It has in fact such acute sight as to be able to pierce the shadows of the night with flashing eyes.»1 When we get to dictionaries such as the one published by the Italian Accademia della Crusca in 1612, we find the definition of cat (entered, with admirable political correctness, in the feminine form gatta, even though the rest is then written using the masculine pronoun): «Known animal, kept in houses, owing to its particular enmity for mice, which it kills.» And that’s that.
As can be seen, once upon a time there were no dictionary-type definitions (except for the traditional «rational mortal animal»). The first attempts in this regard are found in the dictionaries of the perfect languages, such as in the Essay toward a Real Character by John Wilkins (1668), who attempted to define the furnishings of the entire universe by genus and differentia, basing himself on the first attempts at scientific taxonomy. But, after having worked out a table of 40 major Genera, subdivided into 251 peculiar Differences, from which he derived 2,030 Species, Wilkins (if we take, for example, the classification of «viviparous clawed beasts») managed to distinguish the fox from the dog but not the dog from the wolf (see Eco 1993: 242, fig. 12.2).
And then if we want to know what a dog is and what it does, we have to go to consult the Differences, which are not presented as dictionary-type primitives but are authentic encyclopedic descriptions of empirical properties (e.g., rapacious viviparous animals generally have six short, pointed incisors and two long fangs; the «dog-kind» have an oblong head that distinguishes them from the «cat-kind,» which have a round head; and the dog is differentiated from the wolf, because the former «is noted for tameness» and the latter for «wildness and enmity to sheep»). The dictionary schema is an instrument of classification, not an instrument of definition; it is like the Dewey method of librarianship, which allows us to identify a given book from among the thousands of shelves in a library, and to infer its subject matter (if we know the code) but not the specific content.2
Given therefore that scientific taxonomies took on a rough shape in the seventeenth century and were established organically only starting from the eighteenth century, we would seem to be led to the paradoxical conclusion that before then (in the absence of dictionary structures), from the appearance of Homo sapiens up to at least the seventeenth century, since a dictionary competence did not exist, no one managed to use his own language decently (Aristotle and Plato or Descartes and Pascal spoke, but they could not understand each other) and no one managed to translate from one language to another. Since historical experience contradicts this inference, one must conclude that, while the absence of a dictionary competence did not prevent humanity from speaking and understanding for millennia, that absence is, if not irrelevant, certainly not decisive for the purposes of linguistic competence.
Perhaps it would be sufficient to state that the NC is mostly composed of features of an encyclopedic nature, often disorganized, while forms of dictionary competence appear only in representations of MC. But it’s not that simple. The authors of the medieval bestiaries would perhaps fail a zoology exam, but it cannot be denied that in their own way they were trying to constitute categories when they defined the crocodile (in terms of NC) as a water snake, evidently by taking for granted that this category was opposed to that of land snakes.
In addition, if there are semiosic primitives, precategorial distinctions such as that of «animal» (in the sense of animated beings), when we decide to perceive a mosquito as an animal, we collocate it (in a rather confused way) in a categorial order, just as we would put a chicken and an edible mushroom together among «comestible things,» thereby opposing them to a rhinoceros or a poisonous mushroom (dangerous things).
4.2 Files and Directories
Let us try therefore to compare our cognitive processes, from the first perceptions to the constitution of any knowledge, not necessarily scientific, to the organization of our computer.
We perceive things as sets of properties (a dog is a hairy animal that has four legs, a tongue that hangs out, and barks, etc.). In order to recognize or identify things, we construct files (which may be private or public: a file can be our own work, or it may have been communicated to us by the Community). As the file is gradually defined, by our judging similarities or differences, we decide to insert it (or the Community presents it to us as already inserted) in a given directory. Sometimes, when we need to look something up, we call the tree of directories up on the screen and, if we have a vague idea of how the tree is organized, we know that files of a certain type must be in a given directory. As we continue to gather data, we can decide to shift a file from one directory to another. But as the task gets more complex, it becomes necessary to split certain directories up into subdirectories, and at a certain point we may decide to restructure the entire tree of directories. A scientific taxonomy is no more than a tree of directories and subdirectories, and the only difference between the taxonomies of the seventeenth century and those of the nineteenth was that the tree of directories was simply (simply?) restructured on a series of occasions.
But this computer-inspired example conceals a trap. The files in a computer are full (in the sense that they are collections of information), while the directories are empty—in other words, they can be collections of files but, if there are no files, they contain no other information. In a scientific taxonomy, on the other hand (as has already been pointed out in Eco 1984, 2.3), when, let’s say, the CANIDS are inserted among the MAMMALS, saying that dogs and wolves are mammals does not mean only that they are housed in the directory called MAMMALS: the scientist also knows that MAMMALS (be they CANIDS or FELIDS) usually reproduce in a similar manner. This means that the taxonomist cannot open a directory headed, let’s say, CRYPTOTHERIA, and decide to put any old files in there should the need arise: he must have decided what the characteristics (perhaps brand-new) of the CRYPTOTHERIA are, so that—on the basis of the presence of these characteristics in a given animal—he can justify the insertion of the animal’s file in that directory. This ensures that when the taxonomist says that a certain animal is a MAMMAL, he knows what general characteristics it possesses, even though he does not yet know if it looks more like an ox or a dolphin.
Therefore every directory ought to contain a «label» with a series of data on the common characteristics of the objects described in its files. (All we need do is think that it is possible, as is already the case with the files in certain operating systems, to register the name of a directory not as a simple cipher but as a text: in such a case, the MAMMALS directory would be registered as MAMMALS (POSSESSING SUCH AND SUCH REPRODUCTIVE PROPERTIES). As a matter of fact, taxonomic terms such as MAMMAL, OVIPAROUS, FISSIPED, or UNGULATE express a great number of qualities. In the Linnaean system, names such as Poa bulbata contain all the information that Pitton de Tournefort was still obliged to list as «Gramen Xerampelinum, miliacea, praetenui, ramosaque, sparsa canicula, sive xerampelinum congener, arvense, aestivum, gravem minutissimo semine» (see Rossi 1997: 274).
Such a condition is not at all indispensable for a dictionary semantics: if the species of the PRISSIDS were put in the subdirectory of the family of the PROSIDS, let’s say, and if the PROSIDS belonged to the order of the PROCEIDS, it would not be necessary to know the properties possessed by a proceid or a prosid to be able to make (highly accurate) inferences of the type If this is a prissid, then it is definitely a prosid, and it is not possible for something to be a prissid and not a proceid. Unfortunately, while this is the way we reason when performing exercises in logic (laudable activity), and the way in which a zoology student who has memorized the book without understanding the argument (deplorable activity) attempts to answer an examination question, it is not the way we reason in order to understand either the words we use or the concepts that correspond to them, so it would not be unlikely that, on hearing it said that all prissids are prosids, somebody might just ask for some supplementary information.
But even if the dialectic between directory and files can be compared to that between Dictionary and Encyclopedia, or between categorial knowledge and knowledge by properties, this division is not homologous with that between NC and MC. We do in fact also organize directories at the level of NC (by putting cats among the animals and stones among inanimate objects), but the organizational criteria are less strict, and so it is all right, and it has been all right by us for a long time, to put the files on whales in the directory marked «fish,»
