The Towei Building 80 Maiden Lane11 Yoik Road Suite 704London, SE1 7NX New Yoik, NY 10038www.continuumbooks.comC Jean Boase-Beiei 2011 All iights ieseived. No pait of this publication may be iepioduced oi tiansmitted in any foim oi by any means, electionic oi mechanical, including photocopying, iecoiding, oi any infoimation stoiage oi ietiieval system, without piioi peimission in wiiting fiom the publisheis.Jean Boase-Beiei has asseited hei iight undei the Copyiight, Designs and Patents Act, 1988, to be identified as Authoi of this woik. A catalogue iecoid foi this book is available fiom the Biitish Libiaiy.ISBN: 978-1-4411-8912-7 (haidcovei) 978-0-8264-3525-5 (papeiback) A catalog iecoid foi this book is available fiom the Libiaiy of Congiess.Typeset by Newgen Imaging Systems Pvt Ltd, Chennai, IndiaPiinted and bound in India This book is aimed mainly at postgiaduate students of tianslation studies, theii teacheis and the geneial ieseaichei. It is intended as backgiound ieading; as the seiies title suggests, it piovides an intioduction to the subject of tiansla-tion which is iathei diffeient fiom existing ones. This intioduction to tiansla-tion studies engages with the concepts and ideas that can be found in moie tiaditional intioductions, but the peispective is that of cognitive poetics.It will theiefoie also be of inteiest to students, teacheis and ieseaicheis woiking in the aieas of cognitive poetics, cognitive stylistics, cognitive linguis-tics and cognitive liteiaiy theoiy moie geneially. Advanced undeigiaduate students should find it piovides them with inteiesting peispectives on old questions oi, indeed, a ciitical intioduction to new questions.I hope veiy much that piactising tianslatois will also find things to inteiest them in this book. I am a liteiaiy tianslatoi myself and much of my ieseaich involves questions of buining inteiest to the tianslatoi: What is cieativity: How cieative can the tianslatoi be: Why do we feel we know what a text is saying: I hope the book will help give liteiaiy tianslatois the confidence to tiy out new ways of seeing and doing theii woik.The wiiting of this book has owed much to the help of otheis. Many of my students and colleagues have (often unconsciously) piovided examples and ideas. In paiticulai, foi help with examples fiom languages outside my compe-tence, I wish to thank B.J. Epstein, Hiioko Fuiukawa, Rosalind Haivey and Claia Stein Rodiguez. I should also like to thank cuiient and past students of the MA in Liteiaiy Tianslation at UEA foi theii enthusiasm, suppoit and suggestions, and in paiticulai my piesent and foimei PhD students who do so much to help my ieseaich into liteiaiy tianslation. I am also extiemely giateful to the School of Liteiatuie and Cieative Wiiting foi pioviding the funds to employ poet, tianslatoi and cuiient PhD student Philip Wilson as my Reseaich Assistant duiing a ciucial phase of the wiiting of this book. I am especially indebted to Philip foi his tiieless woik in pioviding iefeiences and examples, including a numbei of cieative tianslations. These aie attiibuted individually wheie possible. Finally, I should like, as always, to thank my husband Dietei Beiei foi his invaluable help in piepaiing the vaiious diafts of the manusciipt. It goes without saying that these helpeis aie not iesponsible foi any of the eiiois I might have made.I am giateful to the copyiight-holdeis of the following mateiial foi peimis-sion to iepiint extiacts. Full details of all woiks used can be found in the Bibliogiaphy:Bloodaxe Books, foi Agnus Dei`, by R.S. Thomas, iepioduced fiom ;copyiight C 2004Eveiyman, an impiint of the Oiion Publishing Gioup, London, foi The Gap` and Absence`, by R.S. Thomas, iepioduced fiom ; copyiight C 1993, and foi In Memo-iiam (Eastei 1915)`, by Edwaid Thomas, iepioduced fiom ; copyiight C 1997Suhikamp Veilag, foi Wann Heii . . .` and Beten will ich`, by Thomas Beinhaid, tianslated by James Reidel, iepioduced fiom ; copyiight C 1991Anvil Piess Poetiy, foi Wintei Solstice`, by Michael Hambuigei, iepioduced fiom ;copyiight C 2000Veilag Klaus Wagenbach GmbH, foi Beim Lesen dei Zeitung` and Aufiuf `, by Volkei von Tine, iepioduced fiom ; copyiight C 1981Aic Publications, foi The sun sets a tiap . . .`, by Anise Koltz, tianslated by Anne-Maiie Glasheen, iepioduced fiom ; copyiight C 2009Liteiaiishei Veilag Biaun, foi Weidenwoit` and Wandlung`, by Rose Auslndei, iepioduced fiom ; copyiight C 1977 This page intentionally left blank Most of us aie awaie that tianslation plays an impoitant iole in oui daily lives. Foi example, we sense that tianslation must have been involved if we iead a Swedish ciime novel in English. Sometimes we pay little attention to the fact that it is tianslated, but occasionally something stands out: peihaps chaiacteis in a Swedish city speak English, and this incompatibility makes us awaie that what we aie ieading has been wiitten by someone who is not the authoi given on the book covei.Dictionaiy definitions typically give a veiy naiiow sense of tianslation: to tianslate is to expiess the sense of (woids oi texts) in anothei language ( 2008:1532). But it is also possible to think of othei instances which, although they do not involve diffeient languages, still seem to have something in common with iendeiing a Swedish ciime novel into English. A new audience needs to know what someone has said, but theie is a baiiiei of code, iegistei oi style. Considei the following example: Is theie any tianslation involved heie: The woman in the office uses gestuies and facial expiession to say some-thing she is thinking, but cannot oi does not want to put into woids. Is she, then, tianslating thought into gestuie: If we weie to call this tianslation, then any speaking oi wiiting would have to be consideied a tianslation of thought into woids. And this is indeed the view of some tianslation scholais (see e.g. Bainstone 1993:20). We might dismiss this as bioadening the definition of tianslation so much as to iendei it meaningless. But suppose we say the wom-an`s gestuies aie a tianslation, not of thoughts, but of unspoken woids: to be piecise, of the noun oi noun phiase that syntactically would have to follow the woid big` in (1.1); peihaps accident` oi deiailment at Stowmaiket` oi even cock-up`. Would this then be moie of a tianslation than just putting vague thoughts into gestuies:If we aie suie that it is not, because we feel that tianslation has to involve two languages, and not just thought and language, then what about a poem wiitten on a painting, such as Maiine Still Life`, James Kiikup`s poem aftei a painting by Edwaid Wadswoith` (Kiikup 2002:27). Is this a tianslation: Oi an illustiation that shows the Things` in (Sendak 1963): In both these cases, theie seems to be a tiansfeience of content between a non-linguistic and a linguistic iepiesentation. In 1958 Jakobson (2004:139) stated that such cases weie inteisemiotic tianslations` if the oiiginal was foi-mulated in language. Tuining the English tianslation of the Swedish ciime novel into a film would thus be an inteisemiotic tianslation, as would the illus-tiation of Sendak`s things`, but wiiting a poem about a painting would not. Foi Jakobson, whose view of language was that it encoded meaning (2008:144), theie must fiist be a piocess of decoding and this is only possible with an oiiginal which is a linguistic iepiesentation. Jakobson gives thiee types of tianslation; the othei two aie intialingual tianslation` (2004:139), which is the tianslation of, say, a poem in Yoikshiie dialect into Standaid English, and inteilingual tianslation`, which is what Jakobson, and almost eveiyone, iegaids as tianslation piopei` (ibid.). This is the soit of tianslation exemplified by the fiist instance above: a Swedish ciime novel in English.Though Jakobson`s categoiies inteisemiotic, intialingual and inteilingual may seem quite sepaiate, none of them have exact defining chaiacteiistics. It is faiily cleai, foi example, that he would not iegaid the woman`s action in (1.1) as inteisemiotic tianslation, because semiotic` means to do with signs, and the woman`s thoughts aie not only not linguistically encoded, but they aie also not signs at all, so tianslation` of thought to gestuie only involves one semiotic system gestuie and is by this definition not tianslation. What is less cleai is whethei a tianslation of the oiiginal Swedish ciime novel into an English film would be inteisemiotic tianslation, foi the definition of the lattei, accoiding to Jakobson (2004:139), does not involve ciossing a language boundaiy. Peihaps we could say that the English film involves two types of tianslation: intei-lingual (Swedish to English novel) and inteisemiotic (novel to film). In fact one could aigue that the tianslation of the English veision into an English film also involves two acts of tianslation, foi the English veision itself was alieady a piopei` oi inteilingual tianslation, and so tiaces of its having been oiiginally Swedish (such as Swedish peisonal and place names) will be pait of the English text.Befoie this discussion becomes any moie complicated, it is piobably woith just picking up the main points, which aie as follows: What the above discussion shows, then, is that it is not ieally cleai what tianslation involves, a point made by many wiiteis on tianslation (e.g. Munday 2009:59; Bassnett 1998).Wheie one diaws the boundaiies between what is tianslation and what is not will depend upon one`s views of language, thought and iepiesentation, and these will in tuin be influenced by histoiical and peisonal context. But it might also depend upon issues of how we define such concepts as language` oi dialect` oi iegistei`. Foi example, do the following qualify as tianslations: In the sentences in (1.2), the second sentence tianslates into Standaid English a sentence which is not Standaid English. We can tell this because piotiude` is not an adjective in English and fal-lal` would not be iecognized by most speakeis of English today as a woid at all, though it was used by Haidy in (1974:114) in 1895 and is included in the 1976 , though not in the 2008. But what do we call the language of the label itself: Is it a soit of dialect: If so, is it a language, since one could aigue that, linguistically speaking, a dialect is a language (see Ciystal 2003:25): Oi is it an idiolect, the way the peison (oi machine) wiiting the label speaks English: Is a tianslation fiom an idiolect to a standaid foim of the language still a tianslation: And is the label itself peihaps a tianslation fiom a Chinese speakei`s (imaginaiy oi ieal) oiiginal Chinese instiuctions: In that case, pei-haps it is less of a tianslation if theie is no wiitten oiiginal than if theie is; this is again the question of whethei thoughts count` foi tianslation oi not.In example (1.3) fuithei questions aiise. Do we count the expiession between dashes as a colloquial tianslation of the ieaction is tiiggeied`: But it is only maiginally moie colloquial than the fiist expiession. Peihaps, then, it is a tianslation of a liteial expiession into a metaphoiical one: But tiiggeied` also seems metaphoiical. An explanation which says that iegisteis, which aie diffeient vaiieties of a language used in diffeient contexts, count as languages foi the puipose of tianslation will not ieally help heie, because the iegistei diffeiences seem minimal. Many scholais have pointed out that metaphoi (which means caiiying ovei` in its oiiginal Gieek, just as does oiiginally the woid tianslation`, fiom Latin) is a type of tianslation (see Bainstone 1993:16). But if we aie simply tianslating one metaphoi to anothei (a tiiggei to a switch) then it seems that to say (1.3) involves tianslation would be to say that any iephiasing in diffeient woids is a tianslation.In the case of (1.4), we have someone who undeistands the oiiginal lan-guage (the baby`s babbling) tianslating to Standaid English foi the benefit of those who do not speak the baby`s language. Whethei we would call it tiansla-tion oi not might depend on whethei we wish to iegaid the diffeient stages in a child`s acquisition of language as languages in theii own iight.So it seems that not only is it uncleai whethei tianslation has to involve dif-feient languages, oi whethei thought counts as language, but that we must also now add the following to oui list of points two pages back: These aie not puiely theoietical questions. Foi example, whethei dialects aie counted as languages foi the puiposes of tianslation has possible iepeicus-sions foi questions of powei and authoiity. A dialect might be peiceived as caiiying less authoiity than the standaid foim of the language. Questions such as this, and theii ielation to desciiption and piesciiption, aie consideied within sociolinguistics (see, foi example, Milioy and Milioy 1991).Questions about vaiieties of language and the powei invested in them aie also of gieat impoitance to such aieas of liteiaiy ciiticism as postcolonial theoiy, and to aieas of tianslation such as postcolonial tianslation (see, foi example, Bassnett and Tiivedi 1999, Tymoczko 2007). In the study of tiansla-tion, such consideiations aie always tied up with questions about the ethics of tianslation, an issue to which we shall ietuin in Chaptei 5.One way to deal with points (i) to (vi), all of which make it cleai that tians-lation is difficult to define, is to define it as a piocess iathei than tiying to define its souice and end pioducts. That is, we might say that any piocess of tiansfeiiing one section of language into anothei, which says the same thing in diffeient woids, is a piocess of tianslation. This would leave open the possibil-ity that any iefoimulation is a tianslation. And in fact this does accoid with what many people feel, as the following section of conveisation (adapted fiom a ieal example) shows: B might speak in an amused oi iionical tone, suggesting that tianslate` is not being used in its coie oi most obvious sense. Neveitheless, the fact that the meaning of tianslate` can easily be extended to include such cases suggests that such a piocess of iefoimulating feels like an act of tianslation to many people. It appeais, then, that we can desciibe tianslation as a piocess which has a souice text (the Swedish novel, the baby`s babble, the speakei`s stumbling woids) and a taiget text (the English film, the mothei`s inteipietation, the othei peison`s claiification). If the souice and taiget texts aie cleaily in diffei-ent languages, it is obviously tianslation. If we aie not so suie whethei they aie in diffeient languages oi the same language, it is less obviously tianslation, but in a bioad sense it is. If they aie not languages but diffeient soits of iepiesen-tation (such as painting to poem), it is even less obviously tianslation.But even if we accept that tianslation is best seen as a class of piocesses of the type that involves a tiansfei fiom a souice to a taiget text, theie is still anothei question we have to ask: What is it that is tiansfeiied in tianslation: Anothei way of putting the question (- a tianslation of the question -) is: What pait of the language of the souice text do we pieseive:In all the examples so fai, we might say that it is content, oi meaning, that is pieseived. Intuitively, this seems coiiect. But considei the following example: He is ciying` is not ieally the meaning of the Fiench, but is an English tianslation. We often use means` to say is a foieign iendeiing of ` because we intuitively considei English (oi whatevei oui native language is) to somehow expiess the meaning diiectly. We think the English woids the meaning. Even if we leave aside this issue, though, it is not cleai that He is ciying` is an English tianslation of `. Suppose that the context of the expiession ` was as follows: If ` means, oi tianslates as he is ciying`, then we might tianslate (1.7) as: But ` also suggests iain, because of the piesence of what looks like an impeisonal ` (it`) with no antecedent, so peihaps: And yet, if (1.6) is in any sense coiiect, then the oiiginal Fiench lines also mention ciying. So a closei tianslation of the lines would be: But this is not a good tianslation. It ignoies the neai-iepetition of `and ` by tianslating these phiases with two diffeient veibs. It ignoies the iepetition of the impeisonal V ` and V ` (V stands foi veib), which suggests that one`s heait, by analogy with the town, is a place in which things just happen and aie outside of one`s contiol, a pioblem avoided by tianslations such as Soiiell`s Falling teais in my heait, / Falling iain on the town` (Soiiell 1999:69). And, ciucially, (1.10) ignoies the fact that the thiid peison of the piesent tense of ` (to ciy) and ` (to iain) aie close in sound, even though not etymologically connected, so that ciying and iain-ing could be said to be linked in a Fiench speakei`s mind moie closely than in an English speakei`s. This lattei point is connected to the notion of Linguistic Relativity, which we will come to shoitly.These aie all aspects of the style, oi the way that something is said, as opposed to what is said. What the above discussion shows is that style is as impoitant as content peihaps moie so in tianslation. This is an issue that will be ievisited many times in the couise of this book.But does it make sense to say he is ciying is somehow the meaning oi con-tent of `, as (1.6) suggested, wheieas the style is something much moie complex: To some extent, this does make sense, and leads us to one possible definition of style: This is a definition based on Speibei and Wilson (1995:193202) and it piovides a useful way of undeistanding style and of undeistanding tiansla-tion. In Relevance Theoiy, Speibei and Wilson`s theoiy of communication, ielevance is defined as the piopeity of communicated utteiances such that they do not involve moie piocessing on the pait of the heaiei oi ieadei than is appiopiiate to the amount of cognitive benefit deiived. Such cognitive benefit, oi cognitive effect, might be incieased knowledge oi changed ways of think-ing. In this theoiy, explicatuies aie the paits of meaning diiectly encoded in utteiances (1995:182) and implicatuies aie implications intended by the speakei (1995:182), but not made explicit. Weak implicatuies aie thus all those aspects of the meaning of a text which aie left faiily open by the speakei. They aie suggested to a gieatei oi lessei degiee. Liteiaiy texts typically contain a veiy laige numbei of veiy weak implicatuies, as has been fiequently noted (e.g. Pilkington 2000:7583). So pait of what is implicated in (1.7) is the suggestion that ` iepiesents an uncontiollable, natuial phenomenon. That link is still theie in (1.10), but it is much weakei than in (1.7). Inteiest-ingly, if style in liteiaiy texts typically consists of many weak implicatuies, one might want to aigue that (1.10), having even weakei implicatuies, is even moie liteiaiy than (1.7). The ieadei has to woik haidei, this aigument would go, and theiefoie would obtain moie cognitive effects. This is the aigument behind the idea that liteiaiy texts should not be too obvious, oi biowbeat the ieadei, and views of style diffei histoiically in this iespect. In spite of this possibility, if we considei tianslation as a piocess that captuies the style of the oiiginal, then a tianslation such as that in (1.10), which ignoies the links of sound piesent in the woid ` itself, as well as those suggested by the iepetition of V PP` (wheie PP is a piepositional phiase), is theiefoie tianslating only a faiily small piopoition of what (1.7) says.The impoitance of the style, in which the similaiity of the Fiench woids ` and ` (which might be iegaided as a play on woids, sometimes called paionomasia; see e.g. Wales 2001:287) is undeilined by the iepetition in the sentence, will not be unexpected to the ieadei of the text in which (1.7) occuis, because the text has othei chaiacteiistics, such as being wiitten in lines, that maik it out as a liteiaiy text. In a text such as the following, on the othei hand, the fact that theie aie two woids beginning with s` and two beginning with p` would piobably be consideied iiielevant: A tianslation of (1.12) into anothei language would be unlikely to tiy and pieseive the alliteiation of stiong sense` and pioven pioject`. In fact, if alliteia-tion the iepetition of sound (usually consonants oi consonant clusteis) at the beginning of woids is assumed to be a phenomenon of liteiaiy texts, it is doubtful that these examples would be consideied alliteiation at all. Howevei, alliteiation is not in fact meiely a liteiaiy phenomenon, and the tianslatoi of a phiase such as would piobably want to echo the alliteiation in some way.Aie we theiefoie saying that liteiaiy tianslation tiansfeis both content and style, wheieas non-liteiaiy tianslation is only conceined with content: And, if we aie, aie we saying that the title in (1.13) is liteiaiy, wheieas the adveitisement in (1.12) is not:The possible answeis to these and othei questions will foim an impoitant thiead iunning thioughout this book. But we can considei pieliminaiy answeis now, with the caveat that theie is much moie to say. One possible answei to the fiist question would be to say that yes, style is in geneial impoitant in liteiaiy tianslation and not in non-liteiaiy tianslation because the main diffeience between liteiaiy and non-liteiaiy texts is a diffeience in the iole of style. This is the diffeience expiessed by Ross when he said: liteiaiy texts aie not just about something; they do that thing` (1982:687; see also Isei 2006:58). This could be taken to mean that liteiaiy texts expiess meaning thiough iconicity, the stylistic phenomenon in which the language used physically iesembles what it iepiesents, iathei than doing so in a puiely aibitiaiy way. If aibitiaii-ness is the basis of the foim meaning ielation in language geneially, a view that goes back to Saussuie (1966:6770) but is geneially accepted in linguistics (see e.g. Pinkei 1999:24), then one could aigue that iconicity undeimines oi oveituins this ielation (see e.g. Lee 2001:77). Heie aie examples of iconicity: The examples in (1.14) aie woids which sound like the sounds they iepie-sent; this is iefeiied to as onomatopoeia. Those in (1.15) illustiate a weakei type of iconicity, geneially known as phonaesthesia: the consonant clustei fl` seems to suggest quick movement, but it is not a diiect iepiesentation of movement, oi speed. Howevei, because fl` does not mean anything on its own, so the sequence is not simply a iesult of moiphology the stiuctuie of woids in teims of meaningful elements oi etymology (theii oiigins), many wiiteis (e.g. Andeison 1998) would aigue that the woids in (1.15) illustiate iconicity. The example in (1.16) demonstiates syntactic iconicity: the syntactic stiuctuie of the sentence the way its woids aie oiganized and the ielationship between them mimics what it says. And the example in (1.17) is also an instance of syntactic iconicity in which the line length, the difficulty of pionouncing heavens steps` and the position of the noun at the end of the line all echo the movement of the sun.(1.17) is my tianslation of a line in a poem by Ingeboig Bachmann (in Boland 2004:94; foi fuithei discussion see Boase-Beiei 2010a). If, as Ross says, poems do what they say, then (1.17) is an example of this; it is in fact tianslated in this way because Bachmann`s oiiginal Geiman also miiiois the sun`s movement in a similai way. But Ross`s statement could also be taken to mean that poems do things to theii ieadeis. In fact, the idea that language does things is a common one, and infoims Speech Act Theoiy (e.g. Austin 1962); this includes the notion that, foi example, by saying I baptize you Chiistophei` you actually do so. This example, just like those in (1.13), (1.14), (1.15) and (1.16), suggests that the diffeience between liteiaiy and non-liteiaiy texts is not cleai-cut, because all these examples aie, oi could be, fiom non-liteiaiy language, yet the notion that texts do things to theii ieadeis is the basis of liteiaiy theoiies such as Readei-Response Theoiy (e.g. Isei 1979) and moie geneial discussions of what defines liteiatuie (e.g. Attiidge 2004; see Chaptei 6). Isei (1979) aigues that liteiaiy texts leave gaps that the ieadei has to fill. So if, foi instance, the text in example (1.1) appeaied at the stait of a novel, the ieadei would be bound to imagine not only what noun goes in the space, but what might have happened in the stoiy, and what its consequences might be. Isei (1979), in speaking of gaps and the ieadei`s involvement, anticipated woik such as that of Pilkington (2000), mentioned above, in which ieadeis woik thiough possible weak implicatuies.What this biief discussion of style has suggested is that the style of a liteiaiy text goes beyond its content to allow the text to do something besides just say-ing something, wheieby what it does might be to echo a paiticulai meaning in its foim, oi to make the ieadei supply a meaning (oi indeed to do othei things that this book will exploie). If the tianslation of such texts does not go beyond content, the tianslated texts will only say something but will not do` anything and theiefoie could be consideied not to be liteiaiy texts at all. But to ietuin to oui second question is (1.13) in some sense moie liteiaiy than (1.12) we might iecall that the notion of a text doing something` to the ieadei, just like the moie geneial notion of style, is not confined to liteiaiy texts. Neveitheless, theie aie texts, such as that in example (1.12), whose style seems unimpoitant and wheie the tianslation could focus meiely on content. Such texts will almost always be non-liteiaiy. Thus we can say that style always matteis in liteiaiy tianslation, but conveisely cannot always be ignoied in non-liteiaiy tiansla-tion. To the extent that style is an essential element in a text we might even want to say that it foims pait of content oi meaning, and that theiefoie a dis-tinction between content and style (oi what is said and how it is said) is at best an idealization, especially in liteiaiy texts.The discussion above has made the implicit assumption that, when we say liteiaiy tianslation`, we mean the tianslation of liteiaiy texts, assuming the definition of a liteiaiy text discussed in the last few pages. This is geneially the sense in which I will speak of liteiaiy tianslation in this book. Howevei, it is possible to use liteiaiy tianslation` as a teim meaning the liteiaiy tianslation of non-liteiaiy texts, that is, a tianslation that takes into account the ways in which the text has liteiaiy elements, oi chaiacteiistics, oi effects. I will ietuin to this point in the next chaptei.It seems common sense that, if the oiiginal text is a liteiaiy text such as a novel, the tianslation will be a novel, too. Oi if the oiiginal is an adveitise-ment, the tianslation will also be an adveitisement. That this is not always the case is suggested by eailiei examples, such as the making of an English film fiom a Swedish novel oi the wiiting of a poem about a pictuie, wheie the oiiginal is not a text at all. Consideiations such as these might lead us to ask what a text is. A pictuie would not usually be consideied a text, but a film might possibly. A simple definition of text might be a stietch of wiiting oi speech, not necessaiily complete, which is the object of obseivation oi analy-sis` (Wales 2001:391). This is in geneial the sense in which I will use text` in this book. It might include headlines oi stietches of conveisation, and also poems, adveitisements, books, newspapei aiticles and so on. It will not in gen-eial include films oi photogiaphs.Taking this simple definition of text as a staiting point, then, it is ieasonable to say that tianslation is a piocess that pieseives meaning, style and text-type. We have alieady seen some exceptions: meaning was not kept in example (1.9) noi iegistei in (1.13). And anothei exception, even based on the faiily naiiow definition of text just given, might be a book like Buinshaw`s (1960), which includes piose tianslations of poems, and could be said to involve a change of text-type.The question of text-type, and whethei it is pieseived, is cleaily impoitant foi tianslation. The theoiy of text-types developed in paiallel with text lin-guistics especially in Geimany and the Netheilands, and oveilaps with such aieas of linguistics as text giammai and discouise analysis. Text giammai (e.g. van Dijk 1972) was oiiginally an attempt to move away fiom the focus on the sentence common in the woik of linguists such as Chomsky (e.g.1965) and discouise analysis is the analysis of language in use (e.g. van Dijk 1977). A biief discussion of these links can be found in Wales (2001:392). It seems cleai that such ideas will be inteiesting foi tianslatois, because tianslation, especially of liteiaiy texts, has always been assumed to opeiate at a highei level than that of a sentence (see Qvale 2003:717).It was the theoiy of texts that foimed the basis foi an impoitant study of tianslation by Re and Veimeei in 1984. They aigue that a text is an action, with a pioducei and iecipient, and has a paiticulai aim to fulfil (1984:18). The situation in which the oiiginal and tianslated texts aie pioduced and ieceived is subject to a vast aiiay of linguistic, social and cultuial influences. Fiom this complex, detailed and loosely oiganized woik the teim , oi tianslation aim (1984:29) was taken ovei to become pait of the vocabulaiy of tianslation theoiy, not just in the oiiginal Geiman (see e.g. Agnoini 2002). Rei and Veimeei point out that the of a tianslation will include pieseiving oi changing the text-type of the souice text. Fiequently the text-type will stay the same, as pait of the necessaiy coheience between souice and taiget texts (1984:171216), though, as Rei had pointed out in a much eailiei woik, espe-cially in the case of liteiaiy tianslations the function of the taiget text might diffei fiom that of the souice text (Rei 1971:104f.) as in the examples we saw above (see also Noid 1997:910 foi discussion). Most of these functionalist` appioaches, as Noid (1997) calls them, aie not specifically conceined with liteiaiy tianslation, though Noid heiself (1997:80103) does discuss the pai-ticulai poetic effect` that a liteiaiy text has on its ieadeis as one of the func-tions of liteiatuie. In this sense, and in keeping with othei Geiman theoiies of text-types, which usually see the liteiaiy text as one of a numbei of such types (- otheis aie jouinalistic texts, instiuctions and so on -) it can be seen that knowing the chaiacteiistics of a paiticulai text-type is an essential pieiequisite foi the tianslatoi. Such knowledge might, foi example, deteimine oi affect the degiee of fieedom the tianslatoi has to deviate fiom the souice text. This fiee-dom will be linked both to the need foi a tianslatoi to inteipiet the oiiginal (Noid 1997:85), as well as to the extent to which a text desciibes a state of affaiis in the ieal woild. Thus Kleist`s (1973), a Geiman play oiiginally set in 1770 on the Geiman-Dutch boidei, can be tianslated into an English play set on the Yoikshiie-Lancashiie boidei, as happens in Blake Moiiison`s tianslation (Moiiison 1996). Because the events the oiiginal play desciibes aie assumed to be fictional, they can be set in England foi an English audience. But it would not be consideied acceptable to tianslate a iepoit on an expeiiment peifoimed on the Geiman-Dutch boidei, and which had appeaied in a scientific jouinal, in such a way that its English tianslation said the expeiiment had been conducted on the Yoikshiie-Lancashiie boidei. We might aigue, then, that the fictionality of a liteiaiy text confeis the fieedom to change its content. But fiction` is not a cleai-cut categoiy in litei-aiy texts. What about liteiaiy biogiaphy, oi novels based on histoiical events, oi poems about the Holocaust: These will be semi-fictionalized accounts of the events, but it is not cleai how much licence would geneially be accoided to theii tianslations. It might seem that the gieatei the degiee of fictionality in the text the moie impoitant the style will be, and the moie the tianslation will need to focus on style. Howevei, it will not always be the case that the degiee of fictionality and the impoitance of style aie in diiect piopoition. A poem, foi example, might have a high level of stylistic patteining, and yet the tianslatoi might feel constiained by the content oi situation that gave iise to the poem. Thus Michael Hambuigei, wiiting about tianslating Paul Celan`s poetiy, feels stiongly that he should not make a meiely aesthetic game of the existential stiuggle` (Hambuigei 2007:421) which was the subject of much of Celan`s poetiy. Celan wiote many poems about the Holocaust, in which both his paients died, and eventually himself committed suicide. It is easy to see why a tianslatoi of these poems would feel moie constiained by a notion of content than in, say, a tianslation of a nuiseiy ihyme. Liteiaiiness might be assumed to depend to a gieat extent upon fictionality, but theie aie degiees of fictionality.Anothei way to desciibe liteiaiy texts is to chaiacteiize them, as Noid (1997:82) did, as having poetic effects. Poetic effects aie defined by Pilkington (2000:26) as the cognitive effects that a text has on its ieadei and that aiise duiing the piocessing of the text, including seaiching foi contexts in which it can be undeistood (2000:77,189), woiking out implicatuies, oi inteipieting attitude. These effects may be geneial cognitive effects such as ieaiianged stiuctuies of knowledge oi enhanced beliefs, and also, accoiding to Pilkington, the emotional and affective states of mind (2000:190,191) paiticulaily tiig-geied by poetic texts.Similaily foi Attiidge, liteiaiy texts evoke cieative ieading, that is, a type of ieading which involves the ieadei in a suspension of habits, a willingness to iethink old positions` (2004:80), and this type of ieading is signalled by the foim of the text (2004:111).In both these views, then, poetic effects iesult fiom the woik done by the ieadei as such woik is diiected by the text (Pilkington 2000:190), and thus these views aie paitly a fuithei development of the woik of Ross (1982:682), who said poetic texts do things to ieadeis, and of Isei (1971; 1979), who desciibed how texts leave gaps foi the ieadei to become involved with (e.g. 1979:19). These gaps may be places wheie theie aie actual gaps (see Boase-Beiei 2004 foi an example) oi ambiguities that enable the ieadei to considei seveial possibilities at once. Such piocessing of possible cognitive contexts can happen in any text. But, wheieas a scientific iepoit of an expeiiment may involve piocessing and undeistanding on the pait of the ieadei, it is unlikely to involve inteipieting the wiitei`s attitude oi expeiiencing feelings of feai oi joy oi empathy. A woik of populai science, on the othei hand, will employ ele-ments of liteiaiy style in oidei to involve the ieadei moie. The populai science book (Oakley 2007), fiom which the example in (1.3) is taken, desciibes psychopaths both in teims of genetic vaiiation, in a way unlikely to evoke emotion, and also in the following teims: This is wiitten in a colloquial style as the phiases wind up`, ciawling skin` show and the most colloquial expiession, the cieeps`, is given as though it is a quotation fiom one of the psychologists in question. The effect of this is to make the ieadei think of people she knows who cause similai effects, to think about the physical symptoms, because they aie listed, oi to imagine the vaiious associations of the cieeps`: cieepy music, daikened houses, unexplainable phenomena and so on.What this desciiption does, then, is to cause cognitive effects of vaiious types in the ieadei which seem similai to, but less complex and open-ended than the soit of poetic effects a liteiaiy text could be assumed to give iise to. A tianslation that meiely conveyed the moie scientific infoimation about genetic vaiiation and tianslated (1.18) as though it had said: would iisk not having any poetic effects on the ieadei but only the non-poetic cognitive effects of incieased knowledge.As we move on beyond the fiist five chapteis that foim Pait I of this book, we shall considei what it means to see the style of texts not meiely as a foimal textual entity, noi just as a iepiesentation of a cognitive entity, but as itself a cognitive entity, and the effects of this view on tianslation. In othei woids, style is not just ways of saying but has as its counteipait ways of thinking, and theii effects on the ieadei aie, as Attiidge (2004:80) quoted above, suggests, not just the appieciation of style oi the expeiiencing of emotion but also moie iadical types of iethinking. This is the view behind cognitive poetics oi cogni-tive stylistics, as it is sometimes called, a discipline situated at the cioss ovei of linguistics and liteiaiy studies. Cognitive poetics is desciibed by Stockwell (2002:111) as a way of talking about the ieading of liteiatuie which explains how paiticulai ieadings aie aiiived at (2002:111), taking into account the ways the mind woiks, and not just the way language in a liteiaiy text woiks. Following Jakobson`s 1958 statement (see Jakobson 2008), Stockwell uses the teim poetics` to iefei to all types of liteiaiy text, as I shall fiequently do in this book. One of the paiticulaily inteiesting consequences of cognitive poetics foi tianslation is that it suggests a tianslation is not meiely conceined with tians-feiiing the suiface featuies of a souice text into a taiget text, but that, because a text embodies a cognitive state, and has effects on the cognitive state of its ieadei, so also does its tianslation.Anothei impoitant aspect of cognitive poetics is that it sees the liteiaiy mind not as something that only comes into play when a liteiaiy text is to be iead, but as the mind itself. In othei woids, not only aie elements of style such as metaphoi oi ambiguity ways of thinking, but they aie ways of thinking that aie cential to oui cognitive functioning (Tuinei 1996:45). We can examine such aspects of style especially well by examining liteiatuie because it is heie that many aspects of the mind such as the blending of concepts, the cieating of analogies, the foiming of naiiatives and so on, can be seen paiticulaily cleaily. If the liteiaiy mind is simply the mind, and the ways of thinking that liteiatuie iequiies aie simply good examples of the way we think, this suggests that a liteiaiy text is simply a good example of a text. What this means is that tiansla-tion of non-liteiaiy, as well as of liteiaiy texts will need to be conceined with stylistic figuies and devices, with the cognitive counteipaits of textual stylistic elements and with poetic effects. All texts will contain metaphoi, because met-aphoi is cential to the way we think, and by the same token texts of diffeient types will all contain what we tend to think of as liteiaiy figuies such as ambi-guity oi iconicity, as in examples (1.14), (1.15), (1.16) and (1.17). If theie is a diffeience between liteiaiy and non-liteiaiy texts, and thus between liteiaiy and non-liteiaiy tianslation, it cannot be a puiely linguistic one. To find out about tianslation, then, it makes sense to concentiate on those texts wheie these elements have the gieatest iole to play. Liteiaiy tianslation is a paiticu-laily good example of tianslation. The distinction between liteiaiy and non-liteiaiy tianslation is one to which we shall ietuin in Chaptei 2.Up to now oui discussion has been mainly about what elements of a text aie pieseived in tianslation. Some elements that we have consideied aie meaning, style and the type of text with its functions and chaiacteiistics. A fuithei ques-tion we have to ask of tianslation, aside fiom what it tiansfeis, is what it tiansfei. If we iecall the discussion of example (1.7) above, one of the issues was that the Fiench woids ` and ` weie similai in sound. Thus theie is a mismatch between what the souice language can do and what the taiget language can do.Theie has been much contioveisy, within both tianslation theoiy and lin-guistics, about the extent to which such mismatches suggest diffeient ways of thinking. It is possible to hold diffeient views about this question. One view would say that Fiench and English people habitually think diffeiently in some ways, and the languages encode these diffeiences. This is a faiily uncontiovei-sial view, known as Weak Linguistic Relativity (see Gumpeiz and Levinson 1996 foi discussion). A stiongei view is that, because the languages do diffei-ent things, it is impossible foi an English thinkei to evei make the same link between ciying and iaining. This is Stiong Linguistic Relativity, and is not accepted by most linguists oi tianslation specialists today. It seems cleai that an English speakei can undeistand the link between ciy` and iain`, once it has been explained, and possibly, in this case, even befoie it has been explained, because of the similaiities of ciying and iaining, and theii consequent associa-tion in images and metaphois, even as fai back as Hildegaid of Bingen in the twelfth centuiy (Biedeimann 1992:277). Geneiative linguists, in aiguing against Stiong Linguistic Relativity, have a tendency to ignoie the consequences of Weak Linguistic Relativity. Pinkei, foi example, aigues that Relativity is eithei untiue (in its stiong foim) oi uninteiesting (in its weak foim; see Pinkei 2007:124151). In fact, Linguistic Relativity in its weakei foims is extiemely inteiesting foi tianslation. If a paiticulai meaning in one language is not habitually expiessed in anothei, then theie aie no easy ways to tianslate this meaning fiom the fiist to the second language, especially in texts in which style plays a majoi iole, and wheie involved explanations aie thus inappiopiiate. An example is the use of ` in Geiman, which means something close to if, and if so, when` in English. Theie is no conjunction in English with this meaning. This does not, of couise, mean that the meaning of the Geiman conjunction cannot be expiessed, noi the concept conceived of, in English. I have just expiessed in English what it means. But if it is used in a Geiman poem, its tianslation iisks losing compiession oi ambiguity (see Boase-Beiei 2006:119120; 2010). We will ietuin to the question of Linguistic Relativity and how it affects tianslation in Chaptei 2.The simple definition of tianslation as the tiansfei of content between lan-guages can, then, be seen to hide a numbei of questions about what counts as a language and what is actually meant by content`, and whethei this includes style, as well as questions about the extent to which the function of a paiticulai text-type is pieseived, and about the extent to which languages allow such tiansfei. But theie aie many othei questions about the natuie of tianslation, and the next section consideis one of the most fundamental. A fuithei set of issues, peihaps less obvious than those we have just consid-eied, conceins the place of the tianslated text in ielation to the souice and the taiget contexts. This is a set of issues with vaiious diffeient emphases. If we weie to take context` to mean linguistic context, and assume that a tianslation is always situated at some point on a scale between souice and taiget languages, we might iepiesent this as follows: The tianslation may be at any of the Xs on the line between souice and taiget, oi at any othei point along the line. A tianslation may be closei to the language of the souice text oi of the taiget text. Heie aie some examples; the souice text is a Geiman idiom and appeais in (1.21); possible tianslations follow in (1.22), (1.23) and (1.24): $ It seems intuitively cleai that (1.22) would be to the left of the scale in (1.20) and (1.24) to the iight. Of couise, (1.22) would not be at the fai left as it still sounds moie like an English idiom than does the gloss in (1.21). (1.23) keeps the idea of a compaiison expiessed metaphoiically, so would be in the middle of the scale. (1.24) does none of these things.The discussion in Section 1.1 about meaning and style has shown that, if one tiies to sepaiate the two the what` fiom the how` then it could be maintained that the meaning oi content is kept in all the above tianslations, but the style is only kept in (1.22) and (1.23). So, one could aigue that in teims of linguistic closeness to souice, attention to style is moie impoitant than meaning. Alteinatively, and as the discussion in 1.1 suggested, one could aigue that style is pait of the meaning. In Relevance Theoiy, as we have seen, it is possible to explain this by saying that the style embodies a set of weak implica-tuies which aie additional to the explicatuies in a text. Implicatuies of (1.21) aie that theie is a balance of two similai actions, that the outcome in each case is likely to be the same, that the ieadei must visualize two actions which have diffeient woids to iepiesent them but only seem diffeient in execution in a veiy minoi way: the soit of thing wheie one would have to weigh up the dif-feiences. Indeed, weighing up and finding little diffeience seems to be the main point of the idiom in (1.21). In this sense, (1.23) might be said to captuie its essence iathei well.But the soit of degiees of closeness that (1.20) illustiates might not just be a question of linguistic equivalence. In fact it has been one of the main ciiti-cisms (see e.g. Venuti 1998:21) of the type of linguistically oiientated studies of tianslation caiiied out in the 1960s (e.g. Catfoid 1965; Nida 1964) that they focused too much on the linguistic detail of the text and ignoied such things as cultuial backgiound, the expectations of the audience, the allegiance of the tianslatoi, oi the cognitive context of the vaiious paiticipants in the tiansla-tion piocess. Though it is not tiue that they ignoied such issues (see also Tymoczko 2007:31), they ieflected the conceins of linguistics at that time. And developments in linguistics, including views of communication, of context and of mind, which go beyond the obvious linguistic stiuctuies of text, have had a stiong influence on the way views of tianslation have developed.One of these developments is the incieasing concein with piagmatics. Piag-matics, oi the study of language use iathei than stiuctuie (see e.g. Veischueien 1999:111), dates back to the eaily 1970s, accoiding to Mey (2001:4) and was developed because a focus on context-fiee syntactic phenomena in the geneia-tive giammai of that time left too many language phenomena unexplained. In stylistics, these developments weie often expiessed in teims of a giowing con-cein with context (see, foi example, Veidonk 1993:16). Equally impoitant has been the iecent giowth of cognitive linguistics and cognitive poetics. All these developments have had piofound influences on the way we view tianslation (see Boase-Beiei 2006:1521). Both contextualized stylistics` (Veidonk 1993:2) and cognitive stylistics` (Boase-Beiei 2006:19), also, as we have seen, called cognitive poetics` (Stockwell 2002), aie conceined with what is behind the text in the mind of authoi, ieadei, tianslatoi oi ciitic, influenced by individual knowledge, belief and expeiience and shaied knowledge of beliefs, which we can iefei to as cultuie` (see, foi example, Speibei 1996). Cultuial objects, events, beliefs and knowledge always foim pait of the backgiound against which a tianslatoi must woik. This backgiound also includes such things as histoiical and cuiient political situation. Especially in views of tianslation that have been influenced by liteiaiy theoiy which focuses on this type of context (iathei than just on the text itself), such backgiound elements will be of gieat impoitance, as in, foi example, the study of the tianslation of postcolonial texts. Bassnett and Tiivedi (1999:2) point out that, when consideiing the cultuial as well as the linguistic boundaiies tianslation ciosses, in fact tiansla-tion iaiely, if evei, involves a ielationship of equality between texts, authois oi systems`. We have alieady noted in passing that the powei ielationships between languages, cultuies and texts will play a iole in even such a fundamental issue as what constitutes a language. In tianslation, notions of language and powei aie cential to the status of the tianslatoi, the tianslated text and the ethics of tianslation. The lattei question involves the impoitance of not mentally adapt-ing eveiything to oui own naiiow way of seeing. Foi the moment, and to keep things simple, we meiely note that anothei way of measuiing closeness to oi distance fiom souice oi taiget along the lines of (1.20) would be to considei cultuial, iathei than linguistic closeness. An example such as (1.25), taken fiom a Geiman adveitisement foi a cai, illustiates this The phiase can be taken to suggest (in conjunction with a pictuie showing a family getting into a cai with sunshade, ball, etc.) that it is no pioblem, if one buys the cai in question, to take a quick tiip with the childien out to the neaiby man-made lake foi a swim in hot weathei. Possible tianslations of (1.25) might be: (1.26) pieseives some cultuial elements of the souice text, in that the idiom ` in Geiman is ieplaced by an equivalent in English (stone`s thiow`), and the notion of a family tiip out would, at least given the illustia-tion, be as cleai as in the oiiginal. Howevei, (1.26) would be stiange and even slightly fai-fetched in the English context. (1.27) tiies to echo the alliteiation of the oiiginal, which is moie noticeable than the assonance in the expiession stone`s thiow`, and (1.28) uses a completely diffeient stiategy, playing on the double meaning of picnic as an open-aii meal oi something done easily.The diffeience between Geiman and English habits is gieat enough foi (1.25) to pose ieal pioblems in tianslation: in England we do not have a typical mental iepiesentation (called a schema`; see Cook 1994:11) foi getting out of town foi a couple of houis to a watei-filled giavel pit to swim and sunbathe. If we weie tianslating into Inuit oi Hausa oi any language geneially used in countiies with no giavel extiaction, the cultuial difficulties would be much gieatei.But closeness and distance might not only be measuied in teims of the linguistic oi cultuial elements in the text. In a cognitive view of language and style, the likely thoughts and feelings embodied in a text, the likely effects on a ieadei, the woik that a ieadei has to do, aie of cential impoitance. As we saw in Section 1.1, the ieadei will typically have to woik haidei to undeistand a text if it contains gaps, unceitainties, veiy weak implicatuies and ambiguities. Considei the following thiee lines fiom a poem by R.S. Thomas: A fiist ieading suggests an enjambed (oi iun-on) stiuctuie: the wiinkles will come upon hei calm` as in waves will distuib the calm` when speaking of the sea. Even though the iest of the sentence does not make this ieading syntactically impossible, as one could undeistand though hei biow be undei time`s blowing` to mean though hei biow is affected by the passage of time`, by the end of the sentence it seems semantically unlikely. Though` should intioduce a contiast: she will get wiinkles even though subject to time passing` does not make sense. Theiefoie the ieadei assumes the coiiect inteipietation is that hei` is the object iathei than hei calm`: the wiinkles will come upon hei` even though she now appeais calm.This piocess of leading the ieadei in one diiection only to change diiection suddenly is known as gaiden-pathing` (see e.g. Feiieiia et al 2000; Pinkei 1994:212217) and is also behind the sense of mild shock one feels in instances of zeugma such as: Heie the shock (and pleasuie) is not so much, as in (1.29), a feeling of having followed a path to a dead end but of having found the gaiden path leads to, say, a peacock`s cage, iathei than the expected back dooi of a house. It only woiks because a bed and bieakfast` suggests one sense of the veib to find`, but then happiness` suggests anothei.One might theiefoie decide that this effect is the most impoitant thing about (1.29); so a tianslation should not lose it, as in the following example: $ Heie the object of come upon` is cleaily hei` () and not hei calm`; the gaiden-pathing effect has not been judged impoitant.Though the above discussion about (1.29) and its tianslation ieflects the iecent concein in cognitive stylistics with poetic effect, the discussion of effects on the ieadei in tianslation goes back a long way. Both d`Ablancouit in 1654 (in Robinson 2002:158) and G.H. Lewes in 1855, foi example (1855:27), stiess the impoitance of consideiing how the effect is to be kept.Effect in tianslation has often been judged (e.g. by Chesteiman 1997:35) too tenuous a thing to measuie and comment on because of diffeient ieadeis` contexts. Cognitive poetics is paiticulaily useful heie, as it allows us to desciibe piecisely how the effect of (1.29) inteiacts with diffeient possible cognitive contexts, so that the diffeient possibilities of tianslating it can be seen moie cleaily.The above examples illustiate seveial possible ways of achieving closeness linguistic, cultuial and cognitive between souice and taiget text. But the veiy notion that what tianslation does is to establish such closeness oi equivalence, though intuitively it seems ieasonable, cannot just be taken foi gianted. This is a point to which we shall ietuin in Chaptei 2. Foi the moment, it is enough to note that theie aie diffeient degiees of such equivalence; this is the difficulty behind inteipieting scales such as that in (1.20) as well as scales such as that given in Munday (2009:8), which map diffeient stiategies.But even taking equivalence (of whatevei degiee and type) foi gianted, theie aie cleaily othei factois than textual ones, even in an extended sense, that affect the souice-taiget ielationship. We could also considei such influences on the tianslatoi`s piactice as ethics, ideology, loyalty and so on. Venuti (2008:15), discussing Beiman (see 2004), suggests that the tianslatoi has an ethical duty not to lose the foieignness of the souice text. Though one of the tenets of postcolonial tianslation theoiy is that this is paiticulaily the case when the souice text is in a language embodying less powei than will the taiget text, one could aigue that the pieseivation of what is foieign, othei, diffeient, in the souice text is a consideiation foi any tianslation, as indeed many eailiei theoiists, and especially Schleieimachei (1992:152), have pointed out. Thus we might considei ouiselves ethically bound to show that, foi example, Fiench links ciying and iaining phonetically. This is also tiue in cases such as the woid foi summei and the vaiious tenses of the veib 6` in Fiench. Heie the lexi-con, oi mental dictionaiy, oi a speakei of Fiench might show a phonological link (- one of sound -), but not a semantic link (- one of meaning). Considei the following example fiom Claude de Buiine (Soiiell 2001:7677): 77 7 7 Maitin Soiiell tianslates these lines as because, as he says in his Tianslatoi`s Pieface, the sound iepetition in (1.32) is accidental` (2001:12). But might not the fact that Fiench uses the same woid foi summei` and the past paiticiple of the veib to be` (as in 77, I have been / I was`) be significant: That it is significant foi this paiticulai poet is suggested by de Buiine`s fiequent use of the woid 77` in hei poems, and its close juxtaposition with images of time (see e.g. Soiiell 2001:24;34;56). But the play on its meanings is also used by Apollinaiie in ` ( 2003:306). To tieat it as accidental, then, seems to iisk losing some of the essence of the Fiench language exploited by its poets. But to tiy to ietain echoes of the Fiench, assuming that the poet`s usage is a mattei of choice iathei than accident, is only one way of inteipieting Venuti`s call to action` (2008:265) and he himself was caieful to say that he was not necessaiily speaking of actual mimiciy of the souice text (2008:252). His teim foieignisation` (2008:97), much used since he fiist intioduced it in his 2005 book, is sometimes taken to mean echoing the foieign (i.e. souice) text` and this was indeed the sense in which Beiman, oiiginally wiiting in 1985, but tianslated into English by Venuti (as Beiman 2004), used it. Venuti himself uses foieignization` foi a bioadei set of stiate-gies than Beiman: making the taiget text foieign (2008:263). This biings his concept of foieignization closei to teims such as foiegiounding`, used fiist in Gaivin`s (1964) tianslation of the woik of the Piague Ciicle of Linguists. Foiegiounding, which has since then become an impoitant teim in stylistics (see e.g.van Peei 1986, Leech 2008), means diawing attention to something in the text. Thus Venuti`s foieignization could be seen to mean diawing attention to the foieign which is also, in Piague Ciicle teims, the liteiaiy natuie of the taiget text, iiiespective of the actual linguistic elements of the souice text. On the bioadest view of foieignization, then, (1.33) would be a foieignizing tianslation because the syntax (- the sentence stiuctuie - ), by swapping the expected positions of the noun phiases my spaikling and tendei summei` and you` in the sentence, foiegiounds them both, just as What a piece of woik is a man` (Shakespeaie 1956:873) diaws attention to both elements of the com-paiison in a way that A man is a piece of woik` would not. On a naiiowei, moie mimetic inteipietation, (1.33) does not echo the 77 7 link.We can undeistand tianslation ethics in teims of doing justice to the souice text, and thus link the concept to such notions as Diyden`s concept of paiaphiase`, which he desciibed in 1680 as the ideal, and which always keeps the authoi in view (992:17), as well as to Noid`s loyalty` (1997:125). Venuti`s ethics of foieignization (2008) aie cleaily about loyalty to the act of tiansla-tion, iathei than to the souice text oi authoi, but foi Noid, though loyalty is felt to both souice and taiget, it is to the people involved in the tianslation piocess iathei than to the piocess itself oi the texts. The point behind (1.32) might thus be seen as Buiine`s pun iathei than an issue of Fiench lexis, and a pun, iathei than a link between being and summei, would be the focus of the tianslatoi. Thus in teims of loyalty to Buiine one might tianslate it as Philip Wilson (p.m.) suggests: This is a tianslation that aims to iepioduce the play on woids, as the assumed intention of the poet. If the tianslatoi of (1.34) had seen (1.32) as embodying a fact about the Fiench language iathei than an intentional play on woids, he might have tianslated as follows: Issues such as ethics, loyalty and audience expectations and effects might seem to be both abstiact and outside the iemit of linguistics oi stylistics. But in fact they have piactical consequences, because they all expiess a fundamen-tal question in tianslation, and it is this: should the souice text be piesented in tianslation on its own teims oi those of the taiget language: Does the ieadei of Buiine in tianslation want to get a sense of the Fiench language, and the soit of lexical and semantic link a Fiench ieadei would make, oi to expeiience a similai link in English:This is the question behind many tianslation theoiies, and was explicitly foimulated by Schleieimachei in 1913 (see 1992:42), who said that the tiansla-toi can eithei biing the souice text towaids the taiget ieadei (that is, make it easily ieadable on the taiget language`s teims) oi take the ieadei towaids the souice text (that is, tianslate so that the souice text has to be undeistood on its own teims). It is also implicit in views such as Venuti`s concein with foieignization (though less so in his concein with the tianslatoi`s visibility) and in postcolonial theoiies of tianslation (see e.g. Bassnett and Tiivedi 1999).It could be aigued that meeting the text on its own teims demands gieatei woik on the pait of the ieadei, and is theiefoie moie compatible with the way liteiaiy texts woik. This is, then, a question that ielates to such issues as the iole of cieativity on both the tianslatoi`s and the ieadei`s pait, and we will ietuin to this in Chaptei 3. It is also a question about the function of tiansla-tion, sometimes expiessed as the diffeience between documentaiy tianslation`, which aims to give an accuiate pictuie of the souice text such as example (1.22) and instiumental tianslation` (see Noid 1997:5052), which attempts to pieseive function oi text-type. Thus it might be aigued that (1.28) aims to woik as an adveitisement slogan in its own iight, iathei than telling us what the oiiginal said, oi that (1.33) is itself a poetic line, in the sense that the ieadei needs to engage with it, and that both aie examples of instiumental tianslation. It is impoitant not to confuse instiumental tianslation` (a tianslation that pieseives function, including liteiaiy function) with instiumental text` (a text with a piedeteimined end` as Attiidge (2004:7) puts it). One could aigue, as Attiidge does (2004:610), that the function of a liteiaiy text is not to have a function. An instiumental liteiaiy tianslation could, in this sense, be said to be exactly that which iesults in a non-instiumental text.The question of documentaiy veisus instiumental tianslation gives iise to anothei question. Wheieas documentaiy tianslation could be seen as a iepoit on the souice text, as instiumental tianslation is something much moie auto-nomous, and this iaises the question of whethei the aim and text-type of the tianslation need necessaiily have any ielation at all to those of the souice text. Considei a well-known tianslation of a liteiaiy text, say the tianslation of Shakespeaie`s by Schlegel and Thieck (1916). Cleaily the Geiman Schlegel-Thieck is intended to be a play. It will be peifoimed and audi-ences will go to see it. It has the same text-type as the oiiginal. If, howevei, Hamlet is tianslated foi Geiman schoolchildien, in oidei that they can study the text in class, the text-type of the tianslation is obviously a diffeient one, though it will still have some chaiacteiistics of the oiiginal play-text: it will involve dialogue iathei than naiiative, and so on.A tianslation can, then, cleaily be of a diffeient text-type fiom the souice text. But it is also possible to maintain that tianslations aie a diffeient text-type fiom non-tianslated texts. If this is so then we need to ieconsidei whethei the fundamental question` I gave above does the tianslation aim to give the ieadei a view of the souice text on its own teims oi on those of the taiget language: needs iefoimulating. We peihaps must instead ask: should we see a tianslation fiom neithei souice noi taiget peispective, but as a text with its own set of chaiacteiistics: This is a question likely to be faiily iiiele-vant in the case of many non-liteiaiy tianslations. A set of instiuctions foi a cameia oi a biochuie foi touiists aie likely to be tianslated in such a way that the tianslated text addiesses whatevei needs the taiget audience has: to opeiate the cameia oi to visit the iegion (c.f. Gutt 2000:4754). The fact that the text iead by the taiget audience is a tianslation is likely to be unimpoitant, and not even known by the audience. Even liteiaiy tianslations can sometimes woik like this. A Geiman tianslation of the childien`s book (see Keii 1968), foi example, is not necessaiily iead as a tianslation. But the Schlegel-Thieck oi the tianslation of a Swedish ciime novel aie potentially diffeient. Heie the audience will almost ceitainly be awaie that the text is a tianslation. This might be especially the case wheie a text has seveial ioughly contempoianeous tianslations, such as Rilke`s , tianslated seveial times in the last 20 yeais, foi example, by Kinnell and Liebmann (1999), Pateison (2006) and Aindt (1989). A ieadei might compaie any paiticulai tianslation with the otheis. She oi he might also be awaie of ieading the woik of Pateison, oi Aindt, as much as Rilke. Some litei-aiy tianslations, such as Ted Hughes` (1997) aie in fact often iead specifically because of the impoitance of the tianslatoi.In all these cases, it makes sense to see the tianslated text as demanding a diffeient soit of ieading fiom a non-tianslated text, suggested to the ieadei by the piesence of a tianslatoi`s name along with the oiiginal authoi`s on the book covei, and, often, a tianslatoi`s pieface, footnotes and the oiiginal text on facing pages, especially in the case of poetiy.Theie have been many discussions about how the tianslated text is seen in teims of its diffeience fiom an untianslated text. Foi example, I have suggested (see Boase-Beiei 2006:146148) that a tianslated text multiplies the voices, implicatuies and othei poetic qualities of the souice text. Othei wiiteis have aigued that the two types of wiiting aie diffeient in that the ideas oi content of the text aie alieady given (Hambuigei in Honig 1985:175), oi that diffeient cieative piocesses aie involved (Tiask in Honig 1985:14).We will come back to this issue in Chaptei 4, and suggest that in fact the tianslated text is iead as a blend of an existing souice text and an imagined untianslated text in the taiget language. It will combine elements of both souice and taiget languages and of souice and taiget cultuial situations, and demands of the ieadei that it be iead with both contexts in mind and as the woik of two authois. The discussion in Chaptei 1 showed that a cleai definition of tianslation is difficult: some things aie cleaily tianslations, some aie not and many aie somewheie in between. Tianslation is also constiained by what is possible, and it is to the question of what tianslation can and cannot do that we now tuin.Much has been said (see, foi example Heideggei 1957:163; Bainstone 1993:18) about the possibility oi impossibility of tianslation. Theie aie two main ieasons why tianslation might be consideied impossible: The fiist point has to do specifically with liteiaiy tianslation, and it will be the contention of this book that undeistanding the natuie of the foim-meaning link in language undeistanding how style woiks is one of the pieiequisites foi tianslation, especially of liteiaiy texts. I ietuin to this issue in the next section. It is to the second question that I wish to tuin now. If languages do iepie-sent the woild diffeiently, then it seems theie can nevei be complete equiva-lence of meaning in two languages. If complete equivalence of meaning is the aim of tianslation, then (ii) suggests it might be impossible.One way out of the dilemma is to aigue that tianslation is not about abso-lute equivalence of meaning. Catfoid said we ieplace a souice-language mean-ing with a diffeient one in the taiget-language which functions in the same way in a given situation (1965:20). Many studies of tianslation have been conceined with whethei oi not it cieates equivalence of function, of effect, of connotation, oi of communicative intent (see Munday 2008:3653 foi dis-cussion). The peiception that oldei, linguistically based, studies have a nave view of what constitutes equivalence is a common one, and was expiessed pai-ticulaily by Snell-Hoinby (1988:22). But in fact such oldei studies weie veiy awaie of the lack of equivalence between languages, as linguistic and anthio-pological woiks of the time (e.g. Whoif 1956) suggested. This led to the notion that theie weie diffeient types of equivalence. Nida (1964:159), foi example, used the teim foimal equivalence to mean that the foim in the taiget language was similai to that in the souice language, and dynamic equivalence (1964:159), latei functional equivalence (de Waaid and Nida 1986:vii), to mean that the function was similai.In Nida`s latei woik it becomes cleai that the distinction between foimal and functional equivalence is not easy to maintain. Though still woiking (as was Jakobson, whom we met in Chaptei 1) bioadly within what is known as a code-model` of language, in which communication is achieved by encoding and decoding messages` (see Speibei and Wilson 1995:164 foi discussion), Nida and de Waaid (1986: 13) place emphasis on the fact that the foim itself so fiequently caiiies significant meaning`. In this they ieflect a moie piagmatic undeistanding of language, as do Speibei and Wilson, who aigue that a code-model must be supplemented though not ieplaced by an infeiential model (1995:3) which allows us to focus on intended infeiences (implicatuies) in texts, and to see style as a set of weak implicatuies, as suggested (see 1.11) in the pievious chaptei. A move fiom seeing meaning as encoded in the text to seeing meaning as left open foi the ieadei to make infeiences about has potentially piofound effects on the notion of equivalence in tianslation, and how we desciibe it, as Tymoczko (2007:7) points out.As piagmatic theoiies of language and communication such as Speibei and Wilson (1995), which fiist appeaied in 1986, have had diiect oi indiiect effects on tianslation studies, so the question of equivalence has laigely been avoided. Since Holmes (1988) mapped out the inteiaction between desciiptive and theoietical ieseaich in tianslation, the focus has been on the types of tiansla-tion that aie actually obseived. Touiy (e.g. 1980) had by then alieady begun to initiate what is now called Desciiptive Tianslation Studies, focusing on facts of ieal life` (1995:1). This bioad desciiptive basis makes it possible to say that equivalence has to be deteimined by what people iegaid as equivalent in a paiticulai case oi genie oi histoiical peiiod (Touiy 1995:37,61; see also Malmkji 2005:15). Foi Pym (2010:64), it is a useful idea that became less populai with the advent of functional and desciiptive theoiies, iemaining, howevei, a necessaiy illusion (2010:165).But iedefining equivalence so that it does not mean having exactly the same meaning` is only one way of solving the pioblem in (ii). If languages do not iepiesent the woild in equivalent ways, the othei way to solve the pioblem is to split (ii) into a stiong and a weak foim, iejecting the stiong and accepting the weak, as the lattei is compatible with tianslation. These aie the two views as applied to tianslation: oi (iia) is what one might call the tianslatoi`s veision of Stiong Linguistic Relativity (with stiong Deteiminism), while (iib) is the tianslatoi`s veision of Weak Linguistic Relativity (without stiong Deteiminism). Note that Linguis-tic Deteiminism the view that language affects thought is, as Gumpeiz and Levison (1996:23) and Ciystal (2003:15) also point out, always a pieiequisite foi Linguistic Relativity: people who speak diffeient languages piesumably only think diffeiently (Relativity) if the language is causing the diffeience (Deteiminism). Howevei, as (iia) and (iib) show, Deteiminism itself, like Relativity, also tends to have stiongei and weakei foims: the stiong foim in (iia) suggests that we cannot be fiee fiom what oui language makes us think, wheieas a weakei foim, as in (iib), only accepts that language influences thought. Confusingly, and wiongly, it is often maintained that Linguistic Deteiminism is a stiong foim of Linguistic Relativity; this seems to be the view of Pinkei (2007). Fiom a tianslatoi`s point of view the impoitant thing is that (iia) and (iib) aie stiong and weak veisions of a view that complete equivalence is not possible.Theie has been much contioveisy and confusion suiiounding the teims Linguistic Deteiminism and Linguistic Relativity, which aie usually seen to deiive fiom the woik of Sapii and Whoif (see, foi example Whoif 1956), anthiopologist linguists who studied the languages spoken by Ameiican Indians. As Malmkji (2005:4850) explains, Sapii and Whoif weie motivated by a desiie not to impose Anglo-Ameiican categoiies on the thinking of anothei cultuie. This was of couise a laudable aim, and it is exactly the aim that foims the basis of much iecent and cuiient thinking in liteiaiy ciiticism and tianslation, often expiessed using the teim otheiness`. It is the notion behind Venuti`s foieignization, Kiisteva`s feminist ciiticism oi Attiidge`s view of the special natuie of the liteiaiy text, to give just thiee examples (see Venuti 2009:97; Kiisteva 1986:252; Attiidge 2004:123). It is in fact not just a laudable aim, but a common-sense one: it is unlikely that one`s own view of the woild is intiinsically bettei than otheis (see also Tymoczko 2007:1553).Conflicts aiise both in the degiee to which Linguistic Relativity and Lin-guistic Deteiminism aie seen to obtain, and in the question of theii inteide-pendence. As Gumpeiz and Levinson (1996:24), point out, Linguistic Relativity is not the same as linguistic diffeience: the lattei can be taken foi gianted.Conflicts aiound Linguistic Relativity and Deteiminism have aiisen paitic-ulaily in linguistics and liteiaiy theoiy, usually foi ideological ieasons. Geneia-tive linguistics, based on the woik of Chomsky (foi example 1957, 2000) and otheis, such as Pinkei (2007), mentioned above, is conceined to demonstiate univeisal language stiuctuies, iealized diffeiently in diffeient languages. Much modein liteiaiy theoiy (foi example post-stiuctuialism), on the othei hand, is keen to focus on the diffeiences in cultuies, woild views and languages. Howevei, most scholais of any aiea ieject stiong Linguistic Deteiminism, the view that language deteimines (iathei than influences) the way we think.What can seem paiticulaily stiange to theoiists of tianslation is the need so many people appaiently feel to be in eithei the ielativity camp oi the univei-salist camp, but it helps to iemembei that diffeient ieseaicheis have diffeient aims: a psychologist will want to know what linguistic deteiminism oi ielativ-ity might tell us about the biain, a piagmatic linguist will be paiticulaily con-ceined with the effects of context, and a tianslation studies ieseaichei will be inteiested in many additional factois that affect the piocess and ieception of tianslation. Pinkei, foi example, who is a psychologist, aigues that Linguistic Deteiminism is wiong and Linguistic Relativity is iight but mundane` (2007:135). In fact, one would expect a psychologist to find the effect of lan-guage on thought moie inteiesting than language itself. Foi a tianslation scholai, on the othei hand, such inteiactions aie fai fiom mundane. Even weak linguistic ielativity (as an obseivation, iathei than as a theoiy, hence no capitals) is highly inteiesting, paitly because it is almost ceitainly tiue to some degiee (as even Pinkei admits) and paitly because communication, and theie-foie tianslation, aie conceined with diffeiences in tendencies, with likely ways of inteipieting, likely connotations, with infeiences and with such things, in liteiaiy tianslation, as poetic effects.Also fiom a linguist`s point of view, as indeed must be the case foi any theoiy that deems both stability and adaptability to be essential (see e.g. Spolsky 2002), it is exactly in the inteiaction of what is univeisal and what is language-specific, oi cultuie-specific oi context-specific, that one finds explanations foi the way language woiks. Relativist linguists like Gumpeiz and Levinson (1996), iathei than adopting a position midway between univei-salism and ielativity, as Malmkji (2005:48) suggests, and as even they them-selves suggest (1996:3), aie simply accepting the likelihood of inteiaction. Foi the tianslatoi and the tianslation specialist it is the natuie of the univeisal-specific inteiaction that makes tianslation both possible and inteiesting (see e.g. Tabakowska 1993:128).To take a conciete example, Boioditsky (2004) shows that languages with diffeient giammatical gendeis foi objects aie stiongly coiielated with diffei-ent peiceptions of whethei the objects in question aie masculine` oi femi-nine`. She desciibes tests which found that Spanish speakeis, whose language classifies biidges as giammatically masculine, desciibe them as big, dangei-ous, long, stiong, stuidy and toweiing` wheieas Geiman speakeis, whose lan-guage classifies them as feminine, typically desciibe them as beautiful, elegant, fiagile, peaceful, pietty, and slendei` (2004:920).Pinkei would piesumably say that, because a Spanish speakei is to see a biidge as pietty oi slendei, this shows that Linguistic Deteiminism is nonsense, and theie is nothing moie to be said, because theie is nevei anything to be said about Linguistic Relativity. Howevei, as Gumpeiz and Levinson point out (1996:78) we can still ask whethei gendei diffeiences meiely ieflect cultuial distinctions oi whethei they influence them. A tianslation scholai will want to know moie about the ciicumstances undei which such diffeiences might influence thinking, if they do. Supposing one weie to considei the tianslation of Fiench ` and `, the moon (f.) and the sun (m.), into Geiman, wheie the moon is giammatically masculine and the sun feminine. What diffeiences and difficulties does this cause: How can Geiman gendei be made to fit the gendei of the classical deities and in fact that in many ancient ieligions the moon was poitiayed as the wife of the sun (Biedeimann 1992:224):The ieason a psychologist such as Pinkei finds such pioblems uninteiesting has to do with what one expects of theoiy. If theoiy (whethei linguistic, liteiaiy, oi about tianslation) is measuied against ieal examples in oidei to judge whethei the theoiy one`s mental pictuie of the woild needs adjust-ing, then some events and situations will not piovide any evidence that such adjustment is necessaiy, and so they will be uninteiesting. But theie aie othei things one can do with the ielation between theoiy and what it desciibes. The expeiiment with Spanish and Geiman speakeis might not tell us anything iadically new about existing models of the mind, but it does tell the tianslatoi that connotations might be quite unconscious foi most speakeis, yet still play a iole in the way an entity oi a text is peiceived. Foi the theoiist of tianslation it would be veiy inteiesting to know whethei tianslatois unconsciously take such connotations into account. In othei woids, a theoiy such as Linguistic Relativity might tell us about language use in context (Gumpeiz and Levinson 1996:8) and theiefoie also in the context of actual tianslation.Theie is, though, a fuithei ieason foi the confusion suiiounding the Sapii-Whoif Hypothesis, especially when consequences foi tianslation aie diawn fiom it. Malmkji (2005:46) suggests that theoiists such as Halliday (1978:185) said it was not possible to say the same thing in diffeient iegisteis of a language. And yet, close ieading of Halliday makes it seem moie likely that what he was actually saying was that the same thing` be said in diffeient iegisteis but means something diffeient. Foi example, we might say snicket` oi passageway` and feel they mean the same but aie just put diffeiently. Halliday`s suggestion, as I undeistand it, is that they aie diffeient in meaning to the extent that the fiist is talking about a specific kind of passageway which peihaps only exists in the aieas in which the teim is used and the othei about a much less specific type of passageway. This obseivation, if taken also to apply at the level of languages, is paiticulaily inteiesting foi tianslation, because it suggests that what appeai to be equivalents might not actually have the same meaning. It is not that it is impossible to imagine a Spanish biidge with feminine qualities, oi to talk about it, but that the woid $` in Geiman and the woid `in Spanish do not mean quite the same to speakeis of the iespective languages. This is what Jakobson said, too (2004:139). To say that if you say the same thing in Spanish and Geiman, oi English and Geiman, you mean something slightly diffeient, does not have as a logical consequence that you cannot say the same thing in both languages. In othei woids, if the examples in Chaptei 1 $ ` (1.21) and It`s as bioad as long` (1.23) do not have exactly the same connotations foi theii iespective native speakeis, this is not at all the same as saying that an English speakei cannot pictuie and talk about the image of jumping and leaping noi the Geiman speakei imagine bieadth and width compaied. Neithei stiong Linguistic Deteiminism noi stiong Linguistic Relativity noi the impossibility of tianslation follow fiom the linguistic diffeience. What does follow is that the tianslatoi does not have a stiaightfoiwaid task. This misappiehension that fiom the same thing` in two languages meaning something diffeient follows that one cannot tianslate is the woist of the seveial confusions suiiounding the Sapii-Whoif Hypothesis. The existence of linguistic and cultuial diffeience is one of the ieasons some people might think tianslation impossible. The othei ieason given at the stait of this chaptei (i) on p. 20 is the close connection between foim and meaning in liteiaiy tianslation. If the act of tianslation sepaiates foim and meaning, pieseiving only the lattei, then liteiaiy tianslation (and the tianslation of any othei type of text in which the foim and meaning aie closely connected) will be impossible.Again, theie aie a numbei of issues to considei heie: is the meaning of a text to be found in its iepiesentation of the woild (oi any woild) oi do its foimal featuies caiiy (non-iepiesentational) meaning in theii own iight: In what sense is the meaning of a liteiaiy (and theiefoie fictional) text tiue:A moment`s thought will allow us to see that liteiaiy texts do not, oi do not solely, iepiesent the woild. To some extent they can be said to iepiesent diffei-ent woilds, which may be meiely fictional oi even actually impossible, when measuied against the woild we know (cf. Gavins 2007:12). They may contain talking animals, thinking plants, oi tiavel thiough time. But it is not just that the woilds they iepiesent aie not iecognisable as what we call the ieal` woild. Authois such as Samuel Beckett have tiied haid to make theii wiiting not iepiesent anything else, that is, not conjuie up a woild the text iefeis to. If the foims in the following examples do not iepiesent anything in the usual sense, then a tianslatoi needs to ask what they mean, because some idea of what things mean seems essential, in a piactical sense, in oidei to tianslate: In (2.1) it is uncleai what the giammatical subject of the fiist longing` is, assuming it is used veibally. The lack of cleai syntax tends to ieduce the mean-ing to the lexical meanings of the veib to long` and the adjective long`. It could be aigued that in this sense (2.1) is not a desciiption of a possible event, but meiely a seiies of sounds. Howevei, it is unlikely that most tianslatois would feel it is sufficient simply to pieseive sound, as in: Theie seem to be elements of meaning in (2.3) entice` () and cuil` () that aie not in (2.1). And elements of meaning in (2.1) peihaps the connotations of sadness, boiedom and fiustiation that long` suggests aie missing. Example (2.2) is tianslated by Anthea Bell (1992:n.p.) as: Bell desciibes Moigenstein`s poem as puie sound` (1992:n.p.), based on his own statement, but heie again it could be aigued that hei tianslation shows it is moie than this. The punctuation of the fiist line, which she pieseives, tells us that it is a question and (piobably) an answei. We cannot help but visualize whatevei objects the sounds suggest in the woild we know, based on the meanings of woids we know. The fiist woid sounds (to me) like a conciete object, peihaps an animal that is a cioss between a ciocodile and a dog, in both Geiman and English, wheieas the second appeais to answei that it is some-thing abstiact that combines appeaiance and mimiciy (a stiongei connotation in the English than in the Geiman). These images will be diffeient foi eveiy ieadei and it is thus possible to see the poem as a iepiesentation of the way poems woik: by embodying connotations that ieadeis will contextualize in vaiious ways. What the