The meaning of a syncategorematic term is, therefore, a set (a series, a system) of instructions for its possible contextual insertions and for its different semantic outputs in different contexts (all registered by the code). Can this solution apply to categorematic terms as well? This is, in point of fact, the solution increasingly preferred by context-oriented componential semantics. The forerunners of this type of instructional semantics were Peirce’s logic of relatives (C. P. 2.379, 3.66; see also Eco 1979), the various case grammars (Fillmore 1968; Bierwisch 1970, 1971), the semantic models based on contextual and circumstantial selections (Eco 1976, 2.11), and their reformulation for the disambiguation of the metaphor (see Chapters 2 and 3 of this book). The semantic type is the description of the contexts in which the term can be expected to occur.
If this is the case, then the connoted meaning is possible because, at the first level of signification (where the linguistic sign is primarily at work), inference, rather than mere equivalence, is already present. A linguistic term appears to be based on pure equivalence simply because we do not recognize in it a ‘sleeping’ inference.
The process of recognition of natural events which generates sign-Proposition takes place in the same manner. Perception is always inter-rogative and conditional and is invariably based (even if we do not realize it) on a bet. If certain perceptual data are present, then (there is) Perhaps smoke, as long as other contextual elements authorize me to think that the perceptual interpretation is appropriate. Peirce was aware of the fact that perception is always presumptive evidence, a source of Potential semiosis. The fact that perception takes place without effort does not invalidate its inferential mechanism (see C. P. 5.266—68).
We are left with the problem of the so-called substitutional tables, that is, minimal ciphers where the content-plane is given by the ex-pression-plane of another semiotic system. In Morse code, for instance, /. — / «a», and vice versa, with complete reciprocity. A substitutional table could be viewed as a degraded semiotic system, but in point of fact equivalence appears to be a ‘sleeping’ inference here as well (see Chap-ter 7 of this book).
Therefore, there is no difference in the semiotic structure of first- and second-level signification (we use this distinction because the couple denotation/connotation is equivocal, since ‘denotation’ is used by ex-tensional semantic theories in order to refer to truth-values). The fluc-tuating object, which is commonly called ‘sign’ in so many different cases, exists as a scientifically unified object, constructed by the discipline which studies it, subsuming different phenomena under the same formal scheme p q.
What varies according to the phenomena is the cogency of the infer-ence. If (there is) the first, then (there is) the second. But what is the epistemological value of if and then?
1.10. Strong codes and weak codes
The Stoics’ inference was the Philonian one, the material inference of modern logic. As such, it did not deal with the epistemological value of the link between antecedent and consequent. The Stoics gave various examples: ‘if there is daytime, then there is light’ is an equivalence (biconditional). ‘If it is daytime, then Dion walks’ is an example of ma-terial inference devoid of epistemological value. ‘If she gives milk, then she has given birth’ is an inference from effect to cause based on previ-ous inductions. ‘If a torch is seen, then the enemy is coming’ seems to be a very vague supposition, because the torch could be carried by friends as well.
Sextus interprets this sign as a conventional one, suppos-ing that it is recognized on the basis of a previous agreement. At this point, the epistemological value would depend on social laws rather than on natural laws. By introducing, along with the example, all the com-memorative signs among those founded on an arbitrary correlation, Sex-tus acknowledges the inferential nature of conventional signs. In this case, the epistemological value of if-then assumes the legalistic nature of the norms sanctioned by juridical codes.
Aristotle, who was interested in finding arguments capable of explain-ing the necessary links between facts, posited a number of epistemologi-cal distinctions between necessary signs and weak signs. The Stoics, who were only interested in the formal mechanisms of inference, avoided the problem. Only Quintilian (Institutio oratoria 5.9), who was interested in the reactions of a forensic audience, tried to account for every type of persuasive sign according to an epistemological hierarchy.
Quintilian does not disagree with Aristotle’s classification in the Rhetoric, but he warns that necessary signs can deal with the past (if she has given birth she must have been with a man), with the present (if there is a strong wind on the sea, there must necessarily be waves) and with the future (if one has been stabbed in the heart, one will necessar-ily die).
Clearly, though, these presumed temporal links are in truth different combinations of cause-effect links. The link between giving birth and sexual intercourse (diagnostic sign) goes back from the effect to the cause, while the link between wound and death {prognostic sign) goes from the cause to its possible effects. However, this distinction is not homologous to the distinction between necessary signs and weak signs. Every cause does not necessarily refer us forward to all its possible ef-fects (weak prognostic sign), and not all effects necessarily refer us back to the same cause (weak diagnostic sign). Who carries the torch, the enemy or the allies? A distinction should also be made between necessary causes and sufficient causes.
Oxygen is a necessary cause for combustion (so that if there is combustion, then there is oxygen), but the striking of a match is only a sufficient cause for combustion (in occurrence with other possible causes). One could then say that Aristotle’s weak sign goes from effect to sufficient cause (if one has a hard time breathing, then one has a fever); but, when better examined, the weak sign also exhibits a de-gree of ‘necessity’ — except that this sign refers to a class of causes, rather than to one cause; if there is a torch, then someone must have lit it and must be carrying it. If there is difficulty in breathing, then necessar-ily there is an alteration of the cardiac rhythm (a class of events which includes also fever). These types of signs should have a necessary conse-quent, but the consequent is still too wide and it must be narrowed (passage from the class to a member of the class) on the basis of other contextual inferences (as Hippocrates knew quite well).
In verbal language a similar process takes place, since I can name an entity by synecdoche from genus to species. Instead of saying ‘human beings’, I can say ‘mortals’. The prognostic sign from cause to effect involves a number of prob-lems as well. Thomas Aquinas says (Summa Theologiae 1.70.2—2; 3.62) that the instrumental cause can be a sign of its possible effect: if the hammer, then the operations that it can be expected to perform. This is how the police operate. If weapons are found in an apartment, their possible criminal usage is deduced. This type of sign is obviously open to contextual inferences. The nature of the clue changes, depending upon whether the weapons are found in the house of a presumed terrorist, of a police officer, or of a gunsmith. And why does Aquinas not mention, for instance, the efficient cause? Cannot the presence of a well-known murderer in town be a sign of criminal intent on his or her part? As for the final cause, is it not the basis for the cut prodest type of argumentation?
It appears, then, that all prognostic signs are weak because of the epistemological nature of inference (the link is not a necessary one), whereas all diagnostic signs can be weak because of the generality of the implicatum (the class of the consequents is too wide). Today epistemol-ogy, inductive logic, and probability theory know how to measure the various degrees of epistemological force. One might ask why Aristotle and, above all, Quintilian did not hesitate to list as possible evidence every type of sign, even though they were aware of their different epis-temological force. But, at the rhetorical level, links are mostly based on conventions and common opinions. Quintilian cites as verisimilar the fol-lowing argument: if Atalanta goes walking in the woods with boys, then she is probably not a virgin any more. In certain communities this ver-isimilitude can be as convincing as a necessary sign. It depends on the codes and on the scripts (cf. Eco 1979) which the community registers as ‘good’.
The hiatus between ‘scientific’ certitude and ‘social’ certitude consti-tutes the difference between scientific hypotheses and laws, on the one hand, and semiotic codes, on the other. The necessity of scientific evi-dence has little in common with the necessity of semiotic evidence. Sci-entifically, the whale is a mammal, but in many people’s competence it is a fish. Scientifically, a lemon is necessarily a citrus fruit, and it is not necessarily yellow. But for a reader of poetry (Montale: «The golden trumpets of solarity»), the lemon is a yellow fruit, and its being a citrus is irrelevant.
Therefore, at the semiotic level, the conditions of necessity of a sign are socially determined, either according to
