Book Reviews Book Reviews/Comptes Rendus 293 © Informal Logic Vol. 27, No. 3 (2007): pp. 293-299. Book Reviews A THEORY OF ARGUMENT. BY MARK VOROBEJ. Cambridge: Cambridge University Press, 2006. pp. x, 1-324. ISBN 0-521-67025-X, Pb. US $29.99 A Theory of Argument is an advanced textbook “written for upper-level undergraduate students who have completed at least one prior course in argumentation theory, critical thinking, informal logic, formal logic, or some other related discipline” (ix). This puts Vorobej’s book in a unique position since, to my knowledge, there are no other second course undergraduate textbooks with a non- symbolic focus. (Second course symbolic logic textbooks written for undergraduates, rather than primarily for graduate students, were relatively rare until recently; the past decade has seen a proliferation in such texts.) The book is divided into two parts. The first, comprising Chapters 1-3, is Macrostructure and the second, Chapters 4-6, Microstructure. Each chapter is broken down into various subsections and each subsection, with two exceptions, has its own set of exercises. The exercises are a mix of theoretically oriented problems such as “Suppose that argument A is cogent for person P and not cogent for person P′. Is it possible that, prior to carefully reflecting upon A, P and P′ could share identical epistemic states? Justify your answer” (102), and application oriented problems such as “On the assumption that each of the following passages expresses at least one normal convergent argument, identify the macrostructure and construct a diagram of that argument” (189). Chapter One, titled “Arguments”, lays out the basic framework of the text and defines key terms such as argument, author, audience, argumentation, intentional audience, social audience, propositions, inference, canonical form, static, noise, embryonic argument, principle of charity, etc. With the exception of a heavy emphasis on theoretically oriented exercises, this chapter, the main purpose of which is to make the reader understand what arguments are and to understand the initial steps in extracting arguments from argumentative texts, could easily appear in an introductory text. Vorobej begins his text by writing “An argument is a social activity, the goal of which is interpersonal rational persuasion” (3) and initially talks of arguments occurring. However, Vorobej, without disavowing this commitment to arguments as a sort of act, quickly moves to talking of arguments as composed of propositions and inference claims. It is the propositions and the inference claims that ultimately determine the identity of an argument and it is the set of propositions, in relation to a given argumentative text, that have a canonical form, i.e., a macrostructure, and the propositions that bear evidential relations to each other, i.e., a microstructure. In Chapter Two, “Cogency”, Vorobej presents his choice for an evaluative framework. According to Vorobej, a “cogent argument, we’ll say, is an argument by which you ought to be persuaded. More precisely, an argument A is cogent for some person P, within some context C, just in case it is rational for P, within C, to be persuaded to believe the conclusion of A, on the basis of the evidence cited within 294 A’s premises” (47). Vorobej then discusses four conditions, each of which is necessary and which are jointly sufficient for cogency, viz., having the rational belief that the premise set (i) contains only true members, (ii) is relevant to the conclusion, (iii) grounds the conclusion and (iv) is compact. Given that Vorobej is interested in the rational beliefs that the premises are all true and that the premise set grounds the conclusion, etc., Vorobej spends the next two sections articulating an understanding of rational belief based on Richard Foley’s theory of reflective stability—“if a belief in a proposition is rational for someone, then that person would eventually endorse that proposition after a process of due, or ideal, deliberation” (62). Vorobej then moves on to discuss ‘bad’ cogent arguments and ‘good’ non-cogent arguments; he compares cogency with other properties such as validity, reliability, soundness, trustworthiness, actual truth of the premises, etc. Vorobej finishes Chapter 2 with a brief discussion of the relationship amongst one’s epistemic states, which he construes quite broadly, the argumentative context, and the conditions of rational belief, and concludes that “our own rational beliefs are not privileged over anyone else’s” (105). In the last chapter of Part One, “Normality”, Vorobej introduces the concept of a normal argument as a means of making cogency relevant to the use of the principle of charity for extracting arguments from argumentative texts. According to Vorobej, “We’ll say that an argument A is normal, within a specific context C, just in case, within C, its author consistently believes A to be cogent both for herself and for all the members of her social audience” (111) and “we’ll assume that an argument is normal unless we have evidence to suggest it is not” (119). Given this normality assumption and the desire to represent arguments as conceived by their authors, Vorobej argues that it is appropriate to interpret argument strength in terms of argument cogency. He writes: “Our proposal to interpret argument strength as argument cogency is therefore restricted to contexts within which our primary aim is to represent an argument as it is conceived by its author, and where we have reason to believe—either directly or through an appeal to the normality assumption— that the argument in question is normal” (127). The upshot is that when extracting arguments from texts, we ought to prefer cogent arguments over non-cogent arguments (since the former are stronger than the latter). If we still have multiple cogent argument options to choose from, then more ‘objective’ criteria such as validity and reliability, etc., may come into play when trying to determine what the author’s argument is. Vorobej goes on to supplement the discussion of Chapter Two concerning what can be inferred about cogency on the basis of rational beliefs about validity and reliability. This chapter is the most theoretically oriented chapter of the book and most of the exercises are devoted to demonstrating theoretical understanding. Part Two of the text is devoted to argument diagramming, i.e., representing an argument’s microstructure—“the specific evidential relations that obtain between an argument’s propositional parts” (161). On Vorobej’s account of argument microstructure there are three significant types of arguments: convergent, linked, and hybrid. The three chapters of Part Two deal with each of these types respectively and each chapter comprises a much more sophisticated and detailed discussion of microstructure than is found in introductory texts. Though based on the standard nodes (for propositional parts of the argument) and arrows (for the evidential relations) model of diagramming, Vorobej has added Book Reviews/Comptes Rendus 295 more diagramming tools and gives each piece of the diagram a precise meaning so there can be no ambiguous diagrams. One significant and novel diagramming addition is the ability of the diagram artist to indicate disagreement with various claims of the argument author via squiggly diagrams (see section 4.5). So, for example, if we hold that it is not rational to believe a premise is true, then we can draw a squiggly node rather than a circular node for that premise, or if we hold that it is not rational to believe the premises ground the conclusion then we can draw a squiggly arrowhead rather than a normal arrowhead. The major theoretical divergence from standard accounts of argument diagramming is the introduction of what Vorobej calls hybrid arguments. On standard textbook accounts of argument microstructure there are five sorts of arguments, viz., convergent, linked, single, divergent, and serial (Walton 2006). Elsewhere I have argued that there is no issue with the notions of divergent and serial structures since such diagrams are merely diagramming cases where certain propositions are doing double duty (Goddu 2007a). Hence, serial and divergent ‘arguments’ are plausibly construed as multiple arguments represented via a single diagram and this is exactly how Vorobej construes them (see 180, 183). Vorobej also takes single arguments to be a special case of convergence, which, I have pointed out elsewhere, is purely a definitional matter (Goddu 2007b). Hence the real difference between the standard accounts and Vorobej’s is that where the standard accounts have two classes, viz., convergent and linked, Vorobej has three, viz., convergent, linked, and hybrid (see also Vorobej 1995). One virtue of Vorobej’s idiosyncratic account is, I shall argue, that it provides the tools to explain at least some of the conflicting and muddled accounts of convergence and linkage (see Walton 1996) that have appeared in various first course textbooks. Given an argument A with a premise set S and conclusion C in which S is relevant to C, Vorobej defines three major classes of argument as follows: convergent arguments are such that “each premise within S is independently relevant to C” (172-173); linked arguments are such that “each premise within S is a member of some linked set, i.e., with respect to C” (225), where “a set of premises S forms a linked set with respect to some conclusion C, just in case each of the following three conditions obtains: (i) S contains at least two members; (ii) S is relevant to C; and (iii) no proper subset of S is relevant to C” (225); and hybrid arguments are such that S contains at least one supplementation relation (272), where a “set of premises S supplements a set of premises S′, with respect to a conclusion C, just in case (i) S is irrelevant to C; (ii) S′ is relevant to C; (iii) S and S′ together provide a stronger reason R in support of C, which S′ alone does not provide; and (iv) S and S′ are the smallest sets yielding R that satisfy clauses (i), (ii), and (iii)” (276-277). On Vorobej’s account the evidential relations that are being mapped by diagrams are favorable propositional relevance relations. On standard accounts of mapping relevance relations, some premises are independently favorably relevant to the conclusion on their own (convergence) whereas some work together with others to form favorably relevant reasons (linkage). But the true merit of Vorobej’s approach is a recognition that there are multiple ways that premises might work together to form favorable reasons for the conclusion. Speaking more broadly than Vorobej, though for a purpose that should become clear shortly, irrelevant premises might ‘supplement’ each other to form a favorably relevant reason—this is Vorobej’s linkage; and on the other hand, irrelevant premises might supplement other 296 independently relevant premises or sets of premises to make even stronger reasons— this is Vorobej’s supplementation. Indeed, we might even say that independently relevant premises can supplement other independently relevant premises to form even stronger reasons—this corresponds to what Vorobej calls the ‘pooling’ of premises. But if we adapt Vorobej’s notion of supplementation to cover all the kinds of ‘working together’ Vorobej recognizes, then we can see how various authors come up with competing takes on what the convergent/linked distinction is. To make this clearer, consider four two-premise arguments with differing relevance/supplementation relations. A. The premises are independently relevant and do not supplement each other. B. The premises are independently relevant and do supplement each other. C. One premise is independently relevant and one is not and the latter supplements the former. D. Both premises are independently irrelevant but supplement each other to form a relevant reason. For those who take any supplementation as a sign of linkage, then A only will be convergent and B through D linked. For those who take supplementation by irrelevant premises as a sign of linkage, A and B will be convergent and C and D linked. For those who hold that each premise in a linked set needs to be irrelevant, A through C will be convergent and D alone linked. Hence, given Vorobej’s framework, we can see why so much debate has raged over the linked/convergent distinction, especially over cases falling into the B and C categories—a neat two-fold demarcation is trying to be imposed on something with more options. Vorobej’s more sophisticated account is able to capture and represent these options. Vorobej’s text is a careful and meticulous work clearly suitable as a second course book. It is even plausible as a graduate level text. However, the text does have various shortcomings. For example, this text spends almost no time on the issue of distinguishing non-arguments from arguments; instead it focuses on extracting arguments from argumentative texts and extracting them with fidelity. Perhaps the assumption is that a first course has prepared the students to identify argumentative texts; but then again one might argue that identifying arguments is much more challenging than is often presented in basic texts and so is worthy of further study. In addition, the text is heavily skewed toward accurate and careful representation of arguments as their authors conceive of them with significantly less time devoted to argument evaluation or criticism. Much of what might be construed as concerning argument evaluation ultimately seems more related to determining various measures of argument strength in order to apply the principle of charity fairly when reconstructing and presenting an author’s argument. Direct discussion of the evaluation and criticism of arguments is minimal in the text. Those who are inclined toward objective evaluation criteria such as validity, reliability, true premises, etc., will be frustrated by the terminal focus on cogency and the attention paid toward whether it is rational to believe the premises are true or whether the argument’s author is being epistemically non-culpable or not. Even granting that paying attention to cogency questions is relevant, especially when trying to fairly represent an author’s argument, one might still want, for example, to know whether the premises actually ground the conclusion or not, and not merely whether it is rational for the author and his or her audience to believe so. For example, if we know that applications of the normalcy assumption and what we Book Reviews/Comptes Rendus 297 know of the author’s epistemic state dictates attributing to an author an invalid argument in a context that requires validity, we can reject that argument on the grounds that it fails to meet the required standard regardless of whether the argument is in fact normal and cogent or not. Even if we restrict ourselves to Vorobej’s four cogency conditions, there seems a more elegant set of criteria available. Given Vorobej’s interest in rational acceptance of the conclusion based on some body of evidence, one can easily argue that rational belief in (a) the truth of the premises and (b) the premises grounding the conclusion, is sufficient for rational acceptance of the conclusion. So what of compactness and relevance? Compactness, “each proper subset of its premise set S provides less evidential support for A’s conclusion than does S itself” (51), is meant to force each premise in S to be doing some necessary work. But surely the inclusion of redundant, but relevant, premises does not suddenly make belief in the conclusion irrational. In addition, the inclusion of compactness conflicts with another of Vorobej’s measures of argument strength—i.e., vulnerability. Given an argument in which the premise set S is relevant to A’s conclusion, then A is vulnerable just in case “there is at least one premise P within S such that the elimination of P from S would destroy all relevant support for C”; hypervulnerable just in case “the elimination of any single premise P from S would destroy all relevant support for C”; and invulnerable just in case “there is no single premise P within S such that the elimination of P from S would destroy all relevant support for C” (247). But given a preference for invulnerable arguments, we should also prefer non-compact arguments since they are less likely to be vulnerable (and cannot be hypervulnerable). Hence, why Vorobej insists on compactness as a cogency condition is not clear. The relevance condition is redundant. Since, as Vorobej points out, a set of premises that grounds the conclusion must be relevant to the conclusion, it must be the case that the truth, grounding, and compactness conditions by themselves are individually necessary and jointly sufficient for cogency. If one accepts the non-essential nature of compactness for rational acceptance of the conclusion, then truth and grounding become the two relevant criteria for argument cogency— but then we are quite close to a standard take on argument goodness. But if relevance of the premises is redundant given the nature of grounding, why include it as one of the four necessary cogency conditions? The reason, I suspect, is the second half of the textbook—the half which concerns diagramming various relevance relations. Here, however, we come to the most significant shortcoming of Vorobej’s text—why bother with microstructure at all? I have already acknowledged that Vorobej’s account of argument microstructure is much more sophisticated than any account available, but it too suffers from serious shortcomings. Firstly, and unsurprisingly, given the book’s emphasis on representing arguments in terms of how authors conceive of their own argument, Vorobej holds that “the purpose of diagramming an argument is to arrive at a graphic representation of the evidential relations that are claimed, by the argument’s author, to obtain among its propositional components” (164). So anyone who thinks we ought to be diagramming the actual evidential relations rather than the evidential relations as conceived by the author will once again be frustrated. Secondly, because Vorobej (i) relies on an intuitive notion of propositional relevance that is relative the each individual’s epistemic state and (ii) acknowledges 298 that “we have no grand theory to offer by way of identifying the macrostructure of enthymemes, or for determining whether an unexpressed proposition does or does not function within an argument as a premise” (171), he demonstrates how one and the same argumentative text can be reasonably construed, even from the author’s point of view, as any of the three argument types. He goes on to suggest that this is not a problem because “the different interpretations do not affect the strength of the argument’s grounding claim” (302). Indeed, taking various disparate parts of Vorobej’s text we can see that an argument’s microstructure is largely independent of the nature of the argument’s grounding claim. But if this is so, why is it important to determine an argument’s microstructure at all? The closest Vorobej comes to an answer is the following: Our interest in classifying arguments, according to their structural properties, transcends any mere taxonomical concern. The greater the number of structural options open to an author, the more flexibility she has, in principle, to respond to charges that her argument fails to be cogent. And audience members may all too hastily lay these charges precisely because of a lack of appreciation on their part of the different ways in which an author may conceive of the microstructure of her own argument. (308) But far from supporting the significance of determining argument structure or even argument structure as conceived by the author, this passage seems to suggest that authors are better off if they leave their arguments ambiguous as to argument structure so as to leave more defense options open. Of course, what is not at all clear is how different argument structures leave different defense options open, since, as Vorobej points out several times, cogency cannot be read off argument diagrams regardless of structure. So the question remains: why should we bother trying to map out the relevance relations at all? Vorobej has not provided an answer to this. Hence, I am hard pressed to see why one should spend half a course (given Vorobej’s recommendation for the use of his text (ix)) teaching students how to map out these relations. A Theory of Argument stresses the importance of listening to arguers and representing arguments as conceived by their authors. Without doubt this is good advice and certainly something students of argumentation need to practice. Vorobej’s text excels in this regard. But beyond the extraction and presentation of arguments as conceived by their authors, Vorobej’s text is much more tenuous. Those who are also interested in the objective representation and evaluation of arguments over and above the author’s own conception of his or her argument will need to look elsewhere. By focusing on the relevance relations, Vorobej sidesteps detailed discussion of one of the most important issues of argument theory: understanding how premises ground (or fail to ground) the conclusion.Vorobej’s account of microstructure fails in the same way that all accounts of microstructure fail—it provides no reason why determining an argument’s microstructure is at all important for those concerned with the presentation and evaluation of arguments. Book Reviews/Comptes Rendus 299 References Goddu, G.C. (2007a). “Walton on Argument Structure”. Informal Logic 27, pp. 5-25. Goddu, G.C. (2007b). “Against Making the Linked-Convergent Distinction”. Proceedings of the Sixth Conference of the International Society for the Study of Argumentation, University of Windsor, June 2007. CD-ROM. Vorobej, Mark. (1995). “Hybrid Arguments”. Informal Logic 17, pp. 289-296. Walton, D. (2006). Fundamentals of Critical Argumentation. Cambridge: Cambridge University Press. Walton, D. (1996). Argument Structure: A Pragmatic Theory. Toronto: University of Toronto Press. G.C. GODDU University of Richmond This is a dictionary for students of logic. It covers a range of logical and logic- related topics which may be put in these four categories: (a) brief surveys of logical systems like propositional logic, quantificational logic, belief logic, deontic logic, mereology, modal logic, etc., (b) explanations of common logical terminology such as argument, contrapositive, entailment, fallacy, validity, etc., in addition to (c) very short sketches of the contributions of important logicians like Aristotle, Barcan Markus, Boole, Frege, Kripke, Quine, and Russell, and (d) some entries indicating the role of logic in fields outside logic proper, e.g., logic and God, logic and mind, logic and computers, and formal ethics. I think the entries are all reasonable, as far as they go. Compared with Kneale and Kneale’s The Development of Logic (1962), Gensler’s Historical Dictionary is much less ambitious. The Kneales have much more detail and they often go deeper—but theirs is not a dictionary and since their interest is in logic in the narrow sense (see below), they give us only a history of logic in the narrow sense. Boole’s work, for example, gets over 10,000 words with the Kneales but only about 550 words in Gensler. Frege gets at least 20,000 words from the Kneales and about from 550 words, plus his own one-page bibliography, in Gensler. However, Gensler’s interests, as mentioned, range wider than do the Kneales’, and he has the advantage over them of being able to report on the additional growth of logic over the last forty-five years. Unlike the Kneales’ history, Gensler’s Dictionary is extremely accessible and convenient for someone who just wants to know in outline what, say, deontic logic, or the continuum hypothesis, is. There is something eclectic about Gensler’s Dictionary that invites us to compare it with Quine’s Quiddities (1987). In addition to what we expect (both books contain HISTORICAL DICTIONARY OF LOGIC. BY HARRY J. GENSLER. Lanham, Toronto and Oxford: The Scarecrow Press, 2006. Pp. xxvii, 1-306. ISBN 13 : 978-0-8108-5531-1, Hb. US$70. © Informal Logic Vol. 27, No. 3 (2007): pp. 299-302.