domingo, febrero 26, 2006

The Universality of the Semantic Field

Ludwig von Mises said in his classic “Human Action” that as soon as Marx understood the robust rationality of the arguments of the classical economists, he began a frontal attack against the logical reasoning itself. The classical Marxists tried to prove that social science and science in general were a cultural creation designed to keep the class structure, and a new social class as the proletariat was, needed a new logic. Since then, “polylogism”, that is, the theory that different human groups have different kind of logic has been a growing cultural industry.

The most powerful form of polilogism still alive is the linguistic polilogism. The most developed form of linguistic polilogism is the Sapir-Wharf hypothesis, which states that every language represents a different vision of the world, and that strong syntactic differences make mutual understanding impossible. In fact the most perfect description of the Sapir-Wharf hypothesis could be “every translation is a form misunderstanding”.

Surprisingly, the most powerful enemy of this form of post-modern nihilism was Noam Chomsky. The Chomskyan Universal Grammar was supposed to be a normalized set of underlying syntactic rules that all the languages in the world had in common. The Chomskians considered that beyond the exuberant variety of syntactic structures, there were a few deep rules, which were the real grammar beyond grammar. The equivalence relations of these rules were the Rosetta stone of linguistics. Moreover, for chomskian linguists, the deep syntactic rules were biologically determined.

The empirical support for biological determinism of language has grown in the last years, and there is also an increasing evidence of the natural ability to create syntactic structures. For example, an study of the hybrid languages of the first generation of Antillean slaves proved that slave children created those hybrid languages on the spot.

But the existence of the universal grammar is nowadays not accepted by the majority of linguists. On the other hand the experience of every bilingual speaker (even if you become bilingual in your adulthood, as it is my case) is that translation is possible. So, different languages are not different worldviews.

If not syntactic rules, what do all languages have in common? For me, as a mathematician, long time (ago) involved with logic and model theory, the answer is clear: the semantic field is what all human languages have in common. The translation is something that doesn’t happen between syntactic structures, but between semantic objects. Precisely the difficulties to build high-quality automatic translators come from the fact that translation is not a syntactic but a semantic operation. And I fear our programs are still not manipulating semantic objects: that is, programs don’t understand.

When I say “the semantic field is universal”, I don’t mean that all human beings have the same semantic field in their brains: of course the size and form of the semantic field depends on the cultural background. What I mean is that given two individuals, native speakers of different languages, both of their semantic fields could be extended to contain the semantic objects of the other’s field. That is, the semantic field is not language-dependent.

I could try to prove this strong anti-Shapir-Wharf hypothesis; but the evidence is so overwhelming that I cannot choose where to begin: bilinguism, perfect translations, and literarian influence between languages... (for example, the style of Borges in Spanish is a beautiful mix of Gibbon and Lovecraft.)

I think that as soon as the semantic field universality hypothesis is clearly stated, it becomes obvious. Only the interests of linguistic nationalists and the nihilist preferences of some academics, keep the (racist) Sapir-Wharf hypothesis alive.

The universality of the semantic field has a beautiful literary consequence, I cannot resist discussing: high quality literature should not be written with phonetic regularities or rhythms, but with mental images. Valuable literature is always universal; this implies a sort of Platonism.

Take the beginning of Shakespeare’s Julius Caesar “Lend me ears, Roman citizens”. The beauty of the statement comes not from the sound, but from the intellectually stimulating metaphor. The greatest example of perfectly translatable literature is the Bible, which keeps its beauty in all languages; this is not a accident: the biblical text is not written with sounds [1], but with a rhythmic flow of ideas and a powerful collection of literary archetypes. It is poetry, but not rhythmic poetry.

[1]Unlike the Koran, which is devastated by the translation.


At 9:53 p. m., Blogger Thalasos said...

A ver si me animo con el ejemplar de El Instinto del Lenguaje de Steven Pinker, que adquirí hace poco.
Las teorías del lenguaje, con ser dispares, están limitadas por un hecho previo, la capacidad de hablar y als zonas del cerebbro dedicadas a ello. Y si es verdad que hay enormes diferencias entre algunas lenguas, pero no entre las próximas de origen, porque se influencian unas a otras.
Alguien trabajó sobre las diversas manifestaciones del tiempo en el lenguaje Y provocó ma´s de un disgusto y cambio de hipótesis. Pero hoy día el lenguaje se ah globalizado,a sí que las diferencias desaparecieron. Me gustó su post. Otros no tanto. Ideología difícil de compartir. Pero bueno, compartimos lenguaje, pero no lo que subyace a él, ¿verdad?
Saludos, kantor

At 9:56 p. m., Blogger AMDG said...

Fabuloso Kantor. Déjame apuntar un par de cosas:

- El lenguaje criollo es realmente creado a partir de un lenguaje incompleto e imperfecto como el pidgin (lenguaje de indios). Los niños completan la gramática imperfecta ex nihilo. Bueno, sacándola de las estructuras cerebrales comunes a la especie.

- Pones a Borges como ejemplo. Creo que se dió cuenta de que hablaba dos idiomas distintos cuando tenía x años. Utilizaba uno con cada abuela de forma natural.

- Sobre la Biblia:


At 10:36 a. m., Blogger EVF said...

Hi Kantor,
I agree with most of it. However, I would like to state a few comments:
- You are always choosing examples of indo-european languages to support your hypothesis of "perfect translatability". It is no surprise that cousin languages are possible to translate without loss. But when you arrive to a semitic language, you are forced to admit that the Qu'ran is devastated by the translation. I reckon Confucius is devastated as well.
- "Racism" can be understood as the statement that different races exist, or that some races are better than others. I think you agree with the first one, as it is the basis for the Bayesian Racism theory you stated (and with which I agree). The Sappir-Wharf hypothesis can be racist according to the first definition, but I don't see how would it imply the second one.
- I agree that language is not a racial trait, and that it does not provide a "unique view" of the world. However, the attitude toward language reflects other important features of the culture that uses this language. It is at least funny that the French (as a country, not each one of them) have a rigid, presriptive view on language, they have had many Constitutions, and they are statalist, while the British have a flexible, descriptive approach of language, no Constitution and they are economic liberals.
- My experience is that true bilingualism simply does not exist. I may speak perfect French and Spanish, and more-than-correct English, but in my head "my" language is distinctly Catalan. I may even think in French after a long time in France, but it never really gets in terms of equality with "my" language. All of which, of course, does not affect the translatability of indo-european languages in terms of semantic field. It's just that bilingualism does not exist.

At 8:35 p. m., Blogger Marzo said...

EVF: I was not aware that the Book of Genesis, or the Psalms, or the Song "of Solomon", were originally written in an Indoeuropean language; I thought Hebrew was a Semitic one. But, what do I know...

On the other hand, Borges himself was not at all sure everything is always translatable. "How do you translate into English 'ella estaba solita'?", he asked. (I think it could perhaps be translated into French as 'elle était toute seule').

At 10:56 p. m., Blogger Kantor said...

"But when you arrive to a semitic language, you are forced to admit that the Qu'ran is devastated by the translation"

Yes, of course, but it is devastated because it is written with phonetic regularities instead of ideas. Thats the whole point: if you use the phonetic structure of the language, you become untranslatable. The Bible is poetry writen with ideas, and it is robust to translation. Robustness to translation is in my opnion a very important criterium of literary quality. Literature should happen in the semantic field.

"My experience is that true bilingualism simply does not exist.(...)
All of which, of course, does not affect the translatability of indo-european languages in terms of semantic field"

Of course you feel more comfortable in some language, but the semantic field is the same for every language; indoeuropean or not. I used the Bible and the Koran as examples, because thay are both written in semitic languages, and the Bible is among the most translatable books, while the Koran is not.

At 5:29 p. m., Blogger David Boxenhorn said...

Great post!

The greatest example of perfectly translatable literature is the Bible, which keeps its beauty in all languages

Ha! That's what you think.

At 11:45 p. m., Anonymous Anónimo said...

"Ha! That's what you think."

Don´t tell me I should learn Hebrew!

English has been painful enough...
and it is only another indoeuropean language.



Publicar un comentario

<< Home