But this does not show that the categories assume the form of prototypes. To say that the words cat, Katz, or chien are handier and easier to memorize than the words Felis or Mammal certainly confirms that in everyday experience it is easier for us to identify something as cat rather than as mammal, but it does not tell us whether there is a prototype of the cat or, if there is, what this may be. If anything, the problem of prototypes concerns phenomena such as that of the extensibility of categorial limits (extendible boundaries), so that it is debatable whether certain highly complex irregular polyhedrons are polyhedrons, while there are no doubts about the better known regular polyhedrons, or whether transfinite numbers are numbers or not, while no one doubts that 2 or 100,000,000 are numbers.
But the existence of basic categories is inferred from spontaneous everyday linguistic behavior, whereas an experiment like the one on polyhedrons or numbers requires an interviewer to ask a subject to answer a question that brings complex classifications into play. Therefore the problem is: Can the existence of prototypes be inferred from everyday behavior (not only linguistic but also behavioral, such as felicitous recognition) or from verbal responses to sophisticated questions?
To come to Eleanor Rosch, in a first phase of her experiments (between the sixties and the seventies) prototypes were a matter of perceptual pertinence. In a second phase (the first half of the seventies) the prototypical effects obtainable by experiment were thought to provide a characterization of the internal structure of the category (hence the persuasion that they constituted mental representations). In a third phase (the late seventies) prototypical effects were thought to subdetermine mental representations, but there was no one-to-one correspondence between prototypical effects and mental representations. The effects were not thought to reflect categorial structure. Therefore we might know judgments of prototypicality, but they would tell us nothing about our cognitive processes, and prototypical effects would be superficial.41
In point of fact Rosch (1978: 174ff.) makes it clear that the prototype is neither a member of a category nor a precise mental structure but, rather, the result of an experiment that aims at collecting and quantifying judgments on the degree of prototypicality. What does degree of prototypicality mean? We are said to have an identification of prototypicality when a member of a category is assigned the greatest number of attributes that it shares with other members of the category.
Now, the subjects who attribute to vehicles in general only two properties (of moving and of transporting people), tend to identify a motor car as the prototype of vehicle (with about twenty-five characteristic features) and to put the bicycle or boat on lower levels, while reserving the lowest places in the ranking for the lighter-than-air vehicle and finally the elevator. The elevator is attributed with only two properties (of moving and of transporting people).42
But in that case, the elevator ought to be the prototype of Vehicles, seeing as it presents precisely those properties common to any vehicle and would therefore allow us to relate even the most diverse species and tokens to vehicles. In any categorial order, the superordinate genus must have fewer features than the subordinate species, and the species fewer of the individual tokens that make recognition possible. If the CT for dog provided instructions for «constructing» a Pekinese and nothing else, it would be hard to apply it to an Irish wolfhound. If a prototype (where a classificatory system has already been established) and a CT have anything in common, it would be that both ought to have maximum extension and minimum intension. Instead the prototype has minimum extension and maximum intension.
It seems to me that the notion of prototype has a value for making clear the «borders» of a basic category. If it is decided that the salient features of the superordinate category of birds are beak, feathers, wings, two feet and the ability to fly, it is natural that there be some difficulty over defining the hen fully as a bird, because it does not fly but, at best, flaps about (and yet it is not excluded, because it must be admitted that other birds do not cease being birds when they are not flying). It strikes me that what is more debatable here regards the identification of what the prototype is, because I believe that such identification depends on environmental experiences and that judgments of prototypicality have more value for cultural anthropological research than for the determination of cognitive mechanisms in general.43
3.7.4.3 The mysterious Dyirbal
In any experiment on classification, it is always the experimenter who proposes a subdivision into classes inspired by a certain cultural model, tending not only to obliterate «wild» forms of classification but also to presuppose a classification in which probably there are only morphological accidents devoid of a semantic counterpart.
A curious case of this kind is to be found in Lakoff (1977: 6), where reference is made (on the basis of other research) to the Dyirbal language (Australia), in which every term must be preceded by one of these words:
Bayi: men, kangaroos, possums, bats, most snakes, most fishes, some birds, most insects, the moon, storms, rainbows, boomerangs, some spears, etc.
Balan: women, bandicoots, dogs, platypuses, echidnas, some snakes, some fishes, most birds, fireflies, scorpions, crickets, the hairy mary grub, anything connected with water or fire, sun and stars, shields, some spears, some trees, etc.
Balam: all edible fruits and the plants that bear them, tubers, ferns, honey, cigarettes, wine, cake
Bala: parts of the body, meat, bees, wind, yam sticks, some spears, most trees, grass, mud, stones, noise, language, etc.
Lakoff is surprised that such «categorizations» are used by the natives automatically and almost without their being aware of it, and he seeks semantic and symbolic reasons to justify them. He finds, for example, that birds are classified with women, because they are held to be the spirits of dead women, but he does not manage to explain why the platypus is grouped with women, fire, and dangerous things—as you can see, I am not the only one for whom this animal is a source of continuous worry.
However, Lakoff notes that for speakers of the latest generations, who have lost almost all of the tongue of their fathers, there remain only Bayi for males and nonhuman living creatures, Balan for human females, and Bala for all the rest, and, reasonably, he connects the phenomenon with the influence of the English pronominal system (He, She, It). A correct observation, which nonetheless would encourage one to go further—beyond English, I mean to say. Let is suppose that in a Mediterranean peninsula there lives a singular population whose natives have the curious habit of putting one of two words before every noun: il (with the variant lo) or la, with the following «categorial» effects:
Il is applied to men, kangaroos, bats, many snakes (boa, python, cobra), many fish (bass, pike, swordfish, shark), many insects (hornet, ladybird), sun, guardian, storm, rainbow, boomerang, wagon, rifle, machine, pistol, platypus, rhinoceros.
La is applied to women, sentinel, the tiger, locomotive, some snakes (viper, grass snake), some fish (trout, gilthead), many birds (swallow, great tit), insects (wasp, fly), water, moon, star, armor, pistol, spear, some trees (oak, palm), giraffe, skunk.
As we well know, grammatical gender has nothing to do with sexual gender and not even with any classification that on a conceptual level puts sentinel on the same side as locomotive and moon, and the sun on the same side us guardian and wagon. At the end of the day, we could even suppose that to the north of that peninsula, on the other side of a mountain range, there lives another (extremely barbarous) population that, like the young Dyirbal, put one of three words before every term: der, die, and das (perhaps owing to an effect of «pidginization,» under the influence of the English pronominal system), but that in this case the fact that the sun is die like woman, the moon der like the leopard and the tiger, and that the platypus, the ear, and gold are all das is of no categorial importance.
I am not suggesting in the slightest that something occurs in the Dyirbal language that is similar to what occurs in Italian, German, French, and many other languages. I am merely expressing the suspicion that grammatical phenomena are often discussed as phenomena of classification—which casts a shadow over many investigations in which classifications familiar to the experimenter but not shared by the subjects are presumed, or in which the experimenter vainly struggles to deduce classifications where the subjects do not classify at all and merely follow grammatical automatisms.44
3.7.5. Other Types
I intend to restrict myself to those cases in which objects or events of actual or possible perceptual experience are in question, rather than go further into what happens when we talk of the Bank
