identify
how any one or more of these categories accounts for the distinctiveness of a particular
authors style.
d) Undertake a stylistic analysis of a scene from a play,, paying particular attention to: its
creation of contexts; the structure of its dialogue; the conversational strategies involved
and how these all interrelate or interweave to produce characterisation. For a longer or
more in-depth study, you could analyse more than one scene taken from one play, or
scenes from different plays by the same playwright.
e) For a more corpus-based approach, you could test the application of a stylistic model
against real data, such as the deictic categories outlined in 3.x above in a corpus of
prose texts.
3.11 Where to find out more
For other works on stylistics, see: Carter and Nash (1990); Short (1996); Jeffries (1993);
Werth (1999); Wales (2001); Semino & Culpepper (2002) Simpson (1993 and 2004);
Stockwell (2004); Semino and Short (2004) and Watson and Zyngier (2006). Chapters on
stylistics also feature in general books on and encyclopedias of Applied Linguistics, such
as the one by Clark & McRae in Davies and Elder eds (2004)
Chapter 4 Critical Discourse Analysis
4.1 Introduction
So far, the approaches taken in the preceding chapters have been concerned with
analysing different aspects of spoken and written English. Chapters 1 and 2 have
outlined various issues and debates concerning spoken English; Chapter 1 concentrated
upon linguistic variation in spoken English and Chapter 2 upon the interactive nature of
speech, its context and the ways in which communication involves more than the words
which are actually spoken. Chapter 3 considered the application of linguistic approaches
to the analysis of written language, and literary language in particular. This chapter
considers a different approach to the study of language, that of Critical Discourse
Analysis (CDA). CDA is concerned with structures of power and equality that underlie all
acts of speech and writing, and is thus concerned with the politics of language. It
concentrates upon analysing the ways acts of speaking and writing and the practices
associated with them construct and shape reality and the world in which we live. It is
critical in the sense that it encourages readers to question assumptions and not take
anything for granted, analysing social issues as they are constructed in discourse and
also how discourse contributes to the construction of society. Its methods of analysis
draw upon sociological and critical theory in addition to linguistic theory in taking account
of relations of equality and power, including those of gender, ethnicity and social class.
CDA also differs from other aspects of applied linguistics discussed in this book, in that
its approaches and methods of analysis allow for the study of both spoken and written
language across all genres.
The question of what exactly is the nature of the relationship between language and
reality is a matter of much debate, and is discussed in more detail in TL:TE I.4 and II.2.
That the two are closely intertwined and in many ways inseparable is undeniable, but
how they intersect and act upon one another is unclear. Throughout the twentieth
century, work undertaken in several different disciplines: by linguists, most notably the
French linguist Ferdinand de Saussure (1916), the Russian linguist Mikail Bakhtin (1981),
by anthropologists such as the Americans Edward Sapir (1921)and Benjamin Whorf
(1956), by philosophers such as the German philosopher Ludwig Wittgenstein (1951)
and cultural theorists such as the French theorist Michel Foucault (1972), have all shown
that language plays an important part not only in structuring but also creating reality (If
you want to read more about this, see: TL:TE I.4 and II.2). If it is indeed the case that
language structures and creates, rather than reflects, reality, then the source of power
and control, of who has the authority to determine what is said, when and by whom,
shifts in a seismic way. The authority to construct reality shifts from some external God-
like authority who exists outside of and independently of society, to a community of
language users instead. Linguistic structures, as well as being susceptible to description
and analysis, can thus be studied in ways which attempt to account for underlying
structures of authority and control, ideology and power. It is this dimension of linguistic
analysis which CDA adds to the study of language in action.
This chapter provides a background and overview of CDA. Sections 1, 2, 3 and 4 discuss
major studies in the relationship between language, power, and ideology that have
influenced its development. Section 5 explains the major linguistic tools developed within
a critical linguistic methodology, namely those of transitivity and nominalisation. Section 6
moves on to an explanation of CDA as it is mainly understood today, which purports to
be both a theory and a methodology for research and analysis. Section 7 provides
students with guidance and advice on undertaking their own research into stylistics,
whilst section 8 provides suggestions for further reading.
4.2 Language, society and ideology
Ideology is commonly defined as a set of fundamental beliefs and ideas: that is, the
values and attitudes held by any given society or community which govern behaviour and
social interaction. Much of our understanding of ideology comes from the work of
sociological theory. In sociology, there are two related used of the term ideology. The
first is taken from the work of the German writers Karl Marx and Frederik Engels known
as Marxism, and the second from neo-Marxists such as Louis Althusser. Both related
uses of the word see ideology as a system of beliefs, ideas, speech and cultural
practices of a particular social group. This system of beliefs, ideas and practices disguise
or distort the social, economic and political relations between dominated and dominant
classes. Marxism stresses fundamental differences of interest between social groups that
give rise to conflict, thereby making conflict a common and persistent feature of society,
rather than a temporary disorder after which things go back to normal. Marx (1976)
claimed that the basis of all human organisation is economic, organised in two levels: the
base (or infrastructure) and a superstructure. The base in turn has two aspects to it: the
forces of production and the social relations of production. The forces of production
include the raw materials, technology and people of a particular society; and the social
relations of production refers to the social relations people enter into to produce goods,
such as manager, worker, and models of ownership, private and public, of the means of
production. The superstructure is made up of political, legal and educational institutions,
which are not independent of the base but are shaped by it. Ideology within Marxist
theory thus refers to the set of dominant ideas which emanate from the ruling classes
who control the institutions of the superstructure. Such ideas serve ultimately to justify
the power and privilege of the ruling class, and to hide from all members of society the
ways in which they are exploited and oppressed.
A clear example of this comes from the feudal age in Europe when the dominant
concepts were those of honour and loyalty, which appeared as the natural order,
celebrated in literature and implicit in superstructural institutions like law courts and
education. Similarly, in the capitalist age exploitation is disguised by the ideology of
equality and freedom, which appear to be desirable and natural concepts. Marxists argue
that this conceals the reality or fact that capitalism involves fundamentally unequal
relationships since workers have to work in order to survive, and all they can do is
exchange one form of wage subordination for another. Marx believed that many modern
societies contain basic contradictions which stop them from being sustainable. These
contradictions are based upon the exploitation of one group by another, for example, the
exploitations of serfs or peasants in feudal times, that have to be resolved, since a social
system containing such contradictions cannot survive without changing.
For Marxists, ideology is part of the superstructure rather than the economic base. Neo-
Marxists such as Louis Althusser, whilst agreeing with Marxists on what constitutes
ideology, nevertheless see it as something more fundamental. Althusser (1971) put
forward the view that ideology works by putting people into subject positions, in that all
social relations, including those of what Marx called the base or infrastructure are
subject or subordinate to some power (for example, the Queens subjects). Ideological
processes take place within various organisations which he calls ideological state
apparatuses such as the church, the legal system, the family and most of all, the
education system. Nicos Poulantzas (1976) went further in dividing the state system into
firstly, an ideological apparatus of church, political parties, unions, schools, mass media
and the family, concerned with the promotion and naturalisation of certain values and
beliefs and secondly, a repressive apparatus of army, police, tribunals and sometimes
even a government and its administration to enforce values and beliefs.
Sociological theories of the kind described above have concerned themselves with the
ways in which society is structured, and particularly in how social inequalities are
structured and maintained. What none of these theorists take into account, however, is
the part language, or discourse, plays in the construction of ideology.
4.3 Language, linguistics and ideology
A major influence upon the way the relationship between language and society is
perceived has been the shift in theories of how language functions in society, and
particularly how the meaning of words is determined. This shift is discussed in detail in
TL:TE I.4.4.3 and II 3, 4 and 5. Very briefly, prior to the twentieth century, it was assumed
that the meanings of words were stable and fixed. Such a view has a long history, dating
back to the work of Ancient Greek philosopher Plato, and continued into the writings of
the Bible. In chapter 2 of the Book of Genesis, once God had created all living creatures,
he brought them to Adam for him to give names to all the cattle, and to the fowl of the
air, and to every beast in the field. Names then, are treated as vocables, that is, sounds,
standing in a certain relationship to the things (persons, beasts, plants and so on) of
which they are names. Such a view assumes that the meaning of a word exists
independently and outside of the thing which it names.
4.3.1 Words and meaning
The relationship between words and their meaning has preoccupied various philosophers, initially
by calling into question how a word e.g. gold related to the object it named, that is, its referent
gold. They considered questions such as: the relationship between a word and what it stands for;
whether or not the relationship depended upon a natural connection of some kind; as something
which exists independently in the world, or as an idea in the mind. As linguistics came into being
as a discipline, linguists joined philosophers in subscribing to such a view of language. Neither
philosophers nor linguists, though, questioned the underlying assumption of the relationship
between words and their meaning, which was that the two existed independently of one another.
Under such a view, it is assumed that the individual words of a language name objects, and
sentences are combinations of such names. This picture of language assumes that every word has
a meaning, and that the meaning correlates to the referent or the word. Such a notion of language
and its relationship with meaning came to be increasingly challenged throughout the twentieth
century, most notably by Saussure (see also DL:DE 1.4. and TL:TE:II.3). He took issue with the
notion that a language, reduced to its essentials, is a nomenclature: that is, a list of terms
corresponding to a list of things: a picture-dictionary. He thought that asking what was the
relationship between words and objects was irrelevant, because the question itself assumes that
ideas exist independently of the words. Furthermore, such a view of the nature of meaning did
not make clear whether the name given to an object is a vocal (that is, made up of the sounds we
hear) or a mental (that is, something we imagine) entity. Saussure argued that it was both.
Together, the vocal and mental entity made up the linguistic sign. Consequently, a linguistic
sign is not a link between a thing and a name, but between a concept: that which is signified; and
a sound pattern; that which is the signifier. (See also: TL:TE I.4.3 and II 3.3. and also DLDE 1.4.3
and 7.7.3)
Saussure explained that the relationship between the signifier and signified was not, as had been
traditionally thought, a logical or necessary one, but one which is arbitrary (see DL DE section
1.2.1). . That is, the concept tree and the sounds /((((/ are not logically bound in any way, but
are a matter of convention. Saussures theory of signs, known as semiotics, has
revolutionised the way we perceive the relationship between words and meaning. This is
because such a view of the relationship between words and meaning shifts the allocation
of meaning from an external, Godlike figure outside humanity to humanity itself, and it
becomes the result of social negotiation. In practice meaning in language then, is not
natural but conventional. The relationship between a sound or symbol and what it
represents is fundamentally arbitrary, in that any sound or letter can be used to represent
any concept. Meaning results from shared common understanding amongst a community
of users as much as from anything else.
Take, for example, the ways in which dictionaries construct their definitions. Chapter 1
discussed how Samuel Johnson, in writing the first dictionary of English, was explicit
about the assumptions and prejudices upon which he based his selections and
definitions (see 1.3). Since that time, lexicographers (dictionary compliers) have
endeavoured to make the process of definition much more objective and neutral.
However, given that dictionary definitions are based upon examples of actual language
usage, there is a sense in which dictionaries, no matter how neutral they try to be,
inevitably not only reflect bias but also help to perpetuate it. For example, Hoey
(1996:150-165) analysed definitions of man and woman as defined in the Collins
COBUILD English Language Dictionary. The definitions for this dictionary were
constructed from exampes taken from an extensive corpus or database of contemporary
English. One of its fundamental principles was that all definitions should be based upon
how words were actually used in contemporary English. At the same time, another of its
principles was that there should be no reinforcement of sexist attitudes in and towards
language. Hoeys analysis of examples and definitions given for the terms man and
woman however, discovered that such reinforcement did in fact appear in the dictionary.
Take the following example:
A man is a human being of either sex (1.2)
Hoey (1996:158)
As Hoey points out, firstly, if this definition were strictly accurate, then there would be no
need to specify of either sex. Secondly, if the definition were correct, then man would be
synonymous with human being, and it would be possible to reverse the two noun
phrases without making them nonsensical, which is not the case:
A human being is a man of either sex.
The definition, then, implies that male human beings are sometimes taken as the only
human beings by users of the language, and female human beings are a sub-class of
male human beings. Two other definitions given in the dictionary also make the same
assumptions (Hoey 1996:158):
The man in the street is an ordinary person who is not especially rich or educated or
famous, and who is therefore considered to be a typical representative of public taste
and opinion. (2.1)
Modern man, primitive man, etc means all modern people, primitive people, and so on
considered as a group.
As Hoey says: It is worth noting the democratic implications of the phrase the man in
the street are replaced by sexual ones in the closely parallel phrase a woman in the
street. It would appear that men can in the English language go about in the street and
be regarded as normal, but that women go about the street at the peril of their
reputations(1996:158).
Hoeys analysis shows that, no matter how much lexicographers might aim not to
reinforce sexism in language or indeed or any other kind of prejudice or bias, no corpus
of real language use is value-free or neutral, but inevitably reflects and embodies bias.
The explanation Hoey gives for the bias he found was that the corpus chosen for the
COBUILD dictionary was predominantly that of written, published texts, with a
predominance of male over female writers. Consequently, he concluded that the corpus
strongly reflected the value-system of the male-orientated establishment.
In addition to redefining the nature of the relationship between words and meaning,
Saussure made a further distinction between two different axes of time in linguistic
study. The first is the diachronic axis, that is, the time line of the past, present and future,
and second is the synchronic axis, that is, language as it is used in the present. The fact
that language changes had proved something of a headache to linguists. By
distinguishing between the two different axes, Saussure suggested that we could base
the description of a language on how is used at a particular time, thereby making it
possible to ignore its history. This had the effect of focusing subsequent linguistic
endeavour upon the synchronic, that is, present day, aspects of language and ignoring
the diachronic, that is, developmental aspects of language. However, over long periods
of time in the history of a society, meaning is determined in a way which suits the needs
of that society, usually through serving the needs of the dominant, privileged groups. As
Fowler (1986:31) says: These dominant groups control the means of legitimating the
preferred systems of meanings schools, libraries, the media. Language thus becomes
a part of social practice, a tool for preserving the prevailing order. It does this not only
through propaganda, but also by inertia, the settlement towards stability and resistance
to change An example of legitimating preferred systems of meaning has been the
change in recent years in the meaning of the word gay. Until the 1960s, this word was
synonymous with happy or glad. It began to be applied to homosexual men as a
derogatory term, but as sexual liberation became a dominant theme in Anglo-American
society from the mid nineteen sixties onwards, the homosexual community appropriated
the term for itself, thereby legitimating its use as their own status in mainstream society
altered. Nowadays, the term gay is rarely used to mean happy, and then most likely by
the older generation. It is most commonly used as a synonym for homosexuality with
none of the derogatory overtones its use implied until very recently. Consequently,
language becomes not only a tool for preserving social order but also a means of
acknowledging change within it.
4.3.2 Language and reality
A further influence which has contributed to the altering of linguistic thinking about the
nature of the relationship between language and society is the work of the American
anthropologist Edward Sapir (1884-1939), and his student Benjamin Whorf (1897-1941),
known as the Sapir-Whorf Hypothesis (see: DL:DE 1.4.5). This hypothesis, and the
theories of linguistic relativity and linguistic determinism arising from it, resulted from the
work Sapir and Whorf undertook in making a record of American Indian languages,
particularly Hopi. They compared the grammatical features of American Indian languages
with that of standard average European (SAE) languages, including English, French,
German and other related ones. One of these comparisons was with the American Indian
language of Hopi and published by Whorf (1939). In his study, Whorf focused on verb
tenses, from which he concluded that the ways in which the Hopi Indian used tenses was
radically different from that of SAE, particularly the ways in which concepts of time and
space were talked about. Further more, he also argued that ways in which grammar was
expressed by Hopis related to Hopi culture, just as that of SAE related to European
culture. Take the example of talking about time. In western culture, we mark our verbs to
show whether we are talking about the past, present or future: for example, I wrote this
chapter, I write or am writing this chapter or I will write this chapter. Time then is
basically ordered into three separate periods which progress linearly from the past
through the present onto the future, as represented, for example, in historical time lines.
In Hopi culture, time is expressed very differently; rather than linear stages, it is viewed
as cyclical. Time is expressed through the stages of the human life cycle which goes
from birth, through childhood, adolescence, adulthood to maturity then death. Although
each of these stages are defined, they flow into one another rather than being markedly
defined as yesterday, today and tomorrow are. At each point, an individual is the same
person, although aspects of that individual may have changed, such as appearance and
certain characteristics. Consequently, in Hopi, an expression such as next week or
seven days which reflects a belief that each day is different, is not possible in Hopi,
where: return of the day [is] felt as the return of the same person, a little older but with all
the impresses of yesterday (Whorf in Carroll 1956:156). The Hopi language has no
tenses like European languages do, a difference which comes essentially from the
different world-views of the two types of culture.
Since the two types of culture, European and American Indian, have evolved in different
circumstances and with different influences, (such as the geographic and historic), then
so too, their experience of reality has been different. This worldview informs all cultural
aspects in addition to the linguistic, since, in Whorfs conclusions to his study, language,
culture and behaviour are constantly influencing one another. Sapir (in Lucy, 1992),
claimed that once the language habits of a group have been fixed, then its speakers are
at their mercy, in the sense that we cannot but help acquire these habits as part of
acquiring language. This has been taken to mean that we are passive victims of our
language, and that the language we speak determines our worldview; a kind of linguistic
determinism. In other words, our worldview is so encoded that we cannot think beyond it.
However, what the Sapir-Whorf hypothesis in fact claims is that perceptions and
concepts are encoded in language systems and reinforced by constant and
unquestioned use by their speakers, which does not necessarily mean that no
questioning or alternative to that system is ever possible. Sapir himself said that as our
awareness of language grows, so too must our questioning of it and how it is used. The
Sapir-Whorf hypothesis suggests that language use is not only determined by socio-
cultural factors, but also, and more importantly, determines our very ways of thinking.
This hypothesis has led to he theory of linguistic relativism, and is one which underlies
CDA.
4.4 Language as discourse
The term discourse has several meanings in linguistics. In traditional descriptive
linguistics, discourse is used to describe manifestations of speech and text to describe
manifestations of writing. In discourse analysis it has been used to describe the physical,
aural and visual manifestation of speech and writing, and the study of its management in
relation to structure and grammar, taking account of the interactive nature of speech and
writing between speaker and listener and writer and reader, as discussed in Chapters 2
and 3. There is, however, a further meaning of the word derived from social and cultural
theory, particularly from the work of theorists such as the French social theorist and
philosopher Michel Foucault. Here, the use of the term discourse applies to what is said
and written, but it also applies to the invisible structures and forces through which the
ideas - the ideology - of a society are shaped. These ideas are also subject to change,
and alter across time.
Discourse in this sense has two dimensions: the first is the visible aspect; that is, speech and
writing which is produced from within the structures of language we have at our disposal. The
second is the invisible one of the underlying assumptions and practices which govern our
production of language. Any discourse at any given moment in time is structured as much by the
assumptions about what constitutes it as a discourse as well as the boundaries of language itself.
As an example, Foucault cites institutional practices associated with education. He points out that:
Every educational system is a political means of maintaining or modifying the appropriateness of
discourses with the knowledge and power they bring with them (Foucault 1972 : 46).
For Foucault, discourses are to do with the ways sets of statements are systematically organised in
order to give expression to the meanings and values of an institution or a society. Discourses in
this sense thus define and delimit what it is possible to say or do and not possible to say or do with
respect to the concerns of particular institutions and societies. To take an example cited in 0
Halloran (2003:12), different religions have their own discourses which delimit explanation of
natural behaviour. Catholicism now accepts that the universe began with the Big Bang
(scientific discourse) but believes that the Christian God initiated it (a mixture of scientific and
religious discourse).
Drawing upon social and cultural theories of discourse such as those of Foucault has
given rise in linguistics, and especially within CDA, to different ways of interpreting the
terms discourse and text. So far, the term discourse has been used to describe either the
manifestation of speech and its organisation (discourse and conversation analysis), or as invisible
structures that shape what is said (social and cultural theory). A text is then anything which is
actually produced by language, both spoken and written. In cultural and social theory this notion
is taken even further, and the term text is used to describe any manifestation of human
communication. As the cultural theorist Fornas (1995) describes them, texts are:
... signifying practices (that) combine meaningful signs into complexly structured and ordered
symbolic units. These are referred to as texts, whether they consist of spoken or written words,
images, sculptural or architectural forms, musical sounds, body movements or any combination of
these or other symbolic entities.
Fornas (1995:149).
Consequently, the term text may be extended beyond the written word to include any
visual manifestation in the material world - images, the way we furnish our houses, the
clothes we wear, even our bodies, can all be read as texts. As CDA is primarily
concerned with linguistic analysis, then the texts with which it is concerned are those
produced by speech and writing. Unlike traditional discourse analysis however, in CDA
the use of the term does not differentiate between speech and writing, calling
manifestations of both texts. Within CDA then, discourse (or discursive practices) refer
to the underlying structures that shape what is said: pedagogic discourse, religious
discourse, political discourse and so on, and text to a manifestation or realization of that
discourse: a lesson, a sermon, a speech and so on.
These uses of the term text and discourse have been applied to the analysis of language
by two different recent linguistic traditions: firstly, the work of critical linguists undertaken
during the 1970s by Fowler (1979) and Hodge and Kress (1979). From the 1980s
onwards this work has been subsumed and developed by critical discourse analysts,
most notably Norman Fairclough (1989), a collection edited by Rosa Carmen Caldas-
Coulthard and Malcolm Coulthard (1996) Fairclough and Wodak (1997), Chouliaraki &
Fairclough (1999), and van Dijk (2001).
4.5 Critical linguistics
Critical linguistics resulted from the publication of two seminal books both published in
1979: Language and Control (R Fowler, R Hodge, G. Kress and T Trew) and Language
as Ideology (G. Kress and R. Hodge). Drawing upon sociological and critical theory as
outlined in section 3 above, critical linguists took issue with a purely objective model of
linguistic criticism which claimed that the meaning and value of a text can be derived
solely from its linguistic construction; that is, the words on a page or the sounds made by
speech. They argued that although linguistic structures are themselves objective: their
significances in discourse cannot be read off automatically from the text: a semiotic
assessment in relation to cultural factors is required (Fowler 1986:169). That is, that
communication results not only from what is said or heard, or written and read, but also
from understanding or reading cultural and social norms associated with the acts of
speaking, listening, reading and writing (see also Chapter 2). Consequently, Fowler
describes the central concerns of critical linguistics as the ordering of experience and
with the mediation of social relationships and values(1991:70). Critical linguistics draws
upon cultural theory outlined above to extend the concept of text to include every kind of
writing; newspapers, adverts, leaflets and so on, as well as novels, plays and poetry, and
also extends to spoken speech. Unlike pragmatics then, which focuses mainly upon
spoken speech used in context, or stylistics which focuses mainly upon written,
specifically literary texts, critical linguistics, and CDA, include the whole spectrum of texts
that speech and writing make possible. Such an activity requires a specific model of
linguistics, which for critical linguistics is drawn from systemic-functional linguistics (see:
Halliday 1971&1973). Systemic functional linguistics is concerned with how language is
used in real life, and with how linguistic structure is related to communicative function
and social value. At its heart is the notion of function, a concept used in a general,
global sense. All language performs three functions at the same time: the ideational,
the interpersonal and the textual (see also DL:DE 7.3.2).
The ideational function is concerned with the content of language, as it is used to
express our experience of the external world the inner world of our own consciousness.
The interpersonal function is concerned with language as the mediator of role, as it is
used to express our personalities and personal feelings and our interaction with other
participants in communication. The textual function has the enabling one of creating
text, and is concerned with language in operation as opposed to or as distinct from
strings of words or isolated sentences and clauses: It is this component that enables the
speaker to organise what he is saying in such a way that makes sense in the context and
fulfils its function as a message (Halliday: 1973:66).
As Halliday (1978) makes clear in a later work, the ideational, interpersonal and
textual functions are sets of social options, constrained by social and cultural practices,
rather than areas of voluntary or privileged personal choice. The three functions provide
a way of classifying linguistic structures according to their communicative roles.
Consequently, linguistic details of syntax, vocabulary and so on are conceived of
functionally. That is, they are considered not only as formally different kinds of structure,
but also as the kinds of structure they are based upon the particular jobs or functions
they perform. Approaching the analysis of language in a systemic-functional way has
given rise to a set of linguistic tools that go beyond those normally associated with
grammatical analysis, and include transitivity, syntactic transformations of the clause,
lexical structure and modality. These tools have been described in various works (e.g.
Fowler (1991; Simpson 1993). It is worth, however, explaining transitivity and modality in
more detail, since they exemplify the concept of functionality in language.
4.5.1 Transitivity
The ideational function of language, that is, the content of what is said or written and the writers
or speakers experience that content embodies, is best illustrated by ways of semantically
categorising clauses that reveal speakers or writers experience or point of view. A clause is
essentially a basic unit of syntax which determines the position and sequence of elements (words
and phrases), rather than being concerned with propositional meanings and functions.
Nevertheless, meanings are encoded in clauses because of the way different types of processes
are represented in language. There are three key components of processes:
1. The process itself, typically realised grammatically as the verb phrase. For example:
hit, sees, broke.
2. The participant(s) associated with the process, typically realised grammatically by
the noun phrase. For example: John, she, the vase.
3. The circumstances associated with the process, typically realised grammatically by
the prepositional and adverb phrases, for example, in the room, at breakfast, over the
wall.
Simpson (2004:22-26) identifies six types of processes: 1. material, 2. mental, 3.
behavioural, 4. verbal, 5. relational and 6. existential.
1. Material processes are those concerned with doing, and happen in the physical
world. In this process, there is always an actor or actors, and usually, though not always,
a goal. For example;
(1) John hit the ball.
Actor Process Goal
(2) The vase broke.
Actor Process
2. Mental processes are those concerned with sensing. Unlike material processes
which are concerned with the physical world, mental processes are concerned with
consciousness, involving cognition (encoded in words such as understanding), reaction
(such as despising) and perception (such as tasting).
(3) Lucy puzzled over the problem (cognition)
Sensor Process Phenemenon
(4) Harry cant stand fish (reaction)
Sensor Process Phenomenon
(5) Cathie saw the dress (perception)
Sensor Process Phenomenon
3. Behavioural processes are those which embody physiological actions such as
breathe and sneeze. They can also be expressed as states of consciousness as in cry
or laugh, or processes of consciousness as forms of behaviour, such as dream, or
worry. The participant in behavioural processes is the Behavor, the conscious entity who
is behaving:
(6) She gasped at the sight.
Behavor Process Circumstance
(7) Joan laughed at the clown.
Behavor Process Circumstance
(8) The dog barked..
Behavior Process
4. Verbal processes are processes of saying. The participant roles associated with this
process are the Sayer, who is the producer of speech, the Receiver, who is the entity to
which the speech is addressed, and the Verbiage, which is what gets said. For example:
(9) John told the joke to his friends
Sayer Process Verbiage Receiver(s).
(10) The Judge announced the sentence to the court.
Sayer Process Verbiage Receiver.
Within this category, the verbiage can apply to the content of what is said, as the joke in
example (9) above, or the name of what is said, such as the sentence in (10).
5. Relational processes are those which are to do with being, in the specific sense of
establishing a relationship between two entities. The process is almost always taken from
forms of the verb be (or a very small number of similar verbs such as seem or become),
and the participant roles those of the Identifier and identified. For example,
(11) The show is on all evening.
Identified Process Identifier
(12) Catherines car is a Volvo.
Identified Process Identifier
(13) Harrys dog was in the park.
Identified Process Identifier
6. Finally, existential processes assert that something exists or happens. Typically,
they include the word there as a dummy subject, as in There has been a theft or Has
there been a fight?. They normally contain only one participant role, realised in the
examples as a theft and a fight. Existential processes take us back to material ones, in
that both can answer a question such as what happened? In a material process, an
actor and a goal would be identified, with the participant role replaced by a process. So,
There has been a theft would be something like John stole In the existential version, the
actor and goal disappear, to be replaced by the Existent which is filled by a nominalised
element.
Applying the transitivity model to written texts, especially when analyzing media language
such as newspaper headlines and articles, can be very revealing of ideology. For
example, Trews seminal article on media language (1979) analyses the news coverage
of a civil disorder event in pre-independent Zimbabwe. The headlines and opening texts
of two British newspapers of 2 June 1975 read:
a) POLICE SHOOT 11 DEAD IN SALIBURY RIOT
Riot police shot and killed 11 African demonstrators.
(The Guradian, p.1)
b) RIOTING BLACKS SHOT DEAD BY POLICE
Eleven Africans were shot dead and 15 wounded when Rhodesian police opened fire on
a rioting crowd. (The Times, p1)
In example a), the Guardian headline and first line uses an active construction, a
standard SVO pattern, making the ACTOR, the police, the first element in the clause with
the GOAL, the 11 dead or African demonstrators, appearing at the end and the
PROCESS, shoot, shot and killed. This places considerable emphasis upon the agents
involved in the process. By contrast, the Times employs a passive construction, inverting
subject and object, actor and goal, thereby placing the GOAL, rioting blacks, 11
Africans, at the start of each construction, and thus in a position of prominence. The
ACTOR element, by contrast, is placed in a less prominent position at the end.
Furthermore, in the Times first sentence, agency is actually deleted from the first clause
- Eleven Africans were shot dead and wounded - , and can only be identified by
inference from the second. Trew contends that the effects of the passivity and agency
deletion in the Times are to shift attention away from who did the shooting onto the
victims. In other words, the two messages, whilst reporting the truth, are slanted in
crucially different directions. This slanting can be said to correspond with the political
orientation of the two newspapers; The Guardian reflecting the political left and
The Times the political right. An analysis and interpretation of this kind, based upon
extrapolating from textual analysis to questions of political bias, essentially summarises
the critical linguistic method.
Take a further example, which was the shooting of a suspected terrorist following the
July 2005 terrorist bombings in London, reported by the BBC and the Guardian. The
BBC purports to take a politically neutral stance and the Guardian is associated with the
political left.
The first is:
a) POLICE SHOT BRAZILIAN EIGHT TIMES
The man mistaken for a suicide bomber by police was shot eight times, an inquest into his death
has heard. (BBC News On-Line 25 July 2005).
The second is:
b) SHOT MAN WAS NOT BOMBER POLICE
The man shot dead in Stockwell tube station yesterday was not connected to the attempted
bombings of London on July 21, police said tonight.
(The Guardian On-Line 23 July 2005).
The lexical choices of the two texts are different and clearly significant: the Brazilian of
the BBC becomes Shot Man in The Guardian. These lexical differences are
accompanied by differences in the transitivity structures of the two reports:
POLICE SHOT BRAZILIAN EIGHT TIMES
Actor Process Goal
The man mistaken for a suicide bomber by police was shot eight times,
Goal Actor Process
an inquest into his death has heard.
Receiver Verbiage process.
Although the headline uses an active SVO construction, the lexical choice and
foregrounding of the dead mans nationality as one which is not British is offered as
something of an explanation for the shooting. The London bombers were assumed at
first to be foreign terrorists, and marking the shot man as foreign can thus be interpreted
as some kind of justification for the shooting. However, the statement of the exact
number of times the man was shot emphasises or implies a literal overkill, since it does
not usually take that number of bullets to kill someone. The headline is followed by a
passive construction, inverting subject and object, actor and goal, thereby placing the
GOAL, which is further elaborated as the man mistaken for a suicide bomber in a
prominent position. The fact that he was shot eight times is repeated, again reinforcing
that excessive force was used. The reporting is from accounts made a day after the
shooting, when it became evident that the man shot by police was not a suicide bomber
as had first been reported. Placing the first clause construction of GOAL ACTOR
PROCESS before the reporting clause emphasises the fact that not only did the police
use excessive force, but that such use was unjustified by the outcome, since the man
was not after all a suicide bomber as had first been thought.
The Guardians headline does not mention nationality, instead focusing on the shooting:
SHOT MAN WAS NOT BOMBER POLICE
Verbiage Sayer
The man shot dead in Stockwell tube station yesterday was not connected to the attempted
bombings of London on July 21, police said tonight.
Verbiage Sayer
The Guardians headline and accompanying first line summarises and quotes the
admission made by the police that they had made a mistake. Actual admission of the
mistake is implied, not stated. The Guardian headline does not offer the excuse of
mistaken nationality as does the BBC example, focusing instead upon the fact that the
shooting was an error and a result of at best, over reaction, and at worst, panic. Both,
however, are unequivocal in their condemnation of the action itself. The Guardian report
allows the words of the police to speak for themselves in condemnation of the attack,
whereas that of the BBC presents a more ambivalent stance, highlighting or offering
foreignness as a possible excuse for the shooting, but immediately condemning the
excessive use of bullets.
4.5.2 Modality
Whereas transitivity serves an ideational function, modality serves an interpersonal one:
that is, it expresses comments, attitudes and evaluations. Modality is the way in which
speakers and writers use language to comment or express attitudes and beliefs and
present their own point of view. t is a grammar of explicit comment, and it signals the
varying degrees of certainty we have about what we say or write, and the sorts of, or
degree of, commitment and obligation we express. The way in which modality functions
grammatically is realised most commonly through modal auxiliary verbs, though lexical
verbs, adverbs, adjectives, intonation and even body language can also be modal.
Fowler (1986) identified four different aspects of modality: truth, obligation, desirability
and permission. Truth indicates or implies a commitment to the truth of any proposition
uttered, or a prediction of the likelihood of an event taking place. Modality appears where
the speaker or writer is expressing an opinion on what will, might, could or should
happen. Truth modality varies from a scale of absolute confidence signalled by will to
degrees of lesser certainty signalled by could. It can also be indicated by adverbs such
as certainly and modal adjectives such as likely or unlikely. An example of absolute
confidence is:
Strikes end but airport misery will last for days.
The Times, 13 August 2005
An example of lesser confidence is:
He said the issue could only be resolved by the courts and he predicated that
passengers would bring a case against BA.
The Times, 13 August 2005
Obligation is where speakers and writers express an opinion on what ought to happen,
expressed by ought, should and ought to, as in the following example:
Although the frustrated holiday makers marooned west of London will find it difficult to do
so, there should be sympathy with the workers cause.
The Guardian, 13 August 2005
It is important to note that English, unlike other languages which have future tenses,
has only a modal way of talking about the future, that is, with will. This means that all
future statements are modal, whilst statements about present or past can be categorical.
In the future then, there is no choice about the modality.
Desirability is where the speaker or writer shows approval or disapproval of the state of
affairs that is communicated. This use of modality is widely used in the Press, particularly
in editorials and especially in tabloids and broadsheets with right wing tendencies.
Modality of this kind may be expressed through a range of evaluative adjectives and
adverbs as well as some lexical verbs (e.g. wish, want). For example,
To be sacked by megaphone is callous, even for a part of the service sector well known
for low pay, insecure job tenure and poor career prospects.
The Guardian, 13 August 2005
Permission is where the speaker or writer gives permission to do something, signalled by
may, can and shall. Here, the auxiliaries may and can indicate a degree of cautiousness
or hedging,, as oppoed to shall which is more defibnite, s can be illustrated by the fairy
godmothers utterance to Cinderella that: You shall go to the ball!
Modality is also an important feature in the study of narrative, where Simpson (2004) in
his work on point of view in fiction points out that the degree of modality expressed by the
writer or reader can be categorised as degrees of shading, from positive, through
neutral to negative.
Positive shading is where the writer, narrator or speaker tell you what s/he thinks and
believes by foregrounding expressions, beliefs, opinions and obligations. Modality which
expresses desire and obligation is known as deontic modality: for example, I ought to
visit my Gran or I need a drink. This is the most common use of modality, which
underpins many first and third person narratives. It also underpins the writing of
newspaper Editorials and comment and analysis columns.
Negative shading is where the writer or narrator is uncertain about events and other
characters motivations and is often expressed through structures based upon
perception: as if, it seemed, it appeared to be and so on. This kind of shading is marked
by epistemic modality. This foregrounds a narrators, writers or speakers efforts to
interpret and make sense of what is read or heard, and is often characteristic of Gothic or
existentialist styles of narrative fiction. For example, Maybe the sound came from, or
his horse must have thrown her off
Neutral shading is where the writer or narrator gives no evaluation or interpretation and
thus there is a complete absence of either deontic or epistemic modality. Rather, it is
typified by categorical assertions with the speaker or writer making no evaluative or
interpretive statements. For example, My name is Bill or Bus hits 10 cyclists. This kind
of modality is rare in narrative, but is often to be found in news reporting which aims to be
factual and neutral.
The linguistic tools derived from critical linguistics such as transitivity and modality
described above show how linguistic structures encode both ideational and interpersonal
functions. Tools such as these, together with those which perform a textual function
such as the pragmatic categories of discourse and conversation anlaysis see (2.x) and
stylistic ones such as cohesion and deixis (see 3.x) are all ones upon which CDA draws
in its methods of describing and interpreting both spoken and written texts. Where CDA
differs from other methods or frameworks for the analysis of spoken and written
discourse, it is in its emphasis upon the underlying ideologies of the social and cultural
context within which description and interpretation occur, and which, it is argued, also
shape that description and interpretation.
4.6 A Framework for Critical Discourse Analysis
Taking account of sociological and cultural theory (4.2), changes in theories of language ( 4.3 and
4.4) and developments in functional grammar (4.5 above), have combined to allow insights into
the ways in which language functions in society. They show how points of view, world views and
beliefs are not only socially constructed but also grammatically encoded in language, in ways
which often reveal unequal relationships based upon authority and power. Critical linguistics and
the grammatical tools associated with it (see 4.5 above) have added to the analysis of spoken and
written texts as outlined in chapters 2 and 3 and also extended the range of text studied to non-
literary ones such as newspaper articles and adverts. They also take account of the social element
of how language represents the world. In CDA, this aspect of analysis is taken further. It is
argued that language, as well as playing a part in constructing (or misrepresenting) reality, is also
the primary medium of social control and power, and is therefore fundamentally ideological. For
practitioners of CDA, ideology is all-pervasive in language, and consequently it is with revealing
the ideological nature of language with which they are most concerned.
CDA integrates linguistic analysis of the kind described above and in other chapters opf this book,
with social and cultural theories in order to expose the ideological assumptions and relations at
play in language, which are essentially to do with power. Although critical linguistics recognises
that texts are essentially a form of social practice, the methods of analysis associated with it
remain grounded in the text, as the section above has shown. Critical discourse analysis takes
matters one step further, arguing that the text is itself part of wider discourse practice within
which it is situated. This discourse practice is, in turn, located within wider socio-cultural
practices. Consequently, interaction between all three layers text, discourse practice and socio-
cultural practice - that is, between all aspects of the social use of language, becomes the subject
of analysis. CDA then, acts as both a theory and a method of analysis. As Fairclough argues:
It is not uncommon for textbooks on language to have sections on the relationship between
language and society, as if these were two independent entities which just happen to come
into contact occasionally. My view is that there is not an external relationship between
language and society, but an internal and dialectical relationship. Language is a part of
society; linguistic phenomena are social phenomena of a special sort, and social phenomena
are (in part) linguistic phenomena.
Fairclough: (2001:19)
Figure 4.1 below shows this three-layered model of discourse, which presents discourse as
involving three dimensions at the same time:
Fig 4.1 A three-dimensional model of discourse (from Fairclough 2001:21).
Corresponding to the three dimensions of discourse, Fairclough distinguishes three
dimensions, or stages, of CDA:
1. Description is the stage which is concerned with formal properties of the text.
2. Interpretation is concerned with the relationship between text and interaction -
with seeing the text as the product of a process of production, and as a resource
in the process of interpretation; notice that I use the term interpretation for both the
interacational process and a stage of analysis.
3. Explanation is concerned with the relationship between interaction and social context
with the social determination of the processes of production and interpretation, and their
social effects.
(Fairclough2001:21)
Linguists interested in CDA consider language an important tool in the production, maintenance
and change of social relations of power. Their aim is to increase language users consciousness of
how language contributes to the domination and control of some people by others. It is a means
of: helping people to see the extent to which their language does rest upon common-sense
assumptions and the ways in which these common sense assumptions can be ideologically shaped
by relations of power. (Fairclough:1989:). For the dimensions of description and
interpretation, CDA draws upon linguistic tools that have already been developed as part of
other traditions, such as those of pragmatics and discourse analysis outlined in chapter 2, those
associated with stylistics as outlined in chapter 3 and those of critical linguistics such as
transitivity and modality summarised in 5.5 above. Where CDA differs from these traditions is not
only the linking but embedding of the descriptive and interpretative dimensions within the
evaluative one. It is this interface between linguistic interaction, the social contexts within which
it occurs and underlying ideologies of those contexts which is the most controversial aspect of
CDA (see: Widdowson 2004).
For example, common sense assumptions are evident in the conventions for a traditional type of
consultation between doctors and patients where authority and hierarchy are treated as natural,
and such assumptions are thus embedded in the forms of language used. Another example is in the
relationship between the police and witnesses. Take the following example, which is part of an
interview in a police station involving a witness to a burglary and a policeman. The policeman is
filling out a form and his questions follow the order on which they appear on the form:
P: You say you saw them as they left the shop?
W: Yeah, I saw their faces. There were two of them. I think they were
P: What age would you say they were ?
W: Dunno. One was yunger than the other. Twenties I guess. Mid twenties. The otherFortyish?
I think one of them was carrying
P: Height?
W: medium, Id say. About six foot, maybe less. One was slightly shorter than the other
P: Six foot. Five foot ten. Did you notice the colour of their hair?
W: I think the shorter one had fair hair. Look, how longs this going to take? Ive got to ...
P: Not much longer, no. What about the other one?
W: Cant say, Im sorry.
P: Short? long?
W: Short.
P: What about their clothes?
W: One was wearing a dark, heavy jacket
P: Like a donkey jacket?
W: Yeah.
(authors data)
A discourse or conversation analysis of this exchange would emphasise the verbal utterances that
make up the utterances and the structure of the interaction. Focusing upon, for example, turn-
taking, adjacency pairs, presupposition and politeness (see: 2.5).
A critical discourse analysis of this exchange emphasises what the conversation reveals about the
relationship between the participants and the social conditions which determine the properties of
the discourse. The interviewer, who is the policeman, is in control of the way the interview
develops, and reduced questions such as height? are typical of someone filling out a form. The
witness is forthcoming with information, but the policeman will only deal with each bit as it is
determined by the order on the form, not in the order given by the witness. The witness signals her
frustration by asking: how longs this going to take? The policeman, though, has his task to
perform and the witness, as a law-abiding citizen, complies with the requirement of that task even
if it means being late for a previous engagement. It could also be argued that the sensitive nature
of the situation is also contained or framed by the norms of form-filling. Filling in the form
structures the questions the policeman asks, who ignores any information volunteered by the
witness if it is not relevant to the particular question on the form, even if it is relevant to the
incident as a whole.
In analysing this conversation in this way, it can be seen that the relationship between the police
interviewer and the witness is unequal. The way in which the interview is carried out follows
social conventions based upon the nature of the relationship between the police and members of
the public in our society. In this way, it can be said that social conditions determine the properties
of the discourse. When looking beyond the text itself we can look at the way we produce and
interpret texts and see how these processes are socially shaped and relative to social conventions.
For example, there is complete absence of any acknowledgement by the policeman when the
witness gives information. In a conversation between friends this would not be acceptable,
whereas in this case it generally is.
In terms of the three-dimensional model shown in fig 4.1 above, consider further the example of
the interaction between the policeman and a witness given above:
Text: this involves a linguistic analysis of the actual speech exchange and its
characteristics in terms of conversational properties, politeness phenomena, narrative or
argument structure, characteristics of grammar and accent and so on.
Processes of production and interpretation: this involves considering the type of
discourse produced in the interaction: argument, small talk, political discussion and so
on. Identification of the ways the exchange fits in with genres or speech events.
Social conditions of production and interpretation: this involves consideration of how
the exchange derives from, reinforces or challenges expected relations between the
police and the public and law enforcement as an institution.
As has already been mentioned, linguistic analysis of actual speech exchanges or written
texts within a CDA framework draws upon ones established in other areas of linguistic
activity. It has concentrated more upon developing frameworks or categories for the
remaining two dimensions: processes of production and reception and the social
conditions within which they occur.
5.6.1 Schema theory and members resources
Processes of production and especially of interpretation have been the focus of much
recent linguistic study (see also Chapter 3). Interpretation is generally arrived at through
a combination of what is in any given text and what is in the interpreter. One influential
theory in trying to account for the relationship between the two is schema
theory, derived from cognitive linguistic theory, as proposed by Schank and Abselon
(1977) and developed by Sperber and Wilson (1986). Schema, (or in the plural
schemata), is a mental representation of a particular type of activity, representing models
of social behaviour. Frames represent the entities that occur in the natural and social
world. A frame is a representation of whatever can count as the topic, subject matter or
referent within an activity. Scripts refer to chunks or bits of knowledge which describe a
predetermined, stereotyped sequence of actions that defines a well-known situation
(Schank and Abselon 1977:41). Although scripts refer to what we already know, they are
also subject to modification and change as our range of experience of situations, well-
known or otherwise, grows. A script-based framework of the kind proposed by schema
theory is a way of accounting for how we can understand texts without having everything
within them made explicit linguistically.
Going back to the example of the policeman and witness, applying schema theory makes
the schema being enacted here that of police activity, and the frame is the questioning of
a witness. Scripts represent the subjects who are involved in this activity, the policeman
and the witness, and their relationship with one another. Scripts typify the ways in which
specific classes of subjects behave in particular situations, and explains why the witness
accepts the absence of acknowledgement in this schema but not in another; s/he is
behaving according to the script we set out for ourselves on how a police interview
should be conducted, based on our previous experience real, imagined or hearsay.
Fairclough (2001) absorbs and expands upon schema theory in his notion of members
resources or MR. . Like schemas, MR are the he interpretative procedures in which any
of us engage in any communicative act. MR are essentially interpretative procedures
which help to generate interpretation: ..from the point of view of the interpreter of a
text, formal features are cues which activate elements of interpreters MR, and
interpretations are generated through the dialectical interplay of cues and MR
(2001:118). That is, interpretation results from interaction between the structural content
of any given text and the ways in which this structure activates the readers or listeners
store of interpretative procedures. Fairclough shies away from using the term
background knowledge to describe MR, on the grounds that the term is too restrictive,
and misses the point that many of the assumptions we make in interpretation are
ideological, thus making knowledge a misleading term. MR refers to the representation
each one of us has stored in our long-term memory of, for example, shapes of words,
grammatical forms of sentences, typical narrative structures, the expected sequence of
events in a particular situation and so on. We draw on this MR to both produce and
interpret texts. MR are cognitive in the sense that they are in peoples heads but they are
also social in that they have social origins, being dependent upon social relations and
struggles out of which they were generated. Fairclough (2001: 119-20) identifies four
levels of textual interpretation: surface of utterance, meaning of utterance, local
coherence and text structure and point.
Surface of utterance refers to the processes by which we convert strings of sound or
marks on a page into recognizable words, phrases, sentences and utterances. The
aspect of MR upon which we draw in doing this is commonly known as knowledge of the
language; that is, its phonology, vocabulary and grammar.
Meaning of utterance refers to that aspect of MR concerned with semantic aspects
such as the representation of the meaning of words, including implicit ones. It also
includes drawing upon pragmatic conventions associated with speech acts.
Local coherence refers to the ways in which MR infer connections or cohesive relations
between utterances within a particular part of a text. This is possible even where formal
cohesive cues are absent, where we make implicit assumptions which are often of an
ideological nature.
Text structure and point is to do with working out how a whole text hangs together.
Text structure involves matching the text with one of a repertoire of schemata or
representations of characteristic patterns of organization associated with different types
of discourse (2001:120). For example, once we have worked out that we are taking part
in a telephone conversation, then we can expect the conversation to follow a particular
structure of greeting, establishing a topic, changing a topic, closing off the conversation
and saying goodbye. The point of a text is the overall interpretation of the text as a
whole. For example, a telephone conversation may have included an exchange of
conversation on the participants well-being, recent activities and so on, but the overall
point of the telephone conversation was a request to go out. It is this overall point
which is stored in long term memory and available for recall.
Faircloughs categorization of MR is more specific than that of schema theory, taking into
account or incorporating linguistic processing as part of the process of interpretation, in
addition to the levels of pragmatic and schematic processing. To return to the example of
the policeman and witness, the two participants interpret what is actually being said in
terms of phonology and semantics, surface and meaning of utterance, make sense of
what is implied by local coherence and how to manage the exchange through text
structure and point. This is all in addition to interpreting aspects of the context within
which the exchange is taking place.
In addition to the four levels of interpretation, Fairclough also includes contextual as well
as textual aspects of interpretation, which he identifies as situational context and
intertextual context. Situational context refers to both the physical cues in which an
exchange takes place - its situation, what has previously been said and so on - and to
the ways in which MR interpret these cues: How participants interpret the situation
determines which discourse types are drawn upon, and this in turn affects the nature of
the interpretative procedures which are drawn upon in textual interpretation (2001:121).
Intertextual context refers to the ways in which participants in any given exchange make
assumptions about it based upon their experience of previous exchanges to which the
one in which they are engaged is connected, similar to scripts in schema theory. These
assumptions then determine what can be given and does not need to be re-stated as
part of common experience, what can be alluded to, what disagreed with and so on.
Situational and intertextual contexts thus correspond more closely with the schemas,
frames and scripts of schema theory, whilst the four levels of textual interpretation add a
further dimension. For interpretation to take place, all six levels of the interpretative
process interact with one another in what are clearly complex, as well as dynamic, ways.
Fairclough offers a further diagrammatic example of how interpreters arrive at
interpretations of the situational context, and the way in which this determines decisions
about which discourse type is the appropriate one to draw upon.
Fig 4.2 Situational context and discourse type (Fairclough 2001:122)
In the lower half of the diagram, on the left hand side, are four questions which relate to
four dimensions of the situation: what is going on, who is involved, what the relationships
are and what the role of language is in the situation. On the right hand side are the four
dimensions of discourse type, in the CDA sense of a set of underlying conventions
belonging to some particular order of discourse. If we take again the example of the
policeman and a witness given above, then we can ask these questions:
1. Whats going on? The activity type here is interviewing a witness, and the overriding
purpose is the elicitation of information and accounts of the alleged crime, including its
documentation.
2. Whos involved? Subject positions are multi-dimensional and alter according to situation;
in this case, as it is an interview, the subject positions are that of interviewer and
interviewee. A second dimension is that the institution ascribes social identities to the
subjects who function within it, here a policeman and a member of the public who is
also a witness and a likely victim. A third dimension is that different situations have
different speaking and listening positions associated with them speaker, addressee,
hearer, spokesperson and so on. In this example, speaker and addressee roles alternate
between the policeman and the witness.
3. In what relations? Subject positions here are looked at more dynamically, in terms of
power, social distance and so on. In this case, the focus is on the nature of the relationship
between the police and the public, with the policeman anxious to get his form completed,
and the member of the public wanting to get away and back to schedule.
4. Whats the role of language? Language here is being used in an instrumental way as part
of a wider institutional and bureaucratic objective. Language here determines genre that
of an interview and its channel as spoken or written. Here, the form is being filled out by
the policeman, complicated by an interview and is indicative of the degree of control
which the police exercise over all aspects of the case. The information given by the
witness is mediated and checked by the police; only then is it valid.
The first three discourse types listed on the right hand column of the figure are conventionally
associated with a particular type of situation. The fourth dimension, connections, includes the
ways in which texts connect and are tied to situational contexts and also ways in which
connections are made between parts of a text, both of which can vary between discourse
types. In making judgements about these connections, we draw upon elements of MR
particular to a discourse type. A CDA approach to analysis involves consideration of
social conditions of production and interpretation which is again related to three different
levels: the level of the social situation (the immediate environment), the level of the social
institution, and the level of society as a whole. Fairclough suggests that these social
conditions shape the MR we bring to production and interpretation which shape the way
in which texts are produced and interpreted. The determination of social order in terms of
society and institution is also a matter of interpretation. Observable features of the
physical situation, and the text which has already occurred, that is, the interview between
the policeman and the witness, do not, of themselves, determine the situational context.
They are clues which help the interpreter to interpret it, read in the light of and in
conjunction with the interpreters MR.
In CDA, the notion of language as discourse means analysing the relationship between
all three levels or dimensions of texts, processes and their social conditions, both the
immediate conditions of situational context and more remote conditions of institutional
and social structures, as shown in the example above. It also proposes that discourses
are hierarchical, and that there are thus different orders of discourse. By looking at the
way in which actual discourse is determined by underlying conventions, it is claimed that
these conventions cluster in sets which are called orders of discourse, because they are
more general than a specific type of discourse. For example, in the example given above
between a policeman and a witness, the discourse type given in figure 4.x is an element
in the order of discourse associating policing as a social institution. It contrasts with
others, such as the discourses of making an arrest, charging a suspect and so on. It also
contrasts with others in the discourse of interviewing a witness such as questioning
aimed at teasing out a story. It is the prerogative of the powerful participant to determine
which discourse type is appropriate. In the example above, the choice positions the
witness in a determinate place in the order of discourse and the social order of police
work. It positions the policeman and witness in terms of one of a number of procedures
for dealing with cases that are constituted by a series of discourse types in determinate
orders: for example, information gathering is usually followed by interrogation followed by
perhaps a charge being laid and so on.
The way in which orders of discourse are structured and the ideologies embedded within
them are determined by relationships of power in particular institutions and in society as
a whole. Consequently, it is social structures that determine discourse, and as such
discourse has an effect upon social structures and contributes to the achievement of
social continuity and social change. Take, for example, subject positions in schools. In
school there exists a set of situations where discourse occurs (for example, class,
assembly, playtime, meetings), a set of recognized social roles in which people
participate in discourse (headteacher, teacher, pupil, friend, peer ) and a set of approved
purposes for discourse associated with, for example, learning and teaching, examining,
maintaining social control, and a set of discourse types. Within CDA, occupying a
subject position is a matter of doing certain things. Teachers and pupils each know (or
learn) what they are allowed to say or not to say within particular discourse types, and
occupying particular subject positions also reproduces conventions. Discourse thus
determines and reproduces social structure. Such reproduction can be conservative, in
the sense of sustaining continuity by replication, or transformative, in effecting change.
Since much of this behaviour goes unquestioned and is implicit in our relations with one
another, then the overall aim of CDA is for people to become conscious of the power
relations that exist within and through discourse. By raising self-consciousness, it is
argued that effecting change becomes possible.
4.6.2 CDA, written discourse analysis and stylistics
As the sections in this chapter have explained, CDA is a complex and wide-ranging
theory and method which attempts to analyse texts, both spoken and written, within its
three-dimensional framework of text, interaction between text and its processors and the
social and institutional contexts within which both text and interaction occur. Its
corresponding categories of description, evaluation and interpretation attempt to provide
frameworks within which categories of analysis can be generated and replicated. As a
theory, it draws upon sociological and cultural theory as well as linguistic theory, in ways
which have yet to be fully developed, especially in providing a theory and method for the
ways in which all three aspects of its framework interrelate. Its attraction as a theory and
method is its focus upon exposing the underlying ideologies of a text, be it spoken or
written, and its underlying assumption that discourse is structurally unequal and that
every discourse is based upon internal relations of language and power. Consequently,
such a critical approach can also be adopted within stylistics when analysing written texts
to discover underlying ideological assumptions by drawing upon, for example, models of
transitivity, modality and point of view (see: Simpson 1993; Clark & Zyngier 1998). In
particular a transitivity analysis can reveal much about the various roles portrayed by the
characters and the relationships between them. For example, a transitivity analysis of
passages taken from novels written in the romance genre or fairy tales will usually reveal
that women and girls are more acted upon than acting, with men or boys more often
than not in charge of the action. Van Leeuwen (1996:32-70) has investigated agency
and the ways in which the representation of social actors, in this case in a newspaper
article onimmigration, are represented and positioned.
CDA has also provided a framework for stylistic analysis undertaken from a feminist
perspective, as proposed by Sara Mills (1995:199-202), of which the following is an
adaptation:
1. Context (CDA dimension of explanation): This involves asking questions such
as: what sort of text is it, and to which genre does it belong (novel, advert, newspaper
article and so on)? Is there a tendency for women or men to be associated with this type of
text? What is its status, history, and your reason for analysing it?
2. Gender, reading and writing (CDA dimension of interpretation): This involves
asking questions such as: what assumptions do you have to make about the voice
of the author, and whether it is female or male? Does knowing this serve any
purpose? Does the text address you as male or female, or a universal audience?
Does the text address you directly in other ways? Is background knowledge
drawing upon stereotypical assumptions about men and women?
3. Gender vocabulary, the clause and discourse (CDA dimension of description): This
involves asking questions such as: how are males and females named and/or referred to in
the text? Do any of the terms used have sexual connotations, or taboos associated with
them? Is he used as a generic pronoun? Who acts in the text? Why is the text using
humour? What are the transitivity choices, and are they the same for men and for women?
From whose point of view does the text emanate? What elements are associated with male
and female?
For example, the text of this advertisement for a car which appeared in The Times Magasine
29.04.06, and the analysis which follows it. The text that appears below is given in the bottom
half of a lefthand page, whilst a picture of the car appears on the opposite righthand page.
THE TOPLESS MODEL YOU CAN TAKE SERIOUSLY.
CHRYSLER
Inspiration comes as standard The New PT Cruiser Cabrio.
Context: This is a magazine advert for a car, which appears in a recent The Times
magasine supplement. Although the magazine, aims to appeal to both men and women,
with a mixture of gender-specific and general articles, this particular car advert is aimed
at men. The purposes of analysis is to demonstrate the underlying ideological
assumption about male purchasing power and stereotypical portrayal of females in car
advertising that persist into the twentyfirst century.
Gender, reading and writing: The voice of the author is male, aimed at male readers.
Positioning the text on the left-hand page and the picyure on the right menas that
attention is drawn to the lefthandise and the text first. A female reading the advert is
positioned not as a prospective purchaser of the car, but as a topless model. Although
the pronoun you here is generic, in this instance it addresses men, not women.
Background knowledge draws upon stereotypical assumptions about men and women,
particularly in the reference to women as strippers, men as their voyeuristic audience
and also the underlying assumption that men, not women, buy cars.
Gender vocabulary, the clause and discourse: the most strikingly obvious naming in the text
from the point of view of gender is the topless model, used as a pun on the phrase open-topped
and on the personalities of women who work as topless models, perceived by men as of great
attraction but minimal intelligence. The transitivity process of the sentence the topless model you
can take seriously is a verbal one, with some unknown male addressing his male reader through
the detictic use of you, The modal can is an example of truth modality and a positive shading,
thereby portraying topless models in a positive light which, by association implies an alternative
negative shading. The large typeface of this sentence attracts attention, whilst the factual
information about the car, together with the Chrysler slogan is placed below the head-line
grabbing clause.
Sexual connotations of adverts like this to sell cars are nothing new, but what this
example illustrates is the ways in which gender stereotyping and sexual inequality is still
so ideologically ingrained in sections of Western society, and particularly the car industry,
persist in contemporary Western society. When it comes to car advertising, women are
still portrayed as sexual objects used to sell cars and men as those who have the
purchasing power to buy the car, and, by implication and association, women.
However, discovering underlying ideological assumptions is not the main point of focus for all
analysis undertaken in stylistics as it is for CDA, since stylistic analysis also works within and
draws from other theories and methods such as the cognitive, descriptive and so on. Stylistics may
thus draw upon CDA theory and methodology in undertaking analysis within the third strand
identified in 3.1, that is, that which takes account of the sociocultural and historical context of
both texts and readers/listeners, including the ways in which texts mediate authority, power and
control. Equally, CDA draws upon similar theories and methods as stylistics in the first and
second strands identified in 3.1, that is, the structural, formal or descriptive elements, concerned
with the recognisably formal and linguistic properties of a text existing as an isolated item in the
world, and the psychological: that which refers to the points of contact between a text, other texts
and the way readers/listeners process them. Stylistics and CDA may differ in the text they choose
as their main focus of analysis, but also with the kinds of texts chosen for analysis, with stylistics
most often and commonly (but not always) analyses literary texts, whilst CDA focuses more on
texts such as advertising, natural conversation, political speeches and so on. Nevertheless,
Ideology is still encoded and embedded in literary texts just as much as those of any other kind,
just as ideology can affect a reader in a literary work as it can in any other.
As both theory and method, CDA has been criticised on a number of counts. Firstly, it has often
been criticised for the fact that its data, in focusing upon institutional settings inevitably raises
issues of power and inequality. Consequently, the range of texts chosen for analysis is generally
selected in ways which inevitably throw up ideological considerations: for example, newspaper
reports and doctor-patient conversations (Widdowson, 1995 and 2004, and Toolan, 1997).
However, its practitioners counteract criticism by pointing out that for them, all settings, domestic
and informal as well as institutional, are underpinned by ideological assumptions. Secondly,
CDA has been criticised for the fact that the framework for textual analysis at an ideological level
is nowhere near as fully developed as those which deal with its more formal, linguistic levels. Its
practitioners would argue that no such claim has been made, that CDA is not a closed field, and
that the criticism is not of itself a valid reason for abandoning it. CDA has been the first attempt
so far to formalise a methodology which seeks to articulate the relationship between a text, the
context in which it is produced and its processes of production, reception and interpretation. One
recent development coming from CDA has been the growing areas of forensic linguistics, which
examines the relationship between language and the law. This fascinating field of inquiry not only
has a theoretical dimension, but also an applied one in working with the police and lawyers on
actual cases such as appeals in order to bring a linguistic dimension to them.
4.7 Studying CDA
The aim of any CDA analysis is to expose the ideology embedded in language. Your
task then, will always be one which seeks to expose the ways in which inequalities are
expressed through language and their effect upon a reader. Exposing inequality between
a policeman and a witness is one thing, but showing how a newspaper continually
demonises Muslims, for example, may show their relative powerlessness but also, as a
reader, might affect my own view of muslims. The focus of research in CDA is upon the
nature of interaction as it occurs in everyday speech and writing, taking account of the
contexts within which it occurs and identifying linguistic patterns with the aim of revealing
the underlying ideological assumptions at work that underpin the discourse. CDA, like
conversation analysis, and indeed pragmatics in general, studies the order, organization
and orderliness of social action, particularly as it is located in everyday interaction, but is
above all concerned with the ways in which language structures (and conceals) issues of
power and equality. The aim of research in CDA then, is to uncover issues of power,
authority and control through not only analytic descriptions of the organization of spoken
and written interaction but also by analyzing the underlying social, cultural and
institutional structures through which the interaction occurs.
4 .7.1 Selecting a topic for study
The first thing to decide when undertaking your own research into CDA is the focus of
your study. For example, you may wish to:
> Compare different newspaper reporting of the same event;
> analyse a speech or speeches;
> analyse an advertisement or undertake a contrastive analysis of two advertisements;
> analyse spoken exchanges in a particular setting, for example, between a teacher
and a pupil, a policeman and a witness, or a mother or father and a child.
> compare spoken exchanges between men and women, among women or men of a
similar age or of different ages.
Comparing newspaper reporting of the same event involves you in taking two or three
different newspapers for the same day with different political orientations, for example
The Mirror and The Daily Telegraph, or The Daily Mail and The Guardian, and analysing
their coverage of a particular news event. The aim of your analysis here is to apply
methods of analysis from within each of the three dimensions associated with CDA, in
order to expose underlying ideological assumptions at work within the texts. This also
allows you to take account of not only the words used but also any accompanying
pictures and/or photographs used.
Analysing speeches involves you in taking a speech, extract from a speech or comparing
two speeches in order to expose underlying ideological assumptions at work within the
texts, by applying methods of analysis from within each of the three dimensions
associated with CDA. For example, a student of mine analysed the speeches given by
the founder of scouting, Baden Powell, one given to Girl Guides and one to Boy Scouts.
The analysis revealed an extremely gender-biased view of the role Baden Powell
envisaged for the different sections of scouting.
Like analysing newspaper articles, analysing an advertisement or advertisements
involves you in applying methods of analysis associated with the three dimensions of
CDA in order to expose underlying ideological assumptions at work within the texts,
again taking account of the words used and accompanying pictures and/or photographs.
Analysing spoken exchanges in a particular setting, for example, between a teacher and
a pupil, a policeman and a witness, or a mother or father and a child, or comparing
spoken exchanges between men and women, among women or men of a similar age or
of different ages, involves you in applying CDA methods of analysis in order to expose
underlying ideological assumptions at work within the spoken exchange. The difference
here is that the data with which you are concerned is speech, and draws upon methods
of analysis associated with pragmatics and discourse within a framework of CDA.
Once you have chosen the focus of your study, then you will need to orientate or locate it
within any previously published research. This will help you to form a better picture of
the field of enquiry in which you wish to locate your study, giving you ideas and
information about, for example, the type of data you may wish to collect for your own
study. It will also help you to identify what the arguments are about, what the questions
are and which research methods are most commonly used. This will help you to:
formulate your own question more clearly; give you an idea of what findings you might
expect, and what would be unexpected.
4.7.2 Choosing and collecting data
Since CDA is concerned with both spoken and written texts, then the data you collect will be
either that of naturally occurring speech or of written texts. Research in CDA, as with that in
pragmatics, is usually qualitative in nature; that is, the methods it uses involve collecting data
from a small range which allows for a more in-depth analysis than that which quantitative
methods such as a questionnaire allow (see also 1.4.2). The quantitative dimension in CDA is
provided for in much the same way as for pragmatics and discourse, in that like studies can be
compared with like, and the more studies that are undertaken in a particular area such as
parliamentary speeches, for example, then the larger the research database becomes.
The aim of collecting data for the purposes of studying CDA is to analyse its content to uncover
or expose underlying ideological assumptions. Consequently, your choice of data will be
determined to a large extent by the topic of your research question. If your research question
involves you in analysing coverage of an event in different national newspaper articles for
example, then it is wise to choose newspapers published on a day when a story is likely to be
covered by most of them, such as a shooting, a large-scale robbery, a political scandal or some
other equally newsworthy event. Similarly, if your research question involves you in analysing
adverts with a view to exposing ideological assumptions of gender, then you are more likely to be
successful if you choose adverts for household appliances or motor cars than, say, for furniture or
music.
When it comes to collecting spoken data, then you can choose between collecting your own or
obtaining it from transcripts available from sources such as the Internet or by recording television.
Transcripts of trials by jury in the USA are available on the internet and television programmes
such as Court TV broadcasts live trials, or you could record news broadcast reporting of particular
events.
How much data to collect will again depend upon the topic of your study. The CDA
framework given in 4.6.3 below allows both for a very detailed analysis of the actual
texts analysed and the wider textual and social processes within which it occurs.
Consequently, a great deal can be written about a small amount of data. For example, if
you wish to compare newspaper reporting of the same event, then usually two articles of
between 400 600 words is sufficient. Analysis of a speech or speeches totalling 600
words is again usually enough. Analysis of an advertisement or a contrastive analysis of
two advertisements is more difficult to quantify in terms of word length, but between 50
100 words per advert works as a rough guide. 600 words of transcribed spoken data is
normally sufficient for an analysis of spoken exchanges. Since the focus of your
analysis of spoken data includes categories of analysis from pragmatics and discourse,
then it is best to transcribe your data according to the conventions associated with
spoken discourse (see 2.x ).
4.7.3 Analysing data
Having chosen your text or texts, your analysis will involve applying the three levels or
dimensions of CDA: description, interpretation and evaluation.
1. Description. This involves analysing the text at the level of description by analysing its
formal linguistic properties, such as:
a. subject positioning through transitivity and modality,
b. the relational values between the subjects (i.e., how pronouns and forms of address
are used and the implications of such use),
c. pragmatic analysis in terms of e.g. turn taking,
d. negative and positive sentences which indicate a struggle between the producer of
a text and its audience, especially evident in political texts such as speeches.
For example, if you are comparing newspaper reporting of the same event or analysing
advertisements, then this will involve you in analysis of formal linguistic properties most
usually applied to written texts, such as 1a. and 1b. above: subject positioning through
transitivity and modality; how pronouns and forms of address are used to draw
conclusions about relational values between subject; the use of reported speech or
metaphorical language in constructing relations between subjects and so on. If you are
analysing a speech or speeches, then you would add consideration of aspects such as
1d. to this analysis: that is, including consideration of positive and negative sentences. If
you analyse spoken exchanges of any kind, then applying categories of pragmatic and
discourse analysis as in 1c. above would be more relevant. Your ability to undertake
such analysis will involve you in engaging with MR, specifically in this instance of spoken
exchanges, identifying pragmatic categories such as those associated with turn-taking or
exchange structures.
2. Interpretation: This involves assessing the descriptive analysis undertaken in 1. above
and interpreting it as a process of production and interpretation, identifying discourse types
such as:
a. Discourse type contents: the type of activity going on, its topic and purpose,
b. Discourse type subjects: the subjects involved,
c. Discourse type relations: the relations between the subjects,
d. Discourse type connections: the role of language in what is going on.
(see also Fig 4.2 above).
For example, comparing newspaper reporting of the same event, in addition to analysis
of the linguistic properties of the actual text studied, would also involve considering it in
relation to newspaper reporting as a genre or as whole, in terms of its content, the
subjects involved, the relations between them and how language is used in such
reporting. How this process of interpretation is arrived at involves drawing upon MR,
particularly text structure and point. A newspaper report of a particular case of political
corruption, for instance, would involve locating the descriptive analysis undertaken for 1.
above within the wider sphere of newspaper reporting in general and the reporting of
political events in particular. For example, a newspaper report will usually follow
conventions of newspaper reporting such as headlines and sub-headlines, and possibly
even a house style. Reporting of political events will usually assume a great deal of
implied knowledge on the part of the reader in ways which reporting of say, a natural
disaster in another part of the world does not. It may also use political language, usually
in the form of reported speech, within the report and your analysis would discuss the
relation between the two and how this affects its interpretation. The extent to which you
are able to undertake such analysis and interpretation will depend upon your own
experience of discourse types, in this instance newspaper reporting and political
language.
Similarly, analysis of a speech made at an inauguration of a society of some kind would
involve locating your analysis of its content (the particular inauguration), the subjects
involved (usually the speaker and members of the society), the relation between them (
between the speaker and the members) and the part played by language in establishing
the relation within the wider sphere of speeches in general and inauguration speeches in
particular. Again, the extent to which you are able to undertake such analysis and
interpretation will depend upon your own experience and knowledge of such speeches.
3. Explanation: This involves analysing discourse in terms of the institutional and social
conditions of production and interpretation, and is the least developed in terms of categories of
analysis within the CDA framework. It can involve considering:
a. to which institutional processes does the discourse or discourse types belong, and
how are they ideologically determined. For example, identification of the particular
institution to which the discourse or discourse types belong, such as the political,
educational, social and so on, and identification of underlying institutional ideology,
such as conservative, capitalist, liberal, socialist, Marxist and so on,
b. to which societal processes does the discourse belong, and how is it ideologically
determined. For example, identification of any social institution such as the family,
marriage, religion to which the discourse or discourse type belong, and identification
of underlying social ideology in terms of race, class and gender.
For example, analysing an advertisement would involve identifying the particular discourse
type and institutional ideology underlying it. For example, in a car advertisiemt currently
running on British television, two young boys who are neighbours admire the newly
purchased car of one of their sets of parents, parked on the driveway of their newly
bought, presumably bigger and more expensive, house. This advertisement belongs to
an economic discourse with an underlying capitalist ideology. The discourse also belongs
to the social institution of family and social class, with an underlying social ideology of
equality in terms of race and social class, since one of the boys is white and the other
black, with the two families living side by side in a middle class suburban area.
Analysing spoken exchanges such as the one in 4.x above between a policeman and a
witness belongs to a social and political discourse of law and order with an underlying
ideology of conservatism and capitalism. It also belongs to the social institution of law
enforcement and an underlying social ideology of justice. Also implicit in the exchange is
an underlying structure of authority and control which unbderpins all exchanges between
the police and the public. The witness is doing her duty as a law abiding citizen in
reporting what she saw of the burglary, within the framework allowed by the policeman.
The extent to which you are able to undertake an evaluation of this kind and others will
depend upon your own experience and knowledge of discursive practices and ideology.
Undertaking a CDA analysis as described above involves you in identifying and
describing linguistic properties of a specific text, and in relating that analysis to the
discourse types of which it is a part. In many ways, this is very similar to the kinds of
analysis described in the other three chapters of this book. What makes this a
critical discourse analysis is firstly, the concept of discourse used which does not
distinguish between speech and writing, and secondly, in the added dimension of
evaluation which relates the analysis to underlying structures of authority and control,
ideology and power.
4.7.4 Sample projects
Using a Critical Discourse Analysis approach:
a) Critically analyse one or more than one recent news report(s) in order to explain how
the ideology of the text works. You may choose to concentrate on a newspaper or TV or
radio news report or you may contrast reports of a single event from two or more
sources.
b) Critically analyse one or more advertisements in order to explain how the ideology of
the text works.
c) Critically analyse a spoken exchange between i) a teacher and a pupil, ii) a policeman
and a witness, or iii) a mother or father and a child, to explain underlying ideologies of the
exchange.
d) Critically analyse a spoken exchange between i) men and women or ii) among women
or men of a similar age, identifying and explaining underlying ideologies of the
exchange.
4.8 Where to find out more
Fairclough (1995 & 2001), Wodak & Meyer (2002) and edited collections such as caldas-
Coulthard & Coulthard (1996) Van Dijk (1997) and Toolan (2002) give an overview of
CDA, whilst collections such as Meinhof & Richardson (1994) investigate particular
aspects of social life from a CDA perspective. Widdowson (2004) offers a comprehensive
critique of the approach. whilst Wodak and Chilton (2005) chart recent developments.
Chilton (2004) gives an account of political discourse and Lazar (2005) edits a collection
on feminist approaches to CDA.
References
Aitchison, J (2001) Language Change: Progress or Decay? Cambridge: Cambridge University
Press [1
st
edn 1981]
Althusser, L. (1971) Ideology and ideological state apparatuses, Lenin and Philosophy and other
Essays New Left Books
Austin, J.L. (1962) How to do Things with Words Oxford: Oxford University Press Bakhtin, M
(1981) The Dialogic Imagination Texas: University of Texas Press
Baugh, Albert C. & Cable, Thomas (2002) 4
th
edn. A History of the English Language Harlow:
Pearson [1
st
edn 1951]
Beal, Joan (2006) Language and Region London: Routledge
Biddulph, Joseph (1986)_A Short Grammar of Black Country. Pontypridd: Languages
Information Centre.
Biber, Douglas, Finegan, Edward & Leech, Geoffrey (1999) The Longman Grammar of
Spoken and Written English Harlow: Pearson
Bonvillian, Nancy (1993) Language, Culture and Communication: the meaning of messages New
Jersey:Prentice-Hall Inc
Brown, Gillian and Yule, George (1983) Discourse Analysis Cambridge: Cambridge University
Press
Brown, P. and Levinson, S. (1987) Politeness Cambridge: Cambridge University Press
Burrow, J.W (1982) ed Sir Gawain and the Green Knight London: Penguin [1
st
publication date
unknown]
Burton, Dierdre (1980) Dialogue and Discourse: A Sociolinguistic Approach to Modern Drama
Dialogue and Naturally Occurring Conversation London: Routledge & Kegan Paul
Caldas-Coulthard Carmen Rosa & Coulthard, Malcolm (1996) Texts and Practices: Readings in
Critical Discourse Analysis London: Routledge
Cameron, D. (1997) Performing gender: young mens talk and the construction of heterosexual
masculinity, in Sally Johnson and Ulrike Meinhof (eds), Language and Masculinity Oxford:
Blackwell. Pp.47-64.
Cameron, D. (2001) Working With Spoken Discourse London: Sage Publications Ltd
Carter, R & Nash, W (1990) Seeing Through Language: A Guide to Styles of English Writing
Oxford: Basil Blackwell
Carter, R. & McCarthy M. (2006) The Cambridge Grammar of English Cambridge: Cambridge
University Press
Chambers, J.K & Trudgill, Peter (1980) Dialectology London: Cambridge University Press
Chapman, S. (2006) Thinking about language; Theories of English Basingstoke: Palgrave
Cheshire, J. (1993) Real English: The Grammar of Written Dialects in the British Isles Harlow:
Longman
Cheshire, J. (1991) (ed) English Around the World: Sociolinguistic Perspectives Cambridge:
Cambridge University Press
Chilton, Paul (2004) Analysing Political Discourse: Theory and Practice London: Routledge
Chouliaraki Lilie & Fairclough, Norman (1999) Discourse in Late Modernity: Rethinking Critical
Discourse Analysis Edinburgh: Edinburgh University Press
Clark, Urszula & Zyngier, Sonia (1998) Women beware Women: Detective Fiction and
Critical Discourse Stylistics in Language and Literature, 1998, vol. 7, no. 2: 141-58.
Clark, Urszula (2001) War Words: Language, History and the Disciplining of English Oxford:
Elsevier Science Ltd
Clark, Urszula (2004): The Phonology of the English West Midlands in A Handbook of Varieties
of English Vol 1, ed. B. Kortmann & Edgar W. Schneider Berlin, New York: Mouton de Gruyter
2005 Pp 260-288
Clark, Urszula Stylistics (with J. McRae), in A Handbook of Applied Linguistics, eds. Alan
Davies & Catherine Elder. Oxford: Blackwell 2004
Coates, J & Cameron, D eds (1986) Women in Their Speech Communities Harlow: Longman
Coates, J. (1996) Women Talk Oxford: Basil Blackwell
Coates, J (1997) One-at-a-time: the organization of mens talk, in [Link] and [Link]
(eds) Language and Masculinity Oxford: Blackwell
Cook, G. (1989) Discourse Oxford: Oxford University Press
Coulthard, M. (1977) An Introduction to Discourse Analysis London: Longman
Coulthard, M. and Montgomery M. (eds) (1981) Studies in Discourse Analysis London:
Routledge & Kegan Paul
Crowley, T (1991) Proper English: Readings in Language, History and Cultural Identity London:
Routledge
Crowley, T (1996) Language in History: Theories and Texts London: Routledge
Cruse, A. (2000) Meaning in Language: An Introduction to Semantics and Pragmatics Oxford:
Oxford University Press
Crystal, David (1995) The Cambridge Encyclopedia of English Language Cambridge: Cambridge
University Press
Culpepper, Jonathan (2001) Language and Characterisation London: Longman
Cutting, Joan (2002) Pragmatics and Discourse: A resource book for students London: Routledge
Davies, Alan & Elder, Catherine (2004) A Handbook of Applied Linguistics Oxford: Blackwell
Dickens, Charles (1842 & 2000) American Notes London: London: Penguin
Dickens, Charles (1994) Little Dorrit Harmondsworth: Penguin [1
st
published 1855-1857]
Dillard J.L. (1976) American Talk New York: Random House
Edelsky, Carole (1981) Whos got the floor? Language and Society 10: 383-421
Eckert, Penelope (1988) Adolescent social structure and the spread of linguistic change Language
in Society 17: 183-207
Eckert, Penelope (1989) Jocks and burnouts: social categories and identity in the High School
New York and London: Teachers College Press
Eliot, T.S (2002) Collected Poems: 1909 62 London: Faber & Faber
Elliot, George (2002) Mill on the Floss London: Penguin [1
st
published 1860]
Emmott, C (1997) Narrative Comprehension: A Discourse Perspective Oxford: Oxford
University Press
Fairclough, N. & Wodak, R (1997) Critical Discourse Analysis in [Link] Dijk (ed), Discourse as
Social Interaction London: Sage
Fairclough, N. (2001) Language and Power Harlow: Longman[1
st
edn 1989]
Fludernik, Monika (1993) The Fictions of Language and the Language of Fiction: the Linguistic
Representation of Speech and Consciousness London: Routledge
Foucault, M. (1972) The Archaeology of Knowledge London: Tavistock
Fornas, J. (1995) Cultural Theory and Late Modernity London: Sage Publications
Fowler, R. (1996) Linguistic Criticism Oxford: Oxford University Press[1
st
edn 1986]
Fowler, R. (1991) Language in the News London: Routledge
Fowler, R., Hodge, R., Kress, G. & Trew, T (1979) Language and Control London: Routledge
Francis, Gill & Hunston, Susan (1992) Analysing everyday conversation in Malcolm Coutlhard
(ed) Advances in Spoken Discourse Analysis London: Routledge
Gavins, Jo (2006) Text World Theory:An Introduction Edinburgh: Edinburgh Press
Garfinkel, Harold (1967) Studies in Ethnomethodology Englewood Cliff, NJ: Prentice Hall
Golding, William (1955) The Inheritors London: Faber
Grice, Paul (1975) Logic and conversation in P. Cole and J. Morgan (eds) Speech Acts (Syntax
and Semantics Vol 3) New York: Academic Pres, 113-28Reprinted in Paul Grice (1989) Studies in
the Way of Words Cambridge: Harvard University Press: 22-40
Gumperz, J.J. & Hymes, D. (eds) (1986) Directions in Sociolinguistics: The Ethnography of
Communication New York and Oxford: Basil Blackwell [1
st
edn 1972]
Gumperz, J.J. (ed) (1982a) Language and Social Identity Cambridge: Cambridge University Press
Gumperz, J.J. (ed) (1982b) Discourse Strategies Cambridge: Cambridge University Press
Gutterson, David (2003) Our Lady of the Forest London: Bloomsbury Publishing Plc
Halliday, M.A.K. (1971) Linguistic structures and literary style: an inquiry into the language of
William Goldings The Inheritors, in [Link] (ed) Literary Style: A Symposium New York
and London; Oxford University Press pp. 332-4 quoted in R Fowler (1991).
Halliday, M .A. K. (1973) Explorations in the Function of Language London: Edward Arnold
Halliday, M. A. K. (1978) Language as a Social Semiotic: The social Interpretation of Language
and Meaning London: Edward Arnold
Harris, Thomas (2000) Hannibal London: Arrow Books
Haugen, E (1966) Dialect, language, nation in Pride, J.B. and Holmes, J. (eds) (1972)
Sociolinguistics pp. 97-111. Harmondsworth: Penguin
Hawley, Patrick (2002) What is said Journal of Pragmatics 34: 969-991
Hewings, Ann & Hewings Martin (2005) Grammar and Context London: Routledge
Hillman, Sister Mary Vincent (1961) The Pearl Notre Dame, London: University of Notre Dame
Press [1
st
published 1360-1400]
Hobsbaum Philip (1996) Metre, Rhythm and Verse Form London and New York: Routledge
Hodge, R & Kress, G (1993) Language as Ideology London: Routledge [1
st
edn 1979]
Hoey, Michael (1983) On the Surface of Discourse London: Allen &Unwin
Hoey, Michael (1996) in Rosa Caldas-Coulthard & Malcolm Coulthard (eds) Texts and Practices;
Readings in Critical Discourse Analysis London: Routledge
Hoey, Michael (2005) Lexical Priming: a New Theory of Words and Language London;
Routledge
Hutchby, Ian & Wooffitt, Robin (1998) Conversation Analysis: An Introduction Cambridge:
Polity Press
Hughes, Arthur, Trudgill, Peter & Watt, Dominic (2005) English Accents and Dialects: An
Introduction to Social and Regional Varieties of English in the British Isles London: Hodder
Arnold [1
st
edn 1979]
Hymes, D. (1972) Models of the interaction of Language and Social Life in J.J. Hymes, D. (1974)
Foundations in Sociolinguistics Philadelphia: University of Pennsylvania Press
Jakobson, R. (1960) Linguistics and Poetics in T. Seboek (ed) Style in Language Cambridge,
Mass: Massachusetts Institute of Technology Press
Jeffries, Lesley (1993) The Language of Twentieth Century Poetry Basingstoke: Palgrave
Jeffries, Lesley (1998) Meaning in English Basingstoke: Palgrave Macmillan
Jeffries, Lesley (2006) Discovering Language: Describing English Basingstoke: Palgrave
Johnson, Samuel (1755) A Dictionary of the English Language London
Joyce, James (1992) Ulysses Harmondsworth: Penguin [1
st
published 1922]
Joyce, P. (1991) in Language, Self and Society eds [Link] and R Porter Cambridge:Polity Press
Kerswill, Paul & Williams, Ann (2000) Creating a new town koine: children and language change
in Milton Keynes. Language in Society 29 (1) 65-115
Kerswill, Paul & Williams, Ann (2002) Dialect recognition and speech community focusing in
new and old towns in England: the effects of dialect levelling, demography and social networks.
In Daniel Long & Dennis Preston (eds.) A Handbook of perceptual dialectology Vol 2
Amsterdam: Benjamins
Kiefer, F (1979) What do the conversational maxims explain? In Linguisticae Investigationes 3
(1): 57-74
Kortmann Bernd and Schneider, Edgar W (2004) (eds) . A Handbook of Varieties of
English Vol 1, Berlin, New York: Mouton de Gruyter
Labov, W. (1966) The Social Stratification of English in New York City Washington, DC: Center
for Applied Linguistics
Labov, W. (1972a) Sociolinguistic Patterns Philadelphia: University of Pennsylvania Press
Labov, W. (1972b) The study of language in its social contxt, in Pier Paulo Giglioli (ed.),
Language and Social Context Harmondsworh: Penguin. Pp. 283-307
Labov, W. (1978) ed Sociolinguistic Patterns Oxford: Blackwell
Labov, W. (1979) The logic of non-standard English. In Language in the Inner City: Studies in
the Black English Vernacular Philapelphia: University of Pennsylvania Press, Philapelphia
Labov, W. and Fanshel, D.(1977) Therapeutic Discourse New York: Academic Press
Lakoff, G. & Johnson, M. (1980) Metaphor We Live By Chigago, Il: University of Chicago Press
Lawrence, D.H.(2006) Lady Chatterleys Lover London: Penguin [1
st
published 1928]
Lazar M (2005) ed Feminist Critical Discourse Analysis: Gender, Power and ideology in
Discourse London: Palgrave
Leech, G. (1983) Principles of Pragmatics London: Longman
Leech, G.N., and Short, M.H. (1981) Style in Fiction London: Longman
Leech, Geoffey, N & Thomas, Jenny (1990) Language, Meaning and Context: Pragmatics in
Collinge (Ed), An Encyclopedia of Language London: Routledge 173-206
Leith, D. (1997) A Social History of English London: Routledge & Kegan Paul [1
st
edn 1983]
Levinson S (1983) Pragmatics Cambridge: Cambridge University Press
Lucy, J.A. (1992) Language Diversity and Thought; A Reformulation of the Linguistic Relativity
Hypothesis, Cambridge: Cambridge University Press
McRae, John (1997) Literature with a Small l. London: Macmillan/Prentice-Hall. [1
st
edn
1991]
Marx, K. (1976) Capital: Vol 1 Harmondsworth: Penguin
Mencken, H.L. (1921) 2dn edn The American Language: An Inquiry into the Development of
English in the United States
Meinhof Ulrike & Richardson, Kay (1994) text, Discourse and Context: Representations of
Poverty in Britain London and New York; Longman
Mills, Sara (1995) Feminist Stylistics London: Routledge
Milroy, L. (1987) Observing and Analysing Natural Language: A Critical Account of
Sociolinguistic Method Oxford: Blackwell
Milroy, J. (1992) Linguistic Variation and change Oxford: Blackwell
Milroy, J. & Milroy, L. (1993) Real English: The Grammar of English Dialects in the British
Isles London: Longman
Milroy, L. & Gordon, M (2003). Sociolinguistics: Method and Interpretation. Oxford:
Blackwell.
Montgomery, Martin (1995) An Introduction to Language and Society London and New York:
Routledge [1
st
edn 1986]
Muller, Max (1862) Lectures on the Science of Language London: Royal Institute of Great Britain
Murray, J. (1888-1928) A New English Dictionary Based on Historic Principles Oxford:
Clarendon Press
OHalloran, Kieran (2003) Critical Discourse Analysis and language Cognition Edinburgh:
Edinburgh University Press
Ochs, E., Schegloff E.A and Thompson, S.A. (1996) Interaction and Grammar Cambridge:
Cambridge University Press
Orton, H. (1962). Survey of English Dialects: introduction. Leeds: Arnold.
Pope, Alexander, Churton Collins John (1896) Essays on Criticism London: Macmillan
Poulantzas, Nicos (1976) Classes in Contemporary Capitalism London: New Left Books
Pratt, M.L. (1977) Towards a Speech Act Theory of Literary Discourse Bloomington: Indiana
University Press
Richards, I.A. (1925) Principles of Literary Criticism London: Kegan Paul
Roberts, Celia, Davies, Evelyn and Jupp, Tom (1992) Language and Discrimination: A Study of
Multiethnic Workplaces London: Longman
Sacks, H, Schegloff, E. and Jefferson, G. (1974) A simplest systematics for the organization of
turn-taking in conversation Language 50 (4), 696-735
Sapir, Edward (1921) Language; An Introduction to the Study of Speech Harcourt, Brace &
World, Inc
Saussure de, Ferdinand (1916) Course in general Linguistics London: Duckworth
Schank, R. C. & Abselon,R.P (1997) Scripts, Plans, Goals and Understanding Hillside, NJ:
Lawrence Erlbaum
Schmidt, Norbert (2002) An Introduction to Applied Linguistics London: Arnold
Schiffrin,Deborah (1994) Approaches to Discourse Oxford UK & Cambridge USA:Blackwell
Searle, J.R (1969) Speech Acts: An Essay in the Philosophy of Language Cambridge: Cambridge
University Press
Sebold, Alice (2003) Lovely Bones London; Picador
Semino, E & Culpepper, J (2002) eds Cognitive Stylistics: language and Cognition in Text
Analysis Amsterdam: John Benjamins
Semino, E. & Short, M. (2004) Corpus Stylistics: Speech, Writing and Thought Presentation in a
Corpus of English Narratives London: Routledge
Shaffer, Peter (1975) Equus London: Penguin
Shaw, Sylvia (2000) Language, gender and floor appointment in political debates, Discourse &
Society 11 (3): 401-18
Short, M. (1988) ed. Reading, Analysing and Teaching Literature. Harlow:
Longman.
Short, M. (1996) Exploring the Language of Poems, Plays and Prose London: Addison Wesley
Longman Ltd
Short, M. (1995) in Verdonk, Peter and Weber, Jean-Jaques (eds) Twentieth Century Fiction:
From Text to Context London: Routledge
Simpson, David (1986) The Politics of American English 1776-1850 New York and
London: Oxford University Press
Simpson, Paul (1993) Language, Ideology and Point of View London: Routledge
Simpson, Paul (2004) Stylistics London: Routledge
Sinclair, J. and Coulthard, M. (1975) Towards an Analysis of Discourse: the English used by
Teachers and Pupils London: Oxford University Press
Smith, O (1984) The Politics of Language Oxford: Clarendon Press
Spencer-Oatey, Helen and Wenying Jiang (2003) Explaining cross-cultural pragmatic findings:
moving from polieness maxims to sociopragmatic interactional principles (SIPs) Journal of
Pragmatics 35: 1633-1650
Sperber, D. and Wilson, D. (1995) Relevance: communication and cognition Oxford:
Blackwell [1
st
edn 1986]
Stockwell, P. (2002) Cognitive Poetics: An Introduction London: Routledge
Stubbs, M. (1986) Educational Linguistics London:Blackwell
Swift, J. (1712) Proposal for Correcting, Improving, and Ascertaining the English
Tongue [Link]
Tatalovitch, R. (1995) Nativism Reborn?: The Official English Language Movement and
the American States Lexington, Kentucky: University of Kentucky Press
Thomas, Jenny (1995) Meaning in Interaction: An Introduction to Pragmatics London:
Longman
Toolan, M (1997) What is Critical Discourse Analysis and Why are People Saying Such Terrible
Things About It? Language and Literature 6:2, pp. 83-103
Toolan, M (1998) Language in Literature: An Introduction to Stylistics London: Hodder
Toolan, M (2001) Narrative; A Critical Linguistic Introduction London: Routledge [1
st
edn 1998]
Toolan, M (2002) Critical Discourse Analysis: Critical Concepts London: Routledge
Trudgill, Peter (1974) The Social Differentiation of English in Norwich Cambridge:
Cambridge University Press
Trudgill, Peter (ed) (1978) Sociolinguistic Patterns in British English London: Edward
Arnold
Trudgill, Peter (1983) On Dialect: Social and Geographical Perspectives Oxford;
Blackwell
Trudgill, Peter (1999) Standard English; What it isnt in Bex, T & Watts, R. J. (eds)
Standard English: The Widening Debate London: Routledge
Twain, Mark (1986) The Adventures of Tom Sawyer London; Penguin [1
st
published
1910]
Van Dijk, T (2001) Critical Discourse Analysis in D. Shiffrin, D. Tannen & H. Hamilton
eds. The Handbook of Discourse Analysis: Oxford: Blackwell
Verdonk, Peter and Weber, Jean-Jacques (eds) (1995) Twentieth Century Fiction: From
Text to Context London: Longman
Van Dijk, T (1997) (ed) Discourse as Structure and Process. Discourse Studies: A
Multidisciplinary Introduction, Vol 1 London: Sage
Van Leeuwen, Theo (1996) The Representation of Social Actors in Caldas-Coulthard,
Carmen-Rosa and Coulthard, Malcolm (eds). Texts and Practices: Readings in Critical
Discourse Analysis London: Routledge pp. 32-70
Wales, Katie (1990) 2
nd
edn 2001 A Dictionary of Stylistics Harlow: Longman
Wareing, Shan and Thornborrow, Joanna (1998) Patterns of Language: Stylistics for Students of
Language and Literature London: Routledge
Watson, Greg & Zyngier, Sonia (2006) eds Literary Education: A Critical Reader
Palgrave
Watts, Richard, J. (2003) Politeness Cambridge; Cambridge University Press
Wells, John C. (1999) Accents of English. Cambridge: Cambridge University Press
Welsh, (1993) Trainspotting London: Vintage
Werth, Paul (1999) text Worlds; Representing Conceptual Space in Discourse Harlow:
Longman
West, C. and Garcia, A (1988) Conversational Shift-work: A Study of Topical Transitions.
Social Problems, Y2h 35 (pt. 5), pp. 551-75.
Whorf, B.L. (1939) The Relation of Habitual Thought and Behaviour to Language, in
J.B. Carroll (1956) (ed.) Language, Thought and Reality Massechusetts: MIT Press
Widdowson, J.D.A & Upton, Clive (2006) An Atlas of English Dialects London: Routledge
Widdowson, H.G. (1995) Discourse Analysis: A Critical View in Language &
Literature 4 3 pp 157-72
Widdowson, H.G. (2004) text, Context, Pretext: Critical issues in Discourse
Analysis Oxford:Blackwell
Witman, Walt (2005) Leaves of Grass London: Penguin [1
st
published 1855]
Wittgenstein, Ludwig (1951) Philosophical Investigations Oxford: Blackwell
Wodak R. and Chilton, P (2005) (eds) A New Agenda in (Critical) Discourse Analysis
Amsterdam: Benjamins
Wodak, Ruth & Meyer, Michael (2002) eds Methods of Critical Discourse Analysis
London: Sage
Wolfram Walt & Schilling-Estes (2006) 2
nd
edn American English Malden, MA, USA;
Oxford, UK; Victoria, Australia: Blackwell
Woods Anthony, Fletcher Paul, Hughes Arthur & Anderson S.R. (1986) Statistics in Language
Studies Cambridge: Cambridge University Press
Woolf, Virginia (1978) Night and Day London; Granada Publishing Ltd [1
st
published 1919]
Woolf, Virginia (1999) Mrs Dalloway London: Penguin [1
st
published 1923]
Wright, Laura & Hope, Jonathan (1996) Stylistics; A practical coursebook London:
Routledge
Wyld H.C. (1927) 3
rd
edn A Short History of English with a Bibliography of Recent
Books on the Subject London: Murray