Saturday, February 2, 2013

Umberto Eco Pleads for Communication--Esperanto

The Dream of a Perfect Language
A lecture presented by Umberto Eco
November 26, 1996
Last year appeared the English translation of my book on The Search for a Perfect Language. It is 400 pages long and I have no intention of summarizing here a book which is in itself a summary: Der Turmbau von Babel by Arno Borst, which concerns only the historical discussion on the confusion of tongues (not the search for a perfect one), is 6 big volumes long, and the current literature on perfect languages lists hundreds and hundreds of projects. After the publication of my book I am pursued by people asking me what a perfect language is and how does it work. I repeat that if one writes a book on Donald Duck this does not mean that Donald Duck exists or that the author believes it exists. The search for a perfect language represented and still represents an impossible dream.
However such a dream has obsessed mankind through the centuries and this evening I'll to pick up only some aspects of my story in order to show that a reflection on perfect languages can help us to understand some of the reasons why our natural and abundantly imperfect languages work so well.
The myth of the confusion of tongues can be found in every culture. Sticking to the European culture and to its origins, my story has the advantage of beginning at the Beginning. Genesis (2, 19 ff), says that "out of the ground the Lord God formed every beast of the field, and every fowl of the air: and brought them unto Adam to see what he would call them." It is not at all clear on what basis Adam actually chose the names he gave to the animals. The Vulgate says that Adam called the various animals nominibus suis, that is, "by their own names." Nice question. Did Adam give the animals the names that, by some extra-linguistic right, were already due to them, or did he give them those names we still use on the basis of his conventional decision? The common traditional opinion was that Adam gave the animals names that reflected their inner essence.
For the rest, a long discussion lasted until the XVII century, whether Adam did or did not give names to fishes - since they are not mentioned in the Bible and it is difficult to figure out how God succeeded in bringing them into the Garden.
In Genesis 11 we are told that after the Flood, "the whole earth was of one language, and of one speech". Yet, human beings in their vanity conceived a desire to rival the Lord, and thus to erect a tower that would reach up to the heavens. To punish their pride and to put a stop to the construction of the Babel tower, God confused their languages.
Latin and Greek civilization were not disturbed by the plurality of languages. They simply identified their own tongue as the language of Reason, and considered the Barbarians as stuttering people, speaking no language at all. Even the First Church Fathers did not feel a great embarrassment because of the confusion of tongues. They assumed that Hebrew was the original idiom of mankind, but they peacefully wrote in Greek or Latin.

The confusion of languages became a crucial problem only when Europe faced the birth of the vulgar tongues. We cannot find representations of the Tower of Babel before the fifth or sixth century, and very few until the eleventh century. After this, however, there is a flood of towers. FIGURE Confronted with the drama of linguistic fragmentation, the European culture started speculating about a remedy for linguistic confusion. Some looked backwards, trying to retrieve the language spoken by Adam. Others looked ahead, thinking of Language of Reason possessing the perfection of the lost speech of Eden.
In the seventh century Irish grammarians, in a work titled The precepts of the poets, said that that the Gaelic language was created after the confusion of tongues by the 72 wise men of the school of Fenius, through a curious `cut and paste' operation on all the languages born after the dispersion - so that the best of every language was selected and retained in Irish, which was perfect because preserved the original isomorphism between words and things.
In a way Dante Alighieri tried to do the same: in his De vulgari eloquentia, since the perfect forma locutionis (probably for Dante a sort of universal grammar) had disappeared with the disaster of Babel, he took as his task the creation of a volgare illustre, a noble vernacular, to be chased as a "perfumed panther", which had to display the same revelatory perfection of the lost language of Adam.
It is very difficult to reconcile the invention of a new national language with the mystical persuasion that the only perfect idiom was the Hebrew spoken by Adam. If Dante really believed so, he should have written his Divine comedy in Hebrew. Likewise, in the XV century Pico della Mirandola tried to revive the Kabbalistic tradition by manipulating the biblical Hebrew (badly learned from his friend Flavius Mithridates) but in fact he wrote in Latin his pseudo-Hebrew lucubrations. Perfect languages are very difficult to master.
In the XVI and XVII century many tried to go back to Hebrew. In 1667, Mercurius van Helmont published an Alphabeti veri naturalis Hebraici brevissima delineatio where, in order to teach the deaf-mutes, shows why Hebrew is the only language that can be learned in a natural way: in making the sounds of Hebrew the movements of tongue, palate, uvula, and glottis reproduced the shapes of the corresponding Hebrew letters. Figure
For other authors the problem was to demonstrate that Hebrew was so perfect that its virtues had been transferred to the Western languages. In 1606, Estienne Guichard wrote his L'harmonie étymologique des langues starting from the premise that Hebrew was the simplest language because in it "all words are simple, and their substance consists of but three radicals". Manipulating these radicals through inversion, anagrams, and permutations in the best Kabbalistic tradition, Guichard provided his etymologies. In Hebrew, the verb batar means to divide. How can we prove that Latin dividere comes from batar? Simple: by inversion, batar produces tarab; tarab then becomes the Latin tribus and, from there, turns into distribuo and dividere. Zacen means old. Rearranging the radicals, we get zanec from which derives Latin senex. A further rearrangement and we have cazen, from which derives the Oscan word, casnar, which is the root of the Latin canus, elder. Obviously, by this method we might equally prove that the English head comes from the late Latin testa, since the anagram of: testa gives eatts (Figure).
Such a furious search for etymologies eventually ended by implementing serious work in biblical studies and comparative linguistics, but it also produced a curious "nationalistic" trend. If the language one speaks directly derives from Hebrew, why then not to consider one's national idiom as the most perfect one?
For Giovanni Nanni, or Annius (Commentatio super opera diversorum auctorum de antiquitatibus loquentium of 1498), the Tuscan dialect was a perfect one because, before it was colonized by the Greeks, Etruria had been settled by Noah and his descendants. This thesis was received by

Guillaume Postel, a passionate student of the Hebrew tradition, who simply preferred to transfer the Etruscan heritage to the Celts. In the Flanders Goropius Becanus (Jan van Gorp) in his Origines Antwerpianae of 1569 demonstrated on etymological grounds that the original perfect language was Dutch, particularly the dialect of Antwerp: the ancestors of Antwerpians were the Cimbrians, the direct descendants of the sons of Japheth, who had not been present under the Tower of Babel, and, consequently, had been spared the confusio linguarum. Becanus claimed that his thesis was also proved by the facts that Dutch had the highest number of monosyllabic words, possessed a richness of sounds superior to all other languages, and favored at the highest degree the formation of compound words.
A Swedish candidacy was supported in 1671 by Georg Stiernhielm (De linguarum origine praefatio) and in 1688 his fellow-countryman Andreas Kempe, in his Die Sprachen des Paradises narrates of a conversation between God and Adam, where God speaks in Swedish and Adam in Danish, and Eve is seduced by a French-speaking serpent. This was certainly a parody, but was a symptom of a real debate. Olaus Rudbeck, in his Atlantica sive Mannheim vera Japheti posterorum sedes ac patria of 1675, demonstrated that Sweden was the home of Japheth and his line, and that from this racial and linguistic stock all the Gothic idioms were born. Rudbeck identified Sweden with the mythical Atlantis, the land of the Hesperides, from which civilization had spread to the whole world.
The idea of a German linguistic primacy existed even before Luther, for whom German was the language closest to God. In 1533 Konrad Pelicanus (Comnentaria bibliorum) set out the analogies between German and Hebrew, and in the Baroque period, Georg Philipp Harsdörffer (Frauenzimmer Gesprächspiele, 1641) claimed that the German language:
speaks in the languages of nature, quite perceptibly expressing all its sounds. [...] It thunders with the heavens, flashes lightening with the quick moving clouds, radiates with the hail, whispers with the winds, foams with the waves, creaks with the locks, sounds with the air, explodes with the canons; it roars like the lion, lows like the oxen, snarls like the bear, bells like the stag, bleats like the sheep, grunts like the pig, barks like the dog, whinnies like the horse, hisses like the snake, meows like the cat, honks like the goose, quacks like the duck, buzzes like the bumble bee, clucks like the hen, strikes its beak like the stork, caws like the crow, coos like the swallow, chirps like the sparrow. [...] On all those occasions in which nature gives things their own sound, nature speaks in our own German tongue. For this, many have wished to assert that the first man, Adam, would not have been able to name the birds and all the other beasts of the fields in anything but in our words, since he expressed, in a manner conforming to their nature, each and every innate property and inherent sound; and thus it is not surprising that the roots of the larger part of our words coincide with the sacred language.
Leibniz condamned with heavy irony these and other theories, but he supported the Celto- Scythian hypothesis, according to which there was a Celtic language common both to the Gauls and the Germans, deriving from the Scythians, and in some way having Semitic origins. Thus he concluded that "Teutonic has best preserved its natural and Adamitic aspect" , and that the Germanic language seemed most primitive. We must take into account that every instance of linguistic nationalism always has political reasons. Do not forget that even Heidegger said the only good languages for philosophy are Greek and German (and this is why it is so difficult to translate Heidegger into English).
In the British context, Rowland Jones (The Circles of Gomer, 1771) argued that "the Celtic dialects and knowledge derived their origin from the circles of Trismegistus, Hermes, Mercury or Gomer... [and] the English language happens more peculiarly to retain its derivation from that purest fountain of languages. In the same century Antoine de Rivarol (De l'universalité de la langue française 1784) wondered why to look for a universal language, since a perfect one already existed and was French. In French the word-order (first subject, then verb, and last object) mirrors a natural logic in accordance with the requirements of common sense, its phonetic

system guarantees sweetness and harmony, and in comparison with French German is too guttural, Italian too soft, Spanish too redundant, English too obscure.
Let me now consider another effect produced by the fascination of Hebrew tradition. In order to find the roots of this approach we must go back to an Hebrew treatise written between the third and fourth centuries, the Sefer Jetzirah, or Book of Creation, in which it is outlined the idea that the world was created by God through a manipulation of alphabetic letters.
The principle of the permutation of the letters offered two possibilities. The permutation technique could be applied to the whole text of the Torah (with its thousands and thousands of letters),or to the series of the names of God, and this was typical of the so-called Kabbalah of Names, which permitted ecstatic experiences. But the Sefer Jetzirah also suggested that the permutation could be also implemented on the finished packet of the 22 letters of the alphabet. FIG. (page from Sefer)
Twenty-two foundation letters... And He created with them the whole creation and everything to be created in the future... He fixed them on a wheel like a wall with 231 gates and He turns the wheel forward and backward. (ii, 3)
How did He combine, weigh and interchange them? Aleph with all and all with Aleph; Beth with all and all with Beth...(ii, 5) How did He combine them? Two stones build two houses, three stones build six houses, four stones build twenty four houses, five stones build a hundred and twenty houses, six stones build seven hundred and twenty houses, seven stones build five thousand and forty houses. Begin from here and think of what the mouth is unable to say and the hear is unable to hear" (iv, 16).
Not only mouth and hear are unable to spell out what happens afterwards, but even a personal computer would have severe difficulties in doing it. In mathematics this is called factorial calculus. Figure Kircher. With a finite alphabet one can produce a finite but enormously high number of combinations. It is easy to imagine how this dizzying number of possibilities must have appealed to those who dreamed of a perfect language. If each element of the alphabet corresponded to an idea, then the language would offer the possibility of combining them so producing true philosophical or theological propositions. And if these ideas represented things, then this language would allow one to discover all the connections of reality.
Such was the ars magna of Raymond Lully (XIII-XIV century). FIG. Wheel and rows. To each of the nine letters represented in the wheel we can assign five rows of concepts, namely nine Principia Absoluta, nine Principia Relativa, nine Questions, nine Subjects, nine Virtues and nine Vices. Once assigned a given set of values to the letters, one starts rotating the wheels and can produce a very high number of concepts. Lull thought of his mathematically perfect language as an intellectual instrument for the conversion of the infidel. According to the legend, he went among the Muslims, showed his wheel, and - unconvincing as it method was - was killed.
Lully was however still thinking of a finite system of primitive notions portraying the structure of a rational universe. In fact the definitions he gives of his basic elements severely limits their possibility of combination. If the Ars produced a proposition like the world is eternal, it had to be rejected because it was wrong. Lully acts as one who mixes up the letters ABC in order to produce anagrams. Three items can produce six anagrams but, according to the English lexicon, only CAB and BAC make sense, and the rest must be disregarded as irrelevant. In Lully's ontology, a very structured system of truths limits the chances offered by the combinatorial art.
But Lully's idea went beyond Lully's intentions. Once proposed the possibility that different concepts can be assigned to the letters, why to stop at a finite set of conceptual units? Moreover,

why to bound from the beginning the letters to concepts? Figure Selenus. See for instance how in 1624 Gustavus Selenus conceived of a cipher which could combine 25 series of 24 tuples, where any solution can be correct since the cipher simply transforms symbols into symbols, and the procedure is uninterested in the semantic value of the coupling. If with Lully there was an elementary syntax limited by a strictly limited semantics, now it is possible to think of more complex syntax, without any semantic limitation.
In 1622, Paul Guldin wrote a Problema aritmeticum de rerum combinationibus in which he calculated the number of possible locutions generated by 23 letters. He took into account neither the question of whether the resulting sequences had a sense, nor even if they were capable to being pronounced at all. The locutions could consist of anything from 2 to 23 letters; he did not allow repetitions. He arrived at a result of more than 70,000 billion billions. To write out all these locutions would require more than a million billion billions letters. To conceive of the enormity of this figure, he asked the reader to imagine writing all these words in huge books: each of these books had a thousand pages; each of these pages had a hundred lines; each of these lines could accommodate 60 characters. One would need 257 million billions of these books. Where would you put them all? Guldin then made a careful volumetric study, imagining shelf space and room for circulation in the libraries that might store a consignment of these dimensions. If you housed the books in large libraries formed by cubes whose sides measured 432 feet, the number of such cubical buildings (hosting 32 millions of volumes each) should be more than 8 billions. And where then would you put them all? Even exhausting the total available surface space on the planet earth, one would still find room for more than 7 billions of books!
In 1636 father Marin Mersenne, in his Harmonie universelle asked how many musical sequences that can be generated upon an extension of three octaves, comprising 22 notes, without repetitions (shades of future twelve-tone compositions!). Mersenne observed that to write down all these songs would require enough reams of paper to fill in the distance between heaven and earth, even if every sheet contained 720 songs and every ream was so compressed so as to be less than an inch thick. In fact the number of possible songs amounted to 1.124 billions of billions. By dividing this figure for the 362, 880 songs contained in each ream, one would still obtain a 16 digits figure, whilst the number of inches between the center of the Earth and the stars is only 28 thousands of billions (a 14 digits figure). Anyone who wished to copy out all these songs a thousand per day would have to write for more than 22 billion years.
Mersenne and Guldin were anticipating Borges' Babel Library ad abundantiam. Not only this, Guldin observed that if the numbers are these, who can marvel at the existence of so many different natural languages? The Ars was now providing an excuse for the confusio linguarum. It justifies it, however, by showing that it is impossible to limit the omnipotence of God.
Are there more names than things? How many names, asks Mersenne (Harmonie, II, 72), would we need if we were to give more than one to each individual? If Adam really did give names to everything, how long would he have had to spend in Eden? In the end, the languages of man limit themselves to the naming of general ideas and of species; to name an individual thing, an indication with a finger is usually sufficient (p. 74). If this were not so, it might easily "happen that for every hair on the body of an animal and for each hair on the head of a man we might require a particular name that would distinguish it from all others. Thus a man with 100,000 hairs on his head and 100,000 more on his body would need to know 200, 000 separate words to name them all" (p 72-73). Yet such an artificial language would place man in competition with God, who has the privilege of knowing all things in their individual nature.
This capacity to conceive of a quasi-infinite series of combinations depends on the fact that, unlike Lull, Mersenne and others were no longer calculating upon concepts but rather of simple alphabetic sequences, pure elements of expression with no inherent meaning, uncontrolled by no orthodoxy other than the limits of mathematics itself.

Such was the way followed by Leibniz - at least in some of its proposals for a universal language - a way which led eventually to modern formal logic. Lullian Ars Magna thus became step by step a cogitatio caeca, a blind calculus between free variables, to which an indefinite number of meanings can be bound, so that the combinational laws of the expressions can lead to the discovery of new possible connections between things or ideas. FIG. Leibniz.
The two case I have up to now considered are opposite both in their aims and in their results. (I) By electing a given natural language as the heir of the primeval Hebrew, one makes certainly a strong ideological statement, but one does not find a perfect substitute for a pre-existing natural language. Be it Dutch or German, Sweedish or English, such a privileged language will still display all the flaws of the natural ones and it would be difficult to demonstrate how it really mirrors the nature of things. (ii) By looking for a strictly formalized language one can maneuver, as Lully did, abstract theological concepts, or, as Lebniz proposed, mathematical entities of philosophical ideas, but cannot speak, neither perfectly nor imperfectly, of our everyday affairs, as the imperfect post-Babelic natural languages do.
Is there a perfect language which can really substitute, in a better and more precise way, natural languages? This was the problem of the inventors of so called a priori languages of the XVII century, among whom the most skilled and famous was John Wilkins. I must say that I have a great respect for Wilkins and that in my book I tried to show that he was the first to have (albeit in a confused way) the first notion of a Hypertext. His attempt was a paramount one and his failure was due, among other reasons, to the fact that he was trying to establish a scientific a taxonomy at a time in which natural sciences were still groping their own way. But let me say how, even though Wilkins had implemented such a perfect taxonomy, his project would have collapsed.
In order to understand what happened (and I am sorry to ask you this scholarly effort), let us use as parameter the model for a semiotic system supplied by Hjelmslev (FIGURE)
Every semiotic system can be analyzed in two planes, Expression and Content. Each plane can be subdivided into form and substance, and each arises through organizing a still unshaped continuum. For natural languages, the expression-form is represented by the phonological system, by the lexical repertoire, and by the rules of syntax. Realizing through concrete utterances the possibilities provided by the expression-form, we produce expression-substances, like the words that I am actually uttering.
The content-continuum represents everything we can talk or think about: it is the universe, or reality (physical or mental) to which our language refers. Each language however organizes the way in which we talk or think about reality in its own particular way, through a content-form (let us say very simply that we recognize objects such as trees, animals, square roots, parental relationships and so on).
In order to be able to convey meaning, a natural language must establish a correlation between elements (or units) of the expression-form and elements (or units) of the content-form. In verbal languages, Expression-form and Content-form are not conformal. This means that they are structured according to different criteria: the relationship between the two planes is arbitrary, and variations of expression do not automatically imply a point-to-point variation of the corresponding content. There are also of cases of conformality: think for instance of a clock, where the movement of the hands corresponds to the movement of the Earth around the Sun, but the slightest movement of the hands corresponds to a given movement of the Earth. The two planes are point-to-point conformal. On the contrary, in a verbal language, if, instead of dog, we utter log, we do not mean a different kind of dog, or of animal, but something radically different.
The fact that natural languages are non conformal can be translated in terms of double articulation. The units of first articulation (let me say, words) do convey meaning, while their

phonetical components (units of second articulation) do not. If I draw a trilobate figure assuming that it represent a clover, and then I add a lobe, I have represented a four-leafed clover: the two planes are conformal. If on the contrary I write the word clover and then I re-write it, let me say, with a double final r o with a double l, I have not denoted a four-leafed clover, as well as, if I write cover, by eliminating the l, I have not represented a clover deprived of one leave. The two planes are non conformal. Such an arbitrariness always looked an infelicitous result of the Babel curse, to the eyes of the searchers for a perfect language. How beautiful would be a language where the word for dog not only naturally and evidently portrayed all the properties of a dog, but also where every minimal modification of the world expressed any minimal variation of the nature, the behavior, the properties of a dog.
In order to design characters that directly denote notions (if not the things themselves that these notions reflect), it was necessary to design a system of primitive notions, by composition of which every thought could be expressed. That is why these languages were called as philosophical and a priori.
In his Essay towards a Real Character John Wilkins constructed a table of 40 Major Genera subdivided into 251 characteristic Differences. From these he derived 2,030 Species which appear in pairs. Then Wilkins assigned a real character to every Genus, Species and Difference. In fact, he proposed two types of language: (i) the first, an ideogrammatic form of writing, vaguely Chinese in aspect, was destined to appear in print but never to be pronounced, (ii) the second, were the characters were intended to be pronounced. Since it is easier to understand the latter, I will take from it the few example which follow.
FIGURE represents Wilkins' own illustration of the signs characterizing the 40 Major Genera as well as the signs used to indicate Differences and Species. Genera are represent by expressions such as B , Ba, Be and so on. Differences are marked by nine consonants (B, D, G, P, T, C, Z, S, N). The Species can be expressed by adding a third vowel or diphthong to the Genus+Differences syntagm. Thus, according to the example provided by Wilkins, if De signifies Element, then Deb must signify the third difference of the element, which is Fire, and Deb will denote the first species, which is Flame. Figure.
In this system the choice of the letters is arbitrary, but the disposition of the letters aims at mirroring the composition of ideas. The two planes are so reciprocally isomorphic as to determine each other. If in Deb one replaces the second b with t (thus producing Det ), then a different concept is expressed, namely Rainbow.
Looking back at FIGURE , it is however evident that there are only nine signs or letters to indicate either Differences or Species. To record more than nine species to specify a second group of nine species an L is added after the first consonant in the pronunciation of the name, and to specify a third group a R is added. Therefore if G pe is normally Tulip (third species of the fourth difference of the genus "Grasses classified by leaf"), Gl pe is Ramsom because the addition of the L means that the final e no longer indicates the third species in the genus but the twelfth (FIGURE ).
Yet it is precisely at this point that we come across a curious accident. In the example we just gave, I had to correct Wilkins' text (p. 45). The text uses the normal English terms tulip and ramsom, but designates them in the character by the terms G de and Gl de rather than G pe and Gl pe. One only has to go back and check this in the tables to see that this must be a misprint: in the tables, we see that G de is malted barley. Regardless of whatever botanical affinities the plants might possess, the words tulip and barley are phonetically dissimilar, and thus unlikely ever to be confused with each other. In this philosophical language, however, members of the name species are easy to muddle either phonetically or graphically. Without constant double- checking against the tables, it is difficult to avoid misprints and misunderstandings. This happens because in a language of real characters every element of the expression plane is obliged to

denote a precise content. Wilkins language has only a first articulation. Nothing in it is devoid of meaning, nothing is purely differential.
This means that in a language composed of real characters any alteration of a character entails a change of sense. This makes pretty difficult the creation of neologism, especially for unknown things of which we still ignore the nature. In other words if I find an unknown flower, I can decide to give it a name (let say marigoldus Humberti) before understanding to which family it belongs. I can call a person Mary even without knowing the date and place of her birth, or the name of her parents. On the contrary, in Wilkins language, in order to name something, one should first know the whole of its properties and its precise place in a tree of genera and species.
Suppose we want to transform the character Det (Rainbow) into Den : we would obtain a character that necessarily denote the first species of the ninth difference of the genus Element. Unfortunately there is no such species recorded in Wilkins tables. We can only conclude that the character unequivocally designates an as yet to be discovered content, and that even if the content remains undiscovered, the character has at least told us the precise point where the corresponding idea should be found. But what is that `point'? If the tables were analogous to the periodic table in chemistry, then we really would know what to look for. The periodic table contains boxes which, though momentarily empty, might, one day, be filled. Yet the language of chemistry is rigorously quantitative; the table gives the number and atomic weight of each missing element. An empty space in Wilkins' classification, however, merely tells us that there is a hole at that point; it does not tell us how we should to fill it up, or why the hole appears in one space rather than another. So no one can coin a new word before having rearranged the entire system of knowledge.
Thus we understand why our imperfect natural languages have a double articulation. It is a condition for tentative creativity and a guarantee that many mistakes in speaking or writing can be detected as such - without thinking, in a sort of psychoanalytical obsession, that every lapsus, even a typo, conceals a hidden meaning.
These explorations in the history of perfect languages are not a mere archeological endeavor. The problems faced by Wilkins are reappearing today (albeit in more sophisticated forms) in the framework of many researches in Artificial Intelligence, in the theories of a computational nature of Mind (supposedly articulating a Language of Thought or Mentalese), as well in the researches on mechanical translation.
We know what a predicament translation represents for natural languages. It has been told that every language is a system in itself and that radical translation is impossible, except one is able to find a perfect language of Mind. Such a parameter for every translation was judged as essential by Walter Benjamin: since it is impossible to reproduce all the linguistic meanings of the source-language into a target-language, one is forced to place one's faith in the ideal convergence of all languages. In each language " taken as a whole, there is a self-identical thing that is meant, a thing which, nevertheless, is accessible to none of these languages taken individually, but only to that totality of all of their intentions taken as reciprocal and complementary, a totality that we call Pure Language (reine Sprache)" (Benjamin 1923). This reine Sprache would not be a real language. If we think of the mystic and Cabalistic sources which were the inspiration for Benjamin's thinking, we begin to sense the impending ghost of sacred languages, of something more akin to the secret genius of the Primeval Language than to the ideal of the a priori languages.
In many of the most notable projects for mechanical translation, there exists a notion of a parameter language, which does share many of the characteristics of the a priori languages. There must, it is argued, exist a tertium comparationis which might allow us to shift from an expression in language A to an expression in language B by deciding that both are equivalent to an expression of a metalaguage C. If such a tertium really existed, it would be a perfect language.

The only alternative would be to discover a natural language which is so "perfect" (so flexible and powerful) to serve as tertium comparationis. In 1603, the Jesuit Ludovico Bertonio (Arte de lengua Aymara) described the Aymara language (still partially spoken by Indians living between Bolivia and Peru) as endowed with an immense flexibility and capability of accommodating neologisms, particularly adapted to the expression of abstract concepts, so much so as to raise a suspicion that it was an artificial invention. Later this language was described as the language of Adam, founded upon necessary and immutable ideas", a philosophical language if ever there were, and obviously somebody discovered that it had Semitic roots.
Recent studies have established Aymara is not based on an Aristotelian two-valued logic (either True or False), but on a three-valued logic it is, therefore, capable of expressing modal subtleties which other languages can only capture through complex circumlocutions. Thus there have been proposals to use Aymara to resolve all problems of computer translation. Unfortunately, it has been demonstrated that the Aymara would greatly facilitate the translation of any other idiom into its own terms, but not the other way around. Thus, because of its perfection, Aymara can render every thought expressed in other mutually untranslatable languages, but the price to pay for it is that (once the perfect language has resolved these thoughts into its own terms), they cannot be translated back into our natural native idioms. Aymara is a Black Hole.
It is not this evening that we can discuss the possibility of a new scientific Aymara and how to overcome all the predicaments of an allegedly perfect language. Thus let me conclude with a temporary remark, quoting an Arab writer of the tenth o eleventh century, Ibn Hazm.
According to him, in the beginning there existed a single language given by God, thanks to which Adam was able to understand the quiddity of things. This tongue provided a name for every thing, and a thing for each name. But if such a prior language existed, why should have men undergone the unprofitable task of inventing other idioms? And if it did not exist, which was the source of our natural languages? The only explanation is that there was an original language which included all others. The confusion did not depend on the accidental invention of new languages, but on the fragmentation of a unique tongue that existed ab initio and in which all the other were already contained. It is for this reason that all men are still able to understand the revelation of the Koran, in whatever language it is expressed. God made the Koranic verses in Arabic in order that they might be understood by His chosen people, not because the Arabic language enjoyed any particular privilege. In whatever language men may discover the spirit, the breath, the perfume, the traces of the original polylinguism.
Let us accept that suggestion coming from afar. Our mother tongue was not a single language but rather a complex of all languages. Perhaps Adam never received such a gift in full; it was promised to him. Thus the legacy that he has left to all his sons and daughters is the task of winning for themselves the full and reconciled mastery of the Tower of Babel.
Which means, even in this country where it seems that English is the vehicular universal language, but different people at every corner of New York City speak a different tongue, to be tentatively polyglots is the only chance for mutual understanding.
Once a young American met Roman Jakobson who was starting his teaching in this country and said to him: "Professor, I rushed here to learn from you, but your classes are given in Russian, and I do not understand it." Jakobson (who was told to speak Russian in forty languages) answered: "Try!"
I thank you for having generously tried, this evening, to understand my pidgin English as it were your own perfect language.

No comments:

Post a Comment