From the Tree to the Labyrinth

11

The Language of the Austral Land

The subject of a perfect language has appeared in the cultural history of every people. Throughout the first period of this search, which continued until the seventeenth century, this utopia consisted in the search for the primigenial Hebrew in which God spoke to Adam or that Adam invented when giving names to the animals and in which he had had his first dialogue With Eve. But already in Dante’s De vulgari eloquentia another possibility had been broached: that God had not given Adam primordial Hebrew but rather a general grammar, a transcendental form with which to construct all possible languages.

But this possibility was situated on the two horns of a dilemma. On the one hand, it was possible to conceive of a Chomskian God, who gave Adam some deep syntactical structures common to every language subsequently created by the human race, obeying a universal structure of the mind (without waiting for Chomsky, Rivarol, an eighteenth-century author, had defined French as the language of reason, because its direct order of discourse reproduces the logical order of reality). On the other hand, it could be supposed that God had given Adam some semantic universals (such as high/low, to stand up, to think, thing, action, and so on), a system of atomic notions by means of which every culture organizes its own view of the world.

Until the arrival of Humboldt, even if one accepted the so-called Epicurean hypothesis by which every people invents its own language to deal with its own experience, no one dared prefigure anything similar to the Sapir-Whorf hypothesis: that it is language that gives form to our experience of the world. Thinkers like Spinoza, Locke, Mersenne, and Leibniz admitted that our definitions (of man, gold, and so on) depend on our point of view about these things. Nobody, however, denied that it was possible to design a general system of ideas that somehow reflected the way the universe works.

Still, even before Dante, Ramon Llull had conceived the idea that there were universal notions, present in the language and in the thought of every people; he even believed that, by articulating and combining these concepts common to all men, it would be possible to convince the infidels—namely, the Muslims and the Jews—of the truth of the Christian religion.

This idea was revived at the dawn of the seventeenth century, after the discovery of Chinese ideograms, which were the same in Chinese, Japanese, and Korean (though pronounced differently), for these different peoples referred to the same concepts. The same thing, it was said, happens with numbers, where different words refer to the same mathematical entity. But numbers possessed another attractive aspect: independently of the variety of languages, all peoples (or very many of them) indicated them with the same cipher or character.

The idea that began to circulate, especially in Anglo-Saxon circles, inspired by the Baconian reform of knowledge, was this: postulate a priori a system of semantic universals, assign to each semantic atom a visual character or a sound, and you will have a universal language. As for the grammar, it would be a question, according to the project, of reducing the declensions or the conjugations themselves in order to derive the various elements of speech from a same root, indicating them with diacritical signs or some other criterion of economy.

The first idea of a universal character appeared in Francis Bacon and was to produce in England an abundant series of attempts, of which we would mention only those of George Dalgarno, Francis Lodwick, and John Wilkins. These inventors of languages, which will be called philosophic and a priori, because they were constructed on the basis of a given philosophical view of the world, no longer aimed merely at converting the infidel or recovering that mystic communion with God that distinguished the perfect language of Adam but rather at fostering commercial exchange, colonial expansion, and the diffusion of science. It is no accident that most of these attempts were linked to the work of the Royal Society in London, and many of the results—apparent failures—of these utopists contributed to the birth of the new scientific taxonomies.

But this project, even if abundantly stripped of the mystic-religious connotations of earlier centuries, had another feature in common with the yearned-for perfect language of Adam. It was said of Adam that he had given “proper” names to things, the names that the things should have as they expressed their nature. In earlier centuries and still in the heyday of the occult and the kabbalistic speculations of the seventeenth century (consider, above all, Athanasius Kircher), this kinship between names and things was understood in terms of onomatopoeia, on the basis of far-fetched etymologies. To give an idea of the flavor of these ways of thinking, it suffices to quote Estienne Guichard (L’harmomie étymologique des langues, Paris: Le Noir 1606), where, for example, the author shows how from the Hebrew word batar was derived the Latin synonym dividere (147). Shuffling the letters, the word becomes tarab, and from tarab derives the Latin tribus, which then leads to distribuo and finally to dividere. Zacen means “old”; transposing the radicals one gets zanet, whence the Latin senex, and with a subsequent shift of letters comes zanec, whence in Oscan casnar, from which the Latin canus would be derived (247).

In subsequent attempts, the criterion of correspondence, or isomorphism between word and thing, is, by contrast, “compositional”: the semantic atoms are named arbitrarily, but their combination is motivated by the nature of the designated object. This criterion is similar to that followed by chemistry today: calling hydrogen H, oxygen O, and sulphur S is surely arbitrary, but calling water H₂O or sulfuric acid H₂SO₄ is motivated by the chemical nature of these compounds. If either the order or the nature of the symbols were altered, another possible compound would be designated. Naturally this language is universal because, while each people indicates water with a different linguistic term, all are able to understand and write chemical symbols in the same way.

The search for a priori philosophic languages and the impassioned debates and rejections they inspired are evidenced by those pages in Gulliver’s Travels where Swift imagines an assembly of professors bent on improving the language of their country. The first project, you will recall, was to abbreviate speech, reducing all polysyllables to monosyllables and eliminating verbs and participles. The second tended to abolish all words completely, because it was quite possible to communicate by displaying things (a difficult project because the so-called speakers would be obliged to carry with them a sack containing all the objects they planned to mention).

But even earlier the subject of the philosophic language had rightfully entered the literary genre of seventeenth-century utopias. For that matter, already in the Basel edition (1518) of More’s Utopia, published by Pieter Gilles, there was an illustration with writing in the language of that ideal island; Godwin spoke of the possible language of the Selenites in his Man in the Moone (1638); and Cyrano de Bergerac mentioned other-planetary languages on several occasions, both in Les estats et les empires de la lune (1657) and Les estats et les empires du soleil (1662).

Still, if we want two models of language that echo the a priori philosophical language of the Utopians, we must turn to two novels narrating journeys in the Austral Land, La Terre australe connue (1676), by Gabriel de Foigny, and L’Histoire des Sevarambes (1677–1679), by Denis Vairasse. Here, I will delve only into the first, which to me seems particularly instructive, because, as often happens with good caricatures, the parodistic deformation reveals some essential features of the caricatured object.

La Terre australe connue is naturally a work of the imagination. In distant, unknown lands an ideal community is supposedly discovered. In this ideal community the language, too, is ideal, and it is interesting to remark that Foigny writes in 1676, after the three significant a priori philosophic language projects have appeared: Lodwick’s Common Writing (1647), Dalgarno’s Ars Signorum (1661), and Wilkins’s Essay towards a Real Character (1668).

Foigny’s exposition, precisely because it is incomplete and a burlesque, takes up only a few pages of his ninth chapter, rather than the 500 (in folio!) of Wilkins’s, the most voluminous and complete of all the projects of that century. Yet is worth taking Foigny’s into consideration because, for all its terseness, it illustrates the advantages and limitations of a philosophic language. It reveals and magnifies—as only a parody can—the flaws of its models, but, as they are magnified, the better we are able to distinguish them.

In order to better understand Foigny it is useful to refer to Figure 11.1, where I try to extrapolate from his text a sort of Austral dictionary, along with some grammatical rules. Because the author is often reticent, I have inferred some rules from examples, while others remain unspecific (thus, for example, of thirty-six accidentals, I have been unable to reconstruct only eighteen).

Foigny’s Austral inhabitants,

to express their thoughts, employ three modes, all used in Europe: signs, voice, and writing. Signs are very familiar to them, and I have noticed that they spend many hours together without speaking in any other way, because they are ruled by this great principle: “that it is useless to employ several ways of action, when one can act with few.”

Figure 11.1

So they speak only when it is necessary to express a long series of propositions. All their words are monosyllabic, and their conjugations follow the same criterion. For example, af means “to love”; the present is la, pa, ma; I love, thou lovest, he loves; lla, ppa, mma; we love, you love, they love. They possess only one past tense, which we call the perfect: lga, pga, mga, I have loved, thou hast loved, etc.; llga, ppga, mmga, we have loved, etc. The future lda, pda, mda, I will love, etc., llda, ppda, mmda, we will love, etc. “To work,” in the Austral language, is uf: lu, pu, mu, I work, thou workest, etc.; lgu, pgu, ragu, I have worked, etc.

They have no declensions, no article, and very few words. They express simple things with a single vowel and compound things through vowels that indicate the chief simple bodies that make up those compounds. They know only five simple bodies, of which the first and most noble is fire, which they express with a; then there is air, indicated with e; the third is salt, indicated with o; the fourth, water, which they call i; and the fifth, earth, which they define as u.

As differentiating principle they employ the consonants, which are far more numerous than those of the Europeans. Each consonant denotes a quality peculiar to the things expressed by the vowels, thus b means clear; c, hot; d, unpleasant; f, dry, etc. Following these rules, they form words so well that, listening to them, you understand immediately the nature and the content of what they signify. They call the stars Aeb, a word that indicates their compound of fire and air, united to clarity. They call the sun Aab, birds are Oef, sign of their solidity and their aeriform and dry matter. Man is called Uel, which indicates his substance, partly aerial, partly terrestrial, accompanied by wetness. And so it is with other things. The advantage of this way of speaking is that you become philosophers, learning the prime elements, and in this country, nothing can be named without explaining at the same time its nature, which would seem miraculous to those unaware of the secret that they use to this end.

If their way of speech is so admirable, even more so is their writing … and though to us it seems very difficult to decipher them, custom makes the practice very simple.¹

Instructions in the manner of writing follow; here vowels are indicated with dots marked in different positions, while the thirty-six consonants of the alphabet are little strokes that surround the dots and are recognized by their angles. Foigny mentions these graphic devices obviously making fun of similarly complicated systems, such as, for example, Joachim Becher’s Character pro notia linguarum universalis (1661), which proposes a form of notation capable of completely muddling the reader’s ideas. He then continues, citing composites that can be achieved:

For example: eb, clear air; ic, hot water; ix, cold water; ul, damp earth; af, dry fire; es, white air.… There are another eighteen or nineteen, but in Europe we have no consonants corresponding to them.

The more you consider this way of writing, the more you will discover secrets worthy of admiration: b means clear; c hot; x cold; l wet; f dry; n black; t green; d nasty; p sweet; q pleasant; r bitter; m desirable; g bad; z high; h low; j red; a joined with i, calm. The moment a word is spoken, they know the nature of what it denotes: to indicate a sweet and desirable apple, they write ipm; nasty and unpleasant fruit is ind. I cannot explain all the other secrets that they understand and reveal in their letters.

The verbs are even more mysterious than the nouns. For example, they write and pronounce af, to say “to love”; a means fire, f means the scorching caused by love. They say la to mean “I Love,” which means the secretion that love produces in us; pa, “thou lovest,” sign of the lover’s sweetness; lla, “we love,” the double ll indicating the number of persons; oz means “to speak,” the letter o standing for salt, which seasons out speech, while z indicates the inhaling and exhaling necessary to speech.

When a child is being taught, the meaning of all the elements is explained to him, and when he unites them, he learns both the essence and the nature of all things he is saying. This is a wonderful advantage both for the individual and for society, because, when they have learned to read, as they always do by the age of three, they understand at the same time all the characteristics of all beings.

In this language the single letters are chosen arbitrarily, and each refers to a simple notion or to a thing. When compound entities are denoted, however, the syntax of expression appears isomorphic with reference to the content. Assuming that stars are a compound of fire and clear-colored air, the syntagm aeb expresses “naturally” the nature of the thing. The expression is isomorphic to the content, to such a degree that changing one element of the expression denotes a different content. In fact, aab does not mean stars; it means sun because (in the astronomy of the Austral Land) the sun is obviously a double, clear fire. In this sense the language of real characters is distinguished from the natural languages where, if month means a length of time, the relationship between noun and notion (or thing) in both cases is entirely arbitrary. In other terms, if, by mistake, we write catt, this does not indicate, say, a cat with an extra leg, whereas, if in the Austral language you write, or say, icc instead of ic, probably you want to indicate water not hot but boiling hot.

As I said earlier, the system recalls the language of chemical formulas: if you write H₂Au instead of H₂O in theory you indicate a different chemical compound. But here the first drawback of the system crops up. In chemistry, the system remains, so to speak, open (accommodating neologisms) in case an absolutely new compound has to be named, but the acceptance of the neologism is conditioned by the system of the content. Because in nature the number of known or admitted compounds is limited, one may confidently read H₂Au as a mistake, a misspelling, as it were. But in the Austral language, what happens if one rungs into the syntagm al? Must one admit the possibility that there exists a “wet fire”?

A problem of this sort emerged in connection with the semantic universals that Ramon Llull subjected to combinations and permutations, where the free combination of letters could theoretically produce an utterance repellent to the philosophical bases of the system into which it was introduced (or, in other words, a heretical utterance, such as “truth is false” or “God is lascivious”). But in these cases Llull considered null the theologically unacceptable combination. This also occurred because the letters denoted metaphysical entities that, in the realm of the theology of reference, were precisely defined. Bonitas est magna means that Goodness is great, but as Goodness was already defined in this way, it was impossible to conceive of its opposite, Bonitas est mala (Goodness is evil). Likewise, the Ars did not contemplate the possibility of metaphorical expression or even of periphrases. The primitive terms employed defined the entire universe of what was theologically sayable. Llull, with his perfect theological language, was not interested in talking about stars or hot water.

On the contrary, the Austral language uses a very limited battery of primitives but must serve to express every possible experience, that is, to replace through compositions of primitives the entire vocabulary. Thus, as can be seen from the quotation above, it must employ periphrases that, in Foigny’s satirical version, are highly questionable metaphors: apple becomes sweet and desirable water, and the act of loving is expressed as af (dry fire), or burning derived from the fire of passion. If dry fire means love, then why should wet fire not be able to mean metaphorically some other thing? The problem that arises, analyzing this caricature of language, is a serious problem: if a few primitives must denominate many things, it is indispensable to recur to periphrasis, and this is precisely what happens with the “serious” projects of Wilkins and Dalgarno. And the confines between periphrasis and metaphorical expression can become very hazy. In fact, in Dalgarno’s serious project compounds were introduced on the order of “animal + full-hoofs + spirited” to signify horse and “animal + full-hoofs + huge” to signify elephant.

The equally serious project of Wilkins was based on the fact that all ambiguities of language had to be reduced so that every sign would refer to a single, rigorously defined concept. But some metaphorical operators were introduced to allow the language to express entities for which no terms existed in the philosophical dictionary, whose format had inevitably to be reduced. Wilkins asserts that it is not necessary to have a character for calf because the concept can be reached by combining cow and young; nor does one need a primitive lioness, since this animal can be denoted by combining the sign for female with that for lion. Thus Wilkins develops in his grammar (and then transforms into a system of special signs in the part devoted to the writing and pronunciation of the characters) a system of “Transcendental Particles” intended to amplify or alter the character to which they are applied. The list contemplates eight classes amounting to a total of forty-eight particles, but the criterion that assembles them is not at all systematic. Wilkins harks back to Latin grammar, which makes use of endings/suffixes (that allow the creation of terms like lucesco, aquosus, homunculus); of “segregates” such as tim and genus (allowing the creation, from a root, of gradatim or multigenus); and determination of place (hence vestiarium) and agent (cf. arator). Some of his particles are without doubt of a grammatical nature (for example, those that transform masculine into feminine or adult into young). But Wilkins himself recurs also to the criteria of rhetoric, citing metaphor, synecdoche, and metonymy, and, in fact, the particles in the metaphorical-like category are simply signs of rhetorical interpretation. Thus, adding one of these particles to root one gains original, while adding it to light yields evident. Finally, other particles seem to refer to the cause–effect relation, or container–thing contained, or function–activity, as in the following examples:

like + foot = pedestal

like + blood = crimson

place + metal = mine

officer + navy = admiral

artist + star = astronomer

voice + lion = roar

From the point of view of linguistic precision, this is the weakest part of the project. In fact, Wilkins, who supplies a long list of examples of the correct application of such particles, warns that they are, in fact, examples. Therefore the list is open, and its enrichment depends on the inventiveness of the speaker. It seems almost that Wilkins, concerned about the mechanical quality of his language, is anxious to leave room for its users’ creativity. But once the user is free to apply these particles to any term, it is obvious that ambiguity will be hard to avoid.

And so the artificial language loses its one virtue: that of denoting always and only the same thing with the same character.

The Austral language (like the models it parodies) deliberately rejects the fundamental mechanism of every natural language, namely, double articulation. It is obvious how much double articulation (in which the units of second articulation are without meaning) can contribute to the free formation of neologisms. If, with three meaningless characters (p, c, f), I can compose six syntagma (pot, top, opt, pto, otp, tpo), and only three of these are admitted by the dictionary, the other three remain available for constructing neologisms or indicating the most subtle differences between otherwise similar entities. As long as they remain available, however, if they happen to appear in a context, they may be understood as errors in pronunciation or spelling.

Foigny’s system, on the one hand, allows the creation of neologisms only through metaphor and, on the other, obliges us to seek out a meaning for every syntagm admitted by the ars combinatoria, because even the slightest phonetic or orthographic change immediately reflects on the content and denotes a different (and possible) entity.

Finally, the last limitation of the Austral language is—as occurred with many a priori philosophical languages—the absolute casualness with which the primitives are chosen. We will not speak of the so-called Anonymous Spaniard (Pedro Bermudo), who in his 1654 Arithmeticus nomenclator classified the primitives, subdividing them into:

(1) Elements (fire, wind, smoke, ash, hell, purgatory, and center of the earth). (2) Celestial entities (stars, lightning, rainbow). (3) Intellectual entities (God, Jesus, discourse, opinion, suspicion, soul, stratagem or specter). (4) Secular states (emperor, barons, plebs). (5) Ecclesiastical states. (6) Artificers (painter or sailor). (7) Instruments. (8) Affects (love, justice, lust). (9) Religion. (10) Sacramental confession. (11) Tribunal. (12) Army. (13) Medicine (doctor, hunger, enema). (14) Brute animals. (15) Birds. (16) Reptiles and fish. (17) Parts of animals. (18) Furnishings. (19) Foods. (20) Beverages and liquids (wine, beer, water, butter, wax, resin). (21) Clothing. (22) Silken stuffs. (23) Woolens. (24) Canvas and other textiles. (25) Navigation and spices (ship, cinnamon, anchor, chocolate). (26) Metals and coins. (27) Various artifacts. (28) Stones. (29) Jewels. (30) Trees and fruits. (31) Public places. (32) Weights and measures. (33) Numerals. (34–42) Various grammatical categories. (43) Persons (pronouns, forms of address such as His Eminence). (44) Travel (hay, road, robber) …

But Wilkins himself, though he discussed his list with students of botany, mineralogy, and zoology, put under the heading of Economic Relations not only cases of kinship, in which distinctions appear distorted by criteria such as Progenitor/Descendant, Brother/Half-brother, or Coelebs/Virgin (Coelebs, however, comprises both the bachelor and the spinster, whereas Virgin seems to refer only to a female condition), but also acts that refer to intersubjective relationships, such as Direct/Seduce or Defense/Desertion. Among the Private Relations appear also Provisions, where we find Butter/Cheese but also Butchering/Cooking and Box/Basket.

Note the sly way that Foigny breaks the homogeneity of the list of the four classic elements by adding salt, which, if anything, would belong to another chemical-alchemistic taxonomy, including also mercury and sulfur. But the slyness is not gratuitous precisely because Wilkins added to the four elements a fifth, evident one: the Meteor.

As for the thirty-six accidentals, even if we know only eighteen of them, their heterogeneity is enough for us to infer that the list has prominent omissions. Here Foigny touches palpably the crucial question of the list of the primitives, and he resolves it more in the manner of the Anonymous Spaniard than in that of Wilkins, but only to insinuate (it seems) that, when it comes to incongruity, there is only a difference of degree between the two systems.

The final comic element in the Austral language is that it does not clarify when a letter has a lexical function or when it is morphemathic. It seems that l, m, and p—placed in the first position—function as pronouns. But, in analyzing pa (thou lovest), Foigny speaks of the sweetness of the lover. Thus he assigns two letters with morphemathic functions the meaning they have when they define accidentals. The solution is comic because it allows us to think that lu (I work) must be interpreted with reference to the sweat produced by the earth, but in that case why would there be sweetness in pu (thou workest)?

We cannot tell how consciously Foigny was being ironic about the fact that in the philosophical languages the entire grammar is semanticized, but this mischievousness is not to be overlooked.

Criticism of a priori philosophical languages for the most part appears, as I have shown, in French satirical works. Perhaps this is not an accident: it was in France that the first radical criticism of the project took shape in the serious works of Dalgarno, Wilkins, and Lodwick.

In 1629 the Minim friar Marin Mersenne sends his friend Descartes the project of a nouvelle langue by a certain des Vallées. In a letter to Mersenne on November 20, 1629, Descartes sends his impressions of that proposal. For every language, he says, it is necessary to learn a grammar and the meaning of the words. For the meaning of the words it would suffice to have a good dictionary, but the grammar is difficult. Nevertheless, if a grammar could be constructed free of the irregularities of the natural languages, which have been corrupted by use, the problem would be solvable. Thus simplified, this language would appear primitive compared to the others, which would appear as its dialects. And once the primitive terms were set (of which the terms of the other languages would be synonyms, such as aimer and to love), it would suffice to add the suffixes to obtain, for example, the corresponding substantive. Consequently, a system of universal writing could be developed in which every primitive term would be recorded with a number that would refer back to the synonyms in the different languages.

All the same, there would still remain the problem of the sounds to choose for these terms, inasmuch as certain sounds are pleasant and easy for one people and unpleasant for another. The sounds would thus be difficult to learn: if a speaker used synonyms in his own language for the primitive terms, then he would not be understood by speakers of other nations, except in writing. Yet learning the entire lexicon would require great effort, and if that were necessary, there would seem to be no reason not to use an international language already known to many, such as Latin.

Saying this, Descartes only repeated some ideas that were in the air in those decades. But at this point he saw that the central problem is something else altogether: to be able not only to learn but also remember the primitive nouns, these would have to correspond to an order of ideas, or of thoughts, that would have the same logic as the order of the numbers (where it is not necessary to learn them all but simply to generate them by succession). This problem coincides with another: that of a true philosophy able to define a system and distinct ideas. If a person were able to number all the simple ideas from which are then generated all the ideas that we are capable of thinking and to assign to each of these a character, we could then articulate, as we do with numbers, this sort of mathematics of thought. The words of our languages, on the other hand, refer to confused ideas.

In conclusion, Descartes affirmed: “Now I believe that this language is possible and that the learning on which it depends could be found, by which peasants will be able to judge the truth better than philosophers do now. But I have no faith in ever seeing it used; it presupposes great changes in the order of things, and the whole world would have to be nothing more than an earthly paradise, which can be proposed only in the land of novels.”

The criticism of Descartes was correct. Every attempt to establish an architectonically perfect system of ideas composed of mutual dependences and strict classification from the general to the particular would prove to be a failure. At the end of the eighteenth century Joseph-Marie de Gérando, in Des signes, would isolate the secret termite that was gnawing at all the previous systems: either you create a logical dictionary confined to a very limited notional field or an encyclopedia of all our knowledge, that is, either a necessary but insufficient order of concepts or the flexible, infinitely amplifiable and variable order of a library.

On the other hand, Leibniz would acknowledge (in his Nouveaux essais sur l’entendement humain) that, having to depict the entire system of our learning, we would have a library where the doctrine of spirits could come under logic but also under morality, and all could come under the practical philosophy to the extent that it contributes to our happiness. A memorable story can be placed in the annals of universal history or in the specific history of a country or even in the biography of an individual. Anyone who is organizing a library often encounters the problem of deciding in which section a book should be cataloged.

So the only thing to do would be to essay a polydimensional encyclopedia (a hypertext, as we would say today). We can almost hear, in advance, the project that would be theorized by D’Alembert at the beginning of the Encyclopédie, where he speaks of the Système Général des Sciences et des Arts as a labyrinth. The philosopher is he who can discover the secret routes of this labyrinth, its temporary branches, the reciprocal dependences that compose this enclosure like a globe. Consequently, “one can create as many different systems of human knowledge as there are world maps having different projections.… But often such an object, which because of one or several of its properties has been placed in one class, belongs to another class by virtue of other properties and might have been placed accordingly.”²

The criticism of the Encyclopédíe puts an end to the dream of the grammar of ideas, even though further attempts would follow, down to our own day, when scholars are still studying the possibility of a so-called mentalese, a language written in the very convolutions of our brain, capable of supplying the deep structure of every expression in any natural language.

But as Descartes had announced, it is not impossible to write of ideal languages in the land of novels. Foigny did it, and two and a half centuries later, Borges was to do it, too.

In Other Inquisitions, Borges (1964), studying “the language of John Wilkins” (which, by his explicit admission, he knew only through an encyclopedia entry), recognizes at once the incongruity of the classification of the Wilkinsian semantic primitives (he discusses specifically the subdivisions of stones), and it is in this same brief text that he invents the Chinese classification that Foucault quotes at the opening of Les mots et les choses. In this Chinese encyclopedia, entitled Celestial Emporium of Benevolent Recognitions, “it is written that the animals are divided into (a) those that belong to the Emperor, (b) embalmed ones, (c) those that are trained, (d) suckling pigs, (e) mermaids, (f) fabulous ones, (g) stray dogs, (h) those that are included in this classification, (i) those that tremble as if they were mad, (j) innumerable ones, (k) those drawn with a very fine camel’s hair brush, (l) others, (m) those that have just broken a flower vase, (n) those that resemble flies from a distance.”³ Borges comes to the conclusion that no classification in the universe is not arbitrary and conjectural. But if it has to be arbitrary and conjectura1, why not leave room not for the satire of utopian projects but for the utopia of linguistic fancy?

Borges, on at least two other occasions, returns to the question of ideal languages. In “Dr. Brodie’s Report” Borges (1976) he examines the monosyllabic language of the Yahoos.

Each monosyllabic word corresponds to a general idea whose specific meaning depends on the context or upon accompanying grimaces. The word “nrz,” for example, suggests dispersion of spots and may stand for the starry sky, a leopard, a flock of birds, smallpox, something bespattered, the act of scattering, or the flight that follows defeat in warfare. “Hrl,” on the other hand, means something compact or dense. It stands for the tribe, a tree trunk, a stone, a heap of stones, the act of heaping stones, the gathering of the four witch-doctors, carnal conjunction or a forest. Pronounced in another manner or accompanied by other grimaces, each word may hold an opposite meaning.⁴

This language of the Yahoos is not at all impracticable, as it seems at first glance. Note that the apparent polysemia of the term is, so to speak, held together by certain primitive special signs common to all its meanings. The grimaces that accompany the emission of sound function like the metaphorical operators of Wilkins. For the rest, the language simply carries to extremes the tendency of actual natural languages to contain expressions that mean different things in different contexts, and Borges hastens to remind his readers that this should not be surprising; after all, in English, to cleave means both “to split” and “to cling to.”

Finally, in “Tlön, Uqbar, Orbis Tertius” Borges (1998) speaks of a language structured spatially and not temporally, which proceeds not through agglutinations as in the languages so far examined but only by expressing temporal flow. In this language, nouns do not exist, but only impersonal verbs qualified by monosyllabic suffixes and prefixes with adverbial value. In brief, “there is no word corresponding to the word “moon,” but there is a verb which in English would be “to moon” or “to moonate.” “The moon rose above the river” would thus be written hlör u fang axaxaxas mlö, or literally: “upward behind the onstreaming it mooned” (which sounds like a quote from Joyce’s Finnegans Wake).⁵

The failure of the utopias of the a priori philosophical language has thus produced some interesting experiments in the Land of Novels that, instead of constructing perfect linguistic systems, have demonstrated how our imperfect languages can produce texts endowed with some poetic virtue or some visionary force. I consider this no small achievement.

This translation is a slightly revised version of the essay translated by William Weaver in Eco (1998b).

1. The passage quoted here and subsequent quotations from the same author were presumably translated by Weaver from the Italian version cited by Eco, namely, Gabriel de Foigny. La terra australe, trans. Maria Teresa Bovetti Pichetto. Napoli: Guida, 1978. No page references are provided.

2. English translation: D’Alembert (1963: 46–49).

3. See “The Analytical Language of John Wilkins” in Borges (1964: 103).

4. See “Doctor Brodie’s Report” in Borges (1972: 117).

5. See “Tlön, Ucbar, Orbis Tertius” in Borges (1962: 8).