You may know – or you may not know – that the basic thesis of Ducrot’s theory of argumentation in the language-system (TAL) is that certain argumentative features are inherent to the language as a system. That means that language as a system, as an abstract, general structure (as defined by de Saussure), in itself possesses or contains some argumentative potential, some argumentative force and certain argumentative orientations, and not only language in action, its use in discourse and as a discourse. For example, there are certain language structures that (restrictively) impose certain argumentative orientation on the discourse, or in other words, language as an abstract system (at least partly) controls what discourse can say, and sets its limits. If that sounds too obvious (language controling what discourse can say), let me illustrate what I mean with a few examples. Suppose someone says to us (one of Ducrot’s favourite examples)
(1) It is 8 o’clock.
Is this an argument? Why would anybody be telling us that it is 8 o’clock? Just to let us know what time it is? Not likely, unless we wanted to know what time it was. But suppose we didn’t want to know what time it was, suppose somebody just said to us (1). Why would anybody want to do that? Obviously, because he or she, by saying (1), wanted to tell us something else. But, what possible follow-up(s), what possible conclusion(s) could such an utterance lead to? In a situation where we don’t know what the exact co(n)text is, there are many possibilities:
(1a) It is 8 o’clock Hurry up!
Take your time!
Turn on the radio!
Go brush your teeth!
Now, let us see what happens if we introduce two modifiers to (1), already and only respectively, as in
(1’) It is already 8 o’clock
(1’’) It is only 8 o’clock..
All things equal, from (1’) we can no longer conclude, “Take your time” (as we could from (1)), but only, “Hurry up”; on the other hand, from (1’’) we can no longer conclude, “Hurry up”, but only, “Take your time”. And why is that supposed to be so surprising? Because (1), (1’), and (1’’) refer to the very same (chronological) fact, namely, that it is 8 o’clock: while (1) allows a multitude of conclusions, (1’) only allows conclusions oriented in the direction of lateness, and (1’’) the conclusions oriented in the direction of earliness. How is that possible if (1), (1’) and (1’’) refer to the same chronological fact, if the basis of (1), (1’), and (1’’) is the same state of affairs? Well, this “same state of affairs” is viewed from different angles: in one case, (1’), 8 o’clock is viewed (and represented) as late, in the other, (1’’), 8 o’clock is viewed (and represented) as early.
What makes this differentiation of the same state of affairs possible is simply the introduction of two language particles, in our case, two adverbs.
Only words have the power to differentiate reality from the “facts”, only words can make the sameness different. In example (1’), already orients our conclusion toward lateness, no matter what time of day is mentioned after already; and in (1’’), only orients our conclusion toward earliness, no matter what time of day only is introducing. In other words, the argumentative orientations toward lateness and earliness respectively are inherent to – are written into – those two lexical units of the language-system.
In late 70s and early 80s, Ducrot’s argumentation theory was mainly concerned with language particles (something that some American linguists are trying to reinvent in the 90s) as mediators or vehicles of argumentative orientation. In late 80s and 90s Ducrot’s interest turned to topoi. He is using an Aristotelian term, and he thinks he is more or less faithful to his idea, though he admits he deformed it a little. The aim of this paper is to shed some light on this “deformation”.
It is today almost a commonplace (a topos of its own) that for Aristotle a topos is a place to look for arguments, a heading or department where a number of rhetorical arguments (of the same kind) can be easily found, ready for use. According to Aristotle, topoi are supposed to be of two kinds: general or common topoi, appropriate for use everywhere and anywhere, regardless of situation, and specific topoi, in their applicability limited to different sciences, fields of knowledge, expertise, opinion, situation, etc. Or, as Aristotle (1926/1991: 1.ii 22) puts it:
“By specific topics I mean the propositions peculiar to each class of things, by universal those common to all alike”.
In works on Aristotle (on his theory of rhetoric), there seems to be no unique classification of general topoi, or a consensus how such a classification should look like; what is more or less certain, and agreed upon is that topoi deal with three basic topics (sic!), common to the three kinds of rhetoric:
1. more or less (of something),
2. possible or impossible, and
3. what did happen and what did not.
And, as Aristotle says (1926/1991: 1. ii 21), “those topics will not make a man practically wise about any particular class of things, because they do not deal with any particular subject matter”.
With Romans topoi became loci, and Cicero literally defines them as places, as “the home of all proofs” (1942/1998, 2. xxxviii. 162), “pigeonholes (this “pigeonholes” are product of translators licentia poetica) in which arguments are stored” (1942/1992: ii. 5) or simply “storehouses of arguments” (1942/1992: xxxi. 109). Only with Quintilian (1921/1953: 5. x. 23 sq) do we get some “directions for use” as to how to extract arguments from those places, namely the famous net quis?, quid?, cur?, ubi?, quando?, quomodo?, quibus auxiliis?
For the Ancients, the topoi or loci were therefore places that hid ready-made arguments, but strangely enough, nobody devoted much time or space to the architecture of those places: where those arguments were hidden, how they got there, and why. Topoi were considered as a kind of heuristic devices, something a well-educated person knew how to use, while little people, obviously, didn’t have any need for.
For the New Rhetoric (Perelman 1958/1983: 113) – in this short overview, I’ll have to skip almost 2000 years of (mostly) degeneration of rhetoric – topoi aren’t places that hide arguments any more, but very general premises that help us build values and hierarchies, something Perelman was especially concerned about. But even Perelman left topoi on a somewhat descriptive level, and didn’t go into the technology of their functioning or their architectural design.
Strangely enough, the same year that Perelman and Olbrechts-Tyteca published their New Rhetoric, Stephen Toulmin published his Uses of Argument, probably the most detailed study of how topoi work. I say “strangely enough” because he doesn’t use the term topos or topoi, but somehow judicial term “warrant”. The reason for that seems obvious: he is trying to cover different “fields of argument”, and not all fields of argument use topoi as their argumentative principles or bases of their argumentation. According to Toulmin (1958/1995: 94-107), if we have an utterance of the form, “If D then C” – where D stands for data or facts, and C for claim or conclusion – than warrant would act as a bridge and authorise the step from D to C. But then, a warrant may have a limited applicability, so Toulmin introduces qualifiers Q, indicating the strength conferred by the warrant, and conditions of rebuttal R, indicating circumstances in which the general authority of the warrant would have to be set aside. And finally, in case the warrant is challenged in any way, we need some backing. As Toulmin (1958/1995: 105) puts it:
“… Statements of warrants […] are hypothetical, bridge-like statements, but the backing for warrants can be expressed in the form of categorical statements of fact.”
What about Ducrot, how does he define a topos? He defines it as a principle (or, as some of his followers say, “a messier”), that ensures the validity or the legitimacy of the move from utterance A(rgument) to utterance C(onclusion). Let’s take Ducrot’s another favourite example
(2) It is warm (A). Let’s go for a walk (C).
Topos is supposed to relate two properties here: a first property P (warmth), connected with the argument A, and a second property Q (pleasantness of a walk), connected with the conclusion C.
And what are the characteristics of topos, this tacit, unspoken principle, which is to be found in the background of argumentative discourse-segments? Ducrot’s claim is that it has three characteristics: first, it is general; second, it is represented as a shared belief, that is, a belief that is common to a certain group of people; and third, it is scalar. Topos, ensuring the validity of the move from A to C in (2) could therefore read
(3) T= More it is worm, more it is pleasant to go for a walk
I said “could read” because topoi are no self-subsistent, independent entities per se, like platonian ideas, but should always be reconstructed from a given argumentative string.
And how is the generality of the topos to be understood? It is to be understood that topos is a very general structure or matrix, allowing a multitude of particular conclusions, which are not obligatory or binding in a way, for example, syllogism is (which of course means that topos is not universal). A topos (i.e. summoning a topos or evoking it or using it) can allow some conclusion, but it does not necessarily bind to that conclusion or in other words: if we accept the argument, we aren’t obliged to accept the conclusion as well. For example, in response to (2), which is an invitation for a walk, we could easily say as
(2’) It is warm. But let’s go for a swim instead.
(T= More it is warm, more it is pleasant to get some refreshment in the water)
(2’’) It is warm. But let’s go better play cards in the shade.
(T= More it is warm, more it pleasant to be in the shade).
Which means that in both cases our addressee recognised the validity of the topos used in our conclusion, without actually agreeing with it in that particular situation. He/she found some other topos more appropriate to the situation and used it to support a different conclusion instead.
When we say that topos is general, not universal, we also admit that there might be exceptions to it, but that does not prevent the topos from being valid, which is exactly the point the famous formula attributed to Aristotle makes: “exceptions make it possible to uphold the rule in unforeseen cases”; in such cases, the notion of exception makes it possible to uphold the validity of the rule nevertheless.
How can we prove the general character of the topos? Well, once again we have to consider the refutations of an argument: very often those refutations take into account the generality of the topos. Let us suppose (once again) that it is warm, and that I am using that (once again) as an argument for suggesting a walk. You can object: “It was also warm yesterday and yet it was an unpleasant walk”. That means that you are pointing out that there are exceptions to the rule, which I have used, and in saying that, you are suggesting that perhaps I shouldn’t use that rule for that particular case. But by pointing out that there are exceptions, you recognise that the rule which I have used is a general rule, and at the same time, you are telling me that maybe – according to what you think – I wasn’t in position to use that rule in my particular situation. You do not deny its generality of the rule at all, you are simply showing me that there are exceptions to it and you are suggesting that we may be in one of those exceptional cases.
We also said the topos is represented as a shared belief, a belief that has been accepted beforehand by a community which the locutor and the allocutor (or addressee) belong to. In other words, representing topos as a shared belief means that some community (be it a nation or a small subcultural group) recognises its validity, i.e. validity and justifiability of the conclusions based on it. But, as we have already seen, that doesn’t imply that every member of the community would necessarily use the same topoi in identical situations: the use of some topos, or a conclusion allowed by this topos, can always be refuted by another (generally accepted) topos.
And finally, when we say that the topos is scalar, we are saying two things. First, properties P and Q themselves are scalar. That is to say, that they are properties, which you can have more or less of. Predicates P and Q, whom a topos connects, must therefore be considered as scales. Second, there are different degrees of intensity in the possession of characteristic P and in the possession of characteristic Q. But that does not at all mean that the arguments and the conclusions themselves are scalar. The properties used or mentioned within the topos are scalar, but not the propositions used in discourse as actual arguments or conclusions; they already represent or take as starting point a certain degree on the two scales. Let’s have a look at the following example (I’m deliberately taking all the examples from Ducrot’s last book Slovenian lectures (1996)):
(4) “It’s less than ten degrees, take a coat with you”.
There is no doubt that neither A nor C is scalar: it cannot be more or less ten degrees; it either is or it isn’t ten degrees. And you cannot more or less take a coat; you either take it or you don’t. So, the indications contained in A and in C are not scalar ones. But that does not prevent the topos, which is the warrant for that string, from being describable in scalar terms. The topos here is
(5) T= The colder it is, the warmer you must dress
and it relates one property P, which is the cold, and another property Q, which is, say, garment warmth. The indications contained in discourse segments A and C, “It’s less than ten degrees”, and, “Take a coat with you”, represent degrees within those general properties P and Q, and you will, I’m sure, agree that it can be more or less cold, and that we can wear more or less warm clothes.
There is one other idea about the scalarity of the topos that Ducrot devotes special attention to. The idea is that the relationship which a topos establishes between P and Q is itself scalar. We have already seen that P and Q are scales (it can be more or less cold, we can dress more or less warmly): a topos indicates that there is a scalar relationship between the degrees of property P and the degrees of property Q. Which means that going along the scale of property P in a certain direction also means going along the scale of property Q in a certain direction: if you move up or down one scale, you move up or down the other.
Let us go back to the example (4) for a moment. Suppose it is not less than 10 degrees, but say around 20 degrees. In such a situation one wouldn’t say, “It’s less than 10 degrees. Take a coat”, but rather, “It’s around 20 degrees. Don’t take a coat”, while the topos used would still be the same, maybe just in another form. Which brings us to a yet new idea: the distinction between topos and topical form, a distinction that is closely related to the notion of scalarity
Once more, let’s take a topos relating property P and property Q in a scalar way. We have already seen that when we move along the scale P in one direction, we also move along the scale Q in one direction: when we go up P, we go up Q. It is not difficult to notice that saying: “The more you go up P, the more you go up Q”, amounts to the same thing as saying: “The more you go down P, the more you go down Q”. If, the more you go up the warmth scale, the more you go up the pleasantness scale, it must be the case that, the more you go down the warmth scale, the more you go down the pleasantness scale. So that the same topos, which relates warmth (P) and pleasantness (Q) in a scalar way, can have two forms, which Ducrot symbolises as
Those are the two topical forms, FP’ and FP’’, of the same topos T. The same relationship between warmth and pleasantness can be considered under two forms, positively in one case and negatively in the other. And there is more to that. Consider the following topical forms (where P still stands for warmth, and Q for pleasantness):
Those forms would read, “More it is warm, less it pleasant to go for a walk”, and, “Less it is warm, more it is pleasant to go for a walk”. And we have to admit that in different times, and different situations in our lives (often it is pretty difficult to say exactly when and why) we use both pairs of topical forms, (6) and (7): the former, according to which it is pleasant when it is warm, and the latter, according to which it is not pleasant when it is warm.
At first, Ducrot was using topoi only in that sense, as warrants (in Toulmin’s words) that enable/authorise the passage from the utterance-argument to the utterance-conclusion. For instance, if we take the example (4) again, topos authorising the passage from A to C would be something like (5): “The colder it is, warmer you must dress”. The problem was that topoi had to be reconstructed from the given argumentative strings, which made them look pretty arbitrary. But then Ducrot noticed that they are or that they can be much more than that, that they are in fact discourse fragments contained (written) in (at least some) words of the language-system. Let us take a look at the following four adjectives (I borrow them from Ducrot (1996) as well):
(8) courageous, timorous, prudent, rash.
You will have no problem noticing that in a way those four adjectives belong to a single category, and that they describe the same kind(s) of conduct (or, to be more exact, two related kinds of conduct), but viewed in different ways. Ducrot would say that in the language-system itself, we have two topoi, T1 and T2, for every situation (as we have already seen with warmth and pleasantness): in our present case (8), topos T1 ascribes value to the fact of confronting danger, to the fact of taking risks, and it does so by relating the notion of risk and the notion of goodness. Topos T2, on the contrary, relates the notion of risk and the notion of evil (badness). Therefore, in one case, the fact of taking risks is viewed as something good, in the other, as something evil, and at different times, depending mostly on what our discursive intentions are, we represent risk as worth taking and we have consideration for the person who takes it, and at others, on the contrary, we represent the fact of taking risks as something bad.
It is not difficult to see how those four adjectives might be classified: two of them implement topos T1, and the other two, topos T2. Which ones? Courageous implements topos T1: when one says that someone is courageous, one is attributing some positive value to him, and one is attributing some positive value to him because he dares to take risks; what we have in the adjective courageous is a positive valorisation of risk-taking. In the case of the adjective timorous, the topos used is still topos T1, the topos that values risk-taking positively, but when we say that someone is timorous, and we are attributing some negative value to him. We are attributing some negative value to him because he does not dare take a risk, which implies that risk-taking is good, at least in certain circumstances. Courageous and timorous are therefore based on the same topos T1, but courageous is used to praise those who dare take risks, and timorous is used to criticise those who do not manage to do so.
What about the two remaining adjectives: prudent and rash? They too implement the same topos, this time topos T2, a topos that depreciates risk-taking. When we say that someone is prudent, except if we do so ironically, we ascribe a certain quality to that person, and we praise him because he can keep away from risks: in that way, we consider risk-taking as bad. In the case of rash, the topos used is the same again, T2. But this time, when we describe someone as being rash, we are criticising him, we are blaming him for taking risks in an unacceptable and unjustified way. We are blaming him for not implementing topos T2, just as we are congratulating the prudent person for implementing it.
We can further distinguish courageous and timorous on the one hand, and prudent and rash on the other by making subdivisions within each of those two groups. To obtain those subgroups, we’ll have to bring in the topical forms. As far as topos T1 is concerned, we have two topical forms: FT1‘ and FT1‘’; and similarly, as far as T2 is concerned, we have FT2‘ and FT2‘’. FT1‘ will be something like, “The more one takes risks (+R), the worthier one is (+V)”, and FT1’’ will be the converse of the first topical form, that is, “The less one takes risks (-R), the less one is doing what one should (-V)”. Now that we have distinguished those two forms, we can distinguish courageous and timorous, which both refer to that topos. We will say that courageous implements the topical form FT1‘, “The more one takes risks, the worthier one is”, and timorous the topical form FT1‘’, “The less one takes risks, the less worthy one is”.
The same can be done with the two adjectives involving topos T2, which depreciate risk-taking: FT2‘ (“The greater the risk, the greater the evil”) and on the other hand, FT2‘’ (“The lesser the risk, the lesser the evil”), which are implemented by the two adjectives prudent and rash.
So, according to Ducrot, we would get the following scheme:
+P, +Q (more risk, more good) courageous
-P, -Q (less risk, less good) timorous
+P, +Q (more risk, more evil) rash
-P, -Q (less risk, less evil) prudent
But there is another, better, even more Aristotelian way of representing T2. Namely
+P, -Q (more risk, less good) rash
-P, +Q (less risk, more good) prudent
And why is that way of representing topical forms better? Two reasons, mainly. The first one is methodological and the second one epistemological. Let me explain what I mean, using another group of four adjectives (needless to say I borrowed them from Ducrot as well): generous, avaricious, thrifty, spendthrift. According to Ducrot we would get the following scheme:
T1 (More money you give away, better it is)
+P, +Q (More money, more good) generous
-P, -Q (Less money, less good) avaricious
T2 (More money you give away, worse it is)
+P, +Q (More money, more evil) thrifty
-P, -Q (Less money, less evil) spendthrift
But reformulating T2 as
+P, -Q (More money, less good) thrifty
-P, +Q (Less money, more good) spendthrift
is theoretically more appropriate because it uses the same predicates and the same description for the same variable (“good” for Q) as T1 (with which it compares); it allows us to group different topical forms not only in relation to how they describe, but what they describe. Namely (if we go back to the first four adjectives)
+P, +Q (more risk, more good) courageous
+P, -Q (more risk, less good) rash
for risk-taking, and
-P, -Q (less risk, less good) timorous
-P, +Q (less risk, more good) prudent
Why is that important? Because it lets us see that there are the same extra-linguistic entities that language views as complete oppositions. To the extent that it even coined different expressions for them: courageous and rash for risk-taking and timorous and prudent for risk avoiding.
Obviously, courageous, rash, timorous and prudent are complex or compound predicates (or to put it more modestly, adjectives), consisting of a description of some extra-linguistic entity (I would like to avoid saying “fact”, because I’m not really sure what a fact is) + its evaluation. We could hardly say the same, for example, for “good” or “bad”; in fact, I think they could be described as the building stones of those complex predicates, the pure evaluation.
But then, is it really the same extra-linguistic entities that the language views differently? When we say that someone is courageous, aren’t we saying that he is taking risks, and that we approve of it, while, on the other hand, we label someone as rash when we want to say that he is taking risks, and that we don’t approve of it? And, on the other hand, don’t we say that someone is prudent if we want to say that he is avoiding risks, and that we approve of it, while we label someone as timorous when we want to say that he is avoiding even reasonable and justified risks, and that we blame him for that? If so, are those extra-linguistic entities really the same? And if they are really extra-linguistic, how can we say at all they are the same?
Aristotle (1926/1991). Art of Rhetoric. Cambridge, London: Harvard University Press (Loeb Classical Library).
Cicero (1942/1988). De oratore. Cambridge, London: Harvard University Press (Loeb Classical Library).
Cicero (1942/1992). De partitione oratoria. Cambridge, London: Harvard University Press (Loeb Classical Library).
Ducrot, O. (1996). Slovenian lectures/ Conférences slovènes. Ljubljana: ISH.
Perelman, Ch. & L. Olbrechts-Tyteca (1958/1983). Traité de l’argumentation – La nouvelle rhétorique. Bruxelles: Editions de l’Université de Bruxelles.
Quintilian (1921/1953). Institutio oratoria. Cambridge, London: Harvard University Press (Loeb Classical Library).
Toulmin, S.E. (1958/1995). The Uses of Argument. Cambridge: Cambridge University Press.