Critical Study INFORMAL LOGIC XIV.2&3, Spring & Fall 1992 Dialectics and the Macrostructure of Argument by James B. Freeman ALEC FISHER University of East Anglia Freeman, James B. (1991). Dialectics and the Macrostructure of Arguments: A Theory of Argument Structure. Berlin! New York: Foris Publications (a division of Walter de Gruyter & Co.). Pp. xiv, 1- 273. ISBN 3-11-013390-3. Cloth. DM138/ US $98.60. 1. Introduction. This excellent book is an essay in logi- cal theory which, unusually for such es- says, belongs to the informal logic and argumentation tradition. The central notion of modern deductive logic is that of 'logi- cal form' , so most writings on logical theo- ry are concerned with the internal logical structure of sentences-what Freeman calls the 'microstructure' of argument. Freeman, on the other hand, is mainly con- cerned with how whole sentences function in natural language argumentation, with the kinds of support sentences can offer to other sentences and with the structure of those support relationships; this is what he means when he talks about the 'mac- rostructure' of argument. (p.xi) The tradition to which Freeman's book belongs begins with Beardsley's Practical Logic, and includes as major contributions Toulmin's The Uses of Argument, Scriv- en's Reasoning, Govier's A Practical Study of Argument, Thomas's Practical Reasoning in Natural Language, Johnson and Blair's Logical Self-Defense and Freeman's own Thinking Logically. This, the informal logic tradition, is a reaction against the view that modern deductive logic generally applies to real arguments (arguments of the kind people actually use in order to convince others), and its main objective is to "develop means of analyz- ing and evaluating arguments in natural language" (p.33). A key technique in this tradition is argument-diagramming-using diagrams to display the structure of natural language reasoning, to display how the reasons combine to support their conclusions. Beardsley first introduced this technique, and he and Thomas introduced the associ- ated distinctions between 'divergent' , 'convergent', 'linked' and 'serial' reason- ing which are now so familiar in this tradi- tion. As is well known, Toulrnin's book The Uses of Argument uses quite different fundamental categories from the other works cited above and uses quite different sorts of argument diagrams. Where the other works simply distinguish between the reasons and the conclusion(s) of an ar- gument and construct 'tree' diagrams like this, Basic reason I Basic reason 2 ~~ Intermediate conclusion + Basic reason 3 - ... Main Conclusion Toulmin distinguishes among reasons in terms of how they function in the argu- ment; he distinguishes between data, war- rants, backing, rebuttals (and modalities) 194 Alec Fisher and draws argument diagrams which look like this, IBacking! ~ I Warrant I W IGroundl~ !ModalitYI~ IConclusion I IReb!ttal! Freeman has taken this conflict within the informal logic tradition very seriously and has addressed many of the issues it raises in arriving at his own position. His own position is a fascinating synthesis, drawn partly from Toulmin and partly from what he calls the 'standard view' (the tradition of Beardsley, Thomas, et al.). In arriving at his own position, he has written what is certainly the most scholarly assess- ment of the conflict between Toulmin and the 'standard view' which has been pro- duced so far. In doing this he has written an excellent critique of the details of Toulmin's analysis of argument structure whilst also accepting Toulmin's general contention that we need a dialectical ap- proach to correctly understand argument structure. Besides his critique of Toulmin, he has also written the most thoroughgoing assessment yet to be found in the literature of the problems associated with the 'linked-convergent' distinction. Both these contributions to the development of the field will be essential reading for informal logicians for many years. This is no less true for Freeman's positive, dialectical the- ory, which provides the context for these discussions. 2. What is the Central Goal of Informal Logic? Before entering into the details of Freeman's arguments, I want to raise one question, which I feel must always be kept in sight in this field, and this is 'What is the point of informal logic theory? What is the objective? What problems are we trying to solve?' And the answer is that we want to know how natural language argumentation works especially because we are interested in being able to improve argumentation skills through teaching. As Scriven puts it at the beginning of Reasoning, the first aim of the book is "to improve your skill in analysing and evaluating arguments and presentations of the kind you find in everyday discourse . . . textbooks, and lectures." Freeman himself puts it a bit differently (and doesn't mention the teaching interest): A central goal, if not the central goaL of infonnal logic is to develop means of appraising arguments of ordinary language. (p.xii; see also p.33) Most people who work in the informal log- ic tradition do so because (i) they value Scriven's objective and (ii) they have also realised that teaching formal logic is inef- fective to achieve it (though it is an excel- lent basis, for example, for work in the field of computing and information scienc- es). It is important not to lose sight of this goal, and it would be ironic indeed if infor- mal logicians did so. Unfortunately, it is easy to do this, to become Scholastic and to produce theories which are very com- plex a priori constructions (and hard to fault in terms of their internal consistency) but which are just as ineffective for this purpose as was formal logic-just as inap- plicable to real everyday argumentation as was formal logic. Freeman's theory becomes so complicated that it risks being inapplicable in just this way. 3. The 'Convergent-Linked' Distinction. However, let us leave that thought aside for the moment and attend instead to the details of some of his arguments. The heart of his theory is a method of diagram- ming arguments, so let us begin with some of the elements in his diagrams. Thomas introduced the distinction between 'linked' and 'convergent' reasoning, and though it seemed very natural at first, it has generat- ed endless controversy. Roughly speaking the idea is that two reasons for a conclu- sion are convergent when each gives sup- port to the conclusion independently of the other, whereas if the two reasons have to be taken together for either to be seen to be support for the conclusion, then they are linked. Unfortunately, this rough distinc- tion proves to be difficult to render precise. Consider the following simple example of reasoning: Mary will pass the examination well be- cause she is clever and she has worked hard. She is also lucky in having no complicating factors. like ill-health, exami- nation nerves or difficult personal relation- ships, so I'm confident she will do welL Among writers in this field intuitions vary about whether this is convergent or linked reasoning. Freeman gives an excellent ac- count of the difficulties in explicating the linked-convergent distinction (p.9ff.). Much later he explains his own notion of linked argument structure as follows: Premises are linked when we need to take them together or they are intended to be taken together to see why we have a rele- vant reason for the conclusion. (p.94) But he also praises, Thomas's final and much clearer criterion, that only when a reason would support the conclusion just as well even if the other reasons were false is the argument convergent-otherwise it is linked. (p.1O I) Well, is our previous example convergent or linked? I think it is reasonably clear (from much else that he says besides the quote above) that Thomas would regard this as. linked reasoning (as I would-see Fisher (1988) and as van Eemeren and Grootendorst would-see their (1987), to cite just two other examples). But I think it Review of James B. Freeman 195 is also clear that Freeman would see this as convergent reasoning. Freeman may be right about this; I am simply not sure. He does believe that his dialectical perspective yields a "very clear-cut demarcation of linked from convergent arguments" (p.108) but he is also honest about the fact that this demarcation "is not without controversy" (p.108). What is controver- sial about his conception of the linked- convergent distinction comes out most clearly in his account of inductive generalisation-which he sees as conver- gent, not linked as most authors do. (On this see p.102ff.) However, he propos- es an interesting and plausible solution to the problem on p.160, where he argues that such arguments are convergent with rebuttal conditions). Freeman is very well aware of the ob- jections to his account and he articulates them very clearly; Our account here conflicts with three dif- ferent intuitions. First, if several premises presented to support some one conclusion are individually weak but together make a stronger case, it would seem intuitively ap- propriate to link them together to represent how they support the conclusion. Secondly, and closely related, the fact that an arguer has presented several less than deductively strong reasons for one claim which appar- ently augment one another would intuitively suggest that linked structure appropriately represents the arguer's intentions that these premises be taken together. Finally, if the falsity of one premiss would undercut the force of the remaining premises, intuitively it would seem that their connection is clos- er than mere convergent, and so they should be linked. (p.102) I am not clear whether Freeman answers these objections to his own position, but I do think there is an underlying confusion in the minds of many people who write in this field, which may linger on in Free- man's mind; it concerns what the arrow be- tween reason and conclusion means in the standard tree-diagramming approach. Some people take it to mean 'therefore' 196 Alec Fisher (e.g. Thomas) and others take it to mean 'gives some support to' (eg. Freeman). Freeman is aware of this distinction (see his discussion on p.13) but there is an in- teresting confusion in his last Appendix, p.262, where he suggests that van Eemeren and Grootendorst have "clear analogies of convergent, linked and serial structure" and implies that they are like his. However, this is a mistake; their conceptions are quite different from Freeman's and much more in tune with 'standard' intuitions than Freeman's. This is mainly because they take the arrow to mean "therefore" not "gives some support to" as Freeman does (see p.96). This and other comments of Freeman's make me wonder if there is a lingering unclarity in his mind about the use he makes of arrows in his diagrams- which shows up in the controversy about the linked-convergent distinction. Whether this is the case or not, and whichever intui- tions eventually prevail here, there is no doubt that Freeman's discussion of the is- sues is meticulously researched and very thoroughly argued; it will remain a classic in the field for many years. 4. Freeman's Critique of Toulmin. Let us tum now to Freeman's response to Toulmin's theory in The Uses of Argument. Most writers in the field have rejected Toulmin's data-warrant-backing distinction as unworkable; taking exam- ples other than Toulmin's, they have been unable to decide which reasons are data, which are backing and which are warrants; having rejected the heart of his theory most writers have rejected the overall approach too. Freeman's response is quite different. He accepts the broad outline of Toulmin's approach; indeed Toulmin's dialectical conception of argument is the inspiration for Freeman's whole approach. One result is that, unlike most writers, Freeman gives an excellent, detailed and sympathetic ac- count of how Toulmin divides reasons into data, warrants, backing, and rebuttals (per- haps Freeman is able to do this precisely because he is so sympathetic to the general approach). However, he too rejects the data-warrant-backing distinction, but he does so after the most sustained critique of Toulmin's notion of 'warrant' which I know in the literature. He considers very carefully Toulmin's own arguments for ac- cepting that there are warrants and rejects them. He then considers whether Ryle's conception of conditionals as 'inference- tickets' can provide a satisfactory ration- ale, but again he rejects the idea. Finally he considers whether J. S. Mill's view that universal affirmative propositions are 'memoranda for our guidance' can serve; but when this fails too, he concludes that we must give up on 'warrants' (and having rejected-warrants, Freeman then rejects the associated notions of 'backing' and 'data' for related reasons, though see p.88 for some qualifications). Toulmin's notion of a 'warrant' has been widely criticised be- fore, but the links which Freeman makes with Ryle and Mill build bridges to stand- ard philosophical discussions of condition- als, and ensure that this discussion (mainly in chapter 3) will be of much broader inter- est to philosophers and logicians than most criticisms of Toulmin. 5. Freeman's Dialectical Conception of Argument. Let us now move to Freeman's main contention. It is that one can only under- stand arguments as products through un- derstanding the challenge-response process of arguing. On what we have been calling the 'standard' view, one defines an argument as a set of sentences some of which, the reasons, are presented as supporting others, the conclusions. Thus an argument is the kind of thing that one person can present, it does not essentially require dialogue between two (or more) persons; this conception of argument is what is called the 'monological' concep- tion. But Freeman argues that one can only understand and evaluate this kind of argu- ment by seeing it as derived from a process of dialogue; We see arguments generated through a challenge-response dialogue where the proponent of some thesis answers critical questions posed by a challenger. We may view arguments in the monological sense as products of such dialogical exchange processes. We thus accept the process/ product distinction for argument put forward by a number of authors. And we agree that process is fundamental to understanding argument. Indeed, the fundamental thesis of this book asserts that we can properly motivate and understand the structure of arguments as products through considering the various challenges which may arise in basic dialectical situa- tions, arguments as process. We shall thus present a dialectical theory of argument macrostructure. (p.xiii) The model here is, as with Toulmin, the way the law works in a criminal trial. There is a process of claim, challenge, response and counter-response, which eventually culminates in the judge's sum- ming-up. The summing-up is the argument as product, but the product cannot be un- derstood and evaluated, on Freeman's view, except by reference to the process which produced it. Toulmin described this as replacing the 'geometrical' (monologi- cal) model of argument by the 'jurispru- dential' model. On the jurisprudential model, one cannot evaluate the argument (as product) without knowing that the cor- rect procedures were followed and, more importantly, that the right questions were asked. Of course, in a situation where argu- mentative dialogue is taking place, the parties need not be interested in resolving their conflict by rational argument (they may just be quarrelling). Freeman has no particular interest in this kind of dialogue; he is mainly concerned with • dialectical' situations-dialogical Review of James B. Freeman 197 situations in which the participants aim to resolve their conflict through rational discussion. To use van Eemeren and Grootendorst's terminology, the participants are engaged in a critical discussion, "the purpose of the discussion being to establish whether the protagonist's standpoint is defensible against the critical reactions of the antago- nist." This means that questions will be de- signed to expose or lead to the weaknesses of a claim. (p.19) Thus Freeman's contention is that an argument (as product) is best understood as generated by a dialectical process, in which the protagonist attempts to "convince a sceptical but rational judge of the rightness or rational acceptability of a claim" (p.20ff.). Surprisingly, he uses this idea only to justify his ac- count of argument structure, and he says very little about van Eemeren and Grootendorst's more interesting idea that the norms for appraising and evaluating an argument are then supplied by the rules for conducting such an ideal dialectical discussion. We shall return briefly to this point later. Freeman claims that the central ques- tions in the basic dialectical situation con- cern (i) the acceptability of a premiss, (ii) its relevance to its conclusion and (iii) whether it is a sufficient ground. Freeman developed this idea in Thinking Logically (1988) and acknowledges his debt to Trudy Govier's A Practical Study of Argument (1985) and to Johnson and Blair's Logical Self-Defense (1977). He spells out the basic questions as follows: I. ACCEPTABILITY QUESTION Why should I believe that premise? How do you know that reason is true? II. RELEVANCE QUESTIONS Why is that reason relevant to the claim? How do you get there? (Tollimin's warrant generating question). 198 Alec Fisher Ill. GROUND ADEQUACY QUESTIONS The Central Questions in a Basic Dialectical Situation 1. Can you give me another reason? 2. How sure do your reasons make you of the claim? Given your reasons, how confident should I be of your claim? How sure are you that you'll get there? 3. Why do your premises make you so sure (in light of condition or counter- evidence R)? Why do your reasons make you sure enough to accept your claim? What might prevent you from getting there? (pp.38,39) And how do these questions function? Freeman's contention is that in the basic dialectical situation, where both parties are rational, the proponent puts forward a the- sis and defends it in response to the chal- lenger's questions; the questions articulate what the challenger sees as the logical weaknesses ofthe proponent's argument as so far developed. Freeman's basic contention is that answers to questions from the three different categories function differently, and their different functions show how the different elements in argument fit together and this yields us a correct theory of argu- ment structure (see p.91 for a good sum- mary statement of this position). In short, answers to the acceptability question "Why should I believe that premiss?" generate serial reasoning, answers to the relevance question, "Why is that reason relevant to the claim?" generate linked reasoning, and answers to the first ground adequacy ques- tion, "Can you give me another reason?" generate convergent reasoning (p.93ff.). It is of course within the framework of these ideas that Freeman articulates his ac- count of the convergent-linked distinction which I discussed earlier. If Freeman's ac- count of this distinction is flawed, this will throw his whole approach into question. Generalising this comment yields my judgement on Freeman's general theory; if the details work out (hang together, accord with our intuitions and serve our purposes) this vindicates the general approach, but if they don't the general approach is flawed. Before leaving this subject we should note that Freeman also starts an interesting discussion (see p.39) comparing his basic dialectical questions with Grice's maxims for co-operation in rational discourse ("Logic & Conversation") and the questions which arise in formal disputation (see Rescher's Dialectics); these remarks are in- teresting and deserve further development. 6. Modalities and Rebuttals. Let us turn now to 'modalities' and 're- buttals'. The basic dialectical questions given above include two ground adequacy questions we have not yet discussed. Freeman argues that these generate two further elements in arguments besides rea- sons and conclusions, namely the 'modali- ties' and 'rebuttals' of Toulmin's Uses of Argument. I shall say very little about modalities, words like 'probably', 'certainly' etc. Freeman's position is that their chief func- tion is to make a "claim about just how weighty a case the premiss or premises of an argument make for the conclusion they support" (p.1l2) and that "modal expres- sions are genuinely part of the linking ma- terial between premises and conclusions, and are not properly part of the conclusions themselves." (p.126) His views here seem well-argued and eminently reasonable, and they lead him to diagram modalities in a way similar to Toulmin (see p.127). Rebuttals, on the other hand, are more problematic. How should we identify them? Toulmin indicates that rebuttals are intro- duced into arguments by the word "unless" or similar expressions: unless both Harry's parents were aliens or he has become a naturalized Ameri- can barring accidents, unforeseeable inju- ries, or a more than usual degree of managerial incompetence. (p .13 3) But every logic student knows that 'unless P, Q' makes just one assertion, so why sep- arate out rebuttals? Toulmin's case in The Uses of Argument for the separability of rebuttals rests on his distinction between data, backing, and warrants (with their dif- ferent functions): since Freeman rejects these distinctions his case for separating out rebuttals has to be different. Freeman appears to come to the conclusion that re- buttals are distinguishable elements in ar- gumentation largely on the basis of considerations about defeasible concepts. This may be right, but in that case it is sur- prising that Freeman nowhere mentions Wittgenstein in this connection. On Witt- gensteinian conceptions (which certainly influenced Toulmin's introduction of the notion of rebuttal) rebuttals are exceptional circumstances, but Freeman wants to count any consideration which undercuts the strength of an argument as a rebuttal (see p.152f); We differ [from Toulmin] only in specifically allowing any undercutting circumstances to count as rebuttals. ... Acknowledging rebuttals in the context of an argument, then, is to qualify the argument. It is to qualify the claim about how strongly, with what force, the premises support the conclusion. As we have seen, modalities describe the force of arguments. Rebuttals then modify or quali- fy modalities. In effect, they sharpen modalities. The premises support the conclusion with given strength unless .... (p.155) Freeman compares his conception of 're- buttals' with Govier's 'counter-considera- tions.' For Govier, counter-considerations are points which count against the conclu- sion of an argument, typically of the kind found in 'balance of considerations' argument-'on the one hand ... and on the other hand .. .' . Freeman surprisingly doubts that these "should be counted as elements of arguments." (p.l73) It is not clear to me why Govier's 'counter- considerations' are not rebuttals in Review of James B. Freeman 199 Freeman's sense (though it is clear why many of them will not be rebuttals in Toulmin's sense). My confusion is in- creased by Freeman's discussion of Stephen Thomas's 'pro and con-reasons.' Freeman treats Thomas's 'con-reasons' as rebuttals, but it is hard to see why they are not 'counter-considerations' in Govier's sense. Generally speaking, Freeman discuss- es far too few examples of argumentation throughout the book, and more examples would have been particularly helpful at this point. For example, I am not clear how he would analyse the following argument (does it involve rebuttals, counter- considerations, pro and con reasons, or none of these?); Southpool's traffic congestion can be re- lieved just as effectively by a tunnel as a bridge. However, the environmentalists strongly favour the tunnel which is sub- stantially cheaper to build, but the great ad- vantage of the bridge is that it will promote the growth of SouthpooI's important indus- trial estate, upon which the town's prosper- ity increasingly depends, rather than on its port. It is true that there is pressure to ex- pand the port because of the enlargement of the European Community but there is abso- lutely no more room for expansion at the present docks. If the port is to expand, the new deep-water terminal, proposed by the Port Authority, will have to be built, but that seems a remote prospect at present be- cause local opposition to the only possible site is so strong. There is also strong local opposition to the bridge. This is because many people fear that if the bridge does promote South- pool's industry as predicted, this will pro- vide an irresistible case for building the deep-water terminal (given the EC de- mand) and they are quite right about this. But by far the most important considera- tion is that Southpool needs to attract in- dustry; that is basically why we must build the bridge. (Though it is true that if local objections to the deep-water terminal could be overcome that would be a cost-effective way of attracting industry to South pool too.) 200 Alec Fisher 7. Counterrebuttals-and Their Diagrams. Besides rebuttals, Freeman also intro- duces 'counterrebuttals' as distinct elements in arguments; [A counterrebuttal] functions to show that a possible rebuttal is non-operative, to rule out the possi bility of this rebuttal operating in this case. Given this further evidence, this possibility is no longer something to worry about. We call such. premises counterrebuttals. By answering the third ground adequacy question, they have a distinct function and thus enter into arguments in a structurally distinct way. (p.162) Thus, in the following example, we have both a rebuttal and a counterrebuttal; (A) "Peter is clutching his stomach, he's groaning terribly and there is blood on his hands. He must be. badly hurt." (B) "Unless it's ketchup and he's acting." (A) "No, it's real blood, OK, and he's not acting, he's been stabbed." Although the notion of a counterrebuttal is clear enough, Freeman diagrams it in a strange way. Rebuttals are diagrammed like this (see p.157ff.) or (Conclusions) where the upper diagram is read as saying that P gives some support to C unless R, and the lower diagram is read as saying that P [modal word] gives some support to C unless R. But now Freeman explains the diagramming of counterrebuttals as follows: Sinee counterrebuttals support eonclusions by ruling out possible rebuttals, by elaim- ing either that they do not hold or that they do not undercut the argument in this case, there should be downward directed arrows from the encircled numbers representing these premises to the rebuttal box. Typical- ly the representation of an argument with rebuttal and counterrebuttal would look like this: I Conclusion I The arrow from the counterrebuttal premise does ultimately point to the con- clusion. But it points to it through the re- buttal box to indicate that the rebuttal is non-operative in this case. [etc] (p.163ff.) I find this notation confusing, since the ar- row here cannot be read as Freeman usual- ly reads it, as 'gives some support to'; indeed it means something very like the opposite! This is another case where I'm inclined to wonder whether Freeman is clear about what he means by the arrow. (On p.236 he replaces the arrow between rebuttal and counterrebuttal by a straight line, saying "this is our canonical notation" but also that he understands the two diagrams to be "saying the same thing.") Freeman grants that rebuttals and counter-rebuttals as he conceives them seem to be a "rather rarified element in arguments" (p.l65) but he is convinced that they are distinguishable elements. Indeed, he argues that the role he attributes to re- buttals and counterrebuttals corresponds to dialectical moves recognised in Rescher's Dialectics and he takes it that formal dia- lectic supports the inclusion of rebuttals and counterrebuttals (in his sense) as elements in argument (p.165ff.). On the other hand, Freeman's ideas in this area conflict with those of other authors in various respects, some of which I have already discussed. One further dif- ference is that Freeman allows (unlike Toulmin, Thomas and Meiland) that rebut- tals may tell not only against inferences but also against premisses or conclusions (see p.187). Again, though I am not entirely con- vinced about the correctness of Freeman's ideas on the subject of rebuttals and coun- terrebuttals, no one working in the tield will be able to ignore his careful discus- sion. Those who are unfamiliar with the is- sues in this area, might find it helpful to consider first how they would deal with the following example (due to Scriven); One of the most attractive lines of argu- ment that the Democrats have used in order to justify support for a Democratic candi- date for President in 1976 is the unfortu- nate affair of Watergate. But what guarantee do we have that such an event would not have occurred under a Demo- cratic administration? Looking back over the track record of Democratic administra- tions of the past, it is easy to point to exam- ple after example of corruption, political misjudgement, of impropriety and techni- cal breach of the law. This, like other argu- ments that they have produced can't really be regarded as having any real signifi- cance .... (p.168) Having done this they should then read Freeman's discussion of the same ex- ample. 8. Problems with Complex Argument Diagrams. Let us now return briefly to a remark I made earlier about the point of doing infor- mal logic. Freeman's account of macro- structure gets more and more complicated as the book proceeds. For example, he dia- grams the following argument, I maintain that the U.N. should be disband- ed because it has been unable in general to prevent contlict and war around the world. Review of James B. Freeman 201 One might object that the U.N. has been able to prevent certain conflicts. For U.N. peace keeping forces ended the bor- der fighting between Tyra and Sidonia, U.N. negotiators brought about a cease fire in the Volta War, and they have prevented armed conflict in the Equatorian crisis. I reply that the U.N. has not been able to prevent even particular conflicts, certain- ly not these. For fighting ended between Tyra and Sidonia because the troops ran out of ammunition; internal political problems ended the Volta War; and pressure from al- lies contained the Equatorian crisis. (p.195) as follows (see p.196): ~9~ 5 ~-~~ CV cy ~ un1o;, UN b~ 2 \ t j I been able to ~ prevent certain ~ particular conflicts Perhaps we really have to admit that this simple argument has such a complex struc- ture, but a tension is clearly emerging be- tween theory and utility. This tension grows in subsequent pages. For example (p. 198ff.), We ... represent that [an] objecting argu- ment is not cogent by enclosing it in a box and crossing out the box, viz Freeman also introduces a box notation- very like Thomas's-to handle supposi- tional arguments (p.213ff.) and arrives at 202 Alec Fisher some very complex diagrams for apparent- ly quite simple arguments when he dis- cusses "refutation by logical analogy" (p.203ff., esp.p.206). Consider the following argument, The Democrats will win the next election because the Republicans have made a number of mistakes in office, and after this many years people are just plain tired of them. One may object that the Democrats are totally fragmented. After all, six candidates are running and none has anything close to a majority. But it is far too early in the electoral process for this to be meaningful. (p.201) Freeman's diagram for this argument (p. 203) is, unless Democrats are fragmented 00 CD and one cannot help wondering if this will really serve the objective of informal logic-to help people handle arguments better. Freeman discusses the constraints which any argument-diagramming tech- nique should meet. Very reasonably he re- quires (i) that it "should be generally applicable" to any natural language argu- ment; (ii) that it should be "straightfor- wardly applicable" without too much "reconstruction" and (iii) that it should "mirror the structure of real life argu- ments" (see p.34 and p.247). In his attempt to meet these requirements Freeman is driven to produce surprisingly complex diagrams. This makes me wonder whether argument-diagramming is really the best way to proceed, given what I called the 'teaching' objective of informal logic. Many people panic in the face of diagrams, but anyway perhaps articulating structure in terms of 'key words' would be a simpler way for most people. This observation does not tell of course against any of Freeman's ideas about the macrostructure of argument, merely against representing them diagrammatically. 9. Dialectics, Macrostructure and the Evaluation of Argument. Towards the end of the book Freeman asks the question, Why is this study important for logic?' After all, someone might object, in the ordinary process of evaluating arguments, people do not spend time first analysing them through argument diagrams. (p.255) In short, Freeman's answer is that his dialectical approach to discerning the macrostructure of an argument not only reveals its structure but also reveals its strengths and weaknesses-because one asks the questions a rational critic would ask. Thus he sees the following parallel; when one articulates the microstructure (the logical form) of an argument in the tradition of deductive logic this is the key to deciding its validity; similarly, on Freeman's dialectical account, when one articulates an argument's macro- structure this is the key to deciding its soundness. It is precisely in the light of the dialectical motivation of our structural account that we can maintain that there is [an intimate connection between determining its macro- structure and evaluating an argument]. The dialectical questions which motivate our stock of argument elements and the ways in which those elements can fit together are evaluative questions. In our dialectical model, the challenger asks these questions only because she recognizes a logical deficiency in the argument. We can view an argument text, then, as the product of an attempt to propound some thesis together with answering the anticipated challenges of a rational critic. The structure of the text is basically determined by the particular questions the challenger is expected to ask. Dialectical motivation then connects macrostructure with logical evaluation, the central issue in the logical enterprise. (p.256) Unfortunately he does not show how this works out in any examples, which is a great pity. Indeed, his whole theory of ar- gument macrostructure is presented almost entirely without discussing questions of evaluation. This is disappointing, and it is to be hoped that Freeman will remedy this in a sequel to the present book. 10. Some Concluding Criticisms. Indeed, this is perhaps the point to mention another, and related, overall weakness in Freeman's presentation, and this concerns the lack of worked examples of real rea- soning. Most of his examples are invented to illustrate a theoretical claim- in the time-honoured tradition of formal logicians, (surely informal logicians should smell a rat!). Since Freeman rightly wants his account of argument structure to be comprehensive and to apply to everyday reasoning, he should have tested his ideas on more examples of such reasoning and presented them throughout this book. This is particularly true of the later, more complex ideas concerning rebuttals, counterrebuttals, etc. A further general criticism again con- cerns Freeman's dialectical approach to ar- gument macrostructure and the evaluation of argument. Given the importance of van Eemeren and Grootendorst's approach to argumentation, and given that their their 'pragma-dialectical' approach is motivated crucially by dialectical considerations Review of James B. Freeman 203 very like Freeman's, it is surprising that there is so little discussion of their ideas. It is clear that Freeman owes a good deal to their work and it is to be hoped that in subse- quent writing he will make the connections between his work and theirs-particularly concerning argument evaluation. Let me conclude with some minor points of criticism, and a few misprints, in declining order of importance. On Freeman's conception of the dialectical process, the proponent may only make as- sertions and the challenger may only ask questions; there is a frustratingly brief dis- cussion about these limitations on p.183-4; this must deserve more space. Freeman's note 32 on p.230 about "suppositional as- sertions" seems to me to be confused though I do not think it matters greatly to his theory. There is a lengthy discussion (p.233ff.) of what counts as one argument: it is hard to see that much turns on it: is a chain one physical object or many? There is some muddled explanation on p.12 which should read, "Unless A entails B (and hence entails A & B) the argument from A to A & B is not valid. The situation is symmetric with respect to 'B' and 'A'." There is a substantial omission p.155, 6 lines up. There is a misprint on p.159, line 9, where "Otherwise it is convergent" should read "Otherwise it is linked." 11. Final Verdict. Lest the preceding points of criticism should be misconstrued, let me repeat what I said at the very beginning of this review. This is an excellent book. It is certain to re- ceive a great deal of attention from those working in the field of informal logic and argumentation theory. It is very carefully researched and very thoroughly argued (it is also very carefully annotated with full references, so that researchers will have no difficulty in tracking down sources). It is a major contribution to the development of the theory of argument. 204 Alec Fisher References Beardsley, M.e. (1950) Practical Logic. Scriven, M. (1976) Reasoning. New York: Englewood Cliffs, NJ: Prentice-Hall. McGraw-Hill. Eemeren, EH. van, Grootendorst, R. and Kruiger, T. (1987) Handbook of Argumentation Theory. Dordrecht: Foris. Fisher. A.E. (1988) The Logic of Real Arguments. Thomas, S.N. (1986) Practical Reasoning in Natural Language (3rded.). Englewood Cliffs, NJ: Prentice Hall. Cambridge: Cambridge University Press. Toulmin, S.E. (1958) The Uses of Argument. Freeman. J.B. (1988) Thinking Logically. Cambridge: Cambridge University Press. Englewood Cliffs, NJ: Prentice-Hall. Govier, T. (1992) A Practical Study of Argument (3rd ed.). Belmont, CA: Wadsworth. Wittgenstein. L (1953) Philosophical Investiga- - tions. Tr. G. E. M. Anscombe. New York: Macmillan. Grice, H.P. (1975) "Logic & Conversation" in Cole and Morgan (eds.), Syntax and Semantics 3: Speech Acts, pp. 41-58. __ . (1958) The Blue and Brown Books. Oxford: Johnson, R.H. & Blair, J.A. (1983) Logical Self- Dtifense (2nd ed.). Toronto: McGraw-Hill Ryerson. Meiland. lW (1981) College Thinking: How to Get the Best Out of College. New York: Mentor-New American Library. Rescher. N. (1977) Dialectics: A Controversy- Oriented Approach to the Theory of Knowl- edge. Albany, NY: State University of New York Press. Basil Blackwell. ALEC FISHER DEPARTMENT OF PHIWSOPHY SCHOOL OF ECONOMIC AND SOCIAL STUDIES UNIVERSITY OF EAST ANGLIA NORWICH NRlO 4NZ ENGLAND COMMUNICATION & COGNITION COMMUNICATION &: