ISSA Proceedings 1998 – Evaluating Tests For Reconstructing The Structure Of Legal Argumentation
In legal argumentation, as well as in everyday argumentation, it is often difficult to distinguish between multiple (also called convergent) and coordinatively compound argumentation (also called linked). In legal argumentation the importance of the distinction between these two kinds of complex argumentations becomes clear in complaints about the justification of judicial decisions. Since the interpretation of the relation between arguments can be of influence on the decision, (one of) the parties to the proceedings may criticise the way the judge interprets this relation. Disagreement about the argumentation being multiple or coordinative compound will then be submitted to a higher court.
This was, for example, the case in HR 5 juni 1992, NJ 1992, 539. Mr Van der Vlies, the proprietor of a number of pleasure boats, bought a plot on the Spanish Water. The original owner of the plot, Spanish Water Resort, has, at some time in the past, announced an allotment plan. It was according to this plan that a yacht-basin would be constructed. This yacht-basin has in actual fact never been built. Now Van der Vlies demands that the yacht-basin be built as was agreed. One of the questions that need to be answered by the Court in this case is whether or not there is an actual agreement between the two parties. In order to be able to address this question the Court assesses the six arguments (a through f) with which Van der Vlies justifies his claim. The Court of Appeal concludes that there has never been an agreement between the parties. In his appeal to the Supreme Court Van der Vlies argues that:
(…) in answering the central question the Court of Appeal has, unjustly, limited itself to the assessment of the separate arguments, thereby ignoring their mutual correlation and connection, or so it seems judging by the Court’s decision. Moreover, it is, in the absence of any justification whatsoever, unclear why arguments a, c and e do not play any part at all in the relationship between Spanish Water Resort and Van der Vlies, but that, moreover, even if one or more of these arguments did not play any part when judged on their own merit, it is unclear whether they may play such a part when considered in mutual correlation or connection.
In other words, Van der Vlies is of the opinion that the Court of Appeal has wrongfully reconstructed his argumentation as being multiple which influenced the evaluation of his argumentation negatively. Now the Supreme Court has to decide as to whether the argumentation of Van der Vlies was multiple or coordinatively compound.
Sometimes the text or the context may provide clues that give evidence to interpreting the structure of argumentation. Studies on this type of clues in the fields of argumentation and informal logic often provide useful insights into the reconstruction of legal argumentation. Textbooks in these fields also provide tests that can be of use when textual and contextual clues are not available. A classification of these tests is proposed by Walton in his book Argument Structure: A pragmatic Theory (1996). It is not surprising that Walton concludes that the tests are not to be overestimated, for several authors of the tests are already very modest about the use of the tests. Since legal argumentation does not always provide textual or contextual clues that can be of help, it seems appropriate to find out if and when these tests can be of use to decide on the structure of argumentation.
First I will give a short overview of Walton’s classification of the tests that are proposed in textbooks. Then I will take a closer look at some of the problems to which Walton draws attention when it comes to implementing these tests as well as to the test which he himself regards as superior to the others. Finally I will look at the way(s) in which judges arrive at decisions in actual practice when there are no textual or contextual clues on the basis of which a decision can be made as to whether the argumentation of a party is multiple or coordinative compound.
The context in which legal complex argumentation is presented, is that of a judge or a party to the proceedings who tries to remove doubts that the other party or a higher judge may have with regard to the standpoint. To contest this standpoint successfully, the other party needs to know whether it is necessary to refute only one of the arguments or all of them.
2. Four tests to trace coordinative argumentation
In Argument Structure Walton (1996: 118 e.v.) presents an overview of various tests that are used in textbooks to determine whether argumentation is multiple or coordinative compound. In order to evaluate these tests on their usefulness he classifies them and distinguishes between the following four types of tests.
1. Falsity/ No Support Test: If one premise is false, the conclusion is not given any support.
2. Suspension/ Insufficient Proof Test: If one premise is suspended (not proved, not known to be true), the conclusion is not given enough support to prove it.
3. Falsity/ Insufficient Proof Test: If one premise is false, the conclusion is not given enough support to prove it.
4. Suspension/ No Support Test: If one premise is suspended (not proved, not known to be true), the conclusion is not given any support.
The differences between these four types of tests are based on the different ways in which the premise-requirements and the conclusion-requirements are stated. In some tests, the premise is assumed to be false. Walton calls this premise-requirement the falsity-requirement. In other tests the premise is assumed not to be proved (established, supported, or known to be true). This premise-requirement is called the suspension-requirement.
Apart from these two different premise-requirements, Walton distinguishes two different conclusion-requirements. The first one is called the no-support requirement. This means that the conclusion is not given any support at all when the premise is removed.
The second conclusion-requirement is called the insufficient-proof requirement. This means that the conclusion is not given enough support to prove it when the premise is removed. On the basis of these four requirements, Walton distinguishes four types of tests to be used for determining coordinative compound argumentation, the basic assumption being that the conclusion is supported by two premises. Walton (1996: 121) focuses on the first two tests, the Falsity/No support Test and the Suspension/ Insufficient Proof Test, because they are prevalent in textbooks. Both tests have been discussed and criticised extensively. I will discuss two of his major points of criticism. In doing so, I will make use of some of Walton’s examples as far as they could be examples encountered in legal texts or textbooks.
3. Objections to the Falsity/ No support Test
In Walton’s view (1996: 133), one of the main problems that arise when using the Falsity/ No support Test, is to be expected in case a standpoint is defended by evidence-accumulating argumentation. The following case is an example of this type of coordinative argumentation:
The defendant refuses to adjust appearance
The defendant refuses to address customers properly
Therefore, the defendant is, to a serious degree, not suited for the job.
According to the Suspension/ Insufficient Proof Test, the argumentation is coordinatively compound, because, if the one premise is suspended, the other by itself does not supply sufficient evidence to prove the conclusion. When the Falsity/ No support Test is applied, however, this same argumentation appears to be multiple, because if the one premise is false, the other premise still gives some support to the conclusion.
Since it is characteristic of evidence-accumulating or cumulative argumentation that every one of the arguments give some support to the standpoint, it is not possible to trace this type of argumentation by using the Falsity/ No support Test. Cumulative argumentation will always be analysed as multiple.
The test can however be used to trace complementary argumentation, which is another type of coordinative argumentation. In complementary argumentation one of the arguments supports the standpoint directly. The other argument supports the standpoint indirectly and is advanced to anticipate criticism on the first argument[i]:
The defendant refuses to address customers properly.
The defendant’s job is for 80 % directly concerned with customers.
Therefore, the defendant is, to a serious degree, not suited for the job.
By using the Falsity/ No support Test, the structure of this argumentation is coordinative, because, if one premise (the first one) is false, the other premise gives no support to the conclusion. Although the Falsity/ No support Test is not a very useful test because it is in principle not possible to distinguish between multiple and cumulative argumentation, it can be used to distinguish between multiple and complementary argumentation.
4. Objections to the Suspension/ Insufficient Proof Test
According to Walton (1996: 139, 170) a serious counter-example to the Suspension/ Insufficient Proof type of test is the `bad’ convergent argument[ii]. By this he means argumentation in which both arguments are inadequate, incorrect or irrelevant reasons for accepting the standpoint. He gives the following example:
George appears nervous.
Rodney says that George is guilty.
Therefore, George is guilty.
Here Walton assumes that we know that Rodney has a criminal record, is a habitual liar, and has been bribed to testify against George. We also know that George is terrified that the charges against him will ruin his career and reputation. Walton states that intuitively the arguments are multiple. But it comes out coordinative compound in the Suspension/ Insufficient Proof Test: if one argument is suspended, the other gives insufficient support to prove the conclusion. So, the outcome of the test does not correspond with intuitive analysis, although its not clear on which this intuition is based. But Walton takes us even one step further. Because we are dealing, in this example, with premises of which neither is sufficient to prove the conclusion, the Suspension/ Insufficient Proof Test is not useful at all. Therefore Walton suggests that bad argumentation should be excluded from the test. To confine the range of applicability of the test, he builds in three restrictive conditions: a plausibility, a consistency and a probative relevance condition. The latter is descried as: `one proposition is probatively relevant to another if it gives some reason, justification, or basis for proving the other’. Example (3) should be excluded from the use of the test on the grounds that the premises are not relevant. As Walton states: `The premises are weakly relevant (perhaps), but their probative strength is minimal or even non-existent’.
The implication seems to be that plausibility, consistency and relevance are conditions decisive for the sufficiency of arguments:
only if arguments are sufficient, it is possible to make use of the Suspension/ Insufficient Proof Test. If we take a look at the example (3), however, how are we to decide whether or not the arguments are irrelevant and thereby insufficient?
To answer this question I make use of the distinction between interpretative, analytic and evaluative relevance as proposed by Van Eemeren and Grootendorst (1992b). In order to be able to deal adequately with the concept of relevance, they introduce analytic relevance as an intermediary concept between interpretative and evaluative relevance.[iii] Since both propositions in the example are considered to be arguments, interpretative relevance does not seem to be at stake.[iv] The arguments seem to be analytically relevant, because in principle they are in support of the standpoint and could give some justification for proving the standpoint. The arguments could for example have been brought forward by someone who does not know that Rodney is a liar or that George worries about his reputation. But even without any external information, the arguments in principle give some justification to the standpoint from an analytical point of view. The arguments do, however, appear to be evaluatively irrelevant, because they turn out to be untrue or unacceptable and will be rejected on the ground that we know that Rodney is a liar etc..
In this view the relevance condition which excluded the example from the Suspension/ Insufficient Proof Test refers to analytic relevance. Since in the example (3) the analytic relevance condition seems to be met, it should not be excluded from the test. In Walton’s interpretation, relevance refers to evaluative relevance. In this interpretation the evaluation of the argumentation is incorporated in the analysis.[v]
The next example (4) illustrates that, although both arguments could be considered as bad argumentation and their probative strength might be minimal, the Suspension/ Insufficient Proof Test can be used to decide whether the argumentation structure is coordinative or multiple.
George appears nervous.
In earlier contacts with the police, George was very much at ease.
Therefore, George is guilty.
By using the Suspension/ Insufficient Proof Test, this argumentation would be reconstructed as complementary argumentation, because if one of the arguments is suspended, the other by itself does not supply sufficient justification. Only if taken together, the arguments seem to be analytically relevant. At the same time they probably won’t be effective from an evaluative point of view.
Although both examples illustrate that bad argumentation does not need to be excluded from the Suspension/ Insufficient Proof Test, it is still not clear whether the argumentation in example (3) is multiple or cumulative. If we would use this test the question remains how to decide if an argument is in principle sufficient to prove the conclusion. In example (1) the wording of the standpoint can be seen as an indication to analyse the arguments as cumulative. The use of the intensifier `to a serious degree’ suggests that the standpoint requires strong evidence. It is therefore more likely not to analyse the arguments structure as multiple, but as cumulative coordinative. But if there are no such internal clues, it is difficult to distinguish multiple from cumulative argumentation by using the Suspension/ Insufficient Proof Test.
5. Walton’s Degree of Support Test
After evaluating the various tests, Walton (1996: 181) proposes his own version of the Degree of Support Test as new test which can be used best to determine coordinative argumentation.[vi] One of the advantages he mentions is that this test is not absolutistic: it is not an `all-or-nothing’ kind of test (1996: 121). He emphasizes that the distinction between multiple and coordinative argumentation is rather a question of how well the conclusion was supported before the premise was removed versus how well it is supported once the premise is taken away. This Degree of Support Test works as follows. First you have to block one premise out of your mind and ask what support of degree the other premise by itself gives to the conclusion. Then you repeat this process for the other premise. Next you add these two weights of support together, and you ask what degree of support both premises together give to the conclusion. If there is a significant jump from the first joint degree of support to the second, the argumentation is coordinative.
To examine this test I give the following example of a case in which an employer wants to terminate an employment contract.[vii]
The employee, Mr Jones, misbehaved.
The employee, Mr Jones, improperly charged more hours than he had actually worked.
Therefore: dismissal of Mr Jones is justified.
The judge who has to evaluate this argumentation, has to decide whether the argumentation of the employer is multiple or coordinative. By using the Degree of Support Test to decide on the relation between the arguments, the judge first has to determine what degree of support the misbehaviour (arriving late at his job after a trip to Geneva) of Mr Jones provides to justify his dismissal. Let us assume this argument gets value 2. Then the judge has to determine what degree of support the fact that Jones improperly charged more hours than he had actually worked, provides to the justification of his dismissal. Let us assume this argument gets value 3. The sum of these arguments is 5. Now he has to compare this value 5 to the degree of support both premises give together. Not only is this very difficult to determine, but even if we assume this value to be 7, it is still very hard to decide whether the difference between 5 and 7 is significant or not and whether the arguments are therefore multiple or cumulative coordinative. Encountering problems like these, it is difficult to understand why this test should be better than the other tests. It does not seem to be very useful to resolve the problem of distinguishing between multiple and cumulative coordinative argumentation.
6. If the tests fail
Now the question is how the argumentation structure should be analysed if there are no indications and the tests don’t work. Van Eemeren en Grootendorst (1992: 81) recommend, in what they call `borderline cases’, to make use of the strategy of maximally argumentative analysis. This strategy sets off by analyzing the argumentation as multiple if no good reason can be found to opt for coordinative.
In literature on legal practice this strategy is also recommended to the defence of parties in legal procedures: if a party is not sure whether the argumentation of the opponent is multiple or coordinative, it should be analysed as multiple. By way of this strategy the defence is required to react to each single argument of the court. If the defence would analyse the argumentation as coordinative and react to just one of the single argument, there is a risk that the judge interprets them as independent arguments. Then the defence will fail because one or more independent arguments have not been taken into consideration.[viii]
If we look at the jurisprudence not only the parties to the proceedings but also the court starts by analyzing the argumentation as multiple.[ix] Sometimes this strategy of maximally argumentative analysis is extended. In the example (4) of the argumentation that was put forward to justify the dismissal of Mr Jones, it was unclear whether the argumentation was multiple or coordinative. The judge who had to decide on this, chose the following solution:
(5) Appellant’s misbehaviour is not enough serious that it justifies dismissal. The court considers the fact that appellant improperly charged more hours than he had actually worked, not sufficient to justify dismissal. Even if both grounds for dismissal are considered in conjunction, there is not sufficient justification for dismissal.
Here the judge starts by evaluating each of the single arguments on its own, which implies that they interpreted the argumentation as multiple. After that, he evaluates the argumentation in conjunction, which implies that he interpreted the argumentation as coordinative. By using this strategy, the judge prevents the party that is put in the wrong, to criticise the interpretation of the argument structure in appeal. The judge, as it were, anticipates the criticism that the analysis and therefore the evaluation was incorrect. Does this mean that this extended strategy of the maximally argumentative analysis is advisable in all `borderline cases’ in which the court is unsure about the relation between the arguments to be evaluated?
In the example (5) the evaluation of the arguments are relevant to the argumentative strength of the arguments that are brought forward by the employer. Neither argument seems to be either untrue or irrelevant; they are just not strong enough to accept the standpoint. In other words, the judge criticises the argumentative force of the arguments; not their propositional content. If he would not have accepted the propositional content because Mr Jones did not misbehave and did not charge more hours than he had actually worked, then the maximally argumentative analysis would suffice. It is only useful to extend the strategy if the refutation concerns the argumentative force because of the arguments being insufficient or irrelevant.
Summarizing we could say that all three tests can be useful to distinguish complementary coordinative argumentation from multiple argumentation. But they don’t seem very useful to distinguish cumulative coordinative argumentation from multiple argumentation.
This is particularly the case if there are no internal clues, such as the wording of the standpoint, that indicate what will count as sufficient defence. Although the applicability of the tests is limited, there seems to be no need to exclude bad argumentation beforehand, if relevance and therefore sufficiency are seen from an analytical point of view. In case there are no textual or contextual clues and the tests are not applicable, the strategy or the extended strategy of maximally argumentative analysis seems to be appropriate.
i. Cumulative argumentation is one of the two types of coordinatively compound argumentation that are distinguished by Snoeck Henkemans (1992: 96). The other type is complementary argumentation.
ii. This kind of counter-example also concerns the Falsity/ No support Test. Here I focus on the Suspension/ Insufficient Proof Test.
iii. Van Eemeren and Grootendorst (1992b) demonstrate that in a pragma-dialectical approach of relevance problems the three types of relevance (interpretative, analytic and evaluative relevance) should be clarified in terms of domain, object and aspect these types of relevance refer to.
iv. If we are dealing with utterances that can’t be interpreted or reconstructed as arguments, than there is of course no need to ask whether the structure of the argumentation is multiple or coordinative.
v. See, for example, Freeman (1991: 107) who states that ‘it would be a distinct disadvantage for a diagramming procedure to require us to make evaluative determinations before we could display the structure’. In case this is inevitable ‘we should keep evaluative issues at a minimum’.
vi. Other Degree of Support Tests are advocated by Thomas (1981) and Yanal (1988).
vii. District court Breda, 27 november 1984 (Praktijkgids 1985/ 2253).
viii.This happens, for example, if arguments have the appearance of an obiter dictum (See Plug 1995).
ix. At first sight, the maximally argumentative analysis seems to be in favour of the plaintiff. Whereas the burden of proof lies with the plaintiff, each of his arguments is in principle enough to prove his standpoint. At the same time it should be considered, however, that at least one of his arguments should be resistant to all kinds of criticism (his argument has to be acceptable, relevant, strong, etc.). The defendant for his part can limit his criticism to one point. As for the discussion about the principle of maximally argumentative analysis versus the charity principle, see Van Eemeren and Grootendorst (1992: 81) and Walton (1996: 211).
Eemeren, F.H. van, & Grootendorst, R. (1984). Speech acts in argumentative discussions. A theoretical model for the analysis of discussions directed towards solving conflicts of opinion. Dordrecht/Cinnaminson: Foris Publications , PDA 1.
Eemeren, F.H. van, & Grootendorst, R. (1992a). Argumentation, communication and fallacies. A pragma-dialectical perspective. Hillsdale, NJ: Lawrence Erlbaum Associates.
Eemeren, F.H. van, & Grootendorst, R. (1992b). `Relevance reviewed: The Case of Argumentum ad Hominem’. Argumentation, volume 6, no. 2, pp. 141-159.
Eemeren, Frans H. van, Rob Grootendorst, Francisca Snoeck Henkemans (1996). Fundamentals of Argumentation Theory.