We begin this chapter by analyzing how evidence supporting witness testimony can be modeled as a kind of argumentation. This takes us back to the problems we encountered in Chapter 1 concerning the representation of corroborative evidence using argumentation technology. We continue the chapter by analyzing how argumentation that questions, attacks, or defeats arguments from witness testimony can be modeled. The second task is the more challenging of the two, because, as we have seen in the previous chapters, attacking witness testimony involves scripts and stories, and the kind of plausible reasoning used to support and to attack the arguments in them. In addition, we have seen that the engine for questioning in attacking witness testimony is the examination dialogue, and this type of dialogue has been so little studied in the literature on argumentation, artificial intelligence, and law that any attempt to apply it to witness testimony evidence is pioneering work. At present, the aim of much of this work is to develop systems of argumentation that might lead to applications in law in the not too distant future. However, because the theory of examination dialogue presented in this book is so new, even in argumentation theory, there is an additional task of showing how witness testimony can be formalized in such systems and implemented in computer programs for legal reasoning. The existing systems model arguments as sets of propositions, as premises and conclusions and arguments linked together to form chains of reasoning.

In this book, tools and techniques developed in argumentation theory and artificial intelligence are applied to problems of analyzing and evaluating argumentation used in law. Argumentation is a set of context-sensitive practical methods used to help a user identify, analyze, and evaluate arguments, especially common ones of the kind often found in everyday discourse. In the past it was the prevalent assumption that the deductive model of valid inference was the cornerstone of rational thinking. There has now been a paradigm shift to highly knowledge-dependent models of reasoning under conditions of uncertainty where a conclusion is drawn on a basis of tentative acceptance on a balance of considerations. Argumentation based on this new notion of argument, also called informal logic, is now being widely used as a new model of practical reasoning in computing, especially in agent communication in multiagent systems. Recent work in artificial intelligence and law has recently turned more and more to argumentation as a rich, interdisciplinary area of research that can furnish methods, especially in those areas of law related to evidence and reasoning (Bench-Capon, 1995; Gordon, 1995; Prakken, 2001a; Verheij, 2005; Walton, 2005). Generally, techniques and results of argumentation “have found a wide range of applications in both theoretical and practical branches of artificial intelligence and computer science” (Rahwan, Moraitis, and Reed, 2005, p. I). At the same time, artificial intelligence in law has coincided with the new evidence scholarship in law (Tillers, 2002).

The ad hominem, or personal attack, argument is now highly familiar in politics, especially in the use of negative campaign tactics in elections. This form of argument has been studied previously in the argumentation literature, but it has some special features of interest as a mass media argument strategy. In this chapter it is shown how it has some features comparable to appeals to fear and pity when it is used in mass media as a device of persuasion. This chapter brings out the multi-agent structure of the ad hominem argument as used in rhetorical argumentation, by showing how the proponent mounts a successful strategy in this form of argumentation. It is shown how he or she must use prolepsis by probing into the commitments of the respondent and configuring them in a certain way prior to the attack. These insights into the multi-agent structure of the ad hominem reveal how the proponent must collect evidence proactively, and then use this evidence to attribute a plan to the respondent. The aim is not just to reveal how to use such a tactic of personal attack in the mass media. The analysis is meant to be helpful to both voters and political campaigners, giving them a better understanding of how to deal with rhetorical ad hominem arguments by identifying, analyzing, and critically evaluating them.

The ad hominem argument is not a new phenomenon in American political discourse. A pamphlet was circulated telling of Andrew Jackson's “youthful indiscretions.

A recurring problem for the normative analysis and evaluation of mass media argumentation is the use of the term “propaganda.” One of the most common forms of attack on, or negative criticism of, rhetorical argumentation used in mass media is to label it as propaganda. What frequently happens is that arguments are automatically dismissed as irrational or fallacious, as soon as they are categorized as propaganda. This form of dismissal is especially evident in textbooks on logic and critical thinking, where forms of argument are frequently evaluated as fallacious using the term “propaganda.” In this chapter, I critically question such a policy of automatic dismissal of arguments used in propaganda. I will seek out a better method of evaluating such arguments, so that evaluations can be supported or refuted on a case-by-case basis by employing clearly stated criteria that can be used to assess the textual evidence given in the particular case. This method could be called an evidence-based approach.

One of the thorniest problems is to define the term “propaganda” or at least to grapple with the contradictions that appear in its current usage. One of the results of this chapter is a proposed set of criteria for the identification of propaganda as a type of discourse. Ten defining characteristics (as well as several other typical properties) of propaganda as an identifiable type of discourse will be set out and argued for.

In the kinds of media arguments examined so far, including those classified as appeals to fear and pity and as ad hominem arguments, the argument is partly based on and directed to popular opinion. In the type of argument called argumentum ad populum, as defined in chapter 3, the proponent tries to get the respondent to accept an opinion or perform an action because that opinion is accepted by the popular majority. Of course, much of what we do we learn by watching or following others. And there is a powerful urge not to be singled out or left out of the group. Thus, as already shown in the analysis of propaganda in chapter 3, this form of argument can psychologically be very powerful. But from an evidential point of view, it would appear to be extremely weak. For just because a large number of people accept some proposition, it does not follow that the proposition is true. We have all long been taught that such an inference is erroneous. We are all aware that science has proved many popularly accepted beliefs to be false. How then should one evaluate an appeal to popular opinion? Could there be some legitimate grounds for accepting such an argument in some instances, or is the commonly accepted view in logic right that this type of argument is fallacious?

The three fields of logic, rhetoric, and dialectic are all about arguments, as Aristotle showed, but each takes a different viewpoint on them. Logic is the science of reasoning that studies formal inferential links between sets of propositions designated as premises and conclusion of an argument. Dialectic, usually taken to be a branch of logic, analyzes arguments given in a text of discourse, including fallacious arguments, evaluating them as weak or strong by examining criticisms of them (Kapp 1942; Walton 1998b; Finocchiaro 2005, ch. 13). Rhetoric studies persuasive arguments based on the beliefs, commitments, or values of the target audience to be persuaded (Kennedy 1963; Tindale 1999, 2004; Jacobs 2000). However, the long history of the relationship between logic and rhetoric has been an antagonistic one, characterized by strife and sniping on both sides, beginning with Plato's attack on the Sophists on the basis that they took fees to teach argumentation skills. This attack on rhetoric is visible in many places in Plato's dialogues (Krabbe 2000, p. 206). Aristotle took a balanced view of what he saw as a close relationship between rhetoric and dialectic, but an opposition between the two subjects remained (Hohmann 2000, p. 223). Aristotle thought of dialectic as “a rather pure and theoretically sound method aimed at a cooperative search for cognitive truth” (Hohmann 2000, p. 223), and hence by comparison, rhetoric still had negative implications that are still present.

Appeals to fear and appeals to pity are two types of argumentation widely used in the media in political debates and advertising by advocacy groups, public relations firms, governments, and corporations. Johnson (2000, p. 269) has emphasized that mass media rhetoric, to be effective, needs to take the human emotions, in particular, fear and pity, into account. Both types of rhetorical argumentation can have a tremendous emotional impact on a mass audience, when presented in the right way. Mass media argumentation as a persuasive effort involves strategic maneuvering based on advocacy, audience adaption, and presentational devices, which are used to resolve a difference of opinion in one's own favor (van Eemeren and Houtlosser 1999b, 2000, 2001, 2002). On the other hand, both kinds of arguments are so well known to be subject to exploitation and manipulation that they have been traditionally classified in logic as fallacious. Recently, it has come to be recognized, however, that the traditional blanket condemnation is not warranted (Walton 1994). Appeals to emotion should be generally recognized as having legitimate standing as being, under the right conditions, reasonable arguments carrying some weight in shifting a burden of proof in a balance of considerations case where exact calculation of the outcome is not a practical possibility. But if appeals to fear and pity are sometimes rational arguments, how can we strike the right balance between recognizing their rhetorical power and the logical defects they admittedly have in some instances?

Stevenson's theory of the persuasive definition has been applied only to abstract ethical theory and to philosophical discussions. It has perhaps been seen as not too widely significant when confined to that framework. But its implications are of sweeping significance once it is applied to media argumentation. It is shown in this chapter how many media arguments are based on emotively loaded words or phrases that raise questions about the values, verbal classifications, and definitions of these terms. The persuasive definition, or as it is sometimes called in statistics, the friendly definition, is the definition that is stacked in such a way as to put a positive spin on the argument of the definer. This chapter shows how persuasive definitions are extremely powerful media tools, and how they are often used to influence public policy arguments. Five cases based on the use of persuasive definitions in public policy arguments are presented, analyzed, and evaluated below. In all of these cases, the use of the persuasive definition as an argumentative technique had consequences that changed social policies in a way that led to significant gains in the interests of some advocacy groups and losses in the interests of others.

Using these rhetorical case studies, some basic dialectical problems are posed about the legitimacy of persuasive definitions, about how to reconstruct the chains of argumentation they are based on, and about how to evaluate them.

The subject of this chapter is the meaning of the term “persuasion” as a speech act in argumentation theory. Terms like “persuading” and “persuasion” are pervasive and central in recent work in argumentation, along with closely related terms like “convince” and “convincing.” It is often said that the purpose of an argument, for example, is to persuade or convince someone to accept something as true. Wenzel (1990, p. 13) stated that “the purpose of rhetoric is persuasion.” The problem is to define exactly what is meant by the term “persuasion.” For as Wenzel pointed out (p. 13), simply stating that the purpose of rhetoric is persuasion “evokes all the negative connotations associated with both rhetoric and persuasion.” Both are then linked with deception and sophistical trickery. That line of thought was, of course, Plato's view of both rhetoric and the Sophists, as shown in chapter 1. In this chapter it will be shown that persuasion is a legitimate function of argumentation. What does this remark mean? It seems to imply that persuasion is a distinctive type of communicative act in dialogue, a speech act of some sort. From the previous chapters, many clues can be gathered about what persuasion is, within a multi-agent framework. Obviously, it involves some sort of change of opinion or acceptance of a belief, from an initial state to a new state that is the outcome of the act of persuasion.

Social statistics are needed to conduct intelligent public deliberations and set social policies in a democracy. But activists, the media, and private agencies can and often do use “mutant statistics” as tactics to manipulate public opinion. They can and often do convince people that even the most implausible claims are true by twisting the question wording in a poll (Best 2001, p. 4). But if such polls are based on scientific statistical methods, how is it that their results can be deceptive, misleading, and even deployed fallaciously to support bad arguments? One of the main problems is the use of natural language questions in the polls, containing words and phrases that have emotive connotations that subtly lead respondents toward one answer or away from another (Campbell 1974). This spin factor introduced by the wording of a question is not measured by any indicator of the numerical reliability of the poll typically published alongside the poll findings (Clark and Schober 1992). It is widely recognized that there are biases in scientifically accurate poll results due to the wording of questions used, and that inferences drawn from these biased polls can often be fallacious (Pinto, Blair, and Parr 1993). Critics such as Warnke (1990) and Crossen (1994) have pointed out that exploiting biased language deceptively to twist poll results in a desired direction has become a widely used technique in advertising, cause advocacy, and political push polls.

What initially led me to start work on this project was the observation that the examples of fallacies and questionable argument tactics studied in textbooks of informal logic often featured examples of advertisements and political arguments of the kind that have to do with elections or with public policies. Many of them are media arguments from sources such as political speeches, commercial ads, or Internet blogs. Such arguments are especially interesting when it is evident that they were used – for example, in ads – as rhetorically effective techniques to persuade a mass audience. Formerly (and often still), such arguments tended to be classified in logic as fallacious. But more and more they are now seen as fallible (but slippery) heuristics useful to reach a tentative conclusion under conditions of uncertainty, but subject to critical questioning. The theory put forward in this book strikes a judicious balance between analyzing them as fallible but basically reasonable arguments in some cases, and criticizing them as fallacious arguments used as tactics to unfairly get the best of an opponent or deceive a mass audience in other cases.

More specifically, the kinds of arguments considered throughout the book are ones often used in various communication media, including written texts, television, and the Internet, to attempt to persuade an audience to do something or accept something as true.

The cases studied in the previous chapters have suggested how dialectical and rhetorical methods need to be combined to reveal tactical argument moves and strategies underlying media argumentation. But the respondent-to-dialogue problem posed in chapter 4 remains unsolved. In this chapter, a solution to the problem is presented, using some technology from AI. We begin by reviewing the RTD problem and other unsolved problems brought out in the preceding chapters. These problems continue to be exacerbated by the troubled relationship between rhetoric and dialectic, and one problem is how to integrate the methods of each approach with the other. There is an inevitable tension between them, but paradoxically, one is not much good without the other when it comes to grasping how mass media argumentation works and getting better insights into both the strengths and weaknesses of this important type of argumentation. Solutions to these problems are proposed by putting forward several hypotheses that form a theory of how a proponent attempts to persuade an audience using argumentation across media.

The main elements of the theory are the following propositions. Both the audience and the arguer need to be seen as agents that share common knowledge that enables them to communicate. But some cases are more dialectically complex. In these cases, three agents are involved. Two basic types of media argumentation are distinguished. In direct media argumentation, only the proponent and the audience have the roles of agents.