As for the characteristic (d), the tree can be continually reelaborated and rearranged. Since ‘mortal’ does not contain (necessarily) ‘rational’, why not put ‘mortal’ upon ‘rational’? Boethius (De divisione 6,7) says clearly that this kind of optional division is possible for accidents. We can say that all white are either hard (pearl) or liquid (milk). But we can at the same time say that, given the genus of hard things, some of them are white (pearl) and some of them are black (let us say, ebony); or that, given the genus of liquid things, some are white (milk) and some are black (let us say, ink). Therefore Boethius suggests that (at least for accidents) many trees can be arranged playing upon the same entities, as shown in Figure 2.8.
white things black things liquid things hard things
liquid hard
milk pearl
liquid hard white black
ink ebony milk ink
white black
pearl ebony
FIGURE 2.8
Boethius, however, says something more (De divisione 37), namely, that the same freedom of choice holds as far as every genus is concerned. We can divide numbers between prime and not prime as well as between even and odd. We can divide triangles either according to their sides or according to their angles. «Fit autem generis eiusdem divisio multiplici-ter . . . generis unius fit multiplex divisio . . . genus una quodammodo multarum specierum similitudo, est . . . atque ideo collectivum plurimarum specierum genus est.»
The same is said by Abelard (Editio super Porphyrium 160V.12): «plur-aliter ideo dicit genera, quia animal dividitur per rationale animal et ir-rationale; et rationale per mortale et immortale dividitur; et mortale per rationale et irrationale dividitur» (that is to say, as in Figure 2.9). In a tree composed with pure differences, these differences can be re-arranged according to the description under which a given subject is considered.
mortal rational
rational irrational mortal immortal
FIGURE 2.9
2.2.5. Differentiae as accidents and signs
Differentiae enjoy such a singular status because they are accidents, and accidents are infinite, or at least indefinite, in number.
Differentiae are qualities, and it is not by chance that, whereas genera and species are expressed by common nouns, differentiae are expressed by adjectives. They belong to a tree different from the one of sub-stances. According to Aristotle, their number cannot be known a priori (Metaphysics 8.2.6.lO42b2- 1043a). It is true that he says this of nones-sential differentiae, but who really knows which differentiae are strictly essential or specific? Aristotle plays upon a few examples (such as ra-tional and mortal, followed by the whole medieval tradition), but when he speaks of other natural kinds, such as animals, and of artificial ob-jects, he is more vague, so as to make his readers suspect that he could have never thought of a finite Porphyrian tree.
However, the notion of specific differentia conceals an oxymoron: a specific differentia is an essential accident. Is it possible to solve such a philosophical puzzle? The most striking answer to this question is given by Aquinas. In De Ente et Essentia Aquinas asserts that the differentia specifica corresponds to the substantial form (the difference corresponds to the form, and the genus to the matter, and together they make up the essence of the substance of which they provide the definition). At this point it would appear rather whimsical to identify an accident (a quality) with a substantial form, but Aquinas excogitates a brilliant solution: «in rebus sensibilibus etsi ipsae differentiae essentiales nobis ignotae sunt: unde significantur per differentiae accidentales quae ex essentialibus oriuntur, sicut causa significatur per suum effectum, sicut bipes ponit differentia hominis» (De Ente 6). Essential differences cannot be known directly by us; we know (we infer!) them by semiotic means, through the effects (accidents) they produce, and these accidents are the sign of their unknowable cause.
This idea is repeated, for instance, in the Summa Theologiae (1.29.2—3; or 1.77.1 —7). Thus we discover that differentiae such as rational are not the actual substantial form that constitutes the species as such: Aquinas makes clear that the ratio as potentia animae appears out-side verbo et facto, through external actions (and actions are not substances, but accidents); men are told to be rational because they manifest their rational potency through their activity of knowing, an activity that they perform by internal thought and external discourse (1.79.8ff). In a decisive text (Contra Gentiles 3.46), Aquinas says that human beings do not know what they are (quid est); they know quod est (that they are so), insofar as they perceive themselves as performing a rational activity. What in reality our spiritual potencies are, we know «ex ipsorum actuum qualitate,» through the quality of the acts of which they are the poten-cies.
Thus, ‘rational’ is an accident, and so we are all the differentiae in which the traditional Porphyrian tree dissolves itself.
The tree of genera and species, the tree of substances, blows up in a dust of differentiae, in a turmoil of infinite accidents, in a nonhierarchi-cal network of qualia. The dictionary is dissolved into a potentially unor-dered and unrestricted galaxy of pieces of world knowledge. The dictionary thus becomes an encyclopedia, because it was in fact a dis-
guised encyclopedia.
2.3. Encyclopedias
2.3.1. Some attempts: registering contexts and topics
If a dictionary is a disguised encyclopedia, then the only possible repre-sentation of the content of a given lexical item cannot be provided ex-cept in terms of an encyclopedia. If the so-called universals, or metatheoretical constructs, that work as markers within a dictionary-like representation are mere linguistic labels that cover more synthetic prop-erties, an encyclopedia-like representation assumes that the representa-tion of the content takes place only by means of interpretants, in a process of unlimited semiosis. These interpretants being in their turn interpret-able, there is no bidimensional tree able to represent the global semantic competence of a given culture. Such a global representation is only a semiotic postulate, a regulative idea, and takes the format of a mul-tidimensional network that has been described as the Model Q (Eco 1976, 2.12).
Local representations of the Model Q are implemented every time a given text requires a background encyclopedic knowledge in order to be interpreted. These local representations of the encyclopedic knowledge assume the form of a set of instructions for the proper textual insertion of the terms of a language into a series of contexts (as classes of co-texts) and for the correct disambiguation of the same terms when met within a given co-text. An encyclopedic version of componential semantics should then appear as an Instruktionssemantik which is text-theoretically oriented (see, for instance, Schmidt 1976).
In A Theory of Semiotics I have outlined a model of componential analysis in an encyclopedic format, where the spectrum of the sememe (corresponding to the content of a given expression) was analyzed in terms of contextual and circumstantial selections, I have also maintained that this kind of representation should hold not only for so-called categorematic terms but also for the syncategorematic ones, and I have provided examples of an instruction-like or a text-oriented analysis of not only verbal expressions but also indexical signs, as a pointing finger.
In The Role of the Reader I have insisted on the fact that a sememe is a virtual or potential text and that a text is the expansion of one or more sememes. The encyclopedic representation of the sememe has been reinforced with the reference to frames, scripts, and other instructions concerning coded circumstantial and contextual occurrences. As a conse-quence, it is clear that an encyclopedic representation, insofar as it is text-oriented, must take into account this kind of so-called pragmatic factor as well. In an encyclopedic representation, semantics must trans-late into its own terms most of the phenomena studied by pragmatics. There are many contextual operators which work exclusively in relation with a given co-text, but their co-textual fate must be established, fore-seen, and predicted by coded contextual selections.
Take a co-textual operator such as the Italian syncategorematic ex-pression /invece/, basically translatable as /instead/. When syntagmati-cally linked with /di/ (invece di = instead of) it is basically a sentence operator. Without preposition, /invece/ is an adverb and works as a tex-tual operator, and can be translated as /on the contrary/ or /on the other hand/. As such it seems to express some opposition, but it is doubtful what it is opposed to. Take the following expressions:
(1) Mary ama le mele. John invece le odia. (Mary loves apples. John on the other hand hates them.)
(2) Mary ama le mele. Invece odia le banane. (Mary loves apples. On the other hand she hates bananas.)
(3) Mary ama le mele. Invece John adora le banane. (Mary loves apples. John on the other hand is fond of bananas.)
(4) Mary sta suonando il violoncello. John invece sta mangiando banane. (Mary is playing her ‘cello. John on the other hand is eating bananas.)
We realize that in (i) /invece/ (on the other hand) marks an alternative to the subject and her action; in (2) it marks an alternative to the action and to the object; in (3) it marks an alternative to the subject and the object; in (4) everything seems to be challenged.
Now let us try to insert these expressions into a more comprehensive co-text; let us look at them as the appropriate answers to the following questions:
(la) Do Mary and John love apples? (2a) What kind of fruit does Mary love? (3a) What kind of fruit does John love?
(4a) What the hell are those kids doing? They were supposed to have their music lesson!
It has repeatedly been asserted (see, for instance, van Dijk 1977) that a
