As for images in translation, Miller has, based on her own practice, asserted the importance of image-visualization in literary translation (1986). And Andre Lefevere (1992) has also claimed that translators as rewriters "created images of a writer, a work". And when translators as readers "say that they have ‘read' a book,what they mean is that they have a certain image, a certain construct of that book in their heads." Translation, as Lefevere put it, is to projcct the image of an author and a (series of) work(s) in another culture.

As Bell puts it, the process of transformation of a SLT into a TLT is to synthesize a mental semantic representation into a second language-specific text. We agree with his point but propose a mental artistic image accompanied with to achieve a successful literary translation. Meanwhile we object to the old form-based notions of equivalence, which hindered rather than supported the development of translation theory, for usually, translators who comply with such notion hesitate to diverge from the most obvious features of the text, i.e. the forms, in fear of being disrespectful, esp. when applying to poetic texts for which forms always enjoy a prestigious status.

For some translators and theorists, form-based procedures have a certain appeal, for they provide an orderly, analytic work sequence. In such a view, translating consists of exchanging labels, consulting if necessary a dictionary or an informant. According to Barn-stone (1993), some traditional views in translation hold that the faithful translator performs a word-for-word exercise with the assumption that meaning resides fully in the semantic element of a word's phonic-semantic code. Form-based translating is especially alluring when languages share similar grammatical categories in many areas. We find Catford (1965) speaking of "grammatical translation" and illustrating this by rendering "This is the man I saw”into French as "voici le man que j'ai vue." Small units are replaced by equally small units, and decision can be made in a mechanical fashion. The resulting text is usually not a very faithful reproduction of the original text, nor an acceptable sample of the goal language, for literary texts have aesthetic properties, beyond linguistic meaning which can be translated if we are able to relate them to something manifested by texts as a whole. If focus is placed on word meaning and sentence parsing, surface representation will be conducted, while the image-actualization of a certain gestalt image might bring into light underlying meaning and gestalt qualities and thus reach to dynamic correspondence in translation. Image-building in translation is similar to inference drawing and gist construction, but more than that, for image is built not only on the inherent meaning and general idea, but also on the aesthetic consistency. Doubtlessly, in reading any texts, you may draw inferences and get the general idea. In translation, however, this is not enough, especially in the translating of literary discourse whose overall effect of verbatim information is also important. In reading, attention might be attached to different aspects by different readers. For a good literary translator, the crucial point is to combine both the general idea and overall effect of the verbatim information, so as to represent in the T-text a correspondent text. It is postulated in this study that when the original text is actualized in the translated text in a gestalt image, higher aesthetic equivalence will be acquired. With close attention to the artistic structuration of the original, the translation would create optimal conditions for the transferal of aesthetic properties.