Logos Multilingual Portal

5 - Games theory

"My son, you're playing, and your father is dead! You're playing, you're playing, and your father is dead! Your father is dead! Your father is dead! You're playing, your father is dead!"1.

In mathematics there is a theory, the so called "games theory", that is integrated into that discipline because it provides for the formalization of logical passages occurring during play. According to the type of information that games have (perfect/imperfect, complete, incomplete), they are classified from a logical point of view.

Until 1966, nobody had the idea to apply this branch of mathematics and logics to the study of translation. And, traditionally, translation, since it was considered a subset of linguistics, was considered among the humanistic disciplines, that unfortunately still tend to be considered as separate from the so-called "scientific" ones. If you think about it, between a "humanistic" discipline like translation studies "scientific" discipline like medicine it is not certain that we ever have the possibility to easily determine that one is truly "more scientific" than the other one.

On October 11th, 1966, on occasion of Roman Jakobson's seventieth birthday, many colleagues all over the world compiled a collection of articles to honor him: Jirí Levý's one is an article entitled Translation as a decision process.

Translation is considered a set of game moves, i.e. consecutive situations that force the translator to choose among a given, usually easily definable, number of alternatives. Con-sidering, for instance, the choice of a transla-tant for a single word of the prototext, pre-sumably the translator reviews the possible translatants before deciding what solution is the best.

Levý has the example of a translation from German into English, of the title of Bertold Brecht's drama Der gute Mensch von Sezuan. The translator considers two possibilities:

Der gute Mensch von Sezuan


/ \


The Good Man of Sechuan


The Good Woman of Sechuan

In order to illustrate gradually the whole set of decision processes implied in this translation process, Levý subdivides it into the following components:

in English there is no word precisely matching the German word Mensch; «person» is a partial match, but belongs to another style register; two English words have a semantic field that, in the whole, covers nearly the semantic field of Mensch: they are man and woman;

instruction 1::
the point is to define the class of the possible alternatives: in the case of the example, finding an English word denoting the class of beings that scientifically are defined as homo sapiens;

this component immediately reminds us of Jakobson and the argument on right and left brain hemisphere, syntagmatic and paradigmatic axis; in this case, the paradigm is the whole set of words that, one in alternative to the other, can or better must be used as translatants for Mensch. Levý locates two possible translatants: man, woman;

instruction 2: :
this phase serves to guide the choice from among the alternatives; the choice, always depending on the context and co-text, in this case is based on the whole text of Brecht's drama; the two alternative possibilities are not equivalent between themselves, nor equivalent to the word to be translated; that does not mean, however, that the choice is or can be random; it is always dictated by the context.

This last operation is the one that sends the expert systems that have been created to give the illusion of being able to use machine translation into tilt: machines are not able to consider the context, both because one should create "contextual" data bases within which computers could choose the best solution, and because the logical synthesis of a word and its context is difficult to be defined in formal terms. Levý says:

the interpreter has to choose from a class of possible meanings of the word or motif, from different conceptions of a character, of style, or of the author's philosophical views. The choice is more limited ('easier'), if the number of possible alternatives is smaller, or if it is restricted by context (Levý 1967: 1172).

What Levý emphasizes here is how much a decision is interlinked to all those that follow it, and conclusively prevents a series of potential ensuing decisions. Once the translator, in the example, chooses "man" or "woman", that predetermines some grammatical consequences (the gender-sensible forms, for example) and interpretive questions linked to the poetic view of the whole work.

For these reasons, from the point of view of the mathematical theory of games, translation is a game with complete information. "Every succeeding move is influenced by the knowledge of previous decisions and by the situation which resulted from them". To give a more concrete example, translation is comparable to a chess game, not a card game. In the following diagram, that I transfer directly from the article,

it is possible to see in a graphic form what happens through decision processes in translation. The point where the solid line and the dotted line split corresponds to the moment where a decision or translation choice is made. Solid lines represent choices still possible after the first one, while dotted lines represent ensuing choices that, after the choice that has been made, are no longer possible.

The graphic representation always provides for a choice between two possibilities, but actually in translation practice choices are not always of a binary type: they very often are among a very high number of alternatives. A graphic representation of the many-alternatives procedure would be very complex.

One of the possibilities would consist - in theory, at least: in practice I think it very com-plicated to conceive - in reckoning with all the decisions deriving from a given choice, and then, based on the poetics of the text to be translated, try to set a hierarchical order of the elements by importance, i.e. a sort of hierarchy of dominant plus subdominants. Each choice, and each decision deriving from it, give birth to a different "game", which in translation is called "version".

Levý sees the translation process as a succession of definitional instructions and selective instructions. A definitional instruction produces a paradigm of the possible choices, the selective one implies the choice within the given paradigm.

definitional instruction

/ \

selective instruction


selective instruction

definitional instruction

definitional instruction

selective instruction

selective instruction

selective instruction

selective instruction

The given example in this case is the English translation of the German word Bursche. The instruction system is as follows:

young man

/ \

young man standard


young man substandard

/ \

/ \

young man bookish

young man literary

young man vulgar

young man colloquial

Il corrispondente sistema dei paradigmi è invece il seguente:

boy, fellow, lad, youngster, chap, guy, larh

/ \

boy, fellow, youngster, lad


chap, guy, lark

/ \

/ \

youngster, lad

boy, fellow


chap, guy

In the next unit we will see our way through the workings of such a system on the translation choice.


Bibliographical references

CANETTI ELIAS Die gerettete Zunge. - Die Fackel im Ohr. - Das Augenspiel, München, Carl Hanser Verlag, 1995, ISBN 3-446-18062-1.

CANETTI ELIAS The Tongue Set Free. Remembrance of a European Childhood, translated by Joachim Neugroschel, in The Memoirs of Elias Canetti, New York, Farrar, Straus and Giroux, 1999, ISBN 0-374-19950-7, p. 1-286.

LEVÝ JIRÍ, Translation as a decision proc-ess, in To Honor Roman Jakobson. Es-says on the occasion of his seventieth birthday, Den Haag - Paris, Mouton, 1967, vol. 2, p. 1171-1182.

1 Canetti 1999: 62.