The aim of this article is to explore and develop the claim made in the earlier stages of the Theory of Argumentation in Language, that argumentative topoï are at work at the lexical level. We then claimed that the basic meaning of a lexical item is made of the bundle of topoï that this item allows applying when uttered. Let us recall that our main hypothesis is the following one: in a discursive sequence of the type argument + conclusion, a third factor — a warrant, as Toulmin called it, secures the plausibility of the inference. To honour Aristotle, we will refer to such a warrant as a topos. In the first section, we discuss another basic hypothesis, namely the claim that the topoï have the standard form (±P,±Q), where P and Q are metalinguistic predicates. For instance, in the sequence: Pierre est malade: il ne pourra pas aller travailer. the conclusion is drawn from the premise by means of a topos like (+sick, −work), roughly speaking. More generally, the use of such topoï is intended to build up ideological representations. Sequences like: • Il fait beau. Allons nous promener. • Il fait beau. Je vais aller à la plage. • Il fait beau. Je suis content. are meant to build up such notions as fine-weather-of-going-for-a-walk, fine-weather-of-going-to-the-beach, fine-weather-of-feeling-good, … etc. Let us consider now the following examples: 1. (1) Il y a un problème, mais il est facile à résoudre. 2. (2) ??Il y a un prroblème; pourtant, il est facile à résoudre. 3. (3) Il y a un problème, mais le patron n'est pas là. 4. (4) Il y a un problème; pourtant, le patron n'est pas là. The contrast is rather surprising, all the more as mais and pourtant are exchangeable in most examples. In fact, (1) and (2) diverge from (3) and (4) as for their semantic structure. Being difficult to solve is indeed in the very nature of a problem, whereas there is a priori no linguistic relation between the emergence of a problem and the attendance of the boss. We will refer to the topos involved in the first two examples as an intrinsic topos between problème and difficile à résoudre. In the other two examples, the topos will be referred to as an extrinsic topos between problème et présence du patron. The above mentioned examples then show that mais can rest on an intrinsic topos or an extrinsic one as well, while pourtant admits only extrinsic ones. After an examination of the exact nature of intrinsic topoï, we show the connection between these topoï and the specific subclass of generic sentences made up by proverbs. We then put forward several linguistic criteria that show the difference between intrinsic and extrinsic topoï. Last, several phenomena of scale predication, topicalization, and aspect/time structuration, are accounted for in terms of topoï.