‘There are radical Muslims and normal Muslims’: an analysis of the discourse on Islamic extremism

Abstract

In this article the author develops a multi-perspective and poly-methodical model for discourse analysis and tests it by analysing data generated through focus-group discussions on Muslim–Christian relations in Tanzania and Indonesia. By doing so, he aims to demonstrate the use and usefulness of sociocognitive discourse analysis in religious studies. First he outlines and refines an approach to discourse analysis based on Norman Fairclough. Next he describes how he uses this approach in his fieldwork, focusing on discourses on Muslim extremists in Tanzania and Indonesia. Finally he concludes that the prospects of sociocognitive discourse analysis look promising and discusses some controversial issues that yet have to be resolved.

Keywords:

• sociocognitive discourse analysis
• extremism
• Islam
• Muslim–Christian relations

Introduction

In his article ‘Discourse’ in Braun and McCutcheon's Guide to the Study of Religion Murphy (2000: 403) notes that ‘when it comes to discourse theory and religion, we find ourselves in mostly uncharted territory.’ Ten years later it probably would be more appropriate to say that a fair number of scholars see discursive study of religion as a fruitful alternative to phenomenological or sociological research and as a way out of the opposition between science and humanities approaches (Flood 1999; Jensen 2003; Kippenberg 1983; McCutcheon 2007; von Stuckrad 2003), but that ‘[t]there are no clear rules or conventions as to how to conduct an analysis’ (Granholm 2005: 249, 260). Many studies of discourse and religion remain abstract and theoretical. They seldom analyse concrete discourses in detail nor do they explain and justify their tools and procedures well. The aim of this article is to demonstrate the use and usefulness of discourse analysis in religious studies. We do this by developing a multi-perspective and poly-methodical model for discourse analysis and testing it by analysing data generated through focus-group discussions on Muslim–Christian relations in Tanzania and Indonesia. The structure of this contribution is as follows. First we outline and refine a sociocognitive approach to discourse analysis. Next we describe how we use this model in our research, focusing on discourses on Muslim extremists in Tanzania and Indonesia. Finally we end with a discussion and a conclusion.

Sociocognitive discourse analysis

There are a great variety of approaches to discourse analysis (Fairclough 1992: 12–36; Jørgensen and Phillips 2002: 1–23; Wetherell 2001: 380–383), and a more unified paradigm is yet to be developed in order to facilitate theory formation in discourse studies. This variety notwithstanding, most discourse analysts would draw upon the following theoretical perspectives or research traditions to a greater or lesser extent (Fairclough 1992: 72).

First, there is the influence of linguistic pragmatics. Our primary access to social reality is through language. However, language not only reflects a pre-existing reality but also contributes to the construction of reality (Blommaert and Verschueren 1998: 4–8; Jørgensen and Phillips 2002: 8–9). Second, discourse analysts draw upon cognitive science (Van Dijk 2008: 110).1 The link between discourse (text) and reality (context) is not a direct one (deterministic) but is established (interpretation) via mental models (Van Dijk 2008: 75) or members' resources (Fairclough 2001: 20, 118).2 Third, there is sociology of knowledge or social constructivism.3 Mental models are not individual, but are constructed socially. Not only are the cognitive resources socially determined, but also the conditions of their use (Fairclough 2001: 19–20).4 Fourth, discourse analysts draw upon the archaeology of knowledge. There is an intrinsic link between knowledge production and power relations (Fairclough 1992: 37–61; Jørgensen and Phillips 2002: 12–14).

Underlying these perspectives and traditions are various ontological and epistemological claims deriving from the respective approaches of the sciences and the humanities, using different labels such as explanation versus understanding, function (cause) versus meaning (intention), positivist versus interpretive. We deal with these dilemmas in the discussion and conclusion. For the sake of clarity we first elaborate on the approach to discourse analysis that we use in our research, namely the three-dimensional model developed by Norman Fairclough (1992: 73; 2001: 21). His chief concern is to trace ‘explanatory connections’ (Fairclough 1992: 72, 80, 95) between language use (discourse) and social reality (structure).

Fairclough proceeds on the following assumptions. First, language use is a practice just as any other practice. The only difference is its linguistic form (Fairclough 1992: 72). Second, there is a dialectic relation (Fairclough 1992: 45, 65) between language use (the discursive dimension) and social reality (the non-discursive dimension). Third, the relation between language use and social reality is established via discursive practice (Fairclough 1992: 71, 86). In this introductory section we first explain three perspectives or ‘levels’ of analysis (Fairclough 2001: 20). Then we outline three methods or ‘stages’ of analysis (Fairclough 2001: 21). Finally we design an analytical model and show how we use it in our research.

Levels of analysis

Like other qualitative approaches (e.g., content analysis, conversation analysis), sociocognitive discourse analysis concentrates on participants' perspectives, but it does not do this in a naïve way (Fairclough 1992: 12–36; 2001: 1–14). Unlike these other qualitative approaches, this approach is not interested in participants' perspectives as such. Instead it focuses on how these perspectives relate to the participants' subject positions in social situations or situational contexts (Fairclough 2001: 87, 122), and more particularly to their interests in reproducing or transforming the societal order (Fairclough 1992: 65; Wetherell 2001: 383–387).

Consequently, sociocognitive discourse analysts distinguish between different dimensions of one and the same practice and look at them from different perspectives: the individual dimension or micro perspective, the institutional dimension or meso perspective, and the societal dimension or macro perspective (Fairclough 1992: 25; 164; 2001: 20–21). For example, the interactions between teachers and pupils, doctors and patients, or parents and children (micro level) are not only related to images of what specific institutions such as school, hospital and family are (meso level), but also to processes in the wider society (macro level) such as democratisation or commercialisation (Fairclough 1992: 65–66, 202–224). By speaking about images it is also evident that the ideational (cognitive) and the relational (social) dimensions are linked.

Whereas radical post-structuralists tend to assume that discourse positions the speakers, and discursive psychologists tend to assume that speakers position themselves and others through discourse, Fairclough (1992: 65) and other sociocognitive discourse analysts assume that speakers position themselves and are positioned by the discourse. The relation between the discursive and the non-discursive dimension is dialectical.5 According to Fairclough (1992: 58), Gramsci's concept of hegemony is superior to Foucault's concept of power. This is an important analytical insight which has also implications for the relation between micro (agency) and macro (structure) level of analysis.

Translated into religious discourse, we distinguish between the dimensions of individual believers (micro level), their (full or partial) identifications with their religions or religious institutions (meso level), and the societal context in which these religions or religious institutions operate (macro level). Thus in making an analysis we discriminate between when participants speak for themselves, when they speak as representatives of a religion or as citizens of a country. It is assumed that what believers think and say relates (positively or negatively) to ‘texts’ of the religious institutions that they belong to or (fully or partially) identify with, and to wider societal contexts which they reproduce or transform, for example, processes of democratisation or commercialisation.

Stages of analysis

Sociocognitive discourse analysts distinguish different stages, which are given different names. Fairclough speaks about analysis of linguistic practice, analysis of discursive practice and analysis of social practice. Alternative names are analysis of text, analysis of interaction and of context – or simply description, interpretation and explanation (Fairclough 1992: 198–199; 2001: 21–22).

As noted already, according to sociocognitive discourse analysts discourse is a practice like any other practice. In pragmatic terms, language use is not only a way of saying things (informative), but also of doing things (performative). The only difference from other practices is its linguistic form (Fairclough 1992: 71). Thus the first method is the analysis of the linguistic practice, or the formal properties of the text.

According to sociocognitive discourse analysts the relation between language use and social reality is not direct but is established through discursive practice. Consequently the second method is the analysis of discursive practice; this is the production, distribution and consumption of texts (Fairclough 1992: 71). The discursive practice (i.e., interaction) is crucial, as the dialectic relation between linguistic practice (i.e., text) and social practice (i.e., context) is based on it.

The dialectic relation between language use and social structure is based on the assumption that what participants say is shaped by and shapes social structures by either reproducing or transforming them (Fairclough 1992: 72). In other words, sociocognitive discourse analysts are interested in ideational and relational effects of language use, or in terms of Fairclough (1992: 8, 64), in effects on ‘social identities’ or ‘subject positions’, ‘social relations’ and ‘systems of knowledge and belief’.6 Thus the third method is the analysis of discourse as social practice.

A multi-perspective and poly-methodical model

Fairclough develops discourse analysis as a multi-perspective and poly-methodical approach focusing on three constructive effects of discourse: ideas, identities and relations (Fairclough 1992: 64). The distinction between analytic perspectives and methods is not clear-cut.7 There are overlaps (Fairclough 1992: 73, 231). For example, the term ‘social practice’ is sometimes used for a ‘level’ and sometimes for a ‘stage’ of analysis. Moreover, the distinction between perspectives and analytic methods does not reflect a one-to-one situation. All methods are used for analysing all dimensions of practice, although one method may be more fruitful for analysing a specific dimension than another. Sociocognitive discourse analysts use linguistic methods, but they do not need training in linguistics. They are interested in linking linguistic analysis with (micro and macro) sociological analysis – and within sociological analysis, linking the micro (agency) and the macro (structure) level.

Here is an example from a focus-group discussion among female Muslim workers in Surakarta (Solo), Indonesia:

Interviewee: Before there was a mosque, the tarawih praying [during Ramadan] was in my house. Yes, it was in my house. It was held, but it was in my house. Yes, thanks be to God (alkhamdulillah), [we] can floor [the mosque] including its, what is it called, veranda, that veranda, thanks be to God (alkhamdulillah), from the result of Noordin yesterday, thanks be to God (alkhamdulillah). [It was from] the donation from [the incident of] Noordin yesterday.

Other participants: Ah … ah …

Interviewee: Yes, yes, it is like that …

Interviewer: What do you …?

Interviewee: Noordin brought a blessing (berkah), he brought a blessing. Although Noordin was a disaster (musibah), [at the same time he] brought a blessing. That was the visitors, we circulated boxes of donations (infaq) for the mosque.

Interviewee: It was near to the holiday, Madame?

Other participants: Ah … ah …

Another participant: Yes, it was like a place for recreation [tourism)].

Interviewee: Yes, it was in [the month of] Fasting, after then the holidays. Finally, after it was collected … maybe how many millions, then for flooring [with ceramics] that veranda [of the mosque].

Interviewer: It was from the parking [fees]?

Interviewee: Yes, from the parking [fees], [and also] from the boxes of donations that we collected, then we use it for flooring [with ceramics]. Thanks be to God (alkhamdulillah), [but] also everything comes from God (innalillah).

Other participants: Ah … ah …

Noordin Mohammad Top, a famous terrorist from Malaysia, was moving from one locality to another in Indonesia, but was shot dead by the police in a house that was rented by him in a working-class area in Surakarta. People visit that place for various reasons (leisure). The lady disagrees with Noordin's ideas, describing him as a ‘disaster’ (micro dimension). Nevertheless she sees also some ‘blessings’ in this ‘disaster’, since the visitors' donations enable her to pay for a new floor for the mosque, including its veranda (meso dimension). This practice constitutes and is constituted by the commercialisation of religion in society (macro dimension).

Looking at the linguistic features of the text, the lady uses a strong contrast pair when she describes Noordin as a ‘disaster’ and a ‘blessing’ (linguistic practice). She also draws upon typically Islamic vocabulary associated with matters of life and death: ‘thanks be to God’ and ‘everything comes from God’, expressing God's all-powerfulness and man's duty to resign to His will, even in situations of disaster and misfortune (discursive practice). In so doing she positions herself as a pragmatic believer who sees profit in something that in itself is perceived as a disaster (social practice).8

Religious language, social cohesion and conflict

The author is involved in a research programme on religious discourse, social cohesion and conflict in Indonesia and Tanzania. He is interested in understanding socio-religious transformations in these societies in the post-Cold War era, more particularly in social identity reconstructions through inter-religious (Muslim–Christian) relations. He looks at these social-identity reconstructions from a theory of practice (particularly a theory of communicative practice) point of view (Blommaert and Verschueren 1991; 1998).9

In discourse analysis it is important to select a corpus of (natural or generated) discourse samples for closer inspection (Fairclough 1992: 226–228) and to augment it with data from interviews and panel discussions, websites and other media. In this article we make use of transcriptions of focus–group discussions on Muslim–Christian relations in the above-mentioned countries. In addition to these we also analyse newspapers and audio-visual sources for the sake of triangulation. Although the corpus for analysis comes from focus-group discussions, it is assumed that discourses at micro level constitute and are constituted by discourses at macro level (Van Dijk 1987: 21, 37–38; 2008: 4, 19–20; 2009: 250).10

The focus-group discussions took place between 2008 and 2010 in Dar es Salaam in Tanzania and Surakarta (popularly known as Solo) in Indonesia. Participants were selected using gender, religion, age and education as criteria. With these categories in mind we used snowball sampling facilitated by Muslim and Christian research assistants. We were not able to attract participants from groups that are labelled as extremist; thus our participants are mainly mainstream Muslims and Christians. Moreover, as Muslim–Christian relations are a delicate topic in Indonesia and Tanzania we assume that face-saving took place (Van Dijk 1987: 22). To a certain extent social desirability is unavoidable in social-scientific research and the tendency to maintain harmony and avoid conflict may be overrepresented in our case studies. We come back to this issue in the conclusion and discussion section.

In both towns we selected four male and four female groups, consisting of old and young, educated and uneducated participants, times three: Muslim, Christian and mixed, thus 24 focus groups per case study. There was on average five participants per group. We also interviewed various key informants, such as members of parliament and religious leaders. Taking the two case studies together, in total some 250 participants were involved. Basically we had two questions in mind: How do they (Muslims and Christians) speak about each other, and how do they (Muslims and Christians) speak with each other?

Dar es Salaam has between 2 000 000 and 3 000 000 inhabitants, Surakarta has about 750 000 inhabitants, but could count 1 500 000 daytime town dwellers. The majority of Dar es Salaam inhabitants are Africans, with Arab and Indian minorities. Surakarta's population is mainly Javanese with Chinese and Arab minority groups. In Dar es Salaam, where statistics are hard to get, 45 percent are assumed to be Muslims, 45 percent Christians, and 10 percent others, indigenous believers and Hindus. In Surakarta 77 percent are Muslims, 22 percent are Christians, and others are Hindus, Buddhists or Javanese (indigenous) believers. Both towns have had a series of riots and outbursts of violence in recent times, culminating in the Solo Riots (Surakarta) and the Mwembechai Killings (Dar es Salaam) in 1998.

The analysis will follow the ‘stages’ identified above. The ‘stages’ are distinguished for analytical purpose only. They do not imply chronological order. Data generation and (various stages of) data analysis refer to an iterative process. In each stage we note when participants speak for themselves, as individual believers; when they speak ‘as Christians’ or ‘as Muslims’ (i.e., official or unofficial representatives of a religion), and when they speak as Tanzanians and Indonesians or as Tanzanian Muslims or Indonesian Christians, thus identifying themselves as citizens of a nation or as ‘polyphonic selves’. Specific techniques of sociocognitive discourse analysis will be discussed in the course of the analysis.

Discourse on Muslim extremists

Although the Solo Riots were not religiously inspired, Surakarta is linked to ‘Muslim terrorists’ such as Noordin Mohammad Top who was shot dead by the police in Surakarta, and Abu Bakar Ba'ashir, the leader of the Ngruki pesantren (Islamic seminary) near Surakarta, who was sentenced to jail for the rest of his life. The Mwembechai Killings in Dar es Salaam were said to be fomented by ‘Muslim radicals’ who threatened national security and unity.

There is an on-going rhetoric about Islamic ‘terrorism’, ‘radicalism’ and ‘extremism’, both locally and globally (Abbas 2011; Mamdani 2004). In Tanzania ‘extremist Muslims’ are said to have burned pork butcheries and Christian churches (Wijsen and Mfumbusa 2002). In Indonesia ‘extremist Muslims’ are said to have bombed hotels in Bali and Jakarta (Wijsen and Husein 2009). This rhetoric gives the impression that the extremists who committed these crimes are Muslims (without an attempt at describing who or what a Muslim is, identifying them with a religion), that Islam is inherently violent and that Muslims are necessarily extremists. Our research aims primarily to discover how this rhetoric works in the construction of social reality, not if it is true.

The research strategy we use is a multiple case study of the hierarchical type. We conduct two case studies independently and compare them on a more abstract level. The aim of a case study can be to test generalised theories on specific cases. Our focus is on the social-identity theory (Tajfel and Turner 1986) because this theory is widespread in social-scientific studies of inter-group relations and conflicts. Basically this theory stems from Sumner's functional ethnocentrism theory, formulated more than a century ago: ‘The relation of comradeship and peace in the we-group and that of hostility and war towards others-groups are correlative to each other’ (Sumner 1906: 12).

We are inspired by the comparative study by Van Dijk (1987) of ethnic prejudice in two different states, the Netherlands and California. Comparing case studies in different countries raises various methodological issues that we cannot discuss in detail here. The ground for comparison is the assumption that micro and macro analyses are linked (Van Dijk 1987: 21, 37–38), certainly in an era of globalisation and mass media (Fairclough 1992: 200). Thus intra- and inter-group discourses (e.g., on Muslim–Christian relations) are linked to national and international discourses (e.g., on liberalisation and democratisation). Most movements in the global village are trans-national anyway. For example, there is quite some borrowing between Jama'ah Islamiya in Indonesia and al-Ikhwan al-Muslimun in Egypt, though local varieties exist.

Description

The first stage is description; this is the analysis of the linguistic features of the text (Fairclough 1992: 76–77, 185–194; 2001: 21, 91–116). There are various tools for doing this, but in our research we confine ourselves to vocabulary. ‘It is sometimes useful for analytic purposes to focus upon a single word’ or on ‘culturally salient keywords’, says Fairclough (1992: 185–186). Another focus for analysis is ‘alternative wordings and their political and ideological significance’ (Fairclough 1992: 77).

Words are never neutral. It does make difference if we refer to Palestinians as ‘freedom fighters’ or ‘terrorists’ (Fairclough 1992: 75). Thus we look at wording, over-wording, re-wording and alternative wording. What do Muslims and Christians say about Muslim militants? What words do they use when they speak about ‘the others’?

At the individual level, Muslims and Christians speak about each other as ‘relatives’, ‘brothers and sisters’, ‘children of one father’, ‘children of Abraham’, ‘friends’, ‘neighbours’, ‘comrades’ and ‘fellows’. And they describe the relations between them as ‘good’, ‘collaborative’, ‘respectful’, ‘peaceful’ and ‘harmonious’. Muslims and Christian ‘live side by side’ and the ‘cooperation [between them] is good’.

However, some Muslims are ‘radicals’, ‘fanatics’, ‘extremists’, ‘militants’, ‘hard-liners’, ‘puritans’ or ‘fundamentalists’. They ‘create boundaries between us and them’, ‘they discriminate against us’, ‘they tolerate violence’, ‘they segregate themselves’, ‘they attend the social part but not the religious part of ceremonies’, e.g., funerals. Radicals ‘don't socialise’ and they ‘dress in funny ways’.

At the institutional level, some Christian participants in Tanzania say that there is a link between Islam and violence, ‘as is clear from statistics. There are more Muslims than Christians in prison.’ And a moderate Muslim in Tanzania says about his militant fellows that ‘they are not afraid of using violence for defending their rights’ because to them, ‘a person who uses a cutlass for defending the faith, that person goes to heaven because he is faithful to the scriptures.’

However, the dominant discourse is that ‘there are radical Muslims but there are also normal Muslims’, using the words of a Christian participant in Indonesia. And another one in Tanzania said: ‘There are those [Muslims] who claim to know the faith and there are normal ones.’ That is to say, extremism or radicalism is not normal but ‘excessive’, an exception to the rule. The rule is that Islam means peace. And, ‘Pious Muslims are tolerant.’

Radicalism in Christianity exists also. This is evident in the way these Christians deal with religious texts. And sometimes they are ‘not sensitive to their surroundings’. However, these are not ‘old waves’ but ‘new waves’, neither Protestant nor Catholic but Pentecostal. And ‘those Christians are not Javanese Christians.’

At societal level a Christian woman in a mixed group in Tanzania said, ‘We Christians and Muslims have created a false belief, that is, Muslims have their things and Christians have their things’, thereby referring to segregation in the society. Again, extremism is linked to creating boundaries. Indonesia participants speak about ‘Christian zones’ and ‘Muslim zones’. Muslims are said to be more radical than Christians. This is because Christians embrace Western education and advocate nationalism, secularism and pluralism. ‘We are Tanzanians. We are supposed to love each other.’ Thus being a citizen of a nation comes first, adherence to a religion second.

In speaking about the Solo Riots, in which 28 ‘pro reformation’ students were shot dead by the police, the participants said that these riots were not religious (anti-Christian) or ethnic (anti-Chinese) but social (anti-rich) and political (anti-government), caused ‘just by trouble makers’. And after the riots, people went ‘back to normal’ as soon as possible, resuming their businesses as usual. But they also said that ‘the victims were ethnic Chinese, but Muslims from the ethnic Chinese groups were not victims.’ And, during the riots, ‘some people put a banner in front of their house saying, Muslim Java’, intended to prevent them from being attacked.

Interpretation

The second stage is interpretation; this is analysis of the production, distribution and consumption of texts. When participants produce (communicate) and consume (interpret) texts or talk, they draw on members' resources (Fairclough 2001: 118), or on mental models (Van Dijk 2008: 75) stored in their long-term memory (Fairclough 2001: 8–9). These resources are cognitive in the sense that they are in people's heads; and they are social in the sense that they have social origins and are socially generated (Fairclough 2001: 20).11 So the question is: what members' resources or mental models do the participants draw on to produce (communicate) or consume (interpret) texts?

At the individual level participants remember and refer to good and friendly relations with schoolmates and neighbours. Christians in Surakarta say that they meet with no resistance when building churches in Muslim zones, and among their relatives there are Muslims and Christians. ‘The extended family of my mother is Muslim. The extended family of my father is Christian.’ But some are ‘funny’. One Christian participant in Indonesia remembered that he played with Muslims as a child. ‘I entered their house. I sat on their sofa. The next day the sofa was cleaned and dried in the sun.’ But this kind of separatist behaviour is not based on the Bible and the Koran. Another Christian in Indonesia cites the gospel saying, ‘love your neighbour as yourself.’ The Koran is equally clear on that point: ‘If your neighbour believes in God, you have to honour him.’

But according to an Indonesian Christian the Koran is ambiguous. ‘If they take this verse they love others, if they take another verse they fight others.’ And, he adds, if they love others, ‘They are in line with Bible verses.’ Also, ‘if they love Christians, they will become nationalist Muslims who are tolerant.’ According to some Indonesian Muslims, Muslim permissiveness is misused by Christians to proselytise, especially those who are economically vulnerable. One Muslim respondent in Indonesia refers to Dewi, a Catholic nun who converted to Islam and who says that she was trained to proselytise. And another Muslim referred to Oom Rudi, a Catholic priest, who brought oil and rice to poor Muslims, or Christians offering Muslims free education. Christians are suspected of proselytisation.

At the institutional level participants link their texts to the text of the teachers and preachers of their religions, or to the principles of their organisations, such as Muhammadijah's Himpunan Putusan Tarjih. As one participant in Indonesia said (implicitly referring to Sura 109: 2–6), ‘I have my religion, I practice my religion. You have your religion, please practice your religion.’ In general religious leaders are stricter than ordinary believers, but most of them do not approve the use of violence. Both Christianity and Islam advocate peace. ‘Those who claim to know religion are the ones making trouble. But we, who are just Muslims, we don't pursue religious issues. That's why we compromise with Christians,’ says a female Muslim participant in Tanzania. ‘I myself, I don't know religion. How can I argue with a person,’ she continues. ‘The problem is with the leaders,’ says another female Muslim participant in Tanzania. ‘Between Muslims and Christians there is not a big difference, because they believe in one God.’ Also Muslims and Christians are ‘like children of the same father’.

At the societal level participants are inspired by the fathers of the nation. In Tanzania, participants drew upon the notion of ujamaa. ‘Nyerere always insisted that there must be unity [in the country] and that we must resist ethnic and religious discrimination,’ says a Muslim in Tanzania.12 But since Mwinyi was elected ‘everybody is free to do whatever he likes.’ Participants in Indonesia referred to the pancasila (five principles).13 ‘Each person is free to worship and to practice the religion of his own choice.’ One Christian in Indonesia referred to the formulation of the Jakarta Charter, according to which a Committee of Nine wanted to include the obligation to practice syari'ah for Muslims in the constitution.

International issues also provide background information that is stored in peoples' heads and is drawn on when producing and consuming text. One Indonesian Christian refers to the conflict between Israel and Palestine. ‘Christians are the friends of Israel. We must disagree with that. When we browse the internet we see that Christians support Israel. But they colonise our Muslim friends in Palestine.’ A Muslim in Tanzania refers to the civil war in Somalia. ‘If the government bans open air preaching, we would have a war like our friends in Somalia.’ In general, international institutions are labelled Christian and capitalism is called a ‘kafir system’.

Explanation

The third stage is explanation;14 this is the analysis of the sociocognitive effects of the texts. When participants draw on their cognitive resources these are reproduced (Fairclough 2001: 135). Thus social effects are obtained through members' resources (Fairclough 2001: 137). Van Dijk's study of ethnic prejudice clearly shows how stereotypes are reproduced in everyday talk. But they can also be reinterpreted and thus transformed. So the question is: What are the sociocognitive effects of what the participants say?

At the individual level participants in the focus-group discussions position and thereby constitute themselves as moderate people who are tolerant of believers of other faiths. They try to maintain harmony and avoid conflict. They say that ‘the others’ are among their friends, neighbours and relatives. Those who have extremist or excessive ideas are classified as ‘outsiders’, ‘foreigners’ or ‘strangers’, ‘not from here’. In Dar es Salaam they are said to be ‘people from Pemba’. In Solo extremist Christians are said to be ‘not Javanese’ and extremist Muslims are said to be ‘not from Solo’, or they are classified as ‘funny’ or ‘foolish’.

However, the boundary between moderate and militant is not fixed but fluid. The Muslim woman whom we quoted in the first section of this article did not identify with Noordin as a person, neither with his ideas nor practices, as she described him as a ‘disaster’. She nevertheless saw some profit in his activities, as he was described as a ‘blessing’. The same woman said that this event was ‘thanks to God’, because ‘everything comes from God.’ Some other Muslims in Surakarta said that even if they do not agree with extremists, they are nevertheless ‘heroes of Islam’.

At the institutional level participants represent their religions and their religious institutions as peaceful. In fact, ‘every person who does the will of God is a Muslim,’ thus a Muslim in Tanzania. Consequently, the boundary between clean and unclean, good and evil is not between Christianity and Islam, but within Christianity and Islam. Extremist Muslims are not only against Christians, but also and mainly against their fellow Muslims who are not serious about their faith. And: ‘The religion that discriminates is that of the leaders. We ordinary believers have no problem.’

At the societal level most participants of the focus groups reproduce the official image of Tanzania and Indonesia as being harmonious countries. They describe negative consequences of liberalisation and privatisation (e.g., some people expressing excessive ideas or displaying radical behaviour). But they maintain that this is not mainstream Islam. Muslims who wear long beards and short trousers are ‘funny’ and ‘foolish’, hence ‘not normal’, atypical.

‘It is all about money,’ says a participant in Tanzania, ‘sheer greed’. And this coin has two sides. On the one hand money facilitates inter-religious marriage. If the fiancée has money, there is no objection against Muslims marrying Christians. On the other hand, extremist preaching has become an income-generating project for the preacher. In both cases the participants say that conflict and cohesion are not caused by religion but by economics. As one participant puts it: ‘Who cares about religion?’ Others say that interreligious conflict is a ‘struggle about resources’, and interreligious marriage is ‘regulated by income’.

Another problem is that ‘freedom destroys harmony.’ One participant in Indonesia says, ‘Freedom leads to violence,’ another one in Tanzania said, ‘Freedom brings chaos.’ According to them, too little freedom is bad, but too much freedom is even worse. ‘Previously there was only one stream of Islam. Islam was still close to Java, to the culture of Java. Then our life was so peaceful. Now it is not. Now it is hot even among people of the same religion.’ Maybe the liberalisation processes in Tanzania and Indonesia, which transformed both societies from authoritarian to liberal ones, happened too fast. ‘We must reach a point where we put a limit on the freedom of worship,’ says a Tanzanian Muslim.

Various participants say that extremism is caused by international marginalisation of Islam, feeling threatened, humiliation, frustration, not taken seriously, conspiracy, rumour. A Muslim in Tanzania complained: ‘Muslims don't get assistance because they are linked to terrorism.’ Whereas Christians are allowed to join international (Christian) organisations, Muslims in Tanzania are not allowed to join the Organisation of Islamic Conference. The international community is on the side of Christianity, thus Islam is marginalised. ‘This is why we make noise,’ according to a Muslim in Tanzania. Referring to the bishops' protest against khadhi courts in Tanzania, he said: ‘If the bishops rule the country, the fight will be big.’ The influence of the bishops in Tanzania is disproportionate. ‘For the bishop it is easier to see a minister or even the president than for a member of parliament.’

So extremism starts with creating boundaries between ‘us’ and ‘them’, and using violence against ‘them’. But the picture painted by the participants is that extremism is marginal in Islam, ‘not normal’, even ‘funny’ and ‘foolish’, new and foreign trends, and that it is not specifically religious, although religion is part of it. The causes of extremism are anger, frustration about the kafir system, but also sheer greed, as extremism can be turned to profit and generate income. But the root cause is favouritism, measuring by two standards, one for Christians and another one for Muslims. That makes Muslims angry. ‘You can not have two children and favour one of them.’

Comparison

One way of tracing ‘explanatory connections’ (Fairclough 1992: 72) between language use and social realities is making comparative analyses of alternative wordings (Fairclough 1992) or studying discourse variations in different situations (Van Dijk 2008: 107). A condition for the possibility of comparison is equivalence. Without going into details here we can point to some contextual and conceptual similarities. Tanzania and Indonesia have been part of the Indian Ocean Trading Complex. Business has been going on between East Africa and Southeast Asia for more than 2000 years. The solidarity between the two continents was reconfirmed during the 1955 Africa–Asia Conference at Bandung, from which the Movement of Non-Aligned Countries emerged.

Before and after their independence from colonial rule, the political leaders of both countries formulated ideologies of national unity, ujamaa in Tanzania and pancasila in Indonesia. Their citizens had to become ‘Tanzanians’ and ‘Indonesians’ first. And in both countries, language was used as an instrument to achieve this national unity, Ki-Swahili in Tanzania and Bahasa Indonesia in Indonesia. Both countries are in a transition period from authoritarian rule to an open-market economy and multi-party politics. They try to retain unity in diversity, and have movements that strive for more autonomy in Aceh (Indonesia) and Zanzibar (Tanzania). Both countries struggle with diversities within the religions, between liberals on the one hand and radicals on the other, and they have politicians who before the regime change criticized the political establishment but who are now part of the system, doing the same things that they fought against before.

There are also differences between the philosophies of national unity. Whereas pancasila was formulated before independence and inspired the independence struggle, ujamaa was formulated after independence. Whereas pancasila favours religious pluralism, ujamaa is strictly secular, at least in theory. It has to be acknowledged also that Christianity and Islam are ‘world’ or universal religions with trans-national movements and exchanges of ideas, without suggesting that these movements are part of the same organisation or have close links. The Jama'ah Islamiya, which most likely has its centre in the Ngruki pesantren (Islamic seminary) near Surakarta, was inspired by Egypt's al-Ikhwan al-Muslimun. In the same way Charismatic, Pentecostal and Evangelical movements form trans-national movements within Christianity.

In Tanzania, where Christians and Muslims are equally balanced in the population, some Muslims say that they are ‘second-class citizens’ and are ‘ill-treated by the government’. But also in Indonesia, where Muslims form the vast majority, Muslims are ‘weak in the economy’ and fear Christianisation. Thus on a national level Christians in both Tanzania and Indonesia are perceived as well-to-do people and Muslims as less advanced. On the international level, global institutions are perceived as Christian and Muslims feel that they do not have equal rights. They feel marginalised, threatened, denied their identity by ‘the West’. ‘This is why we make noise,’ says a Tanzanian Muslim. ‘We fight for our [Muslim] rights.’

However, extremism is considered not to be Islamic. In both case studies Muslims and Christians say that ‘pious Muslims are tolerant people.’ Militant Muslims are not normal, as the quotation in the title of this article suggests. They are ‘foolish’ or ‘funny’. They are foreigners, ‘not from here’, and they represent ‘new waves’, not ‘what we used to be’. In Indonesia a Christian protects his Muslim friends by saying that: ‘In fact, they [extremists] are not from Solo, but from outside Solo, from Sukoharjo.’ Throughout the focus-group discussions the mechanism is the same. Participants maintain harmony and avoid conflict by creating clear-cut boundaries between ‘us’ and ‘them’, ‘we’ being ‘normal’ and ‘moderate’ Muslims and Christians, and the ‘others’ being ‘fools’ and ‘extremists’.

What we saw happening in our focus-group discussions is a hegemonic struggle between dominant and peripheral voices about what is considered to be ‘decent’ behaviour and a ‘civilised’ society, or a struggle to dominate the definition of what is considered to be ‘normal’. Dominant is the ‘official’ voice of those who reproduce the images of pancasila and ujamaa, of secular (in the case of Tanzania) or pluralist (in the case of Indonesia) societies. So, you are either a nationalist and thus pluralist or relativist; or you are a believer, and thus an extremist or potential terrorist. There seems no way in between.

At the interpersonal level, peripheral voices were immediately silenced and ridiculed in the focus groups. Extremists were classified as ‘fools’. And according to some participants, this is what is happening at the national (and inter-national) level: Muslims who say that they are ill treated because their rights are violated by their own government; who feel that they are treated as second-class citizens in their own countries; who blame their governments for being allies of the West in its War against Terror; and who fight for their rights by raising their voices are ridiculed, not taken seriously. On the rebound, extremist voices seem to become even more extremist.15 As one participant in Tanzania said, making use of re-wording and over-wording, repeating again and again the same phrase, ‘this is why we make noise.’ In other words, ‘this is why we raise our voices.’

Conclusions and discussion

The aim of a discourse-analytical study of rhetoric on Islamic extremism is not primarily to sort out which statements about reality are true or false, let alone to assess if extremists are good people or bad people in moral terms. Discourse analysis is not about right or wrong, but about the function of language use in the construction of social reality. It is hypothesised above that by silencing and ridiculing extremist voices, extremists tend to become more, not less extremist. And history shows that ‘terrorism has no military solution’ (Mamdani 2004: 229). But it was not the aim of this article to explain or discover causes of extremism. (For more on this see Maalouf 2000). The aim of this article was to demonstrate the use and usefulness of sociocognitive discourse analysis in religious studies. In our view, sociocognitive discourse analysis can help scholars of religion to do what they should do: not to study religion ‘in itself’, but to study why and under what conditions human artefacts and practices are labelled religious; to explore the boundaries of what is considered to be religious, or not religious; or to study how religion is constituted by and constitutes discourse (von Stuckrad 2010: 166).

Theory testing and case study

Case studies can be used to test or modify generalising theories. In our study this was social-identity theory. Tajfel and Turner (1986) discuss two prevalent explanations of inter-group conflict. One is that intra-individual and interpersonal processes lead to prejudice. The other is that competition between group interests cause inter-group conflict. Social-identity theory does not replace the real-group conflict theory but modifies it (Tajfel and Turner 1986: 8). In short, even if there is no incompatible group interest, there is in-group bias (Tajfel and Turner 1986: 15). Experiments show that mere awareness of the presence of an out-group is sufficient to provoke in-group solidarity and out-group hostility (Tajfel and Turner 1986: 13).

A problem with social-identity theorists is their conceptualisation of identity itself (Blommaert 1991: 18–21; Bourdieu 1991: 220–228). According to them, a social (e.g., ethnic, religious) identity is shared by the members of the group. It unites them (essentialism) and distinguishes them from others (exclusivism). In our case studies we found quite some in-group favouritism. But it was not mainly between religions but within religions. And the identity labels that were used were not based on primordial properties but on specific circumstances, and were described in different ways in concrete contexts. A ‘modernist’ Muslim can be ‘orthodox’ in the religious arena and ‘liberal’ in the socio-economic arena. As was shown by the analysis of the discourses about Noordin Mohammad Top's ‘disaster’ which brought a ‘blessing’, and the Solo Riots, which were just social and political, but did have religious and ethnic connotations, the boundaries between what is described as ‘militant’ and ‘moderate’, or ‘good’ and ‘bad’ Muslims, are not fixed but fluid.

Our participants said that pious Muslims are tolerant people, and that active worshippers show ‘great toleration towards other religions’. Thus (religious) in-group solidarity does not necessarily lead to (religious) out-group hostility. Our participants put it even more strongly: radicalism or extremism is ‘not normal’ but ‘excessive’, an ‘exception’ to the rule. Or, as one informant put it, it is a false belief, creating boundaries between ‘us’ and ‘them’. Normal is that Muslims and Christians tolerate each other and love each other as children of one father. A mechanism to avoid conflict is to classify extremists as being ‘not from here’; they are said to be strangers, foreigners. Or they are said to be ‘funny’ and ‘foolish’. By doing so, our participants position themselves as normal, peaceful, tolerant people, who get along with others quite well, visit them in their homes, and participate in their celebrations.

So the appropriate question would be, under what conditions do moderate Muslims turn into militants? Our participants also say that evil and violence are part of life. Muslim participants say that they do not legitimise terrorism. Yet the terrorists are their fellow believers, and brothers in the faith. Moreover, sightseeing in Surakarta at the house where the terrorists were captured by the police brings some benefit or profit. This shows that the boundary between peace and violence is fluid. In most cases people are able to balance good and evil, but sometimes it goes terribly wrong.

Few participants sympathise with extremists.16 They tend to do so when people are denied their dignity, are downgraded as second-class citizens or are ill treated. As one informant in Tanzania says, ‘This is why we make noise.’ In their perception, however, the main enemy is not Christianity but the government and the international community, or capitalism, which is a ‘kafir system’. ‘If the government bans the open air preaching there will be war,’ thus a participant in Tanzania.

In response to the main question of this article, if sociocognitive discourse analysis can be used and is useful in religious and interreligious studies, our answer is affirmative. But some controversial issues have to be explored further (Wetherell 2001: 383). These have to do with effectiveness, ontology and epistemology. How do you measure whether discourses have effects? Does religion exist? And what knowledge claim do sociocognitive discourse analysts make?

Realism or idealism

The main problem is to prove that language use has sociocognitive effects. Fairclough (1992: 238) acknowledges that it is difficult to justify or validate that language is constitutive. He nevertheless assumes that there is a dialectical relation between social reality and cognitive representation of reality (Fairclough 1992: 45, 60, 65). That is his critique of Foucault's ‘determinism’. His main concern is to demonstrate ‘explanatory connections’ (Fairclough 1992: 72, 80, 95) between text and context. But how can a scholar prove empirically that such a relation exists (Jørgensen and Phillips 2002: 89)? Fairclough (1992: 227–228) himself speaks about panels and interviews to augment the corpus with supplementary data. But are these enough?

Van Dijk (2008: 119) claims that there is no direct influence of text on context. Influence is exercised via mental models, being the interpretation of participants. Although Van Dijk (2008: 23) says that a ‘definition of contexts in terms of mental models does not imply that we reduce social influences to mental ones’, he nevertheless holds that ‘there is no such thing as an objective situation, unless I define such objectivity’ (Van Dijk 2008: 119).17 Against such a position, Fairclough (1992: 65) argues: ‘This could easily lead to the idealist conclusion that realities of the social world … merely emanate from people's heads.’ Counter to the idealist position, Fairclough, Jessop and Sayer (2004: 25) advocate a dialectic relation between social reality and its cognitive representation, and opts for critical realism.18 According to Bourdieu (1990: 189) the opposition ‘between a nominalism which reduces social realities, whether groups or institutions, to theoretical artefacts without objective reality, and a substantialist realism which reifies abstractions’ is false.

Fairclough (1992: 71, 86) also says that there is no direct relation between linguistic practice (text) and social practice (context). But in his theory text and context do not communicate through mental models but through discursive practices (interactions). We agree with Van Dijk (2008: 23) and Blommaert and Verschueren (1998: 39) that this relation is seldom explicated. We nevertheless see it as a challenge to overcome theoretical and methodical confusion and lack of clear rules and conventions (Granholm 2005: 249, 260) in discourse analysis and to improve the instruments to prove that text and talk have sociocognitive effects.

If scholars of religion assume that there is a dialectical relation between discourse and reality, they acknowledge that there is something beyond discourse, something that has its own ‘life’ and can turn itself against the speaker. For example, however constructed Hinduism or African religion may be, it would be difficult to deny that they ‘exist’.

Validity and reliability

Fairclough (1992: 193) mentions comparison as a way to prove socio-cognitive effects of language use, but he does not elaborate on this. Constant comparative analysis of various data sources and comparisons between data sources and knowledge sources (Strauss and Corbin 1998) could be used more systematically in discourse analysis. Constant comparative analysis implies triangulation: combining different sources, such as participant observation, focus-group discussion, ethnographic interviewing, collecting written and audio-visual materials, etc. For example, scholars of religion can compare churches that use business language with churches that do not. One can hypothesise that the former correlates with commodification of religion (Fairclough 1992: 207–215) and a ‘prosperity gospel’ and the latter does not.

The reliability of representations is complicated by the distinction between science and ideology, as well as between the discursive and the non-discursive (Jørgensen and Phillips 2002: 20; Wetherell 2001: 395–397). If discourse is considered to be constitutive, as in radical post-structuralism, there is nothing beyond discourse, that is to say, scholars cannot access reality outside discourse. In that case it is difficult, if not impossible, to distinguish between social reality and its cognitive representation, between good and not-so-good descriptions of reality (Jørgensen and Phillips 2002: 196).

Yet the purpose of any scientific research is to produce representations of reality that are as reliable as possible (Jørgensen and Phillips 2002: 182). How can scholars achieve this goal? First there is dialogical research, or close interaction between researchers and the researched. Next there is co-research or teamwork by researchers in the same field. Third, there is experimental and transparent report writing on the data so that others can criticise the claims made. And throughout all these strategies there is the on-going reflexivity of the researchers (Bourdieu 1990).

Interpretation and explanation

However, the term ‘validity’ comes from a different language game, namely that of positivist epistemology and quantitative research, which claims that scientific statements are true if there is a correspondence between reality and its representation.19 But if we accept that there is a dialectic relation between social reality and its cognitive representation, our representations are never mere reflections of reality. That brings us back to our contention that sociocognitive discourse analysis can bridge the gap between interpretation and explanation. Maybe the gap cannot be closed completely,20 but at least interpretation and explanation are no longer competitive but complementary (Fairclough, Jessop and Sayer 2004: 26, 38; see also Jensen 2009).

Fairclough (1992: 72–73, 85–86) combines three analytic traditions: linguistic pragmatics, micro and macro sociology. Not only is the relation between micro and macro level, otherwise called agency and structure (Van Dijk 2009: 250) dialectical, but the relation between the discursive and the non-discursive is also dialectical, and I agree with Fairclough (1992: 45, 65) in this respect. Fairclough (1992: 72) says: ‘I accept the interpretivist claim that we must try to understand how members of social communities produce their “orderly” or “accountable” worlds … ’ I would argue, however, that in so producing their world, members’ practices are shaped in ways of which they are usually unaware by social structures. That is not to say that sociocognitive discourse analysis solves all problems of theory and method in religious studies, but its prospects look promising and deserve to be explored further.

Notes

¹Fairclough (2001: 9) criticises cognitive psychology because it ignores the social origins of members' resources. He nevertheless endorses the notion that mental maps or social cognitions play a role in discursive practices and consequently in mediating between text and context. Because of the ‘sociocognitive nature’ of text production and consumption (Fairclough 1992: 71) I prefer the label ‘sociocognitive approach’ to discourse analysis (Van Dijk 2008), the ‘sociocognitive model’ (Breeze 2011: 494), or simply sociocognitive discourse analysis above critical discourse analysis.

²Fairclough (1992: 80) is not always clear about ‘sociocognitive aspects of production and interpretation’ of texts. As is argued later, it is clear that cognitions do not have a direct influence on reality. Fairclough (1992: 65) criticises the idealist position. Cognitive effects are mediated through discursive practices.

³For further explorations on social constructivism, otherwise called social constructionism, see Granholm (2013) and Moberg (2013).

⁴Keller (2005) claims that discourse analysis stems from linguistics and may be well suited for linguistic research. But it is hardly suitable to grasp macro dimensions of knowledge and power relations. To grasp these dimensions better, Keller draws more on sociology, sociology of knowledge in particular.

⁵Most discourse analysts take Foucault as their starting point (Jørgensen and Phillips 2002: 12–14). Whereas radical post-structuralist discourse analysts apply his principles stringently, critical discourse analysts criticise Foucault (Fairclough 1992: 56–61). According to them the relation between the discursive and the non-discursive is not deterministic but dialectical (Fairclough 1992: 66).

⁶According to Fairclough (1992: 71; see also 80, 81, 97), ‘production and consumption have a partially sociocognitive nature, in that they involve cognitive processes of text production and interpretation’. It is not clear what ‘partially’ means exactly and one can argue about the extent to which ‘cognitive processes’ have an influence. In our approach the influence is big. In this respect our approach is closer to the one of Van Dijk (2008: 119) who argues that influence is exercised through mental models. On the other hand, this influence must not be exaggerated, as will become clear in our conclusions.

⁷As was indicated above when I spoke about different names for methods, Fairclough's terminology is quite ambiguous and his methods are not explained well (Breeze 2011: 501, 504). His ‘three-dimensional conception of discourse’ (Fairclough 1992: 72) or ‘three-dimensional framework for discourse analysis’ (Fairclough 1992: 198) uses a ‘variety of theoretical perspectives and methods’ (Fairclough 1992: 99). He writes about different ‘dimensions’ of analysis in which he aims at different insights and in which he uses different concepts as tools or instruments. In his book Language and Power, Fairclough (2001: 20–21) distinguishes between ‘levels’ and ‘stages’ of analysis. The stages are called ‘description’, ‘interpretation’ and ‘explanation’ (see also Fairclough 1992: 198–199). In his terminology, analysis is not a stage after description, as is usually the case in qualitative research, but description is a stage of analysis.

⁸Fairclough (1992: 62) argues that his model is ‘suitable for use in social scientific research’. However, he limits himself to text analysis; ‘empirical studies are not included in this book’ (Fairclough 1992: 86). Fairclough is not so explicit about the procedures to be followed in social-scientific research. Some discourse analysts argue that we do not need more unification. They prefer discourse analysis to be ‘a loose theoretical framework’ and discourse analytical studies to be ‘designed individually’ (see the reference to Hjelm in the first section of Taira 2013). In our view, if discourse analysis is to be made suitable for empirical studies and to be taken seriously by social scientists, its procedures must be more standardised (see also Breeze 2011: 501). Repeatability is an important criterion for good research.

⁹The author is grateful to Thomas Ndaluka of the Department of Sociology and Anthropology, University of Dar es Salaam, Tanzania, as well as to Suhadi Cholil of the Centre for Religious and Cross-Cultural Studies, Gadjah Madah University, Yogyakarta, Indonesia, for conducting the focus-group discussions and generating the data on which this article is based.

¹⁰As Taira (2013) argues, some discourse analysts prefer naturally occurring data. In our research we use newspapers and audio-visual sources not only to link the micro and the macro level of analysis, but also to reduce our own contribution to the production of the material by comparing the data generated by us through group interviews and naturally occurring data given by these mass media.

¹¹In his book Language and Power Fairclough (2001: 133) equates the ‘members’ resources' that the people draw upon to produce and consume text with their ‘cognitive apparatus’.

¹²Julius Nyerere was the first President of Tanzania. He was a pious Roman Catholic and promoted a form of African socialism, called ujamaa (family ties). His successor was Ali Hassan Mwinyi, a Muslim who started a liberalisation process in Tanzania, referred to as ruksa (liberty).

¹³The five principles (pancasila), namely belief in God, nationalism, humanitarianism, democracy and social justice, were formulated by the first President of Indonesia, Sukarno, and serve as preamble to the constitution. Presently, syari'ah is seen as opposite of pancasila.

¹⁴Fairclough (1992: 72) aims to trace ‘explanatory connections’ between language use (text) and social reality (context), but his notion of explanation is a peculiar one. It remains unclear how effects of texts are measured in his model.

¹⁵This is also the finding of Mamdani (2004). In his view, the classification between ‘good’ and ‘bad’ Muslims is related to the ‘War on Terror’ after ‘9/11’ (p. 15). Classifications such as ‘Arabs’ and ‘Berbers’ in Morocco, or ‘Africans’ and ‘Arabs’ in Sudan, are constituted by political agendas (pp. 32–35).

¹⁶Case studies are never representative. Nevertheless our findings are confirmed by large-scale surveys such as the ones conducted by Esposito and Mogahed (2007) who found that fewer than 7 percent of the Muslims are sympathetic to radical ideas and that they are not specifically religiously motivated but are so politically.

¹⁷Van Dijk minimises the objectification in the social construction of reality (Berger and Luckmann 1966: 70–85). A certain level of ‘reification’ is unavoidable, and constructivism is never total. For this reason, Bourdieu (1990: 123) refers to his work as constructivist structuralism, linking the continuity of the habitus with the flexibility of the market.

¹⁸The opposition between realism and idealism (Fairclough 1992: 8–9) goes under various names. Bourdieu (1990: 124) speaks about objectivism and subjectivism. Whereas Blommaert and Verschueren (1998: 26–27) favour a materialist model, Van Dijk (2008: 118–119) opts for a mental model to explain the correlation between text and context. See also Fairclough (2005) on ‘critical realism’.

¹⁹According to critical discourse analysts, correspondence theory amounts to naïve realism. Of course, epistemology – or rather, the relation between ontology and epistemology – is the crux of the matter. According to Bourdieu (1990: 189) this is the age-old dilemma of nominalism and realism.

²⁰Wiebe (2005: 70, 79–80) is more optimistic about closing this gap. He pins his hopes on experimental psychology and the cognitive sciences to unify religious studies.