As we have seen, approaches to style in translation look at the style of the source text as the translator reads it and also at the sorts of constraints which might affect the translator's stylistic choices in constructing the target text. Some of these were considered in the previous section. But it is also possible to examine the style of the target text itself, and ask whether and how it can help us reconstruct why the translator has made certain choices (Malmkjer 2004:14). This point was touched on in 2.3, and I want to take it up again here. Hermans maintains that there is always a "translator's voice" in every translation; we should not be taken in by the translational ideology which puts across "the illusion of the one voice" (1996:27; see also Schiavi 1996:4) in the translated text. Some examinations of target-text style focus to quite a large extent on the target text in its own right, looking for elements of the translator's style, which Baker describes as "a kind of thumb-print that is expressed in a range of linguistic - as well as non-linguistic - features (2000:244; see also Lech & Short 1981:167). But to some degree all studies of the style of translated texts willrelate this visible presence of the translator to the style of the original text. This may be done explicitly,by looking at obvious interventions by the translator,or less explicitly, by considering choices such as "consistent use of specific strategies", including things like "use of prefaces or afterwords, footnotes, glossing in the body of the text, etc" (Baker 2000:245). Such studies of translated texts, which aim to recreate the choices made by the translator, come under what Malmkjœr refers to as "translational stylistics (2004:15).

Views which focus most clearly on the style of the target text in its own right are often found within theories of translation reception such as have been formulated especially in the framework of polysystem theory (e.g. Even-Zohar 1990:Toury 1980: 1995). Toury (1995:192ff.),for example, discusses the norms governing Avraham Shlonsky's Hebrew translation of Hamlet.

Toury says that, in the various revisions Shlonsky made, the source text played only "a marginal role" (1995:196). The sorts of constraint the translator observed include prosodic (he wrote in iambic hexameter). pragmatic (he complied with "contemporary convention" in his use of "poetic words), and theatrical (he wrote a "script suitable for declamation"). The "aspiration for adequacy, that is, in Toury's terms, the reconstruction of source-text features, was much less important in determining the translator's decisions than considerations to do with acceptability of the target text (1995:149; 203-4).

Views which focus more clearly on the target text's relation to the source text include Vinay & Darbelnet's 1958 book (here 1995) of comparative stylistics of French and English. This tends to concentrate on individual words and expressions, such as the fact that "sauter" in French has an intellectual" equivalent in the word "jump" but also a stylistically subjective equivalent "leap". Vinay & Darbelnet speak of "oblique" translation (not to be confused with Gutt's indirect), which, as opposed to literal, cannot recreate stylistic effects without "upsetting" the syntax or lexis of the target language (1995:31). Vinay & Darbelnet stress the importance of adaptation in such cases. A study like this assumes that it is generally clear what translation is aiming for, namely the preservation of stylistic nuances of the original but always under the assumption that it is clear what is the equivalent of a particular structure in a given context. It can thus consider style and stylistic effects as though they were purely linguistic phenomena in a narrow sense, rather than clues to meaning, and did not in any important way reflect either the ideological choices of the original author or the choices made by the translator.

It is quite possible to envisage studies of translated texts based on stylistics or Critical Discourse Analysis which concentrate on what their style can tell us about the translator's world view, or attitude towards the information conveyed. Baker (2000:246) rightly points out that such studies of translated texts are rare, but she is not correct in asserting that subtle stylistic or linguistic habits which are "largely beyond the conscious control of the writer" (2000:246) have not in general been a feature of literary stylistics per se, for this is exactly the area of concern of those stylistic studies influenced by Critical Discourse Analysis. In principle there is no reason why such analyses should not be performed on translated texts.

Overall, stylistic analyses of translated texts have in general not been numerous as yet. Exceptions are Boase-Beier (2003a) and the articles by Malmkjer (2004), Millán-Varela (2004) and Thomson (2004), all collected in a Special Isue of the journal Language and Literature. Malmkjar (2004:15) points out that the stylistic study of a translated text must take into account the "willing suspension of freedom to invent" on the part of the translator. But it could be argued that exactly in the area of style the translator does have a great deal of freedom to invent. Lowell, in his 1958 anthology Imitations, says that he has striven for "one voice running through many personalities" (1990:xi), a clear indication of stylistic invention. This wish was certainly in part dictated by the need to produce a coherent anthology, but it could equally be argued that translating an anthology is a way of procuring stylistic freedom.

Even in the case of apparently free translations, though, the style of the translation is defined by its relation to the source text, as studies by Malmkjar (2004) and Winters (2004) show. Winters discusses two different translations of Fitzgerald's The Beautiful and Damned and identifies "the distinctive use of modal particles as a textual manifestation of the individual style of each translator" (2004:71). This study both compares target texts and compares each with the source text, discovering that one translator keeps closer to the source text than the other, and that the difference in style between the two translators was especially noticeable in the narrative, as opposed to the translation of direct speech. This suggests that the translators tended to use direct translation for direct speech, and indirect for narrative, leaving their "thumb-print"(Baker 2000:244) more clearly on the latter. Malmkjer (2004) compares a translation of Hans Christian Andersen by Dulcken with its source text, and shows that stylistic clues in the target text point to a split between the heavenly and the earthly which was not present in the original.She speculates as to possible reasons for this, to do with Dulcken's own views, or what he considered to be the religious beliefs of the receiving culture (Malmkjar 2004:22). In Boase-Beier (2003a) two translations of the same poem are considered and stylistic differences seen as evidence for different interpretations on the part of the two translators of the cognitive state embodied in the text. The point is made in that article that a translated text is often felt by translators to be co-authored; the explanation given in the article is of a knowledge basis and world view influenced by that of the original. Though Malmkjœr (2004) does not say this exactly, by arguing for a "translational stylistics" she is arguing for an analysis which takes some degree of amalgamation of source-text and target-text stylistic characteristics into account.