I’ve just read a neat article that summarizes and discusses one of my favorite argumentation theories, namely Anscombre and Ducrot’s (Iten, 2000). The article recounts the historical evolution of A. & D.’s Argumentation Theory (AT).

Let’s start with an example of my own: “August inflation was barely 2%” vs. “August inflation was as high as 2%.” As A. & D. remark in their first publications, both utterances have the same informational (i.e. truth-conditional) content, yet they cannot be used as arguments in favor of the same set of conclusions. Another example: “there’s a little wine left” vs. “there’s little wine left” (conclusions: “we don’t need to buy more wine” in the former case, “let’s go buy a new bottle” in the latter). Same information, opposite argumentative orientations. To explain such phenomena, in their early works A. & D. postulated an integrated pragmatics (pragmatique intégrée). “They call it a ‘pragmatics’ because it is concerned with the sort of meaning that can’t be captured in terms of traditional truth-conditional semantics,” says Iten.

She reviews several of A. & D.’s most famous analyses, such as their treatment of “but” as an “argumentative operator” that connects two utterances with opposite argumentative orientation. My example: “I am enjoying your visit very much, but it’s late and I have to work tomorrow”. Not only do both clauses have opposite orientations; it can also be claimed that the second argument has greater argumentative strength than the first. Or you can take the utterance: “Peter is quite helpful. He didn’t do the dishes but he cleared the table.” According to A & D (1983: 107, as explained by Iten), the but in this sentence “is scalar in nature”, i.e. it not only indicates that the two clauses support contradictory conclusions (or have opposite argumentative orientations) but it also indicates that “Peter has cleared the table” is a stronger argument for “Peter is quite helpful” than “Peter didn’t do the dishes” is for “Peter isn’t helpful”.

Another example Iten brings up is the argumentative operator “nearly”: An utterance containing “nearly” usually has the same argumentative orientation as a corresponding utterance without “nearly”. (“He nearly hit that fence” or “He hit that fence” both warrant the conclusion “he’s a bad driver”). Iten says: “This could be a banal observation if it wasn’t for the fact that, from the point of view of informational content, ‘nearly X’ is equivalent to ‘not X’. This is made even more interesting by the fact that the argumentative orientation of an utterance containing barely (…) is the opposite of that of the same utterance without barely (…), in spite of the fact that ‘barely X’ is informationally equivalent to ‘X’.”

As part of their 1983 formulation, A. & D. also claim that “an act of arguing… is an illocutionary act, which is part of the meaning of every utterance”, a claim that blurs the distinction between the pragmatic and semantic domains.

Iten then discusses the latest formulation of the theory, which relies on the Aristotelian concept of topos or “argumentative commonplace.” “A topos is an argumentative rule shared by a given community.” “This argumentative rule is used to license the move from an argument to a conclusion.” It is an important feature of topoi that they are scalar in nature. For instance, if a topos states that when it’s hot it’s pleasurable to go to the beach, then the hotter it is, the more pleasurable it is to go to the beach (or eat ice-cream, or turn on the air-conditioner, etc.) The general form of a topos is “The more/less object O possesses property P, the more/less object O’ possesses property P’.”

With the introduction of topoi, A. & D. will no longer try to analyze the meaning of utterances in terms of “presupposed contents” (as in the earlier versions of their theory). Rather, the meaning of linguistic predicates is defined by the topoi associated with them, and by the network of possible conclusions they enable. The meaning of a predicate like work, for example, is given by a bundle of topoi involving gradations of work. Some topoi that could be part of the meaning of work are:

• The more work, the more success.
• The less work, the more relaxation.
• The more work, the more fatigue.
• The less work, the more happiness.

“Gradations of work are linked, via different topoi, with a series of other gradations, e.g. of success, relaxation, fatigue and happiness. These gradations, in turn, are themselves linked to different gradations still. For instance, gradations of happiness could be linked with gradations of health, appetite, etc. This network of gradations, linked via an infinite number of topoi, is what A & D (1989: 81) mean by a topical field.”

With this emphasis on topical fields, Iten claims, we arrive at a completely non-truth-conditional semantic theory. “In their own words, A & D (1989: 77/79) move from a position of considering “argumentation as a component of meaning” to one of “radical argumentativism”. This is a position where “the argumentative function of language, and with it argumentative meaning, is primary and the informative function of language secondary. In that sort of an account, any informational (or truth-conditional) meaning would be derived from an underlying argumentative meaning.”

While in its earlier stages AT acknowledged (to some degree) the distinction between semantics and pragmatics, in its later formulations it abandoned this distinction completely. Iten is right in concluding that, in its current state, Anscombre and Ducrot’s theory non-cognitive, non-truth-conditional, and reduces all linguistic phenomena to a semantic-pragmatic level of analysis.

The latest version of D. & A. also includes their beautiful (in my opinion) concept of polyphony. “The idea is that the (usually unique) speaker (locuteur) doing the uttering stages a dialogue inside her own monologue between different points of view (énonciateurs).” The most obvious examples of polyphony in our speech and verbal thinking are: direct and indirect reported speech, ironical utterances, utterances containing but and negative utterances.

Finally, Iten dwells in the extreme consequences of radical argumentativism for linguistic theory, namely: all linguistic meaning can be captured in purely argumentative terms; the meaning of every utterance can be described in terms of a collection of topoi, which constitute different points of view; and there is nothing about language as such that is informative, i.e. language is not cut out to be used to describe states of affairs.

Although Iten might be French, her criticism of A. & D. is too American in my opinion, and also a little unfair. Iten favors a cognitive paradigm according to which the human brain is a computer that takes in information from the outer world (“input”), processes it by applying computational rules, and produces output. The information is encoded in the head of the speaker as representations that can be true or false, that is, that can match or not match the facts of the world. This cognitive paradigm provides Iten with a yardstick for assessing the virtues of A. & D.’s theory, which does not fare well when subjected to such standards: it doesn’t explain how language represents the world; it doesn’t connect input with output; and it doesn’t predict arguments’ conclusions based on their premises.

Yet the merits of A. & D.’s theory lie elsewhere. A. & D. teach us to look at language in a new way. They provide us with tools to understand utterances as actions. One takes positions, defends certain points of view, creates assumptions, and commits to certain conclusions. All these argumentative movements take place on a stage-like mental space, and are carried out by a plurality of characters or actors that A. & D. call enunciators. I like this approach very much, among other reasons because it fits my own emphasis on the sense of justice that animates all argumentation. That’s why A. & D. (unbeknownst to them) also appeal to quasi-legal terminology, such as saying that a certain topos “licenses” the speaker to derive a certain conclusion, etc. In other words: the movement from premises to conclusions cannot be understood in computational terms; it has to be appreciated on the level of free, human action.