ISSA Proceedings 2002 – Different Types Of Evidence And Quality Of Argumentation In Racist Pamphlets

1. Introduction
Every day we are confronted with numerous persuasive texts. Apart from the persuasive texts with a commercial purpose, once in a while we are confronted with a type of text intended to convert us to an ideology or a political stance. We may be encouraged to become a member of the foundation for the protection of the badger, to be sympathetic towards the squatters around the corner, to be against taking gravel out of the river Maas, or to expel all foreigners from the country. This paper addresses a type of text belonging to the group last mentioned; it concerns pamphlets coming from right-wing extremist groups(i). The texts may be written by right-wing extremist groups or by individual pamphleteers. In general, they are spread in the street (and, more recently, by internet).
The contents of these pamphlets can be described as more or less racist, in the sense of  “Treating (members of) a certain group in an unfavourable way, on the basis of their racial or ethnic origin” (Essed, 1985: 20). In most cases, the pamphleteers argue that unfortunate developments in Dutch society can be attributed to members of another race or another religion, and to the government protecting these people. The solution propagated usually consists in barring or expelling “strangers”, or in silencing the government. In general, one might say that the more outspoken and blatant the racist utterances, the harder to find out the identity of the sender.
Although it is not clear what target group the sender has in mind, it is clear that the pamphlets are not intended solely for the members of the group themselves. After all, the texts are of a persuasive nature, trying to persuade people to perform all kinds of actions.

1.1 The analysis of racist texts
There are many studies on prejudice and racism. Some of them are of a text analytic nature, and therefore relevant to this study. I will briefly discuss two of them.
Van Dijk (1992) investigates, amongst other things, racism and argumentation in tabloid editorials. This type of discourse functions to phrase the opinions of newspaper editors on prominent events. It adresses not only the audience, but also, directly or indirectly,  influential news actors, such as the press or politicians. On the grounds of an extensive analysis of two editorials of The Sun and The Mail, exposing the argumentation with respect to content (but not in a schematic way), Van Dijk concludes:  “ In other words, the argumentative structure of the editorials is not only a persuasively formulated opinion about the riots and involvement of blacks. Rather, the editorials have a broader political and socio-cultural function, viz., to argue politically for the control over black people, and for the reproduction of white dominance, that is, for white law and order, the marginalization of black community, the legitimation of white neglect in ethnic affairs, and finding excuses for right-wing racism and reaction ”. (p.258).
Mitten and Wodak (1992) provide another text-analytical study on racist texts. They analyse a letter sent by the anti-semitic politician Hödl to his jewish collegue Bronfman by the so-called ‘discourse-historical method’. The subject of the letter is the question whether the name of the Austrian president Waldheim should be added to an American ‘watch list of undesirable aliens’. Bronfman was in favor of adding him. The letter is scrutinized with the help of questions with respect to the construction of the story, the identity of the speakers, the occurrence of stereotypes, and possible evidence of racist opinions. The study shows that Hödl uses anti-semitic stereotypes in his letter to Bronfman, both of an etnical and of a religious nature. Like Van Dijk (1992), this analysis concerns content matter.
The present study makes use of a specific form of text analysis, which focuses on argumentation structure. Do right-wing extremist texts make use of arguments in support of their claims? If so, which type of arguments? Are they convincing?
Many people still have an intuitive aversion against racist texts, or at least they are aware of the fact that it is not politically correct to agree with racist views. So in fact, the claims made in those texts should be well supported in order to take away the aversion, or even to convince the reader of the point of view propagated.

1.2 Argumentation and persuasiveness
In the socio-psychological literature on the process of persuasion, two possible ways of getting convinced of an idea or the usefulness of an action are distinguished. According to the dual-process model, i.e. the Elaboration Likelihood Model (Petty and Cacioppo, 1986), one of those ways is the so called central route. If the reader is capable and well motivated, he will scrutinize the arguments and come to a balanced conclusion. The second way of reaching acceptance is less concerned with content and follows the so called peripheral route. Here, acceptance is reached with the help of heuristics, or with the help of peripheral cues, such as nice colors, music, pictures, etc.
The texts spread by right-wing extremist groups, generally are low-quality prints, printed on low-quality paper, and peripheral cues, like nice pictures, beautiful photos or nice colors, are sparse. This implies that the texts can only be succesful in convincing the reader by means of good arguments, for lack of peripheral cues. At most, one may be persuaded by the ‘many arguments heuristic’, stating a positive relation between the quantity of arguments and the acceptability of the claim (see O’Keefe, 1990).
Untill recently, most research focused on the effectiveness of peripheral cues and heuristics on the acceptance of a position (see Hoeken, 1998, for a survey). Persuasion via the central route, by balancing the arguments, was much less the trend. In 1995, O’Keefe already argued in favor of an “interchange between argumentation studies and persuasion effect research, with benefits in both directions”(p.16). These last years show a growth of interest for persuasion via the central route. The present study can be seen as an example of this change of focus.
Two recent studies are especially relevant as an introduction to this research. The first one is about forms of argumentation in persuasive texts (Schellens & de Jong, 2000). The second one is about argument quality (Van Dijk, 2000; Van Dijk e.a. in prep.).

1.2 Types of argumentation and argument quality
Before turning to the racist texts, let us take a look at the types of argumentation used in other types of persuasive texts. Schellens and de Jong (2000) report a study on types of argumentation in Dutch information texts, coming from government institutions or non-profit institutions, in some cases supported by the government. These brochures aim at encouraging positive behavior, for example, being physically active for half an hour a day, or discouraging negative behavior, for example, buying toys containing toxics.
Twenty brochures were analysed and the different types of argumentation were categorised according to the categorisation in Schellens and Verhoeven (1994). The form of argumentation occurring most frequently was pragmatic argumentation. This form consists in presenting advantages and disadvantages of the consequences of the encouraged behavior. The authors remark that the desirability in these argumentations often remains implicit: the arguments are supposed to be self-evident. Other types of argumentation occurring frequently were predictive argumentation and argumentation based on examples. All these types of argumentation are of a predominantly descriptive nature. The relation between arguments and claim is ususally based on co-occurrence or causality. Less often occurring types were: argumentation based on rules (a normative / prescriptive form of argumentation, advocating an action or a judgement) and argumentation from authority.

The authors concluded as well that the argumentation in information texts consists in giving information. Schellens and the Jong argue that this has a consequence for our way of thinking about processing via the central route. The reader taking the central route, must be able to reconstruct the arguments from the information given in these persuasive texts, in order to derive some kind of argumentation. And (s)he should be able to evaluate this argumentation. He could arrive at an evaluation by asking himself (some of) the questions belonging to the type of argumentation. This kind of questions is referred to as evaluation questions (Schellens & Verhoeven, 1994). They are defined for each type of evaluation and serve to determinate its quality and validity.
Van Dijk e.a.(in prep.) study the argument quality of the arguments used in Petty and Cacioppo (1986). O’Keefe (1995) already remarks that there is a problem with the concept argument quality as used by Petty and Cacioppo, and many others after them.   “The problem is that in this research ‘argument quality’ has been defined empirically, in terms of observed persuasive effects ” (p. 13). Firstly, argument quality was evaluated intuitively and subsequently, it was determined experimentally in terms of relative persuasiveness. This method resulted in a distinction between strong (persuasive) and weak (hardly persuasive) arguments, which were used in sequel experiments as a variable ‘argument quality’ in order to measure persuasiveness. In conclusion, as O’Keefe remarked: “argument quality is not defined by reference to some independent set of normative standards ” (p. 13).

In the analysis of Schellens and Verhoeven (1994), Van Dijk (2000, in prep.) found a method for evaluating the persuasiveness of argumentation. With the help of the relevant set of evaluation questions, one can distinguish reasonable and unreasonable argumentation, as regards content. Van Dijk analysed the texts used by Petty and Cacioppo by determining the types of argumentation and checking the evaluation questions corresponding to those types. In the results of her analysis, those arguments classified as ‘strong’ by the authors, indeed appear as significantly stronger arguments than those arguments classified by the authors as ‘weak’. Obviously, Petty and Cacioppo’s intuitions on argument quality are in accord with the valuation achieved by a systematic analysis concerning content, as provided by argumentation theory.
The texts used in Schellens and de Jong, were all well written and published professionally. One might assume that the argumentation is reasonable and coherent. In the study of Van Dijk, the texts were also written by experienced authors. In contrast, the study presented here analyses texts that are written by amateur authors.
The texts analysed were selected from a large body of texts, almost all of them provided by the Anne Frank Foundation. The phrasing of the central question is as follows:
1. which type of argumentation is used in right-wing exstremist texts?, and:
2. what would be the quality of the arguments?

2. Method
2.1 Material
A corpus of 17 right-wing exstremist texts was collected. They had to be overtly persuasive (that is, containg a claim), be brief (one page at most), and intended for a group larger than the organization producing the text. Generally, the texts are not dated (they may be very recent, but they may be very old as well), some by an anonymous author or one hard to trace, some written by an accessible political organization. The quality of language use and print varies from very bad to good. Two of the texts are illustrated.

2.2 Instrumentation
The method for analysis is based on Schellens and Verhoeven (1994) and Schellens and de Jong (2000). The method used in these works provides a delicate categorization of forms of argumentation. The basic assumption is that there are four ways to defend a claim/position. The first category of argumentation is referred to as ‘argumentation on the basis of regularity ‘. It is especially suitable for supporting claims or conclusions of a descriptive nature. A regularly occurring relation is argued, often of a causal nature. For example, in a predictive argumentation, a consequence is predicted on the basis of data presented as a cause. In an explanatory argumentation, a probable cause is reconstructed from a consequence. Both argumentation types argue for the probability of a consequence or a cause.
The second category consists of  ‘argumentation on the basis of rules’. A normative claim is defended in this type. Argumentation on the basis of valuation rules is an example of this category. It concerns evaluative judgements, in which the claim is an evaluative utterance about situations, people, objects or ideas, and in which the properties presented lead to the evaluative utterance. Argumentation on the basis of rules of conduct is an example of this category as well. The rules state when a certain kind of behaviour, or a certain action, is called for. In this type of argumentation, a certain situation is presented as justifying some action, some kind of behaviour, or some measure. This normative type of argumentation aims at the acceptation of the desirability of an action or a measure.
The third category is labeled ‘pragmatic argumentation’. This type argues for the desirability of an action or a measure, on the grounds of the desirability and probability of its consequences. Examples of this type of argumentation are: argumentation on the basis of advantage and argumentation on the basis of disadvantage. Two other subtypes of this type of argumentation, making use of (dis)advantages, are argumentation by weighing of alternatives and means/ends argumentation. In the latter type, a goal must be recognizable and the action proposed should be envisaged as leading to this goal.
Finally, Schellens and Verhoeven recognize three types of unbound argumentation: argumentation from authority, which argues for a claim by referring to an authority, that should be expert or qualified with respect to the present issue; argumentation by means of examples, which argues a claim by pointing to one or more examples which should be illustrative for the claim; argumentation by analogy, in which a claim is defended by pointing at a comparable case, that also warrants the conclusion. The types of unbound argumentation may defend both the desirability and the probability of a claim.

Schellens and Verhoeven (1994) and Schellens and de Jong (2000) provide a method of evaluating argumentations with respect to their content. The method consists in answering a set of evaluation questions. For each (sub)type of argumentation, a set of evaluation questions is defined, questioning the correctness of the data and questioning the sensibility of the relation between the data and the claim. Moreover, there is a set of type-specific questions, questioning certain characteristics of the argumentation type. For example, for argumentation from authority, we have questions regarding the actual statements of the authority, and his factual expertise in the matter. For argumentation by means of examples, we have questions regarding the typicality of the examples for the situation described in the current argumentation, and whether counter-examples rejecting the conclusion are available, etc. etc. Furtheron in the paper, I will return to examples of argumentation types and evaluation questions.

2.3 Procedure
In the first place, 17 texts were analysed in terms of the occurrence of different types of argumentation. An individual text could contain several types of argumentations, or several instantiations of one type of argumentation.
This method of assembling data differs from the method used by Schellens and de Jong (2000). They counted the occurrences of types of argumentation in all texts, without paying attention to the amount of occurrences of a certain type of argumentation in an individual text. The reason for their approach lies in the difficulty to decide whether an occurrence is part of one (complex) argument, or constitutes two separate arguments.
Indeed, this is a problem. I did not adopt the method of Schellens and de Jong because in general, right-wing extremist texts are not characterised by a nice, coherent structured. They usually are of a fragmentary nature. In order to do justice to the problem (without really solving it), I analysed each text in terms of a Toulmin-scheme. The arguments supporting one claim were counted as belonging to one type of argumentation. In four cases this approach was not possible: the arguments supporting the claim were of different types of argumentation. Overall, the amount of claims varied from 4 to 11 claims per text.

It is in place to be modest about the analyses. The texts were often badly written and badly structured. Therefore, in certain cases it appeared to be hard to decide which type of argumentation had been used. The analysis was based on an interpretation of the intention of the author. Thus, an apparently incoherent sequence of utterances could be part of an argumentation on the basis of rules of conduct, because they resulted in a call for action. Often, a causal relation was linguistically marked by structure markers, as in: “and that is why we say: stop the inflow of foreigners to in our country.” In conclusion, some of the choices for types of argument made here may be open to question. Two experts in the field of argumentation theory judged the choices.
In order to evaluate the soundness of the argumentations, for each individual one the corresponding evaluation questions were checked. In this process, a problem occurred, which is comparable to the one occurring in the process of classifying the argumentation types. Some facts cannot be evaluated in terms of truth because they are too vague. In those cases, the intuition of the analyser would be decisive. In this respect, the results of the evaluation questions are indicative rather than concrete.
Below, I will present a set of examples of the types of argumentation occurring in the texts. For one of the examples, the evaluation questions are answered for illustration purposes. I will start with an example of a predictive argumentation. It comes from a text by “Burger Belang Nederland”. The first two sentences are arguments for the claim in the third sentence.

1. Landverraders zeggen tegen vreemden: jullie hebben in Nederland internationale rechten. Ga dus naar Nederland en eis van de voormalige kolonialen je recht op. Voorts zegt de godsdienst van de vreemden: onderwerp de heidenen en vestig onze staatsgodsdienst aldaar. Vreemde culturen en geloofjes zullen ons land overspoelen, zoals eertijds de zee ons land overspoelde.
(Traitors tell strangers: in Holland, you people have international rights. So go to Holland and demand your rights from those former colonialists. Furthermore, the religion of those strangers tells them: subdue those pagans and establish our religion over there. Strange cultures and beliefs will overflow our country – as did the sea in former times.)

In this argumentation, a consequence is predicted (that strange cultures and beliefs will overflow our country), on the basis of data (traitors tell strangers to go to Holland to demand their rights and to subdue pagans in order to establish their religion). The evaluation questions corresponding to a predictive argumentation are:
1: Are there reasons to doubt the data?
2: Are the data really relevant for the causal relation appealed to?
3. Is the cause in the data generally sufficient for the plausibility of the following claim?
4: Are there any circumstances in this case decreasing the plausibility of the claim(ii)?

If the answers to question 1 and 4 are “no”, and the answers to question 2 and 3 are “yes”, then all questions are answered evaluation-positive (e-positive). In my analysis of example (1), none of the questions was answered e-positive. The data are debatable: do traitors urge strangers to demand their rights in Holland and to subdue the inhabitants with respect to religion (question 1)? Moreover, if it is the case that people advise foreigners (or strangers) to do something, does that lead to the result described in the claim (question 2)? And would the advice be a sufficient condition for having the country overflow and subdued (question 3)? Would not the Dutch asylum seeker policy have an influence on the situation? Would it not be possible that the ‘stranger’ does not wish to comply with the suggestion (question 4)?

From a text written by the Student’s Front Nijmegen, I took the following examples of an argumentation on the basis of rules of conduct
2. a means/ends argumentation
3. an argumentation on the basis of valuation
4. and an explanatory argumentation
5. The sentence expressing the claim is indicated by (C).

2. Deze uit de kluiten gegroeide sekte vormt een bedreiging voor de vrije Westerse samenleving. Wij vinden dat de Islamitische invloed in Nederland moet worden ingedamd, (C).
(This wildly growing sect is a threat to the free Western society. We think that the Islamic influence in the Netherlands must be stemmed (C).)
The action proposed in the claim, limiting the Islamic influence in Holland, is justified by the data that this sect is threatening Western society.

3. Als jullie van ze af willen dan zijn er geen duizend en een manieren, duizend en een middelen.
Het racisme! Racisme! Racisme!  Racisme! En niet een klein beetje, maar totaal en onbeperkt (C).
(If you want to get rid of them, then you won’t have thousand-and-one ways, thousand-and-one means.
Racism! Racism! Racism! Racism! And not some of it, but complete and unlimited (C).)
In 3, a plea for unlimited racism (the means) in order to ‘get rid of Surinam people’ (the end) is realised by an end/means argumentation. The end is expressed in the first sentence, and the means in the second one. In this case, the means is linguistically marked in the first sentence by “[not] thousand and one means”.

4. Hoe gevaarlijk is de Islam, te gevaarlijk vinden wij (C). Dit snelst groeiend staatsgeloof streeft naar een wereldoverheersing en is zeer onverdraagzaam tegenover andere culturen en mensen met een andere godsdienst.
(How dangerous is Islam really? – In our opinion, too dangerous (C). This fastest growing (state) religion, aims at ruling the world and it is very intolerant towards other cultures and towards people with other religions.)
In 4, the properties attributed to Islam, that it aims at ruling the world and that it is very intolerant towards other cultures and religions, lead to an evaluative claim about Islam. Islam is too dangerous.

5. De inmense(iii) migrantenstroom richting nederland bestaat voor het grootste gedeelte uit moslims die hardnekkig weigeren te intregeren. De multiculturele regering in nederland geeft de islam vrij spel. Ons land is daarom een makkelijke prooi (C).
(The enormous flow of immigrants towards the Netherlands mainly consists of Muslims who stubbornly refuse to integrate. The multi-cultural government in Holland gives the Islam free play. Therefore, our country is an easy prey (C).)
The consequence of our country being an easy prey is explained by a combination of data. The first one consisting in the fact that there is a large flow of immigrants refusing to integrate, the second one consisting in the fact that Dutch government gives Islam free play.
The texts vary with respect to the intensity of racism and hate towards foreigners. In order to demonstrate how intense these feelings can be, I will quote an argumentation on the basis of examples from the text “Amsterdam has got it ”, by an anonymous author:

6. Kijk toch eens rond in Amsterdam, waar joodse bestuurders en zwarte criminelen die straten onveilig maken en het zuiverheidspercentage van ons ras in snel tempo doen afnemen door middel van verkrachtingen en aanrandingen. Tezamen met de vele Turken, Marokkanen, Tamils en andere onvolksen zijn deze vreemdelingen een volksvijandige diepte-investering voor het voortbestaan van ons volk (C).
(Take a look in Amsterdam, where Jewish administrators and black criminals prowl around the streets and diminish the purity of our race at a high rate, by means of rapes and assaults. Together with those many Turkish, Moroccan, Tamil and other non-people, these strangers form the capital deepening of an enemy of our people (C).)
The claim (these strangers form the capital deepening of an enemy of our people) is supported by the examples of Jewish administrators and black criminals, prowling around and endangering the purity of our race.

3. Results
The number of types of argumentation per text, and the number of e-positive answers per type of argumentation were calculated. The added numbers are given in Table 1.
The number of e-positive answers per type of argumentation was obtained by dividing the number of possible e-positive answers by the number of e-positive answers found.
Argumentation on the basis of rules of conduct is observed most frequently in these texts: 31 occurrences. This is 32% of all argumentations (97) found. The number of e-positive answers for argumentation on the basis of rules of conduct is 30%. The low percentage is due mainly to the evaluation question about the credibility of the data and the social acceptability of the action proposed. Notably, in 16 out of the 17 texts, an action is defended which should put an end to the problems presented. The action may vary from voting for a political party to blatant racism.
Less often, we find predictive argumentation: 17 occurrences were found. This is about 17% of all argumentations in the texts. This type of argumentation has a low score on the evaluation questions as well: 25% was scored e-positive. In this case, the credibility of data and the acceptability of a cause-consequence relation are due to the low means.
There are 13 occurrences of explanatory argumentation. Notably, the percentage of e-positive answers (37%) was caused most prominently by various questions other than incredible data (42%). Those questions do not show a pattern. There were 12 occurrences of argumentation on the basis of valuation. Here, a negative evaluation was defended with respect to a group of people or to the government. The percentage of e-positive answers (38%) was caused mainly by incredible data and socially unacceptable rules.
There are 11 occurrences of means/end argumentation. The number of e-positive answers is relatively high: 45%. There are only 9 occurrences of argumentation on the basis of examples. Their score on e-positive answers is only 17%. This is mainly due to the fact that the examples are not typical of the class they are meant to exemplify and that counter-examples, which might threaten the claim, are never given. The low percentage is due not so much to incorrect data.
There are few occurrences of argumentation from authority and argumentation by analogy (2). There are no occurrences in this corpus of argumentation on the basis of advantage, disadvantage or weighing of alternatives.
The mean percentage of e-positive answers is 32%. In argumentation on the basis of rules of conduct, on the basis of valuation rules and means/ends argumentation, we find evaluation questions pertaining to social consensus, for example, “Are the consequences of action A socially acceptable?” Almost all the answers given by the analyser were e-negative. If the answers are given from the perspective of some group or analyser which turns away from social consensus, then the mean value of e-positive answers for all texts increases to 41%.
The argument quality is not constant over texts. For three texts, slightly more than half of the questions were answered e-positive. The lowest percentage (15% and 16%) was found in four texts. The rest of the texts have a score somewhere in between. We may conclude that the low mean percentage of e-positive answers is caused by all texts, rather than a few not well argued texts.

3.1 General comments on the results
Comparing the results to those in Schellens and de Jong (2000), we may note as a striking difference that the type occurring most frequently in their information texts, that is argumentation on the basis of advantages and disadvantages, does not occur frequently in our type of texts. Inversely, the type of argumentation occurring most frequently in the right-wing extremist texts, argumentation on the basis of rules of conduct, does not occur that often in Schellens and de Jong.
This difference might be caused by the difference in focus on the type of claim defended. Schellens and de Jong found the highest percentage of arguments based on advantages and disadvantages. Granted that this type of argumentation is based on desirability arguments, de researchers themselves note that the desirability of the consequences of the measure proposed often remains implicit. In the brochures, one does not argue for the desirability of being healthy and feeling energetic, but simply for the probability of reaching this desired status if one takes exercise for half an hour a day.
In right-wing extremist texts, argumentation on the basis of rules of conduct occurs most frequently. In this type of argumentation, the desirability of some action or measure is argued for. This is also the case in argumentation on the basis of valuation, a type of which we found 12 occurrences. Predictive and explanatory argumentations are probability argumentations. We found 30 occurrences in total. Safe for one, all of these argumentations occur as sub-arguments.
The means/ends argumentations contained either a desirability claim (9), or a probability claim (2). This pattern was inversed for the argumentations on the basis of examples: only 1 was a desirability argumentation, and 8 were probability argumentations. All argumentations on the basis of examples occurred as sub-arguments. In conclusion, about half of the argumentations found resulted in the desirability of a claim, while the other half resulted in a probability claim.
However, focusing on the final claims of the texts leads to another ratio. The final claims classified as probability claims in 2 cases and as desirability claims in 15 cases. One out of the two probability claims was a means/ends argumentation, arguing in favour of the sense in joining an organization. The other one was a predictive argumentation, resulting in a rather threatening prediction about the deplorable condition of our country in a few years. All other final claims pertained to the desirability of an action or a measure. Out of these, 8 were argumentations on the basis of rules of conduct, and 7 means/ends argumentations, arguing in favour of a means or an end.
With respect to argument quality, the results are stunning. Only 31% of the evaluation questions obtained a satisfactory or a positive answer. Judging very mildly, this percentage was 41% at most. This appears to be a very bad result – and so it is. The only standard for comparing these results are the results obtained in Van Dijk e.a.(in prep. 2002). Having analysed the texts used in Petty and Cacioppo, they judge 70% of the weak arguments as e-positive, and 89% of the strong arguments. Even compared to the weak arguments in the Van Dijk study, the percentage (41%) in the right-wing extremist texts is relatively low.

4. Discussion
We may assume that the reader will not process the right-wing extremist texts peripherally. The reason for this assumption lies in the absence of peripheral cues: there are no pictures (except for two small ones) and no colours. There is only text, and often badly written text at that, with a bad quality print and paper.
In the process of persuasion, involvement of the reader is an important factor. Consequence involvement of the reader increases his motivation to process the message via the central route, weighing the arguments (Chaiken et al., 1989). In relation to this, argument quality is important: strong arguments rather than weak ones will convince a highly consequence-involved reader. Value involvement of the reader is of importance as well. If the subject of the text strongly appeals to the values held by a reader (e.g. in relation to environment or religion), he will be influenced less easily than a reader for whom this is not the case.
We can distinguish three categories of readers as recipients of brochures spread by right-wing extremist groups. The first category consists of readers already sharing right-wing extremist convictions. The claims argued for are within their so-called latitude of acceptance (social judgement theory; Sherif & Hovland, 1961). In fact, for this group it is like preaching to the converted. This group can be described as having a high value involvement. The arguments will have little effect on them. This assumption is in agreement with the view of Simpson and Yinger (1985): being prejudiced may lead to a rigid attitude. Probably, readers within the range of acceptance will interpret these texts as a confirmation or strengthening of their attitude.
The second category consists of readers with another conviction. For these readers, a claim like ‘foreigners cause many problems and therefore they should be expelled’ lies within the latitude of rejection. This group must have a high value involvement as well: they must have an aversion towards right-wing extremist texts. Most probably, they will not read the texts. If they do, they will strive to find arguments rejecting the claims (Edwards & Smith, 1996).
For the third category of readers, the claims are within the latitude of non-commitment. Their attitude with respect to those claims is neutral. This category of readers is most promising for the authors of right-wing extremist texts: they may be persuaded to concur with the view or to join the organisation. However, a certain amount of consequence involvement is prerequisite for persuasion.
If readers process texts via the central route, they base their judgements about the claims made on the arguments given. According to the theory, they will adopt a positive attitude towards the claims if the arguments are strong. If the arguments are weak, however, they will end up with a negative attitude (Petty and Cacioppo, 1986; Eagly and Chaiken, 1993). If the theory is correct here, and if right-wing extremist organisations spread their pamphlets in order to increase the number of their members, the authors have two options. Either they should make it easy on themselves and forget about pamphleteering, or they should write texts containing good arguments, supporting their right-wing extremist views.

NOTES
[i] With thanks to Marije Mens for her valuable contribution to this study.
[ii] One question was left aside in the analysis. It is the evaluation question occurring as the last one in each list, questioning the certainty of the claim in relation to the aswers to previous evaluation questions.
[iii] Spelling mistakes in the examples were copied.

REFERENCES
Chaiken, S. (1987). The heuristic model of persuasion. In: M. P. Zanna, J. M. Olson & C. P. Herman (Eds.), Social influence: The Ontario symposium (Vol. 5, pp. 3-39). Hillsdale, NJ: Erlbaum.
Chaiken, S., Liberman, A. & Eagly, A. H. (1989). Heuristic and systematic information processing within and beyond the persuasion context. In: J. S. Uleman & J. A. Bargh (Eds.), Unintended thought (pp. 212-252), New York: Guilford.
Dijk, R. van (2000). De rol van argumentkwaliteit in het Elaboration Likelihood Model. Doctoraal scriptie. Katholieke Universiteit Nijmegen.
Dijk, R. van, Hustinx, L.G.M.M. & Hoeken, H. (in preparation). A normative and empirical approach to Petty and Cacioppo’s strong and weak arguments.
Dijk, T. A. van (1992). Racism and Argumentation: ‘Race Riot’ rhetoric in tabloid editorials. In: F.H. van Eemeren, R. Grootendorst, A.J. Blair, & C.A. Willard. (Eds.),  Argumentation Illuminated (pp. 243-259), Amsterdam: Sic Sat.
Eagly, A. H. & Chaiken, S. (1993). The psychology of attitudes. Fort Worth, TX: Harcourt Brace Jovanovich.
Edwards, K. & Smith, E. E. (1996). A confirmation bias in the evaluation of arguments. Journal of Personality and Social Psychology, 71, 5-24.
Essed, P. (1984). Alledaags Racisme (pp.5-50), Amsterdam: Sara.
Hoeken, H. (1998). Het ontwerp van overtuigende teksten. Wat onderzoek leert over de opzet van effectieve reclame en voorlichting. Bussum: Coutinho.
Hoeken, H. (2001) Anecdotical, statistical, and causal evidence: Their perceived and actual persuasiveness. Argumentation, 15, 4, 425-437.
Mitten, R. & Wodak, R. (1993). On the discourse of racism and prejudice. Folia Linguistica: Acta Societatis Linguisticae Europaeae, 27,  191-216
O’Keefe, D. J. (1990). Persuasion. Theory and research. Newbury Park, CA: Sage.
O’Keefe, D.J. (1995). Argumentation studies and dual-process models of persuasion. In: F.H. van Eemeren, R. Grootendorst, J.A. Blair, & C.A.Willard (Eds.), Perspectives and approaches. Proceedings of the third ISSA Conference on Argumentation (pp. 3-17), Amsterdam: Sic Sat.
Petty, R. E., & Cacioppo, J. T. (1986). Communication and persuasion: Central and peripheral routes to attitude change. New York: Springer.
Schellens, P. J. & De Jong, M. (2000). Soorten argumenten in de voorlichting. Tijdschrift voor Taalbeheersing, 22, 4, 288-308.
Schellens, P. J. & Verhoeven, G. (1994). Argument en tegenargument. Groningen: Martinus Nijhoff.
Sherif, M. & Hovland, C.I. (1961). Social judgement: Assimilation and contrast effects in communication and attitude change. New Haven, CT: Yale University Press.
Simpson, G. E. & Yinger, J. M. (1985). Racial and cultural minorities: An analysis of prejudice and discrimination (pp.3-27), New York: Plenum Press.