Some very important insights for the translator have come from the science of cybernetics, which in its most specific application to translation is generally known as "Information Theory." In terms of this theory it is possible to study the rate of flow of messages in a channel, for the number of messages which can be communicated is mathematically related to the number of signaling units. For example, if there is one such unit, which may be either "on" or "off," as in the case of any electronically controlled device, there аre obviously two "messages" which can be communicated, but if there are two such signaling units, the number of messages, i.e. based on the number of combinations of on-off signals, is four. Similarly, with three such units, the number of messages is eight; with four units, sixteen; with five units, thirty-two, etc.. following an exponential curve to the base 2. In the application of this theory to translation we cannot give precise mathematical formulations to our problems, but we can obtain from some of the basic principles of information theory certain very important insights:

1. The "information" which is communicated by any message is a measure of the unpredictability of the signals which are given. In fact, information in this technical sense is largely equated with unpredictability. This implies, of course, that the more unpredictable the message the greater the channel which is required for the transmission and decoding of the message. In ordinary language this means that one can readily understand a message which is commonplace, but it takes much greater decoding effort to comprehend a message which is unusual, unpredictable, and strange.

2. In order to combine efficiency of communication with an effective guarantee against distortion by noise or other factors, languages tend to be about 50 per cent redundant. That is to say, they seem to reflect a kind of equilibrium between the unexpected and the predictable. Languages differ somewhat in the amount of redundancy they possess, but in formal as well as in semantic structure there seems to be a redundancy of about 50 per cent.

Any message can be communicated through any channel, but it may be necessary to "lengthen" the message, that is, to take more time to communicate it if the channel is too restricted. If the receptive channel is very wide, the message can always be compressed, but if the channel is narrow, the message needs to be drawn out.

"Information" in cybernetics is not to be equated with meaning, but only with unpredictability. That is to say, in information theory we only determine the difficulties of communicating some message, not the relevance or value of the message to the receptors.

The relationship of information (in this technical sense) to unpredictability can be readily understood from a number of very practical situations. For example, if a person always declares that everything is "terrific," whether it is a meal, speech, dress, bouquet of flowers, ball game, movie, or a joke, the word terrific does not carry very much information, for it is largely predictable, Likewise, when an air raid siren goes off each Saturday at noon, it does not carry much information, for the noise is almost entirely predictable.

By saying that languages incorporate a great deal of redundancy into their structures, we are simply saying that languages reflect a good deal of predictability. For example, in the phrase these men are ... we know from the sole occurrence of the plural form these that men and are will be employed rather than the singular forms man and is. Accordingly, we can say that the plural indicators in the forms men and are are redundant. Redundancy, however, is not repetition, for in a phrase such as truly, truly, I say ... the adverb truly is repeated, but the second occurrence is not redundant, for it adds to the total meaning. One can also say truly I say or truly, truly, I say, and the second expression differs from the first in intensity. Nor is redundancy tautology, which in general usage (but not in logic) implies meaningless repetition, e.g. he descended down, in which case down is tautological to descended; but it is not so in the phrase he climbed down, for one can climb up as well as down.

Obviously, we must not confuse unpredictability (or information) with value, meaning, or relevance. One could say, for example, "This is a beautiful pshlong," in which pshlong would be highly unpredictable (and hence carry a great deal of information), but it has no meaning. On the other hand, a message with high unpredictability may have a great deal of meaning, but this depends upon the content of the message and not upon its formal unpredictability.

Before, however, we consider some of the other immediate relationships of information theory to problems of translation we must give further consideration to the redundancy which occurs in language, for only on the basis of a knowledge of such data can we evaluate the problems of impact and the capacity of the decoder to "take in" the message.

责任编辑：admin