To continue my game, I asked Altavista to retranslate the Spanish text into English, and was given:
In the God that began created the sky and the Earth and the Earth was without form, and emptiness; and the dark was on the face of the deep one. And the alcohol of the God moved on the face of waters. There and the God said, let be light: and there was light. And the God saw the light, that that was good: and the God divided the light of the dark. And the God called the slight day, and the dark that it called to Night. And afternoon and the morning was the first day. And the God said, there it let be firmament in means of waters, and it let divide it waters of waters. And the God did firmament, and divided the waters that were under firmament of the waters that were on firmament. And it was so.
This text is more or less a victim of the misunderstandings of the Spanish version, plus some new mistakes (for instance it translates the Spanish medio, middle, as means). Notice that, faced with the Spanish expression dividir las aguas de las aguas, AltaVista, if provided with a reasonable dictionary, should have detected that the Spanish preposition de can be translated either as of or from. AltaVista nonchalantly chose of because it lacks the fundamental information that, in the world we live in, there is nothing like waters of waters.
This is not a mere piece of encyclopaedia information, like the facts that Shakespeare was an author and that God is a spiritual being. It concerns our understanding of the world. Let me avoid a painful semiotic discussion over whether an understanding of the world should or should not be included within an encyclopaedic competence. We shall see later that there are certain pieces of information about the world which are not registered by an encyclopaedia but rather provided on the spot by a given text (like there is a cat on the mat). However, we have succeeded in adding a new stricture to our criteria for good translating competence: translation does not only concern words and language in general but also the world, or at least the possible world described by a given text.
At this point I decided that, by continuing to play, I could probably get some other interesting suggestions, so I asked AltaVista to translate the English text into German, which produced this:
Im Gott, der anfing, stellte den Himmel und die Masse und die Masse war ohne Formular und emptiness her; und die Dunkelheit war auf dem Gesicht vom tiefen. Und der Spiritus des Gottes verschob sich auf dem Gesicht des Wassers. Dort und der Gott sagte, lassend Sie, Licht zu sein: und es gab Licht. Und der Gott sah das Licht, das, das gut war: und der Gott teilte das Licht der Dunkelheit. Und der Gott benannte den geringfügigen Tag und die Dunkelheit, die sie zur Nacht benannte. Und Nachmittag und der Morgen waren der erste Tag. Und der Gott sagte, dort ließ er firmament in den Mitteln des Wassers sein, und er ließ es teilen wässert vom Wasser. Und der Gott tat firmament und teilte das Wasser, das unter firmament des Wassers waren, das auf firmament waren. Und es war so.
The German text accepts the story of a beginning God, of the light day, of a world without void, and that that obviously becomes das das. But it mistranslates Earth as Mass, form as formular, and the divine alcohol becomes Spiritus, which is as ambiguous as spirit. There are other syntactical oddities. In order to translate created AltaVista finds the German verb herstelle, which must be conjugated as Ich stelle her or Ich stellte her. Because of a syntactic rule that was provided, AltaVista ‘knows’ that her should be put at the end of the sentence, but it does not realise that the sentence already ended with Himmel, and so put her much too far ahead. Moreover, apropos waters of waters, the first term is interpreted as a verb and the second as a noun. Such a mishmash could not but give the following result when I asked the computer to retranslate German into English:
In the God, which began, placed the sky and the mass and the mass was without form and emptiness ago; and the darkness was on the face of the deep. And the white spirits of the God shifted on the face of the water. There and the God said, leaving you, to be light: and there was light. And the God saw the light, which, which was good: and the God divided the light of the darkness. And the God designated the slight day and the darkness, which designated it to the night. And afternoon and the morning were the first day and the God said, there let it in the means of the water be firmament, and it left it divides waessert from the water. And the God did firmament and divided the water, which firmament under the water was, which were on firmament. And it was like that.
It is interesting to remark that, faced with stellte . . . her, AltaVista does not (and with good reason) recognise a compound verb; it finds in its dictionary that her alone can also mean ago; and concocts placed . . . ago. (By way of compensation, many other verbs are put at the end of the sentence, as in under the water was.) Spiritus again becomes something alcoholic, and the program is unable to translate waessert.
The conclusion of my experiment is that in order to translate, one must know a lot of things, most of them independent of mere grammatical competence.
But at this point we are encouraged towards another reflection. If one received the different versions of Genesis provided by Babelfish, one would guess that they were bad translations of the King James text – and not, let us say, bad versions of the first adventure of Harry Potter. And if someone who had never heard of the Bible read these versions, I think that even such a naive reader would in some way realise that these texts deal with a God who has created a world (even if it would be very difficult to understand what the hell He actually made).
When I started working for a publishing house I was given the manuscript of a translation from English for a first check, without being able to refer to the English original, which was still in the hands of the translator. The book told the story of the first researches on the atomic bomb in America, and at a certain point it said that, gathered in a certain place, a group of young scientists had started their work by performing corse di treni (which in English means train races). I thought it was pretty peculiar that persons who were supposed to discover the secrets of the atom wasted their precious time in such childish play. I resorted to my world knowledge and inferred that these scientists were certainly doing something else.
I do not remember whether at this point I remembered an English expression that I already knew, or tried to retranslate the Italian expression into English as if I were a bad translator. In any case I immediately realised that these scientists were on training courses, which was more reasonable and less expensive for American taxpayers. As soon as I received the English original I saw that I was right, and I did my best to get the translator fired immediately.
Another time, in the translation of a psychology book, I found that, in the course of an experiment, l’ape riuscì a prendere la banana posta fuori dall sua gabbia aiutandosi con un bastone, that is: a bee succeeded in grasping a banana lying outside its cage with the help of a stick.
My first reaction drew on world knowledge: bees are unable to grasp bananas with a stick. My second reaction used linguistic knowledge. If one cross-references the Italian and English a ‘false friend’ is revealed:
ITALIAN ENGLISH
Ape Bee
Scimmione Ape
It was clear that the original English text spoke of an ape and that the translator believed that ape meant bee in English too. Furthermore, my encyclopaedic knowledge was telling me that apes do grasp and eat bananas.
All this means that, even if a translation is wrong, not only is it possible to recognise the text that it translates badly, but a reasonable interpreter can usually infer, from a wrong translation of an unknown original, what that original was probably saying.
Every text (even the most simple sentence) describes or presupposes a possible world. In the two cases above I made some inferences about the world described by the text, by comparing it with the world we are living in, and trying to figure out how an atomic scientist and a bee should behave. After having made a hypothesis about the probable structure of the world pictured by the original text, a short exploration into the English lexicon helped me to find a reasonable final hypothesis: the scientists were on training courses and the banana was grasped by an ape.
Another example: in Italian we have only one word