a corpus approach to the representation of islam and muslims in the uk press

A Corpus Approach to the Representation of Islam and Muslims in the UK Press

Costas GabrielatosEdge Hill [email protected]

Newcastle Critical Discourse Group

4 December 2013

What is a corpus?

A collection of texts in electronic form.

British National CorpusWritten language section: 90 million words

1200 books

Intuition / introspection vs.

corpus evidence

Why use corpora?

A very small number of textsvs.

A very large number of texts

Perception of usage

(Dictionary definitions)

vs.

Actual usage

(BNC)

Why use corpora? An Example

Lets compare:

egregious

EgregiousDictionary definitions

Conspicuously bad or offensive.

Often of mistakes, extremely and noticeably bad.

An egregious mistake, failure, problem etc is extremely bad and noticeable.

Extraordinary in some bad way; glaring; flagrant: an egregious mistake; an egregious liar.

Conspicuous ; especially : conspicuously bad : flagrant

EgregiousDictionary definitions: Meaning



An egregious mistake, failure, problem etc. is extremely bad and noticeable.


Conspicuous ; especially : conspicuously bad : flagrant

EgregiousDictionary definitions: Use



An egregious mistake, failure, problem etc. is extremely bad and noticeable.


Conspicuous ; especially : conspicuously bad :flagrant

egregious in the BNC

36 instances

bad

conspicuous

However

Action/behaviour/event/result etc. (15)

Person/organisation etc. (12)

Mistake/error etc. (6)

Object (2)

Other (1)

Text vs. Texts: Cumulative patterns

The hidden power of media discourse and thecapacity of . . . power-holders to exercise this powerdepend on systematic tendencies in news reportingand other media activities. A single text on its ownis quite insignificant: the effects of media power arecumulative, working through the repetition ofparticular ways of handling causality and agency,particular ways of positioning the reader, and soforth.

(Fairclough, 1989: 54)

Corpus-based studies look at patterns

words and word groups

grammatical units

meanings

frequencies

co-occurrence

A1: We dont need to. We examinepatterns derived from automatedanalyses.

Q: Corpora contain numerous texts.If we use a corpus approach,how can we read all of them?

A2: We do condensed reading(Gabrielatos, 2005) i.e. targetedreading of particular items/featuresthrough representative samples(concordance lines).

Collocation

The frequent co-occurrence of two words

within a span

usually 4 or 5 words to the left and right

Usually statistically calculated.

List of collocations depends on set parameters (span, statistic etc.)

E.g. adjectives modifying cat (BNC)

Collocations and meaning: Semantic Preference

The relation between a lemma or word formand a set of semantically related words.

(Stubbs 2001, 65).

The examination of the contextual meaning ofstrong collocates can provide a semanticanalysis of a word.

(Sinclair, 1991: 115-116)

Collocations and attitude: Semantic/Discourse Prosody

(Louw, 1993; Stubbs, 2001)

The consistent aura of meaning with which a form isimbued by its collocates.

(Louw, 1993: 157)

A semantic prosody is attitudinal and on the pragmatic side of the semantics/ pragmatics continuum

(Sinclair, 1996: 87, also Stubbs, 2001: 66).

Collocation patterns are often unavailable to intuitionor conscious awareness. They can convey messagesimplicitly and even be at odds with an overtstatement.

(Hunston, 2002: 109).

Qualitative / Quantitative

Many DA studies use various kinds of quantifyingexpressions, such as routinely, regularly, frequently.(ten Have, 2007: 157)

Analysis of aggregates requires a preliminary analysis ofsingle instances. (ten Have, 2007: 157)

In corpus analysis, the quantitative approach forces thelinguist to make constant decisions to achieve a stringentcategorisation. [] Thus the qualitative approach, whichestablishes the basic concepts [], must precede anyquantitative investigation. (Schmied, 1993: 85)

The terms quantitative and qualitative may be morehelpfully regarded as notional methodological extremes.(Baker et al., 2008: 276)

Project

Title: The representation of Islam andMuslims in the UK press, 1998-2009

Aims: Topics related to Islam/Muslims

Attitudes towards Islam/Muslims

Differences between newspapers

Diachronic changes

Funding body: ESRC

Research team: Paul Baker

Costas Gabrielatos

Tony McEnery

Data: source and query

Online database: Nexis UK

Query

Alah OR Allah OR ayatolah OR burka! OR burqa! OR chador! OR fatwa! OR hejab! OR imam! OR Islam! OR Koran OR Mecca OR Medina OR Mohammedan! OR Moslem! OR Muslim! OR mosque OR mufti! OR mujaheddin! OR mujahedin! OR mullah! OR muslim! OR Prophet Mohammed OR Q'uran OR rupoushOR rupush OR sharia OR shari'a OR shia! OR shi-ite! OR Shi'ite! OR sunni! OR the Prophet OR wahabi OR yashmak! AND NOT Islamabad AND NOT shiatsu AND NOT sunnily

Tony McEnery

Corpus

BusinessDaily Express + Sunday Express

Daily Mail + Mail on SundayDaily Mirror + Sunday Mirror

Guardian + ObserverIndependent + Independent on Sunday

PeopleDaily Star + Daily Star Sunday

SunTelegraph + Sunday Telegraph

Times + Sunday Times

Articles: 200,000

Words: 143 million

Spelling normalisation

Sub-corpora: per newspaper per year (1998-2009) broadsheets/tabloids political orientation

Corpus tools and methodology

Tools

WordSmith 5, Sketch Engine

Methodology

Collocation analysis

Word Sketches

Collocates of a word within a grammatical construction Muslim_ADJ + Noun

Muslim_NN as Subject / Object ( Agent / Patient)

Frequency comparisons

Concordance analysis

Approaches informed and supplemented each other

Motivation (1)

Expanded concordance analysis of sketches:

NOUN against/on Islam

n+against-i+N, 905, 35.4

war 340 5.78

crusade 64 9.02

blasphemy 28 8.49

prejudice 23 6.53

conspiracy 20 5.87

fight 20 4.68

struggle 12 4.32

hatred 12 4.16

West 11 7.88

offence 10 4.11

n+on-i+N, 1238, 9.1

war 243 5.29

attack 157 4.62

expert 74 6.32

view 61 5.02

book 46 4.24

comment 36 5.72

lecture 29 7.67

assault 29 5.59

remark 26 6.16

conference 23 4.80

back 19 4.88

slur 10 7.34This is the most salient sketch for Islam_N

Motivation (2)

Recurring juxtaposition of two frames:

Izzadeen told listeners: 'Everyone knows, Muslims and non-Muslims, that the war on terror is a war against Islam, and I'm telling you something, if they don't stop this there is going to be a very strong reaction from the community.

[Daily Mail, 23.09.2006]

But Israel's best interests will not be served by poking sticks in the Palestinians' eye at a time when the West is straining every muscle to persuade the Muslim world that the war against Osama bin Laden is not a war against Islam. The Americans may be very close to Israel, but Mr Sharon will be making a mistake of historic dimensions if he forces them to choose between maintaining their alliance with Israel and winning the wider war against terror.

[The Independent, 26.09.2001]

war etc. against/onIslamterror* / Osama bin Laden etc.

Motivation (3)

Frequent juxtaposition of frames frequent co-occurrence of Islam* and terror* within the same text.

Are the frequent references to terrorism a characteristic of the juxtaposition of these frames, or a more general attraction?

Frequency comparison of Islam* and terror* in the corpus.

Collocation of the forms

025000

50000

75000

100000

125000

150000

175000

200000

Islam* terror*

Islam* is a query term - terror* is not!

Corpus texts containing Islam* tend to also contain terror*

However

Collocation between Islam+ and terror+

terror terrorism terrorist terrorists

Islam ()

Islamic

Islamism -- --

Islamist

Islamists

MI3 and LL 6.63() LL 6.63 (but MI

Collocation between Islam+ and terror+

Proportion of texts in which they co-occur after 9/11 and 7/7

% of texts

11 September 10 October 2001 81.7

7 July 6 August 2005 71.6

terrorist 45818terror 44315terrorism 43564terrorists 34923terrorised 646terrorise 514terrorising 426terrors 218terrorist-related 132terrorism-related 119terror-related 77terrorises 32

Examination of all content types with a frequency of at least 40,000 (40k content types).

The four clearly most frequent forms have an average frequency of around 40,000.

40k content types

Nouns, adjectives, verbs

147 types

0.03% of corpus content types

15.1% of corpus content tokens

85 (58%) are clear indicators of topics or contextual elements in corpus articles (40k indicators)

0.02% of content types

9.5% of content tokens

40k indicators

Reference to ... Types

... armed or violent conflict, and the attendant issue of death

... police / security13

... countries or regions involved in armed conflict of some description - in which religion is, or is perceived to be, directly or indirectly, one of the main causes of that conflict.

29

... governance and leadership, human aspects (e.g. age and sex), time, money, and quantification.

These types usually refer to the (willing or unwilling) participants in the conflict, their representatives/leaders, or the place/time of conflict.

42

40k indicators directly or indirectly refer to issues of conflict ...

... although the query consisted of terms relating to a religion, its believers and attendant customs/practices.

Tip of the iceberg?

Examination of 4k content types

2,981 types0.5% of content types61% of content tokens

Can 85 very frequent content types (representing only 10% of the corpus content tokens) be indicative of the main corpus topics? Yes

4k content types

Three categories:

Types indicating topics in the corpus articles.

Types indicating contextual elements

e.g. participants, place, time.

Types with general meaning / use

Essential for discussing the topics, but not clear topic-indicators themselves (e.g. general adjectives, modality/attitude markers).

Topic indicators: religion, culture, society, education

Topic indicators: Conflict

Topic indicators: Other

4k topic-indicators: tokens

QUERY TERMSRELIGION/CULTURE/

EDUCATION

CONFLICT

HUMAN ASPECT

FINANCE/BUSINESS

TRANSPORT/TRAVEL

SPORT-LEISURE

MATERIALS/RESOURCES FOOD/DRINK

MEDIA/COMMUNICATION

Topic indicators: Tokens and Types

Religion/Society/ Culture/Education

Conflict

Other

0

500000

1000000

1500000

2000000

2500000

3000000

3500000

4000000

4500000

5000000

5500000

0 50 100 150 200 250 300 350 400 450

Toke

ns

Types

The topic or the press?

Is this focus on conflict characteristic of the corpus, or of general newspaper reporting practices?

Frequency comparisons

Frequency comparison (1)

Frequency comparison of the group of CONFLICT tokens in Islam-UK and the news subcorpus of BNC (BNCnews)

Islam-UK(142,962,543)

BNCnews(9,897,378)

Islam-UK%

BNCnews%

Diff. LL

5,278,904 272,131 3.70 2.75 +34.5% 24,618.53

However, BNC articles are from the late 1990s

Frequency comparison (2)

Islam-UK2005-2007

(44,120,772)

BE2006news

(174,963)

Islam-UK2005-2007

%

BE2006news

%Diff. LL

1,894,614 5,626 4.29 3.22 +33.2% 517.78

Frequency comparison of the group of CONFLICT tokens in:

IslamUK_2005-2007

BE2006news (Baker, 2009)

Diachronic development of the proportion (%) of CONFLICT tokens in annual sub-corpora

1998 1999

2000

2001

2002

2003 2004

20052006

2007

2008

2009

3

3.5

4

4.5

1 2 3 4 5 6 7 8 9 10 11 12

Sketch of Muslim used as an adjective

ADJ+n Freq. Salience (6.9)community 7676 10.4world 4928 9.19woman 4484 9.07country 3864 8.74leader 3217 8.8cleric 2279 9.36man 2158 7.5group 1837 7.74population 1799 8.89extremist 1710 8.81school 1220 7.67state 997 7.35girl 966 7.8family 901 7.05faith 823 7.68nation 805 7.66organisation 696 7.45youth 630 7.64fanatic 612 7.71student 579 7.18soldier 519 6.53child 505 6.2

Sketch of Muslim as an adjective-Noun collocates

22 types with freq. > 500- Less that 2% of sketch types- More than 50% of sketch tokens

Less frequent types may, collectively, reveal a different picture (Baker, 2004)

Examination of all sketch types

Categories and sub-categories (when applicable) Examples of noun collocates

CONFLICTextremist, fanatic, terrorist, fundamentalist

RELIGION cleric, faith, festival, preacher

CULTURE

Social practices dress, culture, name, tradition

Education school, teaching, education, college

View/attitude/emotion opinion, anger, voice, attitude, grievance

ETHNIC/NATIONAL

ENTITY

Population community, population, nation, world

Area/country country, state, area, region, land

Governance leader, voter, MP, government, ruler

CHARACTERISING/

DIFFERENTIATING

ATTRIBUTES

Age/sex woman, man, girl, youth, child, teenager

Family/relationship family, parent, brother, friend, wife

Occupation/role officer, patient, doctor, worker, assistant

Ethnicity/race/nationality Briton, Albanian, Malay, Arab

Other house, shop

GROUP /

ORGANISATIONgroup, organisation, association, charity

Collocate categories: frequency and lexical richness

CONFLICT

RELIGION

ETHNIC/NAT. ID.

CHAR./DIFF. ATTR.

CULTURE

GROUP/ORG.

0

5

10

15

20

25

30

35

40

0 5 10 15 20 25 30 35

Fre

qu

en

cy (

toke

ns)

Lexical Richness (types)

FrequentLexically poor

InfrequentLexically poor

FrequentLexically rich

InfrequentLexically rich

Conflict: Examples

Daily Star readers have told Union Jack burning Muslims to sod off. On Saturday,we asked you whether the marauding extremists who torched OUR flag on OURstreets should be kicked out of OUR country. And thousands of you phoned tosay YES. In a record reply to a Daily Star phone and text poll, a staggering 99.7 %told the fanatics to pack up and leave. Thousands of patriotic Brits deluged ourvoteline after we published sickening pictures of British Muslim extremistsburning the Union Jack on the streets of London. The mob tore up an appealfrom fellow Muslims for an end to bloodshed and chanted: "You will pay, binLaden 's on his way. [Star, 05.04.2004]

In February about 400 people attacked and burned a church in the southern cityof Sukkur after accusations that a local Christian had set fire to pages from theQuran. After a similar allegation last November, a Muslim mob wielding axes andsticks set fire to three churches, a dozen houses, three schools, a dispensary, aconvent and two parsonages. The attacks were the worst on Pakistan's Christiancommunity since 2002, when Muslim fanatics led an assault on a church withgrenades on Christmas Day. [Telegraph, 28.04.2006]

The category of conflict indicates instances when issues of armed/violent or social conflict are expressed directly.

However, discourses of conflict are pervasive, and are indexed by all other categories.

Religion (cleric)

WHEN a prominent Muslim cleric, Abu Qatada, was arrested in a South London flat last month after nearly a year on the run, security chiefs in Britain recognised that his supporters might seek their revenge. [Times, 18.11.2002]

THE Muslim teaching assistant who insisted on wearing her veil in the classroom was following fatwa orders. Aishah Azmi, 24, took advice from Mufti Yusuf Sacha, a Muslim cleric in West Yorkshire. It's said the Mufti told her it was obligatory to wear the veil around men who were not relatives. [Star, 30.10.2006]

Culture/Practice

A LEADING barrister says Sharia law in Britain is "inevitable", despite 95% ofDaily Star readers insisting it is wrong in our phone poll yesterday. Most of ourreaders believe that Britons should not have to live under the controversialMuslim code in their own country, which includes stoning to death for adulteryand cutting off hands for theft. But former Chairman of the Bar Council StephenHockman said there was no reason why Sharia law could not be applied here. Hesaid : "Given the world situation and our own substantial Muslim population, it isvital that we now look at ways to integrate Muslim culture into our owntraditions. [Daily Star, 05.07.2008]

Culture/Practice

Reedah Nijabat, a young barrister whose family is from Pakistan [...] is convincedthat what high-flyers like herself lack is their own watering hole - based onSoho's fabled Groucho Club - and has made it her mission to start one. So far, 30-year-old Nijabat has got a lease on three adjoining shops awaiting conversionand been promised bank loans of pounds 200,000. The site of the ArRum club isin the heart of loft-living, warehouse-clubbing, multi-racial Clerkenwell. So whydo Muslims need a separate place to socialise? "Because most of them feel a bituncomfortable about going to a bar or being offered alcohol, Nijabat says. Whileit's true that many second and third generation Muslims here feel moreintegrated and outgoing than their parents, Nijabat reckons their specific needsare not being met by the commercial sector. She's tired, too, of negative imagesof Islam in the media. There's a whole vibrant side of Muslim culture - markedby an interest in art, architecture and literature. [Guardian, 06.12.2000]

Education

Recently, Bradford got its first state-funded Muslim school, sparking a debateover whether such religious schools - which already exist for Christians and Jews- would promote segregation. [Independent, 17.04.2001]

DOZENS of new Muslim state schools could be created in a controversialexpansion of faith education planned by Ministers. Schools Secretary Ed Ballspromised funds to take more than 100 private Muslim schools into the statesector to meet rising parental demand. But there were immediate warnings thatsetting up new faith schools could backfire by increasing religious tensions.Opponents said it could entrench segregation and called them 'plain madness'.And a Muslim leader admitted the biggest obstacle to the opening of new Islamicstate schools was public fears that they would 'produce fundamentalists'. [Mail,11.09.2007]

Community

Those who have lived in Muslim districts frequently remark on the warmth andneighbourly atmosphere of streets which are often among the few in Britainwhere every resident knows the names of his neighbours. They furnish a senseof refuge, and immeasurably buttress the confidence of children and the old,who might otherwise find themselves culturally adrift and isolated. In the longerterm, however, the Muslim community must ask itself whether the demographicmap of the former mill towns of the North is an adequate reflection of its senseof Islam's necessary separateness. [Independent, 15.12. 2001]

Once the bombers' identities had been exposed, the usual platitudes followedthick and fast. They were variously nice, friendly, cricket playing, sports loving,community minded, wonderful with kids. Three lived in the Leeds area where,we are constantly told, the white and Muslim communities rub along togetherjust fine. [Mirror, 17.07.2005]

Muslims presented as a nation or homogeneous population

It also comes just before Mr Obama's visit to Europe and Turkey. It has alreadydrawn warm praise from European leaders, who have been pressing for a newdiplomatic push to underpin the international pressure on Tehran over itsnuclear ambitions. And when he arrives in Istanbul, on his first presidential visitto a Muslim country, he can already show a substantial initiative to improverelations with the Muslim world. [Times, 21.03.2009]

The Palestinian claim to statehood is unanswerable, and with wiser leadership itwould have been flourishing for years. It is tragic that the cause is now being soruthlessly exploited with Jew as a codeword for extremist incitement of hatred ofAmerica and the West. This is jihad. It is aimed at us all, at Europeans who "looklike" Americans because they believe in liberal democracy and are infected byAmerican culture. But its first victims are the Palestinians and the frustratedmasses of the Muslim world. Their leaders have led them into ignominy in threewars. They have failed to reform their corrupt and incompetent societies. Habitsof mind tending to approve terror are becoming ingrained in the Muslim world,sanctioned by the lethargy and prejudice in Europe. [Sun, 29.06.2002]

But not all is negative - at least not intentionally ...

There should be no compromise on supporting universal human rights but theremust be flexibility in how we go about achieving them. Navigating that narrowline demands a mixture of humility and solidarity. Take Islam and women inBritain. Since sexism is omnipresent it would be foolish to assume that it hasescaped Muslim communities or that in some instances the position of someMuslim women is not particularly bad. It would also be arrogant to assume thatMuslim women have not noticed this already or to ignore that some of them aredoing something about it. Finally, it would be negligent to forget that Muslimsare the most vulnerable to racial attack in Britain, which has seen a steepincrease in fascist activity on the streets and at the polls in recent years and isinvolved in an illegal war against a Muslim country. Liberals outside the Muslimcommunity have some choices. We can either condemn the entire community assexist and impose our own priorities on them, thus leaving their communitiesmore embattled and strengthening the conservative forces within them. Or wecan talk to Muslim women's groups and feminists, of whom there are many, tosee what their priorities are and to find some common ground where we cansupport them and them us, as we struggle against our own demons of racismand Islamophobia. [Guardian, 12.07.2004]

Conclusions

The dominant discourses on Islam and Muslims in the UKnational press seem to revolve around issues of conflict - and itsparticipants, logistics and repercussions.

The link between Islam/Muslims and conflict in UK newspapersseems to have intensified after 9/11.

The direct mention of conflict is both very frequent and lexicallyrich.

Other uses of the adjective Muslim are usually embeddedwithin discussions of conflict.

Related to the presentation in a context of conflict are views of

Muslims as a homogeneous group;

Islam as a national/ethnic/cultural identity;

Muslims/Islam as different (Us vs. Them stance).

Main project outputs

Baker, P., Gabrielatos, C. & McEnery, T. (2013). DiscourseAnalysis and Media Attitudes: The representation of Islam inthe British press. Cambridge: Cambridge University Press.

Baker, P., Gabrielatos, C. & McEnery, T. (2013). SketchingMuslims: A corpus-driven analysis of representation around theword Muslim in the British press, 1998-2009. AppliedLinguistics, 34(3), 255278.

Gabrielatos, C., McEnery, T., Diggle, P. & Baker, P. (2012). Thepeaks and troughs of corpus-based contextualanalysis. International Journal of Corpus Linguistics, 37(2), 151-175.

a corpus approach to the representation of islam and muslims in the uk press

Documents