EN:
Translation is usually viewed as a process designed to overcome a deficiency: X translates the words of A for C, because C doesn't have an adequate command of the language in which A expressed himself. Translation so practised is usually, if not always, an entropie process. As I think I have shown in a recent article, there is a strong tendency of the text to run downhill in translation, with a demonstrable loss of ordering and coherency, an inexorable regression of form to formants, of the marked to unmarked1.
There exist, however, cases where translation ceases to be a mere expedient and, far from being entropie, conserves or even enhances the ordering of the texts it brings into play. It is in such cases that translation may be said to function heuristically, by foregrounding structures taken for granted in the source text, by making the information encoded in the text more readily accessible to the target-language group than it was to the source-language group, by enhancing the repertory of esthetic forms available to the target-language group, or by stimulating the creation of new forms.
A number of more or less canonical examples come to mind immediately. Ethno-linguistic translation zeroes in on what I have referred to elsewhere as the grain of the text (i.e. the micro-structures that derive from the linguistic substratum2), in order to provide insights into the functioning of languages very different from that of the target group. In hermeneutic translation, the usually subterranean work of interpretation surfaces quite explicitly in the target text, which thus functions as a gloss, making the message more readily available to the target-language reader than it was to start with in the source text3.
But the heuristics of translation can go far beyond the narrow scope of such undertakings.
